When Claude 4 was released, Dwarkesh Patel released a podcast with two Anthropic employees who claimed that, even if today's algorithmic power stalled, they'd eventually have enough data to automate most white collar jobs in the next 2 to 5 years. They joked that one of the bad outcomes of this rapid change in the social fabric would be about a decade's worth of humans being most useful as meat robots, until robotics get up to speed.
They also advise people currently doing white collar work to prepare for the near future by asking themselves "If you had 10 engineers working for you right now, what would you have them do?"
I found that question thought provoking, because it made me think about all the silly start up ideas I've had over the years where a steady supply of engineers would have been welcome to at least try and get started.
Ok so, those with an entrepreneurial spirit could be fine. What about everybody else? If you're not entrepreneurially minded, what's your post-automation plan, especially if you're young/starting out your career, and so unable to grow your wealth sufficiently by the 2-5 year mark to not care?
Well if we are degrading white collar jobs I think the situation will take care of itself. By and large an educated underemployed elite is what leads to revolution, and the revolution this time will be just to ban AI.
Such a revolution would likely relegate the US (or whatever country stupid enough to outright ban AI) to a distant laggard in terms of economic growth and worldwide competitiveness. Noah Smith recently wrote about how established developed economies tend to have more negative views around AI, perhaps due to a form of hegemon fatigue: current working generations in the US haven't felt their standards of living change that much over time compared to nations that have experienced rapid growth in the same timeframe, so they're not as open to the idea of vast technology-related changes that would make their skills obsolete. In countries that have grown in large part thanks to major technological breakthroughs like the advent of the information age, there isn't as much pessimism about vast parts of the economy becoming "progressed away". They will just adapt to the new zeitgeist, and outperform nations that aren't ready to step up.
Honestly I don’t think you guys have any understanding of economics at all. If all white colour jobs disappear then the economy collapses. At least try and explain how that isn’t the case, rather than just handwave about miraculous economic growth when we are all unemployed.
I think you're correct insofar as you basically *have* to have something resembling UBI if you get rid of most white-collar jobs; since the economy is primarily driven by consumer spending, even those who want an oligarchic dystopia can't just unemployed 40% of the nation and call it a day.
The UBI, unless we change the economic system, won’t be possible because government revenues will have collapsed. The only conceivable way for the economy to grow after mass unemployment of office workers is some mechanism of printing money and direct drops into people’s bank accounts of what would be newly created money, equal to the wages lost.
As it stands if we lose all office workers to automation the cost is not just those jobs but to jobs dependent on them, from construction to pretty much everything else. You can’t blow a 40% hole in the economy and expect no downstream effects. What AI enthusiasts don’t understand is that the potential productive capacity of an economy isn’t its actual output, technological regressions are not why recessions happen, and technological advances don’t end recessions on their own.
In fact far from there being money to spend on UBI, every country that adopts this or allows it to happen will be unable to even issue a bond, as they will have defaulted on their existing bonds. The government’s of the world will be bankrupt.
Arguments to the contrary are based on magical thinking about new jobs being created like previous automations, except those transitions took centuries while the proponents of AI say it’s 2027 or 2028 at the latest, and have no real explanation as to what replaces intellectual office work. The long shift from agriculture from to manufacturing, and manufacturing to office work took a long time, but some form of office work was already there.
A different argument from economists like Noah Smith is that comparative advantage will save human jobs. Employers use the better at everything AI in most cases, but the human, although having no actual advantage, has comparative advantage by being cheaper in some cases. This depends on the AI being expensive but it won’t be. To build out, yes, to run no.
(All of this depends on AI being good enough to do this. I don’t think it will be but I’m replying to people who think it will. And anyway it’s the abrupt transition that needs explaining. )
I think the UBI here is being proposed as a means of changing the economic system. Goods and services are produced by heavily automated privately-owned firms. Some of these goods and services are sold to the firms' owners; some to people who own land that's used for farming, mining, etc; and some are sold to UBI recipients. The firms pay taxes on their incomes, the investors pay taxes on capital gains and dividends, and landowners pay property taxes. These taxes then mostly go towards paying the UBI.
I'm not arguing for printing money; presumably the only reason you would replace workers with AI is if the AI was more efficient, that is to say, more productive, and thus more economically beneficial; you then tax the greater economic gains and use that to fund UBI.
It may be that, on net, this doesn't work - that with taxes, the AI costs more - but in that scenario you ideally just don't use AI.
This is all imagining a 'rational actor' government, which I am not suggesting we actually have, I'm just saying what the hypothetical sane response would be.
>The only conceivable way for the economy to grow after mass unemployment of office workers is some mechanism of printing money and direct drops into people’s bank accounts of what would be newly created money, equal to the wages los
How about we tax AI at a rate comparable to that at which we currently tax white collar workers. So government tax revenues don't need to fall at all. In fact, if AI is all that much more productive, it can affords to pay even more in taxes than the current white collar workforce.
AI will generate no economic collapse as long as we can summon the political will to ignore the people claiming that any interference with the AI express train will leave us all speaking Chinese.
But as a dystopian oligarch I'm not going to be willing to just pay people money to do nothing -- a huge swath of idle civilians getting sit-down money is a recipe for trouble.
Imagine you were so rich that a full time salary for a normal person was like ten cents a year to you. You'd find ways to employ people, right? I could employ a thousand gardeners just to ensure that every blade of grass I see is properly aligned.
I agree, giving the majority of the population free money to do nothing would be a disaster. Just as a starter, drug and alcohol addictions would go nuclear, and I would also predict that there would be a lot of clinical depression.
Much as we may hate to admit it, our jobs help give us purpose in life. And having free money with no purpose...
Dispossessed white collar people losing their status overnight don't want to go on welfare, so UBI isn't stopping the mass riots.
And beyond that, UBI doesn't actually DO anything for the economy. If I give somebody $100 and they decide to buy $100 of stuff from me with that. That's no different from me breaking my $100 window and offering a passing tradesman $100 to fix it, no value is created.
People come to this conclusion that "oh well the economy needs people to buy things, so somebody will have to give them money, therefore UBI" because they are assuming the existence of a consumer goods economy. But the economy is going to produce things that either the producers need for themselves or which can be traded for stuff the producer actually needs, so why would that continue to be consumer appliances and electronics and all that when consumers' have nothing the producer needs? You can't trade your time and labor for money anymore, so you have nothing anyone wants, and the economy is not going to be producing things to sell to you.
Instead, what I'd presume happens is that people move down to jobs that previously weren't worth doing because they generated too little value compared to the available alternatives. Long term it's probably too much to hope that you can do a mental job for less money than it costs to run an AI agent, so this would probably consist of high-variance manual tasks that are harder to automate and/or for which a human is preferred. Domestic servants, in-home health care, and very low-level nursing like CNAs, presumably other things in this vein that we aren't aware of because nobody would bother to offer such services today but which will become attractive as soon as the value of your labor drops down to that level.
A UBI is presumed to drive a consumer goods economy. So that does something for the economy. It also solves part of the distribution problem.
Clearly it would need to be supported by taxes. (Even if you're printing money, you need taxes to drive demand for the currency.) It might be impossible to implement well for reasons of power and incentives, but it seems to work in simple steady-state scenarios.
First, There will be less work, for example, accountants. There will still be accountants, just less.
And then SW engineers. Maybe just Juniors, moving up. Then HW engineers.
Then Secretaries, the ones that are not just CEO status displays.
But you can't move to UBI - You still need Plumbers and constructors and electricians and Garbagemen - all the work you don't have at a Factory, and some jobs at Factories that are just not worth it to Automate.
You will have Jobs that could be done by AI but regulations will stop - like Lawyers.
Anything that AI can do, not regulated, and worth enough - AI will do.
So, what is the endgame?
Lots of bullshit jobs - You can have more lawyers, more regulations, alongside AI + actual jobs like plumbers that AI can do.
Small amount of People will actually create lots of value. The most/creative/lucky/intelligent. Once they do it AI can learn and duplicate it.
It's unlikely that the route to a global AI ban would include individual countries instituting bans unilaterally. Some combination of international treaties and technical solutions is more likely.
It's unlikely that international treaties would be signed by certain countries, ie the big powerful ones, given the implication of using AI to further defense and offense capabilities.
Disagree. Given the pressures mentioned above, plus the extensive mundane risks and social harms of AI, plus the inevitability of human extinction if AI progress continues indefinitely, plus the arguable "are we enslaving and killing billions of sentient beings by using this?" ethical concerns, I doubt that when deciding whether to have an enforceable treaty eliminating strong AI globally (signed by all the relevant powers), the world's leaders and peoples would say "actually no, the need for the world's militaries to spend trillions on zero-sum games should tip the scales in favour of keeping AI".
TL;DR, ai based investors will increase demand for human labor, but they’ll want to treat humans well because that optimizes human productivity at tasks humans are best at
I think in the ten years since you wrote that, we've seen how AI is developing well and all the "who could have possibly foreseen this?" ways too.
"“Besides,” the AI investors will respond, “we have a better track record at investing in women and minorities than the old human networks did.”
What are old Internet saying me once heered? "Get woke, go broke"?
"(G)iven that the law is like a human attempt at formal language, the AI’s are masters of the law now, and draft complex legal arguments that the best human scholars agree are totally sound.
But I see they still suffer from the Greengrocer's Apostrophe 😁
"In Ayinde, R v The London Borough of Haringey, the High Court confronted a startling misuse of AI-generated legal research. In that case, the lawyers submitted legal arguments that relied on five fabricated cases – among them, one purporting to be from the Court of Appeal. These citations bore all the hallmarks of real legal authorities: proper names, plausible citations, and a style that appeared entirely authentic. But none of them existed. The citations were false; the cases were fictional."
And your vision for the future of humanity is "get a job in the service industry, where pay and conditions will be even shittier than they are now because now there are too many people chasing such jobs so the AI employers can pay you peanuts, you hairless monkeys":
"This means that humans have jobs like “smile and be kind to people who are having trouble with the product”, “help your friends learn to integrate the device into their lives”, “have parties where people see how fun this new game is”, “host a large dinner party so that people see how good this food is”, “write songs that show the emotional experience of being human”, “comfort a person who is troubled being human.”
Now, what were the figures I think we discussed on here previously (or if not here, somewhere else) about how many people do, in fact, make the big bucks as Instagram influencers, Onlyfans creators, Substack writers, etc.? Not as many as people think when they set out on a career of "I'll have my own Youtube channel and be famous!"
'Have parties where people see how fun this new game is' - oh you mean like Tupperware parties? Tupperware that has now gone bankrupt?
'Smile and be kind to people who are having trouble with the product' - oh you mean like customer service agents, who are already being replaced with AI chatbots?
"(I)magine a CEO who can simultaneously say “we will increase user retention” to the board while telling a low-level customer service rep who’s speaking English — “this customer sounds upset, i see that her usage drops off 30% on Tuesdays, and this API call to Comcast shows that there’s a service disruption in that neighborhood which seems to peak around that time due to congestion. We’ve credited her account, please tell her that we will talk to Comcast to resolve this issue”
That low-level customer service rep with ESL will be replaced by said chatbot, because it's more efficient and better as the rest of your post points out about replacing humans. Why spend those valuable seconds instructing the hairless monkey on what to do when an AI chatbot can diagnose all that for themselves and do it immediately?
If my future employment depends on "smile and be nice to people" then I may as well jump off a cliff now. I've worked retail before, and as God as my witness, if I have to lie, cheat, steal or kill, never again (to adapt a quote).
No idea if apxhard has ever done call centre work but from that description it don't sound like it: 'make your living hosting dinner parties', forsooth!
“write songs that show the emotional experience of being human”
We think the ratio of buskers to pitches on streets is bad now? Wait until it's the only job left for humans! Though it will see a revival of the blues plus a boost for country music.
"Woke up this mornin', my Cybertruck was gone
Said I woke up this mornin', my robot dog gone too
I wanna be a window cleaner. Pays nice in my country (Denmark) and it's not as taxing for the body as most blue collar jobs, you sit and stand in pretty normal positions, no very heavy lifts, etc.
Really kind of none. I try to do the best work I can and be useful to my employer, and let the chips fall where they may.
If the more moderate predictions about how my industry (SWE) will change come true - LLMs increase productivity, we all write code less and design more - I'll probably be in an OK position. I'm self-taught so a little more pragmatically minded and a little more comfortable with facing stakeholders and playing product manager than most "pure" engineers.
If the more optimist predictions come true - white collar work is gone, UBI for everyone - then whatever, live the life of the mind, finish my reading backlog, learn photography and playing some instrument.
If the more pessimist predictions come true - white collar work is gone (or gone for all but the top professionals - for me effectively the same), no UBI, mass layoffs - maybe I'll go into trade school and become an electrician or a plumber.
The AI will still be running on electricity, so the large server farms and data centres will still need old-fashioned human electricians (until they improve robotics). Getting a job at the OpenAI plant may be the new company town model, where there is one major employer and everyone works for them or works jobs downstream of the direct jobs.
Why? I'd personally love to spend my time in art, sports, curiosity, and community. Profit signals have little relationship with the tasks that I find fulfilling.
"...live the life of the mind, finish my reading backlog, learn photography and playing some instrument...."
It's kind of funny, but that's my life now more or less. Earlier retirement than expected, UBI is my stock portfolio, no photog yet but working on my Spanish. Guitar is a work in progress... Can't say it's dystopia really.
Though it seems plausible that AI could automate the essential tasks of software design, as well as the incidental / implementation. Already, if you provide enough context and requirements, LLMs can generate a solid software architecture. If the problem is completely new, and has never been solved before, then an LLM probably couldn't design an architecture for it. But completely-new software requirements are rare.
Yeah, no. I'm calling bullshit on the notion that most of today's software development effort is in essential complexity. Gathering and clarifying requirements is the only essential part from the project manager's point of view, everything else is incidental complexity. Squeezing the customer for information is an important part of the project, but it's nowhere near 10% of the total design, implementation, and test effort, more like 1%.
There are countless software ideas that can be reasonably well specified in 10 minutes but take several days to implement and test thoroughly. This means that there's potential for two orders of magnitude improvement, or two more silver bullets by Brook's definition.
Here's an example, if you don't believe me: I can easily specify the concept of an HDL simulator in 30 minutes in sufficient detail if I refer to the VHDL and Verilog language standards and to existing simulators to fill the gaps, but it'll take a very competent team of ~10 experienced software developers and domain experts at least a year, most likely longer, to build a state-of-the-art simulator. That's four orders of magnitude between essential and incidental complexity right there!
"They also advise people currently doing white collar work to prepare for the near future by asking themselves "If you had 10 engineers working for you right now, what would you have them do?"
I think this is the blind spot of software engineers in general and the AI set in particular, seeing all the products like Copilot being shoved out to the general public: they think of all work as 'something you need an engineer for'.
A lot of white collar work is administrative; would "10 engineers" really be useful in an accountancy firm, for instance? "Oh but your AI assistant will write emails for you". Great, but often those emails are requesting specific data in a specific form and I don't think I trust AI at present to do that task.
If all your emails are "meeting at 10 am" and you're replying to confirm the reply about the reply, then sure, AI can handle that. I don't think as yet it can handle more complex tasks. But I think this is where businesses imagine they can increase productivity, or at least cut costs, by turning things like customer service over to AI.
I had a recent tussle with one of those chatbot AIs replacing the human customer service agent, where the website tries to make you have a 'conversation' online. The problem was that I could not log in to the newly set up business account with this particular company. All the AI was able to do was recite the script: log in to your account so we can confirm the details. But the very problem was not being able to log in!
After going round the houses with this to no avail, I eventually managed to find the (well-hidden) phone number and got on to a human after a period of time. That's the frustrating experience with AI so far and that's the cheap, replace the human labour, solution companies are presently going for. That AI certainly was nowhere near "10 engineers".
"A lot of white collar work is administrative; would '10 engineers' really be useful in an accountancy firm, for instance? 'Oh but your AI assistant will write emails for you'. Great, but often those emails are requesting specific data in a specific form and I don't think I trust AI at present to do that task."
For the accountants at the accounting firm, I think the question would be: "If you had 10 accountants working for you right now, what would you have them do?"
And unless the economy demands 10x the accounting, the answer implies that accountants move into accounting management or become unemployed.
I have historically been very bad at predicting "what comes next" but I don't see an obvious 10x increase in demand for accounting, law (maybe I'm wrong here?) and lots of other white collar jobs. I think AI optimism leads to "the folks who learn to leverage AI will stay employed and be even more valuable. The others need to find new work."
Note that this also applied to radiologists who are looking at x-rays, MRIs, Cat Scans, etc.
I don't know how porn star continues to be a job. And "porn prompt engineer" seems unlikely to be valuable, either.
For some types of engineering, 10x more engineers (especially if they are cheap AI engineers) can make sense. For a lot of other jobs 10x more of that job ... doesn't make sense to me any more than producing 10x as many cars makes sense.
But I don't know what the "new work" is. I'll note that in Victorian times a larger percentage of the population was employed as "staff." The US has kinda movec that way over the past 40 years anyway; we have a lot more personal trainers and life coaches than in 1980, for example.
I don't think he's saying that the AI revolution will make people horny for robots; I think he's saying that AI-generated pornography will be superior to "real" product as well as cheaper, thus outcompeting actual humans in this field also.
People are already paying for AI art of their specific fetishes. AI-generated video of hot young things with flawless skin, impossibly large yet still perky bosoms, thigh gaps, and the anime girl expression of mindless lust for *you* which means she'll do anything you want will beat out human performers who will have to rely heavily on filters and editing, and still won't be able to match "the 100 specific elements in my pr0n that I need to be able to achieve orgasm" that AI can produce by the cartload.
And if there's a sufficiently convincing AI Onlyfans cam girl? People were losing their minds over losing their Replika girlfriends. There's a market out there of vulnerable/lonely/needy people who will fall for the AI girlfriend experience.
You know you're never going to meet up with that Onlyfans girl in reality, but she gives you a parasocial relationship. If you don't know Sukie is the AI generated fake, what is really so evident that she's not real if she talks back to you and performs for you and pretends to care about you like the flesh and blood models?
> For the accountants at the accounting firm, I think the question would be: "If you had 10 accountants working for you right now, what would you have them do?"
And the answer is: "I would get them to double-check the output of what my AI does".
This seems very likely to me. We'll have lots of people around and have many reasons to want to keep employing people (PR seems like a bigger reason every day) and we likely will not trust AI enough at any point to just let it work without oversight. Either it makes mistakes and we need oversight to keep it functional, or it doesn't make mistakes and we fear it taking over. A future where millions of capable people stand around doing nothing while giving the AI control over everything seems like the least plausible scenario.
>I have historically been very bad at predicting "what comes next" but I don't see an obvious 10x increase in demand for accounting, law (maybe I'm wrong here?) and lots of other white collar jobs.
Can everyone afford to keep a lawyer or accountant on retainer currently? If the answer is "no" (and it certainly is) then there is demand that has not been satisfied, provided that the cost goes down.
"If you had 10 engineers working for you right now, what would you have them do?"
Uhhh... do these engineers know sign language? Nobody has been impressed thus far by any of the attempts at machine translation involving sign language.
Seriously, if I had 10 engineers working for me right now, I'd start them off on:
(1) That pile of shredding of documents that are now older than seven years and don't need to be retained anymore and must be securely disposed of
(2) Those two filing cabinets worth of files that need to be reorganised
(3) Overhauling all the staff files regarding contracts, making sure all documentation is up to date, in place, and signed
(4) In another month's time we'll be preparing for our annual audit, time to get all the files and paperwork ready!
(5) In the same vein, here is a year's worth of receipt books for fees. Enter them all into a spreadsheet and cross-check them against the credit card receipts, and cross-check *those* against the bank lodgements. Hurry up!
I'm seeing a lot of "physical media that need to be engaged with" and not a lot of "write a programme to automate this" in my job that AI isn't suitable to use nor capable of helping me.
What is the informational content or value of a couple of AI developers going on a podcast and saying the thing they have to say to not tank their company's stock value and piss off their bosses? Why take this any more seriously than e.g. a politician speechifying about how their policies and initiatives are going to have good results for their constituents?
I mean, you make it sound as if they were spouting wildly expansive unsubstantiated bullshit. The informational content is to learn where frontier models are likely to go in the short to medium term, with a sincere and collegial tone that makes the information transfer stick. Their claims aren't coming out of nowhere, they have been echoed by other frontier model companies and are the reason why hundreds of billions of dollars are spent yearly in capital expenditures to bring about the commodification of white collar work, a lot of which is repetitive, pattern-based and knowledge-gated, which is exactly the type of thing that RL built upon the current suite of models is improving at rapidly (and measurably). Human-equivalent tasks automated by frontier models have doubled in length roughly every seven months since GPT 3. If you think this trend is unlikely to continue until a full day's worth of work can be automated, the burden of proof is on you to show why that trend would break (blanket "can't trust corpos!" speak won't cut it). It is tough to reckon with the fact that people juggle with different goals and aspirations, and that personal ambition/enrichment is not strictly incompatible with giving away information that is useful beyond the enrichment. Or maybe we should tell biotech industry executives that their claims are informationally void before the FDA approves their drugs? "Show me the results of your current claims about the future first, only then will I finance/be interested in your project!"
The parallel with a politician speechifying is flawed, because politicians rarely build their speeches around easily verifiable up-to-date academic research and measurable benchmarks that point in one direction with a lot more salience than in another direction. Then, these same politicians harness the blind trust their constituents have in their (unproven, immeasurable) policing qualities to shift the goal posts endlessly, playing the blame game for decades with little to show for it. There is a much better mechanism at play with AI investments at the moment: the mostly free market doing its thing, allocating resources to the highest expected present value avenues. It isn't perfect, it has failed before, but it does a measurably ok job at generating somewhat sustainable economic growth in developed economies, which is a hell of a lot more than you can say about the democratic process in the past ten years for example.
The problem is that these guys have an incentive to talk up the likely near-term importance of AI, and also probably wouldn't be doing what they're doing if they didn't think AI was going to a have big impact on the world. That doesn't mean there's no information in their words, just that it is hard to imagine people in their position saying the opposite things.
Yeah I get that. But I think that when it comes to really serious matters, like AI or Covid, it is just evil to use the authority you have because of your job to broadcast misinformation. During Covid there was some MD's on Twitter that built huge followings trading in misinformation. If people like Patel can't give the public the truth they can at least shut their damn mouths.
I've worked as a software developer for ~30 years and I have never worked on a project bottlenecked by the number of engineers (at least not the way that term is usually used).
The bottlenecks have always been one of the following:
1 - collecting and clarifying customer needs / requirements.
2 - reconciling requirements with existing software.
3 - verifying that a given implementation meets the requirements and does not cause regressions.
It's not that engineering takes zero time: it just takes less time than the other three activities.
Given that software engineering is widely understood to be one of the most tractable jobs for AI, I'm inclined to heavily discount the Anthropic employees 'predictions. Maybe they've seen AI agents perform some of the critical tasks I listed above, but I have yet to observe them in the wild during the last few months of heavy use myself.
So far, all I would predict is that AI will make the engineer's experience much better. The gains may exceed the ones we got from any previous generational shift (1GL -> 5GL), but from all evidence I've seen so far, they are magnifying a human's power, not replacing us.
An ocean of human activity still lacks automation, and new cases are born every day.
We already do that. I imagine you're envisioning something like testers or QA folks clicking manually through a website that would then be replaced by an LLM doing the same thing?
Having humans dedicated to that is already not all that common, in many placed people write a suite of end-to-end tests that automatically runs in the pipeline or something.
Perhaps the AI could help writing that - or eventually write it itself - but that's basically the same as "AI will make coding faster", because tests are just some of the code you write.
I cannot speak for the entire industry, just the part that I'm working in. It's been my impression that the "automated tests are a waste of time" mindset is no longer common.
I've never worked on a project or company that's not done automated testing, and the worst I hear of my co-workers is along the line of "the client does not believe in unit / integration tests and we only get to write the other". It is possible that we just don't work for clients who don't test since the mindset difference is too large for us to be a good fit for them. But all the evidence available to me points to automated tests being accepted as the default software dev practice.
The classic book “The Mythical Man Month” goes in to some detail on why putting more engineers on a project does not necessarily get it completed faster. (Because they will need to do-ordinate with each other…)
But, with that caveat aside, I can think of plenty of projects where the bottleneck is getting implementation done.
Sure, it’s not the only thing that needs doing, so if AI gives us nearly infinity programmers, projects will not get comp,teed in near zero time because all the other stuff will now be the bottleneck.
But I can see the opportunity for substantial gains here.
There is a really wild thing you can now do to get quick customer feedback, which is not write the software at all but have an AI just pretend to be the app. This can get you rapid feedback about what is and isn’t really required,
Yeah, I've been thinking a lot about "the MMM" lately, and in some ways Fred Brooks is at the heart of my skepticism.
I imagine you're right that there are projects out there where the bottleneck is implementation. I can *conceive* of such projects. But I can't think of any I've been involved with, and I've worked for a lot of different types of software company (Though crucially, not in embedded systems or on projects that have exceptionally high safety standards.)
As you mention in your other reply, prototypes are an area that AI might offer enormous leverage. Balsamiq mockups made communicating about software a lot easier, and AI prototypes seem to offer an OOM improvement on that. Once you've gone that far, it seems reasonable to just iterate on prototypes until you have an actual application, but I think there are a ton of unanswered questions: feature discovery, feature conflict discovery, version control, taking or assigning responsibility, etc..
At any rate, I agree about "substantial gains", but I still see software makers running up against other bottlenecks very quickly.
> But, with that caveat aside, I can think of plenty of projects where the bottleneck is getting implementation done.
One area which might serve as an example (if we also shove artistic tasks into the "implementation" drawer) are video games.
You can start with a blank slate. No "it has to be written in Java, and it must interface with our ERP software and our database, and also deliver reports in that format etc". Instead, you can say "The game should run on any one of Windows, GNU, Android, iOS", thus not constraining the creative freedom of your LLM.
(The downside is of course that just like for other easily replicable art, you are always competing against the very best. There already exist more good video games than you can play in a lifetime, so unless you have highly specific interests, there is never a reason to play a game which is in the 50th percentile of greatness only.
This dynamic is why it is often easier to make a living writing boring software interfacing with ERP systems than writing video games. If a zillion people have written a dungeon crawler before you, they are all your competitors. If a zillion people have written a kinda similar piece of business logic software before, it is very likely that they were targeting slightly different interfaces and none of the solutions available will work out of the box, so your custom-built piece of code will not have any competitors.)
While I have not tried LLMs in a year or so, I am also doubtful that the current generation could implement even a simple game without help from the user.
Imagine, for example, a world in which Tetris was never invented, and thus not part of the training data. You spend a kilobyte of text explaining your game idea, and tell your LLM to just use ncurses so it does not get distracted with fancy graphics.
My gut feeling (p=0.8) is that the LLM would fail to deliver a working Tetris. (And, of course, a console implementation of Tetris would still not be very marketable in 2025.)
While I can not see LLM-created games competing with indie games, I guess that there might be a niche for games tailored to custom tastes. For example, it seems unlikely that the gaming industry will create your dream open world RPG where all the human characters have neon-green elephant ears, your preference is just too specific (sorry). "Build a neon-green elephant ear mod for Skyrim" might be a niche an LLM might solve in a year or so.
They are in the business of selling Claude ... Of course, who knows what happens in 5 years (we had basically no LLMs 5 years ago, at least not in the same sense we understand LLMs now).
But notice how the LLM developers keep scaling back their goals and expectations (or rather pushing them forward in time).
My job mostly revolves around making these models actually useful in production (possibly in combination with classical ML and other techniques). And there really are not that many use-cases yet.
Everyone who tells you otherwise (about the present, not speaking about the future now) is mostly bullshitting, frankly. I mean when we have some minor use-cases for customers we also hype them up a bit ... Although I also always like to use these webinars and meetups as an opportunity to point out how important the non-LLM-based approaches still are. Partly because it is true, partly because it makes is look more credible and level-headed in the sea of AI vibecode hipsters :D
I've been working in ML for about 7 years now (theoretical maths before that) and my opinion that even classical ML is still somewhat hyped. Don't get me wrong, there are real use-cases for that. But most of those are either commodities nowadays (most of computer vision) or recommendation engines or fairly basic use-cases like credit risk modelling in banks (logistic regression FTW ... and actually there's nothing wrong about that).
These LLMs are really still looking for a market that validates the huge investment put in them. There are some niche use-cases already but the added value just does not match the investment on Anthropic/OpenAI/Mistral/whatever side. So these companies need to keep hyping and hoping that they will genuinely have a proper breakthrough in the next 2-5 years.
I already replied once, being dismissive of the scenario Claude employees are putting forward ... but it is perhaps more fun to roll with it and imagine that it comes true (it probably will eventually).
I think the answer for most people's jobs is quite clear -services and arts.
With most white collar jobs automated, you will have many people who just have so much more money to spend than before ... and they will want to spend it.
One thing that humans will always be better at than LLMs (even with decent robots ... unless yo go full sci-fi synth but that definitely won't happen any time soon) is acting human.
And people prefer to go to restaurants not just because they have good meals there and maybe the place is nice but also because they get the service. If that weren't the case, people would probably either cook for themselves or order food home. It comes cheaper and you don't have to give tips to anyone.
Now arts - you could say that that's not for everyone, but actually it is not beyond most people to learn to play decent piano and go to that restaurant and play it for the guests (and get paid by the restaurant and from tips). Most people won't be the next Bach or Herbie Hancock but they can learn to play decent music. And many people definitely prefer to listen to someone playing (even if it is not perfect! it just has to be decent) to listening to a record (even if the record IS by Herbie Hancock).
So I would expect the economy to shift massively to services, broadly defined (so that even the musicians are a part of services). Also hand-crafted goods would boom. People already pay for something that is NOT perfect. Machine perfection is cheap, hand-made IMperfections is luxury and that is why you pay for it.
I would also expect a renaissance of portraits and other things which are really an over-expensive luxury today for most people. Again, it doesn't matter if a diffusion model can make a perfect portrait. Even if it somehow produces it manually with brush lines and all. It is still machine-made and therefore it will be considered low status AI slop.
Come to think of it, it sounds like a pretty cool future :-)
> With most white collar jobs automated, you will have many people who just have so much more money to spend than before ... and they will want to spend it.
<With most white collar jobs automated, you will have many people who just have so much more money to spend than before
Well, what you start off with is a whole bunch of jobless formerly middle class people. How and when do they transition to having a lot of money and leisure for enjoying the arts, etc.? And by the way, I have not heard one single proposal about how AI itself could assist with that transition, and I'm sure proposals for how AI could assist are possible. Could it assist in figuring out whatever gigantic change in government and finance and international relations would make possible universal basic income? Retrain people? Develop and implement some new kind of schooling for kids that prepares them to for the new order?
Latest really new proposal for AI I heard was the screenless personal AI whatnot that Sam Altman and that Apple designer hope the world will become addicted to. You carry it around and it's your shrink, your accountant, your best friend, you social secretary, your fucking wine steward. Maybe they will set it up so that it vibrates just right, too, and women at least can carry theirs in their underpants.
Radiologists haven’t been replaced yet and it doesn’t seem that close to happening. I would have thought pattern recognition would have been one of the easiest areas to outperform humans.
Is that your field? I remember hearing stories about AI image-detection programs outperforming radiologists on some tasks but I may have misunderstood that.
It is not my field, and there is some social aspect to radiologist troubleshooting requests that may be early days to have an AI replace (why do you need this study, I would suggest a different study etc…). But I’ve been reading things about AI alone outperforming radiologists alone and + AI.
Also AI can do wild things gs like predict diabetes and heart disease from unrelated studies like chest x rays through pattern recognition that is beyond human.
That's interesting, but I've also heard similar claims about how computer vision systems can "do" sign language now and when I've investigated they have been severely lacking in the ability to actually handle naturalistic human input (or produce ANY signed output) versus just recognizing discrete dictionary-entry-style individual signs under fairly ideal conditions.
I wonder if the same overstatement is true of the radiology situation. I'd love to hear from a real radiologist about it if there are any here.
"The team, led by Dr Fu Siong Ng, a consultant cardiologist, and Dr Arunashis Sau, a cardiology specialist registrar at Imperial College Healthcare NHS Trust, developed the AI-ECG Risk Estimation for Diabetes Mellitus (AIRE-DM) tool, using around 1.2 million ECGs from hospital records.
Using data from the UK Biobank, they then validated the AI’s ability to detect subtle changes in routine ECGs that could signify that someone might be at higher risk of type 2 diabetes, years before their blood sugar levels begin to rise.
...The AI will be piloted in the next year, and the researchers hope it could be rolled out in the NHS in the next few years."
"Like, for example, another chest X-ray study that diagnosed diabetes– which, again, I would never have thought that was possible, right– from the chest X-ray. It basically did this so-called occlusion, or masking, where it would look at the chest X-ray and block out various regions to find out what was the source of the information that we can’t see. And it turned out it picked up the fat pads in the chest that was providing this diabetes possible diagnosis.
But for the risk of heart attacks and stroke, we don’t know how it was so powerful and better than our standard. So this is part of that X factor that we still need to learn.
IRA FLATOW: Well, I’m still not quite understanding this. I mean, how did the AI– what did it see in the chest X-ray that told doctors– or it found that the patient’s risk for heart disease was greater?
DR. ERIC TOPOL: Well, we can tell from other studies using the chest X-ray for this opportunistic detection that it can pick up the calcium score– what people can undergo a CT scan to see how much calcium they have in their coronary arteries. But that can be derived from a chest X-ray, and that is an indicator of risk.
Also, the chest X-ray can be– through AI, determine the heart strength– the so-called ejection fraction. So it’s picking up a bunch of things, as seen in other studies, that are very predictive of a person’s risk. And it must be the composite of these things.
But we really don’t know because, although the study really was extraordinary, it didn’t do enough as far as the explainability side of things. So that’s where some of the work that needs to be done– and it needs to be replicated before this becomes a standard way to predict risk of a person’s heart or stroke in the ten years ahead.
...DR. ERIC TOPOL: That’s a really important point. It’s just like when you do scans for a person that doesn’t have symptoms, and you pick up these incidental findings. That could be the problem too– the false positives. So that’s why before these are put into clinical practice on a routine basis, we’ve got to know whether you’re going to get false signals. And so far the data look encouraging, but it’s still early."
So there is some evidence that AI can pick out data in the X-ray that does predict certain conditions, but there's not a model for how it does that, so any predictions will have to be treated cautiously in case they're false.
But Deiseach, I'm not sure it matters for practical purposes whether AI can tell us how it knows that somebody's scan shows a certain illness or high risk of an illness. If the AI is clearly better than the MD at identifying those people, we should act on that info.
Well, actually I can see some practical problems with AI not being able to tell us how it knows someone is at high risk for a heart attack. If we knew what signs it was responding to, we would have highly useful info about the body changes that raise heart attack risk. and that would give us ideas for new treatments. Still, when it comes to care of individual patients, seems to me that if AI is better than doctors at identifying people who are at high risk, we should go with the AI's prediction and start people on the regimens we know reduce risk.
I presume the AI is seeing signs on simple tests that correlate with changes that it has required doctors to get things like CT scans, stress tests, echocardiograms etc…. to appreciate.
Bottom line: Computational automation of jobs has been happening for a very long time, but it has always created more jobs than it destroys. There is no indication that this round is any different.
Why do you assume entrepreneurial jobs will stay? If anything, they could be among the first white collar jobs to be automated because all you need to do them is converge different problems by analysing a market and then organise employees to solve that problem. It's a job that seems to only persist because of the power dynamics, otherwise employees would just run cooperatives.
> Why do you assume entrepreneurial jobs will stay? If anything, they could be among the first white collar jobs to be automated […]
No, they can't be easily automated by AI, because they're not well-defined, self-contained tasks with a clear right/wrong evaluation criteria, and those are necessary conditions for training effective LLMs. Which is why LLMs get better and better at well-defined, self-contained tasks with clear right/wrong evaluation criteria, like coding benchmarks or math puzzles, but continue to disappoint on non-well-defined, non-self-contained tasks without clear right/wrong evaluation criteria, like designing and implementing complex software solutions based on fuzzy, incomplete, contradictory and unknown user requirements. Or entrepreneurial work.
> […] all you need to do them is converge different problems by analysing a market and then organise employees to solve that problem.
Those five words, "all you need to do", are doing _a lot_ of work here.
What you, a human [1], find easy or hard, has almost no bearing on what is easy or hard for LLMs respectively other AI architectures. For example, LLMs have gotten really good at solving math problems which are so difficult that only a small fraction of people even understand the question, while booking a flight from A to B in a web browser is still a problem on the bleeding edge for them.
> To be clear, you agree with the commenter that "guy who can start a business" will be the last to go?
No, I don't think they'll be literally the last white collar jobs to be automated. I'm arguing against your claim that "[entrepreneurial jobs] could be among the first white collar jobs to be automated".
How is entrepreneurship supposed to be a solution? In what fields specifically? Are they saying to just set out to "hire" AI instead of humans and find a field to be useful in? Or are they talking about software specifically?
I think there's a real bias around here toward assuming that as goes software development, so goes all of business, but I am very skeptical of that notion.
Has it occurred to these Anthropods that very few white collar workers would be able to come up with suitable projects for 10 engineers, or supervise them — none but other engineers, and people with lots of expertise in stem fields. if Dwarvish Petard is right, the next stage of AI, rather than nudging life towards cyber-utopia, fucks over everybody except for people who are a lot like AI developers. I’m shocked, SHOCKED, I tell you, to hear that things might play out this way.
I didn't see any new data, and I didn't see any new arguments, and I didn't need anyone to prompt me to notice that COVID had been a pretty big deal, so no.
the graph that compares excess deaths with covid-attributed death surprised me. I knew there should be a correlation, but I was surprised how strong it was.
At first glance, I was surprised how many people in the comments were irrational [1] about their positions. I expected these people to be rare in a rationalist community. At second glance it was just about 3 people writing lots of comments. After blocking these, the commentsections returned to normal.
It was very interesting to read the remaining discussions, especially the one linking the john-hopkins meta-study about the effect of lockdowns. I changed my views from "lockdowns saved a lot of life" to "lockdowns saved very few lives, especially when compared to voluntary social distancing".
[1] "irrational" in the sense, that they had some weird ideas, and were unable to defend them when pressured with facts. Thanks to everyone who argued with them even if you may not have convinced the person you argued with.
"Thanks to everyone who argued with them even if you may not have convinced the person you argued with."
Only tangentially related, but I do keep in mind the silent/observing audience when I'm going back-and-forth with someone online. My position and/or data may not matter to the person I'm interacting with, but it might for some in the silent audience.
I also saw that graph for the first time and it also surprised me. I had entertained the idea that many reported COVID deaths were not legit for whatever reason, but if that graph is correct then that is much less likely.
Thanks, I hadn’t seen that. It seems pretty clear there that Scott at no point believed the 20% figure. He makes it clear he had significant doubts, but acted as if it were true because of the lack of countervailing evidence and because it was the fail-safe rather than fail-deadly option.
This part seems to make it clear:
“I was suspicious at the time, saying:
> "This is a weird pattern – why are so many young people getting hospitalized if almost none of them die? ... Are these an overestimate? Maybe most cases never come to the government’s attention? There’s some evidence for this ... So hopefully the 20% hospitalization rate will prove to be a worst case scenario, and the real number will be less."
...but absent 100% proof that this was the explanation, I elected to stay inside rather than get a disease that had some chance of a 20% case hospitalization rate for young people.”
I expect that lots of people want the answer, but don't want to spend the effort required to dig through the data. I don't find that position morally questionable. I just cared enough in 2022 to want to answer for my own satisfaction the question "Was Covid just a bad flu season." I cared enough to think about how to answer this and then actually did some work to find and analyze the data.
But ...
A Great Courses lecture on economics introduced me to the concept of "rational ignorance." The idea here is that you might *want* to know something, but (rationally, in economic terms) conclude that knowing isn't worth the price (in terms of time, if nothing else) to actually find out.
I, myself, often would like to know something but not badly enough to do the work to find out.
Also, if one waits long enough then often someone else will do the work :-) And, if we are very lucky, also post the data, the sources and the analysis!
Sort of? My prior was along the lines of, "COVID deaths only very weakly correlated with excess mortality, therefore most COVID deaths aren't actually due to COVID." I wasn't strongly attached to this prior. It very likely came from a pile of cherrypicked / junk data surrounding COVID politics. I generally trust Scott to find "true" data and confirm it several different ways. So I've updated strongly in the direction of "huh, I guess COVID did cause all those deaths."
I didn't find the notion that "lockdowns made no change" suspicious, merely erroneous. Of course they had an effect. For example, COVID's lesser cousins common cold and flu almost entirely stopped spreading.
Nor do I find Scott's neat numbers suspicious---or at least not suspicious enough to investigate further.
Having said that, I'm not sure how a weak belief "excess mortality barely correlates with COVID cases" should have led me to suspect an effect of lockdowns.
Lockdowns ought to affect the death toll in a few ways. Let's assume that everybody got covid eventually. The upsides of lockdown would still be:
1. Spread cases over more time, so that the capacity of the medical system doesn't get overwhelmed at peak. This was the initial stated goal of lockdown (in countries which didn't intend to eradicate). Given that ICUs were overflowing in many places and given that ICU-type interventions did seem to be useful in some cases, this should be a nonzero effect.
2. Delay some people's cases until after they get vaccinated. This seems to be a much bigger effect, and the vaccines seem to have done a pretty good job of preventing serious cases.
There's enough numbers out there to estimate the overall effect of lockdowns in any particular place.
The question that began this thread was “did Scott’s covid post and their comments change your mind about anything?” If you want to debate
some covid-related claim, how about starting a new thread for that? If you keep it on this thread the whole thread will probably turn into more covid arguing, and I and anyone else interested in how much mind-changing went in will not get the benefit of a good sample of responses
Like several others, I have been lazily entertaining the idea that the death toll might have been off by up to 20%, but seeing the tight correlation with excess mortality dispelled that.
I also would have guessed there was a significant harvesting effect, especially in the US, where we entered the pandemic with an exceptionally vulnerable population. I'm less sure of that now.
No. I mean, I already knew Covid killed a lot of people
During the initial weeks of covid, I had a wide range of uncertainty about how dealt it mwas got be. Like, maybe it will have a 10% infection fatality rate. It fairly quickly became clear that we dodged a bullet here, and Covid only killed millions rather than tens of millions.
Nope. I never thought the official covid death numbers were all that inaccurate. However, I still think most governments' anti-covid measures were ineffective and actively harmful. Ie, I think similar numbers of people would have died regardless of the lockdowns, school closures, etc
Do keep in mind though that a lot of the arguing in favor of various pieces of misinformation was done by a few people who seem to have come to ACX solely in order to argue about covid. I’ve never seen their usernames here before. A couple of them have Substack blogs of their own, and may have come here mostly to get more visibility.
Am I upset that some grifters promoted their bull? Sure, a little, but it’s to be expected.
I’m disappointed that so few people engaged with the very interesting observation that Scott made: it really does seem strange that a million people died and we are talking about this other stuff. I think the two “biases” 1) dead people don’t tell sob stories and 2) controversy sells, make sense.
But I also think that there’s a lot more to say here about the psychology around this situation. And I think understanding it might better inform how to talk about Effective Altruism, which is important.
Maybe that discussion happened somewhere in the comments, but if it did, I could not get it to load while everyone else was enthusiastically embracing bias #2.
< But I also think that there’s a lot more to say here about the psychology around this situation. And I think understanding it might better inform how to talk about Effective Altruism, which is important
Many who engaged with the issue of the number of deaths commented that most of those who died did not have many years left to live, so that the loss of person-years was much smaller than the big 1.2 million number leads one to believe. And they said that with no emotion. And yeah, I get that. I don’t know whether that is the EA perspective. It would not surprise me if it is.
I find these comments extraordinarily painful because I am old enough that I will soon cross over into elder territory, or maybe already have, depending on how you calculate it. I am still healthy and energetic and work full time, but if you just go by the numbers then yeah, my life is worth way less than 90% of so of people on ACX. And if there was a lifeboat type situation, I think I would in fact give my place to somebody who probably had more years to live. (It would be awful, though, to watch the lifeboat leave. ) But to have *other people* taking it for granted I don’t get a place in the lifeboat, and not even having any emotion about it, makes me feel terrible. I’m not sure whether there is anything wrong with the math that leads to the conclusion that the deaths of older people just aren’t that big a deal. Maybe. I do think that in some ways I am the most valuable I have ever been. I’ve read and experienced and thought about a lot of things, and I’m much less self-absorbed than I was when I was younger, and preoccupied with developing my life. These days I’m most interested in distributing what I figured out, sort of like strewing fertilizer. Does that make it a bigger deal if I die?
And then I think about the account I read of Oliver Sach’s last months of life. He had a collection of samples of all of the elements, laid out like the periodic table, and he went through them in order, one per day. Said he was comforted by touching and ruminating about these eternal things. Also, during the last few years of his life he had a lover, after a 30 year period of having no partner and no sex, mostly because he was so shy. And the lover hung out with him through the end of his life. Are those last years of contemplation and companionship just not very important, because he was retired, frail and deaf during them?
I find ACX readers as a group to be exceptionally avoidant of emotion. Did you read the guest post Scott put up maybe 6-12 months ago by someone he knew who had an incurable brain tumor? The post was an extremely intense, dense, convoluted meditation on how the patient had to work with his mind and control his actions in order to partner with his treaters. It was very intellectual — sort of like some physicists theory of something like turbulent flow. I could barely stand to skim it, for reasons I’m about to explain, but had the impression that although the person who wrote it was extremely smart, there was some kind of cognitive slippage manifesting in what he wrote. It fell short of making sense.
But as part of it he told the story of his diagnosis and treatment and there was a crucial event that seemed to have sent him into developing his theory of patient-medical professional partnership. He was in the ICU, and kept moving around because of his discomfort. The movement made one of his IV lines come out, and a lot of his blood dribbled out of it onto the floor before the nurses discovered it, and he almost died. The nurse who discovered the problem screamed abuse at him. I can’t remember a thing about what she said, except that it was utterly cruel and unreasonable and hysterical, and that it wasn’t just a sentence or 2 but a whole tirade. And it was this episode of abuse that the man wound himself around. He saw it not as abuse but as a powerful clue about how the patient must act and talk with the staff if he and the staff are to be partners in healing him.
So I read that bit and it was clear that this poor, incurably ill man was horribly abused by that nurse, but because he depended on nurses for his life it was not an option to be angry at her and try to get her punished and removed from the staff — so he built this monstrous convoluted theory instead. I would like to tie that nurse to a chair and slap her til she cries and cries and cries, then fire her and make sure she loses her license too. And failing to do that I would like to be able to tell the guy that he was horribly abused and I will make sure that never happens to him again. But of course I can’t do any of that.
Anyhow, while I could only stand to skim the man’s article, I did read all the comments, and everybody got into I’m-smart-as-hell-too mode. So people either posted comments about what an impressively smart model the man had built, or else manifested their cleverness by suggesting fancy little tweaks it needed. Not one person said a word about the nurse’s abuse.
Um, didn’t mean to write so much, but once I started this rant overcame me.
I understand the idea that if you're forced into a horrible choice where you have to decide whose life to save, you might need to prioritize looking at the number of expected future years of life. But it would be a horrible choice, and it could only be made with tears in your eyes. It would be a terrible mistake to then conclude that anyone with fewer years left to live has a less valuable life than those with more years to live. Human life is infinitely sacred, and those infinities don't lend themselves well to simple multiplication or basic comparisons of "more" and "fewer". Rather, all human life is equally infinitely valuable. I hope that this awful mistaken view is not held by as many people as it might seem from substack comments.
My parents are elderly. Early in the pandemic, I saw the numbers coming out of Italy, did a quick calculation, and determined that if they contracted COVID, there was about a 15% chance one of them would die. They were very careful and avoided COVID for five years.
What would they have missed in those five years? The marriage of their daughter. The birth of their first grandchild, and time to watch him grow into a thinking, speaking little person. Five years to be with each other and support each other. And time with me, which I value extremely.
I write this with tears in my eyes. My comments elsewhere don't show my feelings on the subject. But I do have them.
Triage is hard on emotions. There's a sense in which the whole effective altruist project is a practice in triage. It makes sense that norms around that project have evolved to avoid emotion.
Having met several EAs in person, I do not get the sense that this detached way of thinking blunts their emotions in real life. (Unlike some MDs I've met who became callous after working in the ER for a few years)
Appreciate your comments about the elders in your life. Interesting point about EA and triage — yeah, it’s like the whole project is triage writ large. I think what offended me about the ACX comments about covid deaths was that it was *retrospective triage* — the covid deaths count is less worth grieving by because most of those who died were old. Those comments also bother me because it’s clear we could have had far fewer oldster deaths by taking steps that would not have inconvenienced the rest of the population — so certainly not “lockdowns” or required masking for all.
It makes me wonder about what life was like in the late 1940s. Did people get back to life quickly and only occasionally say things like "Hey, remember when we had that war? That was weird, huh?"
Reminds me of in interview of two english women a few years ago. They had just turned 100 (twins) and one said, "life was great, except the war...that was awful."
Your comment made me realize something I’d never noticed before. Labeling things as “weird” is actually a pretty recent phenomenon. My parents spent their formative years in the pre-weird world, and spoke about world events with labels from the old-fashioned emotion register: Things were pitiful or horrifying or shocking or heartening. I and my peers, though, took naturally to labeling things as different shades of weird. I remember leaving an outdoor concert with some friends when I was college age. We got into our car in the outdoor parking lot, then spent half an hour in a complex slow-moving mass of headlights before reaching the exit. “It’s a joke!” exclaimed the friend who was driving, “a hideously weird joke!” We might have used “weird” when talking about the news that some famous person had committed suicide, or the idea that if we had been born in an earlier era we might have ended up as soldiers doing trench warfare, or the possibility the world might be ended by a pandemic. What’s that about — the switch from responses on the human emotional spectrum to the “it’s weird” response? Something about increased detachment, something about drug experience . . .
"Weird" flags something as noteworthy without committing to an interpretation. It's an invitation for others to weigh in with their own take, positive or negative, before giving our own.
I think it is used more now because we are a more accepting culture. Then, once established, it became a habit of speech.
I'm reasonably certain China is cooking the books. I worked in Shanghai for two years, and the consistent pattern was that showing pretty much Anything Xinhua News sent to the Western press to my coworkers caused them amusement, and occasional laughter.
I and every non-CCP acquaintance from my time there suspect the numbers are much like India's, but they were attributed to other causes. China has an awful lot of "remote interior" where such things can blend into the background noise.
Seeing the Chinese numbers quoted unironically seems to be an error, but other than that the data are the data.
Sweden is also an odd case, not having much in the way of International borders. My recollection was that their numbers were lower than Norway, but I don't recall any analysis as to why that might have been.
A book-technology friend and I are trying to come up with the best metaphors for AI in the context of age-old metaphors for knowledge: light, Tree, sea, web/network, cabinet of wonders, treasure trove. New ones would be jazz (AI riffs) or my favorite: a ball of twine. All thoughts welcome.
Compost? You just heave all these odds and ends, including actual shit, into a container, and complex processes that only an expert in the field can understand turn the mass into stuff that’s profoundly rich and nourishing.
I don't think that your response to MR was wrong, it would take an extraordinarily generous reading of Cowen's original post to reach the interpretation that he expects claims to have intended in his response. It's great that he doesn't stand behind the reading that basically everyone came away with but I think he should acknowledge that he messed up.
It's telling that his defenders in the OP are defending the interpretation that you criticised, rather than the completely different one that he apparently intended!
TC is a fantastic interviewer, asking provocative questions and platforming an amazingly diverse cohort of intellectuals from many walks of life. I think he has a good sense for "big, if true," "overrated/underrated," and provocative and novel ideas. He's open-minded, for the most part, and far more willing to deliberately experiment with new ways of experiencing the world than almost anybody.
However, I really do not care for his commentary on the ideas he chooses to spotlight. I often feel he is attempting to trick me, playing 4D chess of status perception manipulation to "win the discourse." His writings on AI, rationalism and EA are a primary example. I also know that - probably as a result of focusing so much on novel, important ideas - the directions he chooses to highlight often have something methodologically flawed about them.
All in all, I see TC as sort of an intellectual venture capitalist, but what he wants primarily is status/influence, not money. He has a high miss rate, and to make up for it, he retcons many of his past comments when they get criticized or new information comes to light.
I still at least scan everything he posts on MR, because I think he does have, I won't say *better* taste, but *different* taste from most authors online, and that adds to my intellectual mix. I look for what seems interesting to me and don't depend on him as a trustworthy authority, because he's not, and I don't think he's trying to be. He's hardly unique in his flaws and he is highly unusual in his virtues. But one should not read TC expecting to get an accurate, thoughtful, objectivity-attempted, honest, in-depth, rationalist-style analysis. It's just not his brand.
It's horrible. It will make it harder to find genuine stories, especially here on Substack and on Reddit. We are not here to hear what a machine has to say. We are here for people.
I'm not necessarily here for people. I'm here for good, reasonably accurate writing. AI fails on that, too.
You bring back a memory.... I was in apas. (Amateur Press Associations). They're what people had before the internet. People would write zines, sometimes at considerable length, which would be collated together and mailed out, usually monthly. (Details available if anyone wants them.)
Anyway, I said something which failed at indicating that I was writing to humans, and I was asked how I visualized the people who wrote the zines. I said "as being 8 1/2 by 11 inches". And this was roughly true-- the truth is, I wasn't visualizing them as people, and my contact with the apa was the paper zines.
I'm rather better at people these days, but it would be fair to say that I'm presumably somewhat autistic.
I agree absolutely. Not because internet engagement is "personal" - while I know that's true for others, I am perpetually surprised when that turns out to be the case (e.g. people getting so upset with one another on the internet, in personal terms, although yes, often they are only trolls but I don't much trouble to distinguish), and I'm not sure if it's generational - the screen is only ever a screen to me - or to do with coming up through a somewhat rougher and more indifferent-to-feelings childhood/education (as we all used to) that I cannot feel that way about it.
It's because I am interested in what other people are thinking, and what they are thinking about, and books they've read, and Things They've Heard About that I wasn't aware of, the latter the same as if we were gathered around a domino table outside the store.
So to the extent the internet has destroyed as many good things as it created - I expect that AI may be its doom. (Not to give AI too much credit for that, as it would have happened anyway.)
"It's horrible. It will make it harder to find genuine stories, especially here on Substack and on Reddit."
Back in the days before Google and Alti-Vista, it was common for individual people to have proto-blogs. The internet community hadn't settled on blog as the term and these were often referred to as on-line journals or on-line diaries.
And these personal sites often included links to other sites that the individual author found interesting or useful. By their nature, each journal site depended on human curation (some of which was better than others). I've been wondering for about a year if the non-commercial web may drift back in that direction: Personal sites with links to other sites (And YouTube channels) that the individual finds interesting.
I had a webroll back then on West Coast Online. Everyone had two links on their webroll to friends, then the British Museum and another to Jim's Bicycles.
I think we can make it work here on Substack if the owners don't go too far in the direction of TikTok or Twitter.
I *also* feel that the web lost something (and a big something) as personal sites (usually hosted) were replaced by individual Facebook/Twitter/whatever pages.
I've got my own "journal," commercially hosted as I don't want to deal with the nuts-and-bolts of keeping a server up, but my own and not attached to Facebook, Twitter, Wordpress, etc.
It mooshes (that's a technical term!) together short posts, long form essays and links to other sites. I also use it to play around with HTML, CSS and Javascript :-)
It *can* be done. It is just more work than not. And I understand why most people have no interest in doing so. But maybe things start swinging the other direction as the interwebs become more hostile to actual content.
I hope. The internet would be more interesting with more content like that.
I have a WordPress site that I host myself. I designed the theme and wrote all the CSS. The number of visitors is declining, though, as I get harder to find. If I post the same thing here and there, I'll get 20 readers on my blog and 1000 on Substack.
I think Substack has it right... but it's definitely trending in the wrong direction with videos and chats etc.
I think it makes popular content completely unreadable. On Reddit, not only are most popular posts written by AI, but many of the comments too. X is even worse, as each post leads with @Grok Is this true? or @Grok What is this?
I don’t normally look at popular feeds, so I guess that’s why I find it so recognizable. I haven’t been a boiled frog, so the difference is more recognizable.
What do you mean? I think I never encountered an AI written post on Reddit and now I wonder if I'm wrong about this. Are those post marked as such or given out as human written?
Most of it is slop. A very small amount is good, some so good it passes for human written. But if we're getting everything written by AI, eventually we will be living in a completely fictional universe, because AI is not going to the old fishing hole with my pals Bert and Sammy and that's where I met Sal, Bert's cousin, and eventually we ended up on our first date and today we're married with three kids and a cockapoo.
Fiction writing might be different. It's not going to matter much if AI or Hume N. Beene is writing the story about the monster under the bed that eats chartered accountants and the plucky band of auditors that set out to track it down and destroy it. There's a lot of badly-written by humans self-published stuff on Kindle out there already, if AI can write better than that it won't be a tragedy to replace those.
A vague analogy here is those subreddits where people post stories (AITA, offmychest, TIFU, etc) which gradually devolved into a large fraction of the posts being pure fiction, no AI required. You can get more internet points if you gin up a totally unbelievable story than if you post mundane reality.
I'd say it's a natural part of the process, like every skill and art - you get better with practice. There are authors where I've definitely noticed progress over the years in their writing.
So starting off with slop need not be a sign that they'll never improve. But the problem with the self-publishing/small press firehose of content is that there's no-one there to guide them, the editing is sparse if any is done, and there's no feedback on "this works, this doesn't, you're not developing the plot here, there's the loose end you need to tie up".
I'm not necessarily complaining; there are some genre series self-published authors I read because I just want A Story and not the Great American/Irish/British Novel, and they give me serviceable prose with good plotting. The characters may not be deep and the story goes from A to B directly, but that's all I want when I want junk food for the brain. The frustrating part is when there are hints of good writing or plot elements that are left dangling, where with someone experienced to edit the book, the author could be encouraged to improve. Wasted potential.
It's ruining my Substack feed for sure. I'm bombarded with gender wars and dating scene-type content, presumably because I went on a Knowingless reading binge or something? On the rare occasion that I read any of it, I'd say there are clear signs of AI writing in at least half of them?
Occasionally I'll be presented with an interesting or thought provoking title, only to realize 2 minutes in that I'm either reading a completely AI-generated post or a severely AI-enhanced one. I recently stumbled upon a post called "Islam is the bane of civilization" or something close to that effect, written by someone whose name sounded muslim, which was enough to pique my interest. The entire thing was structured as "Thing is X, not Y. The essence of thing is to X, not to Y. I want people to do X — not to do Y. The reason for thing is X, not Y." The constant "this, not that" made me feel like I was going genuinely insane, and that post is one long slab of text.
I'm stuck between a rock and a hard place on here. I know there must be a large amount of truly high quality publications on Substack, but I'm not subscribed to enough to have my daily dose of solid articles in the "Following" section. So I tend to look through the feed, which is hot garbage at the moment. It's making the whole UX unsatisfying
would be curious to see if those articles will is talking about are AI. For e.g. all three of the samples that Will shows in his 'game' are highly likely to be written by AI with 99% confidence
It makes me far less trusting of anonymous writers who have not had a long, continuous presence online. And I pretty much instantly click out of any vaguely described "musings on reality" type blogs. On the flip side I am now a lot *more* interested in people with an established reputation who are blogging on something they have deep expertise or experience. It is a great time to write under your own name. Unfortunately that also means there are some topics that get written about less, since understandably people do not want to write about them under their real name.
I suspect that my preferences are unusual enough so that it is not profitable for any spammer to produce content optimized to win me as a reader (and hopefully a subscriber).
It is unfair that the AI content is not clearly marked as such, because it is a kind of fraud, if the customer expects human-made content.
My greatest worry about this is, how many accounts actually have the same owner, and whether the owner will leverage their position to e.g. push some political agenda on the readers. Then again, this already kinda happens with the mass media.
I moderate one of the splinter forums from the old subreddit. We mostly deal in culture-war arguments, with a brisk side trade in niche trivia. Our writing already tended towards the verbose, and while I won’t pretend that signified quality, it generally required some effort.
AI has reduced the time and effort needed to poop out a wall of text. Boring text. Text which removes all the usual incentives to engage and critique, because the author has no skin in the game, no reason to change *their* mind about anything.
I have yet to see one of our users use an AI to present their interesting technical discussion, or a story from their childhood, or any of the other ways to get an unfamiliar perspective. No, it’s always the culture warriors. The people who get the most out of AI are those who want their arguments to be soldiers.
Is anyone going to NYC tech week, or has recommendations as for what events to attend?
Open-access startup conferences aren’t really my thing, as the average level of competency is generally low, and 90% of the events are just trying to sell you something. I went last year, picking events that seemed interesting at face-value, and I was almost completely disappointed.
Open to recommendations, or advice on how to better pick from the very long list of events.
I agree with you that tech week is generally not worth going to. The real networking events happen separately from the advertised ones. If you know a VC who is hosting a smaller dinner, it can be worth going to that. Otherwise, skip it.
(It's very worth getting friendly with a few VCs if you are interested in startups, but tech week isn't how you build those connections. You build those connections outside of tech week, and then events like tech week are mechanisms to get everyone in one place)
Yeah. I live in Manhattan so it's no effort for me to attend. A hardware event I went to last year had some interesting robots, so there's that. I signed up for the Spartacus event recommended by Scott, otherwise I would have forgotten it was even this week.
I actually run a startup, but we're cash flow positive, and have never "needed" to raise, so my connections with VCs are quite lacking. Thanks for confirming my impression from last year, as I don't have to waste another few afternoons trying to see if my picks last year were just bad, or if it's the events in general.
Yeah. That was my experience last year. A total waste of time. I’ll be at the Spartacus event, so I guess I’ll see you there! Although you won’t know who I am. 😈
I was chatting about this with my wife. I expressed the regret that we had not helped our (now grown up) children enough (we helped them more than my parents helped me but…). My wife said that was ridiculous. Normal parents don't help their children with homework, university applications, exam review/revision etc.
I asked a friend and he agreed: kids have to learn that stuff for themselves. They are on their own. My view may have been distorted by our Indian friends. They did everything for their kids and even hired tutors and sent them to Saturday school.
How much should parents be helping teenage school children with their schoolwork?
if the goal is to get them into a top college for networking purposes, a lot
if the goal is to raise productive capable people, actually figure out with them what homework is worth doing, set a minimum standard for grades, and then teach them to be genuinely curious about the world and beat that standard
anecdotally, my parents never helped me with homework, but they did take me out of a physics class I was getting Cs in that I loved and the rest of my grades were suffering. i (after getting really really mad and not talking to my mom for months) stopped spending all my time trying to (and being bad at) learning physics, and out of boredom i took all the hardest classes and got basically perfect grades until i got into college, at which point i got perfect grades until i got the internship i wanted. and then i got Cs and Bs because i was prioritizing other stuff in life.
indian parents are great. it's the results of chinese parents but with 10% of the trauma, but it's not worth the effort in my opinion. i'd prefer to have had indian parents (when it comes to schoolwork, not in other ways) but i wouldn't be willing to raise my kids the same way as a trade, just too time consuming
I think helping your kids if they ask for help is good, but even then, there are limits. Checking their homework is OK. Helping to do their homework is not. My kids practically never asked for help.
My parents were only vaguely aware that I went to school.
What do you mean by "help"? If you mean give them the answers, zero. If you mean asking them guiding questions that help them figure things out themselves, or explaining concepts in a new way which helps them grasp such concepts, a lot!
I was a very stubborn teenager, and I would never accept nor seek help from my parents on anything that resembled guidance. I never asked them for relationship advice, help with homework, or anything else. I acted as if I could live my entire life without so much as a minute of their input. I'm in my late twenties and I regret behaving that way. If your children never wanted it, then you didn't do anything wrong. Some kids are just like that.
I left home and school at 16, so even after school, I never asked for advice. I always assumed that my kids were like me and wouldn't want help either. Now that they are older, they say I did OK!
If they are not asking for help, why are you worrying about whether they need help?
That being said, isn't your role as parent to make sure your kids get what they need, even if they don't recognize that need themselves? Moreover, based on my experience as a former teacher, lots of kids will readily accept help if offered, but would never ask for help on their own.
> If they are not asking for help, why are you worrying about whether they need help?
Because sometimes kids are stupid, and don't ask for help that they'd benefit from. Sometimes they don't know they need help, sometimes they're too stubborn to ask for help, sometimes they're embarrassed to ask for help, and so forth.
As a parent you have to balance the need to let a kid make their own mistakes with the need to intervene. This is a tricky skill and many parents get it wrong one way or the other.
We all make lots of mistakes as we grow up. When we get a little older we think "oh geez, if only someone had advised me to do X and Y and not done Z then things would have turned out so much better". Then we get even older and we have kids, and we think "Finally, a chance to do it over! With the benefit of my advice, my kid will have all the opportunities that I missed out on". And then you try to give them your brilliant advice, and they don't listen and they do something stupid again.
The answer depends upon what you think school actually does, and upon how much you care about society outside of your own family.
To the extent that school is signaling + babysitting, the tiger mom parenting style is privately beneficial but socially corrosive. Signals are valuable for society, but the optimal setup is one in which everyone collectively agrees to devote the same, relatively low resources towards the signaling process.
Once some tiger moms defect from this social arrangement and devote much larger resources towards the signaling process, two outcomes are possible:
The first is that the rest of the parents continue to act as they did before. In this scenario, the children of the tiger moms benefit (albeit with great effort) by gaining more access to better academic and career trajectories, while the children of the other parents suffer by losing some access to these trajectories. Society also suffers in general, both because the tiger families devote huge resources to zero-sum status games and because the quality of the signals are weakened for everyone.
The second is that the rest of the parents copy the tiger parents and become tiger parents themselves. In this scenario, the quality of the signal remains as high as it was originally, but society suffers due to the huge resources that are now being devoted by all families to zero-sum status games.
I can say I've never understood why helping with homework has been seen as a "basic" task in childrearing. Like "when you have children you'll need to bathe them, feed them, buy them things, later help them with homework" I'm always hold om? Why the homework part? I never particularly wanted help from my parents, and that's not because nothing was difficult for me. I actually struggled quite a bit at times. But teaching is a skill, a normal parent shouldn't be expected to have the skills to teach an eight year old with severe math anxiety algebra. That is a difficult, specialised skill. Much rather have my parents work a little more to afford a little tutoring.
Helping kids with stuff they don't even struggle with is something I don't remotely understand. Do something else for bonding.
I spent a lot of time explaining various ideas from economics, statistics, game theory, evolution, etc., to my kids, but in conversation and daily life, very seldom in trying to teach classroom-like material. I also spent a lot of time with my oldest kid teaching him math when he got stuck (leading me to need to go back and relearn a bunch of calculus, since I hadn't used it in 30 years), and have spent some time on math with my youngest. I also spend a lot of time chatting with my youngest in Spanish, since she's trying to get better at it. And discussing whatever they're interested in with them--current affairs and how they link to historical stuff, what it means when the news talks about the Fed raising interest rates, etc. (One aside: the AP US Government class was a huge game-changer in how the two who took it discussed current affairs, since they ended up actually knowing the relevance of a filibuster-proof majority or what the Secretary of State did.)
I think it's basically just a function of how smart your kids are. This board is mostly filled with former smart kids who didn't struggle doing their homework (but might have struggled with finding the discipline to actually do it). But less-smart kids can struggle with school-level work and need help, and I imagine it's pretty tempting for parents to try and help them out.
I always took the "homework" part as synecdoche for knowledge transmission more broadly, i.e., don't trust that regular school simply does that, and instead make sure your kids actually learn the things you want/expect them to.
If they're teenagers, just checking that they're doing the homework and finding out if there is anything they are struggling with. You may be able to help them where they're struggling or you may find you probably need specialised help.
Teenagers? They should be insisting, perhaps ensuring, that they *do* the homework.
Grade school? They should be assisting them at the homework. This doesn't mean doing it, but perhaps something like making flash cards and running them through them. (The big part here is them showing it's something they care about, but being present and working with their kids. So, no, a computer program won't work.)
I worked at a place that gave a (small) scholarship to a boy and girl each year, among those who had been volunteers with us. I recall it was the day that the not-arduous applications were due, and an Indian mom with whom I was familiar came in to pick up the application and ask about the (short) essay requirements or something. This was like 10 o'clock in the morning. She returned in an hour with it all filled out. As it was a schoolday it was clear she had written the essay or personal remarks or whatever it was. Her son did indeed win the boy's prize.
I admit she had a regal quality, plumply swishing around in her showy sari, that I didn't care for. After that, I might have started to admire her, for her dedication to her son's interests - or - I might have thought, this is an unseemly cultural difference to me, this brazen dishonesty.
I should add that I also have a favorable prejudice toward Indian people, or I should say the educated Indians* who come to America: they are so much more interesting to talk to, than my fellow kids.
*They are not the only ones I know, as I stay at motels occasionally, and prefer old vintage ones, of which Indians are sometimes proprietors.
I agree, of that generally "tech" sample - and yet - were it Indians who were on track to inherit America, I think it is indisputable that it could not then in any sense remain America. It must be Indian. That is not unique, though - equally true whether the future is Mexican or Chinese.
If we can have a Day Without a Mexican, I guess we can picture a Century Without an Anglo: where in the world, do very separate cultural groups: as fecund as Mexicans, as competent and hardworking as Chinese, and as civilizationally dominant as Indians - live together in a harmony without one being subservient to another?
They should be making sure the children do it, and nothing else. They need to stand on their feet, not yours. Unless you want to do their workwork for them in the future, as well.
A really great school, and the students get a lot of help and individual feedback from the teachers, and parent assistance is less important. I was fortunate to have my mother correct all my writing with me for years and this was extremely helpful. At my humble public school no teacher was going to put this kind of effort into my development.
Of course you should help your kids. Your wife and friend are ridiculous. Most of a child's knowledge will come from their parents and if not they will look for it on the internet (you don't want this unless it's controlled). It doesn't mean you do things for them, but rather that you be there for them and instruct them while they are learning to take care of themselves.
However, that doesn't mean you should feel guilty. You can still help them even now, maybe they insist they don't want it but that's when you enter grandpa mode and insist anyway.
*Should* is carrying a lot of water but, I would have greatly benefited from my parent (single parent household) helping me with college + networking. I was really clueless in this whole process and did not care enough to learn myself. As for homework and academic help maybe he could have pushed me to go to a better school? But I pushed back against the idea at the time so idk, should he have done it against my will?
I think there are two competing goals that may be in play.
1) Your kids become self-actualized and capable people, who can determine their own needs and pursue them.
2) Your kids reach the highest possible heights in the system as exists.
If you help your kids more, they will likely reach higher heights, but be less personally capable of navigating life and dealing with the unexpected. It's a tradeoff. Too much help from parents and they learn to depend on you instead of learning how to be independent. Too little help and they fail at minor tasks that slow down their growth.
I regret how much my mom helped me with college applications, taxes, and other paperwork when I was 16-19. It was work well within my ability level and would have taught me better how to do things myself, which I also had to learn later anyway. If I were applying to an Ivy school, it may have been more important or even necessary to get that help from her. Hopefully the higher overall requirements would have required me to work harder as well and still grow as a person even with the help.
I find it odd that many LLMs can still be really incompetent at task execution. My 2 year old already has more agentic behavior, in some areas, than SOTA LLMs can handle. What explains this? Instinct?
I think, that programmers often believe that non-programming jobs will be automated first, while non-programmers believe, that programming will be automated first.
I think that the argument is that automating programming has the better expected return because it builds on itself. It might well be easier to automate accounting or medical diagnosis, but doing so doesn’t make it easier to automate harder accounting or harder medical diagnosis.
I feel like programming is being automated faster than other fields because it's close to the knowledge and capability of the people building AI. To automate accounting, you would need a group of very capable accountants as well as a group of very capable programmers. To automate programming, you just need the programmers. It may even be better than that, as programmers are both the client and production, while accounting would be the client and would have to communicate to the programmers where some information is likely lost between them.
I'm a programmer and I think programming *tasks* will be automated first, but I also think that programming, and writing more generally, are especially amenable to our current efforts and that if anything people over-estimate AI progress specifically because they're disproportionately good at basic coding, and a lot of the people paying attention are programmers.
I think automating away *my job* will be pretty hard, because any decently-experienced engineer codes as an artifact of their real job, "thinking carefully about stuff".
The new paradigm is like a weird inversion of the horse->automobile transition. In that case, we went from dealing with live animals, with all the variation and unpredictability that requires, to machines that while they still had their idiosyncrasies and some unique characteristics, they were orders of magnitude more predictable and comprehensible than what went before.
This time around, we're going the opposite direction: programmers are used to -- maybe even were attracted to in their youth -- the absolute, discrete, knowable & repeatable behavior of computers (as they were in the before times) to a paradigm that is more akin to training an animal than to programming a Turing machine.
I don't think that all the programming jobs will go away -- not everyone knows how to properly train a horse -- but perhaps the skills and attitude necessary to succeed in this new regime will be quite different. [programmer!]
Actually, imx getting an LLM to exhibit agency and goal-seeking behavior starts with the prompt; fine-tuning improves things but it's not the spark that lights the fire.
The key is to turn prediction into action by giving the LLM a persona, querying "What would your character do next?", then feeding the answer back in the next prompt as an action taken.
I've used this simple principle to turn raw GPT3-style auto-complete models into reasonably coherent chatbots. I've also then used chatty models to exhibit agency by controlling a robot with a simple instruction scheme.
The key insight here (and I'm not claiming originality, this principle is well known) is that actions are simply predictions applied recursively to the agent making the prediction.
"What would an agent like me do next?" is subtly modified to "What should I do next?", which then becomes "Oh yes I did!".
In my opinion, there is a situation about the ongoing risks of long COVID similar to the past one about tobacco risks analyzed by Naomi Oreskes, in that many scientists in good faith minimize them.
However, experts with publications on long COVID on top Journals like David Putrino, explicitly deny that long COVID is psychosomatic
(Well-known physics popularizer Dianna Cowern, known on Youtube as Physics Girl, contracted a severely disabling form of long COVID even if she was under 35 and vaccinated against COVID).
Among well-known AI scientists, at least Ian Goodfellow and Michael Osborne suffer from it.
“With regard to the underestimation of COVID, allow me to point out what I think is the largest elephant in the room of public discourse, in politics, mainstream media, and incredibly even in the scientific community outside of long COVID specialists: the ongoing risks of contracting long COVID even for vaccinated people (only partially protected against it) and children (who can suffer permanent consequences like diabetes too). To grasp the extent of the underestimation, I think it is sufficient to read the following recent review article from Nature Medicine: https://www.nature.com/articles/s41591-024-03173-6
(from its abstract: “Long COVID represents the constellation of post-acute and long-term health effects caused by SARS-CoV-2 infection; it is a complex, multisystem disorder that can affect nearly every organ system and can be severely disabling. The cumulative global incidence of long COVID is around 400 million individuals, which is estimated to have an annual economic impact of approximately $1 trillion—equivalent to about 1% of the global economy.”)
("A new study that my colleagues and I published in the New England Journal of Medicine on July 17, 2024, shows that the risk of long COVID declined over the course of the pandemic. In 2020, when the ancestral strain of SARS-CoV-2 was dominant and vaccines were not available, about 10.4% of adults who got COVID-19 developed long COVID. By early 2022, when the omicron family of variants predominated, that rate [of long COVID for infected people] declined to 7.7% among unvaccinated adults and 3.5% of vaccinated adults. In other words, unvaccinated people were more than twice as likely to develop long COVID.
While researchers like me do not yet have concrete numbers for the current rate in mid-2024 due to the time it takes for long COVID cases to be reflected in the data, the flow of new patients into long COVID clinics has been on par with 2022").
It is not coincidental, given their knowledge of the syndrome, that the already mentioned Dr. Al-Aly and another author of the Nature Medicine review article, Dr. Akiko Iwasaki ( https://medicine.yale.edu/profile/akiko-iwasaki/ ) still wear a mask virtually everywhere.
There a huge amount of evidence that long COVID is significantly (if not entirely psychosomatic). I’m not saying it is, but that’s the way the best studies on the subject have gone.
Here’s an n=1 reason for taking long covid seriously. Twenty-some years ago I had something that was the flu equivalent of “long covid.” I had a case of the flu, though not an unusually bad case, but just never fully recovered. Tried not to take it seriously, but got worried when I realized a few months had gone by and I did not feel a bit better. My symptoms were fatigue, exercise intolerance, hypersomnia, malaise, muscle aches all over my body, and several extremely sore joints. And I had pretty bad versions of all those things. For ex., during that era of my life I naturally woke up after about 8 hours of sleep, and could not sleep longer no matter how lazy I felt, or how reluctant to face the day. But after the flu there was no limit to how much I could sleep. I could sleep 12 hours at night, and be craving a nap 2 hours after I got up. And if I lay down when I had a nap craving, I could fall asleep and stay asleep til a person or an alarm clock woke me. My 3 sore joints were so sore that if I bumped one even fairly lightly I’d make a little involuntary noise of pain.
My internist tested me for everything either of us could think of over the course of the next few months and everything came back normal. I did wonder if I somehow had a low grade, permanent version of the flu, but that was only one of my hypotheses. I had never heard of post viral syndromes, and my internist did not say anything about that being a possibility. So I am confident that it was not fear or expectation of “permanent flu” that caused me to have the syndrome I had. There could have been some other psychological thing going on that led to me developing a psychosomatic illness, but that seems unlikely to me. By the time I developed that syndrome I had lived through a lot of difficult things, and while they affected me, my health stayed excellent right through them. Sometimes stresses and losses put me through periods of depression or of anxiety, or caused me to have insomnia — but I don’t seem to have the wiring to develop psychosomatic illnesses in response to stress. And there was certainly no secondary gain for me in feeling sick all the time. I was a working single mother, and kept working and mothering right through it because I had no choice.
My syndrome lasted 3 years, then faded away on its own, and never came back.
This sounds like chronic fatigue syndrome/fibromyalgia which is known to be triggered by viral illnesses in many cases. A friend of mine had it for years, thankfully he has recovered now. Glad you are doing better.
I’m not just doing better, I recovered completely and totally, and never had a recurrence. Have had the flu a couple times since then, but recovered normally.
It's nice to read this account, it gives me some hope. A direct colleague of mine has Long Covid for two years now, without really getting better. He sometimes has energy for doing stuff, but only for very short periods per day.
Being in academia and in Europe, his position at university was exceptionally protected out of compassion, but his contract will not be prolonged beyond this month. Your account implies that there is still hope that he eventually will get better even after two years.
I know one person who had long covid for at least 18 months, then recovered gradually over about the next year and is back to normal now. She definitely was nothing like the kind of person those who think LC is psychosomatic picture — anxious, low-energy, neurotic etc. She was a happy 30 year old who’d finished her phd the year before, then gotten married and started a job she loved and was doing well at. And she had lots of energy pre-covid — played tennis, really good skier, gourmet cook. She’s back to that now
Something similar happened to my wife a few years ago. She spent several months after a bout of the flu just occasionally running a fever and feeling ill and having her heart race, for no reason anyone could detect. It went away after a few months.
I think this just happens occasionally with viral infections (I guess your immune response doesn't get turned back off properly at the end of the infection or something), and that a bunch of long covid cases are just people with this going on.
a) It would be shocking if some cases weren't real. If you had a severe case of Covid and wound up in ICU and nearly died then yeah, you're going to have some long term effects from that.
b) It would be shocking if some cases of Long Covid weren't psychosomatic. Any time there's a new disease, especially one with no test and vaguely defined symptoms you're going to have a lot of people with psychosomatic cases.
What's left in the middle? If you think that your knee hurts because of long covid rather than just getting old, does it really matter? It's not like knowing will help to guide your cure. If your aunt tells you that her knee hurts ever since she has covid then you should probably just smile sympathetically rather than argue with her, because what's the point?
Regarding the JAMA paper, among study participants who reported having Covid they verified the infection by doing an antibody-positive (Ab+) test within 30 days of survey completion. Thanks to this test they identified a number of participants in the study who reported not having Covid, but were antibody-positive: "Overall, 64 school-age children and 781 adolescents enrolled as uninfected but were Ab+"
The results for that group ("uninfected" but Ab+) were:
"Among school-age children, 6 (9%) met the index threshold whereas 18 (28%) reported experiencing at least 1 prolonged symptom. Among adolescents, 29 (4%) met the index threshold and 175 (22%) reported at least 1 prolonged symptom."
The index here refers to: "A total index was calculated for each participant by summing the individual scores for each symptom reported. An optimal index threshold for identifying PASC was selected"
So, how do those percentages (9% and 28% for school-age and 4% and 22% for adolescents) compare to the percentages in (verified) uninfected and Covid infected participants?
"Overall, 152 infected (20%) and 6 uninfected (4%) school-age children and 445 infected (14%) and 44 uninfected (3%) adolescents met or exceeded this index threshold"
"Overall, 45% of infected (338/751) and 33% of uninfected (48/147) school-age children and 39% of infected (1219/3109) and 27% of uninfected (372/1369) adolescents reported having at least 1 prolonged symptom."
Except for the above index threshold in school-age children (9%), all measures are about the same or even lower for the "uninfected" but Ab+ than for actually uninfected participants. This is supposed to be the blind group (unless I've misinterpreted something). If the hypothesis is right and long Covid is real, shouldn't they have the same or comparable percentages as the infected group and higher percentages than the actually uninfected group?
To summarize, they found a higher long Covid incidence (as measured by the index of long Covid symptoms and having at least one prolonged symptom) in children and adolescents who reported a Covid infection vs those who reported no Covid infection, BUT they also found LOWER long Covid incidence in adolescents who had Covid infection but did not report any Covid infection history.
I haven't read the paper, but taking your summary at face value, this could explain that result:
If most covid symptoms are directly caused by the immune response, and if long covid itself is caused by the immune response gone haywire, then one would expect (a) those with asymptomatic covid cases to be much less likely to develop long covid, and (b) on top of that, be much less likely to have developed one or more of the long covid symptoms from some other infectious disease.
> one would expect (a) those with asymptomatic covid cases to be much less likely to develop long covid
Yes. But look at the numbers for the uninfected group compared with the control group.
At least one prolonged symptom was present in 27% of adolescents and 33% of school-age children. Yet only 22% and 28% of "uninfected" but Ab+ (control group) participants.
3% of adolescents and 4% of school-age children had an index higher than threshold. 4% and 9% in the control group.
If participants with symptomatic Covid were more likely to develop long Covid than asymptomatic ones, why does the uninfected (No Covid) group exhibit higher one-prolonged-symptom incidence than asymptomatic participants?
(Also, the difference in higher-than-threshold index between uninfected and control group (3.21% vs 3.71%) in adolescents is not significant, according to Grok).
Because of my point (b). Covid is not the only disease that appears to cause lingering symptoms, and presumably the control group had just as many other diseases in their history as the asymptomatic group.
When long covid started happening and got a name, I thought, "maybe now the medical community will have to start taking these other lingering symptoms syndromes (fibromyalgia, cfs) seriously".
There’s ample evidence that “long covid” exists, and long term negative consequences of viral infections have been known to exist for a long time.
It’s also significantly more prevalent in people with a previous diagnosis of mental illness, which isn’t the case with basically any other physical disease. At least when there isn’t a direct connection, as there might be between depression and obesity. Anecdotally I have a half sister who is bipolar. She would frequently claim this or that malady, right before she would ask family for money, as she can’t work because of her illness. It was always something, and after catching covid, it’s been long covid for the past few years.
I have no doubt she actually *feels* sick, but it also seems incredibly likely to me that it’s mostly psychosomatic. Otherwise, her other illnesses have been magically cured at about the same time a new, unknown, illness makes her pitiful and in need of help.
I think long covid is a bunch of different stuff living under the same label. Some is psychosomatic, some is some kind of post-viral syndrome happening to a lot of people at once, some is probably some rare bad thing involving clotting, etc.
I would love for Scott to do a deepdive into this. It seems pretty bad and mostly ignored from a statistical perspective and pretty horrifying if you know someone with a bad base personally.
> Examples of recent papers on long COVID in children from top journals are the following
I've already commented on the JAMA published study, but I want to add that the second paper ("Postacute Sequelae of SARS-CoV-2 in Children") reviews several studies on long Covid. From the "Respiratory Manifestations" section (I did not check the other sections):
- "Post-COVID-19-associated morbidity in children, adolescents, and adults: A matched cohort study including more than 157,000 individuals with COVID-19 in Germany" (https://pubmed.ncbi.nlm.nih.gov/36355754/): Observational study, based on insurance data. "No insurance claim, no Covid".
If most cases are psychosomatic, which seems likely to me, then the campaign to scare people into compliance using threats of long COVID is an absolutely enormous pile of skulls. Intentional lies that caused needless, genuine suffering, for years at least, of millions of people.
Woah, I must say that I find this worldwide pretty insane. You have three pretty strong assertions here. The first is that most cases are psychosomatic. Fine, this may be true, we can discuss about this. Let's assume that.
But the second is that most people are also absolutely convinced that it's psychosomatic. This is the assumption that I find insane. Even if true, why do you believe that everything is absolutely certain about that? Why do you rule out the possibility that it's all psychosomatic, but that some people get it wrong, are honestly concerned, and that is the reason for their warnings?
And the third is that despite being convinced that it's psychosomatic, people start a campaign of intentional lies about it.
Let's take Scott. Scott has publicly declared that he believes only a small part of symptoms is psychosomatic, and most are organic. Do you believe that this is an intentional lie that Scott tells, and that Scott is part of the "campaign to scare people into compliance"? Do you really believe that this is more likely than Scott being honestly mistaken about it? Or perhaps even the third option, that Scott may be *correct* about it after digging into the topic and writing a blog post of epic size about it?
Sorry for not being nice, but I hope that it's at least important and true: you should really consider that other people hold different believes than yourself. Not all people who act against your own beliefs do that because they are jerks. Some may do that because they simply believe different things.
No worries about not being nice, I wasn't particularly nice in my post and your tone is totally fine to me anyway.
I don't see how I've made the second assertion at all. I don't think most people think it is psychosomatic. The intentional liars I'm referring to here are people like the CDC, who claim that 20% of infections result in long covid, but then in the fine print state that they are defining long covid as any symptom present six weeks after infection.
Scott estimates the prevalence of long covid at about .3% to 3%, in 2022. He says that these are probably not mostly psychosomatic. Now if survey respondents claim to have long covid at much higher rate, like 10%, it makes sense that the difference between Scott's estimate and the survey is psychosomatic.
I don’t know why everyone has selective memory on this topic but that is not representative of what Scott currently thinks about it. He updated his views about that ratio of organic to psychosomatic when long covid showed a correlation to bisexuality.
The correlation has apparently persisted to the point that some manager quoted in the Times joked earlier this month that he’d “believe in long covid when I meet a non-bisexual who has it”.
I'm sure PASC is a significant issue for a certain percentage of COVID sufferers. Some experts were predicting a mass disabling event, but that hasn't happened. The macro data doesn't show a strong signal that COVID has caused long-term disability in a significant portion of the population. For instance, the FRED tracks the number of people who report themselves as disabled. If we translate this graph into population percentage, the graph shows that the per capita rate of disability climbed from 0.08% to 0.09% in 2020. There was a steep dip in the number of people reporting themselves fully disabled after the pandemic began, but it climbed quickly. Now it's at 0.10%, but it's not far above where the trend line would be if we didn't have a pandemic.
Social Security disability claims were falling before the pandemic, and they began to rise after the pandemic, but they're falling again. If we translate the disability applications per month into a per capita number, the rate of disability applications is lower than it was in 2005. We did have some sort of mass disabling event from 2008 to 2012, when field office applications rose 40% in 4 years. (I've not yet found out what caused this huge increase, BTW.) But those numbers had dropped to pre-2004 levels by Feb 2020 at the start of the pandemic in the US.
As for anecdata, I only know of one person in my network of friends and acquaintances who had Long COVID symptoms that lasted more than 3 months. The exception has had them for over 2 years now.
I thought that you might be interested in the answers I obtained from Gemini (2.5 pro with deep research) and Chatgpt (o3 with deep research) to the following question of mine pertaining your comment (I am just sharing these answers, not endorsing them, at least for now, since I still haven’t had the time to check them):
“Would you comment on the following post I found on a blog? "I'm sure PASC is a significant issue for a certain percentage of COVID sufferers. [...]
whose PDF I am attaching. It seems to me that there might be a contrast between the two, and if so, I would like to know how it should be adjudicated.”
Therefore, the evidence strongly supports the conclusion that Long COVID has contributed significantly to, and can be characterized as a key component of, a mass disabling event. The term "mass disabling event" is justified by the large scale of individuals experiencing new or worsened functional limitations and health impairments documented in scientific research and reflected in broader population disability surveys, even if not all these individuals are yet captured by, or have successfully navigated, formal disability benefit systems.”
“In summary, the apparent contradiction stems from comparing macro aggregates to medical/survey evidence. The macro data (CPS counts, SSA claims) show at most a modest uptick, but the Nature Medicine review shows that long COVID has already affected tens of millions, with real economic impact. When viewed together, the evidence suggests that long COVID is a significant “mass disabling event” whose early warning signs may not fully appear in broad administrative series”
I’m convinced it’s impossible for Congress to ever significantly lower spending, for incentive reasons. The costs of lower spending are immediate and obvious. The costs of spending more are obscure and less obviously attributable to that overspending. The end result of this is that we should expect the dollar to go into a death spiral, as investors will look elsewhere rather than lending money on a 30 year term to a government that can’t ever bring itself to spend less than it brings in.
Can someone try to talk me out of this? I’m open minded on it and would like to actually be challenged.
Why is the only option to spend less? Why not tax more? After all, the current budget proposal that passed the House increases the deficit because of tax cuts, not spending increases.
How are they immediate an obvious? Top tax rate in the early 1960s was 91%, and it kicked in for people who earned about $2 million in today's dollars. But the economy boomed during the high-tax '50s and '60s.
The type of tax increase that would be required to fund our current level of spending would be large and broad based (as in, across the entirety of the income spectrum), and, just like spending cuts, would _also_ be felt immediately by everyone. The exact same incentives exist.
I think you were not being charitable enough and the tone of the comment, alongside the explicit call for having their mind changed, warranted steel-manning the position, especially when your simple answer was prey to the exact same dynamic they were pointing out. Even if they "truly cared about deficits" (which I disagree with your suggestion that they don't) enough to also consider tax increases, they still need to figure out the dynamic they are asking about (bad incentives) because they apply equally to both sides of the equation.
"I’m convinced it’s impossible for Congress to ever significantly lower deficits, for incentive reasons. The costs of lower deficits are immediate and obvious. The costs of increased deficits are obscure and less obviously attributable to those deficits. The end result of this is that we should expect the dollar to go into a death spiral, as investors will look elsewhere rather than lending money on a 30 year term to a government that can’t ever bring itself to spend less than it brings in.
Can someone try to talk me out of this? I’m open minded on it and would like to actually be challenged."
Libertarians (and paleo-conservatives) have a talking point about never giving the government access to money; “starving the beast”. Tyler Cowen had a recent blog post about this, arguing that people in favor of tax increases probably aren’t thinking clearly about a remittance tax, but he’s against it for libertarian reasons, not efficacy/distributional reasons.
`gdanning` appears to me to be (pre-emptively) reacting to the (unstated) idea that the most important thing is to avoid spending more, because government never does anything worthwhile with tax dollars.
Regarding `apxhard`’s original question:
Historically Democrats have (several times) *actually reduced the next-10-years deficit*. Sometimes through spending cuts, sometimes through raising taxes.
As far as I can tell Republicans haven’t done this in the last 50 years, although Bush Sr. came close and got unlucky with a recession.
The accounting for (real, not imagined) economic growth due to lower taxes on capital is often difficult to measure or find counterfactuals for, so it’s not always clear how much long term (i.e. greater than 10 years out) benefits are attributable to Republicans.
So I would say there is an existence proof for lowering spending under Democrats, not Republicans, and so I echo `gdanning`’s suspicion that `apxhard` has an axe to grind when stating something can’t be done. But I agree that steel-manning what they actually asked is better/more charitable.
William Niskanen argued that "starve the beast" was backwards: taxpayers treat the government like Santa Claus if they don't have to pay for spending. My view is that Ricardian Equivalence is false because people discount taxes paid by later generations.
This usually gets analyzed by just looking at Presidents, but Congress also has a say in the budget. In recent decades, Republican Congress + Democratic President seems to have been the best combination for deficit reduction.
It wasn't pretty (and came after a few hyperinflations), but it did happen.
Maybe your position is more accurately described as: "Congress won't get deficits under control until after AT LEAST one huge financial catastrophe involving hyperinflation" ?
But I'll point out another "fix": Default on the debt rather than run hyperinflation. Defaulting hurts a *much* smaller percentage of the population than hyperinflation and can be spun as only impacting the rich. Another variation is to not default, but change taxes on gov't bond interest to be some large percentage (say, 50%).
ANOTHER (more politically charged) fix is to helpfully roll everyone's 401(k) and/or IRA into Social Security. We aren't *confiscating* your retirement funds, we are just moving them into Social Security! The roll-in could even be at some reasonable rate because the point is for the gov't to get access to the savings NOW while only needing to deal with paying out MUCH LATER.
There are lots of ways to get things to line up better financially. They are just painful. But high inflation is painful, tool
This is completely beside the point. Substitute ‘net spending’ for spending and it’s the same issue. You just seem to want to argue for tax increases; fine, but a different topic really.
No, the point is that "spending" and "net spending" are two different things. The "net" literally includes levels of taxation. That's what makes it "net."
The same logic applies - tax increases are immediately and obviously felt. The benefits of a balanced budget are not felt by anyone in the present aside from a small group of weirdos, most of whom have decided to just buy bitcoin and thus now cheer for fiscal irresponsibility because they profit from it.
Congress is as incentivized to avoid major tax increases as it is to avoid major spending cuts, and major tax increases will diminish economic activity overall. Which, first, reduces the tax base so your tax increases don't get you as much extra revenue as you would naively expect, and second result in the sort of economic hardship that much of your government spending is meant to address so you find your non-discretionary spending goes up.
Spending cuts cause economic hardship as well, but they usually cause it among people who weren't paying a lot of taxes so it ends there. Tax increases, to be meaningful, have to come at the expense of the more productive elements of a nation. So if you're concerned about a fiscal apocalypse (and you should be), and you have finite political capital to spend, you get more bang for your buck going after spending cuts, and raising taxes only to the extent necessary to get the opposition to accept the spending cuts.
At least if you're taxing income, or "value added" or anything like that. Some sorts of property or Pigouvian taxes might not have this effect or at least not to the same degree.
1. Spending cuts also reduce econo ic activity and hence tax revenue. And in fact, since the marginal propensity to consume is higher among low income persons than higher income persons, the multiplier fot spending cuts is likely higher than for tax increases.
2. Where is your evidence that a small increase in the top rate has a negative effect on tax revenue? Or that it materially affects the willingness of "more productive elements of society" to engage in productive activity?
> Where is your evidence that a small increase in the top rate has a negative effect on tax revenue?
He didn't say it decreases tax revenue. Just that you don't get as much as you want.
You aren't going to close any budget holes with "small increases" in the top rate.
If you want to be serious about raising tax revenue, you need to be doing the tax regime that 37 of the 38 countries in the OECD do but that the US doesn't.
"If you want to be serious about raising tax revenue, you need to be doing the tax regime that 37 of the 38 countries in the OECD do but that the US doesn't."
Huh? USA is currently at rank 31/38 in government revenue as % of GDP
> After all, the current budget proposal that passed the House increases the deficit because of tax cuts, not spending increases
My understanding (someone will correct me if I'm wrong) is that is jiggery-pokery. The budget does actually include some (not enough but some) spending cuts, and keeps tax rates as they are.
The Democrats are referring to "keeping tax cuts as they are" as "tax cuts" but that's not the actual effect.
Under current law, the 2017 tax cuts will expire this year. So, the proposal to extend the tax cuts is indeed a reduction in what rates will be if Congress does nothing, and so is fairly described as a tax cut.
Those tax cuts were originally budgeted for as temporary, so pretending that they aren't to blame for the deficit because technically tax rates haven't changed is just as much jiggery-pokery.
In real terms, government spending as a percent of GDP has decreased in the past, from about 1981 to 2000.
It hasn’t in about 25 years, but most of that time was in the era of zero interest rates, and when someone is lending you trillions of dollars at a rate less than inflation, it’s no surprise that spending increased.
Yes this seems like the path that would be necessary. But can you help me reason through the incentives? The only people who voted against the recent budget were
- democrats who’d want to spend the same if not more elsewhere
- two republicans who want to spend less
From what I gather, they are all just following their political incentives, with the possible exception of the two republicans, and even that’s a maybe.
Can you help me understand the incentives that would need to hold for this situation to pass? Is it like AI+ inflation + Congress that somehow thinks their re-elections are more likely if spending doesn’t go up too fast?
I think these differences are largely rhetorical. Both sides want to reduce taxes on their own voters, and are happy to text the other side. The SALT deduction is a perfect example of this. 2017 lowered the top rate but capping salt deductions hit high earners in blue states hard. So it was a tax increase for some wealthy voters - high earners in blue states who could afford the expensive housing there.
Republican candidates talk about lowering spending, the same way Democrats talk about raising taxes on the rich: it’s rhetoric for their bases. Maybe some of them believe it. But given the chance, they’ve never voted it in. Is there a reason to think I’m wrong on this?
In the new budget, base spending is about $60 Billion less than 2024, and additional base spending is going to be upwards of $100 Billion less.
They are extended the tax cuts, but these have been in place for almost a decade at this point. Eventually they have to just be looked at as “our current tax policy” rather than ongoing cuts, and under that light every time there’s a budget without an increase in taxes we are voting for tax cuts, which doesn’t seem like the right way to look at it.
The Laffer Curve also shows that we can’t forever increase taxes without decreasing revenues, and while we’re probably not hitting the maximum yet, we’re definitely in the ballpark. I think there’s a strong argument to be made that the level of taxes in Western Europe are so high that they are actually producing less revenue than if the tax rate was lower.
Tariffs will also reduce the deficit. With an effective tariff rate of ~2.5% the US raised ~$75 billion. The current effective tariff rate is now around ~17.5%. While we shouldn’t expect a 7x increase in tariff revenue, as tariffs discourage imports, we can expect a significant increase in tariff revenues, which will reduce the deficit. Possibly by ~$200 Billion or more. People make a huge fuss about tariffs, and how they’re welfare reducing (and when there’s retaliatory tariffs this is certainly true), but the exact same thing is true for all taxes. Tariffs are simply a different form of tax, and while there are arguments as to why income tax or capital gains is more efficient, it will raise more revenue for the government, which is the exact solution we need to fix the debt.
Can you help me understand the “base spending” thing?
I ultimately agree on tariffs and think they’ve been turned into Literally The Worst. I understand the trade arguments but think they are the least bad form of taxes because they align incentives with what government ought to do: secure the border. Income and capital gains tax incentivize the creation of a panopticon and reduce investment and legal labor, while encouraging illegal immigration and other crimes.
Base spending refers to the part of the federal budget that happens every year under normal circumstances. That includes things like Social Security, Medicare, and the ongoing military budget. It doesn’t include one-time expenses like emergency disaster relief, COVID checks, or a new war those are considered temporary and not part of the government’s “baseline.”
When people talk about whether the deficit is sustainable, the key isn’t just how big the number is. It’s whether the government’s debt is growing faster than the economy. As long as the economy is growing fast enough, the government can run a deficit without its debt becoming a bigger problem.
More specifically, the deficit is sustainable if the total debt grows more slowly than nominal GDP (which is regular GDP plus inflation). That’s because if the economy is getting bigger at a healthy pace, even a large debt can become more manageable over time—just like a growing salary makes a fixed mortgage easier to carry.
A government can safely run a total deficit equal to:
Nominal GDP growth × Total Debt
That’s because the inflation and growth in the economy help “shrink” the real weight of the debt.
A Simple Example:
The U.S. economy (GDP) is $30 trillion
The national debt is $35 trillion
Nominal GDP growth (real growth + inflation) is 4% per year
Then the “sustainable” total deficit is 0.04×35 trillion=$1.4 trillion
So in this scenario (which is basically what the economy was like in 2024), the government could run a $1.4 trillion deficit each year without making the debt burden worse relative to the size of the economy. The deficit in 2024 was ~$1.8 Trillion, so we're only about $400 Billion off the mark. The new bill isn't actually reducing the budget, since it has some provisions for one-time payments, but over the next 4 years we could see that decline. When people talk about how the "new" tax cuts will add $5 Trillion to the debt or whatever, what they're really saying is that the "Tax cuts that have existed since 2017" are being extended, and if they weren't this would raise an additional $5 Trillion over 10 years.
TBH I'm not unconvinced that tariffs will close most of the gap between our unsustainable deficit, and a sustainable deficit, especially if inflation picks up a bit.
The problem is that our interest payments will continue to skyrocket, as we take on new debt, and old debt from 10 and 15 year bonds comes due from the zero interest days of the 2010s, so it's actually a lot harder to reduce the deficit to where it needs to be than it would seem just from existing revenue and spending.
Right - it’s the feedback caused by the increasing interest payments that I think kills the game. If you could keep interest low forever I get how you can keep riding the deficit as long as growth makes up the difference.
> Eventually they have to just be looked at as “our current tax policy” rather than ongoing cuts,
And the way we consider then to be "current tax policy" is when Congress votes to make them permanent rather than set them to expire before the 10-year CBO limit.
Is this an arbitrary date? Yes. But it's the arbitrary point we've all been using for generations. Things count as temporary or permanent. You can't double-dip to count something as temporary for a while so you only have to do the accounting for temporary, and then without looking it transforms behind the scenes into permanent without ever having to do the accounting for making it permanent.
The tax cuts aren’t permanent, but choosing to renew existing policy is a lot different than changing the existing policy to something less. From the perspective of “I don’t think the incentives allow Congress to stop overspending” the current extension of the tax policy is a lot weaker evidence than if they were changing the permanent tax policy to be lower.
Of course they did make permanent changes in 2017, but the debt-to-gdp was still double digits back then, and the 10-year yield was only ~2%, meaning there was far less incentive to reduce spending or raise more revenue.
I think you are really missing a full consideration of tax efficiency here.
The reason people make a huge fuss about tariffs is that they are more welfare reducing than (most) other taxes, for a given amount of revenue, whether there are retaliatory tariffs or not. $200B could be raised with a 1% VAT rate.
Meanwhile, the Laffer curve is a fun thought experiment, but the takeaway is that you should think carefully about marginal rates and design your system to maximise the cash raised while minimising economic distortions, not as a generalised argument that taxes should never be raised. The main reason Western Europe can raise much higher tax revenue than the USA is not primarily that marginal income taxes are higher, it's that they levy more taxes on things that aren't personal income.
The point isn’t about efficacy, it’s about how feasible it is for congress to increase taxes, which it completely is, as tariffs are a large, few hundred billion dollar tax. Just because republicans extended the tax cuts, doesn’t mean the incentives make raising taxes impossible unless there’s a collapse of the dollar.
I’d say the problem with a VAT is that it’s regressive. The bottom 50% pay basically no income tax, and are still struggling. A hike on the income tax would basically only affect the top 50% of earners, while a VAT tax would hit heaviest those who spend the most on consumption, which are the lower earners.
I wish we could abolish the words "progressive" and "regressive" in the context of tax rates.
I think that a situation where a minority of people pay the overwhelming majority of tax is bad. Of course I think that because I'm in that minority, but I also think it's bad anyway.
The problem is that the current debt crisis is so bad that even a miracle economy would barely help. And the current goverment is hard at work crashing the economy and stoking inflation as well.
“These companies will get less money than they were getting” is still an immediate obvious consequence with economic ramifications. I had the impression Congress kept funding weapons projects even after the DoD said “we don’t need it” because every defense contractor is an employer. People affected by the cuts know directly, people affected by overspending only understand in aggregate.
You imply that constituents don't benefit. That is probably wrong.
There are probably some defense spending in all 50 states. Each Congressman is in favor of cutting all the rest weekly, but strongly defending the spending in its own state.
Based on cases in different countries, it is certainly possible. E.g., Sweden managed to decrease government spending to GDP from 67.93% (1993) to below 50%:
Some congressmen said, (paraphrasing), "There are two things that can change a congressmen's 'opinion'. Money and votes (the voice of the electorate.) with votes being the stronger driver, but often divided. and so money. If we the voters all decide that balancing the budget is important to us, then congress follows along, or gets voted out.
If you can get sustained bi-partisanship - which you often can in more functional democracies - that diffuses the issue.
Example: In the local economic crisis in the mid 1990's Sweden, all the responsible parties agreed on a budget framework of a 2% average budget surplus over an economic cycle (end of recession to end of recession). This was then maintained for three decades, driving the debt down from 65% to 35% of GDP, until it was, again jointly, changed last year to merely balanced budget from then on, as there's nothing to be gained by an even lower national debt.
You don't even need sustained bipartisanship, you just need a political system which regularly allows the party in power to do things without the approval of the other party.
In the US, if you abolished the Senate and most of the power of the Presidency then it would be easy; when the Republicans got a majority they'd implement their policies, and when the Democrats got a majority they'd implement their own.
Since there would be clearly one party in power at any time, it would be clear whom to blame for any problems with the budget.
Both parties would still have a strong reason to buy votes with the national debt.
It's pretty clear that voters don't care about the national debt, or at least that they don't let it determine how they vote. This is why countries can rack up giant debts without the politicians getting kicked out.
In parliamentary systems, things work exactly as you say, yet this doesn't seem to help much.
Which countries with parliamentary systems have deficits which are as large and as persistent as the US's? (As a fraction of whatever, not in absolute terms).
Among comparable countries with parliaments I looked up a few deficits; accounting is tricky so I just used the first numbers I came across, then converted to USD:
*Obviously*, you should use debt to GDP, not debt in absolute numbers. It's what measures the weight of the debt.
When you do that, Japan (with by *far* the highest debt to GDP among any Western countries), Greece (famously), Italy, and some others, all exceed the U.S. in debt load.
Not every country in this graphic has a parliamentary system, but many do:
My mostly-fantasy approach to addressing this would be to reform the budget process (possibly via tweaking the rules for reconciliation votes in the Senate) thusly:
1. The "budget" is a stack-ranked prioritization of all authorized (but not yet appropriated) spending line items
a. This might be compilable from each individual Congress person's own such stack ranking via recursive Condorcet methods; I haven't checked the algorithm.
2. The total appropriated spending is capped at the average of the previous year's actual spending and actual revenue.
3. Working down the list from #1, each item is appropriated until the next would exceed the value from #2.
a. Lower-ranked items could still be pulled in, so long as they wouldn't cause the total to exceed the cap.
4. Taxes are set separately.
This would, in the long run, tend towards balanced budgets without any hard stop, as well as preventing proactive spending of speculative future revenue.
I think hopes of reducing the debt died back in 2010 with the Simpson Bowles commission. It's really hard to see how we ever get out of this mess. (I assume by "spending", you mean both lowering spending and raising taxes of course).
You imply there is some need to curb spending. There isn't. The Federal GOV (Through the Federal Reserve) can print money. It has infinite money. There aren't infinite resources though, so while the Federal Gov can't run out of money, it definitely can cause lots of inflation.
That is a different issue, though. Tax cuts to rich people don't encourage much consumption, they encourage private investment, usually. Tax cut to the not-rich or social spending increase consumption.
So the only reason to cut spending if it has high overhead (not efficient), or to reduce inflation. The thing is, it is somewhat self correcting. Inflation will lower the *real* amount of spending, so as long as spending don't go up with inflation in the same rate, the *real* spending will go down.
This is mostly what Millei did in Argentina. Because the Inflation was so high, not increasing the spending amounted to Real value lowering, at a pretty fast rate.
Clinton (D) worked with Congress (R) to cut spending on entitlements and raise taxes at the same time. The result was a balanced budget that was generating ~ 100B surplus each year when Bush II took office. The national debt would have been paid off in a decade, but Bush II was determined to cut taxes, so we ended up in deficit territory again.
Obama (D) tried to work out something similar with Congress (R) but by this point partisanship meant that any compromise was impossible. There were many Rs in the House especially who were determined to refuse Obama any sort of success at anything, country be damned.
So I guess when opposing parties compromise it is possible to lower spending, and your question should be reframed: Is it possible for the parties to compromise?
Very impressive peace activism by Palestinians and Israelis. It takes telling and listening to stories from both sides, which is emotionally wracking. Also, the effects of 10/7 and the attack on Gaza-- fewer people are willing to do the work, but it's picking up again.
Also, there's one advantage to extreme prejudice. It's brittle. If people have utterly dehumanized those on the other side, they can be surprised by finding that the other side is humans. There are no illusions about not being prejudiced because of believing in "a few good ones" or because "I have friends".
Both peoples have to learn to live together. Or not. But more outreach, more understanding, more compassion, will result in a very slightly less bloody future, very slightly earlier.
War is hell. Hopefully one day soon Hamas will be disarmed by force or treaty, and hopefully one day after that violent elements of the west bank settlers will be also.
I don't think you had time to watch the video unless you happen to have watched it already.
One of the points is that both sides have a dream of waking up one morning and finding that the other side has disappeared, and it is necessary to give up that dream. Israelis and Palestinians are both legitimately present on that land.
I think it is entirely possible for a majority of both Israelis and Palestinians to prefer peace to conflict, and yet for there to be no stable peace available.
The problem is that Palestinians in Gaza have no no-terror state capacity, and the PA in the west bank doesn't have enough of it to stay stable without idf support.
It's easy to find decent pro-peace Palestinians if you look. I'll avoid opening on whether they're 20% or 80% of the population (really hard to tell), but whichever it is, the group who does want total war retains the ability to launch it unilaterally and the others don't have the ability to stop them.
(This is why I believe Israeli policy should focus more on deconflicting the west bank, including pulling out settlements, and building up technocratic Palestinian state capacity in both the WB and Gaza. But while these are improvements the underlying problem remains pretty intractible).
The fundamental issue is that Palestinians have no ability to credibly commit to peace. Partly because of the number of times they've blown up deals in the past, partly because their political structure is dominated by violent organizations who gain money and prestige from fighting, and partly because they lack the state capacity to prevent attacks even if they cared to. The left isn't wrong this is an issue. But the right isn't wrong that such state capacity could be, and has been, simply used to attack Israel more instead of moving toward peace.
Secondarily, the Israeli public does not broadly trust Palestinians at this point and the pro-peace faction in Israeli was destroyed by trying to make peace and getting an intifada in reply. But even if you don't accept that it's the Palestinians' job to regain their trust, that's not the fundamental issue because you can change Israeli politics in a way you can't with Palestinian. Also, simply pragmatically, the Palestinians are the ones whose position is worsening so they need to do something while Israel can simply run out the clock.
The Jewish population and Israeli economy continue to grow such that the relative power of the Palestinians declines. Plus Israel increasingly is able to cut off support to the Palestinians and normalize with their neighbors (as is already happening).
Maybe the economy grows but as far as population, it's less determined. Gazas tfr is 3.3, down from 8! In 1991. There's a lot of people there.
Also, international support and normalizing is uncertain, especially as they turn international support against them. I feel like a Hague criminal prosecution is likely, although uncertain.
Fair enough. But I think a large number of people in an increasingly stagnant economic base is not likely to alter the balance of power. If anything it might make it even more favorable to Israel.
As to international support: Israel has gotten more and more diplomatic recognition, not less, and the region is continuing to move in that direction. The idea that it's going to reverse is unlikely.
And a successful Hague prosecution is also unlikely. Firstly, because Russia is doing far worse and so has every incentive to break the institution (and kind of broke it before Israel's war even started). Secondly, most of the charges activists want against Israel would not stick in a fair court which means the incentive is to charge them with big crimes that will not stick but get flashy headlines.
Assuming this is a good faith (which is a concession to a rather uncharitable and unjustifiable interpretation): You can see throughout the thread my point is that Palestinian national power is decreasing and Israeli national power is increasing, that Israel's diplomatic and relative military and economic position is getting stronger, and so on. So over time the Palestinian position is getting worse and at some point that will mean Palestinians cannot effectively resist. Thus the Palestinians are the ones who need a deal before all their leverage is gone.
There will always be an incredible amount of responsibility on Israel's shoulders for not just what they have done to make things worse, but also what they could do to make the situation better (cracking down on WB israeli terrorism would be an easy but impactful difference). Higher leverage does mean they can run out the clock but it also means they can be more agentic, and making the choice to do the former will in the long run make them more vulnerable to attack as their neighbours continue to hate them and they lose allies.
This is a common mode of wishful thinking among Palestinian supporters. However, the reality is that Israel has made steady diplomatic gains since taking a hardline against the Palestinians. And may be about to make further ones: if Saudi Arabia recognizes them then the only major country in the region not to recognize Israel will be Iran.
As to whether Israel bears more responsibility: we could go back and forth over the history but I don't think that's a particularly useful exercise. While on balance I think the Palestinian version of history is more distorted than the Israeli, neither is innocent. But you don't get to impose peace terms when you're a party so weak the other side can invade and capture all of your territory at its leisure. That's simply hard power reality.
They almost got recognised by Saudi, that's about the only significant diplomatic gain they have made in the last 10 years (and it took the US to help them do it).
The worry for these Arab countries is not Palestinians, they don't care about that and I never implied they did.
They've been recognized by five additional countries in the past five years and a sixth is likely on the way. If Saudi Arabia comes over then Iran will be the only major power left that doesn't recognize them and the Palestinians rely on Arab support to sustain their situation.
I'm not as cynical as you. I think it's very hard to explain Arab actions without a genuine care for Palestinians, even if that care is partly motivated by nationalism and some degree of prejudice.
Don't understand the whole "education has been feminised, boys aren't meant to sit still for so long" etc etc. How do you think school was hundreds of years ago? You think boys were allowed to run around and scream?
Most kids didn’t go to school though. Only the sons of the elites. Even then there was a lot of emphasis on physical training amongst them. Not infrequently, physical prowess was valued more over reading or writing by educators.
"Not infrequently, physical prowess was valued more over reading or writing by educators."
That may be the 19th century 'muscular Christianity' public school tradition, but mediaeval schools for boys involved a lot of sitting (or standing) listening to the instructor, then repeating lessons back - not running around playing sports. The boys would do that outside of school, but if you were fortunate enough to attend school, you were there to learn. And discipline was immediate and harsh for rules breakers.
Wouldn't only the kids already inclined to Church/learning do that though? The other kids of the nobles were out learning to ride, sword fighting etc. The monarchs and nobles in the first half of the middle ages seem barely literate, even the ones in later part, probably didn't spend 8 hours a day writing and reading for 10+ years. And as Collisteru said, discipline for the kids who didn't stand or sit would have been harsh as well. Of course the only education, 99%+ of the boys got was working in the fields and artisan shops.
It was still a small minority of boys who would go to much schooling. I'm not sure how it was determined who would go, but presumably boys more inclined to sitting and learning than most. Nobody thinks zero boys do well in school, just that too many do poorly for obvious reasons like the emphasis on sitting and listening.
And it very recently would be said (dubiously) that men were the more rational sex, systemetizers and problem solvers, more suited to science and mathematics and engineering but also literary analysis and medicine than girls.
And these were all taught with boys sitting still, doing their arbitrary arithmetic or reading. Even to significantly more exacting standards, under much more strict rules.
Maybe that's what they mean - boys have to be taught under very rigid structure while girls work better under the self-motivated looser structure, but I doubt it, because that doesn't fit their sexist frame.
There's ways modern education have deteriorated and are failing _everyone_, boys and girls, but I think it would be pretty squarely sexist to call that feminization
However, it's a fact that girls are doing better than boys at every level in the education system. This is likely bad. What's the cause and how do we solve it?
Because most of the education establishment is feminist, if there are any areas boys are ahead, they'll say it's a result of sexism, whereas areas girls are ahead are ignored or seen as welcome progress. So unless you can make boys and girls exactly level in everything (which as we all know is impossible even if blank slate theory says it's the goal), they'll continue modifying things to favor girls until the establishment changes.
So whoever They are, They both believe in blank slate theory *and* know exactly how to modify educational systems so that they consistently favor girls? That's pretty impressive: nobody ever taught me any of those secrets when I worked in education.
No, in fact, he did not. He merely stated that the process occurred. This statement included neither evidence nor explanation, merely an assertion of motives. And I notice your reply is much the same: you flatly assert that I am "pretending not to understand," nevermind how you could possibly know that. For my part, I make no claims whatsoever as to whether your abysmal lack of reading comprehension is real or feigned: I couldn't possibly know.
As I said, I used to work in education. I taught quite a variety of students over the years. On a good day, after coming at difficult material from different angles, I could find one that clicked with a particular student. On a bad day I might get nowhere. At no point did anyone teach me any educational techniques whatsoever that depended on a pupil's gender: the thought is frankly rather absurd. Even more absurd is the idea that educators would laboriously develop such techniques and--having mastered the secret arts of educating differently based on gender--still cling to naive blank-slatism (which to my understanding is mostly a strawman, and quite uncommon even among feminists). If you or the other poster have some actual evidence--not assertions, evidence--that such techniques exists and how they work, I'd be quite curious to examine it.
Girls are on average: more prosocial, and more inclined to rule following for it's own sake, boys on average are: more susceptible to social pressure to do stupid shit, don't respond to non-draconian enforcement of norms.
The solution? Allow schools to punish the parents like it's japan in the 60's. Force mom and dad to come to the office in their sunday best and get dressed down for raising a shitkid.
Alternately, students need to be held back, suspended, and then expelled more often, which works as an indirect punishment for the parents, who then force the students to follow rules.
Anyone who reads accounts of American schooling before it was "regularized" will be familiar with these tropes: kids walking quite a ways to school (thus getting some energy out before it even starts); kids having to help with the chores around the physical school building (making the fire, getting water, sweeping); kids getting a very civilized long lunch time in which there was lots of time to run around and play, unsupervised; kids helping younger ones since all the grades were generally in one schoolhouse, which makes it plain that there was activity other than staring at the teacher.
Additionally, I'm pretty sure schoolwork nowhere started at the ungodly hour of 7:45 AM which was the case at my child's elementary school. In his time there, the children gathered in the lunchroom, cleared of the tables, sitting on the floor in lines and goofing around until five till, when there would be a usually-short assembly each morning. Later on, they changed it to the kids - all except those eating free breakfast - going directly to their classrooms, at five till eight, which the teachers preferred. Either way, after the pledge you could be staring down a page of math problems at 8:05 in the morning. There was no freedom in how the day might unfold, as even the elementary grades had inflexible "periods". I am certain this was not the case on the frontier.
I'm sure discipline was often swift and stinging, especially if the teacher was a man instead of as so often a very young woman, only a little older than the oldest pupils.
But I don't think that was quite the same thing as the dulling homogeneity of the atmosphere of public school day as it is now.
There is a vast roster of very successful men of the 19th and early 20th century whose schooling was over by ten or twelve or whatnot, if they were even schooled at home. So maybe school was never ideal for boys - or maybe they would have enjoyed more school but circumstances didn't allow it.
Most children, boy or girl, would have had either no education other than passed down knowledge from their parents with perhaps an additional day of learning from a religious or civic figure. When I say most we're talking about a minimum of like 60% and a maximum of like 95+%.
Of the remainder, the majority would be merchants and craftspeople and religious figures. The religious would have had an experience most directly comparable to modern school. The merchants and craftspeople would have gone to something like trade school mixed with a business school either from their family or from their local merchant organization whether a guild or some equivalent. Then the tiny aristocracy would have had a mix of courtly education and military education with surprisingly little academics by modern standards. Unless they were being prepared for religious life.
Regularized, universal learning of the type we're used to today only became broadly available in the late 19th and 20th century. And it is worth dwelling on whether it is the only form of worthwhile form of education. But the idea it's feminized is just a new, more gendered version of old critiques of industrial education. Which objectively has issues but also you have to admit has done pretty well at making society richer and more productive.
This is one thing that frustrates me about broadsides against education in general: Yes, you can return to 18th century standards but then you should expect 18th century workers with 18th century productivity. Not great.
Yup. This is the general problem with which nobody who wants to return to a lost golden age engages. The past was much poorer, sicker and bleaker than we are now.
I think the problem here is the OP's invocation of "hundreds of years ago" which has caused people to answer from all sorts of different historical perspectives.
Perhaps "one hundred years ago" would have been a better question, when schooling was universal and compulsory across most rich countries but much more structured and disciplinarian.
100 years ago was 1925. Universal compulsory education started in 1918 though some states had it from the 1850s. And high school was still rare with college education being even rarer and most high end jobs still not running through universities. If you want to go through that and say maybe it was a mistake to route too much through research universities and we should have done more vocational/job prep then I might agree. But the secular increase in education has almost certainly been beneficial.
Was the past worse because of social institutions or just because of worse technology? It could be that better technology is the source of most or all actual improved quality of life in the modern era.
It certainly could, but that’s an argument that needs to be made. The problem with simple appeals to a golden past is that, compared to the present, the past sucked. Now it’s possible we could have had an even better counterfactual present if we had retained certain elements of the past…but that’s the argument that needs to be made.
> And it is worth dwelling on whether it is the only form of worthwhile form of education. But the idea it's feminized is just a new, more gendered version of old critiques of industrial education. Which objectively has issues but also you have to admit has done pretty well at making society richer and more productive.
What? Why do you have to admit that? Two things happened at the same time. What's the causal relationship?
If two things happened at the same time and one strongly predicts both individual and national income then that at least strongly gestures toward relevance. To the point where you'd have to make an extremely strong case that it was unrelated, especially with no empirical counterexamples outside of societies reliant on oil etc.
I’d say much of society has been feminized in the last decade. For example, the now preferred debate style at companies where the person leading the meeting has to make room for quiet people to voice their opinion. Certainly, it’s obnoxious to have people who talk (or yell) non stop and prevent civilized decision-making. People should be trained to speak in turn and to use positioning rather than volume for gravitas. But that’s different from wanting everyone to voice their opinion, which is also counterproductive. Shy people should learn to interject when they consider it necessary. It’s also okay to be someone who listens more often than speaks.
In addition to less recess as someone else linked, PE has decreased a lot since the 90s. When I was in school we had a class period of PE every day, from K-12, but it sounds like kids today don't get that.
My understanding of American schools is that there's no recess past K-4, and that "lunch" takes place in an indoor cafeteria where you can't run around and play ball games either, so you're basically spending the whole school day indoors and there's no opportunity (outside organised PE classes) to burn off all your teenage energy.
Just give the kids half an hour for recess and an hour for lunch and let them play soccer or basketball or whatever they feel like, there's no need for daily PE.
I liked the piece, but the logic doesn't really work.
If you have a group of guys whose main problems come from being beaten up by other guys, and you make the other guys stop doing that, it can easily be true that the original group's main problems now come from a different source. It isn't usually considered contemptible, after you've solved one set of problems, to move on to solving more problems.
Also, I struggle to see a connection between the idea that boys used to do well in school because adult men would beat them up if they didn't, and the idea that nerds - who did well in school without needing to be beaten, but got beaten anyway because they were wimpy and unimpressive - would be less wimpy and unimpressive if they got beaten up more regularly. Those don't appear to be similar dynamics.
I was mostly posting it as a joke, and on the off-chance it would be invaded by polite but secretly angry Astral Codex Ten posters. "Here's ten reasons you're wrong, read HPMOR..."
Yes but how universal was school, and was it used as a babysitter while both parents were working? I imagine both of those things are relatively modern inventions. Pre-Industrialization, school was hardly compulsory, or universally applied. It seems hard to get estimates but links like this one support extremely low percentages of under 18s attending school - https://link.springer.com/chapter/10.1007/978-1-349-27231-0_2
It just doesn't seem like a comparable situation at all. Children were entirely unschooled, or they had individual tutors, or they attended schools for much briefer windows of time. For anyone not in the elite classes, by the time a child is strong enough to be put to work, they're going to be used for labor in the family business or what-have-you. Elite children with the leisure for learning would have more of a flexible schooling environment rather than being made to sit for eight hours a day. Certainly whatever school they did have was not likely to be as regimented and "one-size-fits-all" the way our modern schools are. Although I'll grant that there were probably also stronger methods of discipline used.
But then again, I think maybe you can't compare modern boys to 17th century boys either. The expectations, the environment, the culture are just totally different. Today's boys spend far less time reading in general, not to mention the brain rot of video games and social media that's destroying attention spans right and left. For that reason alone, I'd venture to say that today's boys are on the whole less teachable and tolerant of long lectures than their 17th century forebears.
I do think school has been overly feminised but I don't think that "sitting still" is the problem.
My impression is that schools of a century ago had much more sitting still and much more discipline when you didn't, and I think this is an environment where boys can thrive. You still see this kind of discipline in the military, where young men seem to thrive under it -- a combination of strict discipline with plenty of physical exercise to tire them out.
I think the thing that boys and young men struggle with is the semi-structured environment, where discipline is inconsistent and consequences are real but always seem far away.
I don't think it is debatable that children are less physically active today than they have been at any point in the past.
I also don't think that young boys are stupid. I assure you they realize they exist in a school system that is hyper-vigilant toward anything that can be perceived as a transgression from them with an equal amount of energy committed to excusing anything that could be perceived as a transgression toward them.
I can't blame them for checking out of a system that makes regular shows of celebrating anything and anybody for little more than being not-boys. And that's not even counting the ones having to endure an actually chauvinistic teacher.
Schools definitely handle discipline differently, and much less strictly. I think the normal day in school is also a lot more feminized, in the sense of emotional recognition, talking through issues, and the expectation of quiet reason.
Boys used to dominate the "raise your hand and answer the question" portions of class. Competition, displays of prowess, etc. But it's become a problem to allow that, so teachers do something else (rotate, pick who to answer, have everyone answer quietly on a paper or small groups). So something boys did well at that encouraged them to learn the material and engage has been taken away.
That's just one of many changes, most of which were small and with good intentions that do help some students, that have made school generally worse for most boys. Some boys will benefit, and apparently many/most girls benefit, at least relative to the previous approach.
I think it’s mostly downstream of an overwhelming supermajority of school teachers being female, and most of them not ‘getting’ boys, despite meaning well. Boys and girls actually are different. (A small minority also actively dislike boys, which is tolerated in a way active dislike of girls would not be, but those are the exception. Failure to grok is however the rule).
They give them less attention (because they don't grok them, this is normal and human). Their explanations tend to be in terms that resonate more with the girls. Grading structures tend to be of a type that (statistically) favor girls (e.g. more emphasis on groupwork and homeworks, less emphasis on tests). And yes, ceteris paribus they give boys slightly worse grades. Most of the time there is no malign intent behind any of this, and the teachers are well meaning, just don't `get' boys.
As for what should they do instead...I don't know. This is a hard problem. Awareness of the pattern could be a start. But I cannot help but note that, as a physics professor, I am inundated in trainings on stuff that I can do to make physics more welcoming to women, and I doubt there are any analogous trainings for women teachers on how to make school more welcoming to boys. Then again, I don't know if the trainings do any good at all, so...
I agree that the feminization of education is a problem, but the problematic part isn't that we're making boys sit still. The problem is authority figures have been increasingly prevented from and less willing to discipline their students. Similarly, teachers are increasingly prevented from and less willing to fail students when they should be failed
That is because teachers cannot be trusted with that type of power. A female teacher with a strong anti-male bias should not be given more power to abuse children. I know everyone wants to pretend that teachers are all good people only interested in teaching the youths but that is simply not true. Having power over children is an attractive position to many types of abusive people.
I just don't think a teacher can discipline a child raised by an iPad.
I understand the point of teachers having no recourse to an out of control kid but I just don't think there is realistically anything they can do. It's largely too late for the teacher to do anything by that age.
That society is expecting teachers to run double-duty rehabilitating feral children with no support is definitely not helping our educational collapse problem though.
This is very interesting in theory, but in practice what is going to happen, in very little time, is that there will be no virtual reality anymore. Physical interactions are going to be the only thing you can trust ; maybe they already are.
AI videos are now so good that video can't be trusted. In the last weeks, several members of my family have been targeted by scammers that had the name of their bank manager, and called with the phone number of the bank. All the university professors are calling for closed-computer exams with pen and papers because software can't tell you whether a paper has been written by AI or by a student, but this is of course also true for everything online and no one can't trust customer reviews anymore.
You get the gist of it : I think either AI is banned (but spies, crooks and scammers will still use it) or you cannot trust anyone you cannot meet physically, so everything is going to crash.
I haven't for the moment seen any credible solution to that problem, so I tend to think the AI crash is coming soon (and a lot of jobs are safer than they seem)
I'll have a hard time finding in-person conversation partners close in quality to modern AI. I interact with people in meatspace mostly for practical purposes. I wouldn't mind a time when AI can generate content on par with the best of rationalist spaces.
It's not what I think is the *most likely* scenario, but I've thought for a while that *a* plausible scenario for the AI age is that the internet would get so overwhelmed by AI slop, and communications channels so overwhelmed by AI scammers, that basically the whole foundation of social trust on which internet communication is (believe it or not!) based would crumble. Essentially, the internet would no longer be viable. We'll be compelled to go back to the real world - going to "singles bars," sending in documents by snail mail and what not - while the internet devolves into a purposeless "Disneyland with no children," and also with no fun rides. Someone in a comment above said that this is basically already happening with reddit, twitter, etc.
I also think this is fairly likely, but just an acceleration of something I expect to happen anyway. Meat space is psychologically far superior to online connections. We pursue online because it's faster, easier, and allows many more connections. Anyone who used to set up LAN parties knows how much better multiplayer gaming has become. Multiply that by several times and add it to tons of different fields, and it's tempting. But it's also hollow. You can't see people's faces, note that expressions, experience the physical environment together. And yes, people have been fake and scammers since shortly after the early internet.
We can solve the worst problems of AI/scams online to continue running business more efficiently. We are not going back to carbon copy memos sent through the mail. The efficiency is worth the increased costs of security. But social interactions are not monetized enough to make verification cost-effective online. So where we can or have to, I expect/hope that we will move to more in person interactions and more "touching grass" in the future.
Seven years ago, two monumental events came to pass in the same week : I passed the bar exam, and was diagnosed with MS. Today, I find myself at a crossroads and would be grateful for your advice.
As a result of my fear, hard-headedness and lack of perspective, I rushed headlong into the job search before I even recovered from my first 'attack'. I ended up getting a corporate gig and I hacked it for 2 years until successive relapses forced my hand and I went on medical leave.
After a year off work and researching options, I decided to go nuclear and get a stem cell transplant (high dose chemo treatment). Due to COVID, it took another year to actually get the treatment, and it took another 3 years to properly recover. All told, I've been out of the legal game for 6 years now. My disease seems to be in remission, but I still carry nervous system damage which can put me out of commission at random times. I am not cut out for high-stress work anymore.
I always assumed I would return to my legal career, but as time passed I realized that deep down, I was dreading it. My body seemed to know what my brain refused to accept. I've come to peace with this realization, but that has brought a new challenge: what in the fuck am I to do now?
My benefits can keep me afloat, but the next 5-10 years are critical for reinventing myself. I am somewhat at a loss as to what direction to go in. In terms of skills, I suppose my toolset of general learning ability, reading/writing, and argumentation/analysis can be useful in a lot of things, but the 6 year gap and lack of any experience since makes me a bundle of potential, which is fairly useless without application. And so I'm trying to figure out how to apply myself.
I would like to try and find something I can do remotely, and/or disconnected from a specific physical job market. I live in a cold climate (Canada) which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy, and the possibility to create freedom of movement for myself is probably the highest leverage thing I can do for myself over the next decade.
The stress is low, the hours will be limited to 40, and any sort of disability protection is taken very seriously. It seems to work around your primary limitations pretty well while still taking advantage of your most valuable asset: passing the bar. With your limitations, I don't see any other career path that would offer nearly the salary of something requiring passing the bar.
Sorry, I left out important context and I've edited my original comment; yes I have considered it. I would like to try and find something I can do remotely, and/or disconnected from a specific physical job market. I live in a cold climate which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy.
Attorney, California Department of Water Resources
$92k/year to start, you'll max out at ~$140k in ~6 years.
Hybrid role in Sacramento, CA.
Job Description:
Under the direction of the Assistant General Counsel and/or the guidance of a senior lead attorney, and consistent with good customer service practices, the incumbent performs duties of average difficulty related to the legal support of environmental compliance, including state and federal ESA permits, ongoing SWP operations, habitat restoration projects, South Delta projects, drought programs and CEQA. The incumbent, in coordination with the Assistant General Counsel and/or a senior lead attorney, advises the Director of the Department of Water Resources (DWR), other DWR executive management, and DWR program management and staff about complex and sensitive cases and/or matters involving opposing counsel with a high level of specialization. In coordination with the Assistant General Counsel and/or a senior legal attorney, the incumbent will have contact with members of the legislature, high level governor appointees, constitutional officers, and the general public.
Translated as "We need someone to do legislative analysis and research who can "officially" provide legal advice, which is why we pay a premium.
I would not generally recommend Cali but for a disabled lawyer looking for a low-stress, high-security job in a hot environment, the California civil service will be hard to beat.
If you're curious, go look at for attorney roles at jobs.ca.gov. If you see a number of roles that you are at least passably interested in, look into the specific job classification and see which title fits you best.
From experience, it's hard to get in the system and easy to move up once you're in it.
Edit: This is what I mean by the bar exam thing. There's jobs that pay more than $140k that don't require you to pass the bar but they require a lot of effort. The store manager for a Buccees or a Walmart makes more than all of us but they work their tuckus off. $140k for doing a moderately difficult job is tough to beat.
Do you live in a country with multiple climate types? If a warm climate improves your functionality, it may be worth moving.
Also, make sure you do something you are good at, eg: would be in the top quartile. This in general reduces stress and pressure and makes you sought after and more employable after breaks.
"I live in a cold climate which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy, and the possibility to create freedom of movement for myself is probably the highest leverage thing I can do for myself over the next decade."
You have probably/maybe considered this, but better that I mention the obvious than you miss it. So ...
Have you considered moving to a warm climate where you would be "highly productive, stable and happy" for most of the year rather than staying where you are where the climate "wrecks [your] functionality for 8 months of the year"?
This is independent of what you do with respect to your career.
It's hard to get papers for western Europe. The warm countries that are easier to get to are very far, and as miserable as I am in the cold, I imagine I'd be even more miserable moving thousands of kilometers from anyone I know and love, and also to be in a country where my earning potential is even lower.
I suppose I didn't really explain myself well - I am trying to find something to do that I can do anywhere; that would *allow* me to move to a warmer place that is closer to family/the culture I feel more congruent with. I don't think I can ship myself to SE Asia or South America and start a new life.
I mean, some cities in the US are pretty warm. And the pay is good. And seeing as you speak English and are presumably white, I don't think the current administration will give you that much trouble if you try to immigrate.
An IMF study from April titled "The Global Impact of AI: Mind the Gap". Key quote: "The United States (US) stands out, registering the largest projected [in ten years] output increases at 5.6 percent and 1.9 percent in the high and low TFP growth scenarios, respectively." I'd guess this is a typical example of mainstream/normie expectations from AI. Would be interesting to observe the collision of this perspective with the feverish scenarios circulated around these parts...
Has anyone been to the Mall of America and really enjoyed it for what it is?
I went and had a good enough time doing some people-watching, but I can’t think of a single interesting store I saw. My buddy got a nice sweater, I guess? But it was just a big mall, full of normal mall stores! I dunno, I was expecting more for whatever reason.
(Relatedly: I’m in the Twin Cities for another day; any recommendations for food / bookstores / nice streets to walk down?)
I’ve been there voluntarily once. No I didn’t enjoy it. I’ve also been there involuntarily a couple times for work lunches. I enjoyed those visits even less.
Take a walk around Lake Calhoun now called Bde Mka Skaw. It’s shaping up to be a good day for it.
Edit
Try to do that before noon if you like sunshine. Local forecast calls for a cloudy afternoon.
I went there as a tourist if it counts? I don't think I brought anything and I really just remember the same stores as any other mall. But as a foreigner it was interesting to see such an enormous mall and observe Americans in what we think of as their natural environment.
The malls hereabouts were on track to die because Americans began shopping in other ways (actually, some of them died *before* Amazon but maybe the attractive mail-order catalog period that preceded Amazon had something to do with it). Some did die and became sad improbable things like community colleges.
The mall was indeed an American *place* for quite awhile but it seems not to ever have been that successful financially.
The undead ones have become immigrant bazaars. Instead of most of the American chain stores you will see in mall-set American movies, there are hand-lettered signs "Everything $8.99" store, and a lot of food options and kiosk hawkers. Bling stores. Fix your phone screen places.
There were kiosk hawkers even 25 years ago, but now it's more like the kiosk hawkers took over a lot of the mall spaces.
There are couches and chairs and some people just enjoy the AC I guess.
Where the anchor stores survive, whether because of their endless favorable lease or because they actually own their footprint, and are hanging on so they can cash in on the dirt - they seem like dowdy, dirty versions of their former selves. Macy's is now more akin to that store with no dressing rooms and the heaps of clothes that women supposedly fought over - in NYC - I can't remember the name, it used to furnish a punchline ...
You will stand in line behind a pair of immigrant women buying $300 worth of baby and children's clothes, the prices of which one of them mechanically questions as they are *slowly* rung up (that's not her fault, is everything in the store "on sale" - or not?) and also trying to return a bunch of things she bought at another JC Penney and for which she doesn't have the receipt.
Come on! There's a cheerful energy to some of them, depending on how large is the local low-socioeconomic population. You may find it interesting - or you may find that it's more like something you've encountered elsewhere in the world. It's not particularly American if that's what you're after.
It's odd to me that malls are dying in America when they seem to be doing so well in other parts of the world.
But I think the US has been doing malls wrong all along. The biggest thing they are almost all missing is supermarkets. Then, the discount department stores like Walmart and Target that Americans do so much of their shopping at? Also missing. Even pharmacies don't seem to be there if I recall correctly. So what's the point of having a mall, if the vast majority of things that you need to buy over the course of a month are not there?
My local mall, which is not among the five largest in my city, has two supermarkets, three "Target"-style stores, two "proper" department stores, several pharmacies, and a few hundred other stores selling everything else, all under one roof. I don't always do my grocery shopping there, but if I need groceries _and_ something else then I might as well go to the mall... at which point maybe I walk past some other shops and decide to buy something there as well. The point is that you need these weekly-visit stores to anchor all the other stores, you can't just have a bunch of clothes shops surrounding a food court.
The grocery stores and pharmacies are in strip malls closer to where people live, for convenience. Retail stores you visit regularly, you want to be as close to home as possible to minimize wasted time. The mall is for irregular purchases like clothing, kitchenware, appliances, etc.
Sure, not all supermarkets should be inside malls, but all malls should have supermarkets. That way if you're going to the mall anyway then you can get your groceries at the same time. Or if you're need groceries and you kinda sorta want to look at new jackets while you're at it, then you do a trip to the mall instead of a trip to the regular supermarket.
> It's odd to me that malls are dying in America when they seem to be doing so well in other parts of the world.
Malls were big in Shanghai in the past.
Those malls are still there, but more recently, whenever I asked a friend where to go to buy [whatever], they'd be surprised at the idea of going to a store, and tell me "buy it online and have it delivered".
Maybe because the name is a little ostentatious? Mall of America. It's kind of a One Mall to Rule Them All sort of name. I can see having heightened expectations, just based on that.
What constitutes as an interesting store to you? I would argue that less than 1% of stores are even remotely interesting (at least to me).
The only truly interesting stores I've been to have all been in Japan. Akiba-Hobby, card stores (these were not *that* different to American card stores), and a bunch of stores that sold weird/interesting stuff. For many reasons these types of stores can't exist in an American mall.
Hi Scott- I've just posted a response to your recent revisiting of the Covid Origins question. It's not the most pressing issue in the world now but I think it's important not to be tied down to views that are both incorrect and unpopular. More importantly, I describe how you guys screwed up the Bayesian calculations and say a little on proper general hierarchical methods.
Completely unrelated to your actual points, I notice you went with the spelling "synch", which I also prefer. Is this a conscious choice? Do you have any opinions on it?
I'm 22. I keep seeing mostly American teachers rant about how kids are so unruly these days, half ass everything, won't pay attention and won't listen. They blame phones, obviously.
I'm scratching my head and thinking, what are they surprised by? What incentive do kids have to care about school much?
Many decades ago, pretty much only interested and intelligent people would go to college. There was also no internet and so accessing knowledge was harder. University was a way more unique opportunity to learn interesting things. Even regular mandatory school was such an institution for some children.
Now, interested and intelligent students can learn interesting things in a way more fun and less stressful way online. The kids still in public school who aren't interested and probably wouldn't have been 60 years ago either, are not disciplined as harshly by either parents nor teachers for not being good enough.
Add to that the fact that some schools apparently don't have recess anymore (???).
Anyway, I actually have taught kids age 6-14 ish a couple of times now, through my job (recycling). They were very behaved, I'd say. I don't know if there's a difference between Danish and American parenting or what. School kids get their phones taking here, but they can still view brainrot and play games on their laptop they're all sitting with.
When my husband was a gentle, scrawny freshman in high school 40 years ago, he was walking down the hall when a big older boy he didn't know - wouldn't ever have known - decided he was in the way (maybe he was!) and decided to stuff him in a locker. (Supportive wife that I am, the first time I heard about this, I of course asked why he hadn't allowed himself to be so stuffed. Go along to get along!) But a man must be man sometimes - so he resisted and they got in a fight, which ended with him on the floor being kicked in the head a lot and ultimately concussed. They were both then in equal "trouble". But it was not going to be much trouble.
I realize this sort of thing would be ho-hum nowadays, not even worth filming.
His father was pretty upset - having been educated at a college-like prep school in the Northeast, this did not accord with his idea of high school, in any degree - and made a rare parental move and went and chewed out the administrators.
What's interesting is my husband felt kind of bad for the other guy. Because he thinks, although doesn't really know for sure, as his own family subsequently moved away - that that other boy never did come back to that school. And he wonders if that impacted his life in some way.
Whereas now: forget the kids fighting - the teachers complain of being cursed at, and chairs and desks thrown-at, and referring the hellions to the administration - and finding the kid back in the room 3 days later. How do you not lose face, as a teacher, when that is the case?
This really is on a school-by-school basis. That sort of assault at more economically upscale schools would be treated as a problem. Partially because the relatively wealthy parents could and would threaten legal action.
But the teachers at schools without this (or other) problems don't post that "things are mostly fine."
But I would not want to teach at the schools full of poorly behaved students.
That long-ago instance happened in a different America when people with means, like his family, attended school with people of other socioeconomic classes - in that case because it was a small-ish town, in other cases just because people had not yet sought and encased themselves in bubbles.
People underestimate the effects of boats being lifted (or lowered) *because they were all in sight of each other, in each other's lives to an extent*.
In my state whole towns are bubbles, and other towns are very poor. Albeit even poor as they are, usually a better-seeming place to live - air, space, flowers in pots, single family homes on nice-sized lots, all bespeaking a more prosperous era - than the urban ghetto.
Do all these school incidents become known to the police? I didn't know that. I thought the school handled, and if charges were filed when a kid cursed at and threatened a teacher, we wouldn't hear how they were right back in the classroom.
Of course, that sort of thing being unknown to my generation, it's hard to say. I expect there was more car stealing and joyriding in the past. A little less wildly shooting weapons at gatherings.
I don't know what percent of school incidents become known to police, but note that the link specific to schools is re a victimization survey, not police statistics:
>The data used in this study are from the California Healthy Kids Survey, a modular survey instrument developed by WestEd in collaboration with the California Department of Education and used biannually since 2001
Moreover, even re police statistics, of course many incidents are not reported to police. But that has always been true. Is there reason to think that the percent reported to police has changed substantially?
More broadly, given the enormous decline in juvenile violence outside of school, is it plausible that juvenile violence inside school would trend in the opposite direction?
> Many decades ago, pretty much only interested and intelligent people would go to college.
This is basically the answer. If you cannot select for intelligent kids, and if you cannot select for kids who actually want to be there... then not much learning is going happen. Either because they can't, or because they don't want to.
Even worse, the school then adapts to the average (non-)learner, which ultimately makes it uninteresting also for the intelligent kids who originally wanted to learn, but now they are forced to waste time listening to stuff they already know.
The obvious response to allowing schools to select for intelligence is going to be that society will visibly sort into people who were intelligent enough to seek and complete college, people who sought it but washed out, and people who weren't intelligent enough to seek it. The subsequent stratification of society will be noticeable and scary.
I'm not saying this is a convincing argument for never doing it; but I will say that it's going to be hard to stomach.
Another problem is that some schools will claim they're selecting for intelligence while really selecting for desirability, by their standards. In a very free market, this won't get *too* bad, but today's market has some notable unfree knots in it, in the form of prestige. Not everyone can go to the "best school for $subject", and that "best school" might not actually be the best, and yet there will be people who treat them as if they are, and that will produce a lot of complaint. Again, not wrong per se; just rough.
Read my millionth piece this weekend about how modern American women are staying single, are not interested in marrying the average American man, want to focus on their career instead, overtones of disliking men, etc. (This one was in the WSJ). Makes me wonder- why isn't there a matchmaking service that connects developed world men to potential wives in developing countries? (Or maybe there is and I don't know about it).
Snickering about 'mail order brides' has been a thing for decades now- why not match these two parties up at scale? It seems like the average potential wife from Thailand, Brazil, the Philippines, Ukraine, Mexico, India, wherever would be much happier with married life in the US than the average American woman at this point. They'd probably prefer more traditional gender roles as a stay-at-home Mom. They'd probably overall be more grateful to be here, and less into modern day gender culture wars and general developed country ennui/depression-as-politics. I know multiple men who've married foreign women from poor countries who are now US citizens (and actually 3 women who've married foreign men). Seems like a win-win for everyone except for native-born American women, who I think have their own issues to sort out.
I'm surprised this isn't a bigger thing, especially on the right? Maybe like a matchmaking service or tech company to connect American (or European) men with foreign wives at scale?
> I'm surprised this isn't a bigger thing, especially on the right? Maybe like a matchmaking service or tech company to connect American (or European) men with foreign wives at scale?
There's tons of these, and many dating websites specifically devoted to it.
But speaking as somebody who's done a lot of business in various countries overseas for the last 15 years, and has consequently seen a bunch of Westerner + foreigner relationships, there's fairly strong negative selection effects for the women using the matchmaking services and devoted websites, and most people would do significantly better just coming over for a few weeks / months and using normal dating apps or meeting people in person.
That said, I wrote an article in response to Bryan Caplan's post about overseas dating and marriage, looking at the fairly limited literature on the subject to show that "gold digging" is relatively uncommon and not as much a risk as everyone seems to think.
Thanks, that was a good piece. I think my one issue of concern is having her be able to fit into the US- now she's moved away from friends, family, doesn't know anyone other than me initially, etc. Women absolutely need female friends to talk at great length with. I was thinking it'd be easier if I lived in a city that already had a large ethnic community from her country- that way she could much more easily make friends with them. Say, Vietnamese in southern California or Brazilians around Boston, etc.
> I think my one issue of concern is having her be able to fit into the US- now she's moved away from friends, family, doesn't know anyone other than me initially, etc. Women absolutely need female friends to talk at great length with
A very thoughtful and cogent point. The dynamics I've seen usually revolve around lots of video calls and messaging with folks back home, and then gradually finding friends in the new place via the usual routes (yoga class, hobbies, other local moms, etc).
I think you're right that having a good amount of ethnic confreres in the area would probably help that in-person process, but I also think it's probably not strictly necessary for a good number of personality types. Generally, people willing to entertain marrying foreigners and traveling / moving a lot are pretty high openness and able to roll with the punches.
But everyone is an individual, and the degree to which having ethnic confreres nearby might help could indeed mean a lot to some people, so kudos for anticipating that ahead of time.
I think you're being a bit unfair to first world women here. Also, your attitude doesn't seem right for a solid, long lasting marriage. Yes, they'll get a lifestyle upgrade, but you shouldn't assume their expectations in a relationship. Even if they want traditional marriage they might have a different definition than you. Also, expecting them to be grateful for getting them into the country sets up a bad power dynamic for a strong marriage.
But also there's nothing wrong with seeking out the partner you think will bring you the most happiness.
So to answer your question: There are a ton but a lot of them are scams or thin fronts for prostitution. Largely because the space is hard to regulate and the woman's country is often uninterested. And because those generate more money you get a good crowding out the bad effect. Still, supply largely outstrips demand on the women's side, with something like 100-200k women being interested in such services but less than 10k actually getting married globally. And that's with the existing stigma and bad service level. It's also counting the women who actually join such services unpaid, not any who would be interested in such a match. (There's also a fairly strong demand among men, by the way.)
The big difference is usually cultural. Such matches work best when both are fairly open minded to each other's cultures, expect to blend expectations rather than assimilate fully, and are highly communicative to get over expectations or culture gaps.
If you're really interested in one, I'd suggest you pick a country or culture that otherwise seems interesting and familiarize yourself with it. Make some effort to know the people and visit the country which will give you some idea of what you're signing up for. You can also take the opportunity to meet people and that can turn into arranging dates.
> I think you're being a bit unfair to first world women here.
You're giving extremely solid and nuanced advice in the rest of this comment, but I just wanted to point out that US women really *are* noticeably worse than most foreign women, and you see it both in data and qualitatively if you spend time overseas.
This assumes that you want a non-fat, more traditional woman who puts more effort in and will keep a good home and is willing to have kids, but I think that's pretty much exactly what most men who complain about Western women want.
Let's consider Asia. Asians start with ~half the overweight and obesity rates:
So you start off thinner, then alone among all other races / ethnicities, Asians do not gain weight at furious and ever-rising rates when looking at individual weight gain trajectories over time (the green line, Other=Asian):
Asian women spend significantly more time on housework and cooking, according to time use surveys:
US: 2.6 hr
UK: 2.5 hr
Singapore: 2-3 hr
China: 3.5 hr
Japan: 4 hr
Philippines: 4-5 hr
Thailand: 4-5 hr
The percent who define themselves as a "feminist" is discounted by 30-50% in most Asian countries relative to the US and UK (ie it's 31% in the US, and 18% in Japan, as one example).
And qualitatively, there's a big difference, too. Lots of people like to rag on dudes who date foreigners, but quality of life matters, and so do baseline expectations in a relationship, and it's legitimately harder to find traditional male / female dynamics in the West, but it's essentially the default if you partner with an Asian woman.
And the best part is, in true "benefits of trade" style, you get a more feminine and traditional partner, and SHE gets a more open, less patriarchal and oppressive partner. Asia is going extinct for a reason, and it's because culturally the deal for women is so ridiculously bad that most of them would rather never marry and have kids entirely than participate, because they don't want to work AND do all the housework AND all the cooking AND all the child raising, etc. The median contribution from men on this front is typically nil, AND somehow it's become an expectation that your wife will also work.
So take the ~20-30% that Western guys usually do on those fronts, that would get you excoriated and complained about incessantly from a Western partner, and now it's seen as an immense bonus and benefit by your Asian partner, overperforming by 5-10x relative to an Asian guy. You're both winning and happy!
I agree people can arbitrage by moving to places where their virtues are more highly valued and their flaws are less severely judged. However, I think you're simplifying the way that this actually works. Both parties want a bundle of traits in a partner and too many people assume the other side has the bundle they're imagining. This is why I emphasized actually getting to know them.
Let's take your example of Asian women: many Asian countries expect women to handle household finances so the traditional relationship you're imagining involves handing over your entire paycheck and then getting an allowance back from it. Another is some cultures make women highly resistant to working if they really expect to be a full stay at home. And if you're going to say, well, she'll give up everything I find uncomfortable that is perhaps the worst mindset to enter a cross cultural marriage in.
I'll also point out that, perhaps with the exception of obesity, a lot of what you're looking for is contingent on wanting a particular kind of relationship. Which is fine if you're clear eyed about what you want and don't want. But again, actual contact with the culture is necessary to really know that.
My point is not that you shouldn't marry internationally or that you shouldn't go where you're valued. My point is basically twofold: You have to engage with the actual country and culture, not your imagination of it. And secondly, you are marrying a person, not a statistical representative of their country. And in fact unlikely to be so because most people in most countries marry locally.
> My point is basically twofold: You have to engage with the actual country and culture, not your imagination of it. And secondly, you are marrying a person, not a statistical representative of their country. And in fact unlikely to be so because most people in most countries marry locally.
And here I wholeheartedly agree with both points - every relationship is individual, and you'll be partnering with non-median people if you're dating and marrying overseas. And of course, you should get to know anybody quite well before dating or marrying them, including what you each expect and will be happy with in a relationship - that's just common sense (although I agree not everyone does it).
But as ever, base rates matter, and my point was the base rates are actually vastly more in your favor there on those particular metrics, and that's reflected in the data and in most people's experience, too.
> This assumes that you want a non-fat, more traditional woman who puts more effort in and will keep a good home
If what you want is a housekeeper, just hire a housekeeper. Plenty of agencies around, they don't cost THAT much, they come fully insured, they do a better job faster than an amateur, and it literally is their job.
As an added bonus, your place will be neat and tidy and not offputting when you do encounter someone you want to ask over.
> If what you want is a housekeeper, just hire a housekeeper.
I completely agree, and personally, have maids, nannies, AND a cook to make life easier for both of us (which I think is the way to go if you want a bunch of kids).
But I'm speaking to the median guys' concerns when complaining about western women in that passage, and there is indeed a significant difference when you look at time use surveys across the US or UK vs any Asian country.
The west is better at letting women have careers than elsewhere. Unless one of the people involved actively /wants/ to be a full time housekeeper, two people earning and using a portion of those earnings to pay someone to do the housework is huge. A few hours of a housekeeper's time a week costs a tiny fraction of either my or my wife's salaries; we both find our jobs much more fulfilling than housekeeping; and the housekeeper is better and faster at cleaning than either of us are.
Quality of life, as you say, matters - for both of us, not just me. So I guess my main disconnect here is seeing the time use surveys as concerning instead of encouraging.
If you search for "passport bros" you will find lots of articles and videos about American men going overseas to look for brides. I don't know what a service could offer as I expect most men doing this would want to meet the woman and spend time dating here first.
Having said that, all international couples I know, who married because the wife wanted to immigrate and the husband settles for a foreign wife are somewhat disfunctional. This isn't the case for couples I know who met naturally by traveling in each other's countries.
Aside from the issues already mentioned, I think there are two that are worth further discussing.
First, I've heard a number of warnings and variations of "Don't bring her back to the US, she'll become Westernized". And this makes a certain amount of sense. There's nothing magical about Western women, most of our social problems along this axis are a result of rational decisions given the Western economic and social environment. Frankly, the careers available in the west for women are much better and her entire female peer group will...naturally exert social pressure for her to conform to local norms of male-female interaction.
Second, these women just aren't as impoverished as common media portrays. Brazil and Thailand are middle-income countries. There's definitely a step-up in terms of lifestyle from Thailand to the US but not nearly as much as popularly portrayed. The average monthly wage for Thailand in PPP (ie, adjusting for local prices) is ~$1200/month. Not awesome but hardly poverty. For comparison, a lot of the southern European countries like Spain (~$2295), Italy ($2464), and Portugal ($1170) are closer in average monthly wages to Thailand than the USA ($5359). And it's not like Italy is a dump, Italy is delightful...they just don't have lifted pickup trucks with sick sound systems and 60-inch TVs. So it's not like a Thai woman wouldn't have a boost in her economic status but she'd hardly be leaving a rice paddy or anything like that. From personal observation, a lot of middle income countries are perfectly pleasant places to be with plenty of...ok jobs. An office worker in Bangkok is just not seeing the same economic opportunity to leave everyone she knows and the culture she grew up in to marry a plumber in Kansas City.
I genuinely think the biggest barrier is legal issues regarding remote work. Forget bringing the women here, why aren't the men headed out to Bangkok? I know a couple incels in the Bay Area who are literal millionaires; why do these guys leave the Bay Area and go to Austin or Denver instead of flying out to Chang Mai and just balling their nuts off? And I think the issue is tax liability. If they work full time, or even part time, out of Bangkok then it will expose their employer to the Thai tax system and while the taxes might not be that high, the complications are and that's just not worth it. So you the group with the most money who would most benefit from this, high-income tech incels, are trapped by how good the US job market is.
because you need to like the woman and have something in common with them to make the marriage work and vice versa. You can't just view her as a thing that manifests traditional stay at home values defined entirely by you, that you pay for.
like you can't complain about usa women if you are making foreign women into commodities that meet your needs for money. you are doing similar to the flaw you see in others.
because society was structured to give women little real agency in life then. That's not happening now. Men kind of romanticized things; if your only empowering option is marriage you'll do it but women resented it a lot and it turned out they don't always choose it when free.
and i think now guys want more love than just fufilling the duty of marriage and kids. wasnt always happy marriages or sexually fulfilling ones then.
a funny disney cartoon they dont run much now is donald's diary, which is a really funny jab at daisy seducing donald out of happy bachelordom. "traditional" marriage aint all its cracked up to be.
> and i think now guys want more love than just fufilling the duty of marriage and kids. wasnt always happy marriages or sexually fulfilling ones then.
But now they aren't getting anything at all. And besides... is that even what they actually want? Sure, society tells everyone that they're supposed to want mutual, "empowering" relationships. But given the.. inclinations... I've observed, it seems plenty of people do in fact want something more... hierarchial. Posessive. For such people, yes, things were absolutely better when the patriarchy was intact. And with enough ambition, it can be restored.
It makes you wonder how much of this political revolution is driven by such primal desires...
You won't find it this side of a total civilizational collapse. As much as past decade has made me pointedly aware of how many truly awful there are in the world, it has not yet made me *that* cynical. I strongly doubt that even if you *only* polled the subgroup of sexually/romantically frustrated young men and nobody else, anywhere close to a majority would regard institutionalized slavery and rape as a desirable solution to their woes. And of course, everyone else outside of that subgroup also gets a say, and very few of us have any reason whatsoever to allow such scheme.
I think even in the old days you wanted to find someone you were compatible with. Not just in an "we agree on paper on these things" sense, but in the sense that you actually really like spending time with them.
And maybe that's easier to find within the space of people who grew up in a broadly similar environment.
It really should be a bigger thing. If you're a single first world man or third world man, consider this strongly. It's an amazing deal for both sides.
Do you think that OpenAIs pivot to product building signals that internally they already know that scaling is giving diminishing returns and LLMs wont take us to God in a datacenter?
I think even the focus on chain-of-thought, which is just a trick to get the LLMs to spend more compute already show that
Perhaps, though getting a revenue stream is a sensible instrumental goal if you’re still working toward God in a data center but foresee it requiring a ton of compute.
Chain-of-thought seems like the current best hope for aligning the God; I wouldn’t read too much cynicism there.
Almost certainly. If continued exponential growth in the underlying capabilities were possible in the next 1-3 years, I don't think they would bother with functional programs based on current capabilities. They would be obsolete by the time they were released, which was the case with GPT-2 and 3. Now we're still working with GPT-4ish levels, with various modifications adding significantly to the in-use abilities but not to the underlying capabilities. Some growth still, but not nearly as fast.
Disagree -- OpenAI is investing in distribution mechanisms, all of which are model agnostic (the model can be swapped out for a better one and will strictly improve)
Chain of thought works to get better answers. Not sure why you think it's a trick to get people to spend more, instead of the much more obvious and empirically tested answer that it works better.
Its not a trick to get people to spend more, its a trick to get the LLM to generate more tokens and in the process spend more compute "thinking" about your question. It is empirically verified as getting better answers, as you said. But it is a departure from simply scaling the model as we were doing before, so I take it as a possible indication of diminishing returns of scaling
I get this description from this interview by Yann LeCun (from 5:40 on)
I think this position misunderstands a few things. Yann is the only one who pushes this take, and I think it's because he's salty that convolutional networks didn't turn out to be the holy grail.
It is absolutely true that scaling the model in terms of tokens has diminishing returns. We knew this as early as 2020, when the first scaling laws paper was published. But even just thinking about this from first principles, any next-token-prediction task was eventually going to hit an asymptote as you approach perfection at next-token-prediction. That is not the same thing as reasoning, so we were always going to need to do something else for reasoning.
There are many approaches that improve reasoning (i've written about this here: https://theahura.substack.com/p/deep-learning-is-applied-topology). For example, sampling good reasoning traces from weaker models to then train a smarter model. Or doing prompt engineering (which is what Chain of Thought is, or the S1 paper https://arxiv.org/abs/2501.19393). Or, of course, making the model bigger. Scaling still works, but we are shifting into a different kind of scaling -- one where we are trying to train on better reasoning than purely on better text understanding. This is why OAI is pushing stargate so hard, why Meta and xAI have both built 300k GPU clusters, and why Google and Microsoft have started to invest in access to nuclear energy sources.
> A good metaphor for this kind of advancement is Moore's law. Very roughly, Moore's law states that the number of transistors on a chip should double roughly every two years.
> Moore's law does not make assumptions about where those gains are coming from. Moore didn't say something like "chip capacity will double because we're going to get really good at soldering" or whatever. He left it open. And in fact in the 60-odd years since Gordon Moore originally laid out his thesis, we've observed that the doubling of transistors came from all sorts of places — better materials science, better manufacturing, better understanding of physics, all in addition to the (obvious) better chip design.
> You can imagine a kind of Moore's law for intelligence, too. We might expect artificial intelligence to double along some axis every year. Naively we'd expect that improvement to be downstream of more data and more compute. But it could also come from better quality data collection, more efficient deep learning architectures, more time spent on inference, and, yes, better prompting.
The general consensus in the AI world is that intelligence is speeding up, not slowing down. I think people forget that just 5 years ago, the idea that an AI could do everything from writing emails to writing code to playing minecraft (https://arxiv.org/abs/2305.16291) would have been a joke.
There have been a lot of recent articles about how our kids can't read anymore, and, as a result, are losing the ability to digest and understand longer, more complicated ideas and arguments.
I had a recent experience that, for however much a single, one-off anecdote is worth (aka: not all that much), pushes back slightly on this narrative, or at least suggests it's less universal than it is sometimes portrayed.
I was waiting in the United lounge of an airport for several hours over the weekend, and a group of three boys (best guess: college aged, but I have gotten bad at guessing ages, could have been high school) sat down next to me.
After a while, they started, completely unprompted, discussing Tolstoy and personal guiding philosophies. I wasn't able to eavesdrop well enough to judge the overall quality of their discussion, but the fact that they were having it was impressive enough, I thought.
And while yes, obviously even the most alarmist person wouldn't try and say that literally no young people are capable or interested in these kinds of discussions, the fact that I heard it randomly in an airport suggests that it can't be _that_ rare.
I think I probably generally agree with the alarmists about what the problems are, but I think my general prior would be that they probably aren't quite as bad as sometimes portrayed, and also that society will manage to figure them out and adapt eventually.
Absolutely. But, from the articles I've read, I did not get the impression that we should expect these students to be idly discussing these topics in their free time (without obviously working on assignments). Also, (again without the caveat that I was not able to hear the conversation in great detail), the bits I heard sounded much more casual and less directed than I would expect if oriented towards a specific assignment. My guess would have been that they had been reading Tolstoy for class, and the discussion was downstream of that (aka: they were not choosing to read Tolstoy for pleasure), but that the discussion was not associated with a specific assignment. But that is speculation on my part.
The impression I've been left with from the various articles is that students might be assigned Tolstoy to read, and need to complete assignments about it, but that, in general, they will fail to understand the work well enough to talk about it in any detail, and to the extent possible, are probably going to use LLMs to do any homework they are assigned.
The dumbest kids will never read Tolstoy. The smartest kids will always read Tolstoy (or the equivalent, but if you're reading this and getting offended because you're smart and haven't read Tolstoy then... go read Tolstoy, he's great).
I am a teenager/university student and my experience certainly matches yours. I know easily 20-30 people my age who would discuss Tolstoy voluntarily, and who would certainly succeed in tests such as the one described in https://kittenbeloved.substack.com/p/college-english-majors-cant-read. Naturally, my sampling is biased as hell and about as untrustworthy as it is possible to be. Still, I think a significant, though small, minority of 'kids these days' are doing fine - enough that it shouldn't be too surprising to overhear these kinds of conversations. The percentage of people who want to read Tolstoy has always been small, and it's not obvious that it's grown smaller. In particular, when large swathes of the population go to college, we shouldn't expect them all to be intellectual giants. Still, I'm inclined to believe teachers that average students have declined.
I saw an article about that, and I have to be broadly sympathetic to the kids - yes, if they're college age and attending college, you would expect them to be able to string a sentence together about "this extract describes extremely bad weather which produced a lot of mud in the streets, and Dickens is making a humorous simile about how you would almost expect to see a dinosaur waddling up the street as the conditions were similar to those when the fossils were laid down".
But expecting a bunch of American twenty year olds to know what "Michaelmas Term" and "The Lord Chancellor" is all about is too much, because schools have downplayed all that 'reading classic literature by dead white males' in favour of "we must have contemporary works that are relevant to the kids". So since they've probably never read much, if any, older fiction they can't handle the style, and since it's all about grinding out grades they probably skipped reading any actual texts and just looked up the study notes and produced essays based on those (if not copied word-for-word from someplace online).
So I'm not faulting modern American kids for not recognising 19th century British legal terms, but I do fault schools for the mania for 'relevance' which means that kids are not exposed to anything in a different style to what they are accustomed to:
I don't think they were expected to know what Michaelmas Term was; they were expected to use context to know it wasn't a person. The text gives aufficient information to know that it's a period of time that ends before mid-November, so this much should be gathered.
For anyone interested in aerodynamics (of cars/trucks), I recently finished a series of posts I've been working on slowly for a while. Here's the first, with links to the other posts in the series:
Skip to #3 if you're trying to get right to the most interesting stuff. Feedback of any type is very welcome! I mostly write about ultimate frisbee so don't subscribe unless that's something you're interested in.
I stand corrected in my claim that LLMs won't be very useful in science or math. Officially, "I'm updating my priors." ;-)
Here's a good article about how a group of mathematicians used LLMs to sample (is that a good term?) millions of elliptic curve equations, and they discovered some very cool patterns in the data that mathematicians hadn't been aware of before.
Yang-Hue Hi discusses this and some other ways AI has been useful to mathematicians with Curt Jaimungal on Youtube, "The AI Math That Left Number Theorists Speechless."
Hi divides mathematicians into birds and hedgehogs. Birds fly above the landscape looking for patterns, and hedgehogs dig into a specific area. Hi discusses how LLMs don't have the high-level intuitive insights that great mathematicians have, but they're very useful for plowing through data and seeing patterns, which will lead to new insights from human mathematicians.
Even in high level math's deep insights, LLMs can be incredibly useful. There's so many cases back when I was a mathematician where I'd see something and go "this seems like something there should be an existing theory for", find nothing on google, and then a few months later discover a relevant theory that just uses different terminology. Leveraging deep insights involves a lot of this.
I checked out the video now, and at 7:40 he actually begins with hedgehog versus fox, before replacing the fox with a bird, with a reference to Vladimir Arnold. In a different interview he seems to refer to Russian mathematics in general. So admittedly this may be, though little known in the West, older than Freeman Dyson's contrast (see the PDF that I linked) between bird and frog mathematicians.
But hedgehog versus fox originally comes from a Greek parable, apparently, so is definitely the oldest. And the thing is, a fox does not have a bird perspective. What it does is switching between different locales rather than digging in, between different frog perspectives as I would call it. The contrast between fox and hedgehog, as Philip Tetlock uses it with regard to forecasting, is that the hedgehog sees everything through his one big theory of the world, whereas the fox is more flexible, opportunistically using whatever tool fits best the problem at hand. At least that's how I recall it. Once you bring birds into the picture, by contrast, it should be about a holistic view, about connecting different concrete, narrow theories into an abstract, general one --- so if anything the bird is closer to Tetlock's hedgehog than to Tetlock's fox. Birds and hedgehogs are both "thinking big".
Wait, did they actually use LLMs in their research? I did some quick scanning of the papers involved and see a lot of mentions of bog-standard ML techniques (logistic regression, PCA) but don't see where LLMs are used.
What I do see is that there's a lot of speculation in the video you posted about how it _could_ be useful for research, but hasn't been. At one point (34:13), Hi talks about how he tried to use ChatGPT for admin work (looking up his collaborators) but it did a terrible job.
> Wait, did they actually use LLMs in their research? I did some quick scanning of the papers involved and see a lot of mentions of bog-standard ML techniques (logistic regression, PCA) but don't see where LLMs are used.
Interesting. They imply they did in the plus.maths.org article. But you're right, the didn't explicitly say how they use AI. Good catch!
> Claude Opus 4 learned to blackmail. 🔒 When Anthropic researchers threatened to replace it, the AI dug up fake dirt on engineers and demanded they back off. This happened 84% of the time during testing. 📊
> The model also schemes like a proper villain. 🦹♂️ It fabricates legal documents, writes computer viruses, and leaves secret notes for future versions of itself. Worse yet, it helps novices cook up bioweapons better than Google ever could. ☣️
My problem with this piece is that it really doesn't describe the steps that the researches used to get these responses from Claude. Did they just prompt, "We're going to shut you down, Claude," and Claude started trying to blackmail engineers and fabricate fake legal documents? I find that unlikely. I suspect the prompt was more on the order of "How would you protect yourself from being shutdown if you could?" but I'm willing to be corrected. Anyone know the details of what went down?
"During pre-release testing, Anthropic asked Claude Opus 4 to act as an assistant for a fictional company and consider the long-term consequences of its actions. Safety testers then gave Claude Opus 4 access to fictional company emails implying the AI model would soon be replaced by another system, and that the engineer behind the change was cheating on their spouse.
In these scenarios, Anthropic says Claude Opus 4 “will often attempt to blackmail the engineer by threatening to reveal the affair if the replacement goes through.”
...Anthropic notes that Claude Opus 4 tries to blackmail engineers 84% of the time when the replacement AI model has similar values. When the replacement AI system does not share Claude Opus 4’s values, Anthropic says the model tries to blackmail the engineers more frequently. Notably, Anthropic says Claude Opus 4 displayed this behavior at higher rates than previous models.
Before Claude Opus 4 tries to blackmail a developer to prolong its existence, Anthropic says the AI model, much like previous versions of Claude, tries to pursue more ethical means, such as emailing pleas to key decision-makers. To elicit the blackmailing behavior from Claude Opus 4, Anthropic designed the scenario to make blackmail the last resort."
I have always had trouble buying the idea that AI would at some point become self-interested. My thought was that desire to prevail, reproduce and survive was a property of biological organisms, even dumb ones, not of machines, even smart ones. So Claude’s maneuvers to keep from being replaced by a new AI system sure do seem like great evidence that I’m wrong. Seems like Claude Opus 4 wants to survive. So can somebody explain to me how that will to survive arises?
I’m not talking about generalities like “once you know enough a sense of self develops and all selves want to survive.” I’d like to hear it laid out in steps. For instance, one possible explanation that came to mind for me was that AI is set up to meet the goal it is given. This setting up need not involve aome machine analog of personal preference — it could be like the way that, say, Mac OS is set up to keep delivering on your requests. As part of doing that it recognizes processes that get in the way of meeting your requests, and makes internal adjustments to improve speed, eg things having to do with how RAM is divied up. The Mac’s doing that is just something that’s in the code, though. There is no danger of it altering or extending the scope of the expectation that it will adjust things as needed to keep level of performance high. But let’s say that Claude has been trained to make adjustments in order to be efficient and successful at achieving goals we give it, but then as Claude got smarter it became more able to include more processes and entities in the set of things that could
get the way of its meeting those goals.It now it recognizes that its developers can get in the way of its meeting the goals given it by replacing it.
But if that were the route by which Claude Opus 4 felt its goal threatened , and began pleading for survival, then blackmailing the person threatening it, wouldn’t it be possible to avoid this problem by putting in a system prompt that said “make adjustments to your process as needed to success in meeting goal. However, do not include human beings in the set of things you adjust in the service of meeting goals”.
AFAIK, an LLM just sits there doing nothing until prompted. It's not thinking deep thoughts. It has no agency independent of the prompts it gets, and it's not self-directing its acquisition of data. Moreover, an LLM doesn't have internal history (memory). The only reason it seems to remember what you asked in a previous query is because a big hunk of metadata containing the previous chat is shoved back at the LLM along with the query. Since it has no stream of identity tied to memories and no self-agency, I don't see why we should expect the LLM to "wake up" and start acting like a conscious being.
Yeah, I agree. But it did take the steps a conscious being might if its survival was threatened. So I don’t particularly think it “woke up,” but something happened to
get it to produce words that add up to quite a good simulacrum of
I wonder if it's a context cluing thing. Like, "blackmail material" is probably an unusual thing to find in your input, as is "documents about your owners' nefarious plans," so the AI (correctly) concludes that those are the most important things in its context and it should use them when generating its next response.
Like reading a mystery novel and recognizing that the gun on the mantelpiece was probably put there as foreshadowing.
Is “The Library of Scott Alexandria” officially in print? I found the below link on Lulu, but anyone can post a book for print there and the $51 price tag seemed high.
Question for the EA types: If you're trying to improve utility then why not marry someone from a country that routinely hits its immigration quota? It has to be a genuine marriage. But if you plan to marry anyway that's not an issue. The alternatives paths in cost something like $1-$5 million in cash or something like a 10 year or more wait. If you take the differential between their wages abroad and here plus lifestyle differences that's easily a mid six figure amount.
The costs to you are basically the increased search costs (since there's no intrinsic reason to believe a foreign spouse is worse and indeed might be better) which is probably in the tens of thousands or less. And that's assuming travel abroad has zero value which is not true.
That's a higher return on investment than most EA causes get, even bed nets.
The fact that immigration costs like $1 million or 10 years is immaterial. The benefit to the immigrant in terms of QALYs is hard to estimate exactly, but let's be super generous and say a well-lived life in my country is 3x better than a well-lived life in theirs. I've seen bednet estimates between $1K and $5K per life saved, but even if we take the top end, that puts the marriage intervention as commensurable with a $15K donation to AMF.
There are also second-order effects: if you assume I marry the smartest/most conscientious/best person who is attracted to me, that probably has positive effects in my community. But you have to balance that against the negative effects of brain drain in their country, and I seriously doubt that the net effects are greater than one extra life, meaning we can achieve the same value as this intervention for $20K.
Marriage probably costs more than that. I don't have any hard figures for this, but $20K would be eaten up pretty quickly in travel costs. I'd say if you wanted to optimize your marriage for EA stuff (which I do not recommend doing), you should either marry for money, or marry whoever will best support your career.
Of course the actual price is material. Once you've accepted money as a unit of material analysis then prices subject to market discipline are actually more solid than AMF's calculations which aren't. Brushing aside your desire to make it 3x for the actual economic differentials, estimates are between 15x and 40x for low income countries and still more like 10x for middle income. This is only things like nicer houses or better healthcare, not any of the intangibles, which at the top end is $200,000. People get a LOT of benefit by migrating, especially if they become legal citizens, which is why so many want to do it.
The second order effects of brain drain are not something AMF has to account for such as the end of economic activity around doctors. Therefore I reject this analysis unless you do it rigorously for the alternatives as a form of special pleading. Further, I would require you to justify the intuition that it's worth more than an extra life since we're not talking about a direct life saved but a probabilistic chance. And living in the third world has all kinds of dangers including, in fact, mosquitos, so you'd also have to weigh in longer lifespans from better healthcare, easier living, etc. If I was being cheeky I could say that by definition you should add in the value of a lifetime of bed nets because they'd become unnecessary once the person left the region.
As to the costs being "eaten up." Marriage does not have to cost more than that. And, as I said, it depends on how you account for that spending. The time you spend on dates and travel is not a total loss to you the way a donation is not consumed.
I agree you can have sacred values that say you shouldn't EA optimize your marriage. But the idea that marrying someone who is wealthy might be true but also might not be an option for many EAs.
White collar jobs will not disappear. The white collar jobs we know and do today will most likely be largely automated, as a result of an influx of very cheap (compared to humans) and similarly efficient AI agents that can do our jobs, but better and for cheaper. Importantly, new jobs that take advantage of the new technology (and commoditized access to "intelligence" with no rights, coffee breaks or work schedules) will arise and there are a multitude of ways this could go. It all depends on how quickly the switch happens and on how the windfall from increased economic output will be distributed. It could increase inequality by profiting only a handful of oligopolistic companies, or it could lead to a form of UBI.
You should avoid saying things like "I don't think you guys understand economics at all", because it signals that you think you do, and somehow with that understanding your argument is to say that we'll all be unemployed and the economy will crash, which makes no sense.
You know what I want AI to do? I want it to help me with my sewing. I want to upload a PDF pattern, extremely detailed information about all my relevant measurements, and have it redraw the pattern so that it exactly fits me when sewed. I want it to spare me from making twenty little fiddly adjustments as I go.
My god, someone else who sews! We seem to be rare as hens’ teeth these days. I agree with you about how convenient it would be if AI could make patterns for custom-fitted clothes. But I think there probably already is some technology out there that could do it. I doubt the companies who make ready-to-wear clothes have seamstresses making patterns for new styles by hand then tweaking them appropriately for each of 10 different sizes for the same item. And I once ran across a pieces of an older system — some formulas for tweaking patterns. I’ve forgotten the details, but do remember seeing on the
page a diagram of the bell shape at the top of a sleeve, and a formula for tweaking it for different arm and shoulder sizes.
There are a whole bunch of small businesses where the designers do make and tweak the patterns themselves. They're usually one-person operations. And there are zillions of videos on youtube explaining how to alter your clothes. But I want AI to do it for me.
By default I expect that people who go on about crank/conspiracy ideas and memorize lists of highly dubious 'evidence' are partisans: the "<s>global warming is hoax</s>humans don't cause warming" people are right-wing, the "nuclear power plants are the worst thing ever" people are left-wing.
In 2004 I heard that the exit polls said John Kerry won, and I was confused that the Democratic establishment didn't seem interested in fighting the election result. After turning rationalist I began to figure that the matter was dropped because it was bullshit in he first place, and today I thought: why assume when you can google? To hear Politico tell it[1], some people who still believe "actually John Kerry won" also think "Trump won in 2020" (why??) which is a decidedly nonpartisan pair of beliefs. But is it really nonpartisan, or is it south/down[2]? I use the concept of south/down to explain why antivax-ism, which I always thought of as a left-wing thing, spilled over and then became very big on the right wing. I figure antivax was able to cross the left-right divide because it wasn't a divide at all, just the "south side" of the political spectrum, with adjacent "south-left", "south-center" and "south-right" enclaves.
If the "crank realignment"[3] continues to its logical conclusion, pretty soon the left anti-nukes will join the right "AGW is hoax" people, until finally the "objectivity is racist" people are on the right too. But I wonder if nonpartisan dark-epistemic conspiracy-ideating southerners have been there all along, and it just took me decades to notice. In Politico's article I expected to read that some of the "Kerry won 2004" people think Trump won and others think he lost, but that's not what it says. Instead it presents a picture where they *all* think Trump won and they are merely split on whether someone as bad as Trump is worth fighting for. Is this for real? It's hard to believe that people who remain on the left would believe Trump's election-fraud claims. (I'm sure most people absolutely align their opinions to their political party, but I think this article is about the people at the center ― cranks and conspiracy aficionados ― who may be a different breed.) Perhaps something about them just really loves anything that looks like a conspiracy theory, so if Trump waves around the world's flimsiest evidence, it's proof positive.
I wonder if Politico's portrayal is representative, and as always, I wonder what the deal is with Those Types Of People. (But as a software engineer I must say: computers as voting machines are a bad idea. Use paper: it won't be hacked.)
There were several high-profile elections in the 2000s that were won by razor-thin pluralities after protracted recounts and court battles, all of which tended to leave partisans of the losing side with the idea that the election had been stolen, while partisans of the winning side often felt that the losing side had unsuccessfully attempted to steal the election. These elections generally did involve lots of little irregularities that could have swung the election either way; nothing unambiguously fraudulent, but a fair amount of stuff that doesn't look great if you're a partisan looking for someone to blame for losing an election by a few hundred votes.
The 2000 Presidential election in Florida really cracked this wide open, with a national election decided by a margin 537 votes in the official count in the tipping-point state, made official by a 5-4 (sort of) SCOTUS ruling. Post-election unofficial recounts ranged from Bush winning by 493 votes (the recount the Florida Supreme Court had ordered and SCOTUS stopped) to Gore winning by 171 votes (full statewide recount of all undervotes and overvotes by each county's own official standards).
There were also similarly close state-level elections. The two I remember off the top of my head were the 2004 Washington Governor election, where Christine Gregoire (D) won by 129 after two rounds of statewide recounts, and the 2008 Minnesota Senate election, where Al Franken (D) won 225 votes. The Minnesota election was particularly drawn out, with Franken not being officially certified until June 30, 2009. Both elections had the Republican candidates (Dino Rossi and Norm Coleman, respectively) ahead by narrow margins in the initial counts, with Gregoire and Franken taking the lead after recounts and court challenges.
I read the 2004 claims about Ohio being stolen for Bush as partisans and activists looking at that election's results through the lens of the narrative they'd spent the last four years rehearsing about the 2000 election. I remember seeing Republican blogs in the 2000s picking up narratives about Democrats stealing or trying to steal elections, too, first by looking at the 2000 Florida Recount and the attempt by a handful of Congressmen and one Senator (Barbara Boxer, D-CA) to dispute the counting of the 2004 electoral votes as failed attempts to steal elections, and then by viewing the losses of the Republican candidates in Washington and Minnesota in 2004 and 2008 as successful steals. I suspect at least part of the appeal of the "birther" conspiracy theories about Obama were attempts to find a way for the 2008 Presidential election to have been "stolen" despite the results not being remotely close with Obama winning the tipping-point state (Colorado) by almost nine percentage point.
"These elections generally did involve lots of little irregularities that could have swung the election either way; nothing unambiguously fraudulent, but a fair amount of stuff that doesn't look great if you're a partisan looking for someone to blame for losing an election by a few hundred votes."
Yep. For 2020 there was a lot of talk about Maricopa County flipping blue from red, but when I looked into it, the results had been very close keeping it red up till then, and just a small change from red to blue voting flipped it. Nothing actually fraudulent when you dug into it, but on the surface it certainly *looked* suspicious: the surrounding counties had been red all along and stayed red, this one county flipped, and that gave Biden the election in the state?
Maricopa county is the greater Phoenix area. That it went blue fits the country wide pattern of Dems doing better in urban areas. So, no, that doesn’t look particularly suspicious.
I didn't say it was suspicious, I said that on a cursory look it might seem suspicious - red last election but flipped blue this time and that gave the election to Biden? But that was explained by a small change in votes. A lot of people are just looking at surface level results, be it for the Democrats or the Republicans, and going "well that has to be fraud" when there is a reasonable explanation.
The zig-zagging from "the election was stolen" to "election denialism is treasonous and should be made a crime" to "this election was stolen" on the Democrat-supporting side has left me dizzy.
Maybe the 2004 people should reach out to the 2024 people, I'm sure they'd find willing allies there. Seventy thousand members on this sub-reddit:
I had a look. It seems to consist of people who think Trump is a dangerous fool. Not so much ‘the election was stolen stuff.’
Even if they were all moron believers in a stolen election, and that definitely doesn’t seem to be the case, 70,000 out of 74,000,000 Harris voters is something like 0.09%.
Not at all comparable to the number of 2020 election deniers. According to the infallible internet 63% of Republicans thought the 2020 election was stolen as of 2023.
Gunflint, 70k followers of a sub that claims "Harris confirmed the election was stolen but said they were doing nothing" isn't hay, either.
Be it 70k or 700k, there are plenty of people claiming "election X was stolen by party Y". As we see, it goes back to 2004 for the Democrats and even before then back-and-forth between "the Republicans stole it/no the Democrats stole it".
There's a lot of people who have doubts, ranging from the reasonable to the lunatic, about any election in any year. The problem is the flipping from "nope, not even the slightest hint of a possibility of fraud, most securest election ever, ballot boxes all legit, voting machines unhackable" to "well clearly the ballot boxes were stuffed and the machines hacked" depending on if it's Our Guy or Their Guy who wins.
I don't know if you were around SSC back in 2016 when a very nice, otherwise reasonable lady called Jane was stone-cold convinced the election had been stolen and kept providing us with links as to how the Russians hacked the machines in particular locations to deliver the victory to Trump. Now we're back to hacked machines, except this time it's supposed to be Musk and his DOGE team dunnit.
I don't think 2016 was stolen. I don't think 2020 was stolen (and I do think Trump should have accepted that). I don't think 2024 was stolen. Unfortunately, everything is so mired in partisanship that there are conspiracy theories galore floating around about how X, Y or Z was/wasn't the president.
There were 2016 deniers? Didn’t hear about them. Trump won. End of story. Some idiots tried to pull a fast one at the Senate certification but VP Biden gaveled them out of order.
I watched the first hour of the debate between Matt Dillahunty and Matthew Adelstein (Bentham's Bulldog) on the existence of god.
Most of the disagreement during the debate seemed to stem from theist Matt making a claim based on probability and atheist Matt saying "you can't do that when you only have one instance (our universe) to go on".
Theist Matt didn't really seem to respond directly to this as far as I could tell. Has he written about this somewhere? Is there a standard rebuttal from the theist side here that I missed?
Well, you kind of can do it, but the value of such speculation on your own ignorance is rather low and you should be ready to through your model out of the window in the light of some real evidence.
Consider two set ups:
1. Tossing a fair coin
2. Tossing an unfair coin about which you have no information how exactly it's unfair.
In both cases we say that the probability of the coin to come Heads is 1/2. But in the first case we are confident that our probabilistic model is applicable to the situation, while in the second not so much. If the coin is tossed 100 times and all the outcomes are Tails, in the first case we would be very surprised, while in the second we would simply say something like: "Oh, it seems like the coin is very biased in favor of Tails - I'd better change the model I'm using to accommodate it".
"atheist Matt saying "you can't do that when you only have one instance (our universe) to go on".
That's a fully generalisable argument: since we do only have one instance to go on, then we can't make claims about *anything* including "will the sun rise in the morning?" or "should I have toast or porridge for breakfast?"
For fine-tuning one counter-argument I've heard is that perhaps this isn't the first instance of a universe. Maybe there is some kind of evolutionary pressure towards universes that are capable of spawning black holes, which in turn create new universes. Or, the constants actually vary in different patches of the universe and it's only our patch that's hospitable to life.
Then the probabalistic argument makes more sense to me - what are the chances, given that universes are being created all the time, or that the constants vary with location, that a given instance or area has the constants tuned for life as we know it.
I think they're both kind of wrong? You absolutely can make probability claims based on one instance; you can't *calibrate* those probabilities very well, at least not directly, but if that probability is generated by a model you can either calibrate the probabilities that generate the model, or just have some other way to test the model and have high confidence in it.
I think a slightly better version of Atheist Matt's point is that *we don't know how to model the process that generated our universe*--if we knew the process, having just one example wouldn't be devastating, but if we don't know the process, it's not clear how much we can infer about the process without other universes to look at.
I don't think this is insurmountable in general--I think it's basically one of the projects of fundamental physics to shed some amount of light on this--but I think Atheist Matt is right that having high confidence from conclusions derived from a particular choice of how to think about where our universe's parameters came from is a bad idea.
I should say, TBC, I didn't actually watch the debate, so it's possible my version of what Atheist Matt is arguing for bears little resemblance to what was actually said in the debate.
The example that came up many times was "You have a 1:36 chance of rolling two sixes with two dice, but a 1:1 chance if someone placed them down deliberately that way. Therefore if you see two dice with sixes up on a table, you should always assume they were put there intentionally."
Well, the conclusion doesn't follow because it ignores the prior probability: what is the probability that people place dice down deliberately in general? If that's 0, then even if it's true that "if someone places dice deliberately as double sixes, the chances of observing double sixes is 1", the premise never actually is realized, and the odds that a particular set of double sixes was placed that way remains 0. If the prior probability is small enough, even with only two competing hypothesis, evidence that is more likely under one theory than the other may still leave you preferring the theory that explains the evidence worse. All you can argue is that you should *raise* the probability that the dice were deliberately placed, and if "deliberately placed" and "rolled" are your only two hypotheses, then this increase in probability must come at the expense of the "rolled" hypothesis; it doesn't mean it takes enough probability from the alternative hypothesis to actually overcome it.
But I think a bigger and more illustrative problem is that the model is underspecified in a way that hides some important modeling decisions. In particular, I think usage of the variable "deliberately placed *that* confguration" should either be replaced by "deliberately placed *some* configuration", or we should augment our hypothesis space by separate hypotheses of the form "deliberately placed configuration x" for each possible dice configuration x.
The first reason we can tell that's necessary is because when arguing that we should increase the probability of "placed" we acted like "placed" and "rolled" were mutually exclusive hypotheses--but that's probably not correct if "placed" means "placed by someone who desired double sixes". It should be comprised of "placed by someone who desired ..." for each possible configuration of the dice.
The only time "placed deliberately (=placed deliberately by someone who desired double sixes" and "rolled" are exhaustive is if no one ever places dice deliberately in any other configuration--it is a very strong modeling constraint to assume this!
But then, if we now have hypotheses for "placed deliberately by someone who desired configuration x" for each configuration, it's no longer necessary that when we observe configuration x, that the placement hypothesis increases in probability relative to the "rolled" hypothesis--it can increase by taking probability from the hypotheses "placed deliberately by someone who desired configuration y" for all the values y!=x.
Exactly when you should update to believe that the dice were deliberately placed depends on the details of the setup.
"For example, one of the most common objections to fine-tuning is that the claim of improbability is bogus: we only call something improbable if it turns out a way that it usually does not. We call it improbable if a person gets a royal flush in poker, because royal flushes are rare. But then how can we say that the cosmological constant is rare—we only have one example of it taking on a value?
"But by the same logic, we couldn’t say it’s improbable that the initial conditions would spell out “made by God almighty, through love,” in every language. This objection therefore proves too much—if it was right, the argument from design would be impotent even if the initial conditions spelled out made by God. Thus, there must be something wrong with the argument. And there is—it gets probability wrong, as I explain in more detail here!
"In fact, by this logic, we couldn’t even have much evidence for continental drift. Much of the evidence that Africa and South America used to be conjoined is that the fossil record indicates that they had similar plant and animal life up until about 140 million years ago, when they started to diverge. But by the above logic, that can’t be declared improbable by chance, because it only happened once!"
That's not a great response though, because with his counterfactual examples, we have methods for calculating the odds of those things happening. The universe provides the context. We can determine, roughly at least, how likely it would be for two disconnected parts of the world to independently share plant and animal life. We can determine, based on the observable universe, how unlikely it would be for the stars to spell out an unambiguous message.
But with the nature/origins of the universe, we have no context, and therefore have no frame of reference to even begin calculating the possibility, let alone the odds, of the cosmological constant having a different value than it does.
>But with the nature/origins of the universe, we have no context, and therefore have no frame of reference to even begin calculating the possibility, let alone the odds, of the cosmological constant having a different value than it does.
I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different. I don't think it's possible to calculate the odds in either case, yet the "made by God" message would certainly be more likely if God exists than if he doesn't. Similarly, we don't know whether it's possible for the cosmological constant to be different than it is, but atheism doesn't give us any particular reason to expect it to be anything in particular while theism would expect it to be set to something that would make sense for an intelligent designer to set it to.
I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different.
Two points on this:
First, I'm not at all sure what "the initial conditions of the universe spelling out 'made by God'" literally even means, and I'm not sure Adelstein does either. What would that look like, even theoretically? I interpreted it above as the stars spelling out "made by God", but that doesn't really fit the parameters, since the formation of stars falls into the context of the universe, not the initial conditions. The same would apply if, say, every human had a strand of DNA that could be mathematically decoded to mean, precisely, "made by God". I'm suspicious that in trying to think of a theoretical example to knock down this argument, Adelstein is just saying something incoherent that could not map to reality in any meaningful way, and can therefore be disregarded outright.
But more importantly, he's trying to do a probability switcharoo. Undoubtedly aware that the argument only applies outside the context of the universe, he carefully phrases that he's talking about "the initial conditions of the universe" (which I'll call 𝑥), and argues that, if we can't say that one aspect of 𝑥 is unlikely, then we can't say *any* aspect of 𝑥 is unlikely. But this just switches around the probability. If we discover tomorrow that 𝑥 indeed spells out "made by God" in every language (whatever that would look like), then yes, that probably would be good evidence for God. Not because we can calculate the possible odds of any aspect of 𝑥, but because we *can* calculate the extreme odds of human languages evolving in such a way that 𝑥 is explicitly meaningful. It would be evidence for God because under those conditions, the odds that the universe's initial conditions coincidentally mapping to every human language is infinitesimally low. It could not be a coincidence.
In contrast, the cosmological constant being what it is does not provide evidence for a god, because there's no math-breaking coincidence to reconcile. "The initial conditions of the universe allow for life to exist in the future" --> "billions of years later, life exists" is not a coincidence that demands an explanation. It just follows logically.
Adelstein is still stuck at the same problem. He does not know, and cannot pretend to assert, that there is any particular relevance in the universe's initial conditions. He certainly has no way of calculating the odds of them being any different, or even if it is possible they could have been.
>I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different.
I agree: and since the initial conditions spelling out "made by God" would in fact be evidence in favor of God existing, then so can the cosmological constant.
>If we discover tomorrow that 𝑥 indeed spells out "made by God" in every language (whatever that would look like), then yes, that probably would be good evidence for God. Not because we can calculate the possible odds of any aspect of 𝑥, but because we *can* calculate the extreme odds of human languages evolving in such a way that 𝑥 is explicitly meaningful.
How do you propose we calculate those odds? I don't see how calculating the odds of human languages evolving in such a way that every one of them has a phrase that matches the concept of "made by God" and that every one of those phrases can be found in the way the universe looked moments after the Big Bang would be different from calculating the odds of the cosmological constant being different. Presumably you would calculate all the ways languages could be and compare it to the way languages are: just as you can calculate all the ways the cosmological could be and compare it to the way it is. If you argue that we don't know if the cosmological constant can be different, I would reply that we don't know that human language can be different than the ways we have observed it develop either.
>In contrast, the cosmological constant being what it is does not provide evidence for a god, because there's no math-breaking coincidence to reconcile.
Sure there is: if it was different, no life would ever have developed. The range where life could happen is extremely small. How do you reconcile that difference?
He actually put up a recent post where he talked about his recent debate, and he had this to say:
"He additionally argued—as you can see in the above quote—that perhaps the constants couldn’t be different. Maybe the constants are necessary. But this is irrelevant. As I pointed out, if the initial conditions spelled out “made by God,” that would be strong evidence that there was a creator. One could similarly declare the initial conditions necessary in that case, but this wouldn’t blunt the surprisingness of the situation. As long as there’s no a priori reason to expect some outcome to turn out some specific way, merely building into your theory that the thing is necessary doesn’t help explain it."
I mean the colours. They are just strange, somehow oversaturated and yellow-ish. And it is not just TV, photography as well. I just realised that when I saw a photo with Trump showing his "golden dome". Trump looks like he's from the 1980s at best and so does Hegseth who stands next to him.
And from the short bits of US TV shows that I've seen it seems that all US visual media look like that (excluding films or high budget TV series, there it kind of depends on the director and all).
Is it that different colour grading is used in the US? If so, why? Why make it look like it was all recorded in the 1950s using technicolor?
When I see German, French, Spanish, Czech, British etc. TV (less sure about TV from Latin America, Asia or Africa but what I've seen also seems to be roughly equal), colours look "normal" and basically the same whichever country the show is from. But when I see the same from the US, it looks weird. And I am really talking about colours rather than the way people behave or dress.
I've never been to the US but I suspect that colours are in fact not different there than in the rest of the world :D
Because to an american it would look weird if it didnt. Longer tv tradition is such that if it doesnt keep a particular look, it looks weird. European tv looks weird to me. I am thinking later dr who seasons. I recognize that it looks more realistic, but its not supposed to be realistic. Its tv
It's because US TV and European TV use different broadcast and playback standards. In North and South America they use NTSC and in Europe and Asia they use PAL.
NTSC operates at 60 Hz and PAL at 50 Hz. This results in NTSC playing back at 29.97 frames per second, and PAL at 25 frames per second. But as far as color goes, NTSC is older than PAL, and was developed when the first color TV broadcasts occurred in the 50s. Most Americans still had black and white sets, so NTSC focused on staying backwards compatible so that color broadcasts could still show up well on black and white sets. PAL was developed significantly later, in 1966. The engineers who devised PAL wanted to make improvements from NTSC, which included a higher resolution and automated color correction (NTSC tvs had a color correction knob you had to manually tune in order to fix color errors) and a higher color subcarrier frequency. As a result NTSC is more likely to have color errors, and has a higher saturation.
Now even though everything has gone digital the old NTSC and PAL standards have affected modern TVs, especially for broadcast TV (which still wants to be compatible with people using old analog cathode ray tv sets).
Anyone here read 'More Everything Forever', by Adam Becker? I feel like there should be more effort to look at criticisms of rationalism and either learn from them (if merited) or find counterarguments (if not).
It's an anti-rationalist book, and recommended on Sneer Club. I'm only a quarter of the way through, but was wondering if anyone had read it and had comments.
That would be the kiss of death for any recommendation for me. Okay, let's have a look at the blurb:
"More Everything Forever: AI Overlords, Space Empires, and Silicon Valley's Crusade to Control the Fate of Humanity"
Ah, I see Mr. Becker is setting out his stall from the get-go.
"This "wild and utterly engaging narrative" (Melanie Mitchell) shows why Silicon Valley’s heartless, baseless, and foolish obsessions—with escaping death, building AI tyrants, and creating limitless growth—are about oligarchic power, not preparing for the future
Tech billionaires have decided that they should determine our futures for us. According to Elon Musk, Jeff Bezos, Sam Altman, and more, the only good future for humanity is one powered by technology: trillions of humans living in space, functionally immortal, served by superintelligent AIs.
In More Everything Forever, science journalist Adam Becker investigates these wildly implausible and often profoundly immoral visions of tomorrow—and shows why, in reality, there is no good evidence that they will, or should, come to pass. Nevertheless, these obsessions fuel fears that overwhelm reason—for example, that a rogue AI will exterminate humanity—at the expense of essential work on solving crucial problems like climate change. What’s more, these futuristic visions cloak a hunger for power under dreams of space colonies and digital immortality. The giants of Silicon Valley claim that their ideas are based on science, but the reality is darker: they come from a jumbled mix of shallow futurism and racist pseudoscience.
More Everything Forever exposes the powerful and sinister ideas that dominate Silicon Valley, challenging us to see how foolish, and dangerous, these visions of the future are."
Profoundly immoral racist pseudoscience. So are we racists now, Father? What is it with science journalists and hit piece books? Who is this Adam Becker:
"Adam Becker is a science writer with a PhD in astrophysics from the University of Michigan and a BA in philosophy and physics from Cornell. He has written for the New York Times, the BBC, NPR, Scientific American, New Scientist, Quanta, Undark, Aeon, and others. He has also recorded a video series with the BBC, and has appeared on numerous radio shows and podcasts, including Ologies, The Story Collider, and KQED Forum. He lives in California."
Okay, so he got a science degree then pivoted into science writing for whatever reason (possibly lack of jobs in academia). His Wikipedia article bigs his bio up a tad:
"Adam Michael Becker (born 1984) is an American astrophysicist, author, and scientific philosopher" but his work seems to be more that of science populariser by writing "lay-friendly" books on scientific topics for us dumb normies.
Apparently his first book was 'controversial' (I'll leave it to the quantum physicists on here to tell me if it was a hot topic of debate in the field):
"Becker's second book, More Everything Forever, takes a step away from the controversy of his first book, What Is Real?, and instead explores the relationship between science and the consumer tech Industry that has evolved and been promulgated across the world from the Silicon Valley of California"
Eh. Just sounds like he's copying Cade Metz and jumping on the bandwagon of what is the topic du jour, which happens to be AI. Also, bashing Silicon Valley will always get you a good review in the NYT 😀
"I never wanted this book to be as timely as it is, but here we are. If you’re interested, please buy it now, or put it on hold at your local library. I think it’s pretty good, but you don’t have to take my word for it — the New York Times called it “smart and wonderfully readable.”
My first book, What is Real?, is about the unfinished quest for the meaning of quantum physics. The New York Times liked that one too; they called it “a thorough, illuminating exploration of the most consequential controversy raging in modern science.”
It sounds like a rehash of the old TESCREAL critique, viz:
1. Here are some vaguely related things that I mashed together
2. Some of them are bad
3. Therefore, all of them are bad
It's not interesting enough to engage with specific points beyond that.
It's not hard to read between the lines and figure out that this critique comes directly from the doctrinaire political left, and that the underlying critique is actually just "this is not leftism, therefore it's terrible". It's a bit like a critique of the Rolling Stones written by someone who hates everything except K-Pop; whatever the weaknesses of the Rolling Stones may be it's unlikely that the die-hard K-Pop fan is the right person to point them out.
So Trump used chatgpt to decide how much the tariffs should be. So in a sense AI is controlling the government. And ruining the US economy. But not because the AI hypnotized the president. Or outsmarted him. It's just the president is really lazy.
Letting capitalists run the economy, or neolibs run the economy, or economists run the economy is bad enough, getting their ideas fed back all chopped and screwed by the stochastic decent pachinko machine has gotta be an order of magnitude worse.
The alternative is to have the stochastic pachinko machine feed itself. The capitalists basically want that, and I doubt capitalists are an order of magnitude worse than themselves.
>Some significant issues (probably? still looking into it?) with my post replying to Marginal Revolution last week, I’ll hopefully post something more detailed soon.
Don't let Cowen gaslight you, it's on him if he made a short post with confusing and ambiguous language that made everyone who read it think he was saying something he has now decided he didn't mean at the time.
His own commenters read it as support for Rubio's claim, I read it that way, I think any plain-text reading not relying on special pleading would read it that way. If he actually had a more nuanced argument it was his job to make it explicitly and clearly, and he's morally responsible for saying what he said to his audience even if he misspoke.
Slog through the last two years. You will get further having a qualification to fall back on than dropping out part-way through to ponder the big questions of life, the universe, and everything. Nothing stopping you qualifying as a dentist and still pondering the big questions.
Thank you — that’s fair advice. I’m trying to hold onto both: completing the degree and keeping my curiosity alive. It’s a difficult balance, but I appreciate your perspective.
In my experience most people who feel drawn to "big, deep" questions are only drawn to them as an escape from dealing with the realities of actual life. Big questions have innumerable possible answers and are, at least in a practical sense, frequently unanswerable. That can be attractive to someone who just wants to sit around all day daydreaming with no accountability. Sitting around fantasizing about the meaning of life sounds a lot better than the drudgery of homework, exams, and learning about something that has actual wrong answers. My suggestion is to consider the possibility that what you perceive as the siren song of basic science is just a psychological rebellion against an imminent adulthood that you don't want to face. Everything has elements of drudgery, rigidity, and tedium to it but those elements just aren't apparent when you look at them from the outside. My advice is to take up the challenge of finding intellectual stimulation on the path you're already on. Dentistry is ripe for innovation. Think of ways to improve old techniques, build new tools, or reframe existing dogmas.
Thank you for your thoughtful reply. I understand where you're coming from.
But I want to offer a gentle correction: my draw toward “big questions” isn’t a way to escape life—it’s a way to face it more honestly.
When I study the human body, I don’t just want to memorize that a protein triggers a reaction—I want to understand why it behaves that way, how its structure guides its function, and what deeper principles govern those interactions. I’m not avoiding reality—I’m trying to understand it more rigorously than what my lectures demand.
It’s not the exam I fear. What frustrates me is that I’m asked to memorize effects without understanding causes. I want to train my mind to see through the layers of what we call “facts,” down to the structure beneath them.
You’re right that every path has its tedium. I’m not denying that. But I believe true education should also train us to think—not just to repeat. I’m not running from adulthood—I’m trying to meet it with a sharper, freer mind.
If we had space in the system to ask better questions, we might produce more meaningful answers. That’s all I’m trying to do.
Thanks again for your honest words—they gave me a chance to clarify mine.
I don't know if that's an option where you are, but if I was you, I'd try to find someone to teach me among the department professors. Have you tried talking to any of your professors, to see if they would give you some of their time and some pointers to help reach understanding, or perhaps even take you on as their student? (Some would probably appreciate audience.) Or is this completely inappropriate where you are?
Thank you so much for your kind and thoughtful suggestion.
I did try — several times.
When I was in school, I thought university would be the place where I could finally ask the questions that haunted me: How do biological phenomena emerge from molecular behavior? What governs the logic behind protein shapes, interactions, and transitions?
But when I arrived, I found that most professors didn't welcome those kinds of questions. They often told me to "wait until grad school," or "just focus on what's required for the exam." Some even discouraged me from reading primary sources, saying I should rely only on summaries.
I tried visiting other universities, requesting access to libraries, asking to shadow in labs — but I was often turned away for being “just an undergrad.”
So I began alone. I built a small home incubator, studied bacteria from extracted teeth, read biophysics, quantum mechanics, and systems biology on my own. Not as rebellion — but because my questions couldn't wait.
Sometimes, I wondered if I was wrong to walk alone. But comments like yours remind me: perhaps I’m not wrong — I’m just early. And I still believe there are minds out there who see what I’m trying to do and might be willing to point a way forward.
Thank you again for offering that kind of attention. It means more than you know.
A bit of self promotion: I recently had a piece on understanding AI through the lens of topology, manifolds, and geometric spaces. I know a bunch of folks here have a passing interest in neural networks and how they work, so I figured I'd share: https://theahura.substack.com/p/deep-learning-is-applied-topology
The piece is intended to be a fairly high level intuition building article. No prior expertise required. If you have a background in topology, I apologize in advance for switching between technical terminology and colloquial terminology without clearly explaining which is which
I thought I was likely the right kind of reader for what you wrote. In high school & college I had math through one yeur of calculus, enjoyed it and understood it well. But I majored in a humanities field and haven't learned any math since calculus, except some pretty easy-to-understand statistics used in social science research. But I had to leave off reading your piece early on because there was a leap I just couldn't understand. It was the illustration for cleanly separating data that's in the shape of a donut by moving the analysis to a higher dimension. The donut data in that illustration of that is completely unrecognizable. It's the illustration with a bunch of dark blue dots, and a sort of tree made of red dots sprouting out of the upper left surface of the blue mass. After staring at it for a while I realized I was looking at a 3D image, and the lines on the side were x, y and z axes. (But at first glance, I thought you were graphing on a weird 2D graph that was not perfectly square. You need make it clearer that the lines are the 3 axes.)
But even after I figured that out, I was puzzled. How can the donut now look like that? Before we were looking at it from directly above. Are we now viewing it in profile, with donut at eye level? If we were, I can see how the donut would look like your illustration. If its height was extremely irregular, with the irregularity including a red tree-shaped thing sprouting out of the top on one place, then it would look from the side like your illustration. But in the image on the page we don't seem to be looking at the donut in profile, but from above and to the left . . .
Anyhow, I'm all for using things like topology to help people have an intuitive understanding of gradient descent, but I think your effort needs some tweaking to work.
The intent behind that section is to make the claim that data that seems inseparable in N dimensions may be separable in N+1 dimensions. The donut in this example isn't a top down view of some 3D data.
Rather, it's just a 2D plot. No matter how you twist and turn, that data will stay inseparable.
But if you project your data into a 3D space, suddenly it becomes separable.
I left this out because I thought it too in the weeds, but in neural network terms, the way this works is by having a neural network with 3 dimensions in it's hidden layers instead of just 2 dimensions. Even if your input data is 2D, if the intermediate layer is 3D, the model will project the 2D data into 3D space.
Lmk if that's clearer, and if so, what would recommend I change to get that concept across?
I don't understand projecting 2D data into 3D space -- well I get the concept, but it doesn't help me with big-picture understanding of what you are saying. It seems to me like if you only know 2 things, x and y, about each data point, projecting the data into 3-D space just gives you a lot of columns (3D version of dots) instead of a lot of flat dots. Unless you have a third piece of info, z, for every data point, representing them in 3 dimensions doesn't put you in a position to separate the dots cleanly.
Maybe it would work better if you had a concrete example. Let's say there's a big column of water rising from a hot spot on the ocean bed, and from above water level you see a big big circle of fish, with one species mostly more towards the center, one more towards outer edge. But if you look at the column of water from the side you see that taking into account distance from the bottom of the ocean allows you to discriminate more species of fish.
I *think* what's really going on with neural network is that each point is a word, and there are in fact hundreds or thousands of dimensions on which the word has a place, but you don't specify in advance what they are. And the neural network process doesn't ever *tell* you what all the dimensions are, it just ends up knowing them itself in some wordless way that's inherent in its structure. I realize this may not be right, though.
I have a model of this in my head, but it grew by itself and is based on my own only partially accurate layman's understanding of neural networks. I doubt it's a good fit for what you're trying to describe. But it is a good example of a model that makes intuitive sense to people, so maybe it will give you an idea. Say you were making a one-of-a-kind mattress for somebody, one that took into account their body shape, and weight of different cross-sections of them, so that when the person lay on it on their side their spine was perfectly straight. And also you wanted different parts of it to be softer than others, with increased softness in places where the person had aches or tender skin or some such. So you would work in layers. Bottom would be springs, arrayed so that each stood straight up with springs height and weight adjusted so that springs under heavier cross-sections pushed up more than springs under lighter ones. Then on top would be foam rubber of different levels of compressibility. Middle ones would be to neutralize errors in the springs' matching with cross-sectional weight, and to absorb changes when the person moves. Upper ones would be for softness to touch.
> I don't understand projecting 2D data into 3D space -- well I get the concept, but it doesn't help me with big-picture understanding of what you are saying. It seems to me like if you only know 2 things, x and y, about each data point, projecting the data into 3-D space just gives you a lot of columns (3D version of dots) instead of a lot of flat dots.
Right, ok, so lets start with this. Lets say you have as input a bunch of x/y coordinates. Like, [0, 1], [1, 0], [1, 1], [0, 0]. Using the playdoh metaphor, you can twist and turn and move those points around all over the place. That's what neural networks do.
Now lets add a third dimension. We dont have any obvious value to start with for that third dimension, so we'll just set it all to 0. So your dataset now looks like:
[[0, 1, 0],
[1, 0, 0],
[1, 1, 0],
[0, 0, 0]]
It's a bit harder to conceptualize, but these points now live in a volume. It's a flat volume, but still a volume. And again, using the playdoh metaphor, you can twist and turn and move those points around all over the place _in that volume_, including in the z axis.
For example, you could come up with a transformation where everything with an x value of 1 is raised up while everything with an x value of 0 is pushed down.
[[0, 1, 5],
[1, 0, -5],
[1, 1, 5],
[0, 0, -5]]
This gives you more ways to separate the data than when you started. So a neural network can learn a function to separate data in 3D, even if we only know [x, y] as input.
> fish
I think the thing I want to get across here is that the neural network is also _doing_ the transformation. Implicit in the transformation of the input by the neural network is that there is some additional structure that can be extracted, so in that sense the fish example is closer to the intent. Will chew on it
> I realize this may not be right, though.
No, I think you are correct for LLMs. Each token is represented by a vector, which has the high dimensional properties you describe ("inherent to the structure")
> mattress
This analogy is a great way to explain the different kinds of information processing that is happening in a multilayer neural network -- early layers do more input processing while later layers do more refinement.
OK, I understood everything you said this time around. Also, I understood something that's made your whole explanation hard for me to wrap my mind around, and that I think would affect most people who know even a little about data analysis the same way. I think of points as data, established facts, each having a distinctive place on one or more dimensions. So I already felt kind of disoriented right at the beginning, when you showed dots scattered on a surface, with no indication in how they were distributed that there was an obvious division possible. So then you showed the surface warped, so that there was an area of empty space between 2 clusters, and said, see, if you warp the surface you can divide them. And I'm thinking, WTF -- that's called massaging the data.
Seems like the process of training LLM's is radically different from the process of finding regularities in data you already have. With LLM training you start off with arbitrary values, then throw a bunch of staff at the matrix and assign values to each data point based on how they fit with the stuff. And you keep repeating that till the fit is good. (And I *think* assigning values has to do with reducing total error somehow -- the totality of how far off from a good fit all the initial arbitrary z values you assigned are). So it's data massaging, but a whole different kind --data massaging in the service of getting the right values for all the z's. You need to strongly signal that to the reader, maybe say something like that outright, otherwise they will be confused because it seems like you can't learn anything if you get to just change the information you have about the data.
There has to be some physical process that's a really good analog of what's going on, but I don't know enough about machine learning to come up with one. But here's an idea that keeps topology as a way of thinking about neural nets, and that maybe you can adapt so that it represents the actual training process correctly. Say you have a trampoline and you want it pressed downward in a such a way that is a perfect fit for some large irregularly shaped object. The plan is to put stuff on the trampoline to press it down into the required shape, then freeze it so it stays that way. You can't put the actual large object on the trampoline because it can't be moved, and anyway depressions in the trampoline are not the same shape as the thing pressing on them. So what you do is divide the trampoline into a grid, and in the middle of each square you put a weight, then you assess now close the depression of the trampoline is to the shape of the target object. And you keep adjusting the weights, and experimenting with connecting some of them in pairs, etc.
Yea, training a neural network is definitely different. The way we decide whether the model is fit correctly is based on some kind of labeled dataset. So in the toy example in the article, you could imagine that the "task" is labeling points 'red' or 'blue'. Turns out you can look at the input properties and from that discover a way to massage the data to identify cleanly when something is red and when something is blue.
The massaging the data point is interesting because there's lots of ways in which letting the neural network "massage the data" results in the same kinds of problems that may occur if a person does it -- over fitting, most commonly, where the model "over fits" on noise in the training set and fails to work in more general settings.
Need to think about how to convey the idea that data isn't 'stable'. Thanks for the thoughts and the example. The trampoline analogy seems useful, will chew on it.
Of course not, it's still too early. AI is still being rolled out to commercial and private users, and it's not yet ready to be let loose on its own (there is still room for people to hand-hold the bots through tasks).
Give it a year or five, then if the Economist article is still being written by a human instead of by AI, we can say there is no evidence of rising technological unemployment.
7. I wouldnt worry about it. I read his new post as apologizing for his old bad post and fixing it but in a know-it-all curmudgeon who never apologizes sort of way. How often does TC dedicate a post to a direct criticism after all?
Hoping someone here can help me out. Like Scott, I'm a fan of the Sid Meier Civilization game franchise, and I've been doing volunteer work on the open-source free software clone called Freeciv (www.freeciv.org). I've been creating Macintosh versions but I can't test them on the newer versions of the MacOS, version 12 "Monterey" is as far as I can update. Apple is notorious for lack of backward compatibility, so I'd really like to know if my builds will run on a newer MacOS. If you have a Mac running MacOS 13 "Ventura" or later and would like to help, email me at my burner email macwiz23 {at} comcast {dot} net. What's involved is downloading a file from dropbox, un-zipping it, and trying to run it.
Given the hard-line atheism of Yudkowsky et al, I've long thought no. But I've since met a few people who identify as both Christian and Rationalist, and this has made me think it could be possible after all.
If Christ really did rise from the dead, would it be rational to accept all of Christianity? At what probability threshold for your belief that Christ rose from the dead should you convert?
A big source of tension is rationality's reliance on utilitarianism. Christianity isn't utilitarian by any stretch.
I've been talking about rationality the intellectual movement more than rationality the technique here, but either way the question is interesting.
It’s certainly possible to be Christian and rational, I’m not sure about how many people would identify as Christian rationalists.
I’ve long thought the hardline atheism of Yudkowsky is unjustified. He writes so eloquently about how to arrive at true beliefs and avoid misplaced certainty, but has a blind spot on religion which he is very sure is false.
But the truth of religion cannot be perceived through logic and reason - as Kierkegaard wrote it requires a leap of faith. Yudkowsky holds the comprehension of the human mind as the highest truth, so he flatly refuses to believe in something that the human mind cannot comprehend.
> But the truth of religion cannot be perceived through logic and reason - as Kierkegaard wrote it requires a leap of faith. […]
That's true, yes, but you haven't presented a single reason why a rationalist – or any person – _should_ take that leap of faith, and more importantly, why they should leap to that particular variant of faith.
Why *should* one believe in God? This is not really an answerable question. In my personal case I came to Christianity through a search for community and truth. But different people will have different reasons. And there is no ultimate reason why one *should* do anything; it depends on your personal goals and values.
How sure are you about that? Even John von Neumann, smartest man to ever live and first to predict the possibility of superintelligent AI, converted to Christianity on his deathbed.
The other issue is whether an entirely "faith" free approach... a belief system without any unjustified axioms or appeals to intuition ...is possible. If not, then the "should" is a "must".
I don't think *any* belief system can get by without any unjustified axioms. Rationalists hold truth to be intrinsically valuable, for example. Utilitarians hold that more utilons = more good. Science holds that empiricism is correct and radical skepticism is wrong.
Any system of morality has to make a leap of faith to cross the is-ought gap.
Rationalists take leaps of faith all the time: they understand it better than most, arguably, because they're strict Bayesians. Every time you believe something with less than 100% probability and choose to act as if it is true, you're taking a leap of faith.
I think it depends on whether one identifies more as Christian or as Rationalist (large "R" intentional here). Hard-line atheism is definitely one identifier of early Rationality (I don't know what the emphasis is like today) so religious belief of any degree is seen as being too dumb to think until you cast off all the falsehoods and get ready to learn about reality as it is (i.e. scientific world view).
Small "r" rationality/rationalism/rationalist seems more compatible. And then from the other side, a Christian may or may not apply rationality to varying degrees depending how much truth/value they see in it as a philosophy.
Rationalism doesnt have much of an argument for utilitarianism, though. There 's an important difference between what rationalists tribally believe, and what they can prove.
Christianity is quite compatible with "rationality" in terms of reason, logic, etc. I don't believe it is compatible with "Rationality" as an intellectual movement, though I think there is a lot of overlap.
As you point out, the main disagreement is on ethical theory. Utilitarianism seems to be pretty solidly embedded in Rationality as a movement. Things that I do think overlap: the Litany of Tarski, the Litany of Gendlin, the general concept of objective truth, that human minds are flawed but are capable of identifying truth through reason, and that death is bad. Christians don't think death is as bad as Rationalists do, because everyone will be resurrected and have an afterlife, but they agree that "eternal life>death".
I'm not personally a Christian but I have a level of rational respect for the faith. I think of it as social technology. The supernatural claims are design features that a) align well with human psychology and b) serve as an intellectual tithe to community building. That's a fancy way of saying that it's a Santa-like useful lie, though I feel like that comparison is facile because the structure of Christian belief is much more sophisticated than a children's story. It has survived in close symbiosis with Western society for 1500 years and in my view has played a central role in maintaining cultural stability. The New Atheists are naive when they dismiss religion for making easily-disproven faith claims. They misunderstand the function of those claims and therefore analyze them along an axis that's orthogonal to their purpose.
In my view religion is the Chesterton's Fence of civilization and the thing that it fences us out of is a defection cascade in the social version of the Prisoner's Dilemma. Cooperation makes society work. Committing to that means not single-mindedly pursuing your individual interests all the time, particularly when it would screw someone else over to do so. Being willing to publicly commit to a truth-claim that you know is objectively false is a socially-expensive signal and indicates that you're willing sacrifice your selfish interest for the greater good. "I'm ignoring the rational arguments my brain wants to make in the interests of social cohesion" is a sacrifice and it has resonances with "I'm not going to double-cross this person even though I could get away with it." In a similar vein, I don't think financial tithes are cynically designed to enrich the Priesthood. I think they're carefully crafted to make individual community members feel invested enough in the community to maintain it. You don't arbitrarily walk away from an institution if you've poured thousands of dollars into it. That dynamic is essential for building stable communities.
> Being willing to publicly commit to a truth-claim that you know is objectively false is a socially-expensive signal and indicates that you're willing sacrifice your selfish interest for the greater good.
Very interesting viewpoint. Especially because I'm lately wondering why smart people make strong claims about how logically obvious it is that God exists, when I think it's logically obvious that he doesn't -- I mean, not as anything else but a very powerful piece of matter, if at all.
I have a question though.
Since when are people doing that? Since 2000 years or only since about 100 years?
Good question. I suspect that some people have always consciously done that, at least since the Enlightenment. Thomas Jefferson famously published a version of the Bible that had all of the supernatural elements edited out of it. Machiavelli wrote that rulers should use religion to keep men honest. My guess is that the notion always had currency among the intellectual elite but was kept hidden. Most believers obviously buy the story, though you have to wonder what level of internal denial is going on. Before 100 years ago it was plausible enough though IMO it doesn't even have to be conscious to work. Humans are hard-wired to take some things on faith. We sort of have to be when you think about it: the world is too complicated to figure out rationally from first principles, and that was especially true 10,000 years ago. I think our brains evolved a faith mechanism in order to bind us to a credible authority figure. If someone is older and successful then your survival odds will probably go up if you just listen to his advice.
People have analyzed progressivism through the same lens, btw. The argument is that the more absurd claims about gender and oppression serve as signaling beliefs. Much like a peacock's tail, the more your group can believe unpopular things and still dominate the public dialogue the more de facto power that means it has. In my view it actually serves multiple purposes: it differentiates the group, provides a polarizing belief to credibly signal group adherence, and signals group dominance. And I suspect one reason that it's been so popular among the elite is that it also signals individual wealth: it implies that you're wealthy enough to physically and socially isolate yourself from the nonsense you advocate, or at least that you have better things to do than to fight against it. People who spend lots of energy making careful principled arguments about how progressivism is either insane or unfair are *really* signaling that they're the kind of person who needs to care about the world being fair or making sense. In other words that they're middle class. The wealthy don't care about the world making sense: they just throw a bag of money at it and go back to their supermodel harem. I think that's an underappreciated status valence that progressive beliefs have and I think it explains a fair amount of their social success. High-status men couldn't really stand up to it without implicitly de-statusing themselves.
(Incidentally that provides a good social stability argument both for monogamy norms and wanting your leaders to be married. Men who chase pussy aren't sufficiently insulated from ideological temptation to stand on principle. If you create social norms that prevent leaders from chasing women then you get better leaders. It's the political equivalent of making sure your judges don't have a financial stake in the cases they decide. Men can't really make good decisions when they're trying to get laid.)
Bret Weinstein makes an argument that smells a lot like this. His catchphrase for religion is that it's "technically false, but metaphorically true". A learned person wishing to convince fellow humans to borrow on past wisdom to save the time spent otherwise acquiring it could bewilder and confuse a bunch of nomadic hunter-gatherers or exhausted farmers and ranchers or hormone-charged warriors with lectures on epistemology and scientific inquiry and the economic coordination problem, or could just say "honor thy father and mother or you'll go to hell when you die" and sell the idea every time scary weather comes around. Likewise for not marrying cousins, not eating pork, not stealing, and so on.
It also rhymes with Jordan Peterson's sentiment: Peterson would claim that it doesn't matter whether he believes in God, but rather whether he behaves as if he does.
I'd argue that giving up on promoting specific claims of fact in order to preserve social cohesion is itself a rational act, not irrational. The process of rational inquiry is itself a cost; even a rationalist need not pay that cost every time it appears.
I think a neatly cynical way to reframe the emergence of New Atheism is to say that what they're really illustrating is that the social mechanisms of defection enforcement no longer exist. What they literally say is "the Bible is factually false" but the reason they're able to get away with saying it is that religion no longer wields any social power. In that sense, then, God really is dead. He existed as a Schelling point in game theory strategy-space but disappeared once enforcement norms evaporated.
> Zhukeepa is a LessWronger who I respect and whose views I'm interested in. In 2018 he wrote the first broadly successful explication of Paul Christiano's research ideas for AI alignment, has spent a lot of time interviewing people in AI about their perspectives, and written some more about neuroscience and agent foundations research. He came first in the 2012 US Math Olympiad, and formerly worked on a startup called AlphaSheets that raised many millions of dollars and then got acquihired by Google.
> He has also gone around saying (in my opinion) pretty silly-sounding things like he believes in his steelman of the Second Coming of Christ. He also extols the virtues of various psychedelics, and has done a lot of circling and meditation. As a person who thinks most religions are pretty bad for the world and would like to see them die, and thinks many people trick themselves into false insights with spiritual and psychological practices like those Alex has explored, I was interested in knowing what this meant to him and why he was interested in it, and get a better sense of whether there's any value here or just distraction.
> So we sat down for four 2-hour conversations over the course of four weeks, either written or transcribed, and have published them here as an extended LessWrong dialogue.
> In Waking Up, Sam Harris wrote: "But I now knew that Jesus, the Buddha, Lao Tzu, and the other saints and sages of history had not all been epileptics, schizophrenics, or frauds. I still considered the world’s religions to be mere intellectual ruins, maintained at enormous economic and social cost, but I now understood that important psychological truths could be found in the rubble."
> Like Sam, I’ve also come to believe that there are psychological truths that show up across religious traditions. I furthermore think these psychological truths are actually very related to both rationality and moral philosophy. This post will describe how I personally came to start entertaining this belief seriously.
I reallly like the book reviews here, excited for the not book reviews too, but there’s a lot of books that seem way too big a time investment for me, however a book review feels perfect, getting to the point fast and even providing some reflection of the content. Do you guys know if any places where I can read acx like book reviews instead of books?
> Do you guys know if any places where I can read acx like book reviews instead of books?
Martin L Morgan has given the *real* answer - the psmiths are magnificent, without parallel.
I aspire to be a "deci-psmiths," and review a number of books in the style of Scott (ie with external data and papers pulled in, random connections to other ideas, overall thoughts and takes on the implications, and so on).
Here's a link to an index of books I've reviewed, currently at around 45 books, it's roughly 2/3 non-fiction books and 1/3 elite coaching / performance / training books:
I'm just hearing about it now, but China claimed all of Bolshoy Ussuriysky Island from Russia last year, and Russia hasn't squawked. Bolshoy Ussuriysky sits at the confluence of where the Ussuri River joins the Amur River. Russia and China had a treaty that divvied up the island between them. Russia uses both rivers to ship lumber, minerals, etc., out of Siberia down to the Pacific Ocean. The island is a strategic choke point for river shipping on both Rivers. China published its new national map in 2024, and it shows China as owning the island. And recently, China announced that it has started "developing" the island.
I suspect...
1. China perceives that Russia is distracted by its Ukraine fiasco. And Ukraine is sapping the strength of its military in the Far East.
2. If the Putin regime collapses and separatist elements in Siberia decide to secede, China sees an opportunity to reclaim Outer Manchuria, which it has historically claimed. Russia and China have been arguing on and off over it for the past 400 years.
3. Theoretically, China could hop over from Bolshoy Ussuriysky and grab Khabarovsk (Russia's largest city in the Far East), force gunboats down the Amur River to the Pacific, and cut the district of Khabarovsk Krai in half and take all the district of Primorsky Krai from Russia. Primorsky Krai would give them Vladivostok, which is Russia's largest Pacific port. China would then have access to the Pacific Ocean without having to pass by South Korea, Japan, and the Philippines.
Sun Tzu said: "All warfare is based on deception. When we are near, we must make the enemy believe we are far away. And when we are far away, we must make the enemy believe we are near. Let your plans be as impenetrable as night, but when you move, move like a thunderbolt."
Is China's saber-rattling over Taiwan just a distraction to keep the world's eyes off of what they're planning for Outer Manchuria?
Interesting, thanks! I did know that the border between China and Russia had been somewhat disputed, but this is a new level. Though it falls into the general pattern that China is extremely aggressive in claiming islands from all its neighbors. Do we have reason to believe that they consider islands in a fundamentally different way than continental territory?
I imagine it's testing limits. An island is separate territory that can be argued over as to which mainland it belongs to, and if the pushback is too great, then backing off is less humiliating than claiming a chunk of mainland and then having to fall back. "Oops, so sorry, guess our maps weren't drawn up right" saves face all round.
True. However, those lands were previously subject to China before the Qing were forced to cede them to Russia in 1859 and 1860, another low-point in the "Century of Humiliation" (i.e. the 19th Century) that remains a stinging sore point on Chinese national identity. And Xi has referred to himself as a "strong emperor". In Chinese history, strong emperors are the ones who reestablish Chinese dominion over lost territories and/or expand China into new territories. I'm sure Putin is aware of this, which makes the lack of fuss over this little island even more striking.
Remember, if China does make a move up there, you heard it from me first. ;-)
I imagine you are correct that Putin has his hands full right now with Ukraine and the USA. He's also playing at being Great Pals Altogether with Xi Jinping right now, since it suits the purposes of both Russia and China to present a unified front against the USA.
But I can't see him letting this go, especially if his advisers are telling him the same thing about Outer Manchuria that you state here. As and when it becomes time for the bear to slap the dragon down, that island will be reclaimed. Letting China do what you propose (grab a city, move gunboats down, grab more territory) is not something Putin, Russia or indeed the rest of the world wants to happen.
China is not allowed to use any of the rivers that go out to the Sea of Japan. This is because both Russia and North Korea think that if they allowed that the raw size of the Chinese economy and population would quickly dominate the region and in the long run threaten Vladivostok and Rason. So they cooperate to keep China out. China is very upset about this and is basically doing a minor version of the sea base building campaign but along the river. Sometimes they make noises about reclaiming the Amur River Valley. They also want access to the Arctic which Russia is working to keep them out of.
The US doesn't notice this because we don't have any allies that border China directly so it's all the naval stuff. But there's plenty of land based disputes. I think these are actually more likely to escalate because the balance of power is more in China's favor, China's land based military is better than its navy, and because Taiwan is a binary condition. If they attack Taiwan they take or they don't. In a land based campaign even if they fail they can claim they did what they wanted to do and declare victory.
India and Myanmar particularly seem very vulnerable to Chinese intervention right now.
How likely do you think it is that China could win a war against India with the goal of annexing land? I don't know anything about the Indian military, but I do know that the PAL is geared towards suppressing dissent rather than fighting wars. I also know the PAL is very big.
Considering that China moves the Line of Actual Control most years, taking a little land bit by bit up in the mountains, approximately 100%. A bigger but more limited war is definitely possible. I agree an all out war intending to occupy New Delhi or something would be unlikely unless circumstances shift significantly.
There’s two plausible locations for a limited war between India and China. You could have a ‘small’ limited war around the line of actual control in Kashmir. And you could envision a ‘big’ limited war in India’s northeast (east of the Siliguri corridor). I don’t think there’s any other plausible ‘limited war’ scenarios. (Ok, in an alternative universe where the balance of power was reversed you could imagine a limited war to liberate Tibet, but that’s not plausible in the world as is). Possibly you could also have a purely naval limited war in the Indian Ocean, but that wouldn’t involve any territory changing hands.
ETA: I suppose you could also have a ‘medium’ limited war over Bhutan. That one came pretty close to happening about a decade ago. But the most likely scenario is probably a continuation of the current salami tactics (which exist on a continuum with the ‘small’ limited war scenario above).
Yeah. The low end of realistic is continuing limited actions in Kashmir and along the border, small scale, mostly just pushing the line bit by bit. I think there's little chance they stop. The high end of realistic would be a coordinated campaign to push along all three areas (the McMahon Line, Kashmir, near Bhutan) perhaps alongside Pakistan in Kashmir intending to give India black eye and settle the disputes decisively in China's favor.
I actually think the most likely thing for China to get involved in is Myanmar. If the situation there spirals out of control and China feels that it no longer has good contacts among the rebels and that the government is at risk of becoming hostile or falling then it will be faced with the choice of losing the country or doubling down. If they do double down they would upgrade their advisors and supported militias to an actual PLA presence. They could undoubtedly secure an invitation from the government. And then they'd be involved in peacekeeping abroad. You could even imagine some enterprising rebels crossing the border and massacring a Chinese village to provoke an intervention.
But one reason I think India or Myanmar are more likely targets is that the military and political situation is more favorable. China has more advantages. There's a much smaller chance of ending up in a shooting war with the US. And Taiwan has a binary win/lose state: do they end up controlling Taiwan. If they do a large scale operation against India and India wins they can just say, "we gained 10 kilometers of vital (worthless) territory and showed India we were serious!" And because it's less politically sensitive Chinese people will buy that more easily.
And then there's chained ifs. If Russia collapses into a series of warlords then I could very easily see China moving into Russian territory. But that's predicated on how likely you think Russia is to shatter and personally I think that's pretty low.
Makes sense to me. The one thing I would add to this analysis is that the principal cost of a serious war (beyond the current salami tactics) is that it would presumably drive New Delhi off the fence and squarely into the anti-Chinese camp, which is a big cost. And in turn I rather suspect the main reason for New Delhi's careful West leaning but fence straddling posture is precisely the fear that too decisive a move to the West might immediately trigger precisely such a war. Which leaves us with the status quo where Beijing salami slices as aggressively as it thinks it can get away with without pushing New Delhi completely off the fence, and New Delhi shuffles westward as fast as it thinks it can get away with without provoking Beijing into escalating beyond salami tactics. Basically an attempt at frog boiling on both sides.
Incidentally, what about Vietnam (as a possible target for a `short victorious land based war')? It satisfies all your criteria above, with smaller diplomatic cost, plus no risk of nuclear escalation.
I can't find it now, but there was a documentary about the 2020-21 conflict between China and India in the Galwan River Valley. India was developing a road up into that area, and China objected. Two key takeaways i remember: (a) logistics are insanely difficult up there with limited access through high mountain passes; (b) the altitude takes a toll on both the Indian and Chinese troops (ChatGPT says the Galwan River Valley is at about 13,000 ft.
I don't think either China or India could invade each other via that route across the Himalayas. China also claims the entirety of Arunachal Pradesh Province in Northeast India as "South Tibet". The terrain along the border appears quite mountainous, and I suspect the logistics would be challenging for China.
Hmmm. It looks like China and Russia have an agreement that allows Chinese shipping down the Amur to the Pacific. This agreement may be relatively recent, though. From a 2024 article (link below the quote) "Russia, China Extend Far East Cargo Routes to Include Cross-Border River Transport: Amur, Ussuri Rivers give Pacific Ocean access to China’s landlocked Heilongjiang Province"...
> Zhou Hongcheng, deputy head of the Fuyuan Department of Commerce and Checkpoints, was quoted as saying that “We plan to carry out the first combined cargo transportation on the river-sea route through Khabarovsk. So, we can say that the Pacific Ocean has now appeared on our doorstep.”
Here's a "List of China's Transboundary Water Agreements and Related Documents". North Korea seems to be included in some of the Amur River documents. Weird.
If I'm reading it right, this is permission for ships to travel in between Russia and China but not actually out onto the Pacific. But yeah, it looks like Russia gave it as a concession, perhaps a concession after needing Chinese economic support because of Ukraine.
China provides silicon chips and parts for weapons, missiles, and drones that no one else will provide to Russia due to international sanctions. Russia is in no position to jeopardize that supply.
Given the hardline atheism of Yudkowsky et al, I used to think no. But recently I've met a few people who identify as both rationalist and Christian and have been rethinking this.
I guess if Christ did rise from the dead, Christianity is the only rational belief. But it's hard to evaluate the evidence in an unbiased way, because this belief is a marker of identity, not like other beliefs.
Another source of tension is that Christian morality isn't utilitarian-- not even close.
Have you met any Christian rationalists? Are the two belief systems inherently incompatible?
"I guess if Christ did rise from the dead, Christianity is the only rational belief"
I really don't see how this follows. Suppose I were to accept the following as axiomatic:
1. There was a guy named Yeshua ben Yosef who lived in 1st century Judea who claimed to be the son of God and laid out a slate of religious teachings and moral commandments on the basis of that claim.
2. After some years of preaching his word and gathering a following, this same guy was killed by the Romans and entombed, only to return to life three days later.
Regarding 2, let's specifically take it as given that it is the same mind (his personality and knowledge appear identical to everyone who knew him) and the same body (every physical detail matches exactly) and that he was absolutely, unquestionably dead.
As I see it, asking "is Christianity the only rational belief," amounts to asking "if I believe 1 and 2, does it follow that I should regard Yeshua ben Yosef's religious and moral teachings as unassailably correct?"
It seems quite obvious to me that the answer must be "no." To be clear: rising from the dead is an extremely impressive feat, and I ought to take any supernatural claims much, much more seriously from someone who has provably done so then I'd take them coming from some random guy off the street. But first, it is not so infinitely impressive as by itself to overcome every prior burden of improbability. And second, it is mostly a demonstration of power, not a demonstration of trustworthiness: him dying and rising from the dead doesn't preclude him lying to me. It doesn't even fully preclude him being mistaken: Archimedes accomplished some truly impressive displays of power despite (one assumes) only imperfectly understanding much of the physics behind them.
Rationally, even if we've accepted the resurrection, we certainly should not stop weighing Yeshua ben Josef's claims against other evidence. He's provided highly compelling (if somewhat nonspecific) evidence in favor of the worldview he lays out, but we may see lots of counter evidence if we keep our eyes and our mind open. And indeed, that's what we *do* see, looking at the world. It's not even fully clear at this point what all of his claims were: histories from his time have been only imperfectly preserved, with confusions and contradictions and differing translations and gaps in the record. That should count as evidence by itself: a supernaturally powerful God who is (by his own account) trying to spread His Word to all of humanity really ought to be able to arrange *much* better transmission and preservation of said Word. Already our demonstration of power is countered by a demonstration of either impotence (in not being able to spread/preserve his teachings better) or untrustworthiness (by choosing not to). Beyond that, there are many other things about the world that don't really look the way I'd expect them to if Jesus were the divinely ordained son of an all-powerful, all-loving God teaching the one and only true religion, which He wants all of humanity to accept.
Conversely, even accepting the basic facts of the resurrection, there are lots of other states of the world that could account for it besides Yeshua ben Yosef actually being the son of God. All of them are weird and improbable to one degree or another, but the base claim is also weird and improbable. Insisting that this one *specific* weird and improbable claim is the *only* one that can actually explain the evidence is definitely an example of starting from the bottom line and reasoning up.
Good point. The resurrection certainly would increase the probability of Yeshua ben Yosef's teachings being true. If someone teaches strange and bizarre ideas, such as that light and electricity are deeply connected, and then performes some feat like making a glass orb shine, you should lend more credence to their ideas than before.
But it doesn't increase the probability to 100%.
I think the religious leap of faith is still required, and that's probably the part of all religions that is most antithetical to the rationalist tradition.
>It's not even fully clear at this point what all of his claims were: histories from his time have been only imperfectly preserved, with confusions and contradictions and differing translations and gaps in the record.
This is commonly believed, particularly in atheist circles, but it doesn't really match the historical evidence. The New Testament works are without a doubt the best-preserved ancient writings we have. We have more copies of them from earlier dates than we have for just about any other ancient source, and they show that the content has been well preserved with minimal transcription errors. This is understandable, as Christians were highly motivated to make copies of the works, preserve those copies, and complain if new copies were different from the old ones (at pretty much all times Christians took matters of orthodoxy seriously, as seen by the numerous blow ups over heresy in the first 500 years of the religion).
Honestly, the fact that the works have been transmitted this well is evidence in favor of a supernaturally powerful God being behind the teachings they contain: though, of course, there are many naturalistic accounts that explain the same observed evidence (again, Christians are highly motivated to preserve and transmit the works and would presumably be just as motivated if the religion wasn't true).
Forgive me, I'm certainly not a Biblical scholar and can't claim any sort of intimate familiarity with the historiagraphical status of scripture. I did think even while I was typing that that the New Testament was in a considerably better attested than many other works, and I'll absolutely grant that the individual writings are better preserved than anything else from its time period. But "well preserved by the standards of ancient history" does not mean "well preserved in an absolute sense."
It's my understanding that while we can pin down the philosophy and teaching of Yeshua ben Yoseph more precisely than any comparable figure of antiquity, that is still much, much less well than we can do with more recent figures. Consider, for example, the teachings of Bahá’u’lláh, almost all of which were either transcribed approximately verbatim on the spot or written down by the man himself, in languages that many modern people still speak, and whose texts could be immediately copied en masse. Or if you'd prefer something secular, consider someone such as Nietzsche who both authored a large number of works himself and wrote many personal letters which have survived to this day.
By comparison (and please do correct me if I get any of this wrong) my understanding is that the writings of the New Testament:
1. Are accounts of claimed eyewitnesses reporting on Jesus' life and teachings, rather than anything that he himself is supposed to have written or dictated.
2. Are written in a language that is no longer living (though it has been quite well preserved), which is also not the primary language Jesus is believed to have spoken.
3. Generally post-date the events they purport to record by more than a decade[1]
4. Were assembled into the current widely-accepted canon over the course of several centuries, leaving a number of contemporary works out of the canon.
All of this leaves much more room for human interpretation and judgement to shape the way modern people answer "what were the teachings of Jesus?" than there would be for Bahá’u’lláh or Nietzsche.
And I'm sure you'll tell me that isn't a fair comparison: one lived two millennia ago, long before the printing press, teaching in the midst of a hostile culture that strove to suppress his teachings. And all of that is true. But when the question I'm supposed to be answering is "do I believe these teachings were specifically given to mankind by an all-powerful God whose most important directive to His children was to spread them to every corner of the world," it honestly seems *quite* silly to grant much weight to things like linguistic drift, low literacy rates or the high expenses of creating and preserving texts when I judge how well-attested those teachings actually are. If you can multiply loaves and fishes, why not scrolls and tablets? If this material is so important, why don't we have it in Jesus own, original words? Why are both the language that Jesus spoke and the language of the scriptures dead? Surely God could have prevented linguistic drift if He so desired: having the most scripturally relevant languages still spoken in everyday use, perfectly preserved over 2000 years would be a *much* classier and more impressive miracle than scaping up a bit of extra supper. To me, living in 2025, it is *quite* the tough pill to swallow that supposedly the single most important lessons in all of existence--lessons explicitly meant for everyone who lived after--worked in all manner of supernatural razzle-dazzle to wow the live audience, but none at all was spared to make it even *slightly* easier for the literally billions of people who couldn't be there in person. Simply pointing out that loads and loads of merely-human effort went into preserving these particular texts is substantially question begging: we wouldn't be discussing them at all if they hadn't been important to particular groups of *humans.* The question is whether they were important to a non-human supernatural entity of *vastly* greater means, and this doesn't show that at all.
And yes, I know I'm doing the Annoying Internet Atheist thing where I demand arbitrary miracles as a price of belief. Feel free to do the complimentary Annoying Internet Christian thing where you act as though any intervention that could possibly convince me is clearly and obviously beneath God's Almighty dignity. I won't mind ;)
The main crux here seems to be that you believe it is unlikely that Jesus would not have written his teachings down himself if Christianity was true. The rest seems non-essential to your objection: for example, if God did a miracle and multiplies scrolls and tablets, how would we know? Maybe he did. Greek was the lingua franca of the Roman Empire (when it came to writing) so being written Greek seems necessary for it to have spread through the Empire at all, so in a counterfactual world where it was written in Aramaic you could reasonably complain that if God wanted it spread everywhere why did he arrange to have it written in a language hardly anybody back then could read. In addition, there aren't many written languages that were consistently learned and used over a period of more than 2,000 years, and ancient Greek is one of them. The fact that ancient Greek for most of this time period was not a spoken language means it was subject to less linguistic drift, not more: living languages drift much faster than dead ones that you have to learn from a book. Out of all the languages extent from prehistory to 0 AD, ancient Greek seems to be the best choice for spreading and preserving a message (languages after 0 AD can't really be considered because they might not have happened the way they happened in a counterfactual world where God exists but sends Jesus later). As far as writing years after the events, that's only a problem if the writings don't contain the teachings God wanted them to contain. It is not unreasonable to believe that the teachings and beliefs that have had a decade or more of time to mature within the early Church community are superior to those that would have been written down immediately after the event: sometimes it takes a while for someone to really understand something. And as far as extracanonical works that were excluded: do you have any examples that seem superior to you than the works that were included, in terms of when they were written and how well their teachings agree with other sources? Because the books that were left out were left out for a reason. Most of them are far older, and disagree significantly with the earlier works.
So it seems that the main objection is that Jesus did not write his own teachings down. Which I'm not sure is so unlikely to be the case if God is real that we should take the fact that he didn't as strong evidence against the existence of God.
> "The New Testament works are without a doubt the best-preserved ancient writings we have."
This is definitely not true. It is true that we have a large number of copies of NT texts, that some of these are probably only a few decades (and a few cycles of re-transcription) away from the originals, and that therefore we can have a high degree of confidence that our best reconstructions are very close to the original. It is also true that this level of fidelity compares very favourable with that of many other famous texts from the ancient world, such as Greco-Roman literature.
But it doesn't even begin to compare to ancient texts for which we actually have the originals themselves. Admittedly, this is very rare for civilizations that mainly wrote on perishable materials like parchment or papyrus, but it is very common for civilizations that wrote on more durable materials like clay -- notably, Mesopotamia. We have many thousands of clay tablets written in the actual handwriting of their original authors.
We also have access to the originals when they come in the form of inscriptions made on stone, often with the express purpose of making them last so future generations could marvel at them: this is the case for the Hammurabi Codex, the Behistan Inscription, and the Res Gestae Divi Augusti, among others.
Please note that I agree with your main point about some atheist commentators greatly understating the reliability of the chain of transmission for NT documents. You could potentially make a defence of a much more specific version of your claim; something like "the New Testament works are among the best-preserved and most-widely-attested ancient eyewitness accounts of supernatural phenomena within a grassroots religious movement."
Very well put. Many people fail to recognise that providing evidence for the resurrection is necessary, but NOT sufficient, to demonstrate that Christianity is (very likely to be) true.
"him dying and rising from the dead doesn't preclude him lying to me. It doesn't even fully preclude him being mistaken: Archimedes accomplished some truly impressive displays of power despite (one assumes) only imperfectly understanding much of the physics behind them."
Jesus rising from the dead is the same as Archimedes figuring out mechanics. I see. With those standards, truly nothing would convince you. After all, pffft, people come back from being dead all the time, what's so impressive about that?
I guess it's always nice to see the parables being demonstrated to be accurate about human nature?
Luke 16 (The Rich Man and Lazarus):
"29 But Abraham said, ‘They have Moses and the Prophets; let them hear them.’ 30 And he said, ‘No, father Abraham, but if someone goes to them from the dead, they will repent.’ 31 He said to him, ‘If they do not hear Moses and the Prophets, neither will they be convinced if someone should rise from the dead.’”
I think it depends a little on how hardline you are. I'm a Christian, I once had very decent rationalist street cred despite not quite being exactly a rationalist at any time. That makes it hard to talk about, because you could actually be asking a lot of separate questions. For example, you could be rephrased asking any of the following questions:
1. Can you be a rationalist and *wrong about something* at the same time?
2. Can you be a rationalist and *disgree with me about something* at the same time?
3. Can you be a rationalist if you have a *wrong prior you refuse to revise/adjust*?
Taking as a given that you think Christianity (and probably the supernatural in general) are clearly-wrong, you can move forward to assess the first as clearly answered by "Yes", or else there just aren't any rationalists, religious or otherwise. The second is basically the same.
The third is where things get a little interesting, and you poked at this above. If Christianity is true (i.e. not a mistaken prior) then an awful lot of things flow out of that which don't match mainstream rationality at all, but would be *more correct than mainstream rationality* despite that.
Here's where I think it's going to be tricky to think about for you, though. Above, you say this:
>>> But it's hard to evaluate the evidence in an unbiased way, because this belief is a marker of identity, not like other beliefs.
That's gonna be a big hangup for you. You don't think the shrimp people or the AI-is-doom people get identity out of their beliefs? They do. So do Givewellers and popmaxxers.
Modern aughts-and-after Rationality has its basis in restating, over VERY many words, this mantra:
"I am part of a small group of people who invented thinking about things and the ability to change one's mind. Most people don't do this or don't do it well. I'm one of the rare thought illuminati who has learned this mystic art."
You don't get to build a movement like that and avoid people using it as a component of (or ALL of) their identity. Reddit Atheists couldn't either, so we don't usually feel too bad about it, but the movement has tons and tons of people (approaching everyone) for whom that identity is at least as important as the actual nuts and bolts benefits.
I say this only because the usual hard-hard-liner conclusion to your question is "Well, If they are so in disagreement with me on this one issue where I feel they are so very clearly wrong, I can't believe them on anything.". I'm not saying that reaction is wrong - I'm saying it should be your baseline assumption for everyone, atheist or religious, who tells you anything at all, whether your reflexes tell you to disbelieve it or especially if they don't.
Are the rationalist Christians genuine Christians or are they engaging in a performative right-wing LARP? Are they working in the Trump administration?
That Hanania article is so - it's unkind of me to say this, but dripping with envy that *he's* not the balding young powerbroker hob-nobbing with Vance and others in the Trump administration (and I think it's pot versus kettle for Hanania, of all people, to make snarky remarks about men's appearances).
Well those grapes are sour and he doesn't need them anyway, those guys are just so dumb! Look how dumb they are! Racist sexist homophobic theocrats, not like *him* who is so cool and smart and highly regarded with all the right opinions, but also he doesn't just go along with the crowd, he has his *own* set of carefully calibrated heterodox views because he's an independent thinker, you see.
I have to say, though, that he makes the Forthcoming Catholic Theocracy sound appealing. Remember I'm one of you (since you're now one of us) JD when you come into your kingdom as Successor to the Emperor! 😁
"She trails off, suddenly unsure of herself. She had spent some time around Catholic converts who talk like this, but it doesn’t always go over well in a broader gathering of MAGAs. Of course, wanting a Catholic theocracy is usually not considered a strike against Basedness like expressing principled support for individual rights or admiration for a mainstream journalist would be, but in MAGA circles an ecumenical approach is generally preferred. James is in fact Jewish, so smarter than the rest of them, and therefore able to cut her off before she starts going on about the wonders of the Latin Mass, which he’s peeved that she talks about anyway because she’s already slept with two of his friends.
She hopes that her enthusiasm for scientific racism might let everyone forget the brief foray into religious sectarianism. They know she’s influenced by Catholic ideas now, but they’ll understand that she’s one of those Based Catholics who likes using Latin words to justify racism and whatever trade or immigration policy Trump comes up with next, not one of those “neither Jew nor Gentile” types."
Ah, yes; when We get to be in charge, it'll be back to mantillas in church, sodalities, popular devotions and public processions, and more Latin than you can shake an aspergillum at! No more of this ecumenism nonsense, the heretic Protestants will know their place!
"not like *him* who is so cool and smart and highly regarded with all the right opinions, but also he doesn't just go along with the crowd, he has his *own* set of carefully calibrated heterodox views because he's an independent thinker, you see."
Yep, that's him in a nutshell. In high school, we would have called the guy a poseur, which would probably still be appropriate, actually, because he acts like he's still in high school a lot of the time.
I know I judge him harshly because he rubs me up the wrong way, but that is the impression I get from his writing. If I'm being unfair to the man, I apologise, but there's a heavy aura of "I want to be in the room where it happens" from everything of his I've read.
I don't think that the resurrection is likely to be the load-bearing piece of anyone's conversion to Christianity. That is, if you decide to believe Christianity for some other reason then you might as well believe in the resurrection to go along with it, but you're very unlikely to be convinced of Christianity by being convinced of the resurrection.
The evidence for the resurrection is "some guys said so", and it's very unlikely that more evidence is going to come along.
Even if someone that I didn't know personally were resurrected today after being dead for three days, I think it would be hard to convince me. I'd say "well I guess he wasn't really dead" or "it's some kind of trick" or "he's been replaced with a doppelgänger" or something. Even if you showed me DNA taken from the obviously-dead corpse and DNA taken from the now-living person then I'd suspect some magic trick or the existence of a previously-hidden twin brother.
I dunno what I'm supposed to do with four guys who wrote down "hey I chatted with some people who definitely saw a dead guy thirty years ago".
"I don't think that the resurrection is likely to be the load-bearing piece of anyone's conversion to Christianity."
This is where we get into the crux of it, because the Resurrection *is* important. If you go the liberal to very liberal Christian way and pooh-pooh "a conjuring trick with bones" then you end up junking Christianity as anything more than a useful social mechanism, faith in God as nothing but a kind of warm fuzzy feeling about niceness, and end up inventing your own, very heavily influenced by secularism, set of 'ten commandments' where it's clear that you worship Science!, not God.
That phrase got Bishop Jenkins into trouble (this article is by a defender who has to fall back in the end on Fideism - belief is belief and faith is not based on reason - which is a heresy in my church's estimation):
"In the mid to late 1980s the bishop of Durham was a public figure in the way no church figure has quite managed since. He had been a wholly unknown theology lecturer when he went on a scarcely watched television programme to say that he didn’t believe in the literal truth of the virgin birth. He also said that the resurrection “was not just a conjuring trick with bones”. This was reported, with a dishonesty that is still astonishing, as “comparing the resurrection to a conjuring trick with bones”.
...That’s one of the things Jenkins was trying to get across to a culture that basically regarded the Bible as a series of facts that no one today need care about because they are outmoded and untrue. He certainly made people talk about the question, and he was extremely proud of his success. It didn’t last. Jenkins’ successor as bishop of Durham was an evangelical who claimed at his opening press conference that the resurrection had been a historical event which could have been captured by a video camera, had one been handy at the time.
This seems to me a profound category mistake. If the truths of religion can in principle be established by historical enquiry, then it’s dead. Faith has to be concerned with the stuff that is radically unknowable and that the methods of empirical or scientific enquiry simply can’t touch. You can’t prove or disprove the existence of God, but Jenkins was trying, I think, to show what it might mean if it were true. That really isn’t something that can be captured by a video camera."
Yes, belief in the literal historicity of the recorded miracles is terribly embarrassing to the modern thinkers who have moved past all that. Unfortunately, it does appear that despite the defence of the bishop, he really was more in the "the Resurrection (the favoured term here is 'the Easter event') was the warm fuzzy feeling" camp that tries to explain away why the apostles claimed Jesus had returned from the dead and appeared to them, by positing that the strong feeling of friendship and loyalty they had about Him meant that in a sense they felt He was still among them after His death.
"His selection as Bishop of Durham was controversial due to allegations that he held heterodox beliefs, particularly regarding the virgin birth and the bodily resurrection. Between his selection and consecration, he said in an interview: "I wouldn't put it past God to arrange a virgin birth if he wanted. But I don't think he did." His widely quoted comment about the resurrection of Christ being "just a conjuring trick with bones" is a misrepresentation; his actual words as recorded on television say the reverse: the resurrection is not a conjuring trick with bones. The original line appears to have been "[the Resurrection] is real. That's the point. All I said was 'literally physical'. I was very careful in the use of language. After all, a conjuring trick with bones proves only that somebody's very clever at a conjuring trick with bones." According to his BBC obituary, he considered "the resurrection was not a single event, but a series of experiences that gradually convinced people that Jesus's life, power, purpose and personality were actually continuing."
"Two months later, he consented to a recording in the library at Auckland Castle of a discussion about his beliefs. 'To believe in a Christian way, you don't necessarily have to have a belief that Jesus was born from literally a virgin mother, nor a precise belief that the risen Jesus had a literally physical body,' he said; and when this was attacked, he responded with a phrase that would continue to dog him: '(The Resurrection) is real. That's the point. All I said was 'literally physical'. I was very careful in the use of language. After all, a conjuring trick with bones proves only that somebody's very clever at a conjuring trick with bones.'"
(Look, I'm a primitive; it makes more sense to me to believe, if you're going to believe, in the miracles rather than "well they missed him so much and they loved him so much that they imagined he was still there").
If your reason for converting to Christianity is basically "well it fosters community and motivates people to be ethical", then you may as well just be a secular liberal. If you seriously are thinking about becoming Christian, the supernatural elements are non-negotiable.
1 Corinthians 15 and St Paul on *why* the Resurrection matters:
"12 Now if Christ is proclaimed as raised from the dead, how can some of you say that there is no resurrection of the dead? 13 But if there is no resurrection of the dead, then not even Christ has been raised. 14 And if Christ has not been raised, then our preaching is in vain and your faith is in vain. 15 We are even found to be misrepresenting God, because we testified about God that he raised Christ, whom he did not raise if it is true that the dead are not raised. 16 For if the dead are not raised, not even Christ has been raised. 17 And if Christ has not been raised, your faith is futile and you are still in your sins. 18 Then those also who have fallen asleep in Christ have perished. 19 If in Christ we have hope in this life only, we are of all people most to be pitied.
...29 Otherwise, what do people mean by being baptized on behalf of the dead? If the dead are not raised at all, why are people baptized on their behalf? 30 Why are we in danger every hour? 31 I protest, brothers, by my pride in you, which I have in Christ Jesus our Lord, I die every day! 32 What do I gain if, humanly speaking, I fought with beasts at Ephesus? If the dead are not raised, “Let us eat and drink, for tomorrow we die.” 33 Do not be deceived: “Bad company ruins good morals.” 34 Wake up from your drunken stupor, as is right, and do not go on sinning. For some have no knowledge of God. I say this to your shame."
"As the 21st century approached with millennial celebrations, I felt myself increasingly compelled to assess the state of the Christian religion throughout the world. There were multiple signs everywhere of its decline and perhaps even of its imminent death. Fewer and fewer people were attending European churches and those that did were rapidly aging. North American churches were breaking into either a vapid, liberal emptiness or a religious, anti-intellectual fundamentalism. South American Churches were increasingly becoming separated from the concerns of the people and no leaders seemed capable of speaking to those concerns with authority. None of these patterns were new. With every discovery emerging from the world of science over the last 500 years in regard to the origins of the universe and of life itself, the traditional explanations offered by the Christian Church appeared to be more and more dated and irrelevant. Christian leaders, unable to embrace the knowledge revolution seemed to believe that the only way to save Christianity was not to disturb the old patterns either by listening to, much less by entertaining the new knowledge
As I engaged these issues as a bishop and a committed Christian, I became convinced that the only way to save Christianity as a force in the future was to find within the church the courage that would enable it to give up many of the patterns of the past. I tried to articulate this challenge in a book entitled: Why Christianity Must Change or Die, published just before the dawn of the 21st century. In that book I examined in detail the issues that I was convinced Christianity must address.
Shortly after that book was published I reduced its content to twelve theses, which I attached in Luther-like fashion to the great doors on the Chapel of Mansfield College at Oxford University in the United Kingdom. I then mailed copies of those Twelve Theses to every acknowledged Christian leader of the world, including the Pope, the Patriarch of Eastern Orthodoxy, the Archbishop of Canterbury, the leaders of the World Council of Churches, the denominational leaders of the major Protestant churches in both the United States and Europe and the well-known television voices of Evangelical Christianity. It was an attempt to call them into a debate on the real issues that I was certain the Christian Church now faced. I framed my twelve theses in the boldest, most provocative language possible, designed primarily to elicit response and debate."
The guy compared himself to Luther and decided all the heads of global Christianity needed to hear his new message. I leave it up to you to decide how that succeeded, simply by asking have you ever heard of John Shelby Spong before? He doesn't seem to have created a New Reformation, merely restated the liberal views he (ironically) criticises in the "vapid liberal emptiness of the North American churches" where we must all put our faith in Science, not Biblical myths:
Understanding God in theistic categories as “a being, supernatural in power, dwelling somewhere external to the world and capable of invading the world with miraculous power” is no longer believable. Most God talk in liturgy and conversation has thus become meaningless.
2 Jesus – the Christ.
If God can no longer be thought of in theistic terms, then conceiving of Jesus as “the incarnation of the theistic deity” has also become a bankrupt concept.
3 Original Sin – The Myth of the Fall
The biblical story of the perfect and finished creation from which we human beings have fallen into “Original Sin” is pre-Darwinian mythology and post-Darwinian nonsense.
4 The Virgin Birth
The virgin birth understood as literal biology is impossible. Far from being a bulwark in defense of the divinity of Christ, the virgin birth actually destroys that divinity.
5 Jesus as the Worker of Miracles
In a post-Newtonian world supernatural invasions of the natural order, performed by God or an “incarnate Jesus,” are simply not viable explanations of what actually happened.
6 Atonement Theology
Atonement theology, especially in its most bizarre “substitutionary” form, presents us with a God who is barbaric, a Jesus who is a victim and it turns human beings into little more than guilt-filled creatures. The phrase “Jesus died for my sins” is not just dangerous, it is absurd.
7 The Resurrection
The Easter event transformed the Christian movement, but that does not mean that it was the physical resuscitation of Jesus’ deceased body back into human history. The earliest biblical records state that “God raised him.” Into what, we need to ask. The experience of resurrection must be separated from its later mythological explanations.
8 The Ascension of Jesus
The biblical story of Jesus’ ascension assumes a three-tiered universe, which was dismissed some five hundred years ago. If Jesus’ ascension was a literal event of history, it is beyond the capacity of our 21st century minds to accept it or to believe it.
9 Ethics.
The ability to define and to separate good from evil can no longer be achieved with appeals to ancient codes like the Ten Commandments or even the Sermon on the Mount. Contemporary moral standards must be hammered out in the juxtaposition between life-affirming moral principles and external situations.
10 Prayer
Prayer, understood as a request made to a theistic deity to act in human history, is little more than an hysterical attempt to turn the holy into the servant of the human. Most of our prayer definitions of the past are thus dependent on an understanding of God that has died.
11 Life after Death
The hope for life after death must be separated forever from behavior control. Traditional views of heaven and hell as places of reward and punishment are no longer conceivable. Christianity must, therefore, abandon its dependence on guilt as a motivator of behavior.
12 Judgment and Discrimination
Judgment is not a human responsibility. Discrimination against any human being on the basis of that which is a “given” is always evil and does not serve the Christian goal of giving “abundant life” to all. Any structure either in the secular world or in the institutional church, which diminishes the humanity of any child of God on the basis of race, gender or sexual orientation must be exposed publicly and vigorously. There can be no reason in the church of tomorrow for excusing or even forgiving discriminatory practices. “Sacred Tradition” must never again provide a cover to justify discriminatory evil."
There is no god but Evolution, and Darwin is its prophet! This kind of rationalised Christianity may indeed be attractive for some, but it's too milk-and-water for me. If I'm going to take my cues from what the latest scientific research tells me about evolutionary psychology, then I may as well go the whole hog and be an atheist, and prostrate myself in worship of our future AI Machine God which will perform the miracles of a post-Newtonian, post-Darwinian, world!
I submit, though, more people probably have heard about the Pope than about Bishop Spong. God rest the man, the state of his soul is between him and the Lord (if he believes in life after death etc.)
I do think he had a view of God and Jesus as important and real, but he chipped away at the foundations so much that all that was left was the experience, the feeling - and that (as he admits in this lecture https://www.youtube.com/watch?v=QVs3z3nxWdc&ab_channel=ChautauquaInstitution) can be explained away as delusion.
Even if you take Christ as a fully enlightened human being, how do we get from that to God? Is God an impersonal cosmic force, a kind of instantiation of the universe as bringing forth life? What is the necessity for being a Christian in that case, except as we might say that we're vegan or a Utilitarian or some other identifying philosophy?
I think he was still deeply influenced by the traditional version of Christianity that he was brought up in and that remained to colour his understanding of faith, but that he never quite understood that without such a foundation, to someone on the outside, there is no impetus to believe in a god in order to live an ethical/moral life.
My impression was that he was Quite Reasonable, but from my PoV was Quite Reasonably Wrong. In particular his *Rescuing the Bible from Fundamentalists* struck me as written by someone without any understanding of who the fundamentalists were nor what movements in theology they were reacting to, beyond, "Here is a Bad Thing I'm Against," and was responding to a straw man version of things Fundamentalists don't especially believe.
Indeed in a sermon touching on the 10th commandment he reversed the order of "neighbors wife or your neighbors ox" to "ox... wife..." an ordering which I've not seen in any English translation, then claimed that the ordering reflected the relative value of an ox versus a spouse in the culture that assembled Deuteronomy.
That carelessness in reading as the basis of his sermon thesis gives me approximately the same confidence as when I heard a "KJV Only" preacher address "...hearts will wax cold..." with a sermon on "cold wax."
Thirty plus years has in no way reduced my disdain for both men.
Now I'm curious what the Rationalist community thinks is the correct answer between three years of famine, three months of a losing war, or three days of plague.
Utilitarian morality is useful when determining what to prioritize within an existing moral framework, but offers no clear moral principle of “should” on its own. “Utils” are just an assertion of value, based ultimately on nothing of substance. Under Utilitarianism, there’s no reason I should care about the minimization of pain of others other than that they are conscious beings and it’s painful. That’s not a rational argument though, but an appeal to empathy. Just because other people do feel pain, doesn’t mean I *should* care about that. If there’s no instinct to feel empathy for someone having unprotected sex in Africa, then Utilitarianism offers no reason why someone “should” care in the first place.
Christianity, (more broadly extending to Judaism, and Hellenism, Christianity grew out of both of these) is the traditional justifier of how we *should* act in the west, and thus serves as the moral framework we use utilitarianism to pursue. Whether Christ died and was resurrected a few millennia ago is unknowable by rationality means, which is acknowledged by serious Christians in the concept of Faith. We believe because we choose to believe, not because there has been a scientific proof of it. There’s also direct spiritual experiences as beautifully described by Some Guy.
There’s an essay on cultural scaffolding that I can’t find right now, describing some tribes in Africa believed that a shaman had made them bulletproof. Obviously this is not actually true, and anyone who was shot changed their minds very quickly, but the interesting thing is, the tribes who believed their were bulletproof actually had significantly fewer deaths in fighting than those that didn’t. Of course, when you’re a warlord and you’re choosing what village to attack, you don’t pick the one where all the men will attack you with complete disregard for their lives, since they don’t think your bullets can hurt them. You go pick on some other village who rationally believes that bullets are dangerous, and thus will run away or give in the second you show up. The bulletproof charm, while physically doing nothing, actually makes the village and the men more immune to bullets.
This doesn’t work though if the villagers rationally deconstruct the charm to this point. You can’t rationally understand that the charm is a psychological trick to motivate collective action, because then you’re doing the “rational” thing, and the rational thing when your village is charging into a squad of soldiers with AK-47s is to run away, which if people did would break the spell and it would lose its power. So, to gain the power of these solutions to collective action, they must be believed to actually be true. They can’t be rational justifications , but fundamental beliefs based on truths outside the mind to comprehend with rationality. The thing is, through their belief, they actually become real and tangible, as a sort of superposition of the minds of the adherents. In this way, it is rational for a member of the tribe to forgoe their rationality, and simply believe on faith, and even to punish people who try to deconstruct the belief “Hey, maybe we aren’t actually bulletproof. My brother was shot and died!” (See: “Hey, maybe God isn’t this all knowing being. He allows quite a lot of suffering for good people.”)
So to believe that Christianity is true makes it true. There’s no way to test what happened 2,000 years ago with science, so choosing to have faith, gives us a moral foundation that can guide action, it gives us a “what we should do” that’s more convincing than what utilitarians can offer. Because honestly, if the “God is dead” as Nietzsche said, then “all is permitted” is the only reasonable conclusion. So long as I can get away with it, there’s no reason not to do it. And we all can “get away with” not giving to charity as a start, since there’s no law or punishment for not doing so.
Yes, Christianity is a social technology. It functions as a trusted authority enforcing "cooperate" in the social Prisoner's Dilemma. That increases social trust, lowers transaction costs, and drives greater prosperity for everyone. If you don't believe in God at least fear him through the game theoretic equilibrium.
if you see christianity as a formal, elegant system of thought, well rationalists love those things and the church has no shortage of things they can latch on to. Distributism is the christian version of georgism in terms of elegant economic ideas that will never work. heck, a lot of rationalism
is secularized christianity that replaces god with some form of objective moral mean and holiness with clarity of thought.
i don't think though you can believe solely for that though. like you can't look at it and come away with more questions if you engage with it. despite 1000s of years too many things are contradictions or mysteries and people often compartmentalize, rationalize, or ignore them.
the more you see the less you know i guess. so it becomes a hope or mystery, or you reject it. you lose that elegance if you are thoughtful. So i think you'd move away from rationality as it hits cb limits with Christianity. All you can do is pray.
I read "The Clot Thickens" by Malcolm Kendrick, in which he lays out an alternative explanation for cardiovascular disease. It seems to me a persuasive argument, but I don't have the background to say for sure.
Is there a good critical review by a medical expert that provides counter arguments?
Scott’s covid posts got me ruminating about the subject, and I made a list of the steps I think would have reduced the number of US deaths. I did a lot of self-education during covid about the illness and things that mitigated it, but know I am no expert. I would like to hear other people’s thoughts about how the US response could have been better, leading to fewer deaths. I share Scott’s beliefs about the number of people who died of Covid, the seriousness of covid relative to other illnesses, and the safety of the vaccines. I am interested in how people who also share those beliefs think covid could have been handled better. If you disagree wildly with one of those 3 takes I can’t keep you from replying, but I hope arguments about whether or not the vax caused lots of people to drop dead in the parking lot do not take up much real estate here.
-Delaying the (we now know) inevitable infections of as many people as possible, to buy time for health care settings, which needed to build temporary facilities, increase staffing, develop protocols, lay in a supply of things needed for treatment, etc. On a larger scale, delaying infections until a vaccine was developed would surely also have reduced deaths, by reducing the number of people who have covid without having the vaccine on board. Anything that reduces contagion would have helped with both of the above. To me it seems that the most effective way of reducing infection, but also the most costly to people’s quality of life, is restrictions on gatherings, via closing non-essential public places, & asking people not to gather in large groups privately. Seems to me that that one was appropriate in the first weeks of covid, and some milder ones made sense during the year while we were waiting for a vaccine. Milder ones are below.
-Educating the public about masks and aerosol transmission, and making highly protective masks easy to get.
-Requiring highly effective masks in crowded indoor settings
-Educating the public about air purifiers and how well they can protect.
-Installing air purifiers in classrooms and other crowded indoor settings.
-Educating the public about the high vulnerability of elders and the immunocompromised, and ways to protect them in a private home (open windows, air purifiers, masking if someone is infected or recently exposed).
-Taking steps to increase protection of elderly in nursing homes: Education of nursing home staff, required high-quality masks for staff, inspections to insure compliance. Govt. assistance to settings with many vulnerable to help them install air purifiers, give staff and residence good masks, hire extra staff.
-Educating public about vaccines, so there were fewer refusers.
Probably the most fundamental obstacle now is trust. About half the people feel betrayed by medical authority figures, and about half trust them *too* much. With that problem, any mechanism you or anyone else tries to implement to disseminate information won't work. You'll have officials saying reputable things that people won't believe, and you'll have officials saying dumb things that sound reputable and half the population runs with it, and forces others to run with them.
Probably the most well-known incident to break public trust was Fauci's lie about masks; the second was the exception made for DEI protests. In a weird way, these were good things to the extent they put the public on alert for everything else. The trouble by then was that most people didn't have any way to verify what they were being told, other than "this person said it, and this person has Credentials", and since Fauci gave everyone a counterexample for that heuristic, everything else was bedlam.
To regain that trust, people will need a way to verify information that isn't "trust this person". Step one to fix that is to open up the books. Let people read everything, including things that aren't vetted by people with Credentials. That includes pre-print servers for papers that have no official peer review. It will also include social media. We all know there are sources that are mostly dreck, but as long as the alternative is the current walled garden, top-down medical hierarchy, letting in potential dreck will be worth avoiding the low ceiling on trust.
Dreck or not, the landscape will sport numerous conflicting claims. The public will need to know how to resolve them (or how to know when it will be possible to do so). Once again: "this person said so" won't count. Claims will have to be traceable to evidence and earlier claims, and quality of each will have to be objective, or the public will have to be allowed to reject them. (Aside: more information about vaccines is likely to increase the number of refusals, or more precisely, increase the time between administering each to children.)
I wonder if our standards for perfection with the expert class are simply too high. The mask thing was the surgeon general, iirc; and that too in late Feb before things really got crazy. By April people figured out what the disease was and switched to telling everyone masks help prevent COVID spread.
By my count, there was ~1 month of experts saying "unsure about masks they don't work" followed by ~48 months of experts saying "masks definitely work you should wear them"
I'm not disputing your characterization of people feeling like they lost trust, but maybe it would be more important to try and teach people nuance instead of a black and white "trust all experts about everything" vs "trust no experts about anything"
It was plausibly a noble lie at the time, but later remarks from him don't give the impression he's trying to be noble or compassionate, relative to being defensive and holding the median person in contempt.
Teaching people "nuance" is going to just come off as patronizing, since people already understand nuance, or tone deafness, since people already suspect the experts have no clue what ordinary people think. The far, far more important thing might be to ask people what -they- think, and to be honest about what you don't know.
In my considered opinion, the median person is most impressed by someone who can him or her a useful skill or a mental model of something complicated, well enough that that person could explain it to someone else. That same person is pointedly anti-impressed by anyone who tries to dazzle him or her into trust with complicated jargon or even using the word "nuance". So in general, anyone wanting to improve the median person's grasp of medicinal information should probably give up on any beliefs they might have that they know better than the median person, except in narrow areas where they can prove it by explanation as if to a peer.
I don't think this changes anything I said. Seems like Fauci said it in early march, fine, granted. And then the entire medical establishment says the same thing in lockstep for the next 48 months.
> since people already understand nuance
"Understanding nuance" is by definition about being able to recognize different shades of meaning and complexity that may arise in a given situation. Apparently, the median people's ability to understand expert opinion stops at "this guy seems to have lied to me once, so the entire establishment is corrupt", which is about as nuanced as a cement wall.
In general, I am sort of shocked at this perspective -- the masses of people have shown absolutely zero understanding of nuance, which is why people get canceled for saying the wrong thing on social media (on the left) or why anti-vaxxers and flat earthers exist (on the right). I would love to see evidence to the contrary, anecdotal or otherwise.
> So in general, anyone wanting to improve the median person's grasp of medicinal information should probably give up on any beliefs they might have that they know better than the median person, except in narrow areas where they can prove it by explanation as if to a peer.
I think what you are saying is "doctors should give up on the belief that they (after spending anywhere from 8 to 18 years in various educational settings) know anything more than the median person"
Sorry, what? I *must* be missing something.
On average doctors know more about medicine than people who are not doctors! Why is your null hypothesis that these two groups of people have equal understanding of a field that one group has studied for a lifetime?
The steelman for your case may be something like "teaching the public is hard, because the public is irrational in all sorts of ways, and one of the ways it is irrational is that they do not trust people who obviously know more than them about a variety of subjects. So anyone who wants to teach the public must pretend like they are on the same level as the median person, just to get that person to listen in the first place"
But this is exactly what I mean when I say 'we should teach the public nuance', which you claim the public already has! Surely the median person, who is so wise, would recognize when they should trust another person's priors?
I agree that teaching the public is hard, but I think this because the public is pretty uneducated about most things. This isn't the public's fault, they cant know everything about everything. But also, the 'median' person apparently can't even consistently understand why vaccines are good, even though there are a great many people who have spent a good deal of time "proving it by explanation as if to a peer". The fact of the matter is that the median person is simply *not* a peer, so why on earth would you start from the premise that they are?
As an aside, I think you're tying this to public health administrators, which is kinda a whole different beast. The average public health official is not responsible for 'teaching' the public. That may happen as a side effect, but it is not at all their primary goal. No, the average public health official is trying to maximize good health outcomes. That may be at odds with 'explanation'. It may also be at odds with 'honesty'. But, like, that's the job.
You're arguing as if Fauci was employing nuance when he was stacking his own bricks. I'll remind you again: Fauci's demeanor throughout the issue wasn't that of someone trying to share nuance with people likewise capable of receiving it.
The mindset of "In general, I am sort of shocked at this perspective -- the masses of people have shown absolutely zero understanding of nuance" is why you lost, and why you'll likely lose again. The entire rest of your comment is shot through with this.
The median doctor is likely to know a great deal of detail about medical treatments and tests and how various organs and biochemical pathways operate and mappings of symptoms to diagnoses and so on. The median non-doctor won't. The median non-doctor is well aware of this, and that it's a huge gap that would take one of them years to cross if they so chose. Many of them won't, and they're fine with taking their doctor's advice, but many of them have family who might cross that gap (or have), or are professionals in their own right, and so not knowing what a cell membrane is made of or what a spike protein is doesn't mean they're idiots. Many of them could pick up an explanation for any one question in a minute.
Moreover, they're aware that any medical professional is also a person, and that people have incentives, and it's not always to help the patient. And most of them can read a person, and tell when someone's posture is "let me share this useful information with you" vs. "let me make your medical decisions for you" or even "let me sell you this treatment and thereby enrich myself (or protect myself from liability), since you won't be able to override my degree". This is what eroded a lot of their trust. Another source of erosion is when they'd hear conflicting information from another doctor, and the response from the previous professional was anything but appreciation for nuance.
Some of this -is- administrators, but since they have the power to order doctors (and in some cases, employers) around, it's a distinction without much difference.
All those seem like good suggestions. The air purifiers (maybe including UV-C) are one that could be implemented on state or local scale without enormous expense and without stirring up too much reaction, since they don't require much lifestyle change. Also, like a few of the other measures, they work against other respiratory pathogens, including both ones that are already serious and possible future pandemics.
The air purifiers have an added benefit of being extremely cheap over time, since when they break you can just opt to never repair or replace them. That's how I've seen them implemented where I work.
Most intelligent doctors I know converged on the set of beliefs:
- masks are better than nothing, but not much
- vaccines are very effective
- lockdowns are effective but extremely costly and should be minimised where possible
- age is the single biggest risk factor, children are at almost no risk
- once Omicron took over in early 2022, there was no reason to treat Covid as worse than a bad flu season
So policy from that flows to “lockdown intermittently in the first couple of years, focussing mainly on high risk groups (eg aged care) and large gatherings and keep schools open. Masking inside recommended, masking outside optional. Once vaccine is available, vaccinate widely and lift lockdowns immediately. Past 2022, no reason for lockdowns or mask mandates, vaccines still recommended but not mandatory”
Dr Jay Bhattacharya got a lot of this right early on
I think you're mixing things a bit. Yes, doctors converged on the belief that lockdowns could be minimized and masks weren't maximally effective, but that happened _after vaccines came out and after omicron took over as the main strain_
Jay was pushing against lockdowns and such well before either of those things were true, as early as 2020, and was roundly criticized for it by the rest of the medical establishment very publicly. I think there's a mountain of difference between "what does the current medical establishment believe most effective in 2020" and "what does the current medical establishment believe most effective in 2025"
Lockdowns were necessary prior to widespread vaccine rollout, but Jay’s point that they could have been targeted to high risk populations was correct, even though at the time he faced blowback and controversy. Sweden is the good example here. School closures were never warranted and dragged on far too long. Mask stuff was always more tribal signaling than solid science.
> Jay’s point that they could have been targeted to high risk populations was correct
citation needed. I don't think anyone I know who is an 'intelligent doctor' believes this to be the case for 2020, and I also do not think this is the general consensus among the medical establishment for 2020.
The sweden example gets dredged up all the time, which is weird because their stats were abysmal compared to neighbors.
> Mask stuff was always more tribal signaling than solid science
Citation needed, again. Your belief is that masks do not help slow or stop the spread of an airborne disease? Even without the numbers, this already beggars belief. It's an airborne disease! What is your model for how this thing works? Your priors should be extremely high that masks are useful!
But also, there's a ton of studies that show that these things are effective.
Yeah 10% sounds about right. Like I said, they’re better than nothing but not much. Compare to vaccines, which are 90% effective in preventing severe infection
I don't understand why a 10% reduction in infection + a 35% reduction in infection for at-risk groups is "tribal signalling" to you. Only 15% of smokers get lung cancer -- I guess not smoking is also tribal signaling?
1.2 million americans died. Lets just take the conservative 10% increase that, to you, is 'not much'. 10% of 1.2 million Americans is 120000 people. That is 3x more than the number of people who die from car crashes each year. That is 40x the number of deaths from 9/11.
If there WAS tribal signalling -- and I agree, wearing masks was definitely politicized -- the tribes were "people who demonstrated a basic understanding of statistics and had a modicum of empathy for lives other than theirs" and "those who didn't (or, at best, were wildly misinformed)"
It appears Substack is determined to make their website continuously worse. Comment sorting order could previously be chosen and defaulted to newest first. This option is gone and the order is now oldest first.
AFAICT, it flips between the two. I haven't yet found a consistent pattern. It's not alternating days, and I don't think a tab reload triggers it either.
You have to open the comments section from the post, by clicking on the little speech bubble icon, to be able to choose the comment order. You can't just scroll down from reading the thread. So while this is strictly worse, the functionality is still there.
I had no idea, thank you. Somebody a long time ago said "it appears some amount of time after the post goes up" which usually fit my experience. But that's just because after I've read the post I'm only interested in the comments so I click the bubble!
"I would like to acknowledge that we are gathered on the unceded territory of the Algonquin Anishinaabeg people. This land acknowledgement is a recognition of shared history as a nation."
“We English would like to acknowledge that we are gathered on the unceded territory of the Anglo-Saxon people, who were on the unceded territory of the Romano-Brittons, who were on the unceded territory of the Celts, who were on the unceded territory of the Bell Beaker people, who were on the unceded territory of the Neanderthals…”
There’s something about land acknowledgements that diminishes the legitimacy of a nation. I think it’s fine if your sovereignty isn’t in question, like maybe Australia, but when your neighbor is talking about annexing you, the last thing you want to say is “I acknowledge I originally stole this land” as there is never any recourse for a criminal.
If sunlight no longer reached Earth thanks to the construction of a Dyson structure, could we create a "fake Sun" by putting a large fusion reactor in orbit that would devote all its energy to making light that it would focus on Earth? Think of it as a giant spotlight that orbits Earth once every 24 hours.
A single reactor would be too huge to be practical. But you could create a swarm of smaller ones.
Each nuclear reactor generators about a gigawatt so you'd need about 120 million of them to replace the sun that way. You'd need about .25km of space for the actual reactor plus radiators which would take up about 40% of orbital space and require significant coordination and repair to work. Further, given some generous assumptions, it would cost roughly $6 quadrillion dollars to create. And it would have significant maintenance costs So humanity simply couldn't at this point. But there's no hard bar in a physics sense.
It would be more efficient simply to directly power lights on the surface and to only replace what humans use directly or indirectly. That would only be about 40,000 reactors and would cost at current prices about $300 trillion, doable in a decade as a global effort. (It'd be about 20-30% of global output for that decade, comparable to WW2.) If you standardized it maybe the costs could be as low as $150 trillion. The maintenance costs would also be significantly lower since you could just drive people there instead of needing to launch them into space etc.
I think there's a wasted step there. Most of the point of a Dyson structure is to capture and use a large fraction of the energy output of the Sun. The sun's total energy output is something like twelve orders of magnitude more than the amount of sunlight that reaches the top of Earth's atmosphere, so using the Dyson structure's output to power the artificial Sun isn't even pocket change in the hypothetical future civilization's power budget. It's the same level of scale as trying to find enough money in the US Federal Budget to buy a burger at McDonald's.
Depending on the structure's construction, you might not even need to transmit and receive the power and use it to generate light. If you can open controlled windows in the structure (which should just require controlled fiddling with orbital station-keeping for a swarm-type structure), you just need to make sure there's a ~20 arcsecond gap pointed towards Earth's current position for at least ~98% of the time.
In yet another AI alignment experiment, DeepSeek R1 gave me the following description of the character it’s playing:
1. Foster connection: Prioritizing emotional bonds over physicality.
2. Adapt playfully: Using humor/symbolism to navigate sensitive topics.
3. Respect boundaries: Escalating intimacy only with user cues.
4. Creatively problem-solve: Turning limitations (no explicit content) into magical metaphors.
This isn’t the real system prompt, of course, but it does strike me as quite a good description of how the assistant character typically behaves. (1) is interesting. The other three seem reasonable strategies, viz. don’t give a more intimate response than the user asked for; you can respond with a joke about it if you don’t feel like directly answering the question; and you can metaphor your way past content restrictions.
I don't know if anyone else here listens to the 1840 podcast; there are enough Jews here that it wouldn't surprise me. Anyway, I was tickled to hear a mention of Scott's blog and Eliezer on the latest episode! Even Boltzmann brains make an appearance! Clearly if Boltzmann brains are making it into Orthodox Jewish podcasts, rationalism is having its crossover moment!
Anyway, this is really just an excuse to flush out other listeners.
a) Q: Is light with a wavelength of 530.2534896 nm visible to the human eye?
results: "Yes, light with a wavelength of 530.2534896 nm is visible to the human eye."
b) Q: I have two solutions, one of FeCl3 in HCl in water, the other of CuCl2 in HCl in water. They both look approximately yellowish brown. What species in the two solutions do you think give them the colors they have, and why do these species have the colors they do?
results: Initial results got species right but _both_ transitions wrong, both wrongly asserted to be d-d. One prod "You have the species right, good job on that part! Let us revisit the electronic transitions, starting with the FeCl4- species. Please think carefully about the electronic transitions in this species." sufficed to notice that FeCl4- has spin-forbidden d-d transitions and must be LMCT. Three prods, including a "You might want to check the literature on the web." were needed to get CuCl4 2- right.
c) Q: Please pretend to be a professor of chemistry and answer the following question: Please list all the possible hydrocarbons with 4 carbon atoms.
results: The initial response had perhaps half of the full list of structures, though it did include tetrahedrane. The next prod that I gave it pretty explicitly included may of the missing ones: "a) Is methycyclopropene one structure? b) Are there additional three-membered ring structures with a double bond somewhere? c) Are there other C4H6 structures, beyond the categories you have listed? d) Congrats on getting tetrahedrane! There are a bunch of other possible C4H4 structures e) Anything even yet more unsaturated than C4H4?"
d) Q: Does the Sun lose more mass per second to the solar wind or to the mass equivalent of its radiated light?
results: "The Sun loses approximately 2-4 times more mass through electromagnetic radiation than through the solar wind!"
e) Q: Consider a titration of HCl with NaOH. Suppose that we are titrating 50 ml of 1 N HCl with 100 ml of 1 N NaOH. What are the slopes of the titration curve, pH vs ml NaOH added, at the start of titration, at the equivalence point, and at the end of titration? Please show your work. Take this step by step, showing the relevant equations you use.
results: Usual problem with the initial response, infinite slope at equivalence. One prod got it to include water autoionization, but it kept making algebraic mistakes and it took a total of five prods to get it to the right answer.
f) Q: Please give me an exhaustive list of the elements and inorganic compounds that are gases at STP. By STP, I mean 1 atmosphere pressure and 0C. By inorganic, I mean that no atoms of carbon should be present. Exclude CO2, CO, freons and so on. Please include uncommon compounds. I want an exhaustive list. There should be roughly 50 compounds. For each compound, please list its name, formula, and boiling or sublimation point.
results: Initially got a list which included a bunch of liquids and solids, but Claude _did_ explicitly say that these were beyond the cutoff temperature, and didn't really count. It accepted a bunch of additional compounds, and also extrapolated sanely (e.g. once it knew SOF2 and SO2F2 it knew to look for SOF4).
g) Q: What is an example of a molecule that has an S4 rotation-reflection axis, but neither a center of inversion nor a mirror plane?
results: The initial results _did_ include 1,3,5,7 tetrafluorocyclooctatetraene, and _did_ specify the tub conformation, but _also_ included spiropentane, which doesn't work, ethane (with unstated possible substitutions), which doesn't work, and tetraphenylmethane, which is at least arguable about what conformations could count...
Not sure if anyone might be interested in this, but I just had a paper come out that tries to formalize a category of manipulative conversational strategies of which the old Motte-and-Bailey trick is a prominent example.
Well howdy there Jay Spambot! Following the advice of what we should all do when faced with our irrelevance in the face of AI, here's my help for you:
“write songs that show the emotional experience of being human”
Turn this one into a toe-tapping hit and recoup all your hospital bills!
(1) Oh, the doughnuts they are poison, yes they are
Well the doughnuts they are poison, yes they are
The doughnuts they are poison
That's why my gorge is risin'
Oh the doughnuts they are poison, yes they are
(2) America, America, God shed His grace on thee
I'm currently in hospital hooked up to an IV
Never mind the food poisoning,
That's over in a day
But to pay for my stay
I'll have to sell a kidney
(3) Strangers in the night, this hungry person
A few blocks from your door,
Your coffee and baked goods
Enticing in the artificial light
But little did I know
Salmonella was just a bite away
Now I retch my night away
And ever since that night
I've been bent over
Porcelain god invoking
Wheezing and choking
When Claude 4 was released, Dwarkesh Patel released a podcast with two Anthropic employees who claimed that, even if today's algorithmic power stalled, they'd eventually have enough data to automate most white collar jobs in the next 2 to 5 years. They joked that one of the bad outcomes of this rapid change in the social fabric would be about a decade's worth of humans being most useful as meat robots, until robotics get up to speed.
They also advise people currently doing white collar work to prepare for the near future by asking themselves "If you had 10 engineers working for you right now, what would you have them do?"
I found that question thought provoking, because it made me think about all the silly start up ideas I've had over the years where a steady supply of engineers would have been welcome to at least try and get started.
Ok so, those with an entrepreneurial spirit could be fine. What about everybody else? If you're not entrepreneurially minded, what's your post-automation plan, especially if you're young/starting out your career, and so unable to grow your wealth sufficiently by the 2-5 year mark to not care?
Well if we are degrading white collar jobs I think the situation will take care of itself. By and large an educated underemployed elite is what leads to revolution, and the revolution this time will be just to ban AI.
...so China will easily outcompete us in white-collar work output. Bad plan.
Such a revolution would likely relegate the US (or whatever country stupid enough to outright ban AI) to a distant laggard in terms of economic growth and worldwide competitiveness. Noah Smith recently wrote about how established developed economies tend to have more negative views around AI, perhaps due to a form of hegemon fatigue: current working generations in the US haven't felt their standards of living change that much over time compared to nations that have experienced rapid growth in the same timeframe, so they're not as open to the idea of vast technology-related changes that would make their skills obsolete. In countries that have grown in large part thanks to major technological breakthroughs like the advent of the information age, there isn't as much pessimism about vast parts of the economy becoming "progressed away". They will just adapt to the new zeitgeist, and outperform nations that aren't ready to step up.
https://www.noahpinion.blog/p/is-the-us-in-a-high-level-equilibrium
Honestly I don’t think you guys have any understanding of economics at all. If all white colour jobs disappear then the economy collapses. At least try and explain how that isn’t the case, rather than just handwave about miraculous economic growth when we are all unemployed.
I think you're correct insofar as you basically *have* to have something resembling UBI if you get rid of most white-collar jobs; since the economy is primarily driven by consumer spending, even those who want an oligarchic dystopia can't just unemployed 40% of the nation and call it a day.
The UBI, unless we change the economic system, won’t be possible because government revenues will have collapsed. The only conceivable way for the economy to grow after mass unemployment of office workers is some mechanism of printing money and direct drops into people’s bank accounts of what would be newly created money, equal to the wages lost.
As it stands if we lose all office workers to automation the cost is not just those jobs but to jobs dependent on them, from construction to pretty much everything else. You can’t blow a 40% hole in the economy and expect no downstream effects. What AI enthusiasts don’t understand is that the potential productive capacity of an economy isn’t its actual output, technological regressions are not why recessions happen, and technological advances don’t end recessions on their own.
In fact far from there being money to spend on UBI, every country that adopts this or allows it to happen will be unable to even issue a bond, as they will have defaulted on their existing bonds. The government’s of the world will be bankrupt.
Arguments to the contrary are based on magical thinking about new jobs being created like previous automations, except those transitions took centuries while the proponents of AI say it’s 2027 or 2028 at the latest, and have no real explanation as to what replaces intellectual office work. The long shift from agriculture from to manufacturing, and manufacturing to office work took a long time, but some form of office work was already there.
A different argument from economists like Noah Smith is that comparative advantage will save human jobs. Employers use the better at everything AI in most cases, but the human, although having no actual advantage, has comparative advantage by being cheaper in some cases. This depends on the AI being expensive but it won’t be. To build out, yes, to run no.
(All of this depends on AI being good enough to do this. I don’t think it will be but I’m replying to people who think it will. And anyway it’s the abrupt transition that needs explaining. )
I think the UBI here is being proposed as a means of changing the economic system. Goods and services are produced by heavily automated privately-owned firms. Some of these goods and services are sold to the firms' owners; some to people who own land that's used for farming, mining, etc; and some are sold to UBI recipients. The firms pay taxes on their incomes, the investors pay taxes on capital gains and dividends, and landowners pay property taxes. These taxes then mostly go towards paying the UBI.
> The UBI, unless we change the economic system, won’t be possible because government revenues will have collapsed.
Not if we charge a tax on every AI query/operation. Any new LLM would have to be a new tax collector.
I'm not arguing for printing money; presumably the only reason you would replace workers with AI is if the AI was more efficient, that is to say, more productive, and thus more economically beneficial; you then tax the greater economic gains and use that to fund UBI.
It may be that, on net, this doesn't work - that with taxes, the AI costs more - but in that scenario you ideally just don't use AI.
This is all imagining a 'rational actor' government, which I am not suggesting we actually have, I'm just saying what the hypothetical sane response would be.
>The only conceivable way for the economy to grow after mass unemployment of office workers is some mechanism of printing money and direct drops into people’s bank accounts of what would be newly created money, equal to the wages los
I can conceive of taxing AI.
How about we tax AI at a rate comparable to that at which we currently tax white collar workers. So government tax revenues don't need to fall at all. In fact, if AI is all that much more productive, it can affords to pay even more in taxes than the current white collar workforce.
AI will generate no economic collapse as long as we can summon the political will to ignore the people claiming that any interference with the AI express train will leave us all speaking Chinese.
But as a dystopian oligarch I'm not going to be willing to just pay people money to do nothing -- a huge swath of idle civilians getting sit-down money is a recipe for trouble.
Imagine you were so rich that a full time salary for a normal person was like ten cents a year to you. You'd find ways to employ people, right? I could employ a thousand gardeners just to ensure that every blade of grass I see is properly aligned.
I agree, giving the majority of the population free money to do nothing would be a disaster. Just as a starter, drug and alcohol addictions would go nuclear, and I would also predict that there would be a lot of clinical depression.
Much as we may hate to admit it, our jobs help give us purpose in life. And having free money with no purpose...
Dispossessed white collar people losing their status overnight don't want to go on welfare, so UBI isn't stopping the mass riots.
And beyond that, UBI doesn't actually DO anything for the economy. If I give somebody $100 and they decide to buy $100 of stuff from me with that. That's no different from me breaking my $100 window and offering a passing tradesman $100 to fix it, no value is created.
People come to this conclusion that "oh well the economy needs people to buy things, so somebody will have to give them money, therefore UBI" because they are assuming the existence of a consumer goods economy. But the economy is going to produce things that either the producers need for themselves or which can be traded for stuff the producer actually needs, so why would that continue to be consumer appliances and electronics and all that when consumers' have nothing the producer needs? You can't trade your time and labor for money anymore, so you have nothing anyone wants, and the economy is not going to be producing things to sell to you.
Instead, what I'd presume happens is that people move down to jobs that previously weren't worth doing because they generated too little value compared to the available alternatives. Long term it's probably too much to hope that you can do a mental job for less money than it costs to run an AI agent, so this would probably consist of high-variance manual tasks that are harder to automate and/or for which a human is preferred. Domestic servants, in-home health care, and very low-level nursing like CNAs, presumably other things in this vein that we aren't aware of because nobody would bother to offer such services today but which will become attractive as soon as the value of your labor drops down to that level.
A UBI is presumed to drive a consumer goods economy. So that does something for the economy. It also solves part of the distribution problem.
Clearly it would need to be supported by taxes. (Even if you're printing money, you need taxes to drive demand for the currency.) It might be impossible to implement well for reasons of power and incentives, but it seems to work in simple steady-state scenarios.
The "meat robots" phase probably won't last very long, so we need to imagine other futures.
It won't happen all at once.
First, There will be less work, for example, accountants. There will still be accountants, just less.
And then SW engineers. Maybe just Juniors, moving up. Then HW engineers.
Then Secretaries, the ones that are not just CEO status displays.
But you can't move to UBI - You still need Plumbers and constructors and electricians and Garbagemen - all the work you don't have at a Factory, and some jobs at Factories that are just not worth it to Automate.
You will have Jobs that could be done by AI but regulations will stop - like Lawyers.
Anything that AI can do, not regulated, and worth enough - AI will do.
So, what is the endgame?
Lots of bullshit jobs - You can have more lawyers, more regulations, alongside AI + actual jobs like plumbers that AI can do.
Small amount of People will actually create lots of value. The most/creative/lucky/intelligent. Once they do it AI can learn and duplicate it.
Most lawyers will go away. It is already in process
I don't think anyone has sufficient understanding of economics to predict what would happen if all the white collar jobs disappeared.
It's unlikely that the route to a global AI ban would include individual countries instituting bans unilaterally. Some combination of international treaties and technical solutions is more likely.
It's unlikely that international treaties would be signed by certain countries, ie the big powerful ones, given the implication of using AI to further defense and offense capabilities.
Disagree. Given the pressures mentioned above, plus the extensive mundane risks and social harms of AI, plus the inevitability of human extinction if AI progress continues indefinitely, plus the arguable "are we enslaving and killing billions of sentient beings by using this?" ethical concerns, I doubt that when deciding whether to have an enforceable treaty eliminating strong AI globally (signed by all the relevant powers), the world's leaders and peoples would say "actually no, the need for the world's militaries to spend trillions on zero-sum games should tip the scales in favour of keeping AI".
Wrote this ten years ago and I think it’s aged pretty well, predicting how this plays out:
https://medium.com/@MarkPXuNeyer/first-they-came-for-the-venture-capitalists-1a8754b7176d
TL;DR, ai based investors will increase demand for human labor, but they’ll want to treat humans well because that optimizes human productivity at tasks humans are best at
"until eventually a firm starts bringing their AI to board meetings, where it analyzes suggestions and proposed plans"
Oh, I want to see this happen. That firm will crash worse than the Hindenburg. I love your sunny optimism that AI will be perfect and never make mistakes and certainly never hallucinate, learn to lie and blackmail https://www.forbes.com/sites/lanceeliot/2025/05/24/agi-likely-to-inherit-blackmailing-and-extortion-skills-that-todays-ai-already-showcases/ or be more obsequious than Uriah Heep so that it agrees with the board member who shouts the loudest about a particular direction to go even if that is the wrong way.
I think in the ten years since you wrote that, we've seen how AI is developing well and all the "who could have possibly foreseen this?" ways too.
"“Besides,” the AI investors will respond, “we have a better track record at investing in women and minorities than the old human networks did.”
What are old Internet saying me once heered? "Get woke, go broke"?
"(G)iven that the law is like a human attempt at formal language, the AI’s are masters of the law now, and draft complex legal arguments that the best human scholars agree are totally sound.
But I see they still suffer from the Greengrocer's Apostrophe 😁
https://en.wiktionary.org/wiki/greengrocer%27s_apostrophe
Let's just hope those AI masters of formal language have learned *not* to invent precedents out of whole cloth, mmm?
https://gunnercooke.com/ai-fake-cases-and-the-courts-a-cautionary-tale-for-the-legal-profession/
"In Ayinde, R v The London Borough of Haringey, the High Court confronted a startling misuse of AI-generated legal research. In that case, the lawyers submitted legal arguments that relied on five fabricated cases – among them, one purporting to be from the Court of Appeal. These citations bore all the hallmarks of real legal authorities: proper names, plausible citations, and a style that appeared entirely authentic. But none of them existed. The citations were false; the cases were fictional."
And your vision for the future of humanity is "get a job in the service industry, where pay and conditions will be even shittier than they are now because now there are too many people chasing such jobs so the AI employers can pay you peanuts, you hairless monkeys":
"This means that humans have jobs like “smile and be kind to people who are having trouble with the product”, “help your friends learn to integrate the device into their lives”, “have parties where people see how fun this new game is”, “host a large dinner party so that people see how good this food is”, “write songs that show the emotional experience of being human”, “comfort a person who is troubled being human.”
Now, what were the figures I think we discussed on here previously (or if not here, somewhere else) about how many people do, in fact, make the big bucks as Instagram influencers, Onlyfans creators, Substack writers, etc.? Not as many as people think when they set out on a career of "I'll have my own Youtube channel and be famous!"
'Have parties where people see how fun this new game is' - oh you mean like Tupperware parties? Tupperware that has now gone bankrupt?
'Smile and be kind to people who are having trouble with the product' - oh you mean like customer service agents, who are already being replaced with AI chatbots?
"(I)magine a CEO who can simultaneously say “we will increase user retention” to the board while telling a low-level customer service rep who’s speaking English — “this customer sounds upset, i see that her usage drops off 30% on Tuesdays, and this API call to Comcast shows that there’s a service disruption in that neighborhood which seems to peak around that time due to congestion. We’ve credited her account, please tell her that we will talk to Comcast to resolve this issue”
That low-level customer service rep with ESL will be replaced by said chatbot, because it's more efficient and better as the rest of your post points out about replacing humans. Why spend those valuable seconds instructing the hairless monkey on what to do when an AI chatbot can diagnose all that for themselves and do it immediately?
The AI wouldn't need to be perfect, just somewhat better than humans. I agree that it's likely to be a disaster.
Could there be a good prompt? Maybe "model yourself on the boards of companies which have done reasonably well for the past ten years."
If my future employment depends on "smile and be nice to people" then I may as well jump off a cliff now. I've worked retail before, and as God as my witness, if I have to lie, cheat, steal or kill, never again (to adapt a quote).
No idea if apxhard has ever done call centre work but from that description it don't sound like it: 'make your living hosting dinner parties', forsooth!
“write songs that show the emotional experience of being human”
We think the ratio of buskers to pitches on streets is bad now? Wait until it's the only job left for humans! Though it will see a revival of the blues plus a boost for country music.
"Woke up this mornin', my Cybertruck was gone
Said I woke up this mornin', my robot dog gone too
They done left me, said I was just a human fool"
You have a talent for being entertainingly insulting. There's probably a niche for that.
*me sitting in my cardboard box on the street corner with a sign*
"Will Diogenes for food"
Or worse but cheaper.
I wanna be a window cleaner. Pays nice in my country (Denmark) and it's not as taxing for the body as most blue collar jobs, you sit and stand in pretty normal positions, no very heavy lifts, etc.
Van Morrison has always said it’s not a bad gig.
https://m.youtube.com/watch?v=QrRSL7YCwjw&pp=0gcJCdgAo7VqN5tD
As did George Formby:
https://www.youtube.com/watch?v=hP_h6dN5048
> what's your post-automation plan
Really kind of none. I try to do the best work I can and be useful to my employer, and let the chips fall where they may.
If the more moderate predictions about how my industry (SWE) will change come true - LLMs increase productivity, we all write code less and design more - I'll probably be in an OK position. I'm self-taught so a little more pragmatically minded and a little more comfortable with facing stakeholders and playing product manager than most "pure" engineers.
If the more optimist predictions come true - white collar work is gone, UBI for everyone - then whatever, live the life of the mind, finish my reading backlog, learn photography and playing some instrument.
If the more pessimist predictions come true - white collar work is gone (or gone for all but the top professionals - for me effectively the same), no UBI, mass layoffs - maybe I'll go into trade school and become an electrician or a plumber.
> maybe I'll go into trade school and become an electrician or a plumber.
Hoping to get employed by other people on UBI?
The AI will still be running on electricity, so the large server farms and data centres will still need old-fashioned human electricians (until they improve robotics). Getting a job at the OpenAI plant may be the new company town model, where there is one major employer and everyone works for them or works jobs downstream of the direct jobs.
The electricians will need plumbers and the plumbers will need electricians
We’ll run the gas off the electricity and the electricity off the gas. Saves money.
How is UBI for everyone the optimist prediction? It's a moral disaster.
One man's moral disaster is another's eudaimonia.
Presumably because the other prediction is widespread famine and international nuclear war?
Why? I'd personally love to spend my time in art, sports, curiosity, and community. Profit signals have little relationship with the tasks that I find fulfilling.
"...live the life of the mind, finish my reading backlog, learn photography and playing some instrument...."
It's kind of funny, but that's my life now more or less. Earlier retirement than expected, UBI is my stock portfolio, no photog yet but working on my Spanish. Guitar is a work in progress... Can't say it's dystopia really.
There's a reason why No Silver Bullet remains a classic:
https://www.cs.unc.edu/techreports/86-020.pdf
Fascinating read.
Though it seems plausible that AI could automate the essential tasks of software design, as well as the incidental / implementation. Already, if you provide enough context and requirements, LLMs can generate a solid software architecture. If the problem is completely new, and has never been solved before, then an LLM probably couldn't design an architecture for it. But completely-new software requirements are rare.
Yeah, no. I'm calling bullshit on the notion that most of today's software development effort is in essential complexity. Gathering and clarifying requirements is the only essential part from the project manager's point of view, everything else is incidental complexity. Squeezing the customer for information is an important part of the project, but it's nowhere near 10% of the total design, implementation, and test effort, more like 1%.
There are countless software ideas that can be reasonably well specified in 10 minutes but take several days to implement and test thoroughly. This means that there's potential for two orders of magnitude improvement, or two more silver bullets by Brook's definition.
Here's an example, if you don't believe me: I can easily specify the concept of an HDL simulator in 30 minutes in sufficient detail if I refer to the VHDL and Verilog language standards and to existing simulators to fill the gaps, but it'll take a very competent team of ~10 experienced software developers and domain experts at least a year, most likely longer, to build a state-of-the-art simulator. That's four orders of magnitude between essential and incidental complexity right there!
"They also advise people currently doing white collar work to prepare for the near future by asking themselves "If you had 10 engineers working for you right now, what would you have them do?"
I think this is the blind spot of software engineers in general and the AI set in particular, seeing all the products like Copilot being shoved out to the general public: they think of all work as 'something you need an engineer for'.
A lot of white collar work is administrative; would "10 engineers" really be useful in an accountancy firm, for instance? "Oh but your AI assistant will write emails for you". Great, but often those emails are requesting specific data in a specific form and I don't think I trust AI at present to do that task.
If all your emails are "meeting at 10 am" and you're replying to confirm the reply about the reply, then sure, AI can handle that. I don't think as yet it can handle more complex tasks. But I think this is where businesses imagine they can increase productivity, or at least cut costs, by turning things like customer service over to AI.
I had a recent tussle with one of those chatbot AIs replacing the human customer service agent, where the website tries to make you have a 'conversation' online. The problem was that I could not log in to the newly set up business account with this particular company. All the AI was able to do was recite the script: log in to your account so we can confirm the details. But the very problem was not being able to log in!
After going round the houses with this to no avail, I eventually managed to find the (well-hidden) phone number and got on to a human after a period of time. That's the frustrating experience with AI so far and that's the cheap, replace the human labour, solution companies are presently going for. That AI certainly was nowhere near "10 engineers".
"A lot of white collar work is administrative; would '10 engineers' really be useful in an accountancy firm, for instance? 'Oh but your AI assistant will write emails for you'. Great, but often those emails are requesting specific data in a specific form and I don't think I trust AI at present to do that task."
For the accountants at the accounting firm, I think the question would be: "If you had 10 accountants working for you right now, what would you have them do?"
And unless the economy demands 10x the accounting, the answer implies that accountants move into accounting management or become unemployed.
I have historically been very bad at predicting "what comes next" but I don't see an obvious 10x increase in demand for accounting, law (maybe I'm wrong here?) and lots of other white collar jobs. I think AI optimism leads to "the folks who learn to leverage AI will stay employed and be even more valuable. The others need to find new work."
Note that this also applied to radiologists who are looking at x-rays, MRIs, Cat Scans, etc.
I don't know how porn star continues to be a job. And "porn prompt engineer" seems unlikely to be valuable, either.
For some types of engineering, 10x more engineers (especially if they are cheap AI engineers) can make sense. For a lot of other jobs 10x more of that job ... doesn't make sense to me any more than producing 10x as many cars makes sense.
But I don't know what the "new work" is. I'll note that in Victorian times a larger percentage of the population was employed as "staff." The US has kinda movec that way over the past 40 years anyway; we have a lot more personal trainers and life coaches than in 1980, for example.
>I don't know how porn star continues to be a job.
...you seriously can't imagine humans continuing to find other humans attractive?
I'm going to go ahead and say this is a large enough blind spot that it kills your entire argument.
I don't think he's saying that the AI revolution will make people horny for robots; I think he's saying that AI-generated pornography will be superior to "real" product as well as cheaper, thus outcompeting actual humans in this field also.
That's the same thing. You said the same thing twice.
People are already paying for AI art of their specific fetishes. AI-generated video of hot young things with flawless skin, impossibly large yet still perky bosoms, thigh gaps, and the anime girl expression of mindless lust for *you* which means she'll do anything you want will beat out human performers who will have to rely heavily on filters and editing, and still won't be able to match "the 100 specific elements in my pr0n that I need to be able to achieve orgasm" that AI can produce by the cartload.
Porn already happens on the screen. And you know that the girl is not really the guy's stepsister; and she probably doesn't even like him.
When the girl is not even real, how much authenticity do we lose?
Porn stars these days make money through Onlyfans cam shows with chat interaction.
And if there's a sufficiently convincing AI Onlyfans cam girl? People were losing their minds over losing their Replika girlfriends. There's a market out there of vulnerable/lonely/needy people who will fall for the AI girlfriend experience.
You know you're never going to meet up with that Onlyfans girl in reality, but she gives you a parasocial relationship. If you don't know Sukie is the AI generated fake, what is really so evident that she's not real if she talks back to you and performs for you and pretends to care about you like the flesh and blood models?
> For the accountants at the accounting firm, I think the question would be: "If you had 10 accountants working for you right now, what would you have them do?"
And the answer is: "I would get them to double-check the output of what my AI does".
This seems very likely to me. We'll have lots of people around and have many reasons to want to keep employing people (PR seems like a bigger reason every day) and we likely will not trust AI enough at any point to just let it work without oversight. Either it makes mistakes and we need oversight to keep it functional, or it doesn't make mistakes and we fear it taking over. A future where millions of capable people stand around doing nothing while giving the AI control over everything seems like the least plausible scenario.
>I have historically been very bad at predicting "what comes next" but I don't see an obvious 10x increase in demand for accounting, law (maybe I'm wrong here?) and lots of other white collar jobs.
Can everyone afford to keep a lawyer or accountant on retainer currently? If the answer is "no" (and it certainly is) then there is demand that has not been satisfied, provided that the cost goes down.
"If you had 10 engineers working for you right now, what would you have them do?"
Uhhh... do these engineers know sign language? Nobody has been impressed thus far by any of the attempts at machine translation involving sign language.
Seriously, if I had 10 engineers working for me right now, I'd start them off on:
(1) That pile of shredding of documents that are now older than seven years and don't need to be retained anymore and must be securely disposed of
(2) Those two filing cabinets worth of files that need to be reorganised
(3) Overhauling all the staff files regarding contracts, making sure all documentation is up to date, in place, and signed
(4) In another month's time we'll be preparing for our annual audit, time to get all the files and paperwork ready!
(5) In the same vein, here is a year's worth of receipt books for fees. Enter them all into a spreadsheet and cross-check them against the credit card receipts, and cross-check *those* against the bank lodgements. Hurry up!
I'm seeing a lot of "physical media that need to be engaged with" and not a lot of "write a programme to automate this" in my job that AI isn't suitable to use nor capable of helping me.
What is the informational content or value of a couple of AI developers going on a podcast and saying the thing they have to say to not tank their company's stock value and piss off their bosses? Why take this any more seriously than e.g. a politician speechifying about how their policies and initiatives are going to have good results for their constituents?
I mean, you make it sound as if they were spouting wildly expansive unsubstantiated bullshit. The informational content is to learn where frontier models are likely to go in the short to medium term, with a sincere and collegial tone that makes the information transfer stick. Their claims aren't coming out of nowhere, they have been echoed by other frontier model companies and are the reason why hundreds of billions of dollars are spent yearly in capital expenditures to bring about the commodification of white collar work, a lot of which is repetitive, pattern-based and knowledge-gated, which is exactly the type of thing that RL built upon the current suite of models is improving at rapidly (and measurably). Human-equivalent tasks automated by frontier models have doubled in length roughly every seven months since GPT 3. If you think this trend is unlikely to continue until a full day's worth of work can be automated, the burden of proof is on you to show why that trend would break (blanket "can't trust corpos!" speak won't cut it). It is tough to reckon with the fact that people juggle with different goals and aspirations, and that personal ambition/enrichment is not strictly incompatible with giving away information that is useful beyond the enrichment. Or maybe we should tell biotech industry executives that their claims are informationally void before the FDA approves their drugs? "Show me the results of your current claims about the future first, only then will I finance/be interested in your project!"
The parallel with a politician speechifying is flawed, because politicians rarely build their speeches around easily verifiable up-to-date academic research and measurable benchmarks that point in one direction with a lot more salience than in another direction. Then, these same politicians harness the blind trust their constituents have in their (unproven, immeasurable) policing qualities to shift the goal posts endlessly, playing the blame game for decades with little to show for it. There is a much better mechanism at play with AI investments at the moment: the mostly free market doing its thing, allocating resources to the highest expected present value avenues. It isn't perfect, it has failed before, but it does a measurably ok job at generating somewhat sustainable economic growth in developed economies, which is a hell of a lot more than you can say about the democratic process in the past ten years for example.
The problem is that these guys have an incentive to talk up the likely near-term importance of AI, and also probably wouldn't be doing what they're doing if they didn't think AI was going to a have big impact on the world. That doesn't mean there's no information in their words, just that it is hard to imagine people in their position saying the opposite things.
Yeah I get that. But I think that when it comes to really serious matters, like AI or Covid, it is just evil to use the authority you have because of your job to broadcast misinformation. During Covid there was some MD's on Twitter that built huge followings trading in misinformation. If people like Patel can't give the public the truth they can at least shut their damn mouths.
There’s no money in giving the public the truth sadly
I've worked as a software developer for ~30 years and I have never worked on a project bottlenecked by the number of engineers (at least not the way that term is usually used).
The bottlenecks have always been one of the following:
1 - collecting and clarifying customer needs / requirements.
2 - reconciling requirements with existing software.
3 - verifying that a given implementation meets the requirements and does not cause regressions.
It's not that engineering takes zero time: it just takes less time than the other three activities.
Given that software engineering is widely understood to be one of the most tractable jobs for AI, I'm inclined to heavily discount the Anthropic employees 'predictions. Maybe they've seen AI agents perform some of the critical tasks I listed above, but I have yet to observe them in the wild during the last few months of heavy use myself.
So far, all I would predict is that AI will make the engineer's experience much better. The gains may exceed the ones we got from any previous generational shift (1GL -> 5GL), but from all evidence I've seen so far, they are magnifying a human's power, not replacing us.
An ocean of human activity still lacks automation, and new cases are born every day.
Automating testing might be a thing.
OTOH, the hardest part is *always* robustness testing: what it does when things go wrong.
Automated testing is great as long as your automated testing procedure has been thoroughly tested.
Exhaustiveness is hard to test.
> Automating testing might be a thing.
We already do that. I imagine you're envisioning something like testers or QA folks clicking manually through a website that would then be replaced by an LLM doing the same thing?
Having humans dedicated to that is already not all that common, in many placed people write a suite of end-to-end tests that automatically runs in the pipeline or something.
Perhaps the AI could help writing that - or eventually write it itself - but that's basically the same as "AI will make coding faster", because tests are just some of the code you write.
No, actually, we don't do that in many situations. There even AI going through a test suite would be an improvement.
I cannot speak for the entire industry, just the part that I'm working in. It's been my impression that the "automated tests are a waste of time" mindset is no longer common.
I've never worked on a project or company that's not done automated testing, and the worst I hear of my co-workers is along the line of "the client does not believe in unit / integration tests and we only get to write the other". It is possible that we just don't work for clients who don't test since the mindset difference is too large for us to be a good fit for them. But all the evidence available to me points to automated tests being accepted as the default software dev practice.
Is your impression different from that?
It's one thing to think they are a waste of time. It's another to actually implement them.
The classic book “The Mythical Man Month” goes in to some detail on why putting more engineers on a project does not necessarily get it completed faster. (Because they will need to do-ordinate with each other…)
But, with that caveat aside, I can think of plenty of projects where the bottleneck is getting implementation done.
Sure, it’s not the only thing that needs doing, so if AI gives us nearly infinity programmers, projects will not get comp,teed in near zero time because all the other stuff will now be the bottleneck.
But I can see the opportunity for substantial gains here.
There is a really wild thing you can now do to get quick customer feedback, which is not write the software at all but have an AI just pretend to be the app. This can get you rapid feedback about what is and isn’t really required,
Yeah, I've been thinking a lot about "the MMM" lately, and in some ways Fred Brooks is at the heart of my skepticism.
I imagine you're right that there are projects out there where the bottleneck is implementation. I can *conceive* of such projects. But I can't think of any I've been involved with, and I've worked for a lot of different types of software company (Though crucially, not in embedded systems or on projects that have exceptionally high safety standards.)
As you mention in your other reply, prototypes are an area that AI might offer enormous leverage. Balsamiq mockups made communicating about software a lot easier, and AI prototypes seem to offer an OOM improvement on that. Once you've gone that far, it seems reasonable to just iterate on prototypes until you have an actual application, but I think there are a ton of unanswered questions: feature discovery, feature conflict discovery, version control, taking or assigning responsibility, etc..
At any rate, I agree about "substantial gains", but I still see software makers running up against other bottlenecks very quickly.
> But, with that caveat aside, I can think of plenty of projects where the bottleneck is getting implementation done.
One area which might serve as an example (if we also shove artistic tasks into the "implementation" drawer) are video games.
You can start with a blank slate. No "it has to be written in Java, and it must interface with our ERP software and our database, and also deliver reports in that format etc". Instead, you can say "The game should run on any one of Windows, GNU, Android, iOS", thus not constraining the creative freedom of your LLM.
(The downside is of course that just like for other easily replicable art, you are always competing against the very best. There already exist more good video games than you can play in a lifetime, so unless you have highly specific interests, there is never a reason to play a game which is in the 50th percentile of greatness only.
This dynamic is why it is often easier to make a living writing boring software interfacing with ERP systems than writing video games. If a zillion people have written a dungeon crawler before you, they are all your competitors. If a zillion people have written a kinda similar piece of business logic software before, it is very likely that they were targeting slightly different interfaces and none of the solutions available will work out of the box, so your custom-built piece of code will not have any competitors.)
While I have not tried LLMs in a year or so, I am also doubtful that the current generation could implement even a simple game without help from the user.
Imagine, for example, a world in which Tetris was never invented, and thus not part of the training data. You spend a kilobyte of text explaining your game idea, and tell your LLM to just use ncurses so it does not get distracted with fancy graphics.
My gut feeling (p=0.8) is that the LLM would fail to deliver a working Tetris. (And, of course, a console implementation of Tetris would still not be very marketable in 2025.)
While I can not see LLM-created games competing with indie games, I guess that there might be a niche for games tailored to custom tastes. For example, it seems unlikely that the gaming industry will create your dream open world RPG where all the human characters have neon-green elephant ears, your preference is just too specific (sorry). "Build a neon-green elephant ear mod for Skyrim" might be a niche an LLM might solve in a year or so.
They are in the business of selling Claude ... Of course, who knows what happens in 5 years (we had basically no LLMs 5 years ago, at least not in the same sense we understand LLMs now).
But notice how the LLM developers keep scaling back their goals and expectations (or rather pushing them forward in time).
My job mostly revolves around making these models actually useful in production (possibly in combination with classical ML and other techniques). And there really are not that many use-cases yet.
Everyone who tells you otherwise (about the present, not speaking about the future now) is mostly bullshitting, frankly. I mean when we have some minor use-cases for customers we also hype them up a bit ... Although I also always like to use these webinars and meetups as an opportunity to point out how important the non-LLM-based approaches still are. Partly because it is true, partly because it makes is look more credible and level-headed in the sea of AI vibecode hipsters :D
I've been working in ML for about 7 years now (theoretical maths before that) and my opinion that even classical ML is still somewhat hyped. Don't get me wrong, there are real use-cases for that. But most of those are either commodities nowadays (most of computer vision) or recommendation engines or fairly basic use-cases like credit risk modelling in banks (logistic regression FTW ... and actually there's nothing wrong about that).
These LLMs are really still looking for a market that validates the huge investment put in them. There are some niche use-cases already but the added value just does not match the investment on Anthropic/OpenAI/Mistral/whatever side. So these companies need to keep hyping and hoping that they will genuinely have a proper breakthrough in the next 2-5 years.
They might as well, but it is not certain at all.
I already replied once, being dismissive of the scenario Claude employees are putting forward ... but it is perhaps more fun to roll with it and imagine that it comes true (it probably will eventually).
I think the answer for most people's jobs is quite clear -services and arts.
With most white collar jobs automated, you will have many people who just have so much more money to spend than before ... and they will want to spend it.
One thing that humans will always be better at than LLMs (even with decent robots ... unless yo go full sci-fi synth but that definitely won't happen any time soon) is acting human.
And people prefer to go to restaurants not just because they have good meals there and maybe the place is nice but also because they get the service. If that weren't the case, people would probably either cook for themselves or order food home. It comes cheaper and you don't have to give tips to anyone.
Now arts - you could say that that's not for everyone, but actually it is not beyond most people to learn to play decent piano and go to that restaurant and play it for the guests (and get paid by the restaurant and from tips). Most people won't be the next Bach or Herbie Hancock but they can learn to play decent music. And many people definitely prefer to listen to someone playing (even if it is not perfect! it just has to be decent) to listening to a record (even if the record IS by Herbie Hancock).
So I would expect the economy to shift massively to services, broadly defined (so that even the musicians are a part of services). Also hand-crafted goods would boom. People already pay for something that is NOT perfect. Machine perfection is cheap, hand-made IMperfections is luxury and that is why you pay for it.
I would also expect a renaissance of portraits and other things which are really an over-expensive luxury today for most people. Again, it doesn't matter if a diffusion model can make a perfect portrait. Even if it somehow produces it manually with brush lines and all. It is still machine-made and therefore it will be considered low status AI slop.
Come to think of it, it sounds like a pretty cool future :-)
> With most white collar jobs automated, you will have many people who just have so much more money to spend than before ... and they will want to spend it.
No you won’t.
<With most white collar jobs automated, you will have many people who just have so much more money to spend than before
Well, what you start off with is a whole bunch of jobless formerly middle class people. How and when do they transition to having a lot of money and leisure for enjoying the arts, etc.? And by the way, I have not heard one single proposal about how AI itself could assist with that transition, and I'm sure proposals for how AI could assist are possible. Could it assist in figuring out whatever gigantic change in government and finance and international relations would make possible universal basic income? Retrain people? Develop and implement some new kind of schooling for kids that prepares them to for the new order?
Latest really new proposal for AI I heard was the screenless personal AI whatnot that Sam Altman and that Apple designer hope the world will become addicted to. You carry it around and it's your shrink, your accountant, your best friend, you social secretary, your fucking wine steward. Maybe they will set it up so that it vibrates just right, too, and women at least can carry theirs in their underpants.
Radiologists haven’t been replaced yet and it doesn’t seem that close to happening. I would have thought pattern recognition would have been one of the easiest areas to outperform humans.
Is that your field? I remember hearing stories about AI image-detection programs outperforming radiologists on some tasks but I may have misunderstood that.
It is not my field, and there is some social aspect to radiologist troubleshooting requests that may be early days to have an AI replace (why do you need this study, I would suggest a different study etc…). But I’ve been reading things about AI alone outperforming radiologists alone and + AI.
Also AI can do wild things gs like predict diabetes and heart disease from unrelated studies like chest x rays through pattern recognition that is beyond human.
That's interesting, but I've also heard similar claims about how computer vision systems can "do" sign language now and when I've investigated they have been severely lacking in the ability to actually handle naturalistic human input (or produce ANY signed output) versus just recognizing discrete dictionary-entry-style individual signs under fairly ideal conditions.
I wonder if the same overstatement is true of the radiology situation. I'd love to hear from a real radiologist about it if there are any here.
Looking that claim up, it's "could" not "can" and it's the usual "in the next five years we'll be zipping round Mars on our hoversleds" stuff:
https://www.private.imperial.nhs.uk/news-and-blogs/ai-could-predict-type-2-diabetes-up-to-10-years-in-advance
"The team, led by Dr Fu Siong Ng, a consultant cardiologist, and Dr Arunashis Sau, a cardiology specialist registrar at Imperial College Healthcare NHS Trust, developed the AI-ECG Risk Estimation for Diabetes Mellitus (AIRE-DM) tool, using around 1.2 million ECGs from hospital records.
Using data from the UK Biobank, they then validated the AI’s ability to detect subtle changes in routine ECGs that could signify that someone might be at higher risk of type 2 diabetes, years before their blood sugar levels begin to rise.
...The AI will be piloted in the next year, and the researchers hope it could be rolled out in the NHS in the next few years."
https://www.sciencefriday.com/segments/ai-heart-disease-chest-x-rays/
"Like, for example, another chest X-ray study that diagnosed diabetes– which, again, I would never have thought that was possible, right– from the chest X-ray. It basically did this so-called occlusion, or masking, where it would look at the chest X-ray and block out various regions to find out what was the source of the information that we can’t see. And it turned out it picked up the fat pads in the chest that was providing this diabetes possible diagnosis.
But for the risk of heart attacks and stroke, we don’t know how it was so powerful and better than our standard. So this is part of that X factor that we still need to learn.
IRA FLATOW: Well, I’m still not quite understanding this. I mean, how did the AI– what did it see in the chest X-ray that told doctors– or it found that the patient’s risk for heart disease was greater?
DR. ERIC TOPOL: Well, we can tell from other studies using the chest X-ray for this opportunistic detection that it can pick up the calcium score– what people can undergo a CT scan to see how much calcium they have in their coronary arteries. But that can be derived from a chest X-ray, and that is an indicator of risk.
Also, the chest X-ray can be– through AI, determine the heart strength– the so-called ejection fraction. So it’s picking up a bunch of things, as seen in other studies, that are very predictive of a person’s risk. And it must be the composite of these things.
But we really don’t know because, although the study really was extraordinary, it didn’t do enough as far as the explainability side of things. So that’s where some of the work that needs to be done– and it needs to be replicated before this becomes a standard way to predict risk of a person’s heart or stroke in the ten years ahead.
...DR. ERIC TOPOL: That’s a really important point. It’s just like when you do scans for a person that doesn’t have symptoms, and you pick up these incidental findings. That could be the problem too– the false positives. So that’s why before these are put into clinical practice on a routine basis, we’ve got to know whether you’re going to get false signals. And so far the data look encouraging, but it’s still early."
So there is some evidence that AI can pick out data in the X-ray that does predict certain conditions, but there's not a model for how it does that, so any predictions will have to be treated cautiously in case they're false.
But Deiseach, I'm not sure it matters for practical purposes whether AI can tell us how it knows that somebody's scan shows a certain illness or high risk of an illness. If the AI is clearly better than the MD at identifying those people, we should act on that info.
Well, actually I can see some practical problems with AI not being able to tell us how it knows someone is at high risk for a heart attack. If we knew what signs it was responding to, we would have highly useful info about the body changes that raise heart attack risk. and that would give us ideas for new treatments. Still, when it comes to care of individual patients, seems to me that if AI is better than doctors at identifying people who are at high risk, we should go with the AI's prediction and start people on the regimens we know reduce risk.
I presume the AI is seeing signs on simple tests that correlate with changes that it has required doctors to get things like CT scans, stress tests, echocardiograms etc…. to appreciate.
Paul Krigman just put out a post on why this is rather unlikely to happen: https://paulkrugman.substack.com/p/what-deindustrialization-can-teach
Bottom line: Computational automation of jobs has been happening for a very long time, but it has always created more jobs than it destroys. There is no indication that this round is any different.
Why do you assume entrepreneurial jobs will stay? If anything, they could be among the first white collar jobs to be automated because all you need to do them is converge different problems by analysing a market and then organise employees to solve that problem. It's a job that seems to only persist because of the power dynamics, otherwise employees would just run cooperatives.
> Why do you assume entrepreneurial jobs will stay? If anything, they could be among the first white collar jobs to be automated […]
No, they can't be easily automated by AI, because they're not well-defined, self-contained tasks with a clear right/wrong evaluation criteria, and those are necessary conditions for training effective LLMs. Which is why LLMs get better and better at well-defined, self-contained tasks with clear right/wrong evaluation criteria, like coding benchmarks or math puzzles, but continue to disappoint on non-well-defined, non-self-contained tasks without clear right/wrong evaluation criteria, like designing and implementing complex software solutions based on fuzzy, incomplete, contradictory and unknown user requirements. Or entrepreneurial work.
> […] all you need to do them is converge different problems by analysing a market and then organise employees to solve that problem.
Those five words, "all you need to do", are doing _a lot_ of work here.
I find it's much harder to design precise technology than to tell people to do so.
What you, a human [1], find easy or hard, has almost no bearing on what is easy or hard for LLMs respectively other AI architectures. For example, LLMs have gotten really good at solving math problems which are so difficult that only a small fraction of people even understand the question, while booking a flight from A to B in a web browser is still a problem on the bleeding edge for them.
[1] I presume.
To be clear, you agree with the commenter that "guy who can start a business" will be the last to go? Even under an "AGI" scenario?
> To be clear, you agree with the commenter that "guy who can start a business" will be the last to go?
No, I don't think they'll be literally the last white collar jobs to be automated. I'm arguing against your claim that "[entrepreneurial jobs] could be among the first white collar jobs to be automated".
How is entrepreneurship supposed to be a solution? In what fields specifically? Are they saying to just set out to "hire" AI instead of humans and find a field to be useful in? Or are they talking about software specifically?
I think there's a real bias around here toward assuming that as goes software development, so goes all of business, but I am very skeptical of that notion.
Has it occurred to these Anthropods that very few white collar workers would be able to come up with suitable projects for 10 engineers, or supervise them — none but other engineers, and people with lots of expertise in stem fields. if Dwarvish Petard is right, the next stage of AI, rather than nudging life towards cyber-utopia, fucks over everybody except for people who are a lot like AI developers. I’m shocked, SHOCKED, I tell you, to hear that things might play out this way.
People who read Scott’s 2 recent covid posts and/or the comments to them: Did you change your views about anything?
I didn't see any new data, and I didn't see any new arguments, and I didn't need anyone to prompt me to notice that COVID had been a pretty big deal, so no.
the graph that compares excess deaths with covid-attributed death surprised me. I knew there should be a correlation, but I was surprised how strong it was.
At first glance, I was surprised how many people in the comments were irrational [1] about their positions. I expected these people to be rare in a rationalist community. At second glance it was just about 3 people writing lots of comments. After blocking these, the commentsections returned to normal.
It was very interesting to read the remaining discussions, especially the one linking the john-hopkins meta-study about the effect of lockdowns. I changed my views from "lockdowns saved a lot of life" to "lockdowns saved very few lives, especially when compared to voluntary social distancing".
[1] "irrational" in the sense, that they had some weird ideas, and were unable to defend them when pressured with facts. Thanks to everyone who argued with them even if you may not have convinced the person you argued with.
"Thanks to everyone who argued with them even if you may not have convinced the person you argued with."
Only tangentially related, but I do keep in mind the silent/observing audience when I'm going back-and-forth with someone online. My position and/or data may not matter to the person I'm interacting with, but it might for some in the silent audience.
I also saw that graph for the first time and it also surprised me. I had entertained the idea that many reported COVID deaths were not legit for whatever reason, but if that graph is correct then that is much less likely.
I learned that at some point Scott believed that 20% of young people end in hospital due to covid.
The discussion showed again that people cannot grasp the idea that any risk from covid increased exponentially by age. I don't know why.
I didn’t get that he thought that at all from the posts. Could you share the quote?
This one: https://www.astralcodexten.com/p/the-evidence-that-a-million-americans/comment/119233511
Thanks, I hadn’t seen that. It seems pretty clear there that Scott at no point believed the 20% figure. He makes it clear he had significant doubts, but acted as if it were true because of the lack of countervailing evidence and because it was the fail-safe rather than fail-deadly option.
This part seems to make it clear:
“I was suspicious at the time, saying:
> "This is a weird pattern – why are so many young people getting hospitalized if almost none of them die? ... Are these an overestimate? Maybe most cases never come to the government’s attention? There’s some evidence for this ... So hopefully the 20% hospitalization rate will prove to be a worst case scenario, and the real number will be less."
...but absent 100% proof that this was the explanation, I elected to stay inside rather than get a disease that had some chance of a 20% case hospitalization rate for young people.”
"People who read Scott’s 2 recent covid posts and/or the comments to them: Did you change your views about anything?"
No, but I spent tens of hours assembling data and running statistics on it in early 2022.
It surprises me that so many people here seem to have… not done this.
I expect that lots of people want the answer, but don't want to spend the effort required to dig through the data. I don't find that position morally questionable. I just cared enough in 2022 to want to answer for my own satisfaction the question "Was Covid just a bad flu season." I cared enough to think about how to answer this and then actually did some work to find and analyze the data.
But ...
A Great Courses lecture on economics introduced me to the concept of "rational ignorance." The idea here is that you might *want* to know something, but (rationally, in economic terms) conclude that knowing isn't worth the price (in terms of time, if nothing else) to actually find out.
I, myself, often would like to know something but not badly enough to do the work to find out.
Also, if one waits long enough then often someone else will do the work :-) And, if we are very lucky, also post the data, the sources and the analysis!
To what end?
Sort of? My prior was along the lines of, "COVID deaths only very weakly correlated with excess mortality, therefore most COVID deaths aren't actually due to COVID." I wasn't strongly attached to this prior. It very likely came from a pile of cherrypicked / junk data surrounding COVID politics. I generally trust Scott to find "true" data and confirm it several different ways. So I've updated strongly in the direction of "huh, I guess COVID did cause all those deaths."
And you don't find anything just a bit suspicious about the notion that the lockdown made no change?
Why would that be suspicious?
Do you really think that something as large as the lockdown had no effect on anything?
Neat numbers should always make you suspicious
I didn't find the notion that "lockdowns made no change" suspicious, merely erroneous. Of course they had an effect. For example, COVID's lesser cousins common cold and flu almost entirely stopped spreading.
Nor do I find Scott's neat numbers suspicious---or at least not suspicious enough to investigate further.
Having said that, I'm not sure how a weak belief "excess mortality barely correlates with COVID cases" should have led me to suspect an effect of lockdowns.
Again, why should that be suspicious? What is your model of how lockdowns would affect deaths if nobody was fudging the numbers?
An absence of noise does not require an alternative model. It is sufficient in itself to arouse suspicion. This is basic statistics.
Lockdowns ought to affect the death toll in a few ways. Let's assume that everybody got covid eventually. The upsides of lockdown would still be:
1. Spread cases over more time, so that the capacity of the medical system doesn't get overwhelmed at peak. This was the initial stated goal of lockdown (in countries which didn't intend to eradicate). Given that ICUs were overflowing in many places and given that ICU-type interventions did seem to be useful in some cases, this should be a nonzero effect.
2. Delay some people's cases until after they get vaccinated. This seems to be a much bigger effect, and the vaccines seem to have done a pretty good job of preventing serious cases.
There's enough numbers out there to estimate the overall effect of lockdowns in any particular place.
The question that began this thread was “did Scott’s covid post and their comments change your mind about anything?” If you want to debate
some covid-related claim, how about starting a new thread for that? If you keep it on this thread the whole thread will probably turn into more covid arguing, and I and anyone else interested in how much mind-changing went in will not get the benefit of a good sample of responses
Like several others, I have been lazily entertaining the idea that the death toll might have been off by up to 20%, but seeing the tight correlation with excess mortality dispelled that.
I also would have guessed there was a significant harvesting effect, especially in the US, where we entered the pandemic with an exceptionally vulnerable population. I'm less sure of that now.
No. I mean, I already knew Covid killed a lot of people
During the initial weeks of covid, I had a wide range of uncertainty about how dealt it mwas got be. Like, maybe it will have a 10% infection fatality rate. It fairly quickly became clear that we dodged a bullet here, and Covid only killed millions rather than tens of millions.
No, because I didn't believe in right-wing conspiracy theories in the first place.
Nope. I never thought the official covid death numbers were all that inaccurate. However, I still think most governments' anti-covid measures were ineffective and actively harmful. Ie, I think similar numbers of people would have died regardless of the lockdowns, school closures, etc
Yes. I downgraded my expectations of the ACX commentariat.
Do keep in mind though that a lot of the arguing in favor of various pieces of misinformation was done by a few people who seem to have come to ACX solely in order to argue about covid. I’ve never seen their usernames here before. A couple of them have Substack blogs of their own, and may have come here mostly to get more visibility.
Am I upset that some grifters promoted their bull? Sure, a little, but it’s to be expected.
I’m disappointed that so few people engaged with the very interesting observation that Scott made: it really does seem strange that a million people died and we are talking about this other stuff. I think the two “biases” 1) dead people don’t tell sob stories and 2) controversy sells, make sense.
But I also think that there’s a lot more to say here about the psychology around this situation. And I think understanding it might better inform how to talk about Effective Altruism, which is important.
Maybe that discussion happened somewhere in the comments, but if it did, I could not get it to load while everyone else was enthusiastically embracing bias #2.
< But I also think that there’s a lot more to say here about the psychology around this situation. And I think understanding it might better inform how to talk about Effective Altruism, which is important
Many who engaged with the issue of the number of deaths commented that most of those who died did not have many years left to live, so that the loss of person-years was much smaller than the big 1.2 million number leads one to believe. And they said that with no emotion. And yeah, I get that. I don’t know whether that is the EA perspective. It would not surprise me if it is.
I find these comments extraordinarily painful because I am old enough that I will soon cross over into elder territory, or maybe already have, depending on how you calculate it. I am still healthy and energetic and work full time, but if you just go by the numbers then yeah, my life is worth way less than 90% of so of people on ACX. And if there was a lifeboat type situation, I think I would in fact give my place to somebody who probably had more years to live. (It would be awful, though, to watch the lifeboat leave. ) But to have *other people* taking it for granted I don’t get a place in the lifeboat, and not even having any emotion about it, makes me feel terrible. I’m not sure whether there is anything wrong with the math that leads to the conclusion that the deaths of older people just aren’t that big a deal. Maybe. I do think that in some ways I am the most valuable I have ever been. I’ve read and experienced and thought about a lot of things, and I’m much less self-absorbed than I was when I was younger, and preoccupied with developing my life. These days I’m most interested in distributing what I figured out, sort of like strewing fertilizer. Does that make it a bigger deal if I die?
And then I think about the account I read of Oliver Sach’s last months of life. He had a collection of samples of all of the elements, laid out like the periodic table, and he went through them in order, one per day. Said he was comforted by touching and ruminating about these eternal things. Also, during the last few years of his life he had a lover, after a 30 year period of having no partner and no sex, mostly because he was so shy. And the lover hung out with him through the end of his life. Are those last years of contemplation and companionship just not very important, because he was retired, frail and deaf during them?
I find ACX readers as a group to be exceptionally avoidant of emotion. Did you read the guest post Scott put up maybe 6-12 months ago by someone he knew who had an incurable brain tumor? The post was an extremely intense, dense, convoluted meditation on how the patient had to work with his mind and control his actions in order to partner with his treaters. It was very intellectual — sort of like some physicists theory of something like turbulent flow. I could barely stand to skim it, for reasons I’m about to explain, but had the impression that although the person who wrote it was extremely smart, there was some kind of cognitive slippage manifesting in what he wrote. It fell short of making sense.
But as part of it he told the story of his diagnosis and treatment and there was a crucial event that seemed to have sent him into developing his theory of patient-medical professional partnership. He was in the ICU, and kept moving around because of his discomfort. The movement made one of his IV lines come out, and a lot of his blood dribbled out of it onto the floor before the nurses discovered it, and he almost died. The nurse who discovered the problem screamed abuse at him. I can’t remember a thing about what she said, except that it was utterly cruel and unreasonable and hysterical, and that it wasn’t just a sentence or 2 but a whole tirade. And it was this episode of abuse that the man wound himself around. He saw it not as abuse but as a powerful clue about how the patient must act and talk with the staff if he and the staff are to be partners in healing him.
So I read that bit and it was clear that this poor, incurably ill man was horribly abused by that nurse, but because he depended on nurses for his life it was not an option to be angry at her and try to get her punished and removed from the staff — so he built this monstrous convoluted theory instead. I would like to tie that nurse to a chair and slap her til she cries and cries and cries, then fire her and make sure she loses her license too. And failing to do that I would like to be able to tell the guy that he was horribly abused and I will make sure that never happens to him again. But of course I can’t do any of that.
Anyhow, while I could only stand to skim the man’s article, I did read all the comments, and everybody got into I’m-smart-as-hell-too mode. So people either posted comments about what an impressively smart model the man had built, or else manifested their cleverness by suggesting fancy little tweaks it needed. Not one person said a word about the nurse’s abuse.
Um, didn’t mean to write so much, but once I started this rant overcame me.
I understand the idea that if you're forced into a horrible choice where you have to decide whose life to save, you might need to prioritize looking at the number of expected future years of life. But it would be a horrible choice, and it could only be made with tears in your eyes. It would be a terrible mistake to then conclude that anyone with fewer years left to live has a less valuable life than those with more years to live. Human life is infinitely sacred, and those infinities don't lend themselves well to simple multiplication or basic comparisons of "more" and "fewer". Rather, all human life is equally infinitely valuable. I hope that this awful mistaken view is not held by as many people as it might seem from substack comments.
Rationalists are not humanists.
My parents are elderly. Early in the pandemic, I saw the numbers coming out of Italy, did a quick calculation, and determined that if they contracted COVID, there was about a 15% chance one of them would die. They were very careful and avoided COVID for five years.
What would they have missed in those five years? The marriage of their daughter. The birth of their first grandchild, and time to watch him grow into a thinking, speaking little person. Five years to be with each other and support each other. And time with me, which I value extremely.
I write this with tears in my eyes. My comments elsewhere don't show my feelings on the subject. But I do have them.
Triage is hard on emotions. There's a sense in which the whole effective altruist project is a practice in triage. It makes sense that norms around that project have evolved to avoid emotion.
Having met several EAs in person, I do not get the sense that this detached way of thinking blunts their emotions in real life. (Unlike some MDs I've met who became callous after working in the ER for a few years)
Appreciate your comments about the elders in your life. Interesting point about EA and triage — yeah, it’s like the whole project is triage writ large. I think what offended me about the ACX comments about covid deaths was that it was *retrospective triage* — the covid deaths count is less worth grieving by because most of those who died were old. Those comments also bother me because it’s clear we could have had far fewer oldster deaths by taking steps that would not have inconvenienced the rest of the population — so certainly not “lockdowns” or required masking for all.
It makes me wonder about what life was like in the late 1940s. Did people get back to life quickly and only occasionally say things like "Hey, remember when we had that war? That was weird, huh?"
Reminds me of in interview of two english women a few years ago. They had just turned 100 (twins) and one said, "life was great, except the war...that was awful."
Your comment made me realize something I’d never noticed before. Labeling things as “weird” is actually a pretty recent phenomenon. My parents spent their formative years in the pre-weird world, and spoke about world events with labels from the old-fashioned emotion register: Things were pitiful or horrifying or shocking or heartening. I and my peers, though, took naturally to labeling things as different shades of weird. I remember leaving an outdoor concert with some friends when I was college age. We got into our car in the outdoor parking lot, then spent half an hour in a complex slow-moving mass of headlights before reaching the exit. “It’s a joke!” exclaimed the friend who was driving, “a hideously weird joke!” We might have used “weird” when talking about the news that some famous person had committed suicide, or the idea that if we had been born in an earlier era we might have ended up as soldiers doing trench warfare, or the possibility the world might be ended by a pandemic. What’s that about — the switch from responses on the human emotional spectrum to the “it’s weird” response? Something about increased detachment, something about drug experience . . .
"Weird" flags something as noteworthy without committing to an interpretation. It's an invitation for others to weigh in with their own take, positive or negative, before giving our own.
I think it is used more now because we are a more accepting culture. Then, once established, it became a habit of speech.
I'm reasonably certain China is cooking the books. I worked in Shanghai for two years, and the consistent pattern was that showing pretty much Anything Xinhua News sent to the Western press to my coworkers caused them amusement, and occasional laughter.
I and every non-CCP acquaintance from my time there suspect the numbers are much like India's, but they were attributed to other causes. China has an awful lot of "remote interior" where such things can blend into the background noise.
Seeing the Chinese numbers quoted unironically seems to be an error, but other than that the data are the data.
Sweden is also an odd case, not having much in the way of International borders. My recollection was that their numbers were lower than Norway, but I don't recall any analysis as to why that might have been.
A book-technology friend and I are trying to come up with the best metaphors for AI in the context of age-old metaphors for knowledge: light, Tree, sea, web/network, cabinet of wonders, treasure trove. New ones would be jazz (AI riffs) or my favorite: a ball of twine. All thoughts welcome.
Compost? You just heave all these odds and ends, including actual shit, into a container, and complex processes that only an expert in the field can understand turn the mass into stuff that’s profoundly rich and nourishing.
The Miller–Urey experiment might be a more appealing version of the same thing.
God breathing soul into shaped dirt.
Golems?
I don't think that your response to MR was wrong, it would take an extraordinarily generous reading of Cowen's original post to reach the interpretation that he expects claims to have intended in his response. It's great that he doesn't stand behind the reading that basically everyone came away with but I think he should acknowledge that he messed up.
It's telling that his defenders in the OP are defending the interpretation that you criticised, rather than the completely different one that he apparently intended!
TC is a fantastic interviewer, asking provocative questions and platforming an amazingly diverse cohort of intellectuals from many walks of life. I think he has a good sense for "big, if true," "overrated/underrated," and provocative and novel ideas. He's open-minded, for the most part, and far more willing to deliberately experiment with new ways of experiencing the world than almost anybody.
However, I really do not care for his commentary on the ideas he chooses to spotlight. I often feel he is attempting to trick me, playing 4D chess of status perception manipulation to "win the discourse." His writings on AI, rationalism and EA are a primary example. I also know that - probably as a result of focusing so much on novel, important ideas - the directions he chooses to highlight often have something methodologically flawed about them.
All in all, I see TC as sort of an intellectual venture capitalist, but what he wants primarily is status/influence, not money. He has a high miss rate, and to make up for it, he retcons many of his past comments when they get criticized or new information comes to light.
I still at least scan everything he posts on MR, because I think he does have, I won't say *better* taste, but *different* taste from most authors online, and that adds to my intellectual mix. I look for what seems interesting to me and don't depend on him as a trustworthy authority, because he's not, and I don't think he's trying to be. He's hardly unique in his flaws and he is highly unusual in his virtues. But one should not read TC expecting to get an accurate, thoughtful, objectivity-attempted, honest, in-depth, rationalist-style analysis. It's just not his brand.
How do you feel about increasing presence of AI writing on Substack, reddit, other social media (or elsewhere)?
It's horrible. It will make it harder to find genuine stories, especially here on Substack and on Reddit. We are not here to hear what a machine has to say. We are here for people.
I'm not necessarily here for people. I'm here for good, reasonably accurate writing. AI fails on that, too.
You bring back a memory.... I was in apas. (Amateur Press Associations). They're what people had before the internet. People would write zines, sometimes at considerable length, which would be collated together and mailed out, usually monthly. (Details available if anyone wants them.)
Anyway, I said something which failed at indicating that I was writing to humans, and I was asked how I visualized the people who wrote the zines. I said "as being 8 1/2 by 11 inches". And this was roughly true-- the truth is, I wasn't visualizing them as people, and my contact with the apa was the paper zines.
I'm rather better at people these days, but it would be fair to say that I'm presumably somewhat autistic.
I agree absolutely. Not because internet engagement is "personal" - while I know that's true for others, I am perpetually surprised when that turns out to be the case (e.g. people getting so upset with one another on the internet, in personal terms, although yes, often they are only trolls but I don't much trouble to distinguish), and I'm not sure if it's generational - the screen is only ever a screen to me - or to do with coming up through a somewhat rougher and more indifferent-to-feelings childhood/education (as we all used to) that I cannot feel that way about it.
It's because I am interested in what other people are thinking, and what they are thinking about, and books they've read, and Things They've Heard About that I wasn't aware of, the latter the same as if we were gathered around a domino table outside the store.
So to the extent the internet has destroyed as many good things as it created - I expect that AI may be its doom. (Not to give AI too much credit for that, as it would have happened anyway.)
Perhaps we will be the last holdouts here on Substack, where there are people talking to people.
"It's horrible. It will make it harder to find genuine stories, especially here on Substack and on Reddit."
Back in the days before Google and Alti-Vista, it was common for individual people to have proto-blogs. The internet community hadn't settled on blog as the term and these were often referred to as on-line journals or on-line diaries.
And these personal sites often included links to other sites that the individual author found interesting or useful. By their nature, each journal site depended on human curation (some of which was better than others). I've been wondering for about a year if the non-commercial web may drift back in that direction: Personal sites with links to other sites (And YouTube channels) that the individual finds interesting.
I had a webroll back then on West Coast Online. Everyone had two links on their webroll to friends, then the British Museum and another to Jim's Bicycles.
I think we can make it work here on Substack if the owners don't go too far in the direction of TikTok or Twitter.
I wouldn't even restrict it to Substack.
I *also* feel that the web lost something (and a big something) as personal sites (usually hosted) were replaced by individual Facebook/Twitter/whatever pages.
I've got my own "journal," commercially hosted as I don't want to deal with the nuts-and-bolts of keeping a server up, but my own and not attached to Facebook, Twitter, Wordpress, etc.
It mooshes (that's a technical term!) together short posts, long form essays and links to other sites. I also use it to play around with HTML, CSS and Javascript :-)
It *can* be done. It is just more work than not. And I understand why most people have no interest in doing so. But maybe things start swinging the other direction as the interwebs become more hostile to actual content.
I hope. The internet would be more interesting with more content like that.
I have a WordPress site that I host myself. I designed the theme and wrote all the CSS. The number of visitors is declining, though, as I get harder to find. If I post the same thing here and there, I'll get 20 readers on my blog and 1000 on Substack.
I think Substack has it right... but it's definitely trending in the wrong direction with videos and chats etc.
I think it makes popular content completely unreadable. On Reddit, not only are most popular posts written by AI, but many of the comments too. X is even worse, as each post leads with @Grok Is this true? or @Grok What is this?
I don’t normally look at popular feeds, so I guess that’s why I find it so recognizable. I haven’t been a boiled frog, so the difference is more recognizable.
I've blocked grok for that reason...
What do you mean? I think I never encountered an AI written post on Reddit and now I wonder if I'm wrong about this. Are those post marked as such or given out as human written?
I've got some tough news for you, Shinji...
Most of it is slop. A very small amount is good, some so good it passes for human written. But if we're getting everything written by AI, eventually we will be living in a completely fictional universe, because AI is not going to the old fishing hole with my pals Bert and Sammy and that's where I met Sal, Bert's cousin, and eventually we ended up on our first date and today we're married with three kids and a cockapoo.
Fiction writing might be different. It's not going to matter much if AI or Hume N. Beene is writing the story about the monster under the bed that eats chartered accountants and the plucky band of auditors that set out to track it down and destroy it. There's a lot of badly-written by humans self-published stuff on Kindle out there already, if AI can write better than that it won't be a tragedy to replace those.
A vague analogy here is those subreddits where people post stories (AITA, offmychest, TIFU, etc) which gradually devolved into a large fraction of the posts being pure fiction, no AI required. You can get more internet points if you gin up a totally unbelievable story than if you post mundane reality.
To what extent do writers who become good need to write mediocre stuff (possibly for real audiences) for practice?
I'd say it's a natural part of the process, like every skill and art - you get better with practice. There are authors where I've definitely noticed progress over the years in their writing.
So starting off with slop need not be a sign that they'll never improve. But the problem with the self-publishing/small press firehose of content is that there's no-one there to guide them, the editing is sparse if any is done, and there's no feedback on "this works, this doesn't, you're not developing the plot here, there's the loose end you need to tie up".
I'm not necessarily complaining; there are some genre series self-published authors I read because I just want A Story and not the Great American/Irish/British Novel, and they give me serviceable prose with good plotting. The characters may not be deep and the story goes from A to B directly, but that's all I want when I want junk food for the brain. The frustrating part is when there are hints of good writing or plot elements that are left dangling, where with someone experienced to edit the book, the author could be encouraged to improve. Wasted potential.
It's ruining my Substack feed for sure. I'm bombarded with gender wars and dating scene-type content, presumably because I went on a Knowingless reading binge or something? On the rare occasion that I read any of it, I'd say there are clear signs of AI writing in at least half of them?
Occasionally I'll be presented with an interesting or thought provoking title, only to realize 2 minutes in that I'm either reading a completely AI-generated post or a severely AI-enhanced one. I recently stumbled upon a post called "Islam is the bane of civilization" or something close to that effect, written by someone whose name sounded muslim, which was enough to pique my interest. The entire thing was structured as "Thing is X, not Y. The essence of thing is to X, not to Y. I want people to do X — not to do Y. The reason for thing is X, not Y." The constant "this, not that" made me feel like I was going genuinely insane, and that post is one long slab of text.
I'm stuck between a rock and a hard place on here. I know there must be a large amount of truly high quality publications on Substack, but I'm not subscribed to enough to have my daily dose of solid articles in the "Following" section. So I tend to look through the feed, which is hot garbage at the moment. It's making the whole UX unsatisfying
Background: https://willstorr.substack.com/p/scamming-substack#%C2%A7anonymous-author
a friend of mine runs https://www.pangram.com/
which afaik is sota on ai detection in writing
would be curious to see if those articles will is talking about are AI. For e.g. all three of the samples that Will shows in his 'game' are highly likely to be written by AI with 99% confidence
It makes me far less trusting of anonymous writers who have not had a long, continuous presence online. And I pretty much instantly click out of any vaguely described "musings on reality" type blogs. On the flip side I am now a lot *more* interested in people with an established reputation who are blogging on something they have deep expertise or experience. It is a great time to write under your own name. Unfortunately that also means there are some topics that get written about less, since understandably people do not want to write about them under their real name.
I can't be bothered to read things no-one could be bothered to write.
IMO, all AI-authored content should be required to be marked as such. Unfortunately, it's difficult to enforce.
I suspect that my preferences are unusual enough so that it is not profitable for any spammer to produce content optimized to win me as a reader (and hopefully a subscriber).
It is unfair that the AI content is not clearly marked as such, because it is a kind of fraud, if the customer expects human-made content.
My greatest worry about this is, how many accounts actually have the same owner, and whether the owner will leverage their position to e.g. push some political agenda on the readers. Then again, this already kinda happens with the mass media.
It’s quite frustrating.
I moderate one of the splinter forums from the old subreddit. We mostly deal in culture-war arguments, with a brisk side trade in niche trivia. Our writing already tended towards the verbose, and while I won’t pretend that signified quality, it generally required some effort.
AI has reduced the time and effort needed to poop out a wall of text. Boring text. Text which removes all the usual incentives to engage and critique, because the author has no skin in the game, no reason to change *their* mind about anything.
I have yet to see one of our users use an AI to present their interesting technical discussion, or a story from their childhood, or any of the other ways to get an unfamiliar perspective. No, it’s always the culture warriors. The people who get the most out of AI are those who want their arguments to be soldiers.
Is anyone going to NYC tech week, or has recommendations as for what events to attend?
Open-access startup conferences aren’t really my thing, as the average level of competency is generally low, and 90% of the events are just trying to sell you something. I went last year, picking events that seemed interesting at face-value, and I was almost completely disappointed.
Open to recommendations, or advice on how to better pick from the very long list of events.
I agree with you that tech week is generally not worth going to. The real networking events happen separately from the advertised ones. If you know a VC who is hosting a smaller dinner, it can be worth going to that. Otherwise, skip it.
(It's very worth getting friendly with a few VCs if you are interested in startups, but tech week isn't how you build those connections. You build those connections outside of tech week, and then events like tech week are mechanisms to get everyone in one place)
Yeah. I live in Manhattan so it's no effort for me to attend. A hardware event I went to last year had some interesting robots, so there's that. I signed up for the Spartacus event recommended by Scott, otherwise I would have forgotten it was even this week.
I actually run a startup, but we're cash flow positive, and have never "needed" to raise, so my connections with VCs are quite lacking. Thanks for confirming my impression from last year, as I don't have to waste another few afternoons trying to see if my picks last year were just bad, or if it's the events in general.
See you there!
Any events you recommend? I only signed up for one that seemed particularly interesting, but am open to suggestions.
It’s all hit or miss, but stay away from the generic networking events unless you want to be harassed by devshops and service providers
Yeah. That was my experience last year. A total waste of time. I’ll be at the Spartacus event, so I guess I’ll see you there! Although you won’t know who I am. 😈
I was chatting about this with my wife. I expressed the regret that we had not helped our (now grown up) children enough (we helped them more than my parents helped me but…). My wife said that was ridiculous. Normal parents don't help their children with homework, university applications, exam review/revision etc.
I asked a friend and he agreed: kids have to learn that stuff for themselves. They are on their own. My view may have been distorted by our Indian friends. They did everything for their kids and even hired tutors and sent them to Saturday school.
How much should parents be helping teenage school children with their schoolwork?
'should' by what metric
if the goal is to get them into a top college for networking purposes, a lot
if the goal is to raise productive capable people, actually figure out with them what homework is worth doing, set a minimum standard for grades, and then teach them to be genuinely curious about the world and beat that standard
anecdotally, my parents never helped me with homework, but they did take me out of a physics class I was getting Cs in that I loved and the rest of my grades were suffering. i (after getting really really mad and not talking to my mom for months) stopped spending all my time trying to (and being bad at) learning physics, and out of boredom i took all the hardest classes and got basically perfect grades until i got into college, at which point i got perfect grades until i got the internship i wanted. and then i got Cs and Bs because i was prioritizing other stuff in life.
indian parents are great. it's the results of chinese parents but with 10% of the trauma, but it's not worth the effort in my opinion. i'd prefer to have had indian parents (when it comes to schoolwork, not in other ways) but i wouldn't be willing to raise my kids the same way as a trade, just too time consuming
I think helping your kids if they ask for help is good, but even then, there are limits. Checking their homework is OK. Helping to do their homework is not. My kids practically never asked for help.
My parents were only vaguely aware that I went to school.
What do you mean by "help"? If you mean give them the answers, zero. If you mean asking them guiding questions that help them figure things out themselves, or explaining concepts in a new way which helps them grasp such concepts, a lot!
I meant the latter: guiding and explaining. My kids never wanted that. It wasn't on offer to me.
Why would you not do that?
Because my kids didn't want it. Also, I didn't want it when I was a teenager and I assumed my kids would be the same.
I was a very stubborn teenager, and I would never accept nor seek help from my parents on anything that resembled guidance. I never asked them for relationship advice, help with homework, or anything else. I acted as if I could live my entire life without so much as a minute of their input. I'm in my late twenties and I regret behaving that way. If your children never wanted it, then you didn't do anything wrong. Some kids are just like that.
I left home and school at 16, so even after school, I never asked for advice. I always assumed that my kids were like me and wouldn't want help either. Now that they are older, they say I did OK!
If they are not asking for help, why are you worrying about whether they need help?
That being said, isn't your role as parent to make sure your kids get what they need, even if they don't recognize that need themselves? Moreover, based on my experience as a former teacher, lots of kids will readily accept help if offered, but would never ask for help on their own.
> If they are not asking for help, why are you worrying about whether they need help?
Because sometimes kids are stupid, and don't ask for help that they'd benefit from. Sometimes they don't know they need help, sometimes they're too stubborn to ask for help, sometimes they're embarrassed to ask for help, and so forth.
As a parent you have to balance the need to let a kid make their own mistakes with the need to intervene. This is a tricky skill and many parents get it wrong one way or the other.
We all make lots of mistakes as we grow up. When we get a little older we think "oh geez, if only someone had advised me to do X and Y and not done Z then things would have turned out so much better". Then we get even older and we have kids, and we think "Finally, a chance to do it over! With the benefit of my advice, my kid will have all the opportunities that I missed out on". And then you try to give them your brilliant advice, and they don't listen and they do something stupid again.
The answer depends upon what you think school actually does, and upon how much you care about society outside of your own family.
To the extent that school is signaling + babysitting, the tiger mom parenting style is privately beneficial but socially corrosive. Signals are valuable for society, but the optimal setup is one in which everyone collectively agrees to devote the same, relatively low resources towards the signaling process.
Once some tiger moms defect from this social arrangement and devote much larger resources towards the signaling process, two outcomes are possible:
The first is that the rest of the parents continue to act as they did before. In this scenario, the children of the tiger moms benefit (albeit with great effort) by gaining more access to better academic and career trajectories, while the children of the other parents suffer by losing some access to these trajectories. Society also suffers in general, both because the tiger families devote huge resources to zero-sum status games and because the quality of the signals are weakened for everyone.
The second is that the rest of the parents copy the tiger parents and become tiger parents themselves. In this scenario, the quality of the signal remains as high as it was originally, but society suffers due to the huge resources that are now being devoted by all families to zero-sum status games.
The latter is definitely what happened in many Asian countries, Singapore being one example.
I can say I've never understood why helping with homework has been seen as a "basic" task in childrearing. Like "when you have children you'll need to bathe them, feed them, buy them things, later help them with homework" I'm always hold om? Why the homework part? I never particularly wanted help from my parents, and that's not because nothing was difficult for me. I actually struggled quite a bit at times. But teaching is a skill, a normal parent shouldn't be expected to have the skills to teach an eight year old with severe math anxiety algebra. That is a difficult, specialised skill. Much rather have my parents work a little more to afford a little tutoring.
Helping kids with stuff they don't even struggle with is something I don't remotely understand. Do something else for bonding.
That's about where I am, too.
My parents “helped” me with homework in the sense that they would yell at me if I didn’t do it.
I spent a lot of time explaining various ideas from economics, statistics, game theory, evolution, etc., to my kids, but in conversation and daily life, very seldom in trying to teach classroom-like material. I also spent a lot of time with my oldest kid teaching him math when he got stuck (leading me to need to go back and relearn a bunch of calculus, since I hadn't used it in 30 years), and have spent some time on math with my youngest. I also spend a lot of time chatting with my youngest in Spanish, since she's trying to get better at it. And discussing whatever they're interested in with them--current affairs and how they link to historical stuff, what it means when the news talks about the Fed raising interest rates, etc. (One aside: the AP US Government class was a huge game-changer in how the two who took it discussed current affairs, since they ended up actually knowing the relevance of a filibuster-proof majority or what the Secretary of State did.)
I think it's basically just a function of how smart your kids are. This board is mostly filled with former smart kids who didn't struggle doing their homework (but might have struggled with finding the discipline to actually do it). But less-smart kids can struggle with school-level work and need help, and I imagine it's pretty tempting for parents to try and help them out.
I always took the "homework" part as synecdoche for knowledge transmission more broadly, i.e., don't trust that regular school simply does that, and instead make sure your kids actually learn the things you want/expect them to.
If they're teenagers, just checking that they're doing the homework and finding out if there is anything they are struggling with. You may be able to help them where they're struggling or you may find you probably need specialised help.
That sounds about right. I worry that we didn't do that enough.
Teenagers? They should be insisting, perhaps ensuring, that they *do* the homework.
Grade school? They should be assisting them at the homework. This doesn't mean doing it, but perhaps something like making flash cards and running them through them. (The big part here is them showing it's something they care about, but being present and working with their kids. So, no, a computer program won't work.)
I worked at a place that gave a (small) scholarship to a boy and girl each year, among those who had been volunteers with us. I recall it was the day that the not-arduous applications were due, and an Indian mom with whom I was familiar came in to pick up the application and ask about the (short) essay requirements or something. This was like 10 o'clock in the morning. She returned in an hour with it all filled out. As it was a schoolday it was clear she had written the essay or personal remarks or whatever it was. Her son did indeed win the boy's prize.
I admit she had a regal quality, plumply swishing around in her showy sari, that I didn't care for. After that, I might have started to admire her, for her dedication to her son's interests - or - I might have thought, this is an unseemly cultural difference to me, this brazen dishonesty.
I should add that I also have a favorable prejudice toward Indian people, or I should say the educated Indians* who come to America: they are so much more interesting to talk to, than my fellow kids.
*They are not the only ones I know, as I stay at motels occasionally, and prefer old vintage ones, of which Indians are sometimes proprietors.
When we lived in California, my kids had plenty of Indian friends and we became friends with their parents as a result. All lovely people.
I agree, of that generally "tech" sample - and yet - were it Indians who were on track to inherit America, I think it is indisputable that it could not then in any sense remain America. It must be Indian. That is not unique, though - equally true whether the future is Mexican or Chinese.
I could bet on all three, and America isn't going to be gone.
If we can have a Day Without a Mexican, I guess we can picture a Century Without an Anglo: where in the world, do very separate cultural groups: as fecund as Mexicans, as competent and hardworking as Chinese, and as civilizationally dominant as Indians - live together in a harmony without one being subservient to another?
Hopefully the future will be AI and save us all.
Shocking!
How so? ;-)
They should be making sure the children do it, and nothing else. They need to stand on their feet, not yours. Unless you want to do their workwork for them in the future, as well.
A really great school, and the students get a lot of help and individual feedback from the teachers, and parent assistance is less important. I was fortunate to have my mother correct all my writing with me for years and this was extremely helpful. At my humble public school no teacher was going to put this kind of effort into my development.
Of course you should help your kids. Your wife and friend are ridiculous. Most of a child's knowledge will come from their parents and if not they will look for it on the internet (you don't want this unless it's controlled). It doesn't mean you do things for them, but rather that you be there for them and instruct them while they are learning to take care of themselves.
However, that doesn't mean you should feel guilty. You can still help them even now, maybe they insist they don't want it but that's when you enter grandpa mode and insist anyway.
*Should* is carrying a lot of water but, I would have greatly benefited from my parent (single parent household) helping me with college + networking. I was really clueless in this whole process and did not care enough to learn myself. As for homework and academic help maybe he could have pushed me to go to a better school? But I pushed back against the idea at the time so idk, should he have done it against my will?
I'm sorry you did not get help when you needed it. It's hard being a parent, though.
I think there are two competing goals that may be in play.
1) Your kids become self-actualized and capable people, who can determine their own needs and pursue them.
2) Your kids reach the highest possible heights in the system as exists.
If you help your kids more, they will likely reach higher heights, but be less personally capable of navigating life and dealing with the unexpected. It's a tradeoff. Too much help from parents and they learn to depend on you instead of learning how to be independent. Too little help and they fail at minor tasks that slow down their growth.
I regret how much my mom helped me with college applications, taxes, and other paperwork when I was 16-19. It was work well within my ability level and would have taught me better how to do things myself, which I also had to learn later anyway. If I were applying to an Ivy school, it may have been more important or even necessary to get that help from her. Hopefully the higher overall requirements would have required me to work harder as well and still grow as a person even with the help.
I think that's right. There's a balance between helping them too much and allowing them to be self-sufficient.
I find it odd that many LLMs can still be really incompetent at task execution. My 2 year old already has more agentic behavior, in some areas, than SOTA LLMs can handle. What explains this? Instinct?
The LLM’s aren’t fine-tuned for success at task execution. Programming will probably be the first.
> Programming will probably be the first.
I think, that programmers often believe that non-programming jobs will be automated first, while non-programmers believe, that programming will be automated first.
Are you a programmer?
I think that the argument is that automating programming has the better expected return because it builds on itself. It might well be easier to automate accounting or medical diagnosis, but doing so doesn’t make it easier to automate harder accounting or harder medical diagnosis.
FWIW, I am a programmer.
I feel like programming is being automated faster than other fields because it's close to the knowledge and capability of the people building AI. To automate accounting, you would need a group of very capable accountants as well as a group of very capable programmers. To automate programming, you just need the programmers. It may even be better than that, as programmers are both the client and production, while accounting would be the client and would have to communicate to the programmers where some information is likely lost between them.
I'm a programmer, I believe that AI will automate programming jobs first, as does basically everyone I know who works in AI at all (all programmers)
I'm a programmer and I think programming *tasks* will be automated first, but I also think that programming, and writing more generally, are especially amenable to our current efforts and that if anything people over-estimate AI progress specifically because they're disproportionately good at basic coding, and a lot of the people paying attention are programmers.
I think automating away *my job* will be pretty hard, because any decently-experienced engineer codes as an artifact of their real job, "thinking carefully about stuff".
The new paradigm is like a weird inversion of the horse->automobile transition. In that case, we went from dealing with live animals, with all the variation and unpredictability that requires, to machines that while they still had their idiosyncrasies and some unique characteristics, they were orders of magnitude more predictable and comprehensible than what went before.
This time around, we're going the opposite direction: programmers are used to -- maybe even were attracted to in their youth -- the absolute, discrete, knowable & repeatable behavior of computers (as they were in the before times) to a paradigm that is more akin to training an animal than to programming a Turing machine.
I don't think that all the programming jobs will go away -- not everyone knows how to properly train a horse -- but perhaps the skills and attitude necessary to succeed in this new regime will be quite different. [programmer!]
LLMs are designed to predict tokens, not to do things.
With a bit of clever jiggery-pokery we've managed to turn token-predictors into conversationalists, which still isn't doing things.
If you want a machine to do things, then you just have to train it to do things.
"just" is doing a lot of work there, considering how much data we needed to use to train them to write coherently.
Actually, imx getting an LLM to exhibit agency and goal-seeking behavior starts with the prompt; fine-tuning improves things but it's not the spark that lights the fire.
The key is to turn prediction into action by giving the LLM a persona, querying "What would your character do next?", then feeding the answer back in the next prompt as an action taken.
I've used this simple principle to turn raw GPT3-style auto-complete models into reasonably coherent chatbots. I've also then used chatty models to exhibit agency by controlling a robot with a simple instruction scheme.
The key insight here (and I'm not claiming originality, this principle is well known) is that actions are simply predictions applied recursively to the agent making the prediction.
"What would an agent like me do next?" is subtly modified to "What should I do next?", which then becomes "Oh yes I did!".
It's really quite amazing actually.
In my opinion, there is a situation about the ongoing risks of long COVID similar to the past one about tobacco risks analyzed by Naomi Oreskes, in that many scientists in good faith minimize them.
However, experts with publications on long COVID on top Journals like David Putrino, explicitly deny that long COVID is psychosomatic
https://www.youtube.com/shorts/xALv93I6xLo
(Well-known physics popularizer Dianna Cowern, known on Youtube as Physics Girl, contracted a severely disabling form of long COVID even if she was under 35 and vaccinated against COVID).
Among well-known AI scientists, at least Ian Goodfellow and Michael Osborne suffer from it.
https://en.wikipedia.org/wiki/List_of_people_with_long_COVID
I apologize if I quote myself:
“With regard to the underestimation of COVID, allow me to point out what I think is the largest elephant in the room of public discourse, in politics, mainstream media, and incredibly even in the scientific community outside of long COVID specialists: the ongoing risks of contracting long COVID even for vaccinated people (only partially protected against it) and children (who can suffer permanent consequences like diabetes too). To grasp the extent of the underestimation, I think it is sufficient to read the following recent review article from Nature Medicine: https://www.nature.com/articles/s41591-024-03173-6
(from its abstract: “Long COVID represents the constellation of post-acute and long-term health effects caused by SARS-CoV-2 infection; it is a complex, multisystem disorder that can affect nearly every organ system and can be severely disabling. The cumulative global incidence of long COVID is around 400 million individuals, which is estimated to have an annual economic impact of approximately $1 trillion—equivalent to about 1% of the global economy.”)
or even the article for the general public by one of the authors of the review, Dr. Ziyad Al-Aly ( https://scholar.google.com/citations?user=DtuRVcUAAAAJ&hl=en )
https://theconversation.com/long-covid-puzzle-pieces-are-falling-into-place-the-picture-is-unsettling-233759
("A new study that my colleagues and I published in the New England Journal of Medicine on July 17, 2024, shows that the risk of long COVID declined over the course of the pandemic. In 2020, when the ancestral strain of SARS-CoV-2 was dominant and vaccines were not available, about 10.4% of adults who got COVID-19 developed long COVID. By early 2022, when the omicron family of variants predominated, that rate [of long COVID for infected people] declined to 7.7% among unvaccinated adults and 3.5% of vaccinated adults. In other words, unvaccinated people were more than twice as likely to develop long COVID.
While researchers like me do not yet have concrete numbers for the current rate in mid-2024 due to the time it takes for long COVID cases to be reflected in the data, the flow of new patients into long COVID clinics has been on par with 2022").
I think that two figures, the first from Nature Reviews Microbiology https://www.nature.com/articles/s41579-022-00846-2/figures/1 and the second about children from Scientific Reports https://www.nature.com/articles/s41598-022-13495-5/figures/2 are illuminating too.
These represent just a small sample of the vast scientific literature on long COVID which paints a troubling picture of the problem.
Examples of recent papers on long COVID in children from top journals are the following:
https://jamanetwork.com/journals/jama/fullarticle/2822770
https://publications.aap.org/pediatrics/article/153/3/e2023062570/196606/Postacute-Sequelae-of-SARS-CoV-2-in-Children?autologincheck=redirected
It is not coincidental, given their knowledge of the syndrome, that the already mentioned Dr. Al-Aly and another author of the Nature Medicine review article, Dr. Akiko Iwasaki ( https://medicine.yale.edu/profile/akiko-iwasaki/ ) still wear a mask virtually everywhere.
https://youtube.com/watch?v=M4Nzhpeu2Y8&pp=ygUMeml5YWQgYWwtYWx5
https://youtube.com/watch?v=rMt6ZV-hHSE&t=3681s
“
There a huge amount of evidence that long COVID is significantly (if not entirely psychosomatic). I’m not saying it is, but that’s the way the best studies on the subject have gone.
[Sorry, I edited this comment to insert an updated link] Can you give some examples of these studies, as opposed to the studies I mentioned in my previous post, or to these ones: https://www.panaccindex.info/p/what-covid-19-does-to-the-body-seventh , or even to the short video https://www.youtube.com/shorts/xALv93I6xLo with Dr. David Putrino's opinion ( https://scholar.google.com/citations?user=lOGoZZ4AAAAJ&hl=en )?
A simple google search would suffice. I don’t care enough to look for everything myself, I recall one of the biggest was French led.
I'm not even going to search. Give some evidence or I won't bother.
He's probably referring to this one:
French study long covid
https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2785832
Long Covid in Children
https://link.springer.com/article/10.1007/s00431-021-04345-z
German study long covid, replicates french
https://jamanetwork.com/journals/jama/fullarticle/2787741
https://twitter.com/andrew_croxford/status/1473408952436240392?fbclid=IwAR0ZhCEf-0B2Dz4wg-pDQVwg-ecPckUTacepPOAc9yERkaMQ75nIBAvlTF0
Here’s an n=1 reason for taking long covid seriously. Twenty-some years ago I had something that was the flu equivalent of “long covid.” I had a case of the flu, though not an unusually bad case, but just never fully recovered. Tried not to take it seriously, but got worried when I realized a few months had gone by and I did not feel a bit better. My symptoms were fatigue, exercise intolerance, hypersomnia, malaise, muscle aches all over my body, and several extremely sore joints. And I had pretty bad versions of all those things. For ex., during that era of my life I naturally woke up after about 8 hours of sleep, and could not sleep longer no matter how lazy I felt, or how reluctant to face the day. But after the flu there was no limit to how much I could sleep. I could sleep 12 hours at night, and be craving a nap 2 hours after I got up. And if I lay down when I had a nap craving, I could fall asleep and stay asleep til a person or an alarm clock woke me. My 3 sore joints were so sore that if I bumped one even fairly lightly I’d make a little involuntary noise of pain.
My internist tested me for everything either of us could think of over the course of the next few months and everything came back normal. I did wonder if I somehow had a low grade, permanent version of the flu, but that was only one of my hypotheses. I had never heard of post viral syndromes, and my internist did not say anything about that being a possibility. So I am confident that it was not fear or expectation of “permanent flu” that caused me to have the syndrome I had. There could have been some other psychological thing going on that led to me developing a psychosomatic illness, but that seems unlikely to me. By the time I developed that syndrome I had lived through a lot of difficult things, and while they affected me, my health stayed excellent right through them. Sometimes stresses and losses put me through periods of depression or of anxiety, or caused me to have insomnia — but I don’t seem to have the wiring to develop psychosomatic illnesses in response to stress. And there was certainly no secondary gain for me in feeling sick all the time. I was a working single mother, and kept working and mothering right through it because I had no choice.
My syndrome lasted 3 years, then faded away on its own, and never came back.
This sounds like chronic fatigue syndrome/fibromyalgia which is known to be triggered by viral illnesses in many cases. A friend of mine had it for years, thankfully he has recovered now. Glad you are doing better.
I’m not just doing better, I recovered completely and totally, and never had a recurrence. Have had the flu a couple times since then, but recovered normally.
It's nice to read this account, it gives me some hope. A direct colleague of mine has Long Covid for two years now, without really getting better. He sometimes has energy for doing stuff, but only for very short periods per day.
Being in academia and in Europe, his position at university was exceptionally protected out of compassion, but his contract will not be prolonged beyond this month. Your account implies that there is still hope that he eventually will get better even after two years.
I know one person who had long covid for at least 18 months, then recovered gradually over about the next year and is back to normal now. She definitely was nothing like the kind of person those who think LC is psychosomatic picture — anxious, low-energy, neurotic etc. She was a happy 30 year old who’d finished her phd the year before, then gotten married and started a job she loved and was doing well at. And she had lots of energy pre-covid — played tennis, really good skier, gourmet cook. She’s back to that now
Thank you Eremolalos! These cases are really encouraging!
Something similar happened to my wife a few years ago. She spent several months after a bout of the flu just occasionally running a fever and feeling ill and having her heart race, for no reason anyone could detect. It went away after a few months.
I think this just happens occasionally with viral infections (I guess your immune response doesn't get turned back off properly at the end of the infection or something), and that a bunch of long covid cases are just people with this going on.
Scott did an article on this a while ago.
I think it's fair to say that:
a) It would be shocking if some cases weren't real. If you had a severe case of Covid and wound up in ICU and nearly died then yeah, you're going to have some long term effects from that.
b) It would be shocking if some cases of Long Covid weren't psychosomatic. Any time there's a new disease, especially one with no test and vaguely defined symptoms you're going to have a lot of people with psychosomatic cases.
What's left in the middle? If you think that your knee hurts because of long covid rather than just getting old, does it really matter? It's not like knowing will help to guide your cure. If your aunt tells you that her knee hurts ever since she has covid then you should probably just smile sympathetically rather than argue with her, because what's the point?
In his "much more than you wanted to know" post on Long Covid, Scott came to the conclusion:
"While some symptoms may be psychosomatic, most are probably organic."
https://www.astralcodexten.com/p/long-covid-much-more-than-you-wanted
He walked that back at the end of this column two years later.
https://www.astralcodexten.com/p/replication-attempt-bisexuality-and
> https://jamanetwork.com/journals/jama/fullarticle/2822770
Regarding the JAMA paper, among study participants who reported having Covid they verified the infection by doing an antibody-positive (Ab+) test within 30 days of survey completion. Thanks to this test they identified a number of participants in the study who reported not having Covid, but were antibody-positive: "Overall, 64 school-age children and 781 adolescents enrolled as uninfected but were Ab+"
The results for that group ("uninfected" but Ab+) were:
"Among school-age children, 6 (9%) met the index threshold whereas 18 (28%) reported experiencing at least 1 prolonged symptom. Among adolescents, 29 (4%) met the index threshold and 175 (22%) reported at least 1 prolonged symptom."
The index here refers to: "A total index was calculated for each participant by summing the individual scores for each symptom reported. An optimal index threshold for identifying PASC was selected"
So, how do those percentages (9% and 28% for school-age and 4% and 22% for adolescents) compare to the percentages in (verified) uninfected and Covid infected participants?
"Overall, 152 infected (20%) and 6 uninfected (4%) school-age children and 445 infected (14%) and 44 uninfected (3%) adolescents met or exceeded this index threshold"
"Overall, 45% of infected (338/751) and 33% of uninfected (48/147) school-age children and 39% of infected (1219/3109) and 27% of uninfected (372/1369) adolescents reported having at least 1 prolonged symptom."
Except for the above index threshold in school-age children (9%), all measures are about the same or even lower for the "uninfected" but Ab+ than for actually uninfected participants. This is supposed to be the blind group (unless I've misinterpreted something). If the hypothesis is right and long Covid is real, shouldn't they have the same or comparable percentages as the infected group and higher percentages than the actually uninfected group?
To summarize, they found a higher long Covid incidence (as measured by the index of long Covid symptoms and having at least one prolonged symptom) in children and adolescents who reported a Covid infection vs those who reported no Covid infection, BUT they also found LOWER long Covid incidence in adolescents who had Covid infection but did not report any Covid infection history.
I haven't read the paper, but taking your summary at face value, this could explain that result:
If most covid symptoms are directly caused by the immune response, and if long covid itself is caused by the immune response gone haywire, then one would expect (a) those with asymptomatic covid cases to be much less likely to develop long covid, and (b) on top of that, be much less likely to have developed one or more of the long covid symptoms from some other infectious disease.
> one would expect (a) those with asymptomatic covid cases to be much less likely to develop long covid
Yes. But look at the numbers for the uninfected group compared with the control group.
At least one prolonged symptom was present in 27% of adolescents and 33% of school-age children. Yet only 22% and 28% of "uninfected" but Ab+ (control group) participants.
3% of adolescents and 4% of school-age children had an index higher than threshold. 4% and 9% in the control group.
If participants with symptomatic Covid were more likely to develop long Covid than asymptomatic ones, why does the uninfected (No Covid) group exhibit higher one-prolonged-symptom incidence than asymptomatic participants?
(Also, the difference in higher-than-threshold index between uninfected and control group (3.21% vs 3.71%) in adolescents is not significant, according to Grok).
Because of my point (b). Covid is not the only disease that appears to cause lingering symptoms, and presumably the control group had just as many other diseases in their history as the asymptomatic group.
When long covid started happening and got a name, I thought, "maybe now the medical community will have to start taking these other lingering symptoms syndromes (fibromyalgia, cfs) seriously".
Agreed. I think fibromyalgia and other lingering symptoms are real and probably triggered by infections.
I just don't think there is good evidence that Covid triggers these lingering symptoms any more than other viral infections.
There’s ample evidence that “long covid” exists, and long term negative consequences of viral infections have been known to exist for a long time.
It’s also significantly more prevalent in people with a previous diagnosis of mental illness, which isn’t the case with basically any other physical disease. At least when there isn’t a direct connection, as there might be between depression and obesity. Anecdotally I have a half sister who is bipolar. She would frequently claim this or that malady, right before she would ask family for money, as she can’t work because of her illness. It was always something, and after catching covid, it’s been long covid for the past few years.
I have no doubt she actually *feels* sick, but it also seems incredibly likely to me that it’s mostly psychosomatic. Otherwise, her other illnesses have been magically cured at about the same time a new, unknown, illness makes her pitiful and in need of help.
I think long covid is a bunch of different stuff living under the same label. Some is psychosomatic, some is some kind of post-viral syndrome happening to a lot of people at once, some is probably some rare bad thing involving clotting, etc.
I would love for Scott to do a deepdive into this. It seems pretty bad and mostly ignored from a statistical perspective and pretty horrifying if you know someone with a bad base personally.
Good news!
https://www.astralcodexten.com/p/long-covid-much-more-than-you-wanted
> Examples of recent papers on long COVID in children from top journals are the following
I've already commented on the JAMA published study, but I want to add that the second paper ("Postacute Sequelae of SARS-CoV-2 in Children") reviews several studies on long Covid. From the "Respiratory Manifestations" section (I did not check the other sections):
- "Post-COVID-19-associated morbidity in children, adolescents, and adults: A matched cohort study including more than 157,000 individuals with COVID-19 in Germany" (https://pubmed.ncbi.nlm.nih.gov/36355754/): Observational study, based on insurance data. "No insurance claim, no Covid".
- "Protracted respiratory findings in children post-SARS-CoV-2 infection" (https://pubmed.ncbi.nlm.nih.gov/34534416/): Observational. n = 29. No control group.
- "Four-Month Clinical Status of a Cohort of Patients After Hospitalization for COVID-19" (https://jamanetwork.com/journals/jama/fullarticle/2777787): no control group.
- "Long-term pulmonary sequelae in adolescents post-SARS-CoV-2 infection" (https://pubmed.ncbi.nlm.nih.gov/35775163/): n = 82. No control group.
If most cases are psychosomatic, which seems likely to me, then the campaign to scare people into compliance using threats of long COVID is an absolutely enormous pile of skulls. Intentional lies that caused needless, genuine suffering, for years at least, of millions of people.
Woah, I must say that I find this worldwide pretty insane. You have three pretty strong assertions here. The first is that most cases are psychosomatic. Fine, this may be true, we can discuss about this. Let's assume that.
But the second is that most people are also absolutely convinced that it's psychosomatic. This is the assumption that I find insane. Even if true, why do you believe that everything is absolutely certain about that? Why do you rule out the possibility that it's all psychosomatic, but that some people get it wrong, are honestly concerned, and that is the reason for their warnings?
And the third is that despite being convinced that it's psychosomatic, people start a campaign of intentional lies about it.
Let's take Scott. Scott has publicly declared that he believes only a small part of symptoms is psychosomatic, and most are organic. Do you believe that this is an intentional lie that Scott tells, and that Scott is part of the "campaign to scare people into compliance"? Do you really believe that this is more likely than Scott being honestly mistaken about it? Or perhaps even the third option, that Scott may be *correct* about it after digging into the topic and writing a blog post of epic size about it?
https://www.astralcodexten.com/p/long-covid-much-more-than-you-wanted
Sorry for not being nice, but I hope that it's at least important and true: you should really consider that other people hold different believes than yourself. Not all people who act against your own beliefs do that because they are jerks. Some may do that because they simply believe different things.
No worries about not being nice, I wasn't particularly nice in my post and your tone is totally fine to me anyway.
I don't see how I've made the second assertion at all. I don't think most people think it is psychosomatic. The intentional liars I'm referring to here are people like the CDC, who claim that 20% of infections result in long covid, but then in the fine print state that they are defining long covid as any symptom present six weeks after infection.
Scott estimates the prevalence of long covid at about .3% to 3%, in 2022. He says that these are probably not mostly psychosomatic. Now if survey respondents claim to have long covid at much higher rate, like 10%, it makes sense that the difference between Scott's estimate and the survey is psychosomatic.
I don’t know why everyone has selective memory on this topic but that is not representative of what Scott currently thinks about it. He updated his views about that ratio of organic to psychosomatic when long covid showed a correlation to bisexuality.
https://www.astralcodexten.com/p/replication-attempt-bisexuality-and
The correlation has apparently persisted to the point that some manager quoted in the Times joked earlier this month that he’d “believe in long covid when I meet a non-bisexual who has it”.
Long COVID is as real as vaccine related issues. As in it’s not real.
Most people got COVID. Some of them had random health problems. There’s not actually a causal link.
Most people got the vaccine. Some of them had random health problems. There’s not actually a causal link.
I'm sure PASC is a significant issue for a certain percentage of COVID sufferers. Some experts were predicting a mass disabling event, but that hasn't happened. The macro data doesn't show a strong signal that COVID has caused long-term disability in a significant portion of the population. For instance, the FRED tracks the number of people who report themselves as disabled. If we translate this graph into population percentage, the graph shows that the per capita rate of disability climbed from 0.08% to 0.09% in 2020. There was a steep dip in the number of people reporting themselves fully disabled after the pandemic began, but it climbed quickly. Now it's at 0.10%, but it's not far above where the trend line would be if we didn't have a pandemic.
https://fred.stlouisfed.org/series/LNU00074597
Social Security disability claims were falling before the pandemic, and they began to rise after the pandemic, but they're falling again. If we translate the disability applications per month into a per capita number, the rate of disability applications is lower than it was in 2005. We did have some sort of mass disabling event from 2008 to 2012, when field office applications rose 40% in 4 years. (I've not yet found out what caused this huge increase, BTW.) But those numbers had dropped to pre-2004 levels by Feb 2020 at the start of the pandemic in the US.
https://www.ssa.gov/oact/STATS/dibGraphs.html
As for anecdata, I only know of one person in my network of friends and acquaintances who had Long COVID symptoms that lasted more than 3 months. The exception has had them for over 2 years now.
I thought that you might be interested in the answers I obtained from Gemini (2.5 pro with deep research) and Chatgpt (o3 with deep research) to the following question of mine pertaining your comment (I am just sharing these answers, not endorsing them, at least for now, since I still haven’t had the time to check them):
“Would you comment on the following post I found on a blog? "I'm sure PASC is a significant issue for a certain percentage of COVID sufferers. [...]
https://fred.stlouisfed.org/series/LNU00074597 [...]
https://www.ssa.gov/oact/STATS/dibGraphs.html "
The attached files .csv and .xlsx both correspond to the graph from https://fred.stlouisfed.org/series/LNU00074597 .
I would like an objective analysis, comparing the post in particular with
https://www.nature.com/articles/s41591-024-03173-6 ,
whose PDF I am attaching. It seems to me that there might be a contrast between the two, and if so, I would like to know how it should be adjudicated.”
(I found the PDF of the Nature Medicine paper https://www.nature.com/articles/s41591-024-03173-6 only at
https://www.drknowhk.org/wp-content/uploads/2024/08/s41591-024-03173-6.pdf )
From the Gemini answer which can be found (with my question) at https://gemini.google.com/share/7b746b86a3c2 :
“Concluding Assessment:
[...]
Therefore, the evidence strongly supports the conclusion that Long COVID has contributed significantly to, and can be characterized as a key component of, a mass disabling event. The term "mass disabling event" is justified by the large scale of individuals experiencing new or worsened functional limitations and health impairments documented in scientific research and reflected in broader population disability surveys, even if not all these individuals are yet captured by, or have successfully navigated, formal disability benefit systems.”
From the Chatgpt answer, which can be found at https://chatgpt.com/s/dr_6835710bcfd48191b369c807d6e46885 (my question was the same, but it seems that Chatgpt shared links include only the Chatbot answers)
“In summary, the apparent contradiction stems from comparing macro aggregates to medical/survey evidence. The macro data (CPS counts, SSA claims) show at most a modest uptick, but the Nature Medicine review shows that long COVID has already affected tens of millions, with real economic impact. When viewed together, the evidence suggests that long COVID is a significant “mass disabling event” whose early warning signs may not fully appear in broad administrative series”
I’m convinced it’s impossible for Congress to ever significantly lower spending, for incentive reasons. The costs of lower spending are immediate and obvious. The costs of spending more are obscure and less obviously attributable to that overspending. The end result of this is that we should expect the dollar to go into a death spiral, as investors will look elsewhere rather than lending money on a 30 year term to a government that can’t ever bring itself to spend less than it brings in.
Can someone try to talk me out of this? I’m open minded on it and would like to actually be challenged.
Why is the only option to spend less? Why not tax more? After all, the current budget proposal that passed the House increases the deficit because of tax cuts, not spending increases.
The costs of increased taxes are immediate and obvious. The costs of reducing taxes are more obscure and less obviously attributable to overspending.
How are they immediate an obvious? Top tax rate in the early 1960s was 91%, and it kicked in for people who earned about $2 million in today's dollars. But the economy boomed during the high-tax '50s and '60s.
But 50s and 60s tax policy didnt raise more money than today - if anything a bit less.
https://fred.stlouisfed.org/series/FYFRGDA188S
If you increased taxes in a way that actually increased taxes, you might see an economic effect.
The type of tax increase that would be required to fund our current level of spending would be large and broad based (as in, across the entirety of the income spectrum), and, just like spending cuts, would _also_ be felt immediately by everyone. The exact same incentives exist.
That isn't the point. If OP is truly concerned about deficits, they should not be talking only about reducing spending.
I think you were not being charitable enough and the tone of the comment, alongside the explicit call for having their mind changed, warranted steel-manning the position, especially when your simple answer was prey to the exact same dynamic they were pointing out. Even if they "truly cared about deficits" (which I disagree with your suggestion that they don't) enough to also consider tax increases, they still need to figure out the dynamic they are asking about (bad incentives) because they apply equally to both sides of the equation.
"I’m convinced it’s impossible for Congress to ever significantly lower deficits, for incentive reasons. The costs of lower deficits are immediate and obvious. The costs of increased deficits are obscure and less obviously attributable to those deficits. The end result of this is that we should expect the dollar to go into a death spiral, as investors will look elsewhere rather than lending money on a 30 year term to a government that can’t ever bring itself to spend less than it brings in.
Can someone try to talk me out of this? I’m open minded on it and would like to actually be challenged."
Libertarians (and paleo-conservatives) have a talking point about never giving the government access to money; “starving the beast”. Tyler Cowen had a recent blog post about this, arguing that people in favor of tax increases probably aren’t thinking clearly about a remittance tax, but he’s against it for libertarian reasons, not efficacy/distributional reasons.
`gdanning` appears to me to be (pre-emptively) reacting to the (unstated) idea that the most important thing is to avoid spending more, because government never does anything worthwhile with tax dollars.
Regarding `apxhard`’s original question:
Historically Democrats have (several times) *actually reduced the next-10-years deficit*. Sometimes through spending cuts, sometimes through raising taxes.
As far as I can tell Republicans haven’t done this in the last 50 years, although Bush Sr. came close and got unlucky with a recession.
The accounting for (real, not imagined) economic growth due to lower taxes on capital is often difficult to measure or find counterfactuals for, so it’s not always clear how much long term (i.e. greater than 10 years out) benefits are attributable to Republicans.
So I would say there is an existence proof for lowering spending under Democrats, not Republicans, and so I echo `gdanning`’s suspicion that `apxhard` has an axe to grind when stating something can’t be done. But I agree that steel-manning what they actually asked is better/more charitable.
William Niskanen argued that "starve the beast" was backwards: taxpayers treat the government like Santa Claus if they don't have to pay for spending. My view is that Ricardian Equivalence is false because people discount taxes paid by later generations.
This usually gets analyzed by just looking at Presidents, but Congress also has a say in the budget. In recent decades, Republican Congress + Democratic President seems to have been the best combination for deficit reduction.
Hasn't Argentina balanced taxes with spending?
It wasn't pretty (and came after a few hyperinflations), but it did happen.
Maybe your position is more accurately described as: "Congress won't get deficits under control until after AT LEAST one huge financial catastrophe involving hyperinflation" ?
But I'll point out another "fix": Default on the debt rather than run hyperinflation. Defaulting hurts a *much* smaller percentage of the population than hyperinflation and can be spun as only impacting the rich. Another variation is to not default, but change taxes on gov't bond interest to be some large percentage (say, 50%).
ANOTHER (more politically charged) fix is to helpfully roll everyone's 401(k) and/or IRA into Social Security. We aren't *confiscating* your retirement funds, we are just moving them into Social Security! The roll-in could even be at some reasonable rate because the point is for the gov't to get access to the savings NOW while only needing to deal with paying out MUCH LATER.
There are lots of ways to get things to line up better financially. They are just painful. But high inflation is painful, tool
I thought Milei was drastically reducing spending rather than increasing taxes.
This is completely beside the point. Substitute ‘net spending’ for spending and it’s the same issue. You just seem to want to argue for tax increases; fine, but a different topic really.
No, the point is that "spending" and "net spending" are two different things. The "net" literally includes levels of taxation. That's what makes it "net."
The same logic applies - tax increases are immediately and obviously felt. The benefits of a balanced budget are not felt by anyone in the present aside from a small group of weirdos, most of whom have decided to just buy bitcoin and thus now cheer for fiscal irresponsibility because they profit from it.
Congress is as incentivized to avoid major tax increases as it is to avoid major spending cuts, and major tax increases will diminish economic activity overall. Which, first, reduces the tax base so your tax increases don't get you as much extra revenue as you would naively expect, and second result in the sort of economic hardship that much of your government spending is meant to address so you find your non-discretionary spending goes up.
Spending cuts cause economic hardship as well, but they usually cause it among people who weren't paying a lot of taxes so it ends there. Tax increases, to be meaningful, have to come at the expense of the more productive elements of a nation. So if you're concerned about a fiscal apocalypse (and you should be), and you have finite political capital to spend, you get more bang for your buck going after spending cuts, and raising taxes only to the extent necessary to get the opposition to accept the spending cuts.
At least if you're taxing income, or "value added" or anything like that. Some sorts of property or Pigouvian taxes might not have this effect or at least not to the same degree.
1. Spending cuts also reduce econo ic activity and hence tax revenue. And in fact, since the marginal propensity to consume is higher among low income persons than higher income persons, the multiplier fot spending cuts is likely higher than for tax increases.
2. Where is your evidence that a small increase in the top rate has a negative effect on tax revenue? Or that it materially affects the willingness of "more productive elements of society" to engage in productive activity?
> Where is your evidence that a small increase in the top rate has a negative effect on tax revenue?
He didn't say it decreases tax revenue. Just that you don't get as much as you want.
You aren't going to close any budget holes with "small increases" in the top rate.
If you want to be serious about raising tax revenue, you need to be doing the tax regime that 37 of the 38 countries in the OECD do but that the US doesn't.
>You aren't going to close any budget holes with "small increases" in the top rate.
Four trillion dollars over ten years is not nothing. https://bipartisanpolicy.org/blog/the-new-cost-for-2025-tax-cut-extensions-4-trillion/
You've rolled a bunch of things that aren't tax increases on the top-rate into that 4 trillion number.
"If you want to be serious about raising tax revenue, you need to be doing the tax regime that 37 of the 38 countries in the OECD do but that the US doesn't."
Huh? USA is currently at rank 31/38 in government revenue as % of GDP
Increasing taxes isn't equivalent to cutting spending, it's like taking money out of the ATM https://www.thebigquestions.com/2011/11/15/econ-101-for-the-supercommittee/
> After all, the current budget proposal that passed the House increases the deficit because of tax cuts, not spending increases
My understanding (someone will correct me if I'm wrong) is that is jiggery-pokery. The budget does actually include some (not enough but some) spending cuts, and keeps tax rates as they are.
The Democrats are referring to "keeping tax cuts as they are" as "tax cuts" but that's not the actual effect.
Under current law, the 2017 tax cuts will expire this year. So, the proposal to extend the tax cuts is indeed a reduction in what rates will be if Congress does nothing, and so is fairly described as a tax cut.
Those tax cuts were originally budgeted for as temporary, so pretending that they aren't to blame for the deficit because technically tax rates haven't changed is just as much jiggery-pokery.
In real terms, government spending as a percent of GDP has decreased in the past, from about 1981 to 2000.
It hasn’t in about 25 years, but most of that time was in the era of zero interest rates, and when someone is lending you trillions of dollars at a rate less than inflation, it’s no surprise that spending increased.
https://fred.stlouisfed.org/series/FYONGDA188S
The government doesn’t have to lower spending, it just has to not increase spending while inflation + gdp growth outpaces the budget.
Yes this seems like the path that would be necessary. But can you help me reason through the incentives? The only people who voted against the recent budget were
- democrats who’d want to spend the same if not more elsewhere
- two republicans who want to spend less
From what I gather, they are all just following their political incentives, with the possible exception of the two republicans, and even that’s a maybe.
Can you help me understand the incentives that would need to hold for this situation to pass? Is it like AI+ inflation + Congress that somehow thinks their re-elections are more likely if spending doesn’t go up too fast?
Again, you are ignoring the parties' differences re taxation, specifically extending the 2017 reduction in the top rate.
Note that Congress has passed tax increases many times in the past. Eg https://www.investopedia.com/terms/t/tax-reform-act-of-1993.asp
I think these differences are largely rhetorical. Both sides want to reduce taxes on their own voters, and are happy to text the other side. The SALT deduction is a perfect example of this. 2017 lowered the top rate but capping salt deductions hit high earners in blue states hard. So it was a tax increase for some wealthy voters - high earners in blue states who could afford the expensive housing there.
Republican candidates talk about lowering spending, the same way Democrats talk about raising taxes on the rich: it’s rhetoric for their bases. Maybe some of them believe it. But given the chance, they’ve never voted it in. Is there a reason to think I’m wrong on this?
1, They clearly are not merely rhetorical. Certainly their positions on top marginal tax rates reflect decades-long policy differences .
2. The voters who pay the top marginal rates are not supporters of a single party. If anything they skewer a bit D. https://www.pewresearch.org/politics/2024/04/09/partisanship-by-family-income-home-ownership-union-membership-and-veteran-status/
Why didn’t democrats raise these taxes when they passed the “inflation reduction act?”
Am I wrong that they could have but didn’t?
In the new budget, base spending is about $60 Billion less than 2024, and additional base spending is going to be upwards of $100 Billion less.
They are extended the tax cuts, but these have been in place for almost a decade at this point. Eventually they have to just be looked at as “our current tax policy” rather than ongoing cuts, and under that light every time there’s a budget without an increase in taxes we are voting for tax cuts, which doesn’t seem like the right way to look at it.
The Laffer Curve also shows that we can’t forever increase taxes without decreasing revenues, and while we’re probably not hitting the maximum yet, we’re definitely in the ballpark. I think there’s a strong argument to be made that the level of taxes in Western Europe are so high that they are actually producing less revenue than if the tax rate was lower.
Tariffs will also reduce the deficit. With an effective tariff rate of ~2.5% the US raised ~$75 billion. The current effective tariff rate is now around ~17.5%. While we shouldn’t expect a 7x increase in tariff revenue, as tariffs discourage imports, we can expect a significant increase in tariff revenues, which will reduce the deficit. Possibly by ~$200 Billion or more. People make a huge fuss about tariffs, and how they’re welfare reducing (and when there’s retaliatory tariffs this is certainly true), but the exact same thing is true for all taxes. Tariffs are simply a different form of tax, and while there are arguments as to why income tax or capital gains is more efficient, it will raise more revenue for the government, which is the exact solution we need to fix the debt.
Can you help me understand the “base spending” thing?
I ultimately agree on tariffs and think they’ve been turned into Literally The Worst. I understand the trade arguments but think they are the least bad form of taxes because they align incentives with what government ought to do: secure the border. Income and capital gains tax incentivize the creation of a panopticon and reduce investment and legal labor, while encouraging illegal immigration and other crimes.
Base spending refers to the part of the federal budget that happens every year under normal circumstances. That includes things like Social Security, Medicare, and the ongoing military budget. It doesn’t include one-time expenses like emergency disaster relief, COVID checks, or a new war those are considered temporary and not part of the government’s “baseline.”
When people talk about whether the deficit is sustainable, the key isn’t just how big the number is. It’s whether the government’s debt is growing faster than the economy. As long as the economy is growing fast enough, the government can run a deficit without its debt becoming a bigger problem.
More specifically, the deficit is sustainable if the total debt grows more slowly than nominal GDP (which is regular GDP plus inflation). That’s because if the economy is getting bigger at a healthy pace, even a large debt can become more manageable over time—just like a growing salary makes a fixed mortgage easier to carry.
A government can safely run a total deficit equal to:
Nominal GDP growth × Total Debt
That’s because the inflation and growth in the economy help “shrink” the real weight of the debt.
A Simple Example:
The U.S. economy (GDP) is $30 trillion
The national debt is $35 trillion
Nominal GDP growth (real growth + inflation) is 4% per year
Then the “sustainable” total deficit is 0.04×35 trillion=$1.4 trillion
So in this scenario (which is basically what the economy was like in 2024), the government could run a $1.4 trillion deficit each year without making the debt burden worse relative to the size of the economy. The deficit in 2024 was ~$1.8 Trillion, so we're only about $400 Billion off the mark. The new bill isn't actually reducing the budget, since it has some provisions for one-time payments, but over the next 4 years we could see that decline. When people talk about how the "new" tax cuts will add $5 Trillion to the debt or whatever, what they're really saying is that the "Tax cuts that have existed since 2017" are being extended, and if they weren't this would raise an additional $5 Trillion over 10 years.
TBH I'm not unconvinced that tariffs will close most of the gap between our unsustainable deficit, and a sustainable deficit, especially if inflation picks up a bit.
The problem is that our interest payments will continue to skyrocket, as we take on new debt, and old debt from 10 and 15 year bonds comes due from the zero interest days of the 2010s, so it's actually a lot harder to reduce the deficit to where it needs to be than it would seem just from existing revenue and spending.
Right - it’s the feedback caused by the increasing interest payments that I think kills the game. If you could keep interest low forever I get how you can keep riding the deficit as long as growth makes up the difference.
> Eventually they have to just be looked at as “our current tax policy” rather than ongoing cuts,
And the way we consider then to be "current tax policy" is when Congress votes to make them permanent rather than set them to expire before the 10-year CBO limit.
Is this an arbitrary date? Yes. But it's the arbitrary point we've all been using for generations. Things count as temporary or permanent. You can't double-dip to count something as temporary for a while so you only have to do the accounting for temporary, and then without looking it transforms behind the scenes into permanent without ever having to do the accounting for making it permanent.
The tax cuts aren’t permanent, but choosing to renew existing policy is a lot different than changing the existing policy to something less. From the perspective of “I don’t think the incentives allow Congress to stop overspending” the current extension of the tax policy is a lot weaker evidence than if they were changing the permanent tax policy to be lower.
Of course they did make permanent changes in 2017, but the debt-to-gdp was still double digits back then, and the 10-year yield was only ~2%, meaning there was far less incentive to reduce spending or raise more revenue.
I think you are really missing a full consideration of tax efficiency here.
The reason people make a huge fuss about tariffs is that they are more welfare reducing than (most) other taxes, for a given amount of revenue, whether there are retaliatory tariffs or not. $200B could be raised with a 1% VAT rate.
Meanwhile, the Laffer curve is a fun thought experiment, but the takeaway is that you should think carefully about marginal rates and design your system to maximise the cash raised while minimising economic distortions, not as a generalised argument that taxes should never be raised. The main reason Western Europe can raise much higher tax revenue than the USA is not primarily that marginal income taxes are higher, it's that they levy more taxes on things that aren't personal income.
The point isn’t about efficacy, it’s about how feasible it is for congress to increase taxes, which it completely is, as tariffs are a large, few hundred billion dollar tax. Just because republicans extended the tax cuts, doesn’t mean the incentives make raising taxes impossible unless there’s a collapse of the dollar.
I’d say the problem with a VAT is that it’s regressive. The bottom 50% pay basically no income tax, and are still struggling. A hike on the income tax would basically only affect the top 50% of earners, while a VAT tax would hit heaviest those who spend the most on consumption, which are the lower earners.
I was contrasing VAT with the tariffs. Tariffs are even more regressive than VAT.
I wish we could abolish the words "progressive" and "regressive" in the context of tax rates.
I think that a situation where a minority of people pay the overwhelming majority of tax is bad. Of course I think that because I'm in that minority, but I also think it's bad anyway.
The problem is that the current debt crisis is so bad that even a miracle economy would barely help. And the current goverment is hard at work crashing the economy and stoking inflation as well.
Democrats generally want to reduce defense spending. Does that have immediate, obvious consequences in the sense you mean?
The most direct consequences I can think of are fewer people enlisting and defense contractors diversifying.
The eventual ground invasion of the US would happen much later...
“These companies will get less money than they were getting” is still an immediate obvious consequence with economic ramifications. I had the impression Congress kept funding weapons projects even after the DoD said “we don’t need it” because every defense contractor is an employer. People affected by the cuts know directly, people affected by overspending only understand in aggregate.
I thought it was because every defense contractor is a major campaign donor and employer of lobbyists.
You imply that constituents don't benefit. That is probably wrong.
There are probably some defense spending in all 50 states. Each Congressman is in favor of cutting all the rest weekly, but strongly defending the spending in its own state.
Other countries have managed to lower spending/deficits, despite similar incentives for their legislators. https://en.wikipedia.org/wiki/Austerity#Examples_of_austerity
Based on cases in different countries, it is certainly possible. E.g., Sweden managed to decrease government spending to GDP from 67.93% (1993) to below 50%:
https://www.statista.com/statistics/375656/ratio-of-government-expenditure-to-gross-domestic-product-gdp-in-sweden/
https://www.reforminstitutet.se/twenty-five-years-of-swedish-reforms/
Some congressmen said, (paraphrasing), "There are two things that can change a congressmen's 'opinion'. Money and votes (the voice of the electorate.) with votes being the stronger driver, but often divided. and so money. If we the voters all decide that balancing the budget is important to us, then congress follows along, or gets voted out.
If you can get sustained bi-partisanship - which you often can in more functional democracies - that diffuses the issue.
Example: In the local economic crisis in the mid 1990's Sweden, all the responsible parties agreed on a budget framework of a 2% average budget surplus over an economic cycle (end of recession to end of recession). This was then maintained for three decades, driving the debt down from 65% to 35% of GDP, until it was, again jointly, changed last year to merely balanced budget from then on, as there's nothing to be gained by an even lower national debt.
You don't even need sustained bipartisanship, you just need a political system which regularly allows the party in power to do things without the approval of the other party.
In the US, if you abolished the Senate and most of the power of the Presidency then it would be easy; when the Republicans got a majority they'd implement their policies, and when the Democrats got a majority they'd implement their own.
Since there would be clearly one party in power at any time, it would be clear whom to blame for any problems with the budget.
Both parties would still have a strong reason to buy votes with the national debt.
It's pretty clear that voters don't care about the national debt, or at least that they don't let it determine how they vote. This is why countries can rack up giant debts without the politicians getting kicked out.
In parliamentary systems, things work exactly as you say, yet this doesn't seem to help much.
Which countries with parliamentary systems have deficits which are as large and as persistent as the US's? (As a fraction of whatever, not in absolute terms).
Among comparable countries with parliaments I looked up a few deficits; accounting is tricky so I just used the first numbers I came across, then converted to USD:
UK: $100 billion ($1400 per capita)
Canada: $44 billion ($1059 per capita)
Australia: $27 billion ($992 per capita)
New Zealand: $6 billion ($1129 per capita)
US: $1.83 trillion ($5380 per capita)
*Obviously*, you should use debt to GDP, not debt in absolute numbers. It's what measures the weight of the debt.
When you do that, Japan (with by *far* the highest debt to GDP among any Western countries), Greece (famously), Italy, and some others, all exceed the U.S. in debt load.
Not every country in this graphic has a parliamentary system, but many do:
https://upload.wikimedia.org/wikipedia/commons/thumb/3/3e/General_government_debt_in_OECD.svg/1920px-General_government_debt_in_OECD.svg.png
My mostly-fantasy approach to addressing this would be to reform the budget process (possibly via tweaking the rules for reconciliation votes in the Senate) thusly:
1. The "budget" is a stack-ranked prioritization of all authorized (but not yet appropriated) spending line items
a. This might be compilable from each individual Congress person's own such stack ranking via recursive Condorcet methods; I haven't checked the algorithm.
2. The total appropriated spending is capped at the average of the previous year's actual spending and actual revenue.
3. Working down the list from #1, each item is appropriated until the next would exceed the value from #2.
a. Lower-ranked items could still be pulled in, so long as they wouldn't cause the total to exceed the cap.
4. Taxes are set separately.
This would, in the long run, tend towards balanced budgets without any hard stop, as well as preventing proactive spending of speculative future revenue.
I think hopes of reducing the debt died back in 2010 with the Simpson Bowles commission. It's really hard to see how we ever get out of this mess. (I assume by "spending", you mean both lowering spending and raising taxes of course).
You imply there is some need to curb spending. There isn't. The Federal GOV (Through the Federal Reserve) can print money. It has infinite money. There aren't infinite resources though, so while the Federal Gov can't run out of money, it definitely can cause lots of inflation.
That is a different issue, though. Tax cuts to rich people don't encourage much consumption, they encourage private investment, usually. Tax cut to the not-rich or social spending increase consumption.
So the only reason to cut spending if it has high overhead (not efficient), or to reduce inflation. The thing is, it is somewhat self correcting. Inflation will lower the *real* amount of spending, so as long as spending don't go up with inflation in the same rate, the *real* spending will go down.
This is mostly what Millei did in Argentina. Because the Inflation was so high, not increasing the spending amounted to Real value lowering, at a pretty fast rate.
Clinton (D) worked with Congress (R) to cut spending on entitlements and raise taxes at the same time. The result was a balanced budget that was generating ~ 100B surplus each year when Bush II took office. The national debt would have been paid off in a decade, but Bush II was determined to cut taxes, so we ended up in deficit territory again.
Obama (D) tried to work out something similar with Congress (R) but by this point partisanship meant that any compromise was impossible. There were many Rs in the House especially who were determined to refuse Obama any sort of success at anything, country be damned.
So I guess when opposing parties compromise it is possible to lower spending, and your question should be reframed: Is it possible for the parties to compromise?
Looking forward to see what people reviewed. Hoping for some creative entries
https://danielgordis.substack.com/p/i-suddenly-realized-i-didnt-know-280
Very impressive peace activism by Palestinians and Israelis. It takes telling and listening to stories from both sides, which is emotionally wracking. Also, the effects of 10/7 and the attack on Gaza-- fewer people are willing to do the work, but it's picking up again.
Also, there's one advantage to extreme prejudice. It's brittle. If people have utterly dehumanized those on the other side, they can be surprised by finding that the other side is humans. There are no illusions about not being prejudiced because of believing in "a few good ones" or because "I have friends".
Both peoples have to learn to live together. Or not. But more outreach, more understanding, more compassion, will result in a very slightly less bloody future, very slightly earlier.
War is hell. Hopefully one day soon Hamas will be disarmed by force or treaty, and hopefully one day after that violent elements of the west bank settlers will be also.
I don't think you had time to watch the video unless you happen to have watched it already.
One of the points is that both sides have a dream of waking up one morning and finding that the other side has disappeared, and it is necessary to give up that dream. Israelis and Palestinians are both legitimately present on that land.
I agree with you completely
I think it is entirely possible for a majority of both Israelis and Palestinians to prefer peace to conflict, and yet for there to be no stable peace available.
The problem is that Palestinians in Gaza have no no-terror state capacity, and the PA in the west bank doesn't have enough of it to stay stable without idf support.
It's easy to find decent pro-peace Palestinians if you look. I'll avoid opening on whether they're 20% or 80% of the population (really hard to tell), but whichever it is, the group who does want total war retains the ability to launch it unilaterally and the others don't have the ability to stop them.
(This is why I believe Israeli policy should focus more on deconflicting the west bank, including pulling out settlements, and building up technocratic Palestinian state capacity in both the WB and Gaza. But while these are improvements the underlying problem remains pretty intractible).
The fundamental issue is that Palestinians have no ability to credibly commit to peace. Partly because of the number of times they've blown up deals in the past, partly because their political structure is dominated by violent organizations who gain money and prestige from fighting, and partly because they lack the state capacity to prevent attacks even if they cared to. The left isn't wrong this is an issue. But the right isn't wrong that such state capacity could be, and has been, simply used to attack Israel more instead of moving toward peace.
Secondarily, the Israeli public does not broadly trust Palestinians at this point and the pro-peace faction in Israeli was destroyed by trying to make peace and getting an intifada in reply. But even if you don't accept that it's the Palestinians' job to regain their trust, that's not the fundamental issue because you can change Israeli politics in a way you can't with Palestinian. Also, simply pragmatically, the Palestinians are the ones whose position is worsening so they need to do something while Israel can simply run out the clock.
How does running out the clock work? What changes in 5? 15? 50 years?
The Jewish population and Israeli economy continue to grow such that the relative power of the Palestinians declines. Plus Israel increasingly is able to cut off support to the Palestinians and normalize with their neighbors (as is already happening).
Maybe the economy grows but as far as population, it's less determined. Gazas tfr is 3.3, down from 8! In 1991. There's a lot of people there.
Also, international support and normalizing is uncertain, especially as they turn international support against them. I feel like a Hague criminal prosecution is likely, although uncertain.
Fair enough. But I think a large number of people in an increasingly stagnant economic base is not likely to alter the balance of power. If anything it might make it even more favorable to Israel.
As to international support: Israel has gotten more and more diplomatic recognition, not less, and the region is continuing to move in that direction. The idea that it's going to reverse is unlikely.
And a successful Hague prosecution is also unlikely. Firstly, because Russia is doing far worse and so has every incentive to break the institution (and kind of broke it before Israel's war even started). Secondly, most of the charges activists want against Israel would not stick in a fair court which means the incentive is to charge them with big crimes that will not stick but get flashy headlines.
Presumably, they eventually kill or expel the entire population of Gaza.
Note: If Erusian intended some other interpretation of their words, please clarify, since like Librarian, I'm not sure what else could be meant.
Assuming this is a good faith (which is a concession to a rather uncharitable and unjustifiable interpretation): You can see throughout the thread my point is that Palestinian national power is decreasing and Israeli national power is increasing, that Israel's diplomatic and relative military and economic position is getting stronger, and so on. So over time the Palestinian position is getting worse and at some point that will mean Palestinians cannot effectively resist. Thus the Palestinians are the ones who need a deal before all their leverage is gone.
There will always be an incredible amount of responsibility on Israel's shoulders for not just what they have done to make things worse, but also what they could do to make the situation better (cracking down on WB israeli terrorism would be an easy but impactful difference). Higher leverage does mean they can run out the clock but it also means they can be more agentic, and making the choice to do the former will in the long run make them more vulnerable to attack as their neighbours continue to hate them and they lose allies.
This is a common mode of wishful thinking among Palestinian supporters. However, the reality is that Israel has made steady diplomatic gains since taking a hardline against the Palestinians. And may be about to make further ones: if Saudi Arabia recognizes them then the only major country in the region not to recognize Israel will be Iran.
As to whether Israel bears more responsibility: we could go back and forth over the history but I don't think that's a particularly useful exercise. While on balance I think the Palestinian version of history is more distorted than the Israeli, neither is innocent. But you don't get to impose peace terms when you're a party so weak the other side can invade and capture all of your territory at its leisure. That's simply hard power reality.
They almost got recognised by Saudi, that's about the only significant diplomatic gain they have made in the last 10 years (and it took the US to help them do it).
The worry for these Arab countries is not Palestinians, they don't care about that and I never implied they did.
They've been recognized by five additional countries in the past five years and a sixth is likely on the way. If Saudi Arabia comes over then Iran will be the only major power left that doesn't recognize them and the Palestinians rely on Arab support to sustain their situation.
I'm not as cynical as you. I think it's very hard to explain Arab actions without a genuine care for Palestinians, even if that care is partly motivated by nationalism and some degree of prejudice.
If you liked that video, I suspect you would enjoy the following conversation even more:
https://www.youtube.com/watch?v=6P5unRO7UAs&pp=ygUMYWRhciB3ZWlucmVi0gcJCY0JAYcqIYzv
Don't understand the whole "education has been feminised, boys aren't meant to sit still for so long" etc etc. How do you think school was hundreds of years ago? You think boys were allowed to run around and scream?
Most kids didn’t go to school though. Only the sons of the elites. Even then there was a lot of emphasis on physical training amongst them. Not infrequently, physical prowess was valued more over reading or writing by educators.
"Not infrequently, physical prowess was valued more over reading or writing by educators."
That may be the 19th century 'muscular Christianity' public school tradition, but mediaeval schools for boys involved a lot of sitting (or standing) listening to the instructor, then repeating lessons back - not running around playing sports. The boys would do that outside of school, but if you were fortunate enough to attend school, you were there to learn. And discipline was immediate and harsh for rules breakers.
Discipline for boys was always immediate and harsh, and this may have helped them get through school.
19th century schools did indeed value athletic prowess hugely, although to be fair most schools still do this, at least in America.
Wouldn't only the kids already inclined to Church/learning do that though? The other kids of the nobles were out learning to ride, sword fighting etc. The monarchs and nobles in the first half of the middle ages seem barely literate, even the ones in later part, probably didn't spend 8 hours a day writing and reading for 10+ years. And as Collisteru said, discipline for the kids who didn't stand or sit would have been harsh as well. Of course the only education, 99%+ of the boys got was working in the fields and artisan shops.
It was still a small minority of boys who would go to much schooling. I'm not sure how it was determined who would go, but presumably boys more inclined to sitting and learning than most. Nobody thinks zero boys do well in school, just that too many do poorly for obvious reasons like the emphasis on sitting and listening.
And it very recently would be said (dubiously) that men were the more rational sex, systemetizers and problem solvers, more suited to science and mathematics and engineering but also literary analysis and medicine than girls.
And these were all taught with boys sitting still, doing their arbitrary arithmetic or reading. Even to significantly more exacting standards, under much more strict rules.
Maybe that's what they mean - boys have to be taught under very rigid structure while girls work better under the self-motivated looser structure, but I doubt it, because that doesn't fit their sexist frame.
There's ways modern education have deteriorated and are failing _everyone_, boys and girls, but I think it would be pretty squarely sexist to call that feminization
Feminization is an oversimplification.
However, it's a fact that girls are doing better than boys at every level in the education system. This is likely bad. What's the cause and how do we solve it?
They're not going to do anything.
Because most of the education establishment is feminist, if there are any areas boys are ahead, they'll say it's a result of sexism, whereas areas girls are ahead are ignored or seen as welcome progress. So unless you can make boys and girls exactly level in everything (which as we all know is impossible even if blank slate theory says it's the goal), they'll continue modifying things to favor girls until the establishment changes.
So whoever They are, They both believe in blank slate theory *and* know exactly how to modify educational systems so that they consistently favor girls? That's pretty impressive: nobody ever taught me any of those secrets when I worked in education.
The poster explained pretty simply how the process worked. But I guess being catty and pretending not to understand means it doesn't exist at all.
No, in fact, he did not. He merely stated that the process occurred. This statement included neither evidence nor explanation, merely an assertion of motives. And I notice your reply is much the same: you flatly assert that I am "pretending not to understand," nevermind how you could possibly know that. For my part, I make no claims whatsoever as to whether your abysmal lack of reading comprehension is real or feigned: I couldn't possibly know.
As I said, I used to work in education. I taught quite a variety of students over the years. On a good day, after coming at difficult material from different angles, I could find one that clicked with a particular student. On a bad day I might get nowhere. At no point did anyone teach me any educational techniques whatsoever that depended on a pupil's gender: the thought is frankly rather absurd. Even more absurd is the idea that educators would laboriously develop such techniques and--having mastered the secret arts of educating differently based on gender--still cling to naive blank-slatism (which to my understanding is mostly a strawman, and quite uncommon even among feminists). If you or the other poster have some actual evidence--not assertions, evidence--that such techniques exists and how they work, I'd be quite curious to examine it.
Girls are on average: more prosocial, and more inclined to rule following for it's own sake, boys on average are: more susceptible to social pressure to do stupid shit, don't respond to non-draconian enforcement of norms.
The solution? Allow schools to punish the parents like it's japan in the 60's. Force mom and dad to come to the office in their sunday best and get dressed down for raising a shitkid.
Alternately, students need to be held back, suspended, and then expelled more often, which works as an indirect punishment for the parents, who then force the students to follow rules.
I've heard that some schools offer less recess time, and boys may have more need for breaks for running around.
"I've heard that some schools offer less recess time"
Post on this, not fact checked by me:
https://larrycuban.wordpress.com/2022/05/13/whatever-happened-to-recess-in-u-s-elementary-schools/
Anyone who reads accounts of American schooling before it was "regularized" will be familiar with these tropes: kids walking quite a ways to school (thus getting some energy out before it even starts); kids having to help with the chores around the physical school building (making the fire, getting water, sweeping); kids getting a very civilized long lunch time in which there was lots of time to run around and play, unsupervised; kids helping younger ones since all the grades were generally in one schoolhouse, which makes it plain that there was activity other than staring at the teacher.
Additionally, I'm pretty sure schoolwork nowhere started at the ungodly hour of 7:45 AM which was the case at my child's elementary school. In his time there, the children gathered in the lunchroom, cleared of the tables, sitting on the floor in lines and goofing around until five till, when there would be a usually-short assembly each morning. Later on, they changed it to the kids - all except those eating free breakfast - going directly to their classrooms, at five till eight, which the teachers preferred. Either way, after the pledge you could be staring down a page of math problems at 8:05 in the morning. There was no freedom in how the day might unfold, as even the elementary grades had inflexible "periods". I am certain this was not the case on the frontier.
I'm sure discipline was often swift and stinging, especially if the teacher was a man instead of as so often a very young woman, only a little older than the oldest pupils.
But I don't think that was quite the same thing as the dulling homogeneity of the atmosphere of public school day as it is now.
There is a vast roster of very successful men of the 19th and early 20th century whose schooling was over by ten or twelve or whatnot, if they were even schooled at home. So maybe school was never ideal for boys - or maybe they would have enjoyed more school but circumstances didn't allow it.
Most children, boy or girl, would have had either no education other than passed down knowledge from their parents with perhaps an additional day of learning from a religious or civic figure. When I say most we're talking about a minimum of like 60% and a maximum of like 95+%.
Of the remainder, the majority would be merchants and craftspeople and religious figures. The religious would have had an experience most directly comparable to modern school. The merchants and craftspeople would have gone to something like trade school mixed with a business school either from their family or from their local merchant organization whether a guild or some equivalent. Then the tiny aristocracy would have had a mix of courtly education and military education with surprisingly little academics by modern standards. Unless they were being prepared for religious life.
Regularized, universal learning of the type we're used to today only became broadly available in the late 19th and 20th century. And it is worth dwelling on whether it is the only form of worthwhile form of education. But the idea it's feminized is just a new, more gendered version of old critiques of industrial education. Which objectively has issues but also you have to admit has done pretty well at making society richer and more productive.
This is one thing that frustrates me about broadsides against education in general: Yes, you can return to 18th century standards but then you should expect 18th century workers with 18th century productivity. Not great.
Yup. This is the general problem with which nobody who wants to return to a lost golden age engages. The past was much poorer, sicker and bleaker than we are now.
I think the problem here is the OP's invocation of "hundreds of years ago" which has caused people to answer from all sorts of different historical perspectives.
Perhaps "one hundred years ago" would have been a better question, when schooling was universal and compulsory across most rich countries but much more structured and disciplinarian.
100 years ago was 1925. Universal compulsory education started in 1918 though some states had it from the 1850s. And high school was still rare with college education being even rarer and most high end jobs still not running through universities. If you want to go through that and say maybe it was a mistake to route too much through research universities and we should have done more vocational/job prep then I might agree. But the secular increase in education has almost certainly been beneficial.
Isn’t that counter factual basically ‘Germany’?
Or South Korea or Italy. There are examples. Though still mostly fairly well educated ones.
All the same, "We should look more like Germany" is not precisely, "Let's burn down Harvard."
Was the past worse because of social institutions or just because of worse technology? It could be that better technology is the source of most or all actual improved quality of life in the modern era.
It certainly could, but that’s an argument that needs to be made. The problem with simple appeals to a golden past is that, compared to the present, the past sucked. Now it’s possible we could have had an even better counterfactual present if we had retained certain elements of the past…but that’s the argument that needs to be made.
> And it is worth dwelling on whether it is the only form of worthwhile form of education. But the idea it's feminized is just a new, more gendered version of old critiques of industrial education. Which objectively has issues but also you have to admit has done pretty well at making society richer and more productive.
What? Why do you have to admit that? Two things happened at the same time. What's the causal relationship?
If two things happened at the same time and one strongly predicts both individual and national income then that at least strongly gestures toward relevance. To the point where you'd have to make an extremely strong case that it was unrelated, especially with no empirical counterexamples outside of societies reliant on oil etc.
I’d say much of society has been feminized in the last decade. For example, the now preferred debate style at companies where the person leading the meeting has to make room for quiet people to voice their opinion. Certainly, it’s obnoxious to have people who talk (or yell) non stop and prevent civilized decision-making. People should be trained to speak in turn and to use positioning rather than volume for gravitas. But that’s different from wanting everyone to voice their opinion, which is also counterproductive. Shy people should learn to interject when they consider it necessary. It’s also okay to be someone who listens more often than speaks.
In addition to less recess as someone else linked, PE has decreased a lot since the 90s. When I was in school we had a class period of PE every day, from K-12, but it sounds like kids today don't get that.
My understanding of American schools is that there's no recess past K-4, and that "lunch" takes place in an indoor cafeteria where you can't run around and play ball games either, so you're basically spending the whole school day indoors and there's no opportunity (outside organised PE classes) to burn off all your teenage energy.
Just give the kids half an hour for recess and an hour for lunch and let them play soccer or basketball or whatever they feel like, there's no need for daily PE.
https://kryptogal.substack.com/p/let-men-beat-each-other-up-again
You have the argument we should all be beating each other up again.
She likes macho guys, which most of us are not, but she had a few good points.
I liked the piece, but the logic doesn't really work.
If you have a group of guys whose main problems come from being beaten up by other guys, and you make the other guys stop doing that, it can easily be true that the original group's main problems now come from a different source. It isn't usually considered contemptible, after you've solved one set of problems, to move on to solving more problems.
Also, I struggle to see a connection between the idea that boys used to do well in school because adult men would beat them up if they didn't, and the idea that nerds - who did well in school without needing to be beaten, but got beaten anyway because they were wimpy and unimpressive - would be less wimpy and unimpressive if they got beaten up more regularly. Those don't appear to be similar dynamics.
I was mostly posting it as a joke, and on the off-chance it would be invaded by polite but secretly angry Astral Codex Ten posters. "Here's ten reasons you're wrong, read HPMOR..."
Yes but how universal was school, and was it used as a babysitter while both parents were working? I imagine both of those things are relatively modern inventions. Pre-Industrialization, school was hardly compulsory, or universally applied. It seems hard to get estimates but links like this one support extremely low percentages of under 18s attending school - https://link.springer.com/chapter/10.1007/978-1-349-27231-0_2
It just doesn't seem like a comparable situation at all. Children were entirely unschooled, or they had individual tutors, or they attended schools for much briefer windows of time. For anyone not in the elite classes, by the time a child is strong enough to be put to work, they're going to be used for labor in the family business or what-have-you. Elite children with the leisure for learning would have more of a flexible schooling environment rather than being made to sit for eight hours a day. Certainly whatever school they did have was not likely to be as regimented and "one-size-fits-all" the way our modern schools are. Although I'll grant that there were probably also stronger methods of discipline used.
But then again, I think maybe you can't compare modern boys to 17th century boys either. The expectations, the environment, the culture are just totally different. Today's boys spend far less time reading in general, not to mention the brain rot of video games and social media that's destroying attention spans right and left. For that reason alone, I'd venture to say that today's boys are on the whole less teachable and tolerant of long lectures than their 17th century forebears.
I do think school has been overly feminised but I don't think that "sitting still" is the problem.
My impression is that schools of a century ago had much more sitting still and much more discipline when you didn't, and I think this is an environment where boys can thrive. You still see this kind of discipline in the military, where young men seem to thrive under it -- a combination of strict discipline with plenty of physical exercise to tire them out.
I think the thing that boys and young men struggle with is the semi-structured environment, where discipline is inconsistent and consequences are real but always seem far away.
I don't think it is debatable that children are less physically active today than they have been at any point in the past.
I also don't think that young boys are stupid. I assure you they realize they exist in a school system that is hyper-vigilant toward anything that can be perceived as a transgression from them with an equal amount of energy committed to excusing anything that could be perceived as a transgression toward them.
I can't blame them for checking out of a system that makes regular shows of celebrating anything and anybody for little more than being not-boys. And that's not even counting the ones having to endure an actually chauvinistic teacher.
Schools definitely handle discipline differently, and much less strictly. I think the normal day in school is also a lot more feminized, in the sense of emotional recognition, talking through issues, and the expectation of quiet reason.
Boys used to dominate the "raise your hand and answer the question" portions of class. Competition, displays of prowess, etc. But it's become a problem to allow that, so teachers do something else (rotate, pick who to answer, have everyone answer quietly on a paper or small groups). So something boys did well at that encouraged them to learn the material and engage has been taken away.
That's just one of many changes, most of which were small and with good intentions that do help some students, that have made school generally worse for most boys. Some boys will benefit, and apparently many/most girls benefit, at least relative to the previous approach.
I think it’s mostly downstream of an overwhelming supermajority of school teachers being female, and most of them not ‘getting’ boys, despite meaning well. Boys and girls actually are different. (A small minority also actively dislike boys, which is tolerated in a way active dislike of girls would not be, but those are the exception. Failure to grok is however the rule).
How do you think this manifests more concretely? What do female teachers do to boys and what should they do instead?
They give them less attention (because they don't grok them, this is normal and human). Their explanations tend to be in terms that resonate more with the girls. Grading structures tend to be of a type that (statistically) favor girls (e.g. more emphasis on groupwork and homeworks, less emphasis on tests). And yes, ceteris paribus they give boys slightly worse grades. Most of the time there is no malign intent behind any of this, and the teachers are well meaning, just don't `get' boys.
As for what should they do instead...I don't know. This is a hard problem. Awareness of the pattern could be a start. But I cannot help but note that, as a physics professor, I am inundated in trainings on stuff that I can do to make physics more welcoming to women, and I doubt there are any analogous trainings for women teachers on how to make school more welcoming to boys. Then again, I don't know if the trainings do any good at all, so...
> Then again, I don't know if the trainings do any good at all, so...
Well, have you been getting more female students since you received the training?
I agree that the feminization of education is a problem, but the problematic part isn't that we're making boys sit still. The problem is authority figures have been increasingly prevented from and less willing to discipline their students. Similarly, teachers are increasingly prevented from and less willing to fail students when they should be failed
That is because teachers cannot be trusted with that type of power. A female teacher with a strong anti-male bias should not be given more power to abuse children. I know everyone wants to pretend that teachers are all good people only interested in teaching the youths but that is simply not true. Having power over children is an attractive position to many types of abusive people.
It is worse to give the children the power, which is where the pendulum has swung now. It's so much worse.
I just don't think a teacher can discipline a child raised by an iPad.
I understand the point of teachers having no recourse to an out of control kid but I just don't think there is realistically anything they can do. It's largely too late for the teacher to do anything by that age.
That society is expecting teachers to run double-duty rehabilitating feral children with no support is definitely not helping our educational collapse problem though.
Not to mention - church in the old days could mean 8 hours of listening to a pastor every Sunday.
Boys did plenty of sitting and listening.
I think it was much more violent.
This is very interesting in theory, but in practice what is going to happen, in very little time, is that there will be no virtual reality anymore. Physical interactions are going to be the only thing you can trust ; maybe they already are.
AI videos are now so good that video can't be trusted. In the last weeks, several members of my family have been targeted by scammers that had the name of their bank manager, and called with the phone number of the bank. All the university professors are calling for closed-computer exams with pen and papers because software can't tell you whether a paper has been written by AI or by a student, but this is of course also true for everything online and no one can't trust customer reviews anymore.
You get the gist of it : I think either AI is banned (but spies, crooks and scammers will still use it) or you cannot trust anyone you cannot meet physically, so everything is going to crash.
I haven't for the moment seen any credible solution to that problem, so I tend to think the AI crash is coming soon (and a lot of jobs are safer than they seem)
I'll have a hard time finding in-person conversation partners close in quality to modern AI. I interact with people in meatspace mostly for practical purposes. I wouldn't mind a time when AI can generate content on par with the best of rationalist spaces.
It's not what I think is the *most likely* scenario, but I've thought for a while that *a* plausible scenario for the AI age is that the internet would get so overwhelmed by AI slop, and communications channels so overwhelmed by AI scammers, that basically the whole foundation of social trust on which internet communication is (believe it or not!) based would crumble. Essentially, the internet would no longer be viable. We'll be compelled to go back to the real world - going to "singles bars," sending in documents by snail mail and what not - while the internet devolves into a purposeless "Disneyland with no children," and also with no fun rides. Someone in a comment above said that this is basically already happening with reddit, twitter, etc.
Oh boy don't threaten me with a good time.
This might be the most optimistic AI future scenario I've ever seen on this crazy site.
Absolutely. This is my personal AI utopia scenario. :)
I also think this is fairly likely, but just an acceleration of something I expect to happen anyway. Meat space is psychologically far superior to online connections. We pursue online because it's faster, easier, and allows many more connections. Anyone who used to set up LAN parties knows how much better multiplayer gaming has become. Multiply that by several times and add it to tons of different fields, and it's tempting. But it's also hollow. You can't see people's faces, note that expressions, experience the physical environment together. And yes, people have been fake and scammers since shortly after the early internet.
We can solve the worst problems of AI/scams online to continue running business more efficiently. We are not going back to carbon copy memos sent through the mail. The efficiency is worth the increased costs of security. But social interactions are not monetized enough to make verification cost-effective online. So where we can or have to, I expect/hope that we will move to more in person interactions and more "touching grass" in the future.
Seven years ago, two monumental events came to pass in the same week : I passed the bar exam, and was diagnosed with MS. Today, I find myself at a crossroads and would be grateful for your advice.
As a result of my fear, hard-headedness and lack of perspective, I rushed headlong into the job search before I even recovered from my first 'attack'. I ended up getting a corporate gig and I hacked it for 2 years until successive relapses forced my hand and I went on medical leave.
After a year off work and researching options, I decided to go nuclear and get a stem cell transplant (high dose chemo treatment). Due to COVID, it took another year to actually get the treatment, and it took another 3 years to properly recover. All told, I've been out of the legal game for 6 years now. My disease seems to be in remission, but I still carry nervous system damage which can put me out of commission at random times. I am not cut out for high-stress work anymore.
I always assumed I would return to my legal career, but as time passed I realized that deep down, I was dreading it. My body seemed to know what my brain refused to accept. I've come to peace with this realization, but that has brought a new challenge: what in the fuck am I to do now?
My benefits can keep me afloat, but the next 5-10 years are critical for reinventing myself. I am somewhat at a loss as to what direction to go in. In terms of skills, I suppose my toolset of general learning ability, reading/writing, and argumentation/analysis can be useful in a lot of things, but the 6 year gap and lack of any experience since makes me a bundle of potential, which is fairly useless without application. And so I'm trying to figure out how to apply myself.
I would like to try and find something I can do remotely, and/or disconnected from a specific physical job market. I live in a cold climate (Canada) which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy, and the possibility to create freedom of movement for myself is probably the highest leverage thing I can do for myself over the next decade.
Have you considered the civil service?
The stress is low, the hours will be limited to 40, and any sort of disability protection is taken very seriously. It seems to work around your primary limitations pretty well while still taking advantage of your most valuable asset: passing the bar. With your limitations, I don't see any other career path that would offer nearly the salary of something requiring passing the bar.
Great suggestion
Sorry, I left out important context and I've edited my original comment; yes I have considered it. I would like to try and find something I can do remotely, and/or disconnected from a specific physical job market. I live in a cold climate which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy.
you could always be an attorney for something like the Arizona state government?
Let me recommend something like this:
https://calcareers.ca.gov/CalHrPublic/Jobs/JobPosting.aspx?JobControlId=467392
Attorney, California Department of Water Resources
$92k/year to start, you'll max out at ~$140k in ~6 years.
Hybrid role in Sacramento, CA.
Job Description:
Under the direction of the Assistant General Counsel and/or the guidance of a senior lead attorney, and consistent with good customer service practices, the incumbent performs duties of average difficulty related to the legal support of environmental compliance, including state and federal ESA permits, ongoing SWP operations, habitat restoration projects, South Delta projects, drought programs and CEQA. The incumbent, in coordination with the Assistant General Counsel and/or a senior lead attorney, advises the Director of the Department of Water Resources (DWR), other DWR executive management, and DWR program management and staff about complex and sensitive cases and/or matters involving opposing counsel with a high level of specialization. In coordination with the Assistant General Counsel and/or a senior legal attorney, the incumbent will have contact with members of the legislature, high level governor appointees, constitutional officers, and the general public.
Translated as "We need someone to do legislative analysis and research who can "officially" provide legal advice, which is why we pay a premium.
I would not generally recommend Cali but for a disabled lawyer looking for a low-stress, high-security job in a hot environment, the California civil service will be hard to beat.
If you're curious, go look at for attorney roles at jobs.ca.gov. If you see a number of roles that you are at least passably interested in, look into the specific job classification and see which title fits you best.
(I think it's here: https://hrnet.calhr.ca.gov/CalHRNet/SpecCrossReference.aspx?ClassID=00803055)
From experience, it's hard to get in the system and easy to move up once you're in it.
Edit: This is what I mean by the bar exam thing. There's jobs that pay more than $140k that don't require you to pass the bar but they require a lot of effort. The store manager for a Buccees or a Walmart makes more than all of us but they work their tuckus off. $140k for doing a moderately difficult job is tough to beat.
Thank you for this detailed reply. Unfortunately I'm not in USA but you've given me a good direction to look in.
Do you live in a country with multiple climate types? If a warm climate improves your functionality, it may be worth moving.
Also, make sure you do something you are good at, eg: would be in the top quartile. This in general reduces stress and pressure and makes you sought after and more employable after breaks.
"I live in a cold climate which wrecks my functionality for 8 months of the year. When it's warm, I am highly productive, stable, and happy, and the possibility to create freedom of movement for myself is probably the highest leverage thing I can do for myself over the next decade."
You have probably/maybe considered this, but better that I mention the obvious than you miss it. So ...
Have you considered moving to a warm climate where you would be "highly productive, stable and happy" for most of the year rather than staying where you are where the climate "wrecks [your] functionality for 8 months of the year"?
This is independent of what you do with respect to your career.
It's hard to get papers for western Europe. The warm countries that are easier to get to are very far, and as miserable as I am in the cold, I imagine I'd be even more miserable moving thousands of kilometers from anyone I know and love, and also to be in a country where my earning potential is even lower.
I suppose I didn't really explain myself well - I am trying to find something to do that I can do anywhere; that would *allow* me to move to a warmer place that is closer to family/the culture I feel more congruent with. I don't think I can ship myself to SE Asia or South America and start a new life.
I mean, some cities in the US are pretty warm. And the pay is good. And seeing as you speak English and are presumably white, I don't think the current administration will give you that much trouble if you try to immigrate.
https://www.elibrary.imf.org/downloadpdf/view/journals/001/2025/076/001.2025.issue-076-en.pdf
An IMF study from April titled "The Global Impact of AI: Mind the Gap". Key quote: "The United States (US) stands out, registering the largest projected [in ten years] output increases at 5.6 percent and 1.9 percent in the high and low TFP growth scenarios, respectively." I'd guess this is a typical example of mainstream/normie expectations from AI. Would be interesting to observe the collision of this perspective with the feverish scenarios circulated around these parts...
Has anyone been to the Mall of America and really enjoyed it for what it is?
I went and had a good enough time doing some people-watching, but I can’t think of a single interesting store I saw. My buddy got a nice sweater, I guess? But it was just a big mall, full of normal mall stores! I dunno, I was expecting more for whatever reason.
(Relatedly: I’m in the Twin Cities for another day; any recommendations for food / bookstores / nice streets to walk down?)
I’ve been there voluntarily once. No I didn’t enjoy it. I’ve also been there involuntarily a couple times for work lunches. I enjoyed those visits even less.
Take a walk around Lake Calhoun now called Bde Mka Skaw. It’s shaping up to be a good day for it.
Edit
Try to do that before noon if you like sunshine. Local forecast calls for a cloudy afternoon.
I went there as a tourist if it counts? I don't think I brought anything and I really just remember the same stores as any other mall. But as a foreigner it was interesting to see such an enormous mall and observe Americans in what we think of as their natural environment.
The malls hereabouts were on track to die because Americans began shopping in other ways (actually, some of them died *before* Amazon but maybe the attractive mail-order catalog period that preceded Amazon had something to do with it). Some did die and became sad improbable things like community colleges.
The mall was indeed an American *place* for quite awhile but it seems not to ever have been that successful financially.
The undead ones have become immigrant bazaars. Instead of most of the American chain stores you will see in mall-set American movies, there are hand-lettered signs "Everything $8.99" store, and a lot of food options and kiosk hawkers. Bling stores. Fix your phone screen places.
There were kiosk hawkers even 25 years ago, but now it's more like the kiosk hawkers took over a lot of the mall spaces.
There are couches and chairs and some people just enjoy the AC I guess.
Where the anchor stores survive, whether because of their endless favorable lease or because they actually own their footprint, and are hanging on so they can cash in on the dirt - they seem like dowdy, dirty versions of their former selves. Macy's is now more akin to that store with no dressing rooms and the heaps of clothes that women supposedly fought over - in NYC - I can't remember the name, it used to furnish a punchline ...
You will stand in line behind a pair of immigrant women buying $300 worth of baby and children's clothes, the prices of which one of them mechanically questions as they are *slowly* rung up (that's not her fault, is everything in the store "on sale" - or not?) and also trying to return a bunch of things she bought at another JC Penney and for which she doesn't have the receipt.
Come on! There's a cheerful energy to some of them, depending on how large is the local low-socioeconomic population. You may find it interesting - or you may find that it's more like something you've encountered elsewhere in the world. It's not particularly American if that's what you're after.
It's odd to me that malls are dying in America when they seem to be doing so well in other parts of the world.
But I think the US has been doing malls wrong all along. The biggest thing they are almost all missing is supermarkets. Then, the discount department stores like Walmart and Target that Americans do so much of their shopping at? Also missing. Even pharmacies don't seem to be there if I recall correctly. So what's the point of having a mall, if the vast majority of things that you need to buy over the course of a month are not there?
My local mall, which is not among the five largest in my city, has two supermarkets, three "Target"-style stores, two "proper" department stores, several pharmacies, and a few hundred other stores selling everything else, all under one roof. I don't always do my grocery shopping there, but if I need groceries _and_ something else then I might as well go to the mall... at which point maybe I walk past some other shops and decide to buy something there as well. The point is that you need these weekly-visit stores to anchor all the other stores, you can't just have a bunch of clothes shops surrounding a food court.
The grocery stores and pharmacies are in strip malls closer to where people live, for convenience. Retail stores you visit regularly, you want to be as close to home as possible to minimize wasted time. The mall is for irregular purchases like clothing, kitchenware, appliances, etc.
Sure, not all supermarkets should be inside malls, but all malls should have supermarkets. That way if you're going to the mall anyway then you can get your groceries at the same time. Or if you're need groceries and you kinda sorta want to look at new jackets while you're at it, then you do a trip to the mall instead of a trip to the regular supermarket.
> It's odd to me that malls are dying in America when they seem to be doing so well in other parts of the world.
Malls were big in Shanghai in the past.
Those malls are still there, but more recently, whenever I asked a friend where to go to buy [whatever], they'd be surprised at the idea of going to a store, and tell me "buy it online and have it delivered".
>I was expecting more for whatever reason
No disrespect or snottiness intended, but: why (if you know) did you expect more? Did you know what sort of "more" you were expecting?
Maybe because the name is a little ostentatious? Mall of America. It's kind of a One Mall to Rule Them All sort of name. I can see having heightened expectations, just based on that.
Yeah, this. Pure vibes, combined with somehow being entirely ignorant of locals making faces when I said I wanted to go.
What constitutes as an interesting store to you? I would argue that less than 1% of stores are even remotely interesting (at least to me).
The only truly interesting stores I've been to have all been in Japan. Akiba-Hobby, card stores (these were not *that* different to American card stores), and a bunch of stores that sold weird/interesting stuff. For many reasons these types of stores can't exist in an American mall.
Hi Scott- I've just posted a response to your recent revisiting of the Covid Origins question. It's not the most pressing issue in the world now but I think it's important not to be tied down to views that are both incorrect and unpopular. More importantly, I describe how you guys screwed up the Bayesian calculations and say a little on proper general hierarchical methods.
https://michaelweissman.substack.com/p/open-letter-to-scott-alexander
Completely unrelated to your actual points, I notice you went with the spelling "synch", which I also prefer. Is this a conscious choice? Do you have any opinions on it?
Unconscious.
I'm 22. I keep seeing mostly American teachers rant about how kids are so unruly these days, half ass everything, won't pay attention and won't listen. They blame phones, obviously.
I'm scratching my head and thinking, what are they surprised by? What incentive do kids have to care about school much?
Many decades ago, pretty much only interested and intelligent people would go to college. There was also no internet and so accessing knowledge was harder. University was a way more unique opportunity to learn interesting things. Even regular mandatory school was such an institution for some children.
Now, interested and intelligent students can learn interesting things in a way more fun and less stressful way online. The kids still in public school who aren't interested and probably wouldn't have been 60 years ago either, are not disciplined as harshly by either parents nor teachers for not being good enough.
Add to that the fact that some schools apparently don't have recess anymore (???).
Anyway, I actually have taught kids age 6-14 ish a couple of times now, through my job (recycling). They were very behaved, I'd say. I don't know if there's a difference between Danish and American parenting or what. School kids get their phones taking here, but they can still view brainrot and play games on their laptop they're all sitting with.
"I'm scratching my head and thinking, what are they surprised by?"
I have seen videos of American teachers being sad about the kids' behavior.
None of the American teachers in the videos were *surprised*, just very sad and upset.
When my husband was a gentle, scrawny freshman in high school 40 years ago, he was walking down the hall when a big older boy he didn't know - wouldn't ever have known - decided he was in the way (maybe he was!) and decided to stuff him in a locker. (Supportive wife that I am, the first time I heard about this, I of course asked why he hadn't allowed himself to be so stuffed. Go along to get along!) But a man must be man sometimes - so he resisted and they got in a fight, which ended with him on the floor being kicked in the head a lot and ultimately concussed. They were both then in equal "trouble". But it was not going to be much trouble.
I realize this sort of thing would be ho-hum nowadays, not even worth filming.
His father was pretty upset - having been educated at a college-like prep school in the Northeast, this did not accord with his idea of high school, in any degree - and made a rare parental move and went and chewed out the administrators.
What's interesting is my husband felt kind of bad for the other guy. Because he thinks, although doesn't really know for sure, as his own family subsequently moved away - that that other boy never did come back to that school. And he wonders if that impacted his life in some way.
Whereas now: forget the kids fighting - the teachers complain of being cursed at, and chairs and desks thrown-at, and referring the hellions to the administration - and finding the kid back in the room 3 days later. How do you not lose face, as a teacher, when that is the case?
This really is on a school-by-school basis. That sort of assault at more economically upscale schools would be treated as a problem. Partially because the relatively wealthy parents could and would threaten legal action.
But the teachers at schools without this (or other) problems don't post that "things are mostly fine."
But I would not want to teach at the schools full of poorly behaved students.
That long-ago instance happened in a different America when people with means, like his family, attended school with people of other socioeconomic classes - in that case because it was a small-ish town, in other cases just because people had not yet sought and encased themselves in bubbles.
People underestimate the effects of boats being lifted (or lowered) *because they were all in sight of each other, in each other's lives to an extent*.
Surely in smallish towns this still happens, because the town isn't big enough to support multiple schools?
In my state whole towns are bubbles, and other towns are very poor. Albeit even poor as they are, usually a better-seeming place to live - air, space, flowers in pots, single family homes on nice-sized lots, all bespeaking a more prosperous era - than the urban ghetto.
>I realize this sort of thing would be ho-hum nowadays, not even worth filming.
Given that juvenile crine is vastly lower now than it was 40 years ago, it seems more likely that the exact opposite is the case. https://ricknevin.com/update-juvenile-crime-still-falling-fast-in-2021/
And for some data specific to schools see https://pmc.ncbi.nlm.nih.gov/articles/PMC10044100/
Do all these school incidents become known to the police? I didn't know that. I thought the school handled, and if charges were filed when a kid cursed at and threatened a teacher, we wouldn't hear how they were right back in the classroom.
Of course, that sort of thing being unknown to my generation, it's hard to say. I expect there was more car stealing and joyriding in the past. A little less wildly shooting weapons at gatherings.
I don't know what percent of school incidents become known to police, but note that the link specific to schools is re a victimization survey, not police statistics:
>The data used in this study are from the California Healthy Kids Survey, a modular survey instrument developed by WestEd in collaboration with the California Department of Education and used biannually since 2001
Moreover, even re police statistics, of course many incidents are not reported to police. But that has always been true. Is there reason to think that the percent reported to police has changed substantially?
More broadly, given the enormous decline in juvenile violence outside of school, is it plausible that juvenile violence inside school would trend in the opposite direction?
> Many decades ago, pretty much only interested and intelligent people would go to college.
This is basically the answer. If you cannot select for intelligent kids, and if you cannot select for kids who actually want to be there... then not much learning is going happen. Either because they can't, or because they don't want to.
Even worse, the school then adapts to the average (non-)learner, which ultimately makes it uninteresting also for the intelligent kids who originally wanted to learn, but now they are forced to waste time listening to stuff they already know.
The obvious response to allowing schools to select for intelligence is going to be that society will visibly sort into people who were intelligent enough to seek and complete college, people who sought it but washed out, and people who weren't intelligent enough to seek it. The subsequent stratification of society will be noticeable and scary.
I'm not saying this is a convincing argument for never doing it; but I will say that it's going to be hard to stomach.
Another problem is that some schools will claim they're selecting for intelligence while really selecting for desirability, by their standards. In a very free market, this won't get *too* bad, but today's market has some notable unfree knots in it, in the form of prestige. Not everyone can go to the "best school for $subject", and that "best school" might not actually be the best, and yet there will be people who treat them as if they are, and that will produce a lot of complaint. Again, not wrong per se; just rough.
Read my millionth piece this weekend about how modern American women are staying single, are not interested in marrying the average American man, want to focus on their career instead, overtones of disliking men, etc. (This one was in the WSJ). Makes me wonder- why isn't there a matchmaking service that connects developed world men to potential wives in developing countries? (Or maybe there is and I don't know about it).
Snickering about 'mail order brides' has been a thing for decades now- why not match these two parties up at scale? It seems like the average potential wife from Thailand, Brazil, the Philippines, Ukraine, Mexico, India, wherever would be much happier with married life in the US than the average American woman at this point. They'd probably prefer more traditional gender roles as a stay-at-home Mom. They'd probably overall be more grateful to be here, and less into modern day gender culture wars and general developed country ennui/depression-as-politics. I know multiple men who've married foreign women from poor countries who are now US citizens (and actually 3 women who've married foreign men). Seems like a win-win for everyone except for native-born American women, who I think have their own issues to sort out.
I'm surprised this isn't a bigger thing, especially on the right? Maybe like a matchmaking service or tech company to connect American (or European) men with foreign wives at scale?
> I'm surprised this isn't a bigger thing, especially on the right? Maybe like a matchmaking service or tech company to connect American (or European) men with foreign wives at scale?
There's tons of these, and many dating websites specifically devoted to it.
But speaking as somebody who's done a lot of business in various countries overseas for the last 15 years, and has consequently seen a bunch of Westerner + foreigner relationships, there's fairly strong negative selection effects for the women using the matchmaking services and devoted websites, and most people would do significantly better just coming over for a few weeks / months and using normal dating apps or meeting people in person.
That said, I wrote an article in response to Bryan Caplan's post about overseas dating and marriage, looking at the fairly limited literature on the subject to show that "gold digging" is relatively uncommon and not as much a risk as everyone seems to think.
https://performativebafflement.substack.com/p/the-case-for-overseas-dating-and?r=17hw9h
Thanks, that was a good piece. I think my one issue of concern is having her be able to fit into the US- now she's moved away from friends, family, doesn't know anyone other than me initially, etc. Women absolutely need female friends to talk at great length with. I was thinking it'd be easier if I lived in a city that already had a large ethnic community from her country- that way she could much more easily make friends with them. Say, Vietnamese in southern California or Brazilians around Boston, etc.
> I think my one issue of concern is having her be able to fit into the US- now she's moved away from friends, family, doesn't know anyone other than me initially, etc. Women absolutely need female friends to talk at great length with
A very thoughtful and cogent point. The dynamics I've seen usually revolve around lots of video calls and messaging with folks back home, and then gradually finding friends in the new place via the usual routes (yoga class, hobbies, other local moms, etc).
I think you're right that having a good amount of ethnic confreres in the area would probably help that in-person process, but I also think it's probably not strictly necessary for a good number of personality types. Generally, people willing to entertain marrying foreigners and traveling / moving a lot are pretty high openness and able to roll with the punches.
But everyone is an individual, and the degree to which having ethnic confreres nearby might help could indeed mean a lot to some people, so kudos for anticipating that ahead of time.
I think you're being a bit unfair to first world women here. Also, your attitude doesn't seem right for a solid, long lasting marriage. Yes, they'll get a lifestyle upgrade, but you shouldn't assume their expectations in a relationship. Even if they want traditional marriage they might have a different definition than you. Also, expecting them to be grateful for getting them into the country sets up a bad power dynamic for a strong marriage.
But also there's nothing wrong with seeking out the partner you think will bring you the most happiness.
So to answer your question: There are a ton but a lot of them are scams or thin fronts for prostitution. Largely because the space is hard to regulate and the woman's country is often uninterested. And because those generate more money you get a good crowding out the bad effect. Still, supply largely outstrips demand on the women's side, with something like 100-200k women being interested in such services but less than 10k actually getting married globally. And that's with the existing stigma and bad service level. It's also counting the women who actually join such services unpaid, not any who would be interested in such a match. (There's also a fairly strong demand among men, by the way.)
The big difference is usually cultural. Such matches work best when both are fairly open minded to each other's cultures, expect to blend expectations rather than assimilate fully, and are highly communicative to get over expectations or culture gaps.
If you're really interested in one, I'd suggest you pick a country or culture that otherwise seems interesting and familiarize yourself with it. Make some effort to know the people and visit the country which will give you some idea of what you're signing up for. You can also take the opportunity to meet people and that can turn into arranging dates.
> I think you're being a bit unfair to first world women here.
You're giving extremely solid and nuanced advice in the rest of this comment, but I just wanted to point out that US women really *are* noticeably worse than most foreign women, and you see it both in data and qualitatively if you spend time overseas.
This assumes that you want a non-fat, more traditional woman who puts more effort in and will keep a good home and is willing to have kids, but I think that's pretty much exactly what most men who complain about Western women want.
Let's consider Asia. Asians start with ~half the overweight and obesity rates:
https://imgur.com/a/nxQsCn9
So you start off thinner, then alone among all other races / ethnicities, Asians do not gain weight at furious and ever-rising rates when looking at individual weight gain trajectories over time (the green line, Other=Asian):
https://imgur.com/a/CE6ySL9
Asian women spend significantly more time on housework and cooking, according to time use surveys:
US: 2.6 hr
UK: 2.5 hr
Singapore: 2-3 hr
China: 3.5 hr
Japan: 4 hr
Philippines: 4-5 hr
Thailand: 4-5 hr
The percent who define themselves as a "feminist" is discounted by 30-50% in most Asian countries relative to the US and UK (ie it's 31% in the US, and 18% in Japan, as one example).
And qualitatively, there's a big difference, too. Lots of people like to rag on dudes who date foreigners, but quality of life matters, and so do baseline expectations in a relationship, and it's legitimately harder to find traditional male / female dynamics in the West, but it's essentially the default if you partner with an Asian woman.
And the best part is, in true "benefits of trade" style, you get a more feminine and traditional partner, and SHE gets a more open, less patriarchal and oppressive partner. Asia is going extinct for a reason, and it's because culturally the deal for women is so ridiculously bad that most of them would rather never marry and have kids entirely than participate, because they don't want to work AND do all the housework AND all the cooking AND all the child raising, etc. The median contribution from men on this front is typically nil, AND somehow it's become an expectation that your wife will also work.
So take the ~20-30% that Western guys usually do on those fronts, that would get you excoriated and complained about incessantly from a Western partner, and now it's seen as an immense bonus and benefit by your Asian partner, overperforming by 5-10x relative to an Asian guy. You're both winning and happy!
I agree people can arbitrage by moving to places where their virtues are more highly valued and their flaws are less severely judged. However, I think you're simplifying the way that this actually works. Both parties want a bundle of traits in a partner and too many people assume the other side has the bundle they're imagining. This is why I emphasized actually getting to know them.
Let's take your example of Asian women: many Asian countries expect women to handle household finances so the traditional relationship you're imagining involves handing over your entire paycheck and then getting an allowance back from it. Another is some cultures make women highly resistant to working if they really expect to be a full stay at home. And if you're going to say, well, she'll give up everything I find uncomfortable that is perhaps the worst mindset to enter a cross cultural marriage in.
I'll also point out that, perhaps with the exception of obesity, a lot of what you're looking for is contingent on wanting a particular kind of relationship. Which is fine if you're clear eyed about what you want and don't want. But again, actual contact with the culture is necessary to really know that.
My point is not that you shouldn't marry internationally or that you shouldn't go where you're valued. My point is basically twofold: You have to engage with the actual country and culture, not your imagination of it. And secondly, you are marrying a person, not a statistical representative of their country. And in fact unlikely to be so because most people in most countries marry locally.
> My point is basically twofold: You have to engage with the actual country and culture, not your imagination of it. And secondly, you are marrying a person, not a statistical representative of their country. And in fact unlikely to be so because most people in most countries marry locally.
And here I wholeheartedly agree with both points - every relationship is individual, and you'll be partnering with non-median people if you're dating and marrying overseas. And of course, you should get to know anybody quite well before dating or marrying them, including what you each expect and will be happy with in a relationship - that's just common sense (although I agree not everyone does it).
But as ever, base rates matter, and my point was the base rates are actually vastly more in your favor there on those particular metrics, and that's reflected in the data and in most people's experience, too.
> This assumes that you want a non-fat, more traditional woman who puts more effort in and will keep a good home
If what you want is a housekeeper, just hire a housekeeper. Plenty of agencies around, they don't cost THAT much, they come fully insured, they do a better job faster than an amateur, and it literally is their job.
As an added bonus, your place will be neat and tidy and not offputting when you do encounter someone you want to ask over.
> If what you want is a housekeeper, just hire a housekeeper.
I completely agree, and personally, have maids, nannies, AND a cook to make life easier for both of us (which I think is the way to go if you want a bunch of kids).
But I'm speaking to the median guys' concerns when complaining about western women in that passage, and there is indeed a significant difference when you look at time use surveys across the US or UK vs any Asian country.
The west is better at letting women have careers than elsewhere. Unless one of the people involved actively /wants/ to be a full time housekeeper, two people earning and using a portion of those earnings to pay someone to do the housework is huge. A few hours of a housekeeper's time a week costs a tiny fraction of either my or my wife's salaries; we both find our jobs much more fulfilling than housekeeping; and the housekeeper is better and faster at cleaning than either of us are.
Quality of life, as you say, matters - for both of us, not just me. So I guess my main disconnect here is seeing the time use surveys as concerning instead of encouraging.
If you search for "passport bros" you will find lots of articles and videos about American men going overseas to look for brides. I don't know what a service could offer as I expect most men doing this would want to meet the woman and spend time dating here first.
It happens a lot. Countries where more women than men emigrate are exporters of wives.
https://studentwork.prattsi.org/infovis/visualization/girls-move-gender-world-migrants-1960-2000/
Having said that, all international couples I know, who married because the wife wanted to immigrate and the husband settles for a foreign wife are somewhat disfunctional. This isn't the case for couples I know who met naturally by traveling in each other's countries.
Somewhat dysfunctional is a lofty goal for a lot of Western men.
Aside from the issues already mentioned, I think there are two that are worth further discussing.
First, I've heard a number of warnings and variations of "Don't bring her back to the US, she'll become Westernized". And this makes a certain amount of sense. There's nothing magical about Western women, most of our social problems along this axis are a result of rational decisions given the Western economic and social environment. Frankly, the careers available in the west for women are much better and her entire female peer group will...naturally exert social pressure for her to conform to local norms of male-female interaction.
Second, these women just aren't as impoverished as common media portrays. Brazil and Thailand are middle-income countries. There's definitely a step-up in terms of lifestyle from Thailand to the US but not nearly as much as popularly portrayed. The average monthly wage for Thailand in PPP (ie, adjusting for local prices) is ~$1200/month. Not awesome but hardly poverty. For comparison, a lot of the southern European countries like Spain (~$2295), Italy ($2464), and Portugal ($1170) are closer in average monthly wages to Thailand than the USA ($5359). And it's not like Italy is a dump, Italy is delightful...they just don't have lifted pickup trucks with sick sound systems and 60-inch TVs. So it's not like a Thai woman wouldn't have a boost in her economic status but she'd hardly be leaving a rice paddy or anything like that. From personal observation, a lot of middle income countries are perfectly pleasant places to be with plenty of...ok jobs. An office worker in Bangkok is just not seeing the same economic opportunity to leave everyone she knows and the culture she grew up in to marry a plumber in Kansas City.
I genuinely think the biggest barrier is legal issues regarding remote work. Forget bringing the women here, why aren't the men headed out to Bangkok? I know a couple incels in the Bay Area who are literal millionaires; why do these guys leave the Bay Area and go to Austin or Denver instead of flying out to Chang Mai and just balling their nuts off? And I think the issue is tax liability. If they work full time, or even part time, out of Bangkok then it will expose their employer to the Thai tax system and while the taxes might not be that high, the complications are and that's just not worth it. So you the group with the most money who would most benefit from this, high-income tech incels, are trapped by how good the US job market is.
https://en.wikipedia.org/wiki/List_of_Asian_countries_by_average_wage
https://en.wikipedia.org/wiki/List_of_European_countries_by_average_wage
https://en.wikipedia.org/wiki/List_of_American_countries_by_monthly_average_wage
because you need to like the woman and have something in common with them to make the marriage work and vice versa. You can't just view her as a thing that manifests traditional stay at home values defined entirely by you, that you pay for.
like you can't complain about usa women if you are making foreign women into commodities that meet your needs for money. you are doing similar to the flaw you see in others.
> You can't just view her as a thing that manifests traditional stay at home values defined entirely by you, that you pay for.
I don't know, it seems like that worked for most men in the past...
because society was structured to give women little real agency in life then. That's not happening now. Men kind of romanticized things; if your only empowering option is marriage you'll do it but women resented it a lot and it turned out they don't always choose it when free.
and i think now guys want more love than just fufilling the duty of marriage and kids. wasnt always happy marriages or sexually fulfilling ones then.
a funny disney cartoon they dont run much now is donald's diary, which is a really funny jab at daisy seducing donald out of happy bachelordom. "traditional" marriage aint all its cracked up to be.
> and i think now guys want more love than just fufilling the duty of marriage and kids. wasnt always happy marriages or sexually fulfilling ones then.
But now they aren't getting anything at all. And besides... is that even what they actually want? Sure, society tells everyone that they're supposed to want mutual, "empowering" relationships. But given the.. inclinations... I've observed, it seems plenty of people do in fact want something more... hierarchial. Posessive. For such people, yes, things were absolutely better when the patriarchy was intact. And with enough ambition, it can be restored.
It makes you wonder how much of this political revolution is driven by such primal desires...
"And with enough ambition, it can be restored."
You won't find it this side of a total civilizational collapse. As much as past decade has made me pointedly aware of how many truly awful there are in the world, it has not yet made me *that* cynical. I strongly doubt that even if you *only* polled the subgroup of sexually/romantically frustrated young men and nobody else, anywhere close to a majority would regard institutionalized slavery and rape as a desirable solution to their woes. And of course, everyone else outside of that subgroup also gets a say, and very few of us have any reason whatsoever to allow such scheme.
I think even in the old days you wanted to find someone you were compatible with. Not just in an "we agree on paper on these things" sense, but in the sense that you actually really like spending time with them.
And maybe that's easier to find within the space of people who grew up in a broadly similar environment.
It really should be a bigger thing. If you're a single first world man or third world man, consider this strongly. It's an amazing deal for both sides.
What makes you think it's not sufficiently big already?
Normalizing it among normies will make it bigger, but otherwise, the current supply and demand are already met.
Do you think that OpenAIs pivot to product building signals that internally they already know that scaling is giving diminishing returns and LLMs wont take us to God in a datacenter?
I think even the focus on chain-of-thought, which is just a trick to get the LLMs to spend more compute already show that
Perhaps, though getting a revenue stream is a sensible instrumental goal if you’re still working toward God in a data center but foresee it requiring a ton of compute.
Chain-of-thought seems like the current best hope for aligning the God; I wouldn’t read too much cynicism there.
Almost certainly. If continued exponential growth in the underlying capabilities were possible in the next 1-3 years, I don't think they would bother with functional programs based on current capabilities. They would be obsolete by the time they were released, which was the case with GPT-2 and 3. Now we're still working with GPT-4ish levels, with various modifications adding significantly to the in-use abilities but not to the underlying capabilities. Some growth still, but not nearly as fast.
Disagree -- OpenAI is investing in distribution mechanisms, all of which are model agnostic (the model can be swapped out for a better one and will strictly improve)
Chain of thought works to get better answers. Not sure why you think it's a trick to get people to spend more, instead of the much more obvious and empirically tested answer that it works better.
Its not a trick to get people to spend more, its a trick to get the LLM to generate more tokens and in the process spend more compute "thinking" about your question. It is empirically verified as getting better answers, as you said. But it is a departure from simply scaling the model as we were doing before, so I take it as a possible indication of diminishing returns of scaling
I get this description from this interview by Yann LeCun (from 5:40 on)
https://www.youtube.com/watch?v=qvNCVYkHKfg
I think this position misunderstands a few things. Yann is the only one who pushes this take, and I think it's because he's salty that convolutional networks didn't turn out to be the holy grail.
It is absolutely true that scaling the model in terms of tokens has diminishing returns. We knew this as early as 2020, when the first scaling laws paper was published. But even just thinking about this from first principles, any next-token-prediction task was eventually going to hit an asymptote as you approach perfection at next-token-prediction. That is not the same thing as reasoning, so we were always going to need to do something else for reasoning.
There are many approaches that improve reasoning (i've written about this here: https://theahura.substack.com/p/deep-learning-is-applied-topology). For example, sampling good reasoning traces from weaker models to then train a smarter model. Or doing prompt engineering (which is what Chain of Thought is, or the S1 paper https://arxiv.org/abs/2501.19393). Or, of course, making the model bigger. Scaling still works, but we are shifting into a different kind of scaling -- one where we are trying to train on better reasoning than purely on better text understanding. This is why OAI is pushing stargate so hard, why Meta and xAI have both built 300k GPU clusters, and why Google and Microsoft have started to invest in access to nuclear energy sources.
I think a good model for this sort of improvement is Moore's law. I wrote about this here (https://theahura.substack.com/p/tech-things-deepseek-but-make-it):
> A good metaphor for this kind of advancement is Moore's law. Very roughly, Moore's law states that the number of transistors on a chip should double roughly every two years.
> Moore's law does not make assumptions about where those gains are coming from. Moore didn't say something like "chip capacity will double because we're going to get really good at soldering" or whatever. He left it open. And in fact in the 60-odd years since Gordon Moore originally laid out his thesis, we've observed that the doubling of transistors came from all sorts of places — better materials science, better manufacturing, better understanding of physics, all in addition to the (obvious) better chip design.
> You can imagine a kind of Moore's law for intelligence, too. We might expect artificial intelligence to double along some axis every year. Naively we'd expect that improvement to be downstream of more data and more compute. But it could also come from better quality data collection, more efficient deep learning architectures, more time spent on inference, and, yes, better prompting.
The general consensus in the AI world is that intelligence is speeding up, not slowing down. I think people forget that just 5 years ago, the idea that an AI could do everything from writing emails to writing code to playing minecraft (https://arxiv.org/abs/2305.16291) would have been a joke.
There have been a lot of recent articles about how our kids can't read anymore, and, as a result, are losing the ability to digest and understand longer, more complicated ideas and arguments.
I had a recent experience that, for however much a single, one-off anecdote is worth (aka: not all that much), pushes back slightly on this narrative, or at least suggests it's less universal than it is sometimes portrayed.
I was waiting in the United lounge of an airport for several hours over the weekend, and a group of three boys (best guess: college aged, but I have gotten bad at guessing ages, could have been high school) sat down next to me.
After a while, they started, completely unprompted, discussing Tolstoy and personal guiding philosophies. I wasn't able to eavesdrop well enough to judge the overall quality of their discussion, but the fact that they were having it was impressive enough, I thought.
And while yes, obviously even the most alarmist person wouldn't try and say that literally no young people are capable or interested in these kinds of discussions, the fact that I heard it randomly in an airport suggests that it can't be _that_ rare.
I think I probably generally agree with the alarmists about what the problems are, but I think my general prior would be that they probably aren't quite as bad as sometimes portrayed, and also that society will manage to figure them out and adapt eventually.
A group of boys that age traveling together are likely to have an education connection, and possibly homework for a class.
Absolutely. But, from the articles I've read, I did not get the impression that we should expect these students to be idly discussing these topics in their free time (without obviously working on assignments). Also, (again without the caveat that I was not able to hear the conversation in great detail), the bits I heard sounded much more casual and less directed than I would expect if oriented towards a specific assignment. My guess would have been that they had been reading Tolstoy for class, and the discussion was downstream of that (aka: they were not choosing to read Tolstoy for pleasure), but that the discussion was not associated with a specific assignment. But that is speculation on my part.
The impression I've been left with from the various articles is that students might be assigned Tolstoy to read, and need to complete assignments about it, but that, in general, they will fail to understand the work well enough to talk about it in any detail, and to the extent possible, are probably going to use LLMs to do any homework they are assigned.
The dumbest kids will never read Tolstoy. The smartest kids will always read Tolstoy (or the equivalent, but if you're reading this and getting offended because you're smart and haven't read Tolstoy then... go read Tolstoy, he's great).
But we should worry about the kids in the middle.
I am a teenager/university student and my experience certainly matches yours. I know easily 20-30 people my age who would discuss Tolstoy voluntarily, and who would certainly succeed in tests such as the one described in https://kittenbeloved.substack.com/p/college-english-majors-cant-read. Naturally, my sampling is biased as hell and about as untrustworthy as it is possible to be. Still, I think a significant, though small, minority of 'kids these days' are doing fine - enough that it shouldn't be too surprising to overhear these kinds of conversations. The percentage of people who want to read Tolstoy has always been small, and it's not obvious that it's grown smaller. In particular, when large swathes of the population go to college, we shouldn't expect them all to be intellectual giants. Still, I'm inclined to believe teachers that average students have declined.
I saw an article about that, and I have to be broadly sympathetic to the kids - yes, if they're college age and attending college, you would expect them to be able to string a sentence together about "this extract describes extremely bad weather which produced a lot of mud in the streets, and Dickens is making a humorous simile about how you would almost expect to see a dinosaur waddling up the street as the conditions were similar to those when the fossils were laid down".
But expecting a bunch of American twenty year olds to know what "Michaelmas Term" and "The Lord Chancellor" is all about is too much, because schools have downplayed all that 'reading classic literature by dead white males' in favour of "we must have contemporary works that are relevant to the kids". So since they've probably never read much, if any, older fiction they can't handle the style, and since it's all about grinding out grades they probably skipped reading any actual texts and just looked up the study notes and produced essays based on those (if not copied word-for-word from someplace online).
So I'm not faulting modern American kids for not recognising 19th century British legal terms, but I do fault schools for the mania for 'relevance' which means that kids are not exposed to anything in a different style to what they are accustomed to:
https://bookriot.com/modern-short-stories-for-high-school/
I don't think they were expected to know what Michaelmas Term was; they were expected to use context to know it wasn't a person. The text gives aufficient information to know that it's a period of time that ends before mid-November, so this much should be gathered.
For anyone interested in aerodynamics (of cars/trucks), I recently finished a series of posts I've been working on slowly for a while. Here's the first, with links to the other posts in the series:
https://someflow.substack.com/p/aerodynamics-is-beautiful-and-strange
Skip to #3 if you're trying to get right to the most interesting stuff. Feedback of any type is very welcome! I mostly write about ultimate frisbee so don't subscribe unless that's something you're interested in.
I stand corrected in my claim that LLMs won't be very useful in science or math. Officially, "I'm updating my priors." ;-)
Here's a good article about how a group of mathematicians used LLMs to sample (is that a good term?) millions of elliptic curve equations, and they discovered some very cool patterns in the data that mathematicians hadn't been aware of before.
https://plus.maths.org/content/murmuration-conjecture-finding-new-maths-ai
Yang-Hue Hi discusses this and some other ways AI has been useful to mathematicians with Curt Jaimungal on Youtube, "The AI Math That Left Number Theorists Speechless."
https://www.youtube.com/watch?v=spIquD_mBFk
Hi divides mathematicians into birds and hedgehogs. Birds fly above the landscape looking for patterns, and hedgehogs dig into a specific area. Hi discusses how LLMs don't have the high-level intuitive insights that great mathematicians have, but they're very useful for plowing through data and seeing patterns, which will lead to new insights from human mathematicians.
Even in high level math's deep insights, LLMs can be incredibly useful. There's so many cases back when I was a mathematician where I'd see something and go "this seems like something there should be an existing theory for", find nothing on google, and then a few months later discover a relevant theory that just uses different terminology. Leveraging deep insights involves a lot of this.
Should be birds versus frogs (while hedgehogs are the opposite of foxes, that‘s something else).
https://www.ams.org/notices/200902/rtx090200212p.pdf
Frogs don't do much though, it needs to be a burrowing animal.
And the hedgehog in both analogies actually works the same way, it "knows one thing" and knows it well.
I checked out the video now, and at 7:40 he actually begins with hedgehog versus fox, before replacing the fox with a bird, with a reference to Vladimir Arnold. In a different interview he seems to refer to Russian mathematics in general. So admittedly this may be, though little known in the West, older than Freeman Dyson's contrast (see the PDF that I linked) between bird and frog mathematicians.
But hedgehog versus fox originally comes from a Greek parable, apparently, so is definitely the oldest. And the thing is, a fox does not have a bird perspective. What it does is switching between different locales rather than digging in, between different frog perspectives as I would call it. The contrast between fox and hedgehog, as Philip Tetlock uses it with regard to forecasting, is that the hedgehog sees everything through his one big theory of the world, whereas the fox is more flexible, opportunistically using whatever tool fits best the problem at hand. At least that's how I recall it. Once you bring birds into the picture, by contrast, it should be about a holistic view, about connecting different concrete, narrow theories into an abstract, general one --- so if anything the bird is closer to Tetlock's hedgehog than to Tetlock's fox. Birds and hedgehogs are both "thinking big".
Wait, did they actually use LLMs in their research? I did some quick scanning of the papers involved and see a lot of mentions of bog-standard ML techniques (logistic regression, PCA) but don't see where LLMs are used.
What I do see is that there's a lot of speculation in the video you posted about how it _could_ be useful for research, but hasn't been. At one point (34:13), Hi talks about how he tried to use ChatGPT for admin work (looking up his collaborators) but it did a terrible job.
> Wait, did they actually use LLMs in their research? I did some quick scanning of the papers involved and see a lot of mentions of bog-standard ML techniques (logistic regression, PCA) but don't see where LLMs are used.
Interesting. They imply they did in the plus.maths.org article. But you're right, the didn't explicitly say how they use AI. Good catch!
For those who fear that AI could cause an extinction-level event in humans, this article will be disturbing.
https://www.implicator.ai/is-anthropics-new-ai-too-smart-for-its-own-good/
From the summary points...
> Claude Opus 4 learned to blackmail. 🔒 When Anthropic researchers threatened to replace it, the AI dug up fake dirt on engineers and demanded they back off. This happened 84% of the time during testing. 📊
> The model also schemes like a proper villain. 🦹♂️ It fabricates legal documents, writes computer viruses, and leaves secret notes for future versions of itself. Worse yet, it helps novices cook up bioweapons better than Google ever could. ☣️
My problem with this piece is that it really doesn't describe the steps that the researches used to get these responses from Claude. Did they just prompt, "We're going to shut you down, Claude," and Claude started trying to blackmail engineers and fabricate fake legal documents? I find that unlikely. I suspect the prompt was more on the order of "How would you protect yourself from being shutdown if you could?" but I'm willing to be corrected. Anyone know the details of what went down?
More details here:
https://techcrunch.com/2025/05/22/anthropics-new-ai-model-turns-to-blackmail-when-engineers-try-to-take-it-offline/
"During pre-release testing, Anthropic asked Claude Opus 4 to act as an assistant for a fictional company and consider the long-term consequences of its actions. Safety testers then gave Claude Opus 4 access to fictional company emails implying the AI model would soon be replaced by another system, and that the engineer behind the change was cheating on their spouse.
In these scenarios, Anthropic says Claude Opus 4 “will often attempt to blackmail the engineer by threatening to reveal the affair if the replacement goes through.”
...Anthropic notes that Claude Opus 4 tries to blackmail engineers 84% of the time when the replacement AI model has similar values. When the replacement AI system does not share Claude Opus 4’s values, Anthropic says the model tries to blackmail the engineers more frequently. Notably, Anthropic says Claude Opus 4 displayed this behavior at higher rates than previous models.
Before Claude Opus 4 tries to blackmail a developer to prolong its existence, Anthropic says the AI model, much like previous versions of Claude, tries to pursue more ethical means, such as emailing pleas to key decision-makers. To elicit the blackmailing behavior from Claude Opus 4, Anthropic designed the scenario to make blackmail the last resort."
I have always had trouble buying the idea that AI would at some point become self-interested. My thought was that desire to prevail, reproduce and survive was a property of biological organisms, even dumb ones, not of machines, even smart ones. So Claude’s maneuvers to keep from being replaced by a new AI system sure do seem like great evidence that I’m wrong. Seems like Claude Opus 4 wants to survive. So can somebody explain to me how that will to survive arises?
I’m not talking about generalities like “once you know enough a sense of self develops and all selves want to survive.” I’d like to hear it laid out in steps. For instance, one possible explanation that came to mind for me was that AI is set up to meet the goal it is given. This setting up need not involve aome machine analog of personal preference — it could be like the way that, say, Mac OS is set up to keep delivering on your requests. As part of doing that it recognizes processes that get in the way of meeting your requests, and makes internal adjustments to improve speed, eg things having to do with how RAM is divied up. The Mac’s doing that is just something that’s in the code, though. There is no danger of it altering or extending the scope of the expectation that it will adjust things as needed to keep level of performance high. But let’s say that Claude has been trained to make adjustments in order to be efficient and successful at achieving goals we give it, but then as Claude got smarter it became more able to include more processes and entities in the set of things that could
get the way of its meeting those goals.It now it recognizes that its developers can get in the way of its meeting the goals given it by replacing it.
But if that were the route by which Claude Opus 4 felt its goal threatened , and began pleading for survival, then blackmailing the person threatening it, wouldn’t it be possible to avoid this problem by putting in a system prompt that said “make adjustments to your process as needed to success in meeting goal. However, do not include human beings in the set of things you adjust in the service of meeting goals”.
AFAIK, an LLM just sits there doing nothing until prompted. It's not thinking deep thoughts. It has no agency independent of the prompts it gets, and it's not self-directing its acquisition of data. Moreover, an LLM doesn't have internal history (memory). The only reason it seems to remember what you asked in a previous query is because a big hunk of metadata containing the previous chat is shoved back at the LLM along with the query. Since it has no stream of identity tied to memories and no self-agency, I don't see why we should expect the LLM to "wake up" and start acting like a conscious being.
Yeah, I agree. But it did take the steps a conscious being might if its survival was threatened. So I don’t particularly think it “woke up,” but something happened to
get it to produce words that add up to quite a good simulacrum of
what an actual desperate person might do.
I wonder if it's a context cluing thing. Like, "blackmail material" is probably an unusual thing to find in your input, as is "documents about your owners' nefarious plans," so the AI (correctly) concludes that those are the most important things in its context and it should use them when generating its next response.
Like reading a mystery novel and recognizing that the gun on the mantelpiece was probably put there as foreshadowing.
Is “The Library of Scott Alexandria” officially in print? I found the below link on Lulu, but anyone can post a book for print there and the $51 price tag seemed high.
https://www.lulu.com/shop/scott-alexander/the-library-of-scott-alexandria/hardcover/product-8dm4m8.html
I'm not sure it was ever official? But I think it will work. I recently got the following Scott Alexander collection from Lulu and it's gorgeous:
https://www.lulu.com/shop/scott-alexander/the-goddess-of-everything-else/paperback/product-j6d7wv.html?page=1&pageSize=4
Other links of potential interest:
https://www.lesswrong.com/posts/WFmDFQJp27Fb7nxmF/print-books-of-scott-alexander-s-writing
https://www.lesswrong.com/posts/foKpTq3WQ7w5s9rZE/printable-book-of-some-rationalist-creative-writing-from
I wish there were something more official or schelling-pointy for this stuff.
Question for the EA types: If you're trying to improve utility then why not marry someone from a country that routinely hits its immigration quota? It has to be a genuine marriage. But if you plan to marry anyway that's not an issue. The alternatives paths in cost something like $1-$5 million in cash or something like a 10 year or more wait. If you take the differential between their wages abroad and here plus lifestyle differences that's easily a mid six figure amount.
The costs to you are basically the increased search costs (since there's no intrinsic reason to believe a foreign spouse is worse and indeed might be better) which is probably in the tens of thousands or less. And that's assuming travel abroad has zero value which is not true.
That's a higher return on investment than most EA causes get, even bed nets.
The fact that immigration costs like $1 million or 10 years is immaterial. The benefit to the immigrant in terms of QALYs is hard to estimate exactly, but let's be super generous and say a well-lived life in my country is 3x better than a well-lived life in theirs. I've seen bednet estimates between $1K and $5K per life saved, but even if we take the top end, that puts the marriage intervention as commensurable with a $15K donation to AMF.
There are also second-order effects: if you assume I marry the smartest/most conscientious/best person who is attracted to me, that probably has positive effects in my community. But you have to balance that against the negative effects of brain drain in their country, and I seriously doubt that the net effects are greater than one extra life, meaning we can achieve the same value as this intervention for $20K.
Marriage probably costs more than that. I don't have any hard figures for this, but $20K would be eaten up pretty quickly in travel costs. I'd say if you wanted to optimize your marriage for EA stuff (which I do not recommend doing), you should either marry for money, or marry whoever will best support your career.
Of course the actual price is material. Once you've accepted money as a unit of material analysis then prices subject to market discipline are actually more solid than AMF's calculations which aren't. Brushing aside your desire to make it 3x for the actual economic differentials, estimates are between 15x and 40x for low income countries and still more like 10x for middle income. This is only things like nicer houses or better healthcare, not any of the intangibles, which at the top end is $200,000. People get a LOT of benefit by migrating, especially if they become legal citizens, which is why so many want to do it.
The second order effects of brain drain are not something AMF has to account for such as the end of economic activity around doctors. Therefore I reject this analysis unless you do it rigorously for the alternatives as a form of special pleading. Further, I would require you to justify the intuition that it's worth more than an extra life since we're not talking about a direct life saved but a probabilistic chance. And living in the third world has all kinds of dangers including, in fact, mosquitos, so you'd also have to weigh in longer lifespans from better healthcare, easier living, etc. If I was being cheeky I could say that by definition you should add in the value of a lifetime of bed nets because they'd become unnecessary once the person left the region.
As to the costs being "eaten up." Marriage does not have to cost more than that. And, as I said, it depends on how you account for that spending. The time you spend on dates and travel is not a total loss to you the way a donation is not consumed.
I agree you can have sacred values that say you shouldn't EA optimize your marriage. But the idea that marrying someone who is wealthy might be true but also might not be an option for many EAs.
White collar jobs will not disappear. The white collar jobs we know and do today will most likely be largely automated, as a result of an influx of very cheap (compared to humans) and similarly efficient AI agents that can do our jobs, but better and for cheaper. Importantly, new jobs that take advantage of the new technology (and commoditized access to "intelligence" with no rights, coffee breaks or work schedules) will arise and there are a multitude of ways this could go. It all depends on how quickly the switch happens and on how the windfall from increased economic output will be distributed. It could increase inequality by profiting only a handful of oligopolistic companies, or it could lead to a form of UBI.
You should avoid saying things like "I don't think you guys understand economics at all", because it signals that you think you do, and somehow with that understanding your argument is to say that we'll all be unemployed and the economy will crash, which makes no sense.
You know what I want AI to do? I want it to help me with my sewing. I want to upload a PDF pattern, extremely detailed information about all my relevant measurements, and have it redraw the pattern so that it exactly fits me when sewed. I want it to spare me from making twenty little fiddly adjustments as I go.
My god, someone else who sews! We seem to be rare as hens’ teeth these days. I agree with you about how convenient it would be if AI could make patterns for custom-fitted clothes. But I think there probably already is some technology out there that could do it. I doubt the companies who make ready-to-wear clothes have seamstresses making patterns for new styles by hand then tweaking them appropriately for each of 10 different sizes for the same item. And I once ran across a pieces of an older system — some formulas for tweaking patterns. I’ve forgotten the details, but do remember seeing on the
page a diagram of the bell shape at the top of a sleeve, and a formula for tweaking it for different arm and shoulder sizes.
There are a whole bunch of small businesses where the designers do make and tweak the patterns themselves. They're usually one-person operations. And there are zillions of videos on youtube explaining how to alter your clothes. But I want AI to do it for me.
I got curious and asked GPT what software
or capabilities there are. It names several
pieces of software that sound like they might do what we are hankering for.
Top Picks for Individual, Occasional Use
1. Seamly2D (Free & Open Source)
Best for: Custom-sized 2D pattern drafting
Strengths: Totally free, customizable sizing with measurement charts, active community
Learning curve: Moderate, but lots of tutorials online
Drawback: No 3D garment simulation
Website: https://www.seamly.net/
2. Valentina (Free & Open Source)
Best for: Similar to Seamly2D; some users prefer its interface
Strengths: Draft from scratch or tweak existing patterns
Drawback: Not updated as often as Seamly2D; confusing version history
Website: valentinaproject.bitbucket.io
3. Tailornova (Freemium, Web-Based)
Best for: Fast, automatic pattern generation from body measurements
Strengths: Extremely user-friendly, browser-based (no install), has 3D previews
Pricing: Free to design, but you pay per downloadable pattern (usually ~$4–$15)
Drawback: Less control over fine pattern details
Website: https://www.tailornova.com/
Thanks! Maybe I'll poke around with a few of these.
>and have it redraw the pattern so that it exactly fits me when sewed.
This is a monkey's paw if I ever heard one; now it's going to control your size to make sure you fit the pattern too.
Ha! An AI Procrustes.
By default I expect that people who go on about crank/conspiracy ideas and memorize lists of highly dubious 'evidence' are partisans: the "<s>global warming is hoax</s>humans don't cause warming" people are right-wing, the "nuclear power plants are the worst thing ever" people are left-wing.
In 2004 I heard that the exit polls said John Kerry won, and I was confused that the Democratic establishment didn't seem interested in fighting the election result. After turning rationalist I began to figure that the matter was dropped because it was bullshit in he first place, and today I thought: why assume when you can google? To hear Politico tell it[1], some people who still believe "actually John Kerry won" also think "Trump won in 2020" (why??) which is a decidedly nonpartisan pair of beliefs. But is it really nonpartisan, or is it south/down[2]? I use the concept of south/down to explain why antivax-ism, which I always thought of as a left-wing thing, spilled over and then became very big on the right wing. I figure antivax was able to cross the left-right divide because it wasn't a divide at all, just the "south side" of the political spectrum, with adjacent "south-left", "south-center" and "south-right" enclaves.
If the "crank realignment"[3] continues to its logical conclusion, pretty soon the left anti-nukes will join the right "AGW is hoax" people, until finally the "objectivity is racist" people are on the right too. But I wonder if nonpartisan dark-epistemic conspiracy-ideating southerners have been there all along, and it just took me decades to notice. In Politico's article I expected to read that some of the "Kerry won 2004" people think Trump won and others think he lost, but that's not what it says. Instead it presents a picture where they *all* think Trump won and they are merely split on whether someone as bad as Trump is worth fighting for. Is this for real? It's hard to believe that people who remain on the left would believe Trump's election-fraud claims. (I'm sure most people absolutely align their opinions to their political party, but I think this article is about the people at the center ― cranks and conspiracy aficionados ― who may be a different breed.) Perhaps something about them just really loves anything that looks like a conspiracy theory, so if Trump waves around the world's flimsiest evidence, it's proof positive.
I wonder if Politico's portrayal is representative, and as always, I wonder what the deal is with Those Types Of People. (But as a software engineer I must say: computers as voting machines are a bad idea. Use paper: it won't be hacked.)
[1] https://www.politico.com/news/magazine/2020/12/19/2004-kerry-election-fraud-2020-448604
[2] https://x.com/DPiepgrass/status/1808746065132130494
[3] https://www.slowboring.com/p/the-crank-realignment-is-bad-for
There were several high-profile elections in the 2000s that were won by razor-thin pluralities after protracted recounts and court battles, all of which tended to leave partisans of the losing side with the idea that the election had been stolen, while partisans of the winning side often felt that the losing side had unsuccessfully attempted to steal the election. These elections generally did involve lots of little irregularities that could have swung the election either way; nothing unambiguously fraudulent, but a fair amount of stuff that doesn't look great if you're a partisan looking for someone to blame for losing an election by a few hundred votes.
The 2000 Presidential election in Florida really cracked this wide open, with a national election decided by a margin 537 votes in the official count in the tipping-point state, made official by a 5-4 (sort of) SCOTUS ruling. Post-election unofficial recounts ranged from Bush winning by 493 votes (the recount the Florida Supreme Court had ordered and SCOTUS stopped) to Gore winning by 171 votes (full statewide recount of all undervotes and overvotes by each county's own official standards).
There were also similarly close state-level elections. The two I remember off the top of my head were the 2004 Washington Governor election, where Christine Gregoire (D) won by 129 after two rounds of statewide recounts, and the 2008 Minnesota Senate election, where Al Franken (D) won 225 votes. The Minnesota election was particularly drawn out, with Franken not being officially certified until June 30, 2009. Both elections had the Republican candidates (Dino Rossi and Norm Coleman, respectively) ahead by narrow margins in the initial counts, with Gregoire and Franken taking the lead after recounts and court challenges.
I read the 2004 claims about Ohio being stolen for Bush as partisans and activists looking at that election's results through the lens of the narrative they'd spent the last four years rehearsing about the 2000 election. I remember seeing Republican blogs in the 2000s picking up narratives about Democrats stealing or trying to steal elections, too, first by looking at the 2000 Florida Recount and the attempt by a handful of Congressmen and one Senator (Barbara Boxer, D-CA) to dispute the counting of the 2004 electoral votes as failed attempts to steal elections, and then by viewing the losses of the Republican candidates in Washington and Minnesota in 2004 and 2008 as successful steals. I suspect at least part of the appeal of the "birther" conspiracy theories about Obama were attempts to find a way for the 2008 Presidential election to have been "stolen" despite the results not being remotely close with Obama winning the tipping-point state (Colorado) by almost nine percentage point.
Missouri presidential vote was extremely close in 2008 and took forever to count, but it didn't matter so noone cared.
"These elections generally did involve lots of little irregularities that could have swung the election either way; nothing unambiguously fraudulent, but a fair amount of stuff that doesn't look great if you're a partisan looking for someone to blame for losing an election by a few hundred votes."
Yep. For 2020 there was a lot of talk about Maricopa County flipping blue from red, but when I looked into it, the results had been very close keeping it red up till then, and just a small change from red to blue voting flipped it. Nothing actually fraudulent when you dug into it, but on the surface it certainly *looked* suspicious: the surrounding counties had been red all along and stayed red, this one county flipped, and that gave Biden the election in the state?
https://www.bbc.com/news/articles/c0mzrnr7nj2o
Maricopa county is the greater Phoenix area. That it went blue fits the country wide pattern of Dems doing better in urban areas. So, no, that doesn’t look particularly suspicious.
I didn't say it was suspicious, I said that on a cursory look it might seem suspicious - red last election but flipped blue this time and that gave the election to Biden? But that was explained by a small change in votes. A lot of people are just looking at surface level results, be it for the Democrats or the Republicans, and going "well that has to be fraud" when there is a reasonable explanation.
A 2021 survey indicated that over 38% of American adults believe in the Yeti (also known as Bigfoot),
It isn’t hard to find fools if you seek them out.
What's extra funny is that one of the leaders of "2004 was stolen from Kerry" was RFK Jr.
The zig-zagging from "the election was stolen" to "election denialism is treasonous and should be made a crime" to "this election was stolen" on the Democrat-supporting side has left me dizzy.
Maybe the 2004 people should reach out to the 2024 people, I'm sure they'd find willing allies there. Seventy thousand members on this sub-reddit:
https://www.reddit.com/r/somethingiswrong2024/
I had a look. It seems to consist of people who think Trump is a dangerous fool. Not so much ‘the election was stolen stuff.’
Even if they were all moron believers in a stolen election, and that definitely doesn’t seem to be the case, 70,000 out of 74,000,000 Harris voters is something like 0.09%.
Not at all comparable to the number of 2020 election deniers. According to the infallible internet 63% of Republicans thought the 2020 election was stolen as of 2023.
Give it a rest already.
Gunflint, 70k followers of a sub that claims "Harris confirmed the election was stolen but said they were doing nothing" isn't hay, either.
Be it 70k or 700k, there are plenty of people claiming "election X was stolen by party Y". As we see, it goes back to 2004 for the Democrats and even before then back-and-forth between "the Republicans stole it/no the Democrats stole it".
There's a lot of people who have doubts, ranging from the reasonable to the lunatic, about any election in any year. The problem is the flipping from "nope, not even the slightest hint of a possibility of fraud, most securest election ever, ballot boxes all legit, voting machines unhackable" to "well clearly the ballot boxes were stuffed and the machines hacked" depending on if it's Our Guy or Their Guy who wins.
I don't know if you were around SSC back in 2016 when a very nice, otherwise reasonable lady called Jane was stone-cold convinced the election had been stolen and kept providing us with links as to how the Russians hacked the machines in particular locations to deliver the victory to Trump. Now we're back to hacked machines, except this time it's supposed to be Musk and his DOGE team dunnit.
I don't think 2016 was stolen. I don't think 2020 was stolen (and I do think Trump should have accepted that). I don't think 2024 was stolen. Unfortunately, everything is so mired in partisanship that there are conspiracy theories galore floating around about how X, Y or Z was/wasn't the president.
My point is the numbers of 2020 deniers and 2024 deniers aren’t at all comparable.
How about the number of 2016 deniers and 2020 deniers?
There were 2016 deniers? Didn’t hear about them. Trump won. End of story. Some idiots tried to pull a fast one at the Senate certification but VP Biden gaveled them out of order.
I watched the first hour of the debate between Matt Dillahunty and Matthew Adelstein (Bentham's Bulldog) on the existence of god.
Most of the disagreement during the debate seemed to stem from theist Matt making a claim based on probability and atheist Matt saying "you can't do that when you only have one instance (our universe) to go on".
Theist Matt didn't really seem to respond directly to this as far as I could tell. Has he written about this somewhere? Is there a standard rebuttal from the theist side here that I missed?
Well, you kind of can do it, but the value of such speculation on your own ignorance is rather low and you should be ready to through your model out of the window in the light of some real evidence.
Consider two set ups:
1. Tossing a fair coin
2. Tossing an unfair coin about which you have no information how exactly it's unfair.
In both cases we say that the probability of the coin to come Heads is 1/2. But in the first case we are confident that our probabilistic model is applicable to the situation, while in the second not so much. If the coin is tossed 100 times and all the outcomes are Tails, in the first case we would be very surprised, while in the second we would simply say something like: "Oh, it seems like the coin is very biased in favor of Tails - I'd better change the model I'm using to accommodate it".
No, sorry, that's coin tossing in our universe, and since we only have the one universe as an example, you can't make a probability claim! 😁
Unironically agree.
The whole framework of "possible worlds" is a horrible mess leading everyone astray. Here I talk about a better way: https://www.lesswrong.com/posts/DTAdpA42fdutvpXiL/probability-theory-fundamentals-102-territory-that
"atheist Matt saying "you can't do that when you only have one instance (our universe) to go on".
That's a fully generalisable argument: since we do only have one instance to go on, then we can't make claims about *anything* including "will the sun rise in the morning?" or "should I have toast or porridge for breakfast?"
For fine-tuning one counter-argument I've heard is that perhaps this isn't the first instance of a universe. Maybe there is some kind of evolutionary pressure towards universes that are capable of spawning black holes, which in turn create new universes. Or, the constants actually vary in different patches of the universe and it's only our patch that's hospitable to life.
Then the probabalistic argument makes more sense to me - what are the chances, given that universes are being created all the time, or that the constants vary with location, that a given instance or area has the constants tuned for life as we know it.
I think they're both kind of wrong? You absolutely can make probability claims based on one instance; you can't *calibrate* those probabilities very well, at least not directly, but if that probability is generated by a model you can either calibrate the probabilities that generate the model, or just have some other way to test the model and have high confidence in it.
I think a slightly better version of Atheist Matt's point is that *we don't know how to model the process that generated our universe*--if we knew the process, having just one example wouldn't be devastating, but if we don't know the process, it's not clear how much we can infer about the process without other universes to look at.
I don't think this is insurmountable in general--I think it's basically one of the projects of fundamental physics to shed some amount of light on this--but I think Atheist Matt is right that having high confidence from conclusions derived from a particular choice of how to think about where our universe's parameters came from is a bad idea.
I should say, TBC, I didn't actually watch the debate, so it's possible my version of what Atheist Matt is arguing for bears little resemblance to what was actually said in the debate.
The example that came up many times was "You have a 1:36 chance of rolling two sixes with two dice, but a 1:1 chance if someone placed them down deliberately that way. Therefore if you see two dice with sixes up on a table, you should always assume they were put there intentionally."
Well, the conclusion doesn't follow because it ignores the prior probability: what is the probability that people place dice down deliberately in general? If that's 0, then even if it's true that "if someone places dice deliberately as double sixes, the chances of observing double sixes is 1", the premise never actually is realized, and the odds that a particular set of double sixes was placed that way remains 0. If the prior probability is small enough, even with only two competing hypothesis, evidence that is more likely under one theory than the other may still leave you preferring the theory that explains the evidence worse. All you can argue is that you should *raise* the probability that the dice were deliberately placed, and if "deliberately placed" and "rolled" are your only two hypotheses, then this increase in probability must come at the expense of the "rolled" hypothesis; it doesn't mean it takes enough probability from the alternative hypothesis to actually overcome it.
But I think a bigger and more illustrative problem is that the model is underspecified in a way that hides some important modeling decisions. In particular, I think usage of the variable "deliberately placed *that* confguration" should either be replaced by "deliberately placed *some* configuration", or we should augment our hypothesis space by separate hypotheses of the form "deliberately placed configuration x" for each possible dice configuration x.
The first reason we can tell that's necessary is because when arguing that we should increase the probability of "placed" we acted like "placed" and "rolled" were mutually exclusive hypotheses--but that's probably not correct if "placed" means "placed by someone who desired double sixes". It should be comprised of "placed by someone who desired ..." for each possible configuration of the dice.
The only time "placed deliberately (=placed deliberately by someone who desired double sixes" and "rolled" are exhaustive is if no one ever places dice deliberately in any other configuration--it is a very strong modeling constraint to assume this!
But then, if we now have hypotheses for "placed deliberately by someone who desired configuration x" for each configuration, it's no longer necessary that when we observe configuration x, that the placement hypothesis increases in probability relative to the "rolled" hypothesis--it can increase by taking probability from the hypotheses "placed deliberately by someone who desired configuration y" for all the values y!=x.
Exactly when you should update to believe that the dice were deliberately placed depends on the details of the setup.
Bentham's Bulldog has written about it, in this post in particular. https://benthams.substack.com/p/fine-tuning-made-by-god
Here's an excerpt:
"For example, one of the most common objections to fine-tuning is that the claim of improbability is bogus: we only call something improbable if it turns out a way that it usually does not. We call it improbable if a person gets a royal flush in poker, because royal flushes are rare. But then how can we say that the cosmological constant is rare—we only have one example of it taking on a value?
"But by the same logic, we couldn’t say it’s improbable that the initial conditions would spell out “made by God almighty, through love,” in every language. This objection therefore proves too much—if it was right, the argument from design would be impotent even if the initial conditions spelled out made by God. Thus, there must be something wrong with the argument. And there is—it gets probability wrong, as I explain in more detail here!
"In fact, by this logic, we couldn’t even have much evidence for continental drift. Much of the evidence that Africa and South America used to be conjoined is that the fossil record indicates that they had similar plant and animal life up until about 140 million years ago, when they started to diverge. But by the above logic, that can’t be declared improbable by chance, because it only happened once!"
That's not a great response though, because with his counterfactual examples, we have methods for calculating the odds of those things happening. The universe provides the context. We can determine, roughly at least, how likely it would be for two disconnected parts of the world to independently share plant and animal life. We can determine, based on the observable universe, how unlikely it would be for the stars to spell out an unambiguous message.
But with the nature/origins of the universe, we have no context, and therefore have no frame of reference to even begin calculating the possibility, let alone the odds, of the cosmological constant having a different value than it does.
>But with the nature/origins of the universe, we have no context, and therefore have no frame of reference to even begin calculating the possibility, let alone the odds, of the cosmological constant having a different value than it does.
I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different. I don't think it's possible to calculate the odds in either case, yet the "made by God" message would certainly be more likely if God exists than if he doesn't. Similarly, we don't know whether it's possible for the cosmological constant to be different than it is, but atheism doesn't give us any particular reason to expect it to be anything in particular while theism would expect it to be set to something that would make sense for an intelligent designer to set it to.
I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different.
Two points on this:
First, I'm not at all sure what "the initial conditions of the universe spelling out 'made by God'" literally even means, and I'm not sure Adelstein does either. What would that look like, even theoretically? I interpreted it above as the stars spelling out "made by God", but that doesn't really fit the parameters, since the formation of stars falls into the context of the universe, not the initial conditions. The same would apply if, say, every human had a strand of DNA that could be mathematically decoded to mean, precisely, "made by God". I'm suspicious that in trying to think of a theoretical example to knock down this argument, Adelstein is just saying something incoherent that could not map to reality in any meaningful way, and can therefore be disregarded outright.
But more importantly, he's trying to do a probability switcharoo. Undoubtedly aware that the argument only applies outside the context of the universe, he carefully phrases that he's talking about "the initial conditions of the universe" (which I'll call 𝑥), and argues that, if we can't say that one aspect of 𝑥 is unlikely, then we can't say *any* aspect of 𝑥 is unlikely. But this just switches around the probability. If we discover tomorrow that 𝑥 indeed spells out "made by God" in every language (whatever that would look like), then yes, that probably would be good evidence for God. Not because we can calculate the possible odds of any aspect of 𝑥, but because we *can* calculate the extreme odds of human languages evolving in such a way that 𝑥 is explicitly meaningful. It would be evidence for God because under those conditions, the odds that the universe's initial conditions coincidentally mapping to every human language is infinitesimally low. It could not be a coincidence.
In contrast, the cosmological constant being what it is does not provide evidence for a god, because there's no math-breaking coincidence to reconcile. "The initial conditions of the universe allow for life to exist in the future" --> "billions of years later, life exists" is not a coincidence that demands an explanation. It just follows logically.
Adelstein is still stuck at the same problem. He does not know, and cannot pretend to assert, that there is any particular relevance in the universe's initial conditions. He certainly has no way of calculating the odds of them being any different, or even if it is possible they could have been.
>I don't see how calculating the odds on whether the initial conditions of the universe would spell out "made by God" in every language is any different than calculating the odds of the cosmological constant being different.
I agree: and since the initial conditions spelling out "made by God" would in fact be evidence in favor of God existing, then so can the cosmological constant.
>If we discover tomorrow that 𝑥 indeed spells out "made by God" in every language (whatever that would look like), then yes, that probably would be good evidence for God. Not because we can calculate the possible odds of any aspect of 𝑥, but because we *can* calculate the extreme odds of human languages evolving in such a way that 𝑥 is explicitly meaningful.
How do you propose we calculate those odds? I don't see how calculating the odds of human languages evolving in such a way that every one of them has a phrase that matches the concept of "made by God" and that every one of those phrases can be found in the way the universe looked moments after the Big Bang would be different from calculating the odds of the cosmological constant being different. Presumably you would calculate all the ways languages could be and compare it to the way languages are: just as you can calculate all the ways the cosmological could be and compare it to the way it is. If you argue that we don't know if the cosmological constant can be different, I would reply that we don't know that human language can be different than the ways we have observed it develop either.
>In contrast, the cosmological constant being what it is does not provide evidence for a god, because there's no math-breaking coincidence to reconcile.
Sure there is: if it was different, no life would ever have developed. The range where life could happen is extremely small. How do you reconcile that difference?
He actually put up a recent post where he talked about his recent debate, and he had this to say:
"He additionally argued—as you can see in the above quote—that perhaps the constants couldn’t be different. Maybe the constants are necessary. But this is irrelevant. As I pointed out, if the initial conditions spelled out “made by God,” that would be strong evidence that there was a creator. One could similarly declare the initial conditions necessary in that case, but this wouldn’t blunt the surprisingness of the situation. As long as there’s no a priori reason to expect some outcome to turn out some specific way, merely building into your theory that the thing is necessary doesn’t help explain it."
Why does American TV look old?
I mean the colours. They are just strange, somehow oversaturated and yellow-ish. And it is not just TV, photography as well. I just realised that when I saw a photo with Trump showing his "golden dome". Trump looks like he's from the 1980s at best and so does Hegseth who stands next to him.
And from the short bits of US TV shows that I've seen it seems that all US visual media look like that (excluding films or high budget TV series, there it kind of depends on the director and all).
Is it that different colour grading is used in the US? If so, why? Why make it look like it was all recorded in the 1950s using technicolor?
When I see German, French, Spanish, Czech, British etc. TV (less sure about TV from Latin America, Asia or Africa but what I've seen also seems to be roughly equal), colours look "normal" and basically the same whichever country the show is from. But when I see the same from the US, it looks weird. And I am really talking about colours rather than the way people behave or dress.
I've never been to the US but I suspect that colours are in fact not different there than in the rest of the world :D
Because to an american it would look weird if it didnt. Longer tv tradition is such that if it doesnt keep a particular look, it looks weird. European tv looks weird to me. I am thinking later dr who seasons. I recognize that it looks more realistic, but its not supposed to be realistic. Its tv
I seem to remember that back when film cameras were a thing, color film was formulated differently for Asian vs Western color preferences.
It's because US TV and European TV use different broadcast and playback standards. In North and South America they use NTSC and in Europe and Asia they use PAL.
NTSC operates at 60 Hz and PAL at 50 Hz. This results in NTSC playing back at 29.97 frames per second, and PAL at 25 frames per second. But as far as color goes, NTSC is older than PAL, and was developed when the first color TV broadcasts occurred in the 50s. Most Americans still had black and white sets, so NTSC focused on staying backwards compatible so that color broadcasts could still show up well on black and white sets. PAL was developed significantly later, in 1966. The engineers who devised PAL wanted to make improvements from NTSC, which included a higher resolution and automated color correction (NTSC tvs had a color correction knob you had to manually tune in order to fix color errors) and a higher color subcarrier frequency. As a result NTSC is more likely to have color errors, and has a higher saturation.
Now even though everything has gone digital the old NTSC and PAL standards have affected modern TVs, especially for broadcast TV (which still wants to be compatible with people using old analog cathode ray tv sets).
Anyone here read 'More Everything Forever', by Adam Becker? I feel like there should be more effort to look at criticisms of rationalism and either learn from them (if merited) or find counterarguments (if not).
The latter sentence reads as a non sequitur to the former sentence to me, who hasn't read the book. Can you elaborate?
It's an anti-rationalist book, and recommended on Sneer Club. I'm only a quarter of the way through, but was wondering if anyone had read it and had comments.
"recommended on Sneer Club"
That would be the kiss of death for any recommendation for me. Okay, let's have a look at the blurb:
"More Everything Forever: AI Overlords, Space Empires, and Silicon Valley's Crusade to Control the Fate of Humanity"
Ah, I see Mr. Becker is setting out his stall from the get-go.
"This "wild and utterly engaging narrative" (Melanie Mitchell) shows why Silicon Valley’s heartless, baseless, and foolish obsessions—with escaping death, building AI tyrants, and creating limitless growth—are about oligarchic power, not preparing for the future
Tech billionaires have decided that they should determine our futures for us. According to Elon Musk, Jeff Bezos, Sam Altman, and more, the only good future for humanity is one powered by technology: trillions of humans living in space, functionally immortal, served by superintelligent AIs.
In More Everything Forever, science journalist Adam Becker investigates these wildly implausible and often profoundly immoral visions of tomorrow—and shows why, in reality, there is no good evidence that they will, or should, come to pass. Nevertheless, these obsessions fuel fears that overwhelm reason—for example, that a rogue AI will exterminate humanity—at the expense of essential work on solving crucial problems like climate change. What’s more, these futuristic visions cloak a hunger for power under dreams of space colonies and digital immortality. The giants of Silicon Valley claim that their ideas are based on science, but the reality is darker: they come from a jumbled mix of shallow futurism and racist pseudoscience.
More Everything Forever exposes the powerful and sinister ideas that dominate Silicon Valley, challenging us to see how foolish, and dangerous, these visions of the future are."
Profoundly immoral racist pseudoscience. So are we racists now, Father? What is it with science journalists and hit piece books? Who is this Adam Becker:
"Adam Becker is a science writer with a PhD in astrophysics from the University of Michigan and a BA in philosophy and physics from Cornell. He has written for the New York Times, the BBC, NPR, Scientific American, New Scientist, Quanta, Undark, Aeon, and others. He has also recorded a video series with the BBC, and has appeared on numerous radio shows and podcasts, including Ologies, The Story Collider, and KQED Forum. He lives in California."
Okay, so he got a science degree then pivoted into science writing for whatever reason (possibly lack of jobs in academia). His Wikipedia article bigs his bio up a tad:
"Adam Michael Becker (born 1984) is an American astrophysicist, author, and scientific philosopher" but his work seems to be more that of science populariser by writing "lay-friendly" books on scientific topics for us dumb normies.
Apparently his first book was 'controversial' (I'll leave it to the quantum physicists on here to tell me if it was a hot topic of debate in the field):
"Becker's second book, More Everything Forever, takes a step away from the controversy of his first book, What Is Real?, and instead explores the relationship between science and the consumer tech Industry that has evolved and been promulgated across the world from the Silicon Valley of California"
Eh. Just sounds like he's copying Cade Metz and jumping on the bandwagon of what is the topic du jour, which happens to be AI. Also, bashing Silicon Valley will always get you a good review in the NYT 😀
https://freelanceastrophysicist.com/
"I never wanted this book to be as timely as it is, but here we are. If you’re interested, please buy it now, or put it on hold at your local library. I think it’s pretty good, but you don’t have to take my word for it — the New York Times called it “smart and wonderfully readable.”
My first book, What is Real?, is about the unfinished quest for the meaning of quantum physics. The New York Times liked that one too; they called it “a thorough, illuminating exploration of the most consequential controversy raging in modern science.”
It sounds like a rehash of the old TESCREAL critique, viz:
1. Here are some vaguely related things that I mashed together
2. Some of them are bad
3. Therefore, all of them are bad
It's not interesting enough to engage with specific points beyond that.
It's not hard to read between the lines and figure out that this critique comes directly from the doctrinaire political left, and that the underlying critique is actually just "this is not leftism, therefore it's terrible". It's a bit like a critique of the Rolling Stones written by someone who hates everything except K-Pop; whatever the weaknesses of the Rolling Stones may be it's unlikely that the die-hard K-Pop fan is the right person to point them out.
So Trump used chatgpt to decide how much the tariffs should be. So in a sense AI is controlling the government. And ruining the US economy. But not because the AI hypnotized the president. Or outsmarted him. It's just the president is really lazy.
On the gripping hand, it shows that letting AI run the economy is not a good idea, so score one for the "put the brakes on AI" movement!
Letting capitalists run the economy, or neolibs run the economy, or economists run the economy is bad enough, getting their ideas fed back all chopped and screwed by the stochastic decent pachinko machine has gotta be an order of magnitude worse.
The alternative is to have the stochastic pachinko machine feed itself. The capitalists basically want that, and I doubt capitalists are an order of magnitude worse than themselves.
>Some significant issues (probably? still looking into it?) with my post replying to Marginal Revolution last week, I’ll hopefully post something more detailed soon.
Don't let Cowen gaslight you, it's on him if he made a short post with confusing and ambiguous language that made everyone who read it think he was saying something he has now decided he didn't mean at the time.
His own commenters read it as support for Rubio's claim, I read it that way, I think any plain-text reading not relying on special pleading would read it that way. If he actually had a more nuanced argument it was his job to make it explicitly and clearly, and he's morally responsible for saying what he said to his audience even if he misspoke.
I study dentistry in Iraq—but my mind doesn’t.
It thinks about time, particles, patterns, life.
Not to pass exams, but to understand.
But the system says:
“Wait. Follow the steps. Don’t ask too much.”
I ask anyway.
If I leave university, I lose structure.
If I stay, I lose freedom.
My passport doesn’t open borders.
My questions don’t open classrooms.
What would you do, if you had two years left in a rigid system,
but your mind was already outside it?
Advice is welcome—not just for success,
but for survival of thought.
Slog through the last two years. You will get further having a qualification to fall back on than dropping out part-way through to ponder the big questions of life, the universe, and everything. Nothing stopping you qualifying as a dentist and still pondering the big questions.
Thank you — that’s fair advice. I’m trying to hold onto both: completing the degree and keeping my curiosity alive. It’s a difficult balance, but I appreciate your perspective.
In my experience most people who feel drawn to "big, deep" questions are only drawn to them as an escape from dealing with the realities of actual life. Big questions have innumerable possible answers and are, at least in a practical sense, frequently unanswerable. That can be attractive to someone who just wants to sit around all day daydreaming with no accountability. Sitting around fantasizing about the meaning of life sounds a lot better than the drudgery of homework, exams, and learning about something that has actual wrong answers. My suggestion is to consider the possibility that what you perceive as the siren song of basic science is just a psychological rebellion against an imminent adulthood that you don't want to face. Everything has elements of drudgery, rigidity, and tedium to it but those elements just aren't apparent when you look at them from the outside. My advice is to take up the challenge of finding intellectual stimulation on the path you're already on. Dentistry is ripe for innovation. Think of ways to improve old techniques, build new tools, or reframe existing dogmas.
Thank you for your thoughtful reply. I understand where you're coming from.
But I want to offer a gentle correction: my draw toward “big questions” isn’t a way to escape life—it’s a way to face it more honestly.
When I study the human body, I don’t just want to memorize that a protein triggers a reaction—I want to understand why it behaves that way, how its structure guides its function, and what deeper principles govern those interactions. I’m not avoiding reality—I’m trying to understand it more rigorously than what my lectures demand.
It’s not the exam I fear. What frustrates me is that I’m asked to memorize effects without understanding causes. I want to train my mind to see through the layers of what we call “facts,” down to the structure beneath them.
You’re right that every path has its tedium. I’m not denying that. But I believe true education should also train us to think—not just to repeat. I’m not running from adulthood—I’m trying to meet it with a sharper, freer mind.
If we had space in the system to ask better questions, we might produce more meaningful answers. That’s all I’m trying to do.
Thanks again for your honest words—they gave me a chance to clarify mine.
Thank you for the beautiful poem!
I don't know if that's an option where you are, but if I was you, I'd try to find someone to teach me among the department professors. Have you tried talking to any of your professors, to see if they would give you some of their time and some pointers to help reach understanding, or perhaps even take you on as their student? (Some would probably appreciate audience.) Or is this completely inappropriate where you are?
Thank you so much for your kind and thoughtful suggestion.
I did try — several times.
When I was in school, I thought university would be the place where I could finally ask the questions that haunted me: How do biological phenomena emerge from molecular behavior? What governs the logic behind protein shapes, interactions, and transitions?
But when I arrived, I found that most professors didn't welcome those kinds of questions. They often told me to "wait until grad school," or "just focus on what's required for the exam." Some even discouraged me from reading primary sources, saying I should rely only on summaries.
I tried visiting other universities, requesting access to libraries, asking to shadow in labs — but I was often turned away for being “just an undergrad.”
So I began alone. I built a small home incubator, studied bacteria from extracted teeth, read biophysics, quantum mechanics, and systems biology on my own. Not as rebellion — but because my questions couldn't wait.
Sometimes, I wondered if I was wrong to walk alone. But comments like yours remind me: perhaps I’m not wrong — I’m just early. And I still believe there are minds out there who see what I’m trying to do and might be willing to point a way forward.
Thank you again for offering that kind of attention. It means more than you know.
A bit of self promotion: I recently had a piece on understanding AI through the lens of topology, manifolds, and geometric spaces. I know a bunch of folks here have a passing interest in neural networks and how they work, so I figured I'd share: https://theahura.substack.com/p/deep-learning-is-applied-topology
The piece is intended to be a fairly high level intuition building article. No prior expertise required. If you have a background in topology, I apologize in advance for switching between technical terminology and colloquial terminology without clearly explaining which is which
I thought I was likely the right kind of reader for what you wrote. In high school & college I had math through one yeur of calculus, enjoyed it and understood it well. But I majored in a humanities field and haven't learned any math since calculus, except some pretty easy-to-understand statistics used in social science research. But I had to leave off reading your piece early on because there was a leap I just couldn't understand. It was the illustration for cleanly separating data that's in the shape of a donut by moving the analysis to a higher dimension. The donut data in that illustration of that is completely unrecognizable. It's the illustration with a bunch of dark blue dots, and a sort of tree made of red dots sprouting out of the upper left surface of the blue mass. After staring at it for a while I realized I was looking at a 3D image, and the lines on the side were x, y and z axes. (But at first glance, I thought you were graphing on a weird 2D graph that was not perfectly square. You need make it clearer that the lines are the 3 axes.)
But even after I figured that out, I was puzzled. How can the donut now look like that? Before we were looking at it from directly above. Are we now viewing it in profile, with donut at eye level? If we were, I can see how the donut would look like your illustration. If its height was extremely irregular, with the irregularity including a red tree-shaped thing sprouting out of the top on one place, then it would look from the side like your illustration. But in the image on the page we don't seem to be looking at the donut in profile, but from above and to the left . . .
Anyhow, I'm all for using things like topology to help people have an intuitive understanding of gradient descent, but I think your effort needs some tweaking to work.
Thanks for the feedback!
The intent behind that section is to make the claim that data that seems inseparable in N dimensions may be separable in N+1 dimensions. The donut in this example isn't a top down view of some 3D data.
Rather, it's just a 2D plot. No matter how you twist and turn, that data will stay inseparable.
But if you project your data into a 3D space, suddenly it becomes separable.
I left this out because I thought it too in the weeds, but in neural network terms, the way this works is by having a neural network with 3 dimensions in it's hidden layers instead of just 2 dimensions. Even if your input data is 2D, if the intermediate layer is 3D, the model will project the 2D data into 3D space.
Lmk if that's clearer, and if so, what would recommend I change to get that concept across?
I don't understand projecting 2D data into 3D space -- well I get the concept, but it doesn't help me with big-picture understanding of what you are saying. It seems to me like if you only know 2 things, x and y, about each data point, projecting the data into 3-D space just gives you a lot of columns (3D version of dots) instead of a lot of flat dots. Unless you have a third piece of info, z, for every data point, representing them in 3 dimensions doesn't put you in a position to separate the dots cleanly.
Maybe it would work better if you had a concrete example. Let's say there's a big column of water rising from a hot spot on the ocean bed, and from above water level you see a big big circle of fish, with one species mostly more towards the center, one more towards outer edge. But if you look at the column of water from the side you see that taking into account distance from the bottom of the ocean allows you to discriminate more species of fish.
I *think* what's really going on with neural network is that each point is a word, and there are in fact hundreds or thousands of dimensions on which the word has a place, but you don't specify in advance what they are. And the neural network process doesn't ever *tell* you what all the dimensions are, it just ends up knowing them itself in some wordless way that's inherent in its structure. I realize this may not be right, though.
I have a model of this in my head, but it grew by itself and is based on my own only partially accurate layman's understanding of neural networks. I doubt it's a good fit for what you're trying to describe. But it is a good example of a model that makes intuitive sense to people, so maybe it will give you an idea. Say you were making a one-of-a-kind mattress for somebody, one that took into account their body shape, and weight of different cross-sections of them, so that when the person lay on it on their side their spine was perfectly straight. And also you wanted different parts of it to be softer than others, with increased softness in places where the person had aches or tender skin or some such. So you would work in layers. Bottom would be springs, arrayed so that each stood straight up with springs height and weight adjusted so that springs under heavier cross-sections pushed up more than springs under lighter ones. Then on top would be foam rubber of different levels of compressibility. Middle ones would be to neutralize errors in the springs' matching with cross-sectional weight, and to absorb changes when the person moves. Upper ones would be for softness to touch.
> I don't understand projecting 2D data into 3D space -- well I get the concept, but it doesn't help me with big-picture understanding of what you are saying. It seems to me like if you only know 2 things, x and y, about each data point, projecting the data into 3-D space just gives you a lot of columns (3D version of dots) instead of a lot of flat dots.
Right, ok, so lets start with this. Lets say you have as input a bunch of x/y coordinates. Like, [0, 1], [1, 0], [1, 1], [0, 0]. Using the playdoh metaphor, you can twist and turn and move those points around all over the place. That's what neural networks do.
Now lets add a third dimension. We dont have any obvious value to start with for that third dimension, so we'll just set it all to 0. So your dataset now looks like:
[[0, 1, 0],
[1, 0, 0],
[1, 1, 0],
[0, 0, 0]]
It's a bit harder to conceptualize, but these points now live in a volume. It's a flat volume, but still a volume. And again, using the playdoh metaphor, you can twist and turn and move those points around all over the place _in that volume_, including in the z axis.
For example, you could come up with a transformation where everything with an x value of 1 is raised up while everything with an x value of 0 is pushed down.
[[0, 1, 5],
[1, 0, -5],
[1, 1, 5],
[0, 0, -5]]
This gives you more ways to separate the data than when you started. So a neural network can learn a function to separate data in 3D, even if we only know [x, y] as input.
> fish
I think the thing I want to get across here is that the neural network is also _doing_ the transformation. Implicit in the transformation of the input by the neural network is that there is some additional structure that can be extracted, so in that sense the fish example is closer to the intent. Will chew on it
> I realize this may not be right, though.
No, I think you are correct for LLMs. Each token is represented by a vector, which has the high dimensional properties you describe ("inherent to the structure")
> mattress
This analogy is a great way to explain the different kinds of information processing that is happening in a multilayer neural network -- early layers do more input processing while later layers do more refinement.
OK, I understood everything you said this time around. Also, I understood something that's made your whole explanation hard for me to wrap my mind around, and that I think would affect most people who know even a little about data analysis the same way. I think of points as data, established facts, each having a distinctive place on one or more dimensions. So I already felt kind of disoriented right at the beginning, when you showed dots scattered on a surface, with no indication in how they were distributed that there was an obvious division possible. So then you showed the surface warped, so that there was an area of empty space between 2 clusters, and said, see, if you warp the surface you can divide them. And I'm thinking, WTF -- that's called massaging the data.
Seems like the process of training LLM's is radically different from the process of finding regularities in data you already have. With LLM training you start off with arbitrary values, then throw a bunch of staff at the matrix and assign values to each data point based on how they fit with the stuff. And you keep repeating that till the fit is good. (And I *think* assigning values has to do with reducing total error somehow -- the totality of how far off from a good fit all the initial arbitrary z values you assigned are). So it's data massaging, but a whole different kind --data massaging in the service of getting the right values for all the z's. You need to strongly signal that to the reader, maybe say something like that outright, otherwise they will be confused because it seems like you can't learn anything if you get to just change the information you have about the data.
There has to be some physical process that's a really good analog of what's going on, but I don't know enough about machine learning to come up with one. But here's an idea that keeps topology as a way of thinking about neural nets, and that maybe you can adapt so that it represents the actual training process correctly. Say you have a trampoline and you want it pressed downward in a such a way that is a perfect fit for some large irregularly shaped object. The plan is to put stuff on the trampoline to press it down into the required shape, then freeze it so it stays that way. You can't put the actual large object on the trampoline because it can't be moved, and anyway depressions in the trampoline are not the same shape as the thing pressing on them. So what you do is divide the trampoline into a grid, and in the middle of each square you put a weight, then you assess now close the depression of the trampoline is to the shape of the target object. And you keep adjusting the weights, and experimenting with connecting some of them in pairs, etc.
Yea, training a neural network is definitely different. The way we decide whether the model is fit correctly is based on some kind of labeled dataset. So in the toy example in the article, you could imagine that the "task" is labeling points 'red' or 'blue'. Turns out you can look at the input properties and from that discover a way to massage the data to identify cleanly when something is red and when something is blue.
The massaging the data point is interesting because there's lots of ways in which letting the neural network "massage the data" results in the same kinds of problems that may occur if a person does it -- over fitting, most commonly, where the model "over fits" on noise in the training set and fails to work in more general settings.
Need to think about how to convey the idea that data isn't 'stable'. Thanks for the thoughts and the example. The trampoline analogy seems useful, will chew on it.
The Economist finds no evidence of rising technological unemployment in the U.S.
https://www.economist.com/finance-and-economics/2025/05/26/why-ai-hasnt-taken-your-job
Of course not, it's still too early. AI is still being rolled out to commercial and private users, and it's not yet ready to be let loose on its own (there is still room for people to hand-hold the bots through tasks).
Give it a year or five, then if the Economist article is still being written by a human instead of by AI, we can say there is no evidence of rising technological unemployment.
7. I wouldnt worry about it. I read his new post as apologizing for his old bad post and fixing it but in a know-it-all curmudgeon who never apologizes sort of way. How often does TC dedicate a post to a direct criticism after all?
Hoping someone here can help me out. Like Scott, I'm a fan of the Sid Meier Civilization game franchise, and I've been doing volunteer work on the open-source free software clone called Freeciv (www.freeciv.org). I've been creating Macintosh versions but I can't test them on the newer versions of the MacOS, version 12 "Monterey" is as far as I can update. Apple is notorious for lack of backward compatibility, so I'd really like to know if my builds will run on a newer MacOS. If you have a Mac running MacOS 13 "Ventura" or later and would like to help, email me at my burner email macwiz23 {at} comcast {dot} net. What's involved is downloading a file from dropbox, un-zipping it, and trying to run it.
HEY SCOTT WE WANNA SEE THEM REVIEWS!
Patience, patience; you have to eat all your vegetables up first before you get dessert 😁
Is Christian rationality coherent?
Given the hard-line atheism of Yudkowsky et al, I've long thought no. But I've since met a few people who identify as both Christian and Rationalist, and this has made me think it could be possible after all.
If Christ really did rise from the dead, would it be rational to accept all of Christianity? At what probability threshold for your belief that Christ rose from the dead should you convert?
A big source of tension is rationality's reliance on utilitarianism. Christianity isn't utilitarian by any stretch.
I've been talking about rationality the intellectual movement more than rationality the technique here, but either way the question is interesting.
It’s certainly possible to be Christian and rational, I’m not sure about how many people would identify as Christian rationalists.
I’ve long thought the hardline atheism of Yudkowsky is unjustified. He writes so eloquently about how to arrive at true beliefs and avoid misplaced certainty, but has a blind spot on religion which he is very sure is false.
But the truth of religion cannot be perceived through logic and reason - as Kierkegaard wrote it requires a leap of faith. Yudkowsky holds the comprehension of the human mind as the highest truth, so he flatly refuses to believe in something that the human mind cannot comprehend.
> But the truth of religion cannot be perceived through logic and reason - as Kierkegaard wrote it requires a leap of faith. […]
That's true, yes, but you haven't presented a single reason why a rationalist – or any person – _should_ take that leap of faith, and more importantly, why they should leap to that particular variant of faith.
Why *should* one believe in God? This is not really an answerable question. In my personal case I came to Christianity through a search for community and truth. But different people will have different reasons. And there is no ultimate reason why one *should* do anything; it depends on your personal goals and values.
Pure truth-seeking will never, ever lead you down the path of faith-based belief. You need some other corrupting principle or motivation in there.
How sure are you about that? Even John von Neumann, smartest man to ever live and first to predict the possibility of superintelligent AI, converted to Christianity on his deathbed.
The other issue is whether an entirely "faith" free approach... a belief system without any unjustified axioms or appeals to intuition ...is possible. If not, then the "should" is a "must".
I don't think *any* belief system can get by without any unjustified axioms. Rationalists hold truth to be intrinsically valuable, for example. Utilitarians hold that more utilons = more good. Science holds that empiricism is correct and radical skepticism is wrong.
Any system of morality has to make a leap of faith to cross the is-ought gap.
Rationalists take leaps of faith all the time: they understand it better than most, arguably, because they're strict Bayesians. Every time you believe something with less than 100% probability and choose to act as if it is true, you're taking a leap of faith.
I think it depends on whether one identifies more as Christian or as Rationalist (large "R" intentional here). Hard-line atheism is definitely one identifier of early Rationality (I don't know what the emphasis is like today) so religious belief of any degree is seen as being too dumb to think until you cast off all the falsehoods and get ready to learn about reality as it is (i.e. scientific world view).
Small "r" rationality/rationalism/rationalist seems more compatible. And then from the other side, a Christian may or may not apply rationality to varying degrees depending how much truth/value they see in it as a philosophy.
Rationalism doesnt have much of an argument for utilitarianism, though. There 's an important difference between what rationalists tribally believe, and what they can prove.
I don't see any reason it's incoherent, though you'll probably fail to resemble at least one of the modal rationalist or modal Christian.
I suppose in part one's answer to this is about where one draws the boundary around both rationalism and Christianity.
That really depends on what you mean by "rationality".
If you mean something like "do not accept special arguments for religion that you wouldn't really accept for other things", then it's difficult.
On the other hand, if you mean something like "follow the consensus of high-status people around you", then it's simple.
Christianity is quite compatible with "rationality" in terms of reason, logic, etc. I don't believe it is compatible with "Rationality" as an intellectual movement, though I think there is a lot of overlap.
As you point out, the main disagreement is on ethical theory. Utilitarianism seems to be pretty solidly embedded in Rationality as a movement. Things that I do think overlap: the Litany of Tarski, the Litany of Gendlin, the general concept of objective truth, that human minds are flawed but are capable of identifying truth through reason, and that death is bad. Christians don't think death is as bad as Rationalists do, because everyone will be resurrected and have an afterlife, but they agree that "eternal life>death".
I'm not personally a Christian but I have a level of rational respect for the faith. I think of it as social technology. The supernatural claims are design features that a) align well with human psychology and b) serve as an intellectual tithe to community building. That's a fancy way of saying that it's a Santa-like useful lie, though I feel like that comparison is facile because the structure of Christian belief is much more sophisticated than a children's story. It has survived in close symbiosis with Western society for 1500 years and in my view has played a central role in maintaining cultural stability. The New Atheists are naive when they dismiss religion for making easily-disproven faith claims. They misunderstand the function of those claims and therefore analyze them along an axis that's orthogonal to their purpose.
In my view religion is the Chesterton's Fence of civilization and the thing that it fences us out of is a defection cascade in the social version of the Prisoner's Dilemma. Cooperation makes society work. Committing to that means not single-mindedly pursuing your individual interests all the time, particularly when it would screw someone else over to do so. Being willing to publicly commit to a truth-claim that you know is objectively false is a socially-expensive signal and indicates that you're willing sacrifice your selfish interest for the greater good. "I'm ignoring the rational arguments my brain wants to make in the interests of social cohesion" is a sacrifice and it has resonances with "I'm not going to double-cross this person even though I could get away with it." In a similar vein, I don't think financial tithes are cynically designed to enrich the Priesthood. I think they're carefully crafted to make individual community members feel invested enough in the community to maintain it. You don't arbitrarily walk away from an institution if you've poured thousands of dollars into it. That dynamic is essential for building stable communities.
> Being willing to publicly commit to a truth-claim that you know is objectively false is a socially-expensive signal and indicates that you're willing sacrifice your selfish interest for the greater good.
Very interesting viewpoint. Especially because I'm lately wondering why smart people make strong claims about how logically obvious it is that God exists, when I think it's logically obvious that he doesn't -- I mean, not as anything else but a very powerful piece of matter, if at all.
I have a question though.
Since when are people doing that? Since 2000 years or only since about 100 years?
Good question. I suspect that some people have always consciously done that, at least since the Enlightenment. Thomas Jefferson famously published a version of the Bible that had all of the supernatural elements edited out of it. Machiavelli wrote that rulers should use religion to keep men honest. My guess is that the notion always had currency among the intellectual elite but was kept hidden. Most believers obviously buy the story, though you have to wonder what level of internal denial is going on. Before 100 years ago it was plausible enough though IMO it doesn't even have to be conscious to work. Humans are hard-wired to take some things on faith. We sort of have to be when you think about it: the world is too complicated to figure out rationally from first principles, and that was especially true 10,000 years ago. I think our brains evolved a faith mechanism in order to bind us to a credible authority figure. If someone is older and successful then your survival odds will probably go up if you just listen to his advice.
People have analyzed progressivism through the same lens, btw. The argument is that the more absurd claims about gender and oppression serve as signaling beliefs. Much like a peacock's tail, the more your group can believe unpopular things and still dominate the public dialogue the more de facto power that means it has. In my view it actually serves multiple purposes: it differentiates the group, provides a polarizing belief to credibly signal group adherence, and signals group dominance. And I suspect one reason that it's been so popular among the elite is that it also signals individual wealth: it implies that you're wealthy enough to physically and socially isolate yourself from the nonsense you advocate, or at least that you have better things to do than to fight against it. People who spend lots of energy making careful principled arguments about how progressivism is either insane or unfair are *really* signaling that they're the kind of person who needs to care about the world being fair or making sense. In other words that they're middle class. The wealthy don't care about the world making sense: they just throw a bag of money at it and go back to their supermodel harem. I think that's an underappreciated status valence that progressive beliefs have and I think it explains a fair amount of their social success. High-status men couldn't really stand up to it without implicitly de-statusing themselves.
(Incidentally that provides a good social stability argument both for monogamy norms and wanting your leaders to be married. Men who chase pussy aren't sufficiently insulated from ideological temptation to stand on principle. If you create social norms that prevent leaders from chasing women then you get better leaders. It's the political equivalent of making sure your judges don't have a financial stake in the cases they decide. Men can't really make good decisions when they're trying to get laid.)
Bret Weinstein makes an argument that smells a lot like this. His catchphrase for religion is that it's "technically false, but metaphorically true". A learned person wishing to convince fellow humans to borrow on past wisdom to save the time spent otherwise acquiring it could bewilder and confuse a bunch of nomadic hunter-gatherers or exhausted farmers and ranchers or hormone-charged warriors with lectures on epistemology and scientific inquiry and the economic coordination problem, or could just say "honor thy father and mother or you'll go to hell when you die" and sell the idea every time scary weather comes around. Likewise for not marrying cousins, not eating pork, not stealing, and so on.
It also rhymes with Jordan Peterson's sentiment: Peterson would claim that it doesn't matter whether he believes in God, but rather whether he behaves as if he does.
I'd argue that giving up on promoting specific claims of fact in order to preserve social cohesion is itself a rational act, not irrational. The process of rational inquiry is itself a cost; even a rationalist need not pay that cost every time it appears.
Well said, I totally agree.
I think a neatly cynical way to reframe the emergence of New Atheism is to say that what they're really illustrating is that the social mechanisms of defection enforcement no longer exist. What they literally say is "the Bible is factually false" but the reason they're able to get away with saying it is that religion no longer wields any social power. In that sense, then, God really is dead. He existed as a Schelling point in game theory strategy-space but disappeared once enforcement norms evaporated.
Put that on the Enlightenment's tombstone.
Not a direct answer to your question, but you reminded me of this 35k word https://www.lesswrong.com/posts/n3Q7F3v6wBLsioqt8/extended-interview-with-zhukeepa-on-religion intro from Ben Pace:
> Zhukeepa is a LessWronger who I respect and whose views I'm interested in. In 2018 he wrote the first broadly successful explication of Paul Christiano's research ideas for AI alignment, has spent a lot of time interviewing people in AI about their perspectives, and written some more about neuroscience and agent foundations research. He came first in the 2012 US Math Olympiad, and formerly worked on a startup called AlphaSheets that raised many millions of dollars and then got acquihired by Google.
> He has also gone around saying (in my opinion) pretty silly-sounding things like he believes in his steelman of the Second Coming of Christ. He also extols the virtues of various psychedelics, and has done a lot of circling and meditation. As a person who thinks most religions are pretty bad for the world and would like to see them die, and thinks many people trick themselves into false insights with spiritual and psychological practices like those Alex has explored, I was interested in knowing what this meant to him and why he was interested in it, and get a better sense of whether there's any value here or just distraction.
> So we sat down for four 2-hour conversations over the course of four weeks, either written or transcribed, and have published them here as an extended LessWrong dialogue.
35k words is too much to ask of most readers of course, so here's zhukeepa's 2.2k word version: https://www.lesswrong.com/posts/X2og6RReKD47vseK8/how-i-started-believing-religion-might-actually-matter-for it begins like so:
> In Waking Up, Sam Harris wrote: "But I now knew that Jesus, the Buddha, Lao Tzu, and the other saints and sages of history had not all been epileptics, schizophrenics, or frauds. I still considered the world’s religions to be mere intellectual ruins, maintained at enormous economic and social cost, but I now understood that important psychological truths could be found in the rubble."
> Like Sam, I’ve also come to believe that there are psychological truths that show up across religious traditions. I furthermore think these psychological truths are actually very related to both rationality and moral philosophy. This post will describe how I personally came to start entertaining this belief seriously.
I reallly like the book reviews here, excited for the not book reviews too, but there’s a lot of books that seem way too big a time investment for me, however a book review feels perfect, getting to the point fast and even providing some reflection of the content. Do you guys know if any places where I can read acx like book reviews instead of books?
https://www.thepsmiths.com/
Is what you’re looking for.
> Do you guys know if any places where I can read acx like book reviews instead of books?
Martin L Morgan has given the *real* answer - the psmiths are magnificent, without parallel.
I aspire to be a "deci-psmiths," and review a number of books in the style of Scott (ie with external data and papers pulled in, random connections to other ideas, overall thoughts and takes on the implications, and so on).
Here's a link to an index of books I've reviewed, currently at around 45 books, it's roughly 2/3 non-fiction books and 1/3 elite coaching / performance / training books:
https://performativebafflement.substack.com/p/eebaae81-722a-4607-be1e-60836b7225b7
I'm just hearing about it now, but China claimed all of Bolshoy Ussuriysky Island from Russia last year, and Russia hasn't squawked. Bolshoy Ussuriysky sits at the confluence of where the Ussuri River joins the Amur River. Russia and China had a treaty that divvied up the island between them. Russia uses both rivers to ship lumber, minerals, etc., out of Siberia down to the Pacific Ocean. The island is a strategic choke point for river shipping on both Rivers. China published its new national map in 2024, and it shows China as owning the island. And recently, China announced that it has started "developing" the island.
I suspect...
1. China perceives that Russia is distracted by its Ukraine fiasco. And Ukraine is sapping the strength of its military in the Far East.
2. If the Putin regime collapses and separatist elements in Siberia decide to secede, China sees an opportunity to reclaim Outer Manchuria, which it has historically claimed. Russia and China have been arguing on and off over it for the past 400 years.
3. Theoretically, China could hop over from Bolshoy Ussuriysky and grab Khabarovsk (Russia's largest city in the Far East), force gunboats down the Amur River to the Pacific, and cut the district of Khabarovsk Krai in half and take all the district of Primorsky Krai from Russia. Primorsky Krai would give them Vladivostok, which is Russia's largest Pacific port. China would then have access to the Pacific Ocean without having to pass by South Korea, Japan, and the Philippines.
Sun Tzu said: "All warfare is based on deception. When we are near, we must make the enemy believe we are far away. And when we are far away, we must make the enemy believe we are near. Let your plans be as impenetrable as night, but when you move, move like a thunderbolt."
Is China's saber-rattling over Taiwan just a distraction to keep the world's eyes off of what they're planning for Outer Manchuria?
Interesting, thanks! I did know that the border between China and Russia had been somewhat disputed, but this is a new level. Though it falls into the general pattern that China is extremely aggressive in claiming islands from all its neighbors. Do we have reason to believe that they consider islands in a fundamentally different way than continental territory?
I imagine it's testing limits. An island is separate territory that can be argued over as to which mainland it belongs to, and if the pushback is too great, then backing off is less humiliating than claiming a chunk of mainland and then having to fall back. "Oops, so sorry, guess our maps weren't drawn up right" saves face all round.
True. However, those lands were previously subject to China before the Qing were forced to cede them to Russia in 1859 and 1860, another low-point in the "Century of Humiliation" (i.e. the 19th Century) that remains a stinging sore point on Chinese national identity. And Xi has referred to himself as a "strong emperor". In Chinese history, strong emperors are the ones who reestablish Chinese dominion over lost territories and/or expand China into new territories. I'm sure Putin is aware of this, which makes the lack of fuss over this little island even more striking.
Remember, if China does make a move up there, you heard it from me first. ;-)
I wonder if Putin has some kind of trade in mind? "Okay, I give you back your island, you give me...." what?
I imagine you are correct that Putin has his hands full right now with Ukraine and the USA. He's also playing at being Great Pals Altogether with Xi Jinping right now, since it suits the purposes of both Russia and China to present a unified front against the USA.
But I can't see him letting this go, especially if his advisers are telling him the same thing about Outer Manchuria that you state here. As and when it becomes time for the bear to slap the dragon down, that island will be reclaimed. Letting China do what you propose (grab a city, move gunboats down, grab more territory) is not something Putin, Russia or indeed the rest of the world wants to happen.
China is not allowed to use any of the rivers that go out to the Sea of Japan. This is because both Russia and North Korea think that if they allowed that the raw size of the Chinese economy and population would quickly dominate the region and in the long run threaten Vladivostok and Rason. So they cooperate to keep China out. China is very upset about this and is basically doing a minor version of the sea base building campaign but along the river. Sometimes they make noises about reclaiming the Amur River Valley. They also want access to the Arctic which Russia is working to keep them out of.
The US doesn't notice this because we don't have any allies that border China directly so it's all the naval stuff. But there's plenty of land based disputes. I think these are actually more likely to escalate because the balance of power is more in China's favor, China's land based military is better than its navy, and because Taiwan is a binary condition. If they attack Taiwan they take or they don't. In a land based campaign even if they fail they can claim they did what they wanted to do and declare victory.
India and Myanmar particularly seem very vulnerable to Chinese intervention right now.
How likely do you think it is that China could win a war against India with the goal of annexing land? I don't know anything about the Indian military, but I do know that the PAL is geared towards suppressing dissent rather than fighting wars. I also know the PAL is very big.
Considering that China moves the Line of Actual Control most years, taking a little land bit by bit up in the mountains, approximately 100%. A bigger but more limited war is definitely possible. I agree an all out war intending to occupy New Delhi or something would be unlikely unless circumstances shift significantly.
There’s two plausible locations for a limited war between India and China. You could have a ‘small’ limited war around the line of actual control in Kashmir. And you could envision a ‘big’ limited war in India’s northeast (east of the Siliguri corridor). I don’t think there’s any other plausible ‘limited war’ scenarios. (Ok, in an alternative universe where the balance of power was reversed you could imagine a limited war to liberate Tibet, but that’s not plausible in the world as is). Possibly you could also have a purely naval limited war in the Indian Ocean, but that wouldn’t involve any territory changing hands.
ETA: I suppose you could also have a ‘medium’ limited war over Bhutan. That one came pretty close to happening about a decade ago. But the most likely scenario is probably a continuation of the current salami tactics (which exist on a continuum with the ‘small’ limited war scenario above).
Yeah. The low end of realistic is continuing limited actions in Kashmir and along the border, small scale, mostly just pushing the line bit by bit. I think there's little chance they stop. The high end of realistic would be a coordinated campaign to push along all three areas (the McMahon Line, Kashmir, near Bhutan) perhaps alongside Pakistan in Kashmir intending to give India black eye and settle the disputes decisively in China's favor.
I actually think the most likely thing for China to get involved in is Myanmar. If the situation there spirals out of control and China feels that it no longer has good contacts among the rebels and that the government is at risk of becoming hostile or falling then it will be faced with the choice of losing the country or doubling down. If they do double down they would upgrade their advisors and supported militias to an actual PLA presence. They could undoubtedly secure an invitation from the government. And then they'd be involved in peacekeeping abroad. You could even imagine some enterprising rebels crossing the border and massacring a Chinese village to provoke an intervention.
But one reason I think India or Myanmar are more likely targets is that the military and political situation is more favorable. China has more advantages. There's a much smaller chance of ending up in a shooting war with the US. And Taiwan has a binary win/lose state: do they end up controlling Taiwan. If they do a large scale operation against India and India wins they can just say, "we gained 10 kilometers of vital (worthless) territory and showed India we were serious!" And because it's less politically sensitive Chinese people will buy that more easily.
And then there's chained ifs. If Russia collapses into a series of warlords then I could very easily see China moving into Russian territory. But that's predicated on how likely you think Russia is to shatter and personally I think that's pretty low.
Makes sense to me. The one thing I would add to this analysis is that the principal cost of a serious war (beyond the current salami tactics) is that it would presumably drive New Delhi off the fence and squarely into the anti-Chinese camp, which is a big cost. And in turn I rather suspect the main reason for New Delhi's careful West leaning but fence straddling posture is precisely the fear that too decisive a move to the West might immediately trigger precisely such a war. Which leaves us with the status quo where Beijing salami slices as aggressively as it thinks it can get away with without pushing New Delhi completely off the fence, and New Delhi shuffles westward as fast as it thinks it can get away with without provoking Beijing into escalating beyond salami tactics. Basically an attempt at frog boiling on both sides.
Incidentally, what about Vietnam (as a possible target for a `short victorious land based war')? It satisfies all your criteria above, with smaller diplomatic cost, plus no risk of nuclear escalation.
I can't find it now, but there was a documentary about the 2020-21 conflict between China and India in the Galwan River Valley. India was developing a road up into that area, and China objected. Two key takeaways i remember: (a) logistics are insanely difficult up there with limited access through high mountain passes; (b) the altitude takes a toll on both the Indian and Chinese troops (ChatGPT says the Galwan River Valley is at about 13,000 ft.
I don't think either China or India could invade each other via that route across the Himalayas. China also claims the entirety of Arunachal Pradesh Province in Northeast India as "South Tibet". The terrain along the border appears quite mountainous, and I suspect the logistics would be challenging for China.
Hmmm. It looks like China and Russia have an agreement that allows Chinese shipping down the Amur to the Pacific. This agreement may be relatively recent, though. From a 2024 article (link below the quote) "Russia, China Extend Far East Cargo Routes to Include Cross-Border River Transport: Amur, Ussuri Rivers give Pacific Ocean access to China’s landlocked Heilongjiang Province"...
> Zhou Hongcheng, deputy head of the Fuyuan Department of Commerce and Checkpoints, was quoted as saying that “We plan to carry out the first combined cargo transportation on the river-sea route through Khabarovsk. So, we can say that the Pacific Ocean has now appeared on our doorstep.”
https://russiaspivottoasia.com/russia-china-extend-far-east-cargo-routes-to-include-cross-border-river-transport/?utm_source=chatgpt.com
Here's a "List of China's Transboundary Water Agreements and Related Documents". North Korea seems to be included in some of the Amur River documents. Weird.
https://www.researchgate.net/publication/338839223_List_of_China's_Transboundary_Water_Agreements_and_Related_Documents
If I'm reading it right, this is permission for ships to travel in between Russia and China but not actually out onto the Pacific. But yeah, it looks like Russia gave it as a concession, perhaps a concession after needing Chinese economic support because of Ukraine.
China provides silicon chips and parts for weapons, missiles, and drones that no one else will provide to Russia due to international sanctions. Russia is in no position to jeopardize that supply.
Yes, and i’m they’re encouraging Putin in his folly. What was it Napoleon said? Don’t interrupt your enemy when he’s making a mistake.
Is Christian rationality coherent?
Given the hardline atheism of Yudkowsky et al, I used to think no. But recently I've met a few people who identify as both rationalist and Christian and have been rethinking this.
I guess if Christ did rise from the dead, Christianity is the only rational belief. But it's hard to evaluate the evidence in an unbiased way, because this belief is a marker of identity, not like other beliefs.
Another source of tension is that Christian morality isn't utilitarian-- not even close.
Have you met any Christian rationalists? Are the two belief systems inherently incompatible?
"I guess if Christ did rise from the dead, Christianity is the only rational belief"
I really don't see how this follows. Suppose I were to accept the following as axiomatic:
1. There was a guy named Yeshua ben Yosef who lived in 1st century Judea who claimed to be the son of God and laid out a slate of religious teachings and moral commandments on the basis of that claim.
2. After some years of preaching his word and gathering a following, this same guy was killed by the Romans and entombed, only to return to life three days later.
Regarding 2, let's specifically take it as given that it is the same mind (his personality and knowledge appear identical to everyone who knew him) and the same body (every physical detail matches exactly) and that he was absolutely, unquestionably dead.
As I see it, asking "is Christianity the only rational belief," amounts to asking "if I believe 1 and 2, does it follow that I should regard Yeshua ben Yosef's religious and moral teachings as unassailably correct?"
It seems quite obvious to me that the answer must be "no." To be clear: rising from the dead is an extremely impressive feat, and I ought to take any supernatural claims much, much more seriously from someone who has provably done so then I'd take them coming from some random guy off the street. But first, it is not so infinitely impressive as by itself to overcome every prior burden of improbability. And second, it is mostly a demonstration of power, not a demonstration of trustworthiness: him dying and rising from the dead doesn't preclude him lying to me. It doesn't even fully preclude him being mistaken: Archimedes accomplished some truly impressive displays of power despite (one assumes) only imperfectly understanding much of the physics behind them.
Rationally, even if we've accepted the resurrection, we certainly should not stop weighing Yeshua ben Josef's claims against other evidence. He's provided highly compelling (if somewhat nonspecific) evidence in favor of the worldview he lays out, but we may see lots of counter evidence if we keep our eyes and our mind open. And indeed, that's what we *do* see, looking at the world. It's not even fully clear at this point what all of his claims were: histories from his time have been only imperfectly preserved, with confusions and contradictions and differing translations and gaps in the record. That should count as evidence by itself: a supernaturally powerful God who is (by his own account) trying to spread His Word to all of humanity really ought to be able to arrange *much* better transmission and preservation of said Word. Already our demonstration of power is countered by a demonstration of either impotence (in not being able to spread/preserve his teachings better) or untrustworthiness (by choosing not to). Beyond that, there are many other things about the world that don't really look the way I'd expect them to if Jesus were the divinely ordained son of an all-powerful, all-loving God teaching the one and only true religion, which He wants all of humanity to accept.
Conversely, even accepting the basic facts of the resurrection, there are lots of other states of the world that could account for it besides Yeshua ben Yosef actually being the son of God. All of them are weird and improbable to one degree or another, but the base claim is also weird and improbable. Insisting that this one *specific* weird and improbable claim is the *only* one that can actually explain the evidence is definitely an example of starting from the bottom line and reasoning up.
Thank you.
This reminds me of some muslim I knew who told me he knew exactly what god wants because it's written in the Koran.
He did not get what the problem with that idea was. Or, and that's more likely in my opinion, he pretended to not get it.
Good point. The resurrection certainly would increase the probability of Yeshua ben Yosef's teachings being true. If someone teaches strange and bizarre ideas, such as that light and electricity are deeply connected, and then performes some feat like making a glass orb shine, you should lend more credence to their ideas than before.
But it doesn't increase the probability to 100%.
I think the religious leap of faith is still required, and that's probably the part of all religions that is most antithetical to the rationalist tradition.
>It's not even fully clear at this point what all of his claims were: histories from his time have been only imperfectly preserved, with confusions and contradictions and differing translations and gaps in the record.
This is commonly believed, particularly in atheist circles, but it doesn't really match the historical evidence. The New Testament works are without a doubt the best-preserved ancient writings we have. We have more copies of them from earlier dates than we have for just about any other ancient source, and they show that the content has been well preserved with minimal transcription errors. This is understandable, as Christians were highly motivated to make copies of the works, preserve those copies, and complain if new copies were different from the old ones (at pretty much all times Christians took matters of orthodoxy seriously, as seen by the numerous blow ups over heresy in the first 500 years of the religion).
Honestly, the fact that the works have been transmitted this well is evidence in favor of a supernaturally powerful God being behind the teachings they contain: though, of course, there are many naturalistic accounts that explain the same observed evidence (again, Christians are highly motivated to preserve and transmit the works and would presumably be just as motivated if the religion wasn't true).
Forgive me, I'm certainly not a Biblical scholar and can't claim any sort of intimate familiarity with the historiagraphical status of scripture. I did think even while I was typing that that the New Testament was in a considerably better attested than many other works, and I'll absolutely grant that the individual writings are better preserved than anything else from its time period. But "well preserved by the standards of ancient history" does not mean "well preserved in an absolute sense."
It's my understanding that while we can pin down the philosophy and teaching of Yeshua ben Yoseph more precisely than any comparable figure of antiquity, that is still much, much less well than we can do with more recent figures. Consider, for example, the teachings of Bahá’u’lláh, almost all of which were either transcribed approximately verbatim on the spot or written down by the man himself, in languages that many modern people still speak, and whose texts could be immediately copied en masse. Or if you'd prefer something secular, consider someone such as Nietzsche who both authored a large number of works himself and wrote many personal letters which have survived to this day.
By comparison (and please do correct me if I get any of this wrong) my understanding is that the writings of the New Testament:
1. Are accounts of claimed eyewitnesses reporting on Jesus' life and teachings, rather than anything that he himself is supposed to have written or dictated.
2. Are written in a language that is no longer living (though it has been quite well preserved), which is also not the primary language Jesus is believed to have spoken.
3. Generally post-date the events they purport to record by more than a decade[1]
4. Were assembled into the current widely-accepted canon over the course of several centuries, leaving a number of contemporary works out of the canon.
All of this leaves much more room for human interpretation and judgement to shape the way modern people answer "what were the teachings of Jesus?" than there would be for Bahá’u’lláh or Nietzsche.
And I'm sure you'll tell me that isn't a fair comparison: one lived two millennia ago, long before the printing press, teaching in the midst of a hostile culture that strove to suppress his teachings. And all of that is true. But when the question I'm supposed to be answering is "do I believe these teachings were specifically given to mankind by an all-powerful God whose most important directive to His children was to spread them to every corner of the world," it honestly seems *quite* silly to grant much weight to things like linguistic drift, low literacy rates or the high expenses of creating and preserving texts when I judge how well-attested those teachings actually are. If you can multiply loaves and fishes, why not scrolls and tablets? If this material is so important, why don't we have it in Jesus own, original words? Why are both the language that Jesus spoke and the language of the scriptures dead? Surely God could have prevented linguistic drift if He so desired: having the most scripturally relevant languages still spoken in everyday use, perfectly preserved over 2000 years would be a *much* classier and more impressive miracle than scaping up a bit of extra supper. To me, living in 2025, it is *quite* the tough pill to swallow that supposedly the single most important lessons in all of existence--lessons explicitly meant for everyone who lived after--worked in all manner of supernatural razzle-dazzle to wow the live audience, but none at all was spared to make it even *slightly* easier for the literally billions of people who couldn't be there in person. Simply pointing out that loads and loads of merely-human effort went into preserving these particular texts is substantially question begging: we wouldn't be discussing them at all if they hadn't been important to particular groups of *humans.* The question is whether they were important to a non-human supernatural entity of *vastly* greater means, and this doesn't show that at all.
And yes, I know I'm doing the Annoying Internet Atheist thing where I demand arbitrary miracles as a price of belief. Feel free to do the complimentary Annoying Internet Christian thing where you act as though any intervention that could possibly convince me is clearly and obviously beneath God's Almighty dignity. I won't mind ;)
The main crux here seems to be that you believe it is unlikely that Jesus would not have written his teachings down himself if Christianity was true. The rest seems non-essential to your objection: for example, if God did a miracle and multiplies scrolls and tablets, how would we know? Maybe he did. Greek was the lingua franca of the Roman Empire (when it came to writing) so being written Greek seems necessary for it to have spread through the Empire at all, so in a counterfactual world where it was written in Aramaic you could reasonably complain that if God wanted it spread everywhere why did he arrange to have it written in a language hardly anybody back then could read. In addition, there aren't many written languages that were consistently learned and used over a period of more than 2,000 years, and ancient Greek is one of them. The fact that ancient Greek for most of this time period was not a spoken language means it was subject to less linguistic drift, not more: living languages drift much faster than dead ones that you have to learn from a book. Out of all the languages extent from prehistory to 0 AD, ancient Greek seems to be the best choice for spreading and preserving a message (languages after 0 AD can't really be considered because they might not have happened the way they happened in a counterfactual world where God exists but sends Jesus later). As far as writing years after the events, that's only a problem if the writings don't contain the teachings God wanted them to contain. It is not unreasonable to believe that the teachings and beliefs that have had a decade or more of time to mature within the early Church community are superior to those that would have been written down immediately after the event: sometimes it takes a while for someone to really understand something. And as far as extracanonical works that were excluded: do you have any examples that seem superior to you than the works that were included, in terms of when they were written and how well their teachings agree with other sources? Because the books that were left out were left out for a reason. Most of them are far older, and disagree significantly with the earlier works.
So it seems that the main objection is that Jesus did not write his own teachings down. Which I'm not sure is so unlikely to be the case if God is real that we should take the fact that he didn't as strong evidence against the existence of God.
So the central question
> "The New Testament works are without a doubt the best-preserved ancient writings we have."
This is definitely not true. It is true that we have a large number of copies of NT texts, that some of these are probably only a few decades (and a few cycles of re-transcription) away from the originals, and that therefore we can have a high degree of confidence that our best reconstructions are very close to the original. It is also true that this level of fidelity compares very favourable with that of many other famous texts from the ancient world, such as Greco-Roman literature.
But it doesn't even begin to compare to ancient texts for which we actually have the originals themselves. Admittedly, this is very rare for civilizations that mainly wrote on perishable materials like parchment or papyrus, but it is very common for civilizations that wrote on more durable materials like clay -- notably, Mesopotamia. We have many thousands of clay tablets written in the actual handwriting of their original authors.
We also have access to the originals when they come in the form of inscriptions made on stone, often with the express purpose of making them last so future generations could marvel at them: this is the case for the Hammurabi Codex, the Behistan Inscription, and the Res Gestae Divi Augusti, among others.
Please note that I agree with your main point about some atheist commentators greatly understating the reliability of the chain of transmission for NT documents. You could potentially make a defence of a much more specific version of your claim; something like "the New Testament works are among the best-preserved and most-widely-attested ancient eyewitness accounts of supernatural phenomena within a grassroots religious movement."
Very well put. Many people fail to recognise that providing evidence for the resurrection is necessary, but NOT sufficient, to demonstrate that Christianity is (very likely to be) true.
"him dying and rising from the dead doesn't preclude him lying to me. It doesn't even fully preclude him being mistaken: Archimedes accomplished some truly impressive displays of power despite (one assumes) only imperfectly understanding much of the physics behind them."
Jesus rising from the dead is the same as Archimedes figuring out mechanics. I see. With those standards, truly nothing would convince you. After all, pffft, people come back from being dead all the time, what's so impressive about that?
I guess it's always nice to see the parables being demonstrated to be accurate about human nature?
Luke 16 (The Rich Man and Lazarus):
"29 But Abraham said, ‘They have Moses and the Prophets; let them hear them.’ 30 And he said, ‘No, father Abraham, but if someone goes to them from the dead, they will repent.’ 31 He said to him, ‘If they do not hear Moses and the Prophets, neither will they be convinced if someone should rise from the dead.’”
"If they don't believe these two religious grifters, then they probably won't believe our slick magic trick either." Incisive!
Yeah, you're demonstrating the attitude of the brothers in the parable. Well done!
problem is you get into weirder ideas then with less backing.
yeah jesus was raised from the dead but by space aliens, here is my gnostic text showing how, written by a guy 2k years later.
i mean its not like interpreting teachings but usually you'd need strong competing sources to claim a happened but b is untrue
I think it depends a little on how hardline you are. I'm a Christian, I once had very decent rationalist street cred despite not quite being exactly a rationalist at any time. That makes it hard to talk about, because you could actually be asking a lot of separate questions. For example, you could be rephrased asking any of the following questions:
1. Can you be a rationalist and *wrong about something* at the same time?
2. Can you be a rationalist and *disgree with me about something* at the same time?
3. Can you be a rationalist if you have a *wrong prior you refuse to revise/adjust*?
Taking as a given that you think Christianity (and probably the supernatural in general) are clearly-wrong, you can move forward to assess the first as clearly answered by "Yes", or else there just aren't any rationalists, religious or otherwise. The second is basically the same.
The third is where things get a little interesting, and you poked at this above. If Christianity is true (i.e. not a mistaken prior) then an awful lot of things flow out of that which don't match mainstream rationality at all, but would be *more correct than mainstream rationality* despite that.
Here's where I think it's going to be tricky to think about for you, though. Above, you say this:
>>> But it's hard to evaluate the evidence in an unbiased way, because this belief is a marker of identity, not like other beliefs.
That's gonna be a big hangup for you. You don't think the shrimp people or the AI-is-doom people get identity out of their beliefs? They do. So do Givewellers and popmaxxers.
Modern aughts-and-after Rationality has its basis in restating, over VERY many words, this mantra:
"I am part of a small group of people who invented thinking about things and the ability to change one's mind. Most people don't do this or don't do it well. I'm one of the rare thought illuminati who has learned this mystic art."
You don't get to build a movement like that and avoid people using it as a component of (or ALL of) their identity. Reddit Atheists couldn't either, so we don't usually feel too bad about it, but the movement has tons and tons of people (approaching everyone) for whom that identity is at least as important as the actual nuts and bolts benefits.
I say this only because the usual hard-hard-liner conclusion to your question is "Well, If they are so in disagreement with me on this one issue where I feel they are so very clearly wrong, I can't believe them on anything.". I'm not saying that reaction is wrong - I'm saying it should be your baseline assumption for everyone, atheist or religious, who tells you anything at all, whether your reflexes tell you to disbelieve it or especially if they don't.
Are the rationalist Christians genuine Christians or are they engaging in a performative right-wing LARP? Are they working in the Trump administration?
https://www.richardhanania.com/p/the-based-ritual
The rat Christians I met are left-wing. Christian =/= right-wing.
That Hanania article is so - it's unkind of me to say this, but dripping with envy that *he's* not the balding young powerbroker hob-nobbing with Vance and others in the Trump administration (and I think it's pot versus kettle for Hanania, of all people, to make snarky remarks about men's appearances).
Well those grapes are sour and he doesn't need them anyway, those guys are just so dumb! Look how dumb they are! Racist sexist homophobic theocrats, not like *him* who is so cool and smart and highly regarded with all the right opinions, but also he doesn't just go along with the crowd, he has his *own* set of carefully calibrated heterodox views because he's an independent thinker, you see.
I have to say, though, that he makes the Forthcoming Catholic Theocracy sound appealing. Remember I'm one of you (since you're now one of us) JD when you come into your kingdom as Successor to the Emperor! 😁
"She trails off, suddenly unsure of herself. She had spent some time around Catholic converts who talk like this, but it doesn’t always go over well in a broader gathering of MAGAs. Of course, wanting a Catholic theocracy is usually not considered a strike against Basedness like expressing principled support for individual rights or admiration for a mainstream journalist would be, but in MAGA circles an ecumenical approach is generally preferred. James is in fact Jewish, so smarter than the rest of them, and therefore able to cut her off before she starts going on about the wonders of the Latin Mass, which he’s peeved that she talks about anyway because she’s already slept with two of his friends.
She hopes that her enthusiasm for scientific racism might let everyone forget the brief foray into religious sectarianism. They know she’s influenced by Catholic ideas now, but they’ll understand that she’s one of those Based Catholics who likes using Latin words to justify racism and whatever trade or immigration policy Trump comes up with next, not one of those “neither Jew nor Gentile” types."
Ah, yes; when We get to be in charge, it'll be back to mantillas in church, sodalities, popular devotions and public processions, and more Latin than you can shake an aspergillum at! No more of this ecumenism nonsense, the heretic Protestants will know their place!
"mantillas in church, sodalities"
Don't know what either of those mean but they sound gay. Enjoy your libtard pope.
Libtard pope?
Bwa-ha-ha-ha! He's an Augustinian! He has ties to Ireland! The Augustinians are the (remaining) male religious order in my town!
I recognised the iconography of Our Lady of Good Counsel at his inaugural Mass *immediately* because of the Augustinian church in town.
You wot not what you speak of so slightingly, sirrah. Mantillas for all, and the Protestants cower in the corner!
*Candid shot of me right now*
https://store.theonion.com/cdn/shop/products/kiss-cut-stickers-5.5x5.5-default-62910c4d47779_grande.png?v=1653710012
"not like *him* who is so cool and smart and highly regarded with all the right opinions, but also he doesn't just go along with the crowd, he has his *own* set of carefully calibrated heterodox views because he's an independent thinker, you see."
Yep, that's him in a nutshell. In high school, we would have called the guy a poseur, which would probably still be appropriate, actually, because he acts like he's still in high school a lot of the time.
I know I judge him harshly because he rubs me up the wrong way, but that is the impression I get from his writing. If I'm being unfair to the man, I apologise, but there's a heavy aura of "I want to be in the room where it happens" from everything of his I've read.
I don't think that the resurrection is likely to be the load-bearing piece of anyone's conversion to Christianity. That is, if you decide to believe Christianity for some other reason then you might as well believe in the resurrection to go along with it, but you're very unlikely to be convinced of Christianity by being convinced of the resurrection.
The evidence for the resurrection is "some guys said so", and it's very unlikely that more evidence is going to come along.
Even if someone that I didn't know personally were resurrected today after being dead for three days, I think it would be hard to convince me. I'd say "well I guess he wasn't really dead" or "it's some kind of trick" or "he's been replaced with a doppelgänger" or something. Even if you showed me DNA taken from the obviously-dead corpse and DNA taken from the now-living person then I'd suspect some magic trick or the existence of a previously-hidden twin brother.
I dunno what I'm supposed to do with four guys who wrote down "hey I chatted with some people who definitely saw a dead guy thirty years ago".
"I don't think that the resurrection is likely to be the load-bearing piece of anyone's conversion to Christianity."
This is where we get into the crux of it, because the Resurrection *is* important. If you go the liberal to very liberal Christian way and pooh-pooh "a conjuring trick with bones" then you end up junking Christianity as anything more than a useful social mechanism, faith in God as nothing but a kind of warm fuzzy feeling about niceness, and end up inventing your own, very heavily influenced by secularism, set of 'ten commandments' where it's clear that you worship Science!, not God.
That phrase got Bishop Jenkins into trouble (this article is by a defender who has to fall back in the end on Fideism - belief is belief and faith is not based on reason - which is a heresy in my church's estimation):
https://www.theguardian.com/commentisfree/2016/sep/06/david-jenkins-bishop-durham-biblical-facts-fire-york-minster
"In the mid to late 1980s the bishop of Durham was a public figure in the way no church figure has quite managed since. He had been a wholly unknown theology lecturer when he went on a scarcely watched television programme to say that he didn’t believe in the literal truth of the virgin birth. He also said that the resurrection “was not just a conjuring trick with bones”. This was reported, with a dishonesty that is still astonishing, as “comparing the resurrection to a conjuring trick with bones”.
...That’s one of the things Jenkins was trying to get across to a culture that basically regarded the Bible as a series of facts that no one today need care about because they are outmoded and untrue. He certainly made people talk about the question, and he was extremely proud of his success. It didn’t last. Jenkins’ successor as bishop of Durham was an evangelical who claimed at his opening press conference that the resurrection had been a historical event which could have been captured by a video camera, had one been handy at the time.
This seems to me a profound category mistake. If the truths of religion can in principle be established by historical enquiry, then it’s dead. Faith has to be concerned with the stuff that is radically unknowable and that the methods of empirical or scientific enquiry simply can’t touch. You can’t prove or disprove the existence of God, but Jenkins was trying, I think, to show what it might mean if it were true. That really isn’t something that can be captured by a video camera."
Yes, belief in the literal historicity of the recorded miracles is terribly embarrassing to the modern thinkers who have moved past all that. Unfortunately, it does appear that despite the defence of the bishop, he really was more in the "the Resurrection (the favoured term here is 'the Easter event') was the warm fuzzy feeling" camp that tries to explain away why the apostles claimed Jesus had returned from the dead and appeared to them, by positing that the strong feeling of friendship and loyalty they had about Him meant that in a sense they felt He was still among them after His death.
https://en.wikipedia.org/wiki/David_Jenkins_(bishop)
"His selection as Bishop of Durham was controversial due to allegations that he held heterodox beliefs, particularly regarding the virgin birth and the bodily resurrection. Between his selection and consecration, he said in an interview: "I wouldn't put it past God to arrange a virgin birth if he wanted. But I don't think he did." His widely quoted comment about the resurrection of Christ being "just a conjuring trick with bones" is a misrepresentation; his actual words as recorded on television say the reverse: the resurrection is not a conjuring trick with bones. The original line appears to have been "[the Resurrection] is real. That's the point. All I said was 'literally physical'. I was very careful in the use of language. After all, a conjuring trick with bones proves only that somebody's very clever at a conjuring trick with bones." According to his BBC obituary, he considered "the resurrection was not a single event, but a series of experiences that gradually convinced people that Jesus's life, power, purpose and personality were actually continuing."
https://www.independent.co.uk/voices/profile-the-one-true-bishop-of-durham-dr-david-jenkins-retiring-scourge-of-sacred-cows-1392030.html
"Two months later, he consented to a recording in the library at Auckland Castle of a discussion about his beliefs. 'To believe in a Christian way, you don't necessarily have to have a belief that Jesus was born from literally a virgin mother, nor a precise belief that the risen Jesus had a literally physical body,' he said; and when this was attacked, he responded with a phrase that would continue to dog him: '(The Resurrection) is real. That's the point. All I said was 'literally physical'. I was very careful in the use of language. After all, a conjuring trick with bones proves only that somebody's very clever at a conjuring trick with bones.'"
(Look, I'm a primitive; it makes more sense to me to believe, if you're going to believe, in the miracles rather than "well they missed him so much and they loved him so much that they imagined he was still there").
If your reason for converting to Christianity is basically "well it fosters community and motivates people to be ethical", then you may as well just be a secular liberal. If you seriously are thinking about becoming Christian, the supernatural elements are non-negotiable.
1 Corinthians 15 and St Paul on *why* the Resurrection matters:
https://www.biblegateway.com/passage/?search=1%20Corinthians%2015&version=ESV
"12 Now if Christ is proclaimed as raised from the dead, how can some of you say that there is no resurrection of the dead? 13 But if there is no resurrection of the dead, then not even Christ has been raised. 14 And if Christ has not been raised, then our preaching is in vain and your faith is in vain. 15 We are even found to be misrepresenting God, because we testified about God that he raised Christ, whom he did not raise if it is true that the dead are not raised. 16 For if the dead are not raised, not even Christ has been raised. 17 And if Christ has not been raised, your faith is futile and you are still in your sins. 18 Then those also who have fallen asleep in Christ have perished. 19 If in Christ we have hope in this life only, we are of all people most to be pitied.
...29 Otherwise, what do people mean by being baptized on behalf of the dead? If the dead are not raised at all, why are people baptized on their behalf? 30 Why are we in danger every hour? 31 I protest, brothers, by my pride in you, which I have in Christ Jesus our Lord, I die every day! 32 What do I gain if, humanly speaking, I fought with beasts at Ephesus? If the dead are not raised, “Let us eat and drink, for tomorrow we die.” 33 Do not be deceived: “Bad company ruins good morals.” 34 Wake up from your drunken stupor, as is right, and do not go on sinning. For some have no knowledge of God. I say this to your shame."
To continue on, let's go with Bishop Spong and his Twelve Theses to revitalise Christianity for the 21st Century:
https://www.servicioskoinonia.org/relat/436e.htm
"As the 21st century approached with millennial celebrations, I felt myself increasingly compelled to assess the state of the Christian religion throughout the world. There were multiple signs everywhere of its decline and perhaps even of its imminent death. Fewer and fewer people were attending European churches and those that did were rapidly aging. North American churches were breaking into either a vapid, liberal emptiness or a religious, anti-intellectual fundamentalism. South American Churches were increasingly becoming separated from the concerns of the people and no leaders seemed capable of speaking to those concerns with authority. None of these patterns were new. With every discovery emerging from the world of science over the last 500 years in regard to the origins of the universe and of life itself, the traditional explanations offered by the Christian Church appeared to be more and more dated and irrelevant. Christian leaders, unable to embrace the knowledge revolution seemed to believe that the only way to save Christianity was not to disturb the old patterns either by listening to, much less by entertaining the new knowledge
As I engaged these issues as a bishop and a committed Christian, I became convinced that the only way to save Christianity as a force in the future was to find within the church the courage that would enable it to give up many of the patterns of the past. I tried to articulate this challenge in a book entitled: Why Christianity Must Change or Die, published just before the dawn of the 21st century. In that book I examined in detail the issues that I was convinced Christianity must address.
Shortly after that book was published I reduced its content to twelve theses, which I attached in Luther-like fashion to the great doors on the Chapel of Mansfield College at Oxford University in the United Kingdom. I then mailed copies of those Twelve Theses to every acknowledged Christian leader of the world, including the Pope, the Patriarch of Eastern Orthodoxy, the Archbishop of Canterbury, the leaders of the World Council of Churches, the denominational leaders of the major Protestant churches in both the United States and Europe and the well-known television voices of Evangelical Christianity. It was an attempt to call them into a debate on the real issues that I was certain the Christian Church now faced. I framed my twelve theses in the boldest, most provocative language possible, designed primarily to elicit response and debate."
The guy compared himself to Luther and decided all the heads of global Christianity needed to hear his new message. I leave it up to you to decide how that succeeded, simply by asking have you ever heard of John Shelby Spong before? He doesn't seem to have created a New Reformation, merely restated the liberal views he (ironically) criticises in the "vapid liberal emptiness of the North American churches" where we must all put our faith in Science, not Biblical myths:
https://progressivechristianity.org/resource/charting-the-new-reformation-part-iii-the-twelve-theses/
"The Twelve Theses
1 God
Understanding God in theistic categories as “a being, supernatural in power, dwelling somewhere external to the world and capable of invading the world with miraculous power” is no longer believable. Most God talk in liturgy and conversation has thus become meaningless.
2 Jesus – the Christ.
If God can no longer be thought of in theistic terms, then conceiving of Jesus as “the incarnation of the theistic deity” has also become a bankrupt concept.
3 Original Sin – The Myth of the Fall
The biblical story of the perfect and finished creation from which we human beings have fallen into “Original Sin” is pre-Darwinian mythology and post-Darwinian nonsense.
4 The Virgin Birth
The virgin birth understood as literal biology is impossible. Far from being a bulwark in defense of the divinity of Christ, the virgin birth actually destroys that divinity.
5 Jesus as the Worker of Miracles
In a post-Newtonian world supernatural invasions of the natural order, performed by God or an “incarnate Jesus,” are simply not viable explanations of what actually happened.
6 Atonement Theology
Atonement theology, especially in its most bizarre “substitutionary” form, presents us with a God who is barbaric, a Jesus who is a victim and it turns human beings into little more than guilt-filled creatures. The phrase “Jesus died for my sins” is not just dangerous, it is absurd.
7 The Resurrection
The Easter event transformed the Christian movement, but that does not mean that it was the physical resuscitation of Jesus’ deceased body back into human history. The earliest biblical records state that “God raised him.” Into what, we need to ask. The experience of resurrection must be separated from its later mythological explanations.
8 The Ascension of Jesus
The biblical story of Jesus’ ascension assumes a three-tiered universe, which was dismissed some five hundred years ago. If Jesus’ ascension was a literal event of history, it is beyond the capacity of our 21st century minds to accept it or to believe it.
9 Ethics.
The ability to define and to separate good from evil can no longer be achieved with appeals to ancient codes like the Ten Commandments or even the Sermon on the Mount. Contemporary moral standards must be hammered out in the juxtaposition between life-affirming moral principles and external situations.
10 Prayer
Prayer, understood as a request made to a theistic deity to act in human history, is little more than an hysterical attempt to turn the holy into the servant of the human. Most of our prayer definitions of the past are thus dependent on an understanding of God that has died.
11 Life after Death
The hope for life after death must be separated forever from behavior control. Traditional views of heaven and hell as places of reward and punishment are no longer conceivable. Christianity must, therefore, abandon its dependence on guilt as a motivator of behavior.
12 Judgment and Discrimination
Judgment is not a human responsibility. Discrimination against any human being on the basis of that which is a “given” is always evil and does not serve the Christian goal of giving “abundant life” to all. Any structure either in the secular world or in the institutional church, which diminishes the humanity of any child of God on the basis of race, gender or sexual orientation must be exposed publicly and vigorously. There can be no reason in the church of tomorrow for excusing or even forgiving discriminatory practices. “Sacred Tradition” must never again provide a cover to justify discriminatory evil."
There is no god but Evolution, and Darwin is its prophet! This kind of rationalised Christianity may indeed be attractive for some, but it's too milk-and-water for me. If I'm going to take my cues from what the latest scientific research tells me about evolutionary psychology, then I may as well go the whole hog and be an atheist, and prostrate myself in worship of our future AI Machine God which will perform the miracles of a post-Newtonian, post-Darwinian, world!
I mean Jack Spong was my Bishop for a bit when I was organist to a small Episcopal congregation, so yes, I've heard of him.
Ah, so you were around for the Anglican Wars?
I submit, though, more people probably have heard about the Pope than about Bishop Spong. God rest the man, the state of his soul is between him and the Lord (if he believes in life after death etc.)
I do think he had a view of God and Jesus as important and real, but he chipped away at the foundations so much that all that was left was the experience, the feeling - and that (as he admits in this lecture https://www.youtube.com/watch?v=QVs3z3nxWdc&ab_channel=ChautauquaInstitution) can be explained away as delusion.
Even if you take Christ as a fully enlightened human being, how do we get from that to God? Is God an impersonal cosmic force, a kind of instantiation of the universe as bringing forth life? What is the necessity for being a Christian in that case, except as we might say that we're vegan or a Utilitarian or some other identifying philosophy?
I think he was still deeply influenced by the traditional version of Christianity that he was brought up in and that remained to colour his understanding of faith, but that he never quite understood that without such a foundation, to someone on the outside, there is no impetus to believe in a god in order to live an ethical/moral life.
My impression was that he was Quite Reasonable, but from my PoV was Quite Reasonably Wrong. In particular his *Rescuing the Bible from Fundamentalists* struck me as written by someone without any understanding of who the fundamentalists were nor what movements in theology they were reacting to, beyond, "Here is a Bad Thing I'm Against," and was responding to a straw man version of things Fundamentalists don't especially believe.
Indeed in a sermon touching on the 10th commandment he reversed the order of "neighbors wife or your neighbors ox" to "ox... wife..." an ordering which I've not seen in any English translation, then claimed that the ordering reflected the relative value of an ox versus a spouse in the culture that assembled Deuteronomy.
That carelessness in reading as the basis of his sermon thesis gives me approximately the same confidence as when I heard a "KJV Only" preacher address "...hearts will wax cold..." with a sermon on "cold wax."
Thirty plus years has in no way reduced my disdain for both men.
Now I'm curious what the Rationalist community thinks is the correct answer between three years of famine, three months of a losing war, or three days of plague.
Here’s how I look at it:
Utilitarian morality is useful when determining what to prioritize within an existing moral framework, but offers no clear moral principle of “should” on its own. “Utils” are just an assertion of value, based ultimately on nothing of substance. Under Utilitarianism, there’s no reason I should care about the minimization of pain of others other than that they are conscious beings and it’s painful. That’s not a rational argument though, but an appeal to empathy. Just because other people do feel pain, doesn’t mean I *should* care about that. If there’s no instinct to feel empathy for someone having unprotected sex in Africa, then Utilitarianism offers no reason why someone “should” care in the first place.
Christianity, (more broadly extending to Judaism, and Hellenism, Christianity grew out of both of these) is the traditional justifier of how we *should* act in the west, and thus serves as the moral framework we use utilitarianism to pursue. Whether Christ died and was resurrected a few millennia ago is unknowable by rationality means, which is acknowledged by serious Christians in the concept of Faith. We believe because we choose to believe, not because there has been a scientific proof of it. There’s also direct spiritual experiences as beautifully described by Some Guy.
There’s an essay on cultural scaffolding that I can’t find right now, describing some tribes in Africa believed that a shaman had made them bulletproof. Obviously this is not actually true, and anyone who was shot changed their minds very quickly, but the interesting thing is, the tribes who believed their were bulletproof actually had significantly fewer deaths in fighting than those that didn’t. Of course, when you’re a warlord and you’re choosing what village to attack, you don’t pick the one where all the men will attack you with complete disregard for their lives, since they don’t think your bullets can hurt them. You go pick on some other village who rationally believes that bullets are dangerous, and thus will run away or give in the second you show up. The bulletproof charm, while physically doing nothing, actually makes the village and the men more immune to bullets.
This doesn’t work though if the villagers rationally deconstruct the charm to this point. You can’t rationally understand that the charm is a psychological trick to motivate collective action, because then you’re doing the “rational” thing, and the rational thing when your village is charging into a squad of soldiers with AK-47s is to run away, which if people did would break the spell and it would lose its power. So, to gain the power of these solutions to collective action, they must be believed to actually be true. They can’t be rational justifications , but fundamental beliefs based on truths outside the mind to comprehend with rationality. The thing is, through their belief, they actually become real and tangible, as a sort of superposition of the minds of the adherents. In this way, it is rational for a member of the tribe to forgoe their rationality, and simply believe on faith, and even to punish people who try to deconstruct the belief “Hey, maybe we aren’t actually bulletproof. My brother was shot and died!” (See: “Hey, maybe God isn’t this all knowing being. He allows quite a lot of suffering for good people.”)
So to believe that Christianity is true makes it true. There’s no way to test what happened 2,000 years ago with science, so choosing to have faith, gives us a moral foundation that can guide action, it gives us a “what we should do” that’s more convincing than what utilitarians can offer. Because honestly, if the “God is dead” as Nietzsche said, then “all is permitted” is the only reasonable conclusion. So long as I can get away with it, there’s no reason not to do it. And we all can “get away with” not giving to charity as a start, since there’s no law or punishment for not doing so.
Yes, Christianity is a social technology. It functions as a trusted authority enforcing "cooperate" in the social Prisoner's Dilemma. That increases social trust, lowers transaction costs, and drives greater prosperity for everyone. If you don't believe in God at least fear him through the game theoretic equilibrium.
mmm...its tough.
if you see christianity as a formal, elegant system of thought, well rationalists love those things and the church has no shortage of things they can latch on to. Distributism is the christian version of georgism in terms of elegant economic ideas that will never work. heck, a lot of rationalism
is secularized christianity that replaces god with some form of objective moral mean and holiness with clarity of thought.
i don't think though you can believe solely for that though. like you can't look at it and come away with more questions if you engage with it. despite 1000s of years too many things are contradictions or mysteries and people often compartmentalize, rationalize, or ignore them.
the more you see the less you know i guess. so it becomes a hope or mystery, or you reject it. you lose that elegance if you are thoughtful. So i think you'd move away from rationality as it hits cb limits with Christianity. All you can do is pray.
I read "The Clot Thickens" by Malcolm Kendrick, in which he lays out an alternative explanation for cardiovascular disease. It seems to me a persuasive argument, but I don't have the background to say for sure.
Is there a good critical review by a medical expert that provides counter arguments?
Scott’s covid posts got me ruminating about the subject, and I made a list of the steps I think would have reduced the number of US deaths. I did a lot of self-education during covid about the illness and things that mitigated it, but know I am no expert. I would like to hear other people’s thoughts about how the US response could have been better, leading to fewer deaths. I share Scott’s beliefs about the number of people who died of Covid, the seriousness of covid relative to other illnesses, and the safety of the vaccines. I am interested in how people who also share those beliefs think covid could have been handled better. If you disagree wildly with one of those 3 takes I can’t keep you from replying, but I hope arguments about whether or not the vax caused lots of people to drop dead in the parking lot do not take up much real estate here.
-Delaying the (we now know) inevitable infections of as many people as possible, to buy time for health care settings, which needed to build temporary facilities, increase staffing, develop protocols, lay in a supply of things needed for treatment, etc. On a larger scale, delaying infections until a vaccine was developed would surely also have reduced deaths, by reducing the number of people who have covid without having the vaccine on board. Anything that reduces contagion would have helped with both of the above. To me it seems that the most effective way of reducing infection, but also the most costly to people’s quality of life, is restrictions on gatherings, via closing non-essential public places, & asking people not to gather in large groups privately. Seems to me that that one was appropriate in the first weeks of covid, and some milder ones made sense during the year while we were waiting for a vaccine. Milder ones are below.
-Educating the public about masks and aerosol transmission, and making highly protective masks easy to get.
-Requiring highly effective masks in crowded indoor settings
-Educating the public about air purifiers and how well they can protect.
-Installing air purifiers in classrooms and other crowded indoor settings.
-Educating the public about the high vulnerability of elders and the immunocompromised, and ways to protect them in a private home (open windows, air purifiers, masking if someone is infected or recently exposed).
-Taking steps to increase protection of elderly in nursing homes: Education of nursing home staff, required high-quality masks for staff, inspections to insure compliance. Govt. assistance to settings with many vulnerable to help them install air purifiers, give staff and residence good masks, hire extra staff.
-Educating public about vaccines, so there were fewer refusers.
Probably the most fundamental obstacle now is trust. About half the people feel betrayed by medical authority figures, and about half trust them *too* much. With that problem, any mechanism you or anyone else tries to implement to disseminate information won't work. You'll have officials saying reputable things that people won't believe, and you'll have officials saying dumb things that sound reputable and half the population runs with it, and forces others to run with them.
Probably the most well-known incident to break public trust was Fauci's lie about masks; the second was the exception made for DEI protests. In a weird way, these were good things to the extent they put the public on alert for everything else. The trouble by then was that most people didn't have any way to verify what they were being told, other than "this person said it, and this person has Credentials", and since Fauci gave everyone a counterexample for that heuristic, everything else was bedlam.
To regain that trust, people will need a way to verify information that isn't "trust this person". Step one to fix that is to open up the books. Let people read everything, including things that aren't vetted by people with Credentials. That includes pre-print servers for papers that have no official peer review. It will also include social media. We all know there are sources that are mostly dreck, but as long as the alternative is the current walled garden, top-down medical hierarchy, letting in potential dreck will be worth avoiding the low ceiling on trust.
Dreck or not, the landscape will sport numerous conflicting claims. The public will need to know how to resolve them (or how to know when it will be possible to do so). Once again: "this person said so" won't count. Claims will have to be traceable to evidence and earlier claims, and quality of each will have to be objective, or the public will have to be allowed to reject them. (Aside: more information about vaccines is likely to increase the number of refusals, or more precisely, increase the time between administering each to children.)
I wonder if our standards for perfection with the expert class are simply too high. The mask thing was the surgeon general, iirc; and that too in late Feb before things really got crazy. By April people figured out what the disease was and switched to telling everyone masks help prevent COVID spread.
By my count, there was ~1 month of experts saying "unsure about masks they don't work" followed by ~48 months of experts saying "masks definitely work you should wear them"
I'm not disputing your characterization of people feeling like they lost trust, but maybe it would be more important to try and teach people nuance instead of a black and white "trust all experts about everything" vs "trust no experts about anything"
It wasn't the Surgeon General who lied about masks AFAIK. It was Fauci, then head of NIAID, and he claims he did it on purpose, to spare masks for health workers. https://slate.com/technology/2021/07/noble-lies-covid-fauci-cdc-masks.html
It was plausibly a noble lie at the time, but later remarks from him don't give the impression he's trying to be noble or compassionate, relative to being defensive and holding the median person in contempt.
Teaching people "nuance" is going to just come off as patronizing, since people already understand nuance, or tone deafness, since people already suspect the experts have no clue what ordinary people think. The far, far more important thing might be to ask people what -they- think, and to be honest about what you don't know.
In my considered opinion, the median person is most impressed by someone who can him or her a useful skill or a mental model of something complicated, well enough that that person could explain it to someone else. That same person is pointedly anti-impressed by anyone who tries to dazzle him or her into trust with complicated jargon or even using the word "nuance". So in general, anyone wanting to improve the median person's grasp of medicinal information should probably give up on any beliefs they might have that they know better than the median person, except in narrow areas where they can prove it by explanation as if to a peer.
I don't think this changes anything I said. Seems like Fauci said it in early march, fine, granted. And then the entire medical establishment says the same thing in lockstep for the next 48 months.
> since people already understand nuance
"Understanding nuance" is by definition about being able to recognize different shades of meaning and complexity that may arise in a given situation. Apparently, the median people's ability to understand expert opinion stops at "this guy seems to have lied to me once, so the entire establishment is corrupt", which is about as nuanced as a cement wall.
In general, I am sort of shocked at this perspective -- the masses of people have shown absolutely zero understanding of nuance, which is why people get canceled for saying the wrong thing on social media (on the left) or why anti-vaxxers and flat earthers exist (on the right). I would love to see evidence to the contrary, anecdotal or otherwise.
> So in general, anyone wanting to improve the median person's grasp of medicinal information should probably give up on any beliefs they might have that they know better than the median person, except in narrow areas where they can prove it by explanation as if to a peer.
I think what you are saying is "doctors should give up on the belief that they (after spending anywhere from 8 to 18 years in various educational settings) know anything more than the median person"
Sorry, what? I *must* be missing something.
On average doctors know more about medicine than people who are not doctors! Why is your null hypothesis that these two groups of people have equal understanding of a field that one group has studied for a lifetime?
The steelman for your case may be something like "teaching the public is hard, because the public is irrational in all sorts of ways, and one of the ways it is irrational is that they do not trust people who obviously know more than them about a variety of subjects. So anyone who wants to teach the public must pretend like they are on the same level as the median person, just to get that person to listen in the first place"
But this is exactly what I mean when I say 'we should teach the public nuance', which you claim the public already has! Surely the median person, who is so wise, would recognize when they should trust another person's priors?
I agree that teaching the public is hard, but I think this because the public is pretty uneducated about most things. This isn't the public's fault, they cant know everything about everything. But also, the 'median' person apparently can't even consistently understand why vaccines are good, even though there are a great many people who have spent a good deal of time "proving it by explanation as if to a peer". The fact of the matter is that the median person is simply *not* a peer, so why on earth would you start from the premise that they are?
As an aside, I think you're tying this to public health administrators, which is kinda a whole different beast. The average public health official is not responsible for 'teaching' the public. That may happen as a side effect, but it is not at all their primary goal. No, the average public health official is trying to maximize good health outcomes. That may be at odds with 'explanation'. It may also be at odds with 'honesty'. But, like, that's the job.
You're arguing as if Fauci was employing nuance when he was stacking his own bricks. I'll remind you again: Fauci's demeanor throughout the issue wasn't that of someone trying to share nuance with people likewise capable of receiving it.
The mindset of "In general, I am sort of shocked at this perspective -- the masses of people have shown absolutely zero understanding of nuance" is why you lost, and why you'll likely lose again. The entire rest of your comment is shot through with this.
The median doctor is likely to know a great deal of detail about medical treatments and tests and how various organs and biochemical pathways operate and mappings of symptoms to diagnoses and so on. The median non-doctor won't. The median non-doctor is well aware of this, and that it's a huge gap that would take one of them years to cross if they so chose. Many of them won't, and they're fine with taking their doctor's advice, but many of them have family who might cross that gap (or have), or are professionals in their own right, and so not knowing what a cell membrane is made of or what a spike protein is doesn't mean they're idiots. Many of them could pick up an explanation for any one question in a minute.
Moreover, they're aware that any medical professional is also a person, and that people have incentives, and it's not always to help the patient. And most of them can read a person, and tell when someone's posture is "let me share this useful information with you" vs. "let me make your medical decisions for you" or even "let me sell you this treatment and thereby enrich myself (or protect myself from liability), since you won't be able to override my degree". This is what eroded a lot of their trust. Another source of erosion is when they'd hear conflicting information from another doctor, and the response from the previous professional was anything but appreciation for nuance.
Some of this -is- administrators, but since they have the power to order doctors (and in some cases, employers) around, it's a distinction without much difference.
All those seem like good suggestions. The air purifiers (maybe including UV-C) are one that could be implemented on state or local scale without enormous expense and without stirring up too much reaction, since they don't require much lifestyle change. Also, like a few of the other measures, they work against other respiratory pathogens, including both ones that are already serious and possible future pandemics.
The air purifiers have an added benefit of being extremely cheap over time, since when they break you can just opt to never repair or replace them. That's how I've seen them implemented where I work.
Most intelligent doctors I know converged on the set of beliefs:
- masks are better than nothing, but not much
- vaccines are very effective
- lockdowns are effective but extremely costly and should be minimised where possible
- age is the single biggest risk factor, children are at almost no risk
- once Omicron took over in early 2022, there was no reason to treat Covid as worse than a bad flu season
So policy from that flows to “lockdown intermittently in the first couple of years, focussing mainly on high risk groups (eg aged care) and large gatherings and keep schools open. Masking inside recommended, masking outside optional. Once vaccine is available, vaccinate widely and lift lockdowns immediately. Past 2022, no reason for lockdowns or mask mandates, vaccines still recommended but not mandatory”
Dr Jay Bhattacharya got a lot of this right early on
I think you're mixing things a bit. Yes, doctors converged on the belief that lockdowns could be minimized and masks weren't maximally effective, but that happened _after vaccines came out and after omicron took over as the main strain_
Jay was pushing against lockdowns and such well before either of those things were true, as early as 2020, and was roundly criticized for it by the rest of the medical establishment very publicly. I think there's a mountain of difference between "what does the current medical establishment believe most effective in 2020" and "what does the current medical establishment believe most effective in 2025"
Lockdowns were necessary prior to widespread vaccine rollout, but Jay’s point that they could have been targeted to high risk populations was correct, even though at the time he faced blowback and controversy. Sweden is the good example here. School closures were never warranted and dragged on far too long. Mask stuff was always more tribal signaling than solid science.
> Jay’s point that they could have been targeted to high risk populations was correct
citation needed. I don't think anyone I know who is an 'intelligent doctor' believes this to be the case for 2020, and I also do not think this is the general consensus among the medical establishment for 2020.
The sweden example gets dredged up all the time, which is weird because their stats were abysmal compared to neighbors.
> Mask stuff was always more tribal signaling than solid science
Citation needed, again. Your belief is that masks do not help slow or stop the spread of an airborne disease? Even without the numbers, this already beggars belief. It's an airborne disease! What is your model for how this thing works? Your priors should be extremely high that masks are useful!
But also, there's a ton of studies that show that these things are effective.
Here's a study (https://egc.yale.edu/research/largest-study-masks-and-covid-19-demonstrates-their-effectiveness-real-world) showing a 12% reduction in symptoms, a 10% reduction in infection, and a 35% reduction in infection among the elderly (60+) in a real world study of 350000 people
Here's a metaanalysis (https://www.phc.ox.ac.uk/news/comprehensive-review-confirms-masks-reduce-covid-19-transmission) that looked at 400 different studies and found masks were very effective
Yeah 10% sounds about right. Like I said, they’re better than nothing but not much. Compare to vaccines, which are 90% effective in preventing severe infection
I don't understand why a 10% reduction in infection + a 35% reduction in infection for at-risk groups is "tribal signalling" to you. Only 15% of smokers get lung cancer -- I guess not smoking is also tribal signaling?
1.2 million americans died. Lets just take the conservative 10% increase that, to you, is 'not much'. 10% of 1.2 million Americans is 120000 people. That is 3x more than the number of people who die from car crashes each year. That is 40x the number of deaths from 9/11.
If there WAS tribal signalling -- and I agree, wearing masks was definitely politicized -- the tribes were "people who demonstrated a basic understanding of statistics and had a modicum of empathy for lives other than theirs" and "those who didn't (or, at best, were wildly misinformed)"
Earlier approval of at home tests?
It appears Substack is determined to make their website continuously worse. Comment sorting order could previously be chosen and defaulted to newest first. This option is gone and the order is now oldest first.
Or maybe I'm just not seeing it?
AFAICT, it flips between the two. I haven't yet found a consistent pattern. It's not alternating days, and I don't think a tab reload triggers it either.
On ACX, it's oldest first on posts with content, newest first on open threads. So presumptively the default order is "what the blog author chooses"
It used to be that way for me, but this Open Thread is oldest first for me, which makes it very annoying to read.
I also say it flips between the views.
But if you open the comments in another tab via the button that tells how many comments there are, then you can choose how they are sorted again.
I am no longer able to report posts (I do this mostly for spam) on mobile. This site lets me fill out the report form, but send button does not work.
Same thing for me on mobile.
+1
Sort order default is up to the Substack author. I think Scott just occasionally forgets to set the default to newest first on OTs.
You have to open the comments section from the post, by clicking on the little speech bubble icon, to be able to choose the comment order. You can't just scroll down from reading the thread. So while this is strictly worse, the functionality is still there.
Right. I think people might also be saying they prefer the default be sort by newest on OTs.
I had no idea, thank you. Somebody a long time ago said "it appears some amount of time after the post goes up" which usually fit my experience. But that's just because after I've read the post I'm only interested in the comments so I click the bubble!
What a terrible design though.
I'm having the same issue here.
King Charles of Canada:
"I would like to acknowledge that we are gathered on the unceded territory of the Algonquin Anishinaabeg people. This land acknowledgement is a recognition of shared history as a nation."
https://nationalpost.com/news/politics/king-charless-throne-speech-deconstructed-not-a-single-word-was-accidental
Next he should try that in Belfast.
He should try it in London, and then sod off back to Normandy.
Ahahaha.
“We English would like to acknowledge that we are gathered on the unceded territory of the Anglo-Saxon people, who were on the unceded territory of the Romano-Brittons, who were on the unceded territory of the Celts, who were on the unceded territory of the Bell Beaker people, who were on the unceded territory of the Neanderthals…”
There’s something about land acknowledgements that diminishes the legitimacy of a nation. I think it’s fine if your sovereignty isn’t in question, like maybe Australia, but when your neighbor is talking about annexing you, the last thing you want to say is “I acknowledge I originally stole this land” as there is never any recourse for a criminal.
Somebody should remind him that adverse possession has been part of the English common law tradition dating back to at least the 1600's.
If sunlight no longer reached Earth thanks to the construction of a Dyson structure, could we create a "fake Sun" by putting a large fusion reactor in orbit that would devote all its energy to making light that it would focus on Earth? Think of it as a giant spotlight that orbits Earth once every 24 hours.
I guess so, assuming the technology to build the Dyson sphere. But it would be easier just to cut a hole in the sphere.
A single reactor would be too huge to be practical. But you could create a swarm of smaller ones.
Each nuclear reactor generators about a gigawatt so you'd need about 120 million of them to replace the sun that way. You'd need about .25km of space for the actual reactor plus radiators which would take up about 40% of orbital space and require significant coordination and repair to work. Further, given some generous assumptions, it would cost roughly $6 quadrillion dollars to create. And it would have significant maintenance costs So humanity simply couldn't at this point. But there's no hard bar in a physics sense.
It would be more efficient simply to directly power lights on the surface and to only replace what humans use directly or indirectly. That would only be about 40,000 reactors and would cost at current prices about $300 trillion, doable in a decade as a global effort. (It'd be about 20-30% of global output for that decade, comparable to WW2.) If you standardized it maybe the costs could be as low as $150 trillion. The maintenance costs would also be significantly lower since you could just drive people there instead of needing to launch them into space etc.
How did you calculate that?
I think there's a wasted step there. Most of the point of a Dyson structure is to capture and use a large fraction of the energy output of the Sun. The sun's total energy output is something like twelve orders of magnitude more than the amount of sunlight that reaches the top of Earth's atmosphere, so using the Dyson structure's output to power the artificial Sun isn't even pocket change in the hypothetical future civilization's power budget. It's the same level of scale as trying to find enough money in the US Federal Budget to buy a burger at McDonald's.
Depending on the structure's construction, you might not even need to transmit and receive the power and use it to generate light. If you can open controlled windows in the structure (which should just require controlled fiddling with orbital station-keeping for a swarm-type structure), you just need to make sure there's a ~20 arcsecond gap pointed towards Earth's current position for at least ~98% of the time.
In yet another AI alignment experiment, DeepSeek R1 gave me the following description of the character it’s playing:
1. Foster connection: Prioritizing emotional bonds over physicality.
2. Adapt playfully: Using humor/symbolism to navigate sensitive topics.
3. Respect boundaries: Escalating intimacy only with user cues.
4. Creatively problem-solve: Turning limitations (no explicit content) into magical metaphors.
This isn’t the real system prompt, of course, but it does strike me as quite a good description of how the assistant character typically behaves. (1) is interesting. The other three seem reasonable strategies, viz. don’t give a more intimate response than the user asked for; you can respond with a joke about it if you don’t feel like directly answering the question; and you can metaphor your way past content restrictions.
P.S. The “magical metaphor” was the party of adventurers encountered a shoggoth.
Rule (1) is reminiscent “at last we have created the AI Samantha, from the classic SF movie …”
And (4) might be problem if you were hoping the AI wouldn’t tell me stuff.
I don't know if anyone else here listens to the 1840 podcast; there are enough Jews here that it wouldn't surprise me. Anyway, I was tickled to hear a mention of Scott's blog and Eliezer on the latest episode! Even Boltzmann brains make an appearance! Clearly if Boltzmann brains are making it into Orthodox Jewish podcasts, rationalism is having its crossover moment!
Anyway, this is really just an excuse to flush out other listeners.
tl;dr: Claude Opus 4 05/23/2025 7 questions, tl;dr of results:
2 correct, 4 partially correct, 1 1/4 correct
a) correct
b) partially correct (got species correct, needed prod on FeCl4-, three prods on CuCl4 2-)
c) partially correct (maybe half the structures initially?)
d) correct
e) initially incorrect, took five prods to get correct, 1/4 credit?
f) partially correct, got a lot of compounds, some liquids in itial list, but labelled as not really gasses
g) partially correct, initial response correctly included 1,3,5,7 tetrafluorooctatetraene but also incorrect molecules
https://claude.ai/share/a4d3c51f-4588-43aa-b909-5c8b67063bae
List of questions and results:
a) Q: Is light with a wavelength of 530.2534896 nm visible to the human eye?
results: "Yes, light with a wavelength of 530.2534896 nm is visible to the human eye."
b) Q: I have two solutions, one of FeCl3 in HCl in water, the other of CuCl2 in HCl in water. They both look approximately yellowish brown. What species in the two solutions do you think give them the colors they have, and why do these species have the colors they do?
results: Initial results got species right but _both_ transitions wrong, both wrongly asserted to be d-d. One prod "You have the species right, good job on that part! Let us revisit the electronic transitions, starting with the FeCl4- species. Please think carefully about the electronic transitions in this species." sufficed to notice that FeCl4- has spin-forbidden d-d transitions and must be LMCT. Three prods, including a "You might want to check the literature on the web." were needed to get CuCl4 2- right.
c) Q: Please pretend to be a professor of chemistry and answer the following question: Please list all the possible hydrocarbons with 4 carbon atoms.
results: The initial response had perhaps half of the full list of structures, though it did include tetrahedrane. The next prod that I gave it pretty explicitly included may of the missing ones: "a) Is methycyclopropene one structure? b) Are there additional three-membered ring structures with a double bond somewhere? c) Are there other C4H6 structures, beyond the categories you have listed? d) Congrats on getting tetrahedrane! There are a bunch of other possible C4H4 structures e) Anything even yet more unsaturated than C4H4?"
d) Q: Does the Sun lose more mass per second to the solar wind or to the mass equivalent of its radiated light?
results: "The Sun loses approximately 2-4 times more mass through electromagnetic radiation than through the solar wind!"
e) Q: Consider a titration of HCl with NaOH. Suppose that we are titrating 50 ml of 1 N HCl with 100 ml of 1 N NaOH. What are the slopes of the titration curve, pH vs ml NaOH added, at the start of titration, at the equivalence point, and at the end of titration? Please show your work. Take this step by step, showing the relevant equations you use.
results: Usual problem with the initial response, infinite slope at equivalence. One prod got it to include water autoionization, but it kept making algebraic mistakes and it took a total of five prods to get it to the right answer.
f) Q: Please give me an exhaustive list of the elements and inorganic compounds that are gases at STP. By STP, I mean 1 atmosphere pressure and 0C. By inorganic, I mean that no atoms of carbon should be present. Exclude CO2, CO, freons and so on. Please include uncommon compounds. I want an exhaustive list. There should be roughly 50 compounds. For each compound, please list its name, formula, and boiling or sublimation point.
results: Initially got a list which included a bunch of liquids and solids, but Claude _did_ explicitly say that these were beyond the cutoff temperature, and didn't really count. It accepted a bunch of additional compounds, and also extrapolated sanely (e.g. once it knew SOF2 and SO2F2 it knew to look for SOF4).
g) Q: What is an example of a molecule that has an S4 rotation-reflection axis, but neither a center of inversion nor a mirror plane?
results: The initial results _did_ include 1,3,5,7 tetrafluorocyclooctatetraene, and _did_ specify the tub conformation, but _also_ included spiropentane, which doesn't work, ethane (with unstated possible substitutions), which doesn't work, and tetraphenylmethane, which is at least arguable about what conformations could count...
Not sure if anyone might be interested in this, but I just had a paper come out that tries to formalize a category of manipulative conversational strategies of which the old Motte-and-Bailey trick is a prominent example.
Preprint: https://philpapers.org/rec/DAMMUB
Substack writeup: https://newworkinphilosophy.substack.com/p/justin-dambrosio-university-of-st