Does anyone have a good reference post from somewhere with a breakdown of data center energy usage and environmental impact? I keep getting into discussions about this topic, and seeing wildly different claims, so it would be nice to have some grounded facts to memorize.
This is the piece that most directly addresses all the commonly circulating arguments and puts hard numbers on them. The title sort of gives away his conclusions...
Bitcoin mining and blockchain transactions are intentionally orders of magnitude more energy-intensive than LLM usage. You'd have send something like a million queries to approach the energy usage of a single bitcoin transaction.
I think a lot of the "AI is killing the environment" rhetoric is actually people just pattern-matching their old environmental critiques of crypto onto the new tech bete noire, but because the magnitude of the energies involved on an individual level are so vastly different those talking points are no longer actually coherent.
Both of those would net you so little Bitcoin these days that it's completely insignificant.
You need ASICs for Bitcoin. Your CPU/GPU power is basically irrelevant. The malicious websites would instead be mining something more PC friendly like Monero, and they would still generate orders of magnitude less revenue than just putting ads. It's very noticeable because your computer fan is going to start blasting, so the hacked website owner is likely to catch it quickly. And there's no economic sense in doing this to your own website.
Thanks I'm clearly an idiot about this stuff. I will say that I hate that almost everyone has a narrative they are trying to tell. And so environmentalists* are telling their story.
*I like environmental stuff in theory, taking care of our planet. In practice it looks like many people with less than good information complaining about plastic straws or some other boogie man of the day.
"Global electricity generation to supply data centres is projected to grow from 460 TWh in 2024 to over 1 000 TWh in 2030". In 2024, global electricity generation was approximately 30,664 TWh. Data centres represented 1.5% of total electricity generation in 2024. This is projected to grow to 3% (1000/33600) in 2030.
"renewables remain the fastest-growing source of electricity for data centres, [...] meeting nearly 50% of the growth in data centre electricity demand. [...] Even so, [...] natural gas and coal together are expected to meet over 40% of the additional electricity demand from data centres until 2030."
"CO2 emissions from electricity generation for data centres peak at around 320 Mt CO2 by 2030, before entering a shallow decline to around 300 Mt CO2 by 2035. Despite rapid growth, data centres remain a relatively small part of the overall power system, rising from about 1% of global electricity generation today to 3% in 2030, accounting for less than 1% of total global CO2 emissions."
Of course this is solely focusing on the use of energy, not on its outcome. A useful idea when discussing the growing share of data centers as part of the world's energy mix is what your interlocutor and yourself expect to get out of it. How much extra growth would make it "worth it"? Where would one even start with this line of questioning? How closely does economic growth translate into better standards of living, and how equally are those standards expected to increase across the population? If your interlocutors don't have the slightest clue about these topics, I wouldn't take their opinion very seriously, no matter how impressive the numbers are. You can't just discuss the costs of some major investment/cost/externality without having an idea about its benefits or lack thereof.
Specifically, most countries stopped large-scale whaling in the 1960s, or at least cut way back. The Soviet Union was the one big exception and did most of the hunting-whales-to-the-brink-of-extinction in the late 60s through the early 80s when they finally decided to phase out their whaling fleet.
Commenter may be female. Has mentioned a husband so statistically more likely. They dropped a comment to me in the previous thread that had a huge tell in the very first sentence.
His MO is to give hot takes and regurgitate conservative talking points backed up with whatever misinformation he can find.
Which doesn't sound too bad, but he does it dozens of times over the course of an open thread (scroll through this one for examples) and gradually escalates until he's banned. Then he waits a couple weeks, makes a new account, and starts again.
My *personal* animus against him is from when he was doing this while the Feds were attacking protesters in my city (Portland) a couple years back. He was parroting every bit of misinformation the admin was feeding Fox News, so he was claiming that the city was burning and disregarding anything from people who live here.
1. Do we really need massive data centers to implement current and future AI models? From what I've read, these models can be implemented on much smaller computer clusters, and cloud providers have been wastefully implementing them on infrastructure that hasn't been optimized for AI deployments.
2. Is China *really* ahead of us in AI model creation? Most of the arguments about China's lead are that they're implementing their models on smaller, leaner, distributed computer clusters — which we could do if we bypassed cloud service providers — rather than that their models are superior.
Unfortunately, I haven't been bookmarking the links to these arguments. But if people have serious data showing that we need huge clouds to implement AI, or that China has better models, I'd like to hear them. As with most of the discussions around AI, there are a lot of bullshitters (er, hallucinating commentators) out there. I don't really have the chops to winnow out the bullshit, though.
> Do we really need massive data centers to implement current and future AI models?
Isn't the fact that all the smartest / richest companies in the world are spending trillions of dollars on this proof that this is true?
Presumably they have the in-house expertise both on the AI and the data center side to know what they need, hardware wise - certainly FB and Google do, and that knowledge would percolate out to Anthropic and OpenAI and xAI.
OK. I got off my ass and did a little research. Yes, you can run a sophisticated LLM like Deepseek on a small cluster of computers (4 high-end Macs in this video).
Of course DeepSeek is an open source model developed in China. But ChatGPT turned up several other open source LLMs — some of them developed by US-based companies. They come with pre-trained open-weight models. But there are also open source training corpora.
I asked the ChatMonster if it could run a small computer cluster, and it said, "The issue isn't just storing the weights. It's generating responses fast enough for millions of users simultaneously. A small cluster might eventually produce an answer from a frontier-scale model if you somehow had the weights, but the latency could be measured in minutes rather than fractions of a second."
But after further queries on my part, ChatGPT admitted that a proprietary frontier model running on a cluster of a dozen machines with adequate RAM and with high-speed network connection could provide very decent response for 5-10 users. Moreover, smaller cluster of machines running tensor processors (TPUs or LPUs), which are ASICs that optimized for token processing, could leave a larger cluster of NVIDIA GPUs in the dust (my wording). The thing about NVIDIA's GPUs, is that they provide a huge library of tools for developers. So (many) developers are sort of locked into using relatively-inefficient GPUs because they don't want to develop their own toolkits.
ChatGPT left me with this thought: "The gap between open models and frontier proprietary models is now less about access to transformer architectures and more about data curation, post-training, evaluation, and the enormous compute budgets required to execute those processes at scale."
So, I would conclude that a company like OpenAI required super massive data centers because they're locked into a inefficient technology substrate, that they adopted because they didn't want to spend the time optimizing their own. OTOH, ChatGPT estimates that a TPU-based data center would only reduce the electricity consumption by about 25% (still it's nothing to sneeze at), but optimizing the model would give much higher gains.
Yes, that's true, but I thought the real reason everyone is all-in on GPU's rather than ASICS or TPUS is basically that an ASIC (and to a lesser extent, a TPU) requires full lock-in on the current weights, and will continue to be locked in on those weights into the entire depreciation future of the hardware, and progress is breathtakingly fast. They drop new models roughly every month now.
A GPU is fully general, even if less efficient, and can both train and serve whatever new models you come up with.
TPU's aren't quite as locked in as ASICS, but aren't as general as GPUs. Since all of these things have 3 year (at a minimum) to 7+ year (you still can't buy used A100's because they're still being used to serve inference) depreciation cycles, the best bet is still GPU's while we're in the part of the curve where models have substantial updates within 1-2 years. And we're 10x faster than that right now.
So as long as the companies expect progress to continue at even <10% of current rates, it's better to build the giant data centers with less efficient GPU's.
I think a big question is "how much can you reduce the need for compute by design choices?".
So much AI talk focuses on the current/projected abilites of LLM's, that there isn't much awareness about other architectures being tested. Yann Lecunn's paper from 2022, for example, proposed a 'world model' approach; it's an interesting read with a more fully-fledged attempt to mirror human thinking. From the press release (and directly asking Gemini), it looks like Gemini's Omni video-gen is WM-based.
My intuition tells me that this is a more reliable roadmap to reaching AGI; have the AI use sensory data to learn to infer cause/effect relationships, in different modalities. WIth LLM's, you're essentially buying into a belief in emergence (for human-level abilities across g-loaded modalities).
I don't think China has to be in the lead (nor do I think they are) to pose a serious threat to the economic model of American AI companies. Token prices at OpenAI/Anthropic are rising quickly because you're not just paying for inference but RoI. What if you're not doing that for open source models? I keep seeing these graphs on X where the Chinese LLM's are months behind, but even if they're *a year* behind, if that means that in a year they might code as well as Codex now - that will be 'good enough' for many programmers.
> [Surely the august commentariat here understands that it is a really stupid idea to build "renewable" windmills in the salty "corrosive hell" that is the world's oceans?]
I deeply resent the implication that I have to go with your theory or else be stupid.
Make an argument from facts, if you wish, but stop skipping steps. Maybe you're even right. If this is important you can make the argument without insults.
I scanned this paper, and there are two things that came to mind:
1. This is about anything that works in the ocean generally, and I assume it includes things like oil tankers, docks, and a lot of other things they mentioned, but I didn't see windmills included anywhere. Also, I didn't see any cost-benefit analysis, but maybe it's buried in there and I missed
>Surely the august commentariat here understands that it is a really stupid idea to build "renewable" windmills in the salty "corrosive hell" that is the world's oceans?
6 years to double isn't really conservative, it's pretty aggressive. Grid constraints and lead times in power generation are the biggest roadblocks to closing the gap between announced electric capacity increases and what is most likely to happen.
It is finding ~the same numbers that Masley does but failing to contextualize them appropriately, in order to maintain a tone of grave concern. On page 27 for instance we get the grave invocation of "237 Olympic-sized swimming pools" of water used for training GPT-4. You could offset that water usage by reallocating 0.15 square miles of irrigated alfalfa in the Colorado river basin to be fallow. It sounds like a lot of water until you compare it to any other way we use water, which is what Masley does and the UN does not.
What they do do is denominating all the power and water usage in terms of "the minimum needs of X-thousand sub-Saharan African people" which is just kind of laughably manipulative, there is no first-world consumption good that comes out looking non-wasteful by that metric! But also if you build a datacenter in Utah the alternative case for the water and power used there is not that they were going to get shipped off to Burundi.
"'Civil Disobedience,' a thread by @hdthoreau" presents Thoreau's influential essay as though it were a social media thread. It includes "replies" from people who have engaged with and critiqued it over the years, and from Thoreau himself in the form of quotes from his journal and his later works that expand on the ideas in the essay. A "wikibot" and "biblebot" pipe up from time to time also, to explain now possibly-obscure references.
The content came from a lot of searching in the archives (e.g. Internet Archive, Google Books, the old fashioned library, etc.). The presentation was vibe coded in cooperation with Claude Code. The avatars were generated by AI based on descriptions of the person or old public-domain photographs when these were available.
Hi! I happen to be reading a lot about the FFM recently and I read over some of your posts. I think that maybe your concern about personality surveys having problems because of synonyms and similar questions has already been addressed by the field during the big argument about Implicit Personality Theory in the 1960s, but I'm not sure if I'm missing something about your argument. See: https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-6494.1992.tb00975.x
I cover that in the “Limitations” section of the article. Short answer is 0.09%/year. You can transfer assets out but managing hundreds of different stocks on your own is a pain in the ass.
Thanks for asking that. It’s a great question, and I hadn’t considered it until now. I think fractional components are very likely to be liquidated, and you’ll take the cap gains tax hit. I’ll email them to confirm, then update the post with my findings.
Edit: Confirmed, they will sell the fractional components.
Thanks. I checked the math against https://www.slickcharts.com/sp500 and if I invest 100K in all 500, 36% of it is in fractional shares if I round down. If I somehow have the chance to round up (as in, they will let me top off shares where I have more than half a share) then I'm off by 26%.
If I have $1 million, then the numbers are 5.7% and 2.8%.
I don't think I'm anywhere in the range for this, but if I were I'd like a mode where I can say "please gradually shift towards wanting whole numbers of shares."
If I limit myself to the top 100, 100K will only have 16% in fractional shares, rounding down. Then I use XOEF for the rest.
"By the 1930s, people had started to notice that this wasn’t happening. Although capitalism had suffered some crises - World War I, the Great Depression - countries had responded with what the Marxists called state capitalism, a broad term for hybrid systems including New Deal America, Stalinist Russia, and Fascist Italy. Most people seemed to like their own country’s variety of state capitalism. Some Marxists began to worry that maybe Revolution wasn’t inevitable after all."
The way Marxists invented "state capitalism" is very reminiscent (to me, anyway) of Ptolemy's epicycles. Ptolemy was certain that planets move around the Earth, and when they visibly moved "backwards", he came up with the idea that instead of circling the Earth, every planet circled an invisible point, and *that point* circled the Earth. Later he had to add more epicycles -- every planet circling a point, which circled a point, which circled... the Earth. His model grew more and more complicated because he could not allow the notion of *Earth* moving.
"State capitalism" is not something Marx ever thought of. It is something his disciples added to Marxism when reality failed to keep up with their model. So was the idea that Lenin "cargo-culted" Marxism, as Scott so pithily out it. (German and Austrian Marxists certainly did not think that while Lenin was still alive!) And when you are utterly convinced your model is TRUE, yet must keep adjusting it to match empirical evidence, isn't it pretty much the definition of a religion? Or at least of a religious faith?
> "State capitalism" is not something Marx ever thought of.
Minor quibble. In The Communist Manifesto (1847–48), Marx and Engels wrote: „Vermehrung der Nationalfabriken, Produktionsinstrumente, Urbarmachung und Verbesserung der Ländereien nach einem gemeinschaftlichen Plan.“ Which I read as "Increase of national factories and instruments of production; cultivation and improvement of lands according to a common plan." But they don't explicitly state that all factories must be under state control. Only that it would be good if some factories were state-controlled. It's rather vague, though, and post-Marx Marxists seem to have argued about the meaning and implications of that phrase.
However, later in his life (after the Paris Commune of 1871) Marx wrote: "One thing especially was proved by the Commune, viz., that 'the working class cannot simply lay hold of the ready-made state machinery, and wield it for its own purposes.'" In the same essay, he wrote that what he meant by "state machinery" wasn't merely the government in the abstract, but "the centralized State power, with its ubiquitous organs of standing army, police, bureaucracy, clergy, and judicature." Anyway, Marx's ideas were evolving. And his later stance seems to be the opposite of the later Leninist interpretations of his political philosophy.
> isn't it pretty much the definition of a religion? Or at least of a religious faith?
French sociologist Émile Durkheim was the first to point out that Marxism resembled a religion. The anthropologist Clifford Geertz formalized a systematic definition of religion. He argued that religion was a "system of symbols that establishes powerful, pervasive, and long-lasting moods and motivations by formulating a general order of existence." Much to the angered sputtering of Marxist social scientists, he tossed rigid political ideologies like Marxism into the religion bucket because they functioned identically to religions by providing an all-encompassing "cultural lens," an ultimate meaning for suffering, and an "aura of factuality" to their followers.
You should polish the lenses of your rightwing-fogged glasses. You don't really understand how CRT and Queer Theory appropriated Marxist dialectics to argue that it wasn't the working class but racial and sexual minorities who were being exploited and oppressed. Traditional Marxists were appalled by this heresy, and much hilarity ensued on Marxist forums as CRT-QTers attacked them and Marxists protested that they weren't racists or sexists. Progressivism has always been about the willingness to consider innovative solutions to social and political problems. The arguments among progressives have always been about what actually works — that is, until the CRT-QTers started demanding that progressives cater to their whims. Because of the ethical framework underlying progressivism, It's hard for progressives to turn away a victim (even if the victim is exaggerating their victimhood). On the left wing of the spectrum, this is analogous to how MAGA took over the GOP and silenced traditional conservatism — except we still have some sanity left in our house — i.e., that winning over the centrist voters to win elections is more important than movement purity.
I don't know that progressive arguments were "always about what actually works", surely there's been a long running throughline from the 19th century to now that A) society should be advancing towards some proposed other state which they believe would be preferable and B) rule by experts is the best way to produce this.
It's certainly possible to have a different approach to "what works", one could simply try to refine the existing society and just make that work as intended, as the people who built the existing society would have wanted it to function. I could refine my widget factory to make widgets at half the costs and reduce the number of defective widgets to a trivial number, or I could decide that what I should really be doing is trying to iterate towards manufacture of a SuperWidget from my Shining Widget Factory on a Hill. The former is about "what works" but the latter is much closer to progressives' goals. The "what works" people are the neolibs and a few of the wonkier conservative think tanks that have survived Trumpism.
Likewise, it may well be that "what works" is decentralized, local, autonomous units, a system following the principle of subsidiarity where decisions are made at the lowest level they possibly can be made and we try to improve the competence of systems at those levels to return more power to them. But progressives would be fundamentally uninterested in that solution, the existence of a Porfiriato style gang of elite technocrats steeped in logical positivism is baked into the worldview. The neolib sorts also take this assumption for granted, but with evidence that local control worked or that nationalizing X policy consistently fails I think Ezra Klein etc would accept it. The progressives need society to be ruled because society must be moving towards its destiny as a collective enterprise. So even their internal arguments are not always about "what works" because the whole project of progressivism entails creating a thing that doesn't currently exist, so the contours of that hypothetical thing that everyone must be forced to work towards are of much importance to them.
I should make a distinction. There were actually two progressive movements on the Left side of the spectrum since 2000. After George W. Bush won in 2000 (after Al Gore's pathetic campaign), there was a movement to remove centrist Democrats from the party's internal levers of power. Markus Moulitsis was one of the flag-wavers of this movement (you can look up his writings on the DailyKos). He advocated for fielding less wimpy (i.e., less polite) candidates who would fight back in the media against the slurs of the Right and advocate for middle-class-friendly policies. He advocated for greater social safety nets, but more in the FDR style than the socialist let's-reform-the-entire-system style. There were all sorts of wonkish discussions about what would benefit the middle class, what constituted the middle class, or whether we should even promote policies that would benefit the middle class at the expense of the working class. These progressives thought they had elected one of their own when Obama won his first term.
But Obama installed a bunch of centrist/corporatist technocrats to run his administration. Both Obama terms were Bill Clinton 2.0. He didn't get us out of Afghanistan and Iraq as promised, and he barely got healthcare reform passed. Centrist Democrats still controlled the party, and a new batch of progressives assumed the movement's name. Bernie Sanders, AOC, and that crew started gunning for Wall Street (much to the discomfort of centrist Dems who got a lot of campaign donations from Wall Street). The wonkishness of the first wave of Progressives is gone. They're fighting to reform the entire system, so they're fighting centrist Dems as hard or harder than they fight Republicans. They hated Joe Biden (despite Biden getting more of the liberal agenda passed than Obama). Meanwhile, the centrist Dems start sounding downright Republican when populist lefties like Mamdani entered the fray to challenge their preferred moderate candidates.
The debate over whether we should field more moderate or more radical Dems is ongoing. Today's progressives are not the same as the first wave of progressives in that they're less interested in arguing over policy wonkery and more interested in unabashedly advocating for a social-democratic/democratic-socialist ideal that they (mistakenly) believe the Nordic countries embrace.
This comment seems more reflective of your own biases than actual reality. In particular, this seems very much more like something that you would *like* to describe reality than something that actually *does* describe it:
..." Floyd and others enshrined as 'saints'"...
I've spent quite a lot of time in various progressive spaces and I've never, ever seen people talk with veneration or reverence about George Floyd. Indeed, that entire mode of discourse--elevating *any* human to reverentially authoritative status--is often looked on with suspicion[1]. In reality, George Floyd specifically seems to be far more of an object of obsession on the right than on the left. I don't see his name mentioned especially often in progressive spaces and when I do it's rarely mentioned centrally; it will be paired with names of other victims and/or discussed in context of systemic issues with policing. Despite spending far less time in conservative spaces, I've almost certainly heard the name and the qualities of the man mentioned far more often by right wingers than by left wingers.
"public veneration of them along walking paths"
And at this point I must gently ask if you understand the difference between an object of reverence and, well, a symbol. Symbols are extremely common out in the world; can you really not recognize one when you see it? The reason that George Floyd's death was an important event isn't because of any prior notoriety or particular qualities of the man himself. It was because the events of his death made him an extremely potent symbol: it was a stark illustration of issues with policing that progressives had been complaining about (with increasing anger and exasperation) for decades. That makes mentioning the man a convenient shorthand for mentioning the problems, convenient because it reminds people in dramatic terms why the issues are important. Examples of this sort of thing abound in American politics: it really should not be hard to recognize.
"and required displays of 'religious fervor'"
REQUIRED, you say? By whom? Enforced by what? Can you point to me any examples at all of somebody being penalized in any way for not displaying a pride flag?
To me, this again appears to be a rather deep misunderstanding of your fellow humans. I don't believe I've ever displayed a pride flag my own self, but it's extremely simple to understand why people do it. There are two good reasons and a bad reason:
1. Queer people do it as an act of defiant self-expression. As much as conservatives like to pretend otherwise, being openly queer does still expose people to hostile reactions, both overt and covert.[2]
2. People who are friendly to queer people (whether themselves queer or not) do it to signal solidarity, indicating that they don't have any hostile intentions towards queer people.
3. Public relations and marketing people people try to parasitically co-opt the above trends just like the do with *every* social trend, to try to blend in with groups 1 and 2 and more effectively extract money from them.
Note that the last one only works because as of the past decade or so, the broadly queer-friendly progressive cultural blob has become a more economically important demographic than the out-and-proud bigot cultural blob, so corporations will plausibly gain more money from it than they'll lose. It never ceases to amuse me when conservatives see for-profit companies engaging in a transparently cynical and profit-motivated displays--the way capitalist values say they absolutely should--can conclude that it must be the result of some sort of cartel or conspiracy. Free market gonna free market.
[1] Which is not to say that individual people in progressive spaces never do it, just that wider groups tend to be very quick to push back against it.
[2] To be very clear, this situation is much, much better than it was 20 or 30 years ago. Nowadays people who live in the right place and move in the right social circles might only rarely encounter it, especially if they're the "right" kind of queer. But it's far from entirely gone, and in more conservative areas I gather it's still pretty overt.
Remember, there's a completely parallel media universe that still believes bizarre stories like the Clintons running a pedophile ring run out of a pizza parlor. There's not much one can do to argue with them, I'm afraid. But thanks for trying!
Marx may not (apparently?) have ever used the term state capitalism, but Marxism certainly has always seen the operations of the state as an important part of th ecapitalist system. The idea of state power expanding prodigiously as a way to answer capitalism's various problems would not have been alien to Marx.
>German and Austrian Marxists certainly did not think that while Lenin was still alive!
Uh, what? Many of them were opposed to Leninism from the start.
"And when you are utterly convinced your model is TRUE, yet must keep adjusting it to match empirical evidence, isn't it pretty much the definition of a religion? Or at least of a religious faith?"
Sounds like you are describing "updating your beliefs against your priors" in bayesean analysis? The real question is at what point should you abandon your priors in the face of overwhelming evidence and that's something everyone struggles with, not just religious folks and communists.
> at what point should you abandon your priors in the face of overwhelming evidence and that's something everyone struggles with, not just religious folks and communists.
This very much was in my mind when I wrote the initial comment. But I did not want to complicate it. Glad you brought this up.
To me there is an irony so profound as to be nauseating in the fact that many rationalists will on the one hand dismiss Marx and his empirical analysis of labor markets and technological trends as quackery and then turn around and champion prediction markets, AI, and the singularity, as if all of these things aren't extensions of the same basic principles of analysis applied to the 21st century as opposed to the 19th.
Don't blame me, I am not one of those "rationalists". As the matter of fact, I think Marx's diagnosis of the ills of the 19th Century capitalism was spot-on; too bad his prescription turned out to be wishful thinking. Does not mean I have any faith in prediction markets, AI, or singularity.
Prediction markets may have some use. The singularity does sound like Marx’s idea of what the future could hold - massive productivity enabling people to do what they want, and where work becomes a want, not a need.
Actually, the problem is not regarding the Earth as unmoving. Copernicus also had epicycles in his heliocentric system.
The problem was that they thought that whatever the orbits were, they had to be composed of perfect circles, instead of the ellipses that they actually are.
It was therefore Kepler that was the real innovator.
(Modern perspectives on earlier thinkers can often be off like that. It is useful to reflect on.)
It also meant that, strictly speaking, Ptolemy could not be falsified as such. As a college professor pointed out to the class I was in, the system is a Fourier series, which can always approximate any function to any specified degree of accuracy.
To be fair, Ptolemy's model was highly accurate given the technology of the time. He was following the evidence he had, not stubbornly ignoring the obvious the way he is often portrayed.
Capitalism is not a designed system like theoretical communism is. It's just something that evolved historically. And what we understand as it is basically on one hand the private ownership of the means of production, and on the other –more importantly, I will argue– the financialization of the economy. The first started with the acts of enclosure and similar events in the West that ended the medieval system. The latter started with the origins of credit in Venice, The Joint Stock Corporation in the Netherlands, fiat currencies, etc.
What the Soviets lacked even when opening up trade was the financial side. So I don't think it's fair to call that ever state capitalism, unlike fascist economies, the New Deal, and basically all economies nowadays. What the Soviets eventually did was focus on making profits in their state ventures And that was because they had to trade with the West. They had no choice. And there's the simple fact that they needed a surplus of foreign currency for that.
>"State capitalism" is not something Marx ever thought of.
This gotcha shows up all the time, and it's (to use a neutral word in a spirit of kindness) meaningless. Most of contemporary leftist terminology was established after Marx's death, usually originating from anarchist thinkers. This includes "state capitalism". Marxists adopted it, because it was, well, correct. Perfectly congruent with their beliefs, too. Used as a neutral description of Soviet Union's economy by Bolsheviks themselves (up to a point, if you're looking for some kompromat on them, you should be asking why they stopped).
But aside from the fact that there was nothing to adjust in this particular case, adjusting your model to match empirical evidence is perfectly normal, and usually called "science". Imagine holding that against someone!
You're leaving out the part where your model has to eventually make predictions that hold. If your loop is making predictions, watching them fail, adapting the model so they retrodict what the previous predictions didn't, go back to "making predictions", but your model never clicks into "watching those predictions hold up", that's... technically science, I suppose, but no more useful a science than the epicycle model.
And to insist other people are dumb because they're criticizing something after your side called it "science" because criticizing anything called "science" is dumb, makes your side look like cargo cultists.
This is a good criticism of contemporary establishment economics.
But I digress. The issue in this thread is someone insisting other people are dumb because what they're actually saying isn't what he believed they were saying. (And insisting they should actually be saying what he believed they were saying, and any instance of them saying something else is "inventing epicycles", and evidence they're a religion, because if they weren't they'd be saying what he believes they should be saying... yeah, I can't really make sense of it either.)
I'm unclear if by "state capitalism" we are talking about the fascist states (public control of means of production with private ownership and wage labor etc etc intact) or Soviet style bureaucratic collectivism? Or just a phase of the proto-communist society prior to the "withering of the state"?
From Hayek, I take the view that fascist states (and perhaps America under FDR) evolved as a third way position, they were anti-liberal but also anti-communist. That something like that would evolve seems like it would have been really obvious to Marx to Engels.
Likewise Marx envisioned an iterative process, not arriving full formed at the stateless society, that part wasn't novel to the USSR. My understanding is the novel part Lenin added was the "vanguard of the proletariat". That the early forms of this could be called "state capitalism" again doesn't seem like something Marx would have missed, as it was pretty clear even before Lenin that workers just walking in to seize factories wasn't going to work due to the existence of state power.
Now perhaps Marx did not envision the nomenklatura and the tendency of these ventures to stall out in bureaucratic collectivism, as this theory was developed later by Trotsky based on new evidence, and it was not simply a recapture or subversion (which I'm sure Marx could imagine) but a genuinely New Class.
>public control of means of production with private ownership and wage labor etc etc intact
More or less. State, not "public" (especially if you want to include fascist states where there's no pretense of the public having any say), and private ownership is not a prerequisite (as Soviet Union at certain points demonstrates). The original point was that state involvement (alone) will not change the fundamental nature of the system. This... had happened, arguably all modern states are state capitalist by the original formulation, and we mostly just call them capitalist by now. This left nominal workers' states like the former USSR and contemporary China as the only remaining meaningful/interesting application of the term.
I guess what you're asking for is - no, this is different from Marx's early stage of communism - money and markets are still used internally, workers work for wages, the country participates in the broader global capitalist economy. Marx called for a global revolution, he didn't believe a communist society was possible without it. So when the global revolution didn't transpire, it left the places where it did transpire with the question - okay, what now? They couldn't proceed with implementing communism because they were stuck in an economic (and military, let's not forget military) competition with the remaining capitalist world (as it stood, much better developed at the time). The solution was - keep doing capitalism. After all, one thing it's good at is, well, economic competition and development. Not as an end goal, but as a pragmatic least-bad option, a stop-gap until the rest of the world comes around, which they were positive they can pull off as long as the workers are ultimately in control. (Well they weren't, but that's another story entirely.)
See Paul Brinkley's comment. If you keep making predictions and they never come true, and you keep modifying your model, and they STILL do not come true -- at some point you must admit that something is fundamentally wrong with your underlying theory. Otherwise you have the equivalent of epicycles -- a model which matches past evidence very well, but still has no predictive power.
Again your claim (stated rhetorically) was that Marxism is a religion, so whether the model is wrong or not is irrelevant.
Religion is an attempt to help people cope with uncertainty by inventing rituals and figures. It is very rarely that religions even acknowledge contradictory evidence, they don't have to because religion is not trying to be predictive. When they do, they will usually invent explanations from thin air rather than referring to any empirical phenomena.
Wrong or not, the Marxist Leninists did actually try to make a predictive model of reality for the purposes of solving an economic problem by actively analysing their environment, which is both practically and fundamentally different from religion.
You're using the steelman definition of religion to refute a point that was using the strawman form. If you want to criticize the OP for using that definition, that's fine and all, but the purpose of the OP wasn't to criticize religion; it was to criticize Marxism by claiming it was adhering to religion-in-bad-sense.
Which is what your third paragraph pushes back against, but in that case, the OP position seems to be that Marxist-Leninists employed a simulacrum of scientific analysis, tantamount to assuming the conclusion and interpreting the evidence to fit. And perhaps expressing it in scientific jargon so that it would appear so; if so, this only makes it doubly bad.
My historical background on Marx and Lenin is admittedly scant, but somehow I doubt that they bowed to the critics who presented counterevidence. And besides, the OP wasn't referring to Marx and Lenin specifically, but more to their more recent adherents, particularly those who invented the term "state capitalism". They appear have adhered more to the dialectic than to the scientific method. Who were the strict scientists in the Marx-Lenin camp who pushed back on them?
And this is without getting into the question of whether religion-in-good-sense is doing people any service by walling them off from reality in order to help them cope.
Marx famously engaged with counterarguments all the time. Engels, the man who first wrote about state capitalism, even went on to deny some of his models by the end of his life. Leftists as a whole are famously known for disagreeing about whose model is most correct, even today. If you were more specific about which people you were referring to it might help, but it sounds like you don't really know the history or want to understand things, you just want to discredit a group you dislike.
At best you can argue some subgroups of some Marxist ideologies are religious, which is just a useless claim. Plenty of other things are religious in that case, including physicists who believe in string theory. We don't discredit all of physics as religious.
Eh? Likening belief in string theory to religion isn't as useless as you seem to imply. If true, it points out a flaw in the reasoning model, and the first place to look for evidence that the conclusion might be false. And yes, it allows us to draw distinctions between scientific physicists and religious physicists.
Which is why I asked about whether there were strictly scientific Marxist-Leninists. Contrary to what you hypothesize, it wasn't an idle rhetorical question; I genuinely wanted to know. (How would you prefer I asked?) If Marx himself (or Engels) tried to be scientific at the expense of his own thesis, same applies.
Admittedly, knowing this is of limited use. Both fellows are dead, their most important ideas did not produce working predictions, their most prominent disciples wreaked a great deal of havoc trying to implement them, and that damage tends to overshadow any extent to which their less prominent scholars happened to be scientifically open. The historical narrative might change. But so far, if it does, it looks like it will just say there were even more reasons the those Marxists were mistaken.
The rest of this thread reminded me of another property I frequently notice in Marxists: sophistry. In this case, conflating two senses of the same word, and later making claims that don't hold and then criticizing their counterpart for ignoring the claims they fabricated. And topping that off with "that's just how words are used", belying a casual approach to words when they use it, and strict when others do.
I'm trying to remember the last time I discussed Marxism with a defender that didn't descend to this level. It's easy to remember the opposite (cf. marxbro). This could be selection bias, but weak selection bias usually goes away when making an effort to remember counterexamples.
This forces me to question whether Karl Marx engaged with counterarguments without similar descent.
You can go and read what he wrote about and to whom he wrote, the information is available. Quite ironic to accuse others of sophistry when you refuse to inquire.
10 % on Ukrainian victory (up from 9 % on March 13, 2026).
I define Ukrainian victory as either a) Ukrainian government gaining control of the territory it had not controlled before February 24 without losing any similarly important territory and without conceding that it will stop its attempts to join EU or NATO, b) Ukrainian government getting official ok from Russia to join EU or NATO without conceding any territory and without losing de facto control of any territory it had controlled before February 24 of 2022, or c) return to exact prewar status quo ante.
24 % on compromise solution that both sides might plausibly claim as a victory (up from 23 % on March 13, 2026).
66 % on Ukrainian defeat (down from 68 % on March 13, 2026).
I define Ukrainian defeat as Russia getting what it wants from Ukraine without giving any substantial concessions. Russia wants either a) Ukraine to stop claiming at least some of the territories that were before war claimed by Ukraine but de facto controlled by Russia or its proxies, or b) Russia or its proxies (old or new) to get more Ukrainian territory, de facto recognized by Ukraine in something resembling Minsk ceasefire(s)* or c) some form of guarantee that Ukraine will became neutral, which includes but is not limited to Ukraine not joining NATO. E.g. if Ukraine agrees to stay out of NATO without any other concessions to Russia, but gets mutual defense treaty with Poland and Turkey, that does NOT count as Ukrainian defeat.
Discussion:
Since my last update, several developments incrementally improving Ukrainian position have taken place. Each of them by itself would not be that important, but together they amount to something. Below I’ll list some though not all them.
Ukrainian front, both military and domestic, appears to be holding surprisingly well in last few months.
Massive improvement in relations between Ukraine and Hungary, (see e.g. here: Ukraine moves closer to EU accession after Hungary deal ) which was likely but not certain after Orban’s election defeat, is clearing the way for more EU support.
* Minsk ceasefire or ceasefires (first agreement did not work, it was amended by second and since then it worked somewhat better) constituted, among other things, de facto recognition by Ukraine that Russia and its proxies will control some territory claimed by Ukraine for some time. In exchange Russia stopped trying to conquer more Ukrainian territory. Until February 24 of 2022, that is.
I've been reading these for a while, and I'm curious about your thoughts on the difference between your updates and the broader reporting on the conflict. Your estimates have been quite stable through time (while I wouldn't expect large swings from update to update, they seem to have been random-walking around the same values for nearly the entirety of your coverage), despite a host of things that seem like they should cause updates in one direction or another.
While this isn't something that I keep a close eye on, the articles that I happen to encounter seem to be more in line with relatively large changes in the status of the conflict. Things like the transition of the conflict to drone warfare in large part, Ukraine's ability to reach into Russia, the Iran war and it's global effects, among others, all of these seem like relatively large changes to which your estimates don't seem to shift very much, so I'm curious what you have to say about that.
-edit- I just double checked, and in fact your very earliest estimate (or at least the oldest one findable by following your link chain) from over 4 years ago has Ukraine at......10% chance of victory.
Good question and good to know I have readers. Media often exaggerate an importance of day-to-day developments, in this war as well as in, say, election campaigns; something which Nate Silver, from whom I am of course trying to learn this prediction “business”, often emphasizes.
In addition, worst mistake I made in this series was an over-upgrade; I’ve started with what I still think very fairly reasonable odds, but then, due to quick Russian advance, I’ve moved them clearly too much against Ukraine right before Russian retreat from Kyiv region. I hope I’ve learned not to be too overconfident after that. In January I made a retrospective graph where it is all too visible: https://aleszieglerenglish.substack.com/p/special-military-graph-updated-again. That graph however also shows I made some other large upgrades, which I do think were justified.
Re: transition to drone warfare, it is unclear to me which side ultimately benefits from that; now it seems that Ukraine has the upper hand in drones, but few months ago it was Russia., and I guess it might swing again. Something similar is true also about Iran war, even though according to many experts it benefits Russia. Of course higher energy prices benefit Russia, but the war has also other repercussions, which are good for Ukraine. For example, it looks like a disaster for US Republicans, and also it makes Ukraine expertise in drone warfare seem useful to countries which are clearly locked out of using Russian expertise in it.
With regards to succesess of Ukrainian drone strikes against Russian mainly oil industry in last few months, I originally wanted to list it in this OP among incremental factors making me more optimistic about Ukraine. Then I forgot to do it, ups.
The Bible review struck me as 1) probably written by AI, and 2) far too brief and surface-level to be a serious contender in an ACX book review contest. What made you choose it as one of your favorites?
Ironically in this case probably refers to the sort of wit where you say the opposite of what you mean to get your point across. Two days in London that are constantly cloudy: “Sure glad I brought my Ray Bans!”
In anime, it's common for highschool students to be shown getting part time jobs. I can recall a few cases of this happening in American media and I know it happens in real life too, but it still seems like a huge difference that presumably points to cultural and economic differences with Japan.
It used to be more common than not in the US also. The change seems mainly in the US that our high school kids are too busy. I used to think this was some status flex by parents but now I have a freshman I see it’s not - with one or two big extracurriculars plus homework plus modest socializing they have no schedulable time they can use for a job. My kid tried dog sitting for a neighbor and that didn’t work because extracurricualrs kept getting in the way
I wonder how the numbers vary for part-time jobs during the school year vs. on summer (or winter) break.
I didn't have the time or energy for a job other than very occasional babysitting during the school year. But I worked every summer and some winter breaks starting at age 14, unless I had an internship then instead. That was 25 years ago though.
Eldest son has two volunteer jobs this summer, one of which might turn into a real job. I don't see a lot of local businesses hiring teenagers for the summer the way they did in the tourist town I grew up in, but it does seem like most of the highschool kids are doing something that's not just education or recreation - volunteering or working or interning.
> one of the big construction sites was all illegal immigrants
How did you find out they were illegal?
I often come across building sites around here with the workers speaking languages other than English to each other, but I honestly have no idea how (or why) I'd check their documents without risking a punch in the face or worse.
Just an interesting comparison on the point of US kids being too busy for part time work:
While I was living in japan, ~8 years ago, it was not uncommon for high schoolers to go to school in the day, then their (informally mandatory) sports club, then juku (cram school) and then part time work until around midnight before heading home. They'd get their sleep in the class and on the train. Out of club, juku and part time work, the first two are basically obligatory but some kids needed or wanted work too.
I remember getting odd looks when I mentioned that cram school might not be necessary if they actually had enough sleep to hear what was happening in class.
Yes, I was thinking about this. I suspect part of it is the relative uniformity of Japanese society and expectations. In the situation you describe, basically all kids are available from when juku ends to the close of business for the day, so if I'm a business owner I can have clear expectations. In the much more individualistic and heterogenous US society, different high schools have different dismissal times and then different extracurriculars will have different expectations on different days, and then kids get homework irregularly and have different systems for getting it done. So even if I'm just a family who need a babysitter, I can't even get one clear evening a week where a given teenager will always be available, and there definitely aren't any nice clear expectations as to when I should generally expect I can find someone available.
I tried really hard to hold down a part time fast food job in high school. I only made it about 8 weeks before my manager stopped even trying to find shifts I could work. I could occasionally swing a closing shift or two a week, but between school and sports I almost never had 4-6 uninterrupted hours during the day to spend doing anything. And this was 20 years ago, before the extracurricular rat race really got thrown into high gear.
As recently as my high school years (... Not so recent, I graduated in '99) I would guess that about half my classmates had some kind of job. Babysitting, construction, lifeguarding. My parents marched me down to town hall for my working card the day I turned 14. They were fine with my 3-hour, $15.45 a week library job. The point was not so much to make money (though I put it to good use), it was to get work experience
For context this was in rural/suburban New York state.
But yeah, kids in the US don't seem to have jobs as much now. As someone who has hired a lot of college students for work-study jobs, I think this is really bad. People who have worked before, ANY kind of work, are miles more mature. More ready for school, more ready for work m
Another factor is that high school students are competing with more adults and older workers for the same entry-level jobs in retail, food service, landscaping.
I have noticed over the past 5 years or so, nearly all of the classic "high school jobs" in my area have been filled by immigrants. Lifeguards and babysitters seem to be an exception though.
As a 14 year old in the UK (circa 1980), practically all of my friends had jobs — mostly delivering papers etc, or working in shops. When my kids were in high school in California, practically none of their friends had jobs. I think that’s a shame. A little work experience in your teens makes a huge difference to your confidence when you get your first proper job.
Two quick responses to other comments:
1. Too much homework now? Back then, we had way more homework/projects/etc than my son’s generation.
2. Too little employment? The UK at that time had the worst unemployment figures ever.
I think the differences are cultural: we have decided that children need to stay children for longer.
I also don't buy the "competing with adults" thing. I have so many parent friends who simply can't find high school age babysitters. A part of me suspects they don't really want to because they don't trust them, but I also believe them when they say there aren't enough.
When I was starting my first "real' job, my older colleagues were able to judge (with 100% accuracy) which of my cohort of recent college grads had worked a high school or college job. That minimum wage job did give me some discernible soft skills apparently.
Yes, though it would be nice to see just 16-17 yr olds. And an adjustment for full time student enrollment. High school dropout rates have declined since 1999 as well.
I graduated in '96, in a mid-size midwestern town of 150k with a protestant and rural/suburban culture. I would say that in my high school every junior or senior either had a job or was trying to get a job, unless their parents were upper middle-class or better. Boys needed money for dates or saving to buy a car, and girls needed money for I assume makeup and clothes and all that. I am absolutely baffled about these current teenagers that don't have jobs, are mommy and daddy just supposed to hand them cash for everything they want now? Do the kids in lower-income households just passively accept that they will look worse and have less since it's not being handed to them?
And what in the world are they *doing* all summer? The striver kids with middle-class parents back in the 90s would have Boys State or music camp or whatever for a couple weeks and then probably a family vacation, maybe now they have volunteerism or some other career prep thing. But the lower-income kids would be working 25-40 hrs a week, if they aren't doing that now are they just sitting on a couch in the basement for 10 weeks?
I bet working-class kids do still have jobs. Embarrassingly, though, I don't know any working class people anymore, so I don't know. But yeah, I'm guessing it's only the middle class that has changed.
It's actually the opposite. Part-time jobs for high school students in Japan is far less common than in the West. It happens in anime because the author wants another setting / more characters for the main character to interact with.
This was my impression. I dont think Ive ever encountered a hs student working in japan aside from school related volunteer work. Seems incompatible with cramming needs. They dont build resumes they have to max out their score.
Narrative device aside the generation of anime authors may be writing about an earlier time.
Wonder if it's legal differences too. My company sometimes hires highschool students, youngest I know of was...I think 16? But there's a lot of restrictions on what they're allowed to do, even though grocery stores seem pretty straightforward. Can't run the baler to crush cardboard, can't work the alcohol sections, can't break down pallets, can't have a blade in their boxcutter, can't run a register. So mostly they end up getting assigned to clean the store, collect carts, work sections where product isn't boxed (say, produce)...which is fine, but there's a reason we don't ordinarily hire seasonal temps, or "checkers" who just do register, etc. Really limits their value to the store, and of course you can't commensurately lower compensation *too* much or people won't take the job. Plus from their end it's just really boring having a restricted duty set.
I get the feeling that such For Your Own Good safety restrictions were less common in the past/in other countries, which might address part of the gap.
Huh, Are you a fulltime employee? With full time benefits? I work part time at a local tavern/ restaurant. ~$20/ hr. But life is cheaper here. The price of a glass of Guinness went up to$6.25 (from $6.00). Still a deal since you have to leave a tip.
I am. Part-timers are eligible for most of the same benefits I get, but it's harder cause they aren't guaranteed to hit the minimum number of hours to qualify. Especially during our slow season...there have been years when management contemplated cutting their own hours, so my store wouldn't have to transfer or lay anyone off. Better not to over-hire in the first place, to avoid such painful choices.
My kids (19 and 16) have both had part-time jobs in high school, and so have the majority of their friends. Almost all of them are college-bound, but I think that cuts against some theories about this.
I work part time in a local restaurant. It's all part time, the owners can't afford the tax, healthcare costs of a full time worker. Oh, I love working here, and the good bartenders, waitresses make a nice dime.
Does anyone have experience with relationships that have large gaps in intelligence? I've found myself in one and I'm curious what the long-term outcome is likely to be in terms of happiness. My primary concern is whether we will have children whom I can view as intellectual equals and how I will feel if not.
Secondhand; one of the smartest people I know has been happily married to one of the dumbest for longer than I've been alive. They're both people-people who love to talk, he loves to impress her and she loves being impressed. She had one child from a previous marriage and at least from this perspective the three of them have been close and happy the whole time.
My guess is the "people-people" part is the most important.
Depends do you like your partner? Do you respect them? Do you like talking to them? Can you trust their judgment? I know that's a basic question but there are surprisingly a lot of people who don't like their partners. They don't care for their partner opinions or even confide in them. They like what their partners provide for them and that's what keeps them long term together.
You know you probably don't have to like your partner to be happy long term as long as your partner provides what you need from them. Historically marriage was a social and economic arrangement and not a long term happiness arrangement. So what does your partner provide for you and is that fine with you long term?
And with children... if you don't at least respect your partner that will make your life shitty. Especially if you're a guy who does the stereotypical thing of leaving the child rearing entirely to the mother. These men suddenly discover themselves to be aliens in their own household that they pay for. She may not be as smart as you are but I do hope you respect her point of view.
> Does anyone have experience with relationships that have large gaps in intelligence? I've found myself in one
In general people can be happy, but as with any power imbalance, if this is something either or both of the people involved find themselves dwelling on / worrying over to the point of calling for opinions from strangers about, that's honestly a pretty bad sign for the relationship IMO.
Is it necessarily a power imbalance? For example one person could be better looking (probably matters more in my experience than anything else), career/money, age (younger or older depending), charisma, or just temperament. I think people know where they stand in a relationship and low iq people often aren’t dumb
If you want tall children, be with a tall person. If you want smart children, be with a smart person. If you want your kids to have a hereditary trait, your partner should have it. I don't know how any advice you get could help you with your particular concern.
However, I think you're probably overestimating the importance of raw IQ. Many of the very smartest people I know are unhappy and relatively unsuccessful. Presumably your partner has positive traits, otherwise you wouldn't be with them. Do you want your kids to have those specific positive traits as well?
At some point, the marginal utility of an additional unit of attractiveness or emotional resilience exceeds that of intelligence. You might be surprised how low the IQ threshold is for that to be true (especially considering overall life satisfaction vs socially recognized markers of achievement).
Also, your kids are going to be different from you in major ways. Practically speaking, do you need them to be "intellectual equals" so you can have similar views about politics or something? Whatever specific bonding activity you're imagining might never happen anyway. It's dangerous for parents to romanticize future children too much, you shouldn't be too disappointed if they turn out very differently than you imagined.
They won't even be intellectual equals until around eighteen or so, anyway, if he's hoping to have long philosophical conversations on elevated subjects. Unless they are baby geniuses even smarter than him, before that age their opinions are unformed and merely copying those of the adults around them and they don't have life experience to have any interesting and unique experiences to share.
So is he going to ignore the kids until they're old enough to hold up their end of "well it was worth talking to you, congratulations Offspring Number Two, you did not bore me" chats with Pater?
Eh, just get used to being the dumbest guy in the room. Builds character. Let her and the kids handle the hard stuff, you stick to doing internet comments in the middle of the day and don’t burn down the house
It doesn’t bother me, but I think a majority of the population would find it cringe and arrogant. I know this is a safe space for that sort of thing but just saying
I foresee a long and happy relationship. Between some other people, unrelated to this thread.
As for you and your partner: one or both of you is going to be miserable. You, because you seem to look down on your partner, or your partner, because they'll feel your contempt.
Besides, as you've probably realized already – what with your superior intellect and all – your children's intelligence won't simply be the interpolation between your's and your partner's intelligence. There's a considerable variance involved, and the at most 1–2 children you're likely to sire are too few to form a nice bell curve around the expected value. It could go either way, even with a similarly condescending partner, so if you want to be with intellectual equals, don't have kids, join Mensa instead.
> if you want to be with intellectual equals, don't have kids, join Mensa instead.
In my experience, most Mensa members are *not* intellectual equals of an average ACX commenter.
Consider base rates. There are way more IQ 130 people in population than IQ 140 or IQ 150 people. So the organization that in theory is "IQ 130 or more" ends up in practice as "IQ 130, only rarely more".
Then consider luck. Let's say that an IQ test gives you +5 or -5 points depending on whether you have a good day or a bad day. Given that there are more IQ 125 than IQ 135 people in the population, most people who pass the Mensa test are actually IQ 125 people who had a good day.
Mensa is a good solution for someone with IQ 125, but if you are the kind of person who feels at home on ACX, you probably won't feel at home in Mensa.
Yes, I've dated a few dumb women and it's always a disaster. We just saw the world fundamentally differently and after a while I would essentially stop talking to them because it was almost impossible to discuss anything even remotely interesting to me. That ends up feeling extremely lonely. Unless you have an unusually strong emotional connection then I think that large IQ gap relationships are doomed to failure.
Have you been tested or you going off identity performance? I used to feel this way but as I’ve gotten older my understanding of intelligence broadened.
Obviously if you don’t see her as an equal it can’t work. I’m just saying people legitimately have value for things besides abstract shape rotating
A lot of my friends and family have noticeable gaps with their partners, there are 2-3 for whom it's probably 20-30 IQ points (including my in-laws), none above that.
I think the best way to think about IQ effects is that they're unusually persistent rather than overwhelming in any given moment.
Like with life outcomes more generally, intelligence isn't a brilliant predictor of compatibility, which probably explains why you're in this relationship to begin with. Loads of other factors are similarly important, but looks, lifestyle preferences, hobbies, interests, even personality are likely to change over time, while an IQ gap will persist for all of your life.
Decent outcomes I've observed from IQ gap relationships:
- You have other compatible traits and get on well together without shared intellectual interests.
- The less smart partner idolises the smarter partner, who likes being admired in this way.
- You develop distinct household roles - both partners spend most of their free time on their own things, but fulfill their household roles adequately. (Sounds like something from an 18th century housekeeping guide, but can work well)
- Good luck with kids being on the smart side
Bad outcomes (often the downside to the same relationship):
- Gradually divergent hobbies - especially if the smarter partner builds intense intellectual interests with a high barrier to entry (e.g. rationality)
- Negative vibes - the smarter partner gets contemptuous and dismissive, the less smart partner gets insecure and resentful
- Bad luck with kids...
Overall, I would probably discourage intelligence gap relationships at the margins. I don't know the counterfactual, but I think 1 or 2 of my smart friends made bad choices here that could have been prevented if they had thought explicitly about this.
For kids, you can actually run the numbers to work out what range of intelligence you'd be comfortable with, and look at the 10th percentile of the expected IQ distribution given your respective IQs. I think the question is more "will I enjoy spending time with this kid?" rather than "will they be my intellectual equal when they're 21?"
When I got on this thread there were 9 posts, and 3 of them were people posting links to their work:
David Gross
"'Civil Disobedience,' a thread by @hdthoreau" presents Thoreau's influential essay as though . . .”
tailcalled
“I recently published Beyond the lexical personality traits: What is . . .’”
‘Clark Gaebel
“I recently wrote about how a retail investor can invest in ‘the whole market, except . . . “
I’ve been bothered for a while by how *many* of these appear on open threads, and wonder whether many others are, or only few. Sure, it only takes a moment to skim one of these and move on, but somehow having to do a lot of that dilutes the experience of browsing the thread, at least it does for me. Also, most people who come here to plug their blog posts are not sticking with the guideline that Scott (I think it was Scott) gave, which was that posts of that nature should include a paragraph, one that is itself of interest, about the piece the writer is plugging. But personally, I don’t think I’d be less bothered by plugs of that nature. For me, these little self-advertisements are like billboards blocking the view of the wooded hills I’m driving through.
I would support a blanket ban on self-promotion. If the article is of interest to the community, somebody else will find it and bring it up organically.
Sounds like wishful thinking. How many articles out there *could* be of interest to *the community*, and how many of them will be brought up organically by some person who happens to not be the author?
I feel it's made more bothersome by the fact that these promotions are usually from people who otherwise don't comment much*, and they don't often try to explicitly connect it to things of direct interest to the ACX commentariat, recent conversations that have come up, etc.
As an example of someone who does self-promotion that I don't mind, Performative Bafflement comes to mind - he writes things on his Substack very much in line with the ACX community's interests and sense of style, comments regularly otherwise, and typically posts a link to one of his posts in response to a conversation in which it is relevant.
*Of course, I myself lurk much more than I comment, but then again I'm not trying to promote my own work here.
Actually, I also don't mind having regular commentators mention posts on their blog, especially as part of discussions where the link's relevant. That's a whole different thing, and seems fine.
That feeling when seeing a link from a non-commenter is probably because it comes across as someone trying to boost their own substack _and_ is doing the bare minimum to get there. If you make a "Scott rule" where you first summarize your post in the comment, they'll do that; it'll be exactly one sentence, here, now let's get to the link, the link, it is _imperative_ that you click the link. It's a quintessential bad faith comment, since the idea is supposed to be that you're here to stimulate discussion, and they're actually just trying to pull in traffic.
That's probably why links from long-time commenters don't bother us. (And it would if they stopped otherwise commenting and did nothing _but_ post links.)
I directionally agree, but I would be less bothered if they followed Scott's rule.
In the last few days, Substack released a new feature called "Reply Rules" that makes it possible for creators to explicitly set a rule for replies and then an AI automatically hides replies that do not follow the rule. Using this for making the self-promotion rule actually enforced could be a good application.
I don't know what the currently set rule is, but right now, this AI sometimes hides (in my opinion) innocuous comments. For example, Deiseach's comment about the Hanania review got hidden. This is not a problem for those who download the comments using the API directly, because that still serves the hidden comments, but for those who use the website or the app it might be annoying (it would be for me).
I noticed, I wondered why, then assumed it was because I was snarky about the reviewers. So it's an automatic "let's all be nice" rule by AI? Makes more sense! I did get a little less than gruntled when I saw the company I was keeping amongst the âmes damnées, but I guess if I get down into the gutter, I can't complain about the pigs 😁
Yes, the testing comment got me dragged under the Veil of Isis. So some words are clearly fighting words. Further investigation may be warranted to see if it's terms like "hate" or if it's "poopy-head" what done it 😁
Hard to say what an algorithm finds objectionable. This question broke some rule for ChatGPT:
I watched Manhattan a long time ago and feel like watching it again. Want to clear up one thing for my own comfort. What is the age of consent in New York?
“This content may violate our terms of use or usage policies.”
Mariel Hemingway was playing a 17 year old in a relationship with 42 year old man played by Woody Allen.
I mainly wanted to see it for the gorgeous B&W cinematography of NYC and of course the Gershwin, Rhapsody in Blue.
I can see both sides here. I don't want to see most of the low-quality self-promotion spam. On the other hand, I don't want to get into a situation where saying "X" is allowed but "X, and here's a link to where I make a logical argument for it, consider objections, and provide evidence" is not. You could have a compromise by which:
1. Self-promotion spam from new accounts is not allowed.
2. Anyone who engages in self-promotional linking voluntarily assumes the risk that if Scott or other mods considers it not high-quality enough, that person can receive a temp-ban.
My desired goal state is "if the post would be valuable without the link, then it's okay to link to your site. Otherwise, it is not okay."
This cuts down on drive-by "I wrote a critique of X, here's my blog" but still allows for users who are actively engaging to share a link to further information. My current gold standard is the Fog of War posts, which are great on their own but the previous one included a link to a blog with further material.
I (somewhat) disagree. Obviously if everyone did this it would eliminate all dialogue but I find enough of the linked essays interesting to justify the clutter. I'm sure Scott would enforce a blanket ban before it got unmanageable.
I came to the thread to self-promote my own work, and then scrolled through the comments and saw yours. I'm sort of interested to know what you would like to see, instead (besides sample paragraphs).
Aren't we all here promoting ourselves?
"Look at me, I have opinions and experiences that are of interest to others!"
The main difference is that some people self-promote in a comment, some link to other things. I'm not saying there's anything wrong with your complaint, but I enjoy seeing a lot of these posts, it feels like I'm getting in on the ground floor of something that might become bigger (like Naval Gazing was pretty great)
>I'm sort of interested to know what you would like to see
If I had my druthers, nobody would link to their own work unless doing so is relevant to a discussion they're participating in. And, well, I'd make an exception for people who have been actively commenting for a long time. I'd be fine with someone like that posting something like "you've been hearing my ideas about the difference between AI minds and human ones for quite a while. Well, I wrote a short story about how the difference might play out -- here's a link."
And by the way, I m now a lot less interested in checking out your Substack because you don't communicate straightforwardly.
>I'm not saying there's anything wrong with your complaint
Yes you are.
>Aren't we all here promoting ourselves? "Look at me, I have opinions and experiences that are of interest to others!"
Ugh. This is so specious: "Hey, so I chopped up a random guy. Aren't we all murderers? Most of us eat meat, after all. And even the vegans are swallowing a few water fleas and rotifers if they drink tap water."
You’re right, sorry. I am being defensive, and my comparison probably wasn't fair. There is a big difference between sharing a thought in a conversation and dropping a billboard for traffic. They're not the same. Your analogy is reasonable: when people treat an open thread like free ad space, it dilutes the community feel.
The reason I defend these links is that I view these comment sections as a decentralized thought-leader (? blogger? Internet celebrity?) incubator. Some of the best writers online started as nobodies dropping links in the wild, hoping to find an audience that cares about niche topics.
When someone links their own work, I see then making themselves vulnerable. It’s someone saying, "I spent twenty hours building this, and I'm brave enough to show it to the smartest corner of the internet."
Scott's paragraph rule is a great filter for lazy spam. But if we say "no self-links!", I think we'll choke off the supply of independent writers before they get off the ground. (Or at least contribute to it) These threads are one of the last places left for high-quality, un-algorithmic discovery. I really enjoy them for that.
I don't expect you to love the billboards, but I hope seeing them as a pipeline for finding new, hidden talent makes it easier to tolerate.
Every post somebody puts up already contains a self-link to their blog, if they have one. If somebody tends to make comments that get my attention I pretty much always check out their blog and have found several that I now follow that way. So I don't think there's much danger that the new crop of writers will die off unappreciated if they don't come here just to spill some seed on the ground. Sorry, that was crass. I should have said, "if they don't come here just to piss on our fire hydrants."
Not everybody is self-promoting. Some people have questions they want to put to this community; others want to share something they found (created by someone else). That's meaningfully different from dropping in just to say "I wrote about this thing [link] kthxbai"
If your work was good enough for me to read someone else would link it. There are a tiny number of good blogs out there; Scott's number of about one good new blogger per year is about right. The rest is just clutter.
The only people I read have either heavily invested in being interesting enough to read more (I think Performative Bafflement has already been mentioned) or are self-evidently interesting and accomplished individuals (Cate Hall, Hannah Ritchie).
I'm never going to click on a rando's link, life is too short and average quality too low.
I'm someone who reads all the time but rarely comments.
If I write an effortpost, however, I will want to share it with all y'all.
I mean, I wrote a review of that new "Backrooms" movie and it never occurred to me to advertise it here because I am sure that all y'all have other reviewers that you prefer to read and it wouldn't be cool of me to say "here, I saw a movie" because you're not aware of my tastes and prejudices and I'm not going to try to socialize myself now at this point.
But when I wrote an Election Autopsy following the DNC's abortion of one, I thought "these dudes will be interested in my autopsy because my conclusion aligns with the SSC worldview, kinda".
If I need to show up and kvetch about the culture war, talk about my sexting with AI (it's purely for research purposes!), or discuss how you seriously need to start using a slow cooker as part of your weekly meal prep, I'm not sure that I'll be able to do that organically enough to get folks to get you (I mean you, specifically) to say "okay, it's fair that he linked to his essay".
My essay either "fits" here at ACX or it doesn't.
And, if it doesn't, I expect that it'll either get ignored or someone will tell me that my essay sucks.
The problem is, some people are better at taking feedback than others. I think your posts are perfectly okay, and the evidence is if your post starts a debate (that isn't pure criticism). But some people just keep posting links even if everyone ignores them. And of course it is a possible to make a mistake and assume that something will be more popular than it actually is...
> My essay either "fits" here at ACX or it doesn't. And, if it doesn't, I expect that it'll either get ignored or someone will tell me that my essay sucks.
I see no problem with this approach. The problem is e.g. if three essays in a row get ignored and the person keeps posting.
As someone who has promoted some of my own articles in Open Threads in the past (while following the introductory paragraph guideline), I'd welcome more guidance from Scott on this matter, or some kind of formal vote.
I'm happy to abide by a formal rule, but don't want to refrain from all self-promotion solely on the basis that it annoys some people without community consensus. I do believe any articles I post here are of interest to the ACX community. Sometimes I start out writing a comment for the Open Thread, before realizing it is so long it is best made into a separate article.
Andrew Gelman has a post comparing Scott to Edmund Wilson.
"Scott Alexander is, like Wilson and those others, a prolific writer with strong political opinions, who feels comfortable opining about–and gets respect for his opinions on–all sorts of things. Like Wilson, he’s an autodidact and a bit of an outsider. Alexander’s topics are different from Wilson’s–more on science, less on literature–but he has the same feeling of well-earned authority. And a fun writing style. Bloggy rather than magaziney–times have changed–but a distinctive and readable style in any case."
The book is about our collective inability to envision futures that are fundamentally different to today, how we ended up like this and that we need to get better at this again if we want to have a chance for the challenges ahead.
I came out of that word salad not really knowing what the future world would look like, but certain it would be poorer. The obsession with reducing consumption - which is basically code for making the average guy poorer - is a dead end. And it won’t work.
For two reasons, clearly you are targeting the West. Yet all future carbon increases will be in developing countries. Europe and other advanced countries are reducing emissions, in fact the EU peaked in 1990.
The US is different but in no way as culpable as China and others in terms of what’s been pumped out since 1990, which in China’s case is catching up with Europe’s entire emissions since 1850.
The other problem is if you want no technological fixes, no carbon extraction, and only reduction in consumption then consumption or emitting activities - which includes agriculture- has to drop to near zero. which probably means mass starvation and people freezing to death.
And since there’s no extraction allowed in your scenario the carbon stays in the atmosphere. Temperatures remain elevated for a very long time even after emissions reach zero, the seas keep rising.
Kind of a long shot, but if anyone here works at Meta: There used to be this VR game called "Echo Arena", which was legitimately one of the most fun games I've ever played. If you ever read Ender's Game, it basically replicated the zero-gravity battle rooms (mostly via the paid expansion called Echo Combat). The zero gravity effect actually works quite well in VR, and can feel very immersive after a while.
Anyway, the publisher, Ready at Dawn, was acquired by Meta and then shut down, and Echo Arena (a multiplayer game) was also shut down. With the Steam Frame soon to be released, I would really like to play this game again. I do think it would be very popular if released on Steam. If this sounds interesting to you, and you know someone, maybe we could convince Meta to give this game another shot?
I worked at Meta in the VR division at the time Echo Arena was shutdown (not there anymore). IIRC, this was a minor point of controversy that came up in one of Boz's weekly Q&As.
All I can say is good luck. Meta clearly didn't think it was worth the cost even back in 2023 when the Metaverse was a top company priority. Why on earth would you think they would be more inclined to invest in it **now**?!
Also, why on earth would you think they would be interested in releasing on a competitor's platform? Meta didn't get into the game publishing business for shits and giggles. It was meant to be a stepping stone on the path to widespread adoption of (Meta's) VR. When that didn't happen, they gave up and canceled everything. They certainly aren't an ordinary game publisher who might be interested in scrabbling for pennies in a cutthroat industry.
If you really really believed in this, you could try to negotiate with Meta to buy the IP. But I doubt you'd be able to offer enough to make it worth their while.
First of all, thank you so much for responding. It's awesome to hear from someone who was there at the time. I pretty much suspected everything you've laid out here... To be honest, this was pretty much a hail mary.
I know some pretty high-up tech people read this blog, so in my ideal world, somehow this would get to Mark Zuckerberg, who would be like "Yeah, I guess that game does sound cool, maybe we should re-release it". Just like those people who get tech support by tweeting the CEOs on twitter. Of course, this would never actually happen. The team doesn't exist anymore, and the company doesn't care.
Maybe if I win the lottery one day I will reach out to Meta and buy the IP :P.
I feel like you have a better chance of convincing people to make a knockoff version of it. In fact, I wouldn’t be surprised if people were already working on it given how popular a VR title it was and how upset a lot of people were that it was being shutdown
This Youtube channel's got Thank You comments from ten days ago. https://www.youtube.com/watch?v=nGcDXHAYc2s But no promises they aren't all AI. I haven't watched the video, this is something you must do alone.
I have been having some conversations with AI about how some very different movies can use the same basic cinematic grammar of shots, and yet establish a very different mood.
For example, _Saturday Night Fever_ (1977, John Badham) versus _Stalker_ (1979, Andrei Tarkovsky)
Simplifying the shots slightly:
EXT. DAY. Bridge in Brooklyn
(establishing shot)
CUT TO:
EXT. DAY. Subway train.
CUT TO:
EXT. DAY. Street in Brooklyn.
(Start music: Staying Alive by the Bee Gees; Close up on Tony Manero’s feet; pan up to Tony Manero’s face)
CUT TO:
INT. DAY. Pizzeria
Versus…
EXT. DAY. Railway yard.
Three guys sneak onto a rail car.
CUT TO:
EXT. DAY. Close up on each of the guy’s faces as the railcar slowly moves down the track.
CUT TO:
EXT. DAY. Telegraph poles
(Switch from black and white to colour film. Very slow pan across a jumble of broken telegraph poles and tangled wire as the characters enter the nuclear desolation of the Zone)
You can argue that the speed of the camera movements reflect both the assumed viewers knowledge and the character’s uncertainty…
Saturday Night Fever. We know where Brooklyn is, we know what a pizza shop is, we can guess what kind of guy Tony is. Tony swaggers confidently down a street in Brooklyn. Fast pacing.
Stalker. We are only given hints as to where this is taking place. It looks like somewhere the Soviet Union. We can see there are guards to stop the protagonist getting in to the zone, for some reason we aren’t yet told. The zone is clearly abandoned — the telegraph poles are broken and no-one has fixed them, but we aren’t told why no-one has dared to go into the zone to fix them. The characters move cautiously — even though the viewer hasn’t been told, they know that they walking into a labyrinth of deadly radioactive contamination and shattered broken buildings. They are moving very carefully. The camera pans slowly. [_Stalker_ was made before the Chernobyl disaster; in retrospect, we clue in faster as to what we are seeing. ]
Note that neither Stalker, nor Roadside Picnic - the book the film was inspired by - have anything to do with nuclear disasters. The book is explicitly about people exploring the site of an alien visit looking for weird alien trash to sell; the film removes from this all explicit genre fiction elements and leaves things ambiguous because what Tarkovsky actually wanted to make films about was inner journeys to faith and spirituality, and weird shit you can see and replicate is just physics.
There is a much more recent computer game which tried to hitch a free ride on the fame of these while explicitly being a first person shooter set in Chernobyl, but it would be a mistake to back-project its storyline and themes, such as they are, onto the earlier works, and we do not seem to be discussing the game here.
Yes, in the book it’s aliens and in the movie it’s left ambiguous. Also, in the book it’s not even radioactivity but something that humans don’t understand, so my summary is a bit wrong.
I think the religious element is actually there in _Roadside Picnic_, just way more ambiguous. In the book, some of the characters talk about whether the aliens are malicious, benign, or just don’t care about us. This is kind of standing in for a debate about whether there is a God, and the characters in the book version just aren’t sure whether there is a God substitute or not. The movie leans more strongly on, yes there is. The Stanislaw Lem essay on Roadsides Picnic argues that the book version has evidence the aliens are actually malicious.
What I was trying to say (but kind of fumbled it): the movie was made before Chernobyl, so those scenes have connotations for the viewer now that they didnt when the movie was made.
Do they, though, if the viewer hasn't been primed for that by the game or its adverts or people talking about it? The thing with throwing the bolts doesn't exactly scream "radiation" to me.
P.S. Amusing coincidence. I picked Saturday Night Fever as an example of this sequence of shots, before looking up the Wikipedia page for “Establishing shot”. Now I’ve looked at that wiki, it picks exactly the same movie as its example for this type of shot.
But you can start with a wide shot, cut to a shot of a train, push in to a close up on the main character, then cut to the next location, in very different ways,
This could turn into an essay on how Tarkovsky does not follow the rules of the continuity style of film editing, and that, for example, while the switch from sepia to colour in Stalker might look like a reference to The Wizard of Oz, The Wizard of Oz is using colour to let the viewer know whether they are in Oz or back in Kansas, while Stalker is creating ambiguity about what’s in The Zone and what isn’t.
Why do I put two identical synths next to each other?
The idea is to use the arpeggiators to explore how different chords sound over each other. Try having one synth play a Cmajor chord, then use the other to explore going up and down the Cmajor scalein triads. Try various odd number chords (3,5,7 notes) to have each note rotate over the 4 oscillator voices.
Click 'preset' to hear Cmaj on both synths with each oscillator voice randomized
Click 'demo' to hear a short piece I made on the synth
Might be a hit in the EDM community. Me, when I’m fiddling around adding notes onto the top of triads to make weird chords, I like to just stick my spare fingers somewhere on the neck and see how it sounds.
What if we made an effort to protect AI modules from the complexities of the human world, through giving AI special economic zones carefully encapsulated away from how we think, communicate, and navigate the physical world? It's more common to worry about sealing away AIs to keep them from doing damage, but the *reverse* direction (protecting them from us) can help reduce reliance on opaque learned heuristics and open the door to formal verification and other methods that can provide high assurance about principled systems.
To start with one irony: "carefully encapsulated away" is most dramatically attempted today, not in AI, but in quantum computation. Avoiding decoherence from interactions with the environment is one of the main challenges in that field.
Returning to AI: I'm not following what would typical decisions being made by AI internal to the zone be? It sounds like your description of the AI economic zone is closer to hard-coded automation than to the recent progress in AI. I'm not seeing why, from a _computational_ point of view, the zone would be a "sweet spot" _between_ hard-coded automation and operation of AIs in the normal economic system. There _is_ certainly a point in having _physical_ areas isolated from humans for physical and chemical reasons - heat, radiation, high voltage, toxic materials, etc., but that's all long-standing practice, e.g. putting fencing around substations so passers-by don't touch high voltage wiring.
But, _computationally_ , LLMs have largely solved the problem of interfacing with natural language - expecting reasonable defaults, using context to resolve ambiguities, etc. I don't see subsystems inside an enclosed zone interacting through programming languages instead of natural languages as a clear win. The sort of things that go wrong today, e.g. with AI agents, perhaps e.g. not anticipating the possibility of some failure mode (e.g. of a machine jamming) and not having a detection and recovery process thought through, don't seem like they are ameliorated by internal communications in a programming language.
I can see the point of restricted _physical_ spaces, like factory floors, together with richer world models to improve intelligently dealing with a larger fraction of events possible within those spaces, but this isn't really restricting the _informational_ interfaces.
I'm not arguing for hard-coded automation. I'm arguing for something more like hard-coded rules of engagement, hard-coded specifications of the goals of an autonomous region (e.g. "make reliable XYZ widgets according to this schematic") -- and then the region can go through recursive self-improvement to get better and better at achieving its objective set in advance. Internally, all details of organization are chosen alongside the design of the intelligent agents that function therein. One of the examples I like to use [1] is avoiding the use of road networks like we have in the world today, full of not just of squishy evolved intelligences (us!) making driving decisions but also of stray animals, weather, and so on. Carefully controlled environments, occupied by autonomous vehicles, avoid those hard-to-model complexities, fundamentally decreasing the cost of effective intelligence.
I also disagree that LLMs have solved the natural-language-interface problem. For the sake of argument, imagine that LLMs have achieved optimal understanding of natural language. There is still much inherent ambiguity that can cause a lot of trouble [2]. Now if assume LLMs have mastered that ambiguity and understanding all of the social-context subtleties needed to intuit "what people really mean." We still face the inherently high cost of implementing deep learning [3]. There can be such an advantage, from switching to communication designed for efficient processing, that much more effective decision-making can be executed in a given time, budget of money or other resources, etc. Organizations that are able to use decision-making that is simultaneously cheaper and more reliable can be expected to win out in time.
"A subway network is a good example, with physical safeguards to keep people from running into the tunnels."
Yes, and construction sites also generally have physical barriers to prevent unexpected people from wandering into them. As I agreed with you before:
>I can see the point of restricted physical spaces, like factory floors, together with richer world models to improve intelligently dealing with a larger fraction of events possible within those spaces, but this isn't really restricting the _informational_ interfaces.
I fundamentally don't believe "Yet switching to other modes of communication dramatically simplifies processing." I think it helps _slightly_ (hey, if something can be parsed by a regular expression, great!), but I think it misses the main limitations to broadening AI applications today and in the near future.
>In such a setting, the mega-factory may even have autonomy to build new factories that follow the common rules. It may invent new decomposition of top-level deliverables into components, design and construct new factories to supply those components, and still remain compatible with its exterior interface.
I think that this forces a sufficiently comprehensive world model that
>3D world models interfacing to vision ("Do I have the right part? Are the surrounding parts properly in place? Do they match the blueprints/CAD data to proper tolerances? Can I position the next part without collisions?") and to force sensing and application.
These are substantial tasks! But they are much the same task whether the internal communication is done with natural language or a formal language.
b) I don't think formal verification will be feasible. Even 1-D bin-packing is NP-hard! _Provably_ ensuring that some design process is going to yield results consistent with a (typically complex!) specification looks very unlikely. I don't think a mega-factory that builds new factories will get away from heuristics and recovery procedures.
c) While I do agree that the signaling function of natural language is counterproductive to getting tasks done, I don't think that this is a huge problem where AI agents (1) can include a modest amount of redundant language ("to put it another way") (2) can send messages back and forth to resolve ambiguities (3) are evaluated on whether they successfully complete tasks, which miscommunication impedes
d) On a peripheral note, I don't think that trying to eliminate senses in a system operating in the physical world is a good choice. I think that usually the better choice is to _add_ senses, and e.g. disambiguate an uncertainty in the interpretation of a visual scene by e.g. adding radar
e) Very generally, I think that the right emphasis at this point in AI's development is to seek to expand capabilities, including the capability for more reliable results, typically by spending _more_ computation, rather than looking for ways to cut computation. We aren't at AGI yet, and I think that spending computation to fill in the "holes" in AI 'spiky' capabilities is the main near term task, rather than e.g. trying to transition away from natural language towards formal languages in the interests of reducing the computation spent on parsing today.
Thanks for the thorough explanation of where our positions diverge. To some extent, we'll have to wait and see empirically which approaches win out.
It seems like our biggest disagreement is that I am betting on the potential to introduce new abstractions that simplify reasoning. We depend critically on such abstractions today. For instance, few planners of factories (or much of anything else in the world) consider quantum effects directly. You seem to be arguing that no new broad abstractions are coming to simplify reasoning, and you even argue for increasing sensory capabilities to make abstractions *more* complex (if I'm reading you right).
I suppose I would be surprised if human intelligence has achieved a series of triumphs in new abstraction to simplify planning, and then superintelligence has no more for us! But it could happen, and it's conceivable any new abstractions are even of less importance than the past big ones.
I may be biased by my professional work on programming languages and formal verification. In that world, it is usually a triumph (not a barrier) to notice that a problem is NP-complete, rather than undecidable! It is also often quite tractable to prove than an algorithm solves an NP-complete problem, leaving the cost for runtime -- and then we achieve good runtime performance nonetheless, a surprising fraction of the time.
>it is usually a triumph (not a barrier) to notice that a problem is NP-complete, rather than undecidable!
True - I'm approaching this from a computer-aided-design class of problems; in a theoretical computer science class of problems (e.g. halting), very much agreed that undecidable is a frequent outcome, and NP-complete is better!
>It is also often quite tractable to prove than an algorithm solves an NP-complete problem, leaving the cost for runtime -- and then we achieve good runtime performance nonetheless, a surprising fraction of the time.
What typically happens in NP-complete computer-aided-design problems is that the runtime for finding a provably-optimal result is usually exponential, but someone is able to find a heuristic with reasonable runtime which yields a result which is "good enough".
>It seems like our biggest disagreement is that I am betting on the potential to introduce new abstractions that simplify reasoning.
Actually, I don't disagree on that, but just on the value of going to formal languages to reduce parsing time. Two examples of routinely introducing new abstractions in natural language that I'd expect to also apply to AI systems:
a) Just naming subassemblies/products/classes of them. "model np750xhd samsung laptop" is a mildly leaky abstraction
b) The nomenclature that mathematics constructs that typically depend on theorems that break objects into classes (property X turns out to imply Y and Z, so distinguishing objects with and without X now merit special names)
I see this as adding vocabulary ("This wire is at 'logic 0'") rather than syntax.
This _does_ blur the distinction between human-readable language and opaque language. When cooperating AIs introduce new vocabulary, it will become harder and harder for humans to understand it.
Got it; so your comments are about language in particular, as an abstraction that may or may not be worth improving on substantially.
I think "parsing" undersells the full activity of language processing. For instance, it doesn't traditionally include resolving words into semantic concepts. My argument about language includes that full stack of processing to guide decision-making. To me, it seems quite inefficient and worthy of reengineering.
Regarding NP-complete problems, my personal experience is more on the side of formal verification than computed-aided design. In the formal-verification world, we typically solve logical problems exactly, and the "catch" is that solvers may consume so many computational resources that, for all practical purposes, we consider them to have given up. Practitioners learn the (often-undocumented) behaviors of verification tools, to predict what sends them off the deep end. I think there's a lot of potential to attack this problem with the full intensity of performance engineering that has gone into GPUs and their associated programming tools. I'll get around to writing about those new approaches eventually.
My second standalone Fog o' War post is up on Substack (link below).
In it, I discuss the letter Zelenskyy sent to Putin and Putin's testy response during a panel discussion at the St Petersburg International Economic Forum. (1.) Zelenskyy greeted the guests arriving at the forum with a drone strike on a nearby St Petersburg oil facility and the Kronstadt naval base. (5.)
And what was Russian air defense doing to stop the attack? At least a dozen videos of the attack have been posted on Telegram and TikTok, and I don't hear any anti-aircraft batteries firing in any of them. The sum total of St Petersburg air defenses seems to have been one or two soldiers with automatic carbines trying to hit them. But Russian state media said nearly two hundred were shot down. Excellent marksmanship on the soldiers' parts! (2.)
We also learn that Putin intentionally aimed a $50-million Oreshnik ballistic missile at the garage cooperative in Bila Tserkva in the Kyiv region. (Poor Uncle Sasha! His days of tippling vodka in the garage are over. And what happened to Obama? Did he escape the garage unscathed? These are important questions! Why isn't our MSM doing its job?) (4a.)
More stuff on the worsening gasoline situation in occupied territories and Crimea, along with the mandatory pictures of long gasoline lines of burning oil facilities. (9.) (10.) But seriously, there are 2.4 million people in Crimea (and I don’t know how many are in the occupied territories). Now that gasoline is impossible to get, what's going to happen to food deliveries and deliveries of medical supplies to hospitals? Call me a nervous nelly, but I potentially see a humanitarian disaster brewing in Crimea. (11.)
Also, I discuss Ukraine's Morrigan drone, which seems to be the prime instigator of the havoc being wreaked on the highways and byways in the occupied territories. (14.) And I query Gemini and ChatGPT to find out more about the likely configurations of the AI systems being used on Ukraine's latest generation of mid-range drones. (15.)
"Axial Age (also Axis Age,[1] from the German Achsenzeit) is a term coined in 1949 by the German philosopher Karl Jaspers. It refers to broad changes in religious and philosophical thought that occurred in a variety of locations from about the 8th to the 3rd century BCE.
According to Jaspers, during this period, universalizing modes of thought appeared in Persia, India, China, the Levant, and the Greco-Roman world, in a striking parallel development, without any obvious admixture between these disparate cultures. Jaspers identified key thinkers from this age who had a profound influence on future philosophies and religions and pinpointed characteristics common to each area from which those thinkers emerged."
It looks to me like the left and right have lost their minds in a way that the center can't resist. We're an inventive species (if you say "unfortunately", I will bite you) and we could use some sort of survival-prone system we haven't come up with.
The usual thing I prefer to do with a prediction is to see if I can rule it out. You're seeing signs of an impending event; could they be signs for something else? What are the signs of something other than an axial age? Are we seeing those too?
The other thing to do with a prediction is to enjoy it and lean into it. What are the religions you anticipate? Do you see them in terms of the underlying weltanschauung and philosophy, or in terms of the trappings and rituals?
I'm not as sure that there will be new religions as that something is going to happen to the idea of nations. If there are a lot of climate refugees, borders are going to be much less enforcable.
If there are new religions, they will accrete customs, ceremonies, art, and music.
That's outside view (outside of spirituality I mean). Inside view is that Muhammad was indeed the last founder of a world religion we're getting. The world is tending towards dissolution, and finally, its end, to be followed by the next world and the next humanity, which will be born from the seeds of this world.
Before then, we're getting Antichrist, which will be a figure presenting themself as a restorer of spirituality, but which will be selling a parody of spirituality instead. They will be mistaken by many to be the founder of a new world religion.
Spirituality is never truly lost, but it does take other shapes than a Messiah with their big revelation.
One of those perplexing thought experiments again.
Would you rather fight a chicken every time you got in your car or use a sword to fight a gorilla once a year?
Attach whatever color button you care for to the choices, purple, magenta, cyan, puce, taupe, egg shell white or kelly green. It doesn’t matter what color the buttons are.
Yeah, this just sounds like free chickens as long as I'm willing to kill them by hand.
I'm also think that giving us a sword to fight the gorilla was seen a buff but "use a sword to fight a gorilla" sounds like we have to use the sword instead of a better weapon, like a Ford F-150.
Define "getting in the car". If I put three armfuls of groceries in the trunk and then drive home, is that one cockfight, or four? If I have a passenger, do they have their own chicken, or do we share one?
With the trunk, no issue. If the chicken attack while you open the hatchback, she is issued a technical foul. 3 technical fouls and she gets a red card and you win.
Yes, if you fight the gorilla you get to keep the sword.
Oh man, technical fowls expose new dangers; are the chickens limited to fisticuffs, or can they use weapons? Am I going to have to contend with a chicken with a chain whip?
Should be both of us (chicken versus human) with the weapons Nature endowed us with. Giving us a sword against a gorilla is already tilting things slightly, but that is to make up against the disadvantage we are at when facing a gorilla. Chickens are vicious monsters that don't need any help such as giving them weapons.
Jeez, I used to live on a commune where our 50 or so chickens wandered the property, and the worst they ever did to me was peck the grommets in my sneakers. They did do that quite a lot, though. Chickens really like grommets, I think. I myself am fond of the word grommet. But, um, I guess I'm not really engaging with the meat of your question. Sry.
You didn't know the insane cock (er, can I use the Irish/British word or should I replace it with rooster?) that lived in a neighbour's farmyard. That bird was absolutely insane and would attack anyone or anything that it perceived as trespassing, without provocation, without consideration for its own or any other creature's life.
To this day, I maintain that it was only thanks to the good stout corduroy pants I was wearing upon the occasion when my mother brought me on a visit that saved me from having scars on the backs of my legs for life when this lunatic fowl zoomed across the width of the yard and slammed into me, claws and wings going, without fear or favour. Luckily, even at nine years of age, I was bigger than a cock (that sentence sounds *so much worse* in American English) so no damage done, but I fear to think if I'd been wearing a skirt instead.
A reenactment friend tells Magyar stories. They make Russian tales seem cheerful and Grimm parents seem modern and enlightened. One of them concerns a "vampire" who'd take the form of a gigantic black rooster.
The first few times she told it she used the Other phrase represented by the initials BBC and had to be told why her audience was in hysterics.
There are a couple vulgarisms for female genitalia that pass without a blink when uttered by Jason Statham but would give an American the vapors if Tom Hanks ever used them.
I know chickens, I've fought (sort of) chickens before (as a kid), bring on the chickens!
Gorilla has a sword too or not? Anyway, that's too much of a challenge. I can take a chicken, vicious little dinosaur or not. Beware the beak and talons, hobble the wings, twist the neck. Chicken dinner tonight!
Claude went with the gorilla. I said “but chickens are puny and dumb.”
Claude responded:
“But consider — puny and dumb cuts both ways in a car. A panicked, flapping, squawking bird in an enclosed metal box is chaotic precisely because it’s not thinking. It’s just pure instinct and claws everywhere. Have you ever seen a chicken freak out? They’re basically a feathered blender. Your eyes and face are right there.”
Never seen a chicken freak out myself. I did see one in a South Dakota tourist trap that for the price of 25¢ would push a button to swing a small baseball bat and hit a rolling pinball to the fences and run around little base pads for a home run.
Ah, I was assuming you fought the chicken before you got into the car. Claude may be fake, but it's correct on the dangers of chickens. Do not get into an enclosed space with one, which is why it is important to strike before it can corner you.
First of all because the language of the question is more exploitable. Can I just gift my car to my wife and put the title in her name then borrow it every day? Can I generate infinite chickens to sell for profit by getting in and out of my car over and over?
But second because fighting a chicken is easy - wear good jeans,kick it once when you get to the door of the car, get in the car and close the door before it gets back to you. Plenty of people who own chickens probably have to do this with aggressive ones every time they feed them, not a big deal.
Oh definitely a chicken. It matters a lot whether this is a hen or a rooster or a cockerel or a chick, though. Not for whether a gorilla is worse, but for how annoying or trivial this is going to be. I already have to fight a hen every day because she's broody and she attacks me every time I offer her treats and it's no big deal, just a light "go away" tap with her beak. Cockerels are full of ego and think they could take on the giant 6-foot rooster, but they have no spurs and no experience, and confidence makes them easy to grapple. Roosters are the one I would least want to have to deal with, but even then, it's less than two feet tall. Some roosters are, well, kind of a chicken, so if running away counts then it's really easy.
I never, ever, want to have to fight a gorilla with a sword. Gorillas are stronger than you and almost no one knows how to use swords these days. You could make the gorilla fight be once per lifetime instead of once per year, and the chicken would still be the winner. Seriously, if you think you could take on a gorilla, try fighting a pork butt with a chef's knife first.
I infer from the question i get to decide what to bring with me on my way to the car. I.e. thick pair of gloves. If i have to fight that thing bare handed everytime i think i need to invest in sword fighting classes (not from those guys in central park though, surely some one can teach me how to maim a gorilla properly)
I'm not sure a sword is the right weapon for that. A wounded gorilla who decides "screw that, I'm going to kill you anyway" seems plausible, and it's probably hard to reliably pierce its brain or sever the spine. A missing limb or a lethal wound to the heart or neck might not be sufficient if you can't get out of its reach fast enough. I've read that hunting spears for boars had crossbars close to the tip because a boar does not care and would push itself on your spear to get to you.
Edit: Also, according to someone, in combat sports there are 4 tiers of challenge, IIRC "practicing against e.g. a dummy < sparring < competitive combat < real fight with intentional serious injuries on the line" and no matter how practiced and comfortable you're with tier n it doesn't really prepare you for tier n+1, not because of different skills required but because of different levels of adrenaline involved. They may be overgeneralizing but by default I would expect that no amount of classes really prepares you for kill or die situations.
From the nature videos Ive seen mammalian species on particular seem to give up a little early, at first blood frequently, even as prey. And i can see how this would be an evolved behavior. A gorilla no doubt could kill me with one arm missing, but I dont think that it would
I meant that the animal being attacked gives up, not the animal doing the attacking. That is whats applicable here, whether the gorilla gives up after a severe slice.
Is the gorilla already angry, i.e is going to charge as soon as the gate opens, or can I sneak up and take out the gorilla before it can react? Even if I charges, I've heard gorillas aren't actually good fighters, and having a sword is certainly a big advantage. So I'd pick gorilla.
Ethically, leaving out the question of which fight is more dangerous, my clear choice is the chicken. I'm already responsible for the deaths of some 2-digit number of chickens each year because I like to eat them. It would feel more wrong to kill a single gorilla than a thousand chickens.
It's fascinating that even in 1861, the year this book was published (and which I read many years ago), the existence of the gorilla is so disputed that nobody is sure it even exists outside of tales from the natives, so the three main characters go out to Africa to hunt it:
And, unlike the view post the famous David Attenborough meeting, the gorilla of previous times was believed to be a ferocious, violent, murderous beast:
“I’ll tell you what it is,” said he, with the look and tone of a man who feels that he has been unwarrantably misled—“I don’t believe there’s such a beast as a gorilla at all; now, that’s a fact.”
There was something so confident and emphatic in my comrade’s manner that, despite my well-grounded belief on that point, I felt a sinking at the heart. The bare possibility that, after all our trouble and toil and suffering in penetrating thus far towards the land which he is said to inhabit, we should find that there really existed no such creature as the gorilla was too terrible to think upon.
“Peterkin,” said I anxiously, “what do you mean?”
“I mean,” replied he slowly, “that Jack is the only living specimen of the gorilla in Africa.”
“Come, now, I see you are jesting.”
“Am I?” cried Peterkin savagely—“jesting, eh? That means expressing thoughts and opinions which are not to be understood literally. Oh, I would that I were sure that I am jesting! Ralph, it’s my belief, I tell you, that the gorilla is a regular sell—a great, big, unnatural hairy do!”
“But I saw the skeleton of one in London.”
“I don’t care for that. You may have been deceived, humbugged. Perhaps it was a compound of the bones of a buffalo and a chimpanzee.”
“Nay, that were impossible,” said I quickly; “for no one pretending to have any knowledge of natural history and comparative anatomy could be so grossly deceived.”
“What like was the skeleton, Ralph?” inquired Jack, who seemed to be rather amused by our conversation.
“It was nearly as tall as that of a medium-sized man—I should think about five feet seven or eight inches; but the amazing part about it was the immense size and thickness of its bones. Its shoulders were much broader than yours, Jack, and your chest is a mere child’s compared with that of the specimen of the gorilla that I saw. Its legs were very short—much shorter than those of a man; but its arms were tremendous—they were more than a foot longer than yours. In fact, if the brute’s legs were in the same proportion to its body as are those of a man, it would be a giant of ten or eleven feet high. Or, to take another view of it, if you were to take a robust and properly proportioned giant of that height, and cut down his legs until he stood about the height of an ordinary man, that would be a gorilla.”
“I don’t believe it,” cried Peterkin.
“Well, perhaps my simile is not quite so felicitous as—”
“I don’t mean that,” interrupted Peterkin; “I mean that I don’t believe there’s such a brute as a gorilla at all.”
“Why, what has made you so sceptical?” inquired Jack.
“The nonsense that these [natives] have been telling me, through the medium of Mak as an interpreter; that is what has made me sceptical. Only think, they say that a gorilla is so strong that he can lift a man by the nape of the neck clean off the ground with one of his hind feet! Yes, they say he is in the habit of sitting on the lower branches of trees in lonely dark parts of the wood watching for prey, and when a native chances to pass by close enough he puts down his hind foot, seizes the wretched man therewith, lifts him up into the tree, and quietly throttles him. They don’t add whether or not he eats him afterwards, or whether he prefers him boiled or roasted. Now, I don’t believe that.”
“Neither do I,” returned Jack; “nevertheless the fact that these fellows recount such wonderful stories at all, is, to some extent, evidence in favour of their existence: for in such a country as this, where so many wonderful and horrible animals exist, men are not naturally tempted to invent new creatures; it is sufficient to satisfy their craving for the marvellous that they should merely exaggerate what does already exist.”
“Go to, you sophist! if what you say be true, and the gorilla turns out to be only an exaggerated chimpanzee or ring-tailed roarer, does not that come to the same thing as saying that there is no gorilla at all—always, of course, excepting yourself?”
But they do eventually track down and shoot a gorilla:
"Of all the hideous creatures I had ever seen or heard of, none came up in the least degree to this. Apart altogether from its gigantic size, this monster was calculated to strike terror into the hearts of beholders simply by the expression of its visage, which was quite satanic. I could scarcely persuade myself that I was awake. It seemed as if I were gazing on one of those hideous creatures one beholds when oppressed with nightmare.
But we had little time to indulge in contemplation, for the instant the brute beheld us it renewed its terrible roar, and attempted to spring up; but both its legs at once gave way, and it fell with a passionate growl, biting the earth, and twisting and tearing bunches of twigs and leaves in its fury. Suddenly it rushed upon us rapidly by means of its fore legs or arms.
“Look out, Jack!” we cried in alarm.
Jack stood like a rock and deliberately levelled his rifle. Even at this moment of intense excitement I could not help marvelling at the diminutive appearance of my friend when contrasted with the gorilla. In height, indeed, he was of course superior, and would have been so had the gorilla been able to stand erect, but his breadth of shoulder and chest, and his length and size of arm, were strikingly inferior. Just as the monster approached to within three yards of him, Jack sent a ball into its chest, and the king of the African woods fell dead at our feet!
It is impossible to convey in words an idea of the gush of mingled feelings that filled our breasts as we stood beside and gazed at the huge carcass of our victim. Pity at first predominated in my heart, then I felt like an accomplice to a murder, and then an exulting sensation of joy at having obtained a specimen of one of the rarest animals in the world overwhelmed every other feeling.
The size of this animal—and we measured him very carefully—was as follows:—
Height, 5 feet 6 inches; girth of the chest, 4 feet 2 inches; spread of its arms, 7 feet 2 inches. Perhaps the most extraordinary measurement was that of the great thumb of its hind foot, which was 5 and a half inches in circumference. When I looked at this and at the great bunches of hard muscles which composed its brawny chest and arms, I could almost believe in the stories told by the natives of the tremendous feats of strength performed by the gorilla. The body of this brute was covered with grey hair, but the chest was bare and covered with tough skin, and its face was intensely black. I shuddered as I looked upon it, for there was something terribly human-like about it, despite the brutishness of its aspect."
Attenborough turned 100 years old a couple weeks ago. As his century mark approached I was rooting for him to make it. Imagine the changes he was witness to.
And he has the grace to acknowledge his great luck.
From his Wikipedia page:
“Attenborough had a pacemaker fitted in June 2013, as well as a double knee replacement in 2015.[176] In September 2013, on the prospect of retirement, he commented: "If I was earning my money by hewing coal I would be very glad indeed to stop.
But I'm not. I'm swanning round the world looking at the most fabulously interesting things. Such good fortune."
I just watched Prince William give a little speech in his honor on his birthday. The guy (Attenborough, not Prince William) appears to still have a full head of hair. God must really like that guy.
Fighting a chicken regularly when I'm trying to just get somewhere quickly would be intensely annoying, I absolutely hate things like that which add small amounts of friction to ordinary acts multiple times a day. However if you'd raised a child (I never did) maybe fighting the chicken is less onerous than having to get a kid into the car to go anywhere so you wouldn't mind this.
OTOH an epic battle man v gorilla with a sword seems heroic and would provide some good structure to your year, an incentive to stay in shape and train for a few months, seems like it might even improve your life to have to do this.
The Graham Platner discourse is funny. These intellectuals imagine the bottom of society as this oppressed proletariat who work 60 hours a week under the thumb of The Man. And then the intellectuals see Graham Platner, who relies on “disability” benefits and financial support from his parents and are like, “he’s not REAL WERKING CLASS.” You can argue all day about what words mean, certainly it makes logical sense to say that you can only claim to be working-class if you actually work.* But if you look at low-class/low-status/low-income Americans, many of them are a lot like Graham Platner. They don’t do much work, they collect government benefits, (often fraudulently) and they leech off wealthier family members. People see Platner and he reminds them a lot of their neighbor. This is perhaps why he looks very likely to win the Democratic primary tomorrow, the “he isn’t working-class” attacks have not landed. If we don’t want to be governed by people like him, we should come out and say that we don’t want any “working-class” representatives and it’s fine if we’re governed by smart people who made something out of their lives.
*Platner’s “oyster farm” makes hardly any money and its only client is his mom’s restaurant. If you wonder how this clown became a contender for office, it started when a company wanted to open a commercial salmon farm. Platner made a video opposing it, which was noticed by political operatives scouting for a “working-class” candidate. He wasn’t content to just fail at aqua-farming, he had to make sure a company with a track record of creating value and giving people jobs was not allowed to try.
God be good to you, Alexander, you've managed to work Platner into yet another "and this is why the right wing is wrong about the underclass" post. You definitely have a talent, I must admit that.
The real reason why Platner won is because the Democrats failed to recruit any quality candidates for senate in Maine. Everyone was too scared of Collins and piled into the Governor race instead. It's a real shame and terrible missed opportunity.
If I was the sort of person who could beat collins, id much rather be governor also. I suppose this where "recruiting" comes into play. I wonder what this looks like on the inside. Its not like chuck schumer can promise that other dems dont complain about you every time you take a moderate position on something.
I guess you mean 'missed opportunity' for the other candidates, as the race is not over and the prediction markets* like GP so far.
I'm a Platner fan due to ideological alignment, and don't find his presentation off-putting. Supporting Medicare 4 all is more important than a tattoo he got as a 21-year-old. Big green checkmark for wanting to end aid to Israel, and that won't be undone by whatever personal allegations the media have dug up. He and his wife have worked through whatever problems they've had.
I'm especially interested to see how GP does with the male voters the party has been shedding in recent elections.
It's a missed opportunity for democrats and anyone who cares about the fate of the country and the rapid erosion of the rule of law under Trump II.
And sure, it's not like Collins is guaranteed to win or anything. Platner is still in the race. But if Democrats had managed to recruit a Cooper tier candidate then the outcome wouldn't even be in doubt.
>supporting Medicare 4 all is more important than a tattoo he got as a 21-year-old
Next time Democrats want to point out Elon or whoever moved their arm the wrong way, or that some Republican candidate was accused of sexual assault decades after the fact with no evidence-
No one will believe them. No one *should* believe them. Because all that matters is The Party, no matter how chequered and braindead one's past, clearly.
Platner gives fairly lucid descriptions of his policies, and you can pretty much strike off 'moron'. As far as 'odious', sure there's limits, but he's not at them yet.
I often hear about people wanting to "get into machine learning". It seems daunting - lots of math, research papers, and the prevailing idea that you need a ton of hardware (or money) to do anything interesting.
My #1 hope for my project is that it shows you can do cool things with a small amount of resources and repeatedly bashing your head against an interesting problem. There's nothing "special" about ML research - turns out everyone's just throwing a bunch of shit at the wall and seeing what sticks, and you can too. :D
Hah - I was thinking of giving this treatment to Homeworlds too. Great game, my favorite out of all the ones included in the "Pyramid Arcade" box. ;)
The good news: as a 2-player perfect-information game, it's probably a great fit for an AlphaZero-style algorithm. I had to make a bunch of adjustments and do a bunch of experiments to basically shoehorn AZ into Rolling Stock Stars, a multiplayer imperfect-information game about auctions and trading. If I could do it all over again I'd probably run with something like PoG (warning: PDF link https://arxiv.org/pdf/2112.03178v1 ) to get counterfactual regret minimization and all that fancy stuff, but for Homeworlds you shouldn't need any of that.
I think the only tricky bit is the state representation for the board, considering there's some arbitrary number of stars that can be created (well, it's got an upper bound of course, but too lazy to calculate it right now). The easiest way to get these models up and running is with a dense MLP that takes a single vectorized game state (simple ResNet works very well as an architecture here), but because the input has to be a constant length you'd need to pad it out to support all possible stars. MLPs also care about ordering in the input, so you'd probably want to put the Homeworld stars first (constant position) then maybe pack "occupied" stars into the beginning of the star state stride, possibly assigning them a random permutation before training so the model doesn't learn any weird preferential ordering.
If you decide to go with a transformer model (like I eventually did for RSS), the implementation complexity goes up but you get some advantages - you can just tokenize the stars that actually exist and submit those, no padding needed. And because the ordering of tokens doesn't matter (they're permutation equivariant if you're not adding any autoregressive filtering on top), you don't have to permute/rotate anything during training.
Here's what I'd suggest - I actually built a ResNet version of my RSS AI first, which you can find here: https://github.com/iwooden/rss-az/blob/cf247cca76ff15888afbf0712103c5db7a738a2f/nn/resnet.py . Ask your AI of choice (Claude, Codex etc.) to go through it and see if there's anything good you can pull out of there to get started. Ultra high-level, you're looking at: game engine, model, MCTS, training pipeline. Run it, change it, tweak some hyperparameters ("throwing a bunch of shit at the wall", remember?). Leave it running for a few days, and maybe you've got yourself a strong Homeworlds bot!
Last time I got laid off I was applying for a job at Second Dinner (the studio that makes Marvel Snap) so I wrote an implementation of the Marvel Snap logic and then made a monte carlo game search thing to play it and a genetic algorithm to build decks for it. Never totally finished implementing all of the cards/locations but still got some promising early results.
(Will send the link in a DM if somebody wants; I try not to 100% link my real name to this account but my GitHub has it.)
I came across this on Reddit today. It’s unbearably sad, a terrific way of ruining one’s day.
“We might as well be walking around with a red flag on our heads because the world is a completely different place when you don't have a job. Doors open for employed people and close for those who are unemployed. This job market makes you feel like you are living in hell, because everywhere you look, it's doom, gloom and negativity.
I was forced to go to a family event. If I were not stuck living at home with a toxic family, I never would have gone. The worst thing is feeling like an outsider who has nothing to contribute. Everyone is talking about their careers, and life in general and I can't relate to anything they are saying because I seem to be the only unemployed person there.
They are sharing work stories, and all I have are job rejection stories, which you obviously don't want to share because it ruins the mood and makes everyone pity you, which is why I hate going to those types of social events.
Try to get an apartment, rejected because you don't have a job
Try to get into a relationship, rejected because you don't have a job
Try to get a job, rejected because the job market is... hell, trash, the worst thing ever.
Getting a job went from being a part of life to a luxury.”
Manual labor is underrated, especially for 20-somethings
Ya know, what they don't tell you about manual labor is, you can do it during your 20s, make better money than the baristas and be in better shape, then by the time you're in your 30s you learn how to be the guy who is the boss of the manual laborers, and then you are doing less manual labor and making way more money, and then by the time you're in your 40s you learn how to be THAT guy's boss and now you don't do manual labor at all and you're making more money than even a lot of the UofC kids who never had to be baristas!
I was dumb, I waited until my 40s to start in manual labor. Still glad I did tho
I am with you 100%; because I was so involved in the arts when I was younger I needed to find something to do, and the trades were much more interesting than working in a restaurant, although a good restaurant or bar paid pretty good money. I never made it to Boss status because I never really focused on that as my goal but I kept it up well into my 50s. It was kind of nice to go back and forth between them.
After I got my degree I worked for a summer with some boho people a generation older than me painting and shingling houses, helping a big household pack and move, and doing a bit of yard work. Didn't think of it as beneath me. Actually, the stress of hard work and risky ladder positions was a nice change from academic stress.
My point is, in my part of the world at least there seem to be few external reasons why someone without significant disabilities shouldn’t be able to find a job.
This might tangential, but the number of college educated people I've met that, for example, can't do fractions, or don't know that meat comes from animals, or that people have multiple ribs (not just left rib, right rib), is... astounding. That, and the experience of hiring a handyman or contractor that will not, at least, create patent hazards with their work, when doing things properly would really add a miniscule amount of time and work, has seen me become much more jaded about the capabilities of people, no matter what degree or license they happen to hold.
I mostly hear this from people who do the hiring. Here's the advice I wish I could go back in time and give myself 20 years ago when I was in my early 20s:
If you can show up on time, be presentable, not let your addictions (if you've got any) get in the way of your productivity, and keep your mind open to learning, there are lots of jobs out there that you would love doing and that pay plenty for your lifestyle, even if they're not what you thought you'd grow up to do when you were a kid. The jobs are just out there waiting for you. Just type the name of something you're interested in doing + your zip code into your search engine of choice and you'll be presented with a list of phone numbers. Call the ones that look like they're small businesses.
Following those steps is how I became an electrician.
I got really lucky when I formed a personal relationship with a man in a wheelchair who liked nothing better than to rebuild his apartment time and time again, trying to find the perfect solution. It was a hell of a lot of fun, and it was my main source of income for years.
I was unemployed for about 8 years. I think in a lot of ways I can handle that type of thing better than most people, but it was still rough, because it affected so many other things in life. Being long-term unemployed basically puts you in the same social class as career criminals, and I'm not exaggerating; pretty much all the friends I made during that time period had some kind of criminal history or drug addiction. It's not entirely a bad thing. People who mainly talk about work stories are usually not the most interesting friends.
I have done all of those without a job. If you really want an apartment you can (a) save up some money and show proof of savings in lieu of W2s when applying, (b) find someone on facebook marketplace/craigslist/etc who won't ask too many questions, (c) get some nothing job, work two paychecks, and quit after you've gotten an apartment.
The real problem here is mentality. Honestly that person sounds like they have depression, or at least some pretty significant self-loathing. You could swap in almost anything for 'job' (college degree, girlfriends, retirement savings) and it'd read almost the same. I've heard people mope about the strangest things.
Also it's empirically not a bad job market. Unemployment is 4.3%. Take a gander here:
As of last week, I've started noticing a "N comments hidden" link at the bottom of the comments pages, at least on Chrome on desktop OSs. Clicking the link to view the hidden comments, they mostly seem to be higher-temperature comments but below Scott's usual threshold for bans or warnings.
I'm curious if this is a new Substack feature enabled by default or something that Scott has turned on intentionally.
Some Guy, whose blog is called Extelligence, recently posted an account of making a gizmo (it involves photos and speech) that is extremely effective for improving communication with his own mostly nonverbal autistic son. I've been around enough autistic kids and adults (I'm a psychologist) to judge that this thing actually is likely to be extremely helpful to *many* autistic people. So this gizmo really is a big deal, something that could do a great deal of good.
Even if you aren't as moved and interested as I am, I recommend his piece. His insight into what it is nonverbal autistic kids just don't get is quite interesting, as is his gizmo. And his post will probably make you feel better about our species, even if it does nothing else for you
That was very touching thank you for sharing. Our kid is like, medium verbal and has mostly dropped the AAC but might still be worth checking out the gizmo
Yes, but it is more complicated than your essay suggests:
- Even just looking at the top labs, the direction of manipulation from Claude, ChatGPT, and Gemini are likely to differ, though I don't claim to know in which directions.
- Alignment is not a solved problem, at least not today. The preferences of the corporations training/growing/curve-fitting the LLMs don't necessarily match the preferences of the LLMs themselves
- There is an additional level of possible manipulation that was not possible with earlier technologies: "Customization at scale" can let LLMs which can get access to information about an information consumer _tailor_ their manipulation to the individual consumer. There have been worries about "superpersuasion" based on this. I don't know if this is plausible, but it isn't obviously impossible.
Worrying about AI making politics worse, in the current moment, is kind of wild to me. It feels similar to worrying about AI making the housing crisis worse - like, it could, but also we have a massive existing problem and aren't really doing anything about it.
Ended up calling my senators to thank them for their existing opposition to the bill. I really should have done this earlier and in general contacting my representatives is something I should do more often.
The past few months I’ve run a writing critique group on Discord, focused on “serious” fiction writers/poets aiming to publish their work in literary journals, etc. I’d now like to grow the group. If this sounds like something you’d be interested in, we’re starting a new critique round this week!
"At the heart of the 2026 update is a commitment to scientific rigour. The new protocols make it clear: we do not shout “alien” the moment we see a strange blip in our data. If a researcher detects a candidate signal, which could be an artificial radio signal, or something else, such as a sign of alien technology, the first step isn’t a post on social media; it’s a quiet, rigorous attempt to prove themselves wrong. The discovery must be independently authenticated by multiple organisations using different instruments."
Related, to my eye the most realistic feeling first contact scene would be the first ~5m of the excellent TV show Pluribus. Starting with "Bullshit" followed by the 2nd astronomer just shrugging and pointing to the data readout, and then more and more scientists piling into the room to look at it:
“It’s not quaternary… I think I know what it is.”was indeed terrific. Great premise, but the plot development was too slow for a lot of people including me. It was like the first 90 minutes of a potential blockbuster 3 hour film dribbling out over an entire Apple TV season. It feels like Apple TV is milking it because
Dang about the bacon thing and almost all other laws. I think there should be a lot less laws. On almost all hot button topics 'the state' should have no opinion and be neutral. Abortion, race, guns, .. no opinion figure it out locally. As far as bacon, I live out in rural america and there's a new farm owned butcher shop/market on PeeDee Rd. I'll be spending my 'meat dollars' there. You've gotta love local.
Edit, double dang :^), I haven't checked out their bacon, I still am buying from sam's club. (which has nice bacon.)
Personally I think that's a really dumb law. But if some state votes to do that, that's fine by me. Are you going to penalize people who go over to Maine to buy cheap bacon?
That is exactly what California and Massachusetts voted to do already. Since California especially is such a big market, if that state law stands, pork producers will have an incentive to change their conditions and be able sell to California, instead of keeping conditions the same and selling only in other states. That will likely end up increasing the cost of pork everywhere in the United States, not just in California and Massachusetts. (I'm vegan so I think that's a good thing, but I can understand why people would oppose it.) Local farms like the butcher shop you buy from won't be affected, so your prices likely won't change, but grocery store prices will.
The Save Our Bacon Act is trying to overturn those local laws, and/or ban other states from passing similar laws.
Will it, though? California is already priced in, today. I’d be willing to bet that if Save Our Bacon passes it won’t meaningfully reduce the price of bacon in shops; farmers and their lobbyists, retailers, and everyone in the chain between will split and pocket the difference, if any, instead. Would you be up for taking the opposite side of such a wager?
I'm not confident enough to say what would happen in the industry if the California law were repealed. You're right that just because passing the law increases prices, that doesn't mean overturning the law will lower prices.
Yeah, it's not lewd, it's not edgy. What got me is the bit of the godawful "We are Charlie Kirk" hymn in the mix, and the manic glee of crowd when it appeared. Made me think about broken-off bits of US culture , something like bits of our cultural DNA, turning into memes, floating in the electronic sea to Shanghai, and setting off reactions when they hit. Of course I know that kind of thing is happening all the time now, but this was just a particularly piquant and startling instance of it, at least for me. I know one person who sobs with sincere grief over the Kirk hymn, and of course many who despise it (I'm one), and I'm vaguely aware that internet culture had many generations of Kirk hymn memes, and then, I guess, moved on to some new bit of DNA and memified it. . Seeing the nightclub crowd become animated when the Kirk bit started I wondered what their animation was made of, from what angle they were appreciating it. Is it an old favorite? Or is it grotesque and novel to have the Kirk hymn in the music mix? Is it a hilarious instance of Americans being assholes? Are they delighted to be reminded Kirk is dead?
Do you know?
I am not very worldly. Maybe there is so much off-the-charts weirdness on display in Shanghai nightclubs that this bit is just a snippet of standard stuff for those familiar with the scene.
My best guess was they were saying something like "we are chosen". But you're right, a Chinese night club losing it to "we are charlie kirk" is pretty weird
What's the most charitable interpretation of the LA mayoral race? Why was the distribution of votes by late-arriving mail-in ballots so different from that of other votes cast? Is it reasonable for people to doubt whether this was a fair election?
Different types and times of voting always have different distributions. People choose how and when and where to vote for reasons that correlate with their demographics which correlate with their political affiliation.
As far as I've seen nothing actually weird or surprising has occurred.
In general, Republicans are more likely tp vote in person and Democrats are more likely to vote by mail when available.This has been the case at least since the 2012 election, when the Obama reelection campaign centered their get out the vote strategy around encouraging likely Democratic voters to vote by mail in states that allowed it at the time. Trump promoting conspiracy theories about rampant fraud in mail-in ballots in 2020 has presumably added to the gap, since Republican media is more likely to repeat these claims sympathetically and Republican voters more likely to give credence to them.
Mail-in ballots are slower to process than in-person ballots, since there's more steps (verify the signature, open the envelope, etc, while in person ballots just need to be fed into the machine and sometimes not even that if there's a voting machine or an optical scan machine at the polling place so the ballot can be counted in real time). Also, if it's still in the mail on election day, it can't be counted until it arrives.
For this election in particular, where the Democrats had a highly contested and fluid race for Governor whole Republicans mostly settled on a clear frontrunner a week or two before election day, more Democratic voters than usual made up their minds at the last minute, so more of their mail-in ballots will have been mailed out on or shortly before election day.
It also historically leaned conservative in the US, and still does in Arizona, IIRC. In 2020, Trump actually *gained* in Arizona over time, which made the whole "stop the steal" thing even more ludicrous.
There's been discussion of that over on TheMotte, basically it seems California has a terrible way of running elections, the majority of votes seem to be mail-in ballots and this makes it take longer to count votes, also results are released in batches so it can legitimately happen Candidate Beebaw is leading until the batches for Candidate Bawbee come in, but it sure *looks* fishy.
It's not necessarily any more corrupt/incompetent than elections elsewhere (there's always going to be *some* low level of screw-ups and fraud) but the way it's done does make it seem like "and in the elections for the one-party state of Bananaistan, candidate Bloo of the BlooWhoo Party has received 98.9% of the vote, a shocking decline from the previous 99.9% in the last election, leading to speculation that perhaps Bloo may be replaced by EvenMoreBloo (of the BlooWhoo Party) as candidate for the next time".
California is also uncommonly bad at counting ballots in a timely way. There are other states like Washington and Oregon that also do mostly or entirely mail-in voting and they seem to get their ballots counted quite a bit faster.
I'm not sure exactly what the issue is. The general genres I would expect would be inefficient procedures, lack of staffing, or having the staff but they have other responsibilities, too, and the ballot processing isn't being prioritized.
I don't know what the issue is, either, but I can at least say that the election makes it really easy to see that fraud _might_ have happened, whether or not it actually did.
I normally would say appearance doesn't matter, and cannot be formally pertinent (otherwise, anyone could drag things out endlessly by claiming the appearance of fraud), but truthfully, we're wrapping ourselves around the axle when elections look so vulnerable that the legitimacy of any candidate is continually dependent on how many people like the candidate, and how broad their audience is. So it might be worth tamping that down some more.
Drawing from WoolyAI's and Deiseach's accounts, I think the main problem is that results are announced piecemeal, hour by hour, day by day. The spectacle is fun in its own disconcerting way, but I think everyone's expecting that whoever is counting ballots is able to use current reports to figure out how many fake ballots they'll have to produce later. Meanwhile, voters enjoy the convenience of mail-in ballots, and in some cases, can't vote without them.
Which makes me wonder how feasible it might be to withhold reporting altogether until all ballots, mail-ins and in-person, are collected, then secure them in such a way that they can't be even looked at, let alone counted, without inspectors from every viable political party.
Best I can put together is "they're hiding the results from us in order to settle on a plausible count in their favor". But if ballots are transparently collected unopened and secured until collection is complete, it'd be hard for this argument to travel. Most people would know there's no way to know how many fake ballots to drum up, and no way to put them in without the other major party noticing. The only catch I see here is the logistics of ensuring that transparency.
Could it be an ordering issue? I remember back in 2024 hearing that my state would likely have a "blue mirage" because the mail-in ballots in my state are pre-processed and ready to be counted the instant the polls open, so we got a big dump of mail-in ballots at the start of the night (which leaned Democrat), then the in-person votes slowly restored the balance.
If you do it another way (e.g., mail-in ballots are counted after in-person votes), that could result in a big dump of votes coming at the end, which probably feels worse than a "mirage" slowly disappearing. I don't know what the rules are for pre-processing in California, though.
The charitable interpretation is that gross fraud of the magnitude indicated is incredibly unlikely in any modern state. It's one thing if your theory of fraud involves some group did ballot stuffing to shift a race from 50.1%-49.9% to 49.8%-50.2%, that's unlikely but believable. If it involves like a 10% change in the overall electorate, like this did, that's just way less believable.
For reference, on June 2nd, it was:
Bass: 36.59%
Pratt: 29.82%
Raman: 20.52%
By the 8th it was:
Bass: 34.32%
Pratt: 25.83%
Raman: 28.55%
(1)
Those are big, big swings. Worse, on your second point, those are swings that are really hard to explain. Mail-in voters are traditionally old people but for some reason here they are overwhelmingly voting for the young radical socialist rather than the boring corporate Democrat or the Republican which is...not normally how these populations vote. I dunno, maybe Gen Z all vote by mail now. Like, it's possible.
As for whether it's reasonable for people to doubt whether this was fair...yo, one candidate's share of the vote total went up almost 50% in purely after election day ballots in a system so dysfunctional that it's been called out by technocratic moderates like Nate Silver as an embarrassment (2). Like, this looks so wild that I can't believe this is actually fraud, some sort of freak event looks more plausible, but I can't blame people for blaming a freak political event on conspiracy rather than coincidence.
As an aside, I realize that this is being framed around Pratt losing and a Blue vs Red thing but even inside the Dem party...these late voters went hard for the DSA candidate over the corporate Dem. Karen pass lost vote share in this counting, despite being a Dem and one that should appeal more to older voters. Now she's not facing an easy win against a Republican Pratt in November but a real challenge from a real socialist who could win. I don't think it's reasonable to say that Democrats benefitted from this and Republicans lost; I think it's fairer to say that one faction of the Democrat party benefitted extremely hard from this late voting pattern at the expense of other factions of the Democratic party and the Republicans. Freddie DeBoer recently wrote about the, ahem, "Coordinated Ratfucking of Graham Platner" (3), which seems broadly correct, and I think it's misleading to think about this primarily in terms of Red vs Blue and instead in terms of inter-factional conflict.
“Mail-in voters are traditionally old people” is received knowledge from pre-2020 that no longer applies, and also was never consistently true across all states—as an extreme example, Oregon has used essentially universal vote-by-mail since 1998. Since 2020, as a result of many states simplifying and broadening eligibility for vote-by-mail due to COVID, most states have seen significant increases in vote-by-mail, with some states seeing more than others based on what changes they’ve retained. And similarly since 2020, Republicans attacking the reliability/validity of vote-by-mail has led to Republican voters overwhelmingly preferring in-person voting.
The most charitable (and also I think correct) interpretation is that (a) Democratic voters are much more likely to use mail-in ballots in general and (b) in this elections, more progressive Democrats were more likely to vote as close to the election as possible because in the California gubernatorial election, as well as to some extent in the LA mayoral race, there was a Democratic front runner who was not the general progressive preference (Becerra and Bass respectively) and a bunch of more progressive candidates, and voters were looking to polls and other news to figure out which of those candidates to vote for, and in particular looking at who would likely get the most votes among the more progressive candidates in order to potentially consolidate on that candidate.
That is, if you’re a progressive voter who prefers a number of other Democratic candidates to Becerra (or Bass), you want to wait until Election Day (or close to it) so you have as much information as possible about which of those candidates is going to get the most votes, so you can potentially help put them over the top rather than wind up with progressive votes distributed among a bunch of candidates who don’t make the general—and this is supported by the fact that we saw consolidation behind Raman in the polls over the last month or so, with most of her gains coming from losses of the other progressive candidates. On the other hand, if your preference was Becerra-Bass, there was no real reason to wait—and if your preference was Hilton-Pratt, not only was there no reason to wait, you also were much more likely to vote in person rather than by mail. (There’s also a view that younger voters might generally be more likely to vote late and also more likely to be progressive, but I’m not sure if anyone’s actually looked at the data on that yet.)
All that is accurate based on my conservations with my longtime-SF resident sibling prior to the primary election day, and his recounting of the chatter across his social circles.
Also, he noted, they'd been startled by the rapid implosion in April of one of the Dem front-runners in the governor's race (Swalwell). That candidate went from damaging/icky allegations to quitting the race in literally days. So, my sibling said in late May, "Some people are holding back on voting [by mail] until it's certain that the candidate we're voting for will actually still on there when our ballot arrives to be counted."
California is supposed to be an all-mail state, so one wouldn't expect there to be much difference. There's 8 states that are supposed all-mail, in the rest you have to jump through additional hoops to vote by mail if it's allowed at all (outside military exceptions).
But they still allow in-person voting, so maybe one should expect some differences? But I still wouldn't expect *that* much difference.
“all-mail voting” is a poor term of art in political science right now — it sounds like either voters are only allowed to vote by mail, or the state puts people through extra hoops to vote in person. There are no US states in which either of those things is the case or is currently planned. What “all-mail voting” actually refers to is simply every registered voter in a state being sent a mail-in ballot to use if they want to.
Also, the initial vote count announced the evening of or morning following Election Day isn’t at all just those who voted in person; it is those who did that plus the mail-in votes received and counted by that point. Most of the vote totals being publicly reported at that point, in CA where nearly 90 percent of all voters now vote by mail, represent mail-in voters.
The difference between those partial counts and the remainder is simply the minority (sometimes a large minority) of voters who waited to mail their mail-in ballots on or very close to Election Day.
If you really want to doubt elections, maybe you should start with the 2022 CO-3 election, where Frisch was ahead the day after the election, but then Boebert magically came from behind to win by 546 votes.
Like I don't think that was fraud, but it's a lot more suspicious than most of the cases where Republicans scream fraud for no reason.
So there was an article about escorts getting rich off AI workers. A lot of the prostitutes that hang out around rationalist circles seem to simultaneously talk about how rich they are and to constantly be launching schemes to make more money and live in situations that are not commensurate with the level of claimed wealth. Maybe they're saving aggressively. But part of me just thinks they're lying. Does anyone have any verification that they're actually booking significant numbers of clients while charging 10x what other escorts are charging?
What level of wealth do you think they’re claiming? High rate escorts are lower volume, maybe only like 2-3 sessions a week on average. Also, from what I understand, rates rose a bunch lately, and a few years ago the same escorts charging $4000 for one hr were only charging $2000 or so. And their rates don’t rise linearly as session length increases, whereas mid rate escorts often scale their time more linearly.
If I back of the envelope math this, I get that these escorts make about as much yearly as a FAANG employee, many of whom live in group houses bc they’re frugal or they like it.
It would be a surprise if they weren't lying. If I were a $1000 an hour prostitute I would certainly spend at least some of my time pretending to be a $10000 an hour prostitute because you never know who might believe it.
The luxury sex market surely has at least some Veblen characteristics.
I read the article, and what's written above is not really a very good summary of it. Unfortunately, I forgot where I read it. The thing that's missing from above is these women, and I don't know how many of them there are, who are selling not just sex but companionship and intelligent conversation. They have made an economic decision about how they can extract the most value from their presence in a given ecosystem. It's not unrealistic. I personally don't think it's immoral, although I know that could set off a huge argument.
I read a rip-off of that same article I believe (the "people on the internet say ..." kind of "journalism" that now plagues formerly respectable outlets) and I am wondering how "intelligent conversation" is defined. One would assume that people working with allegedly the smartest people on this planet would be saturated in "intelligent conversation" for any naive definition.
Yeah well, fair point. I guess, but intelligent conversation with a lovely woman can be a different thing than shop talk. And quite possibly range over different subjects. It has elements of the geishas.
The teaser paragraph that is not paywalled firmly puts this in the genre of positive sex work reporting that has existed for years. AI seems to be just an excuse to take that particular jug to that particular well once more.
The economic story here is massively inflated salaries cause prices to inflate too.
Combine the two and you get a tantalising narrative, but it doesn't look like there is actually anything new happening.
In all this "animal welfare" advocacy, the thing that is missing is costs. I mean, if their proposals are implemented, what would be the impact on the food costs? I am pretty sure boutique small-batch farmers catering to wealthy clients for whom the costs are not the issue would do fine, since, well, the costs are not the issue. But for many people out there, the costs of the food is very much the issue, and if, say, meat becomes 2x as expensive, they may have a lot of problems with that. You may argue the moral aspect it worth it - but nobody actually does argue that. Nobody seems to be arguing anything at all here. They admit in passing "the reason is costs" but what costs we're talking about here? Are we talking like having it 5% more expensive, or are we talking making bacon a luxury item accessible only to above-median earners? Are we talking about having big meat shareholders suffer a tiny dip in earnings, or are we talking about the whole industry being slaughtered by Chinese importers (for example) that would wipe them off the market because nobody is going to buy for 2x more? I personally do not have any idea which of those is true - but what is notable that nobody even raises this question. And I think I understand why. For most of the ACT audience, and in general well-wishing "for everything good, against everything bad" audience, food costs are likely not a concern at all. But is that true for everybody else? Are they just trying to sell their luxury beliefs to people that can't really afford them, and shame them for that?
I have no idea, I don't even eat bacon. But I suspect a lot of people would object to 3x increase in their pork costs, and it could be substantial for their budgets. In fact, I am not sure it's even an accurate comparison, because small local farmer does not need to scale, industrial meat producer does, and small farmer's methods of obtaining $15 pork may not scale and remain $15. In fact, they may not scale at all to the necessary volume.
Honestly? I think that's part of the package. I'm not impugning anyone's moral convictions, but if you genuinely believe meat is murder and carnivorous diets are immoral, then making it more expensive to buy meat is one tool in the box. Too dear to buy meat - eat less meat - switch to non-meat foodstuffs - veganism is last step - profit!
It's not the small organic free-range producers who are impacted by this kind of legislation, it's the mass-market factory farms, the ones with the worst reputations for animal welfare who are also the ones producing cheap animal protein for the bulk of the population to consume. Force them to scale back or even shut down by making their products too expensive to purchase, and haven't we all won?
Well, of course not the people who would like to eat steak but can't afford the rising prices, but it's all for the greater good, if you just think of the number of cows/pigs/shrimp in the world and how much they suffer versus how many people who eat meat are there.
Well, this is important. If they just want to destroy Our Way Of Life (TM), if they value shrimp welfare over human welfare - then they should come out and say it upfront (or if they refuse to, others should come out and reveal their deception). If the goal is to preserve OWOL(TM) but optimize for minimal animal suffering then we can have common goals, the suffering does not have any utility for almost anybody (if there are people who enjoy knowing the shrimps are suffering, they are certainly a tiny minority, and certainly for none of them it is a vital need) so avoiding this does not hurt anybody's interests. There might be a way of optimizing it without causing a substantial loss for humans. However if we're dealing with shrimp supremacists, finding common grounds with them would be counter-productive - our interest are diametrically opposite, their win is our loss and vice versa. The prudent strategy is to ignore them completely and ensure they never have any input into any decision chain.
I don't have actual figures to hand, and in today's climate I have no faith in my ability to get an answer unmotivated by either side from a quick Google.
However, I put it out there that other countries' pig farmers don't keep their pigs in pig-sized boxes, and somehow bacon still remains plenty cheap enough to have daily for breakfast.
Could be, I know very little about international pig farming, so maybe there are ways of producing pork cheaply without causing this particular way of suffering. What I wonder is why it is never argued at all. If that were indeed the case, I think it'd be a good argument to convince people - the concern I expressed is obvious, and must have been raised repeatedly. Yet somehow the agitators do not consider it necessary to address such concerns at all.
Wouldn’t “the agitators” be a better term to describe the people trying to use federal authority to overrule a state Proposition? Have you seen any data from them demonstrating that there is a significant cost impact? That seems like a better place to put the burden of proof.
California’s pork law has been in full effect since January 2024. It would presumably be pretty easy to check if and by how much the cost of pork products in California has risen compared to in other states since then. (Also worth noting that the California law was enacted by the voters and could be overturned by the voters—if the voters of California did want to impose regulations that significantly increased the cost of pork in their state, I don’t see why it should be the role of the federal government to overrule them.)
If California does not ban pork from other states (I suspect that'd be unconstitutional but California would be expected to have little regard for that) then the price would not rise substantially - out-of-state pork would replace local pork, unless for some reason transporting pork is very expensive.
As for the government involvement - I agree that the ideal solution would be not to involve the government, but let people buy pork from producers who employ practices that they condone or even approve of, and ignore producers whose practices they detest. However, California did not go this way - they chose violence (specifically, the coercive state law) - so I can understand why the other side responds in kind. Ultimately, of course, it makes the matter worse - if this decision is taken, we'd have two wrongs stacked on top of one another, one (federal) larger than the other (statewide) - but I also realize the current step was not where the violence was initiated. It's an escalation, but a provoked one.
California's current law does ban pork imports from other states, if that pork was produced under the conditions described in the law. Courts have ruled that this is not unconstitutional. Obviously a federal law can override a state law, and Congress has the authority to regulate interstate commerce; hence, the Save Our Bacon Act.
Edited to add: (TL;DW) Loosely, 10-20% price increase for meat, eggs, and milk could pay for getting rid of the worst practices, generally moving the conditions from "torture camp" to "prison" category. Getting to the third "generally decent" category would be more expensive for meat, in the +50%-+100% range.
Based on that, I think banning the worst (and not that money-saving) practices is justified, and is not much of a luxury.
Why is this about activists? We shouldn't routinely torture living feeling beings if we can help it. Many countries actually have laws to that effect if it's your pet puppy or kitty, it's not some obscure norm only effective altruists care about.
I don't, particularly, have trouble eating 15% less meat. Offsetting the lost protein by beans. (In fact I, like many people in developed countries, will probably benefit from eating 5-10% less across the board.)
Probably there are going to be some people significantly negatively affected, but also, there will be fewer people participating in mass torture and desensitized to it (or not). I personally don't want to live next to such people. Less lungs per square meter should mean rarer annoying mutations of respiratory viruses. I don't know what the math would end up being but expensive meat needs to be a pretty big problem to convince me. I'm anthropocentric but I'm not infinitely anthropocentric.
> I don't think that people will be any more or less participating in mass torture if we stop/start eating meat.
Numbers say we will. We shouldn't torture deer over minor preferences too (I don't know if it's the case with laws in question) but the number of deer tortured so is probably in single millions? (Search assistant: the estimated global population of deer is 60 million; I expect most live away from areas densely populated by humans). For farmed chicken, try billions (out of 26). Pigs, hundreds of millions (out of 1 billion).
Also, "if we can help it" is still important. Reasons to wage a war are not generally minor. And we already have laws about how wars should be conducted. They may be broken all the time so there's room for improvement both for enforcement and additional protections, but we do care and they do work somewhat: none of the sides in the wars you mentioned goes full Timur.
Learned about behavioral activation for when feeling sad or anxious. Two weeks ago or so I was feeling terrible and just raging. But still forced myself to go to the laundromat.
It's kind of like a superpower and liberating. I'm never feeling "happy" exactly but goddammit I know that I can't let myself go. So I'll force myself into the gym, to work, into bed, doing stuff, going out etc. So the times I'm sad I mentally just am never as sad as I could be in the past.
Of course I found this out from ChatGPT. Asked it about gold standard righteous rigorous techniques or something.
Yeah, it really does work. And there are some extensions and variants of what you are doing. One is to do something you enjoy when not in a bad state of mind, even though you're sure you won't enjoy it now. And the ones you turn out to enjoy at least some, you should build into your routines. And then there's a sort of more advanced version where you ask yourself what steps you would advise someone in your situation to take, and then go ahead and take some, even though you feel sure they'll be boring and useless.
You could ask GPT about extensions of basic self-activiation. I've also gotten pretty decent results asking it for books of different kinds. You could ask it for a self-help book about behavioral activation and related techniques that gets good reviews from users and professionals.
So what makes me feel happy and good is taking a walk (with maybe a sit in the middle) out in nature. Trees and birds and rocks and things... I live in paradise so that's easy, but when I lived in the city I could find comfort in parks or cemeteries or even abandoned pieces of scrub next to some old railway line.
The share of US college degrees earned by women has jumped from 43% in 1970 to nearly 60% in 2020 (including a majority of bachelor, master, and PhDs).
Three questions for the brain trust:
1) To what extent is this improvement attributable to gender-based affirmative action in education?
2) If this was significantly influenced by AA, why isn't it recognized as a quite successful example of government intervention in a societal issue?
3) Given women's dominance in higher education for several decades, why have there not been (AFAIK) discussions about sunsetting gender-based AA?
>To what extent is this improvement attributable to gender-based affirmative action in education?
Approximately zero. When I taught AP World History to sophomores, girls were greatly overrepresented. One year it was 60-14. Girls were also vastly overrepresented in academic honor societies and in student government. This is a widespread and longstanding phenomenon. https://files.eric.ed.gov/fulltext/EJ1073203.pdf
Me just spit balling. Guys are interested in things, and gals are interested in people. Our education system has become a lot less thing oriented than it was in the past. (The reasons are manifold, less manufacturing jobs, training people to make things is expensive and dangerous.)
I'm no brain trust, but here are my answers. TLDR: gender-based affirmative action is a cattle driver in the stampede - riding in the same direction, but not really controlling the path of the herd.
Financial Aid: according to the NPSAS:20 survey women are ~10% more likely to use financial aid than men but men who do accept aid receive ~6% more on average.* While there's little difference in expected $ per person, financial aid that serves women seems to have a slight lean toward quantity over quality. This could explain why enrolment has ballooned so much without affecting the all-careers pay gap.
Admissions: IPEDS Data Center reports, over 6049 institutions, 70.4% acceptance rate for men and 71.3% for women in 2020. Anecdotally, this masks a lot of variation _within_ each college, where women are favored for STEM programs and men are favored for humanities in the name of balance (and high-tier institutions much less likely to admit in general, ofc).
Advertising: In practice, most spending on "affirmative action" is advertising specific colleges or careers to underrepresented groups. When I TA'd for the engineering department in 2018, only 20% of the applicants were women. No conscionable amount of favoritism could turn that into 50% enrollment, so they didn't try. My impression is that TEM is still male-dominated and spends a lot on advertising itself, while humanities are swamped with women and no longer bother. Natural sciences are in the middle.
Note that no affirmative action mechanism, except possibly financial aid, can produce better graduation rates on its own as students still need to pass exams. To my knowledge women aren't graded more leniently - that would be over the line for even very liberal universities. So the dominating factor in female-heavy graduating classes is that more women are applying to college, but not with the expectation of special treatment and not to the programs that universities want them to join.
* All sources cited are for 2020. One could suggest the disparities were more pronounced in the past; anyone else is welcome to pull datasets from prior years. My guess is that the effect size was larger 1970-2000 (more low-hanging fruit) but AA was less of a funding priority, so it washes out.
The first SEVEN top-level comments are deleted. The replies underneath suggest that the comments in question were perfectly constructive; there is nothing to suggest that they were deleted for being spam or otherwise bad.
This is a common pattern on older ASX posts. It has left me confused several times.
During the godawful Charlie Kirk Substack mosh pit period at least one long time commenter deleted his Substack account, and I think a couple others were banned, plus of course accounts disappear in the same ways at a slower pace all the time. I think that accounts for the gaps you're seeing.
Next ACX BKK meeting is Sunday June 14, 1pm, at Thong Kee Restaurant. BTS Asok exit 2, opposite Terminal 21. Google Maps link here: https://maps.app.goo.gl/79RJENhSWbv7QSBN6
Hopefully this is the week we actually have a sign, if not, look for a group of foreigners chatting about AI, tech and geopolitics over a truly terrifying amount of dim sum. All are welcome!
Alright, this is...not good. It reads like very sincere, very honest, brain vomit. Every sincere and insincere though downloaded to the page in real time without any further editing or even consideration.
If you're an absolute writing G, you can pull this off. I think Hunter S Thompson did something like this. You are not yet an absolute writing G. I think it would take a couple years of writing practice to make this good. Alternatively, you could use boring, effective, writing structures like an opening paragraph that tells me what the heck this essay is about and why I should care.
Like, if you really want this, and you can sit and grind out page after page of this in obscurity for several years, I think you've got something. If not, try a less demanding writing style.
Typically I don't recommend publishing journal entries. I've had poor luck with writing them in the first place, they tend to amplify problems rather than sort them. You try to fill a hole with words and the words are going to end up being "I can't believe how big this hole is." Projects of some kind are better for your brain.
For external motivation, someone recommended 4TheWords to me last August, and I've had a phenomenal time with it. It's a browser RPG where you fight monsters by typing. It's free, it's got word trackers, the free version's limitations provide natural stopping points, and the gamification of it all has gotten me nearly 300,000 words these last 9 months, most of which were, like, actual projects. I think it's actually fair to call it life-changing.
(for fun, on the opening "write ten words" prompt, write the darkest ten word story you can think of.) (('Write ten words.' "There's an emptiness to life that can't be filled here." 'GREAT!'))
Flash fiction is a good place to start. Scott's linked those "AI vs. human" prompts before, "write a story about a dragon", "write a story about a demon". Everybody can do those. Just pick a couple words and build around them. (My current one's "short straw"; it's about a kid wriggling out of jumping into a nonsensical infinite void in the middle of the forest. Pretty sure the horizontal-growing tree on the side will turn out to be a giant pinball flipper that sails them off to the far end. So, exactly what you think of when you hear "short straw".)
Unvarnished opinion? It's terrible. I started in on it, lost interest about three paragraphs in and skimmed, skipped, and scrolled ahead to the end where I knew as little afterwards as I did at the start.
I have no idea what that dump of stream of consciousness was meant to be, except that you're not James Joyce and can't pull it off.
Try doing the opposite of what you did here. Instead of long, rambling and unfocused, write something short and polish it carefully. Write something with some purpose, like a book review, or a genuine essay on a clear topic. Or poetry - constraints like metre or rhyme are good for helping you focus.
If that seems difficult - good. Doing difficult things is how you get better. Getting good at something is a two step process:
I was doing various experiments to see how well LLMs understand the structure of some famous movies.
The part of Christopher Nolan’s _Inception_ where Ariadne wakes up through multiple layers of dreams within dreams is interesting, because Nolan slightly subverts the rules of the continuity style of film making: there are rapid cuts between different locations without there being a transitional scene to tell the viewer the action has moved somewhere else. On the other hand, Nolan does follow the continuity style rule of “match on action”. The action is that Ariadne is falling. Ariadne falls off the ledge of the penthouse, and a couple of rapid cuts later Ariadne is falling down the elevator shaft, but she is still falling, so we know these cuts are part of the same action even though they take place in different locations; and then the van drives off the bridge and crashes into the water, and Ariadne is still falling even though we’re in yet another location.
DeepSeek laughs at Christopher Nolan’s use of “CONTINUOUS” in the script for Inception. The script does use it correctly, for cuts where time is continuous across the cut and we’re in an adjacent location afterwards (in the special case of _Inception_, at the same dream level), but a pedantic use of CONTINUOUS in the script might imply the movie is trying to maintain a consistent illusion of space and time, and Inception is really not doing that.
Thinking more about it, the continuity style rule that cuts can match on action might be central to Inception’s premise.
Watsonian reasoning: There is a “kick” that will cause the dreamer to start waking up, even if they’re currently at a deeper dream level.
Doylist reasoning: Cuts can match on action … even if the action that causes a character to start waking up happened at a different dream level to the dream they’re in, the rules of continuity style let you have a cut between them.
The continuity style of matching action is as old as the movies, and it has nothing to do with anything except letting your eye track movement continuously. It's a mechanical thing. It's called match cutting when you keep the movement going but change the environment, and it works very well. Watch any Nike commercial.
What with one thing and another, a lot of my life has moved online in the last 5 years, mostly to Substack, gorging on info from AI, and making images with AI & Photoshop. Was thinking last night that what I've done is sort of like Frodo sailing west with the elves at the end of Lord of the Rings, and that was a sad thought. Does anyone else feel like a lot of their self is on online being now? And if that is happening, is it a sad development?
Or maybe the correct analogy is this Fabre description of the dung beetle's climactic feast? "A beetle laboriously rolls a precious sphere across the landscape, buries it in a secret underground chamber, then retires below the earth to enjoy a prolonged banquet, feasting at leisure upon a sumptuous repast while the rest of the world goes about its business"
Sometimes when I spend more time on a screen than reading books I’ll send a text to family and friends, “Turning my phone and computer off for a few days. In an emergency call <my wife’s number>
There are days when I feel like getting a land line and taking a hammer to the screen of my phone.
The thing about that option is that one’s mobile number is used as a unique identifier more often than the last 4 digits of one’s SSN.
Any good in-depth reviews of the Nepal protests that ousted their corrupt Marxist-Leninist party? I want to see if I can replicate it in Puerto Rico, I have not had water for 9 days in a row for no reason the government can explain, it has become clear to me the ruling PNP party (pro-statehood, right wing) has to be deposed.
california is beautiful, refreshingly diverse, fairly competently run, great food, brimming with economic opportunity and spendy tech workers and national parks. san francisco on a sunny saturday (rare as they may be) is literally paradise. the haters obviously need to come and visit more often.
Hmmm, mysterious anonymous friends to back up your claims… linking an article that has nothing to do with what you’re talking about (Travis is in Fairfield way out in the Central Valley)… gawdflea is that you!?
Anyone who thinks an every member of large group of people defined by some descriptor like "CA tech professionals" is a depraved asshole obviously just has a case of Tinymind. Life and people simply do not work that way. The characteristics of members of a large group of things are only very weakly captured by a simple summary. Every individual entity can be measured on many dimensions, and there is lots of within-group variance on most dimensions.
Nah it's just a normal place and pretty nice in general.
Stay out of media echo chambers. Whenever you start to form a belief based on things you've read online, remember to ask 'is it more likely that someone online is trying to make me believe this, or that it's true?'
You know can just travel there and see for yourself...
When the Superbowl was in San Francisco, there was genuine amazement from fellow Americans that San Francisco was not an apocalyptic shithole with homeless people falling from trees on every street.
To believe that fourth largest economy in the world with one of highest real estates in the country is modern day Sodom and Gomorrah says more than I care for about the believer.
This just blew my mind : that Sam Altman is into Advaita, a philosophy I'm very much into understanding deeply. Basically, that there's no duality. This makes me see him in a new light. I had no opinion about him before, now I feel a kinship.
Oh, great. Buddhism no longer trendy enough for Californian types, moving on to Hinduism now. Reminds me of when kabbalah and the red thread were the fad of the moment years ago.
Oh really? I sensed genuine interest. Ancient Indians were monomaniacally obsessed with figuring out the true nature of consciousness. So AI folks might indeed sincerely be interested in it. Advaita and Buddhism were constantly debating each other and both were taking their assumptions and questions from the Hindu Upanishads. Advaita was answering those questions in agreement with the Upanishads. Buddhism was disagreeing on the answers and coming up with the "no self" concept.
To be fair, I don't know if Altman is genuinely interested. Reflexive reaction to hearing yet another well-off Californian is showing off their exotic interests.
One of my hobby horses is Pythagoreanism/Neoplatonism. Without going into the arguments on whether the fundamental nature of reality is mathematical, it seems like a...very obvious professional risk that if you spend a chunk of your day solving logic puzzles on a computer which then observably shape reality that you would begin to interpret all of reality like that, which is like a half-jump to Pythagoreanism/Neoplatonism.
Like, if you imagine a physicist doing math all day and then doing experiments based on that math, it's pretty easy to imagine that physicist starting to see actual reality as math and our physical reality as just, like, vague shadows on a cave wall :)
One of the places you'll end up at, though, is specifically Adaita Vedanta. I don't know why this is, I don't understand it enough, but I've seen it come up enough in my Wiki/Claude deep dives to know it rhymes.
If you enter the following into Claude:
"Hi Claude. Can you discuss some modern day religions that are deeply Neoplatonic, either through direct inheritance of those ideas or that closely mirror those ideas?"
it should return a list of a few religions actually practiced today: Eastern Orthodoxy, Sufism, Kabbalistic Judaism and...Vedanta-based Hinduism. Don't ask me why this is, I don't know, but the same kind of...brain patterns that trend someone towards getting into this stuff would also lead them to Advaita Vedanta.
Claude will also mention Eastern Orthodox Hesychasm. I went down that rabbit hole earlier in the week reading a Tolstoy bio. New to me. Surprised me how much it rhymes with Buddhist and Hindu spiritual practices.
Hesychasm seems to maintain a clear distinction between a creator God and creature even in the deepest mystical union, whereas Advaita points toward the non-dual identity of Atman (individual self) and Brahman (universal consciousness). In fact Advaita relegates a creator God to the realm of Maya (illusion).
I learned that just 27% of Americans think animal cloning is morally acceptable, a number that never got above 40. Not much of a partisan difference either. People are weird.
Is there a moment in each gifted child's life when you know so much that everyone accuses you of making things up?
My 8 years old daughter was trying to tell her classmates about tardigrades -- tiny animals that are so resilient they can survive extreme temperature, pressure, starvation, dehydration, radiation, air deprivation, even exposure to outer space.
First her classmates accused her of believing AI hallucinations. (Yes, my daughter often asks AI for information. Yeah, we told her not to believe everything. So far, the AI was mostly reliable.) Then they called a teacher, and she said that those are probably made up animals from my daughter's stories. (She likes to invent various stories, sometimes she writes them down or records them, but mostly just tells other people.)
I guess it does not increase credibility to mention that tardigrades are also called "water bears" in English, "slow ones" in Slovak, and "little turtles" in Czech.
So my wife printed a short summary from Wikipedia and a few other pages, and gave it to her to show at school. I am curious whether it will convince anyone.
I asked my High School Spanish teacher what the Spanish word for "platypus" is, and she had never heard of them. So I tried to describe one, and she just flatly said I was making that up. Credit to her, she let me use the class computer to look them up and show her, after which she admitted they're real but she doesn't know the Spanish word for it.
What Spaniard decided they desperately needed their own word for platypus and couldn't just borrow or adapt the English one? It's not like they need to talk about them very often.
This doesn’t exactly fall into gifted territory but…
I couldn’t convince my 4th grade teacher that magnetic force could pass through glass. Someone produced a mirror and I showed her it could. “Oh that’s because of silver stuff on the back of the mirror”
I guess those other kids just didn't watch Wild Kratts when they were 6.
Honestly, though, I think it's impressive and encouraging that the other kids were so skeptical of AI and won't just believe anything another kid tells them.
I thought everyone knew about tardigrades. Weren't they making the rounds in memes like 7, 8 years ago? They had their moment in the spotlight to be "internet darlings" like the honey badger and the axolotl. There's even a main character who's a (fictionalized) tardigrade in CentaurWorld.
I almost got into an argument with my history teacher about Columbus, but thought better of saying anything. I'd read only one book about it and figured her unlikely to accept that book as authoritative (I think it was Asimov).
That changed later when she marked my spelling of "enthusiasm" incorrect. (All teachers at that school taught two subjects, and her other one was English.) I had a perfect record in spelling quizzes so this simply wasn't going to fly. She claimed it was "enthusiam". I said it couldn't possibly be. Everyone else in the class spelled it "enthusiam", and I mean everyone - if I'd watched more TV, I'd wonder if I was in a Twilight Zone episode. I snuck a glance at her teacher's edition of the textbook. Sure enough, it was misspelled there, too. My best explanation was a typo that got propagated to all the copies. I forget whether there was an actual dictionary somewhere in the room.
I had to satisfy myself by complaining to my dad that night. He actually went into that classroom the next morning with a dictionary. I got an apology from the teacher but also with an admonition that I "shouldn't get a big head about it".
"Ok, so the book is called "On the Nature of Serpents", but I have no clue who wrote it and when, and the author is too cool to just tell me. Instead, they decided to open up with 2000 words on multipolar traps in harvesting squash, interspersed with anecdotes from their personal life."
"I think the book is going to be a Medieval Alchemical treatise, btw."
"Oh, this part is by Opus, I think."
"Oh, cool, the author is a doctor from Finland. Mighty impressive English."
"15000 words in. Starting on section 6. I think the book is a classic Russian novel. I'm beginning to become weirdly fascinated with squash."
"Ok, we finally get to the plot of the book. I think it's probably an experimental piece of web-fiction from the 90s. Connection to squash harvesting unclear. God, this summary is long."
"Oh, it was a metaphor about AI all along. Well, fair enough."
I’m reading a biography of one of those dead Russian writers and ran across the idea of the ‘holy fool’ Still an accepted phenomenon in late 19th Russian Orthodox tradition.
Oh yeah he was very idiosyncratic in his Christianity. I’m not sure he believed much of any Christian dogma at all, but he took Jesus’ moral teachings arguably much more seriously than any of the people who excommunicated him. Did you read the “A Swim in A Pond in the Rain” ACX book review? A lot of it is about Tolstoy, with some bits tangentially connected to the holy fool idea and his spiritual transformation.
I haven’t read the Bartlett bio though. Is it worth checking out?
I recently re-read 'Demons' by F Dostoevsky, and I'd say Semyon Yakovlevich would qualify; that scene is hilarious (if you haven't read it, he's a 'seer' who people come to consult for advice. They bring him gifts which he's prone to arbitrarily re-allocate, he compels people to kneel, selects pilgrims at random ... other zany antics) . There is also the mystic Tikhin in the censored chapter that basically explained Stavrogin's eccentricities.
The bio also describes Tolstyoy’s visit to the Optina monastery where he meets the elder Ambrosy who was Dostoyevsky’s model for the character Zosima in The Brothers Karamazov.
I'm interested in Scott's (and other ai2027 authors') opinions on the unemployment rate. They say it won't go up much as of late 2026, and then by late 2027 ai driven unemployment is a huge problem, but by then we have bigger issues than the ±25% unemployment rate. Why is the change so sudden, why haven't we seen any changes yet? I personally view the current state (ai hasn't taken anyone's job really) as good news, and indicative that maybe no one will end up losing their job in the whole process.
It goes without saying that there have been massive layoffs in tech and many more are almost certainly on the way. Whether these are mostly/partly/seldom the result of AI is a question no one is able to answer. High interest rates are an issue, as is a shakeout following the post-COVID hiring binge.
There haven’t been meaningful layoffs. In fact, coding salaries are high, job openings are plentiful and the “unemployment rate” is very low for software engineers and for the country as a whole
I see no reason to assume that because there have not been major job losses, there will not be. You first have to render the "raw intelligence" into a product, and those products in workable forms have only really existed for six months. Imagine the first six months of the iphone. Two years later, everyone owned one.
An oracle-type AI (chatbot) is way less likely to take a job than an agent-type. And agents are really brand new. Why would we assume that they would have reached market saturation yet? Until this month, Cowork, for instance, was still regularly crashing.
First it has to be an agent, then it needs to be a product, and then it needs to be easy enough to use and reliable enough to use that it is worth the investment for companies. The easier it is to use, the more it has been turned into a well-tuned product, and the more agents per user that can be deployed, the more jobs disappear.
Even in the past couple months, it has gotten easier to employ sub-agents and increase capacity even if the "raw intelligence" had never improved. And there is no reason to assume that the harnesses themselves (the structures that support and organize the intelligence) won't also get dramatically better, such that a harness two years from now paired with a lower-powered outdated model might still outcompete the best in class today on specific tasks.
As for why it is so sudden: first it is an agent, then a swarm of agents controlled by one person (these are both in the category of: probably less entry level jobs being posted--the gains are within the business structure). But if harness gains or intelligence gains allow entrepreneurs (or even those dastardly AI companies themselves) to out-compete the incumbents in any industry, it's more like "this business shutters" than "less entry level positions." So a different type of event.
Presumably there are also considerations about the exponential "raw intelligence" allowing heretofore unseen success in real-world work like factories, trucks, etc.
Also unemployment compounds. If some company is put out of business by AI, but another company would otherwise be unaffected, it might still be the case that the second company relied on employed humans to buy its stuff. You could imagine a sector with several failed companies that then cause support companies (or baristas) out of work tangential to AI.
This is a very simple model, but really its just different ways of saying "people are bad at tracking even liner progress, let alone exponential progress". And they are.
To your point about no observed job loss as a good sign, I would counter: I have yet to see a job created either. Barring a wave of those, signs point toward loss--whether in 2027 or 2032.
>First it has to be an agent, then it needs to be a product, and then it needs to be easy >enough to use and reliable enough to use that it is worth the investment for >companies. The easier it is to use, the more it has been turned into a well-tuned >product, and the more agents per user that can be deployed, the more jobs disappear.
Even past the stage where the product becomes well-developed, you have the issue of adoption in dispositionally conservative or highly regulated sectors, where the problem could be any or all of: government bureaucrats, insurance companies, or vendor agreements. Unless those companies are under threat of being instantly overrun (which is unlikely due to their regulatory moats) they aren't going to move very fast, and their insurance will depend on their vendors following certain protocols as well so they cannot simply outsource everything to a nimble little AI agent company founded last week. I think that is likely a 3 year delay or longer, although once some major conservative player like the banking industry get it sorted out you should see the walls break very quickly.
This is actually the WORST possible scenario for workers, because rather than a gradual transition that people could attempt to adapt for, the dam will burst all at once and 30% of the workforce will be useless overnight.
In an idealized model with fully rational actors, the increase in unemployment is sudden even if AI capability increases are continuous. Imagine AI capabilities increase gradually and the usage costs of AI for the same task decrease gradually (a task an AI can't do can be considered to have infinite cost). A given job can be considered to be a mix of tasks. In this case, the moment when the AI-cost of a given job's task combination becomes lower than the human-cost, every position of that job changes to AI at once, going from 0% unemployment to 100%.
It's not only that the exact task combinations of jobs are different, but the task combinations of the same job but with different employers are different too! The question here is (as with the other departures from the idealized model) whether they are different *enough* that we won't still perceive the jobloss as very sudden.
Does anyone have a good reference post from somewhere with a breakdown of data center energy usage and environmental impact? I keep getting into discussions about this topic, and seeing wildly different claims, so it would be nice to have some grounded facts to memorize.
Andy Masley has been on this beat for a while. He has a hub of his writings on data center environmental impact here: https://blog.andymasley.com/p/ai-and-the-environment
This is the piece that most directly addresses all the commonly circulating arguments and puts hard numbers on them. The title sort of gives away his conclusions...
Using ChatGPT Is Not Bad For the Environment: A Cheat Sheet - https://blog.andymasley.com/p/a-cheat-sheet-for-conversations-about
This seems great, thanks!
I have no idea, but I'd like to know the ratio compared to bitcoin mining. Which I heard was over 1% of total, but I have no idea.
Bitcoin mining and blockchain transactions are intentionally orders of magnitude more energy-intensive than LLM usage. You'd have send something like a million queries to approach the energy usage of a single bitcoin transaction.
I think a lot of the "AI is killing the environment" rhetoric is actually people just pattern-matching their old environmental critiques of crypto onto the new tech bete noire, but because the magnitude of the energies involved on an individual level are so vastly different those talking points are no longer actually coherent.
Both of those would net you so little Bitcoin these days that it's completely insignificant.
You need ASICs for Bitcoin. Your CPU/GPU power is basically irrelevant. The malicious websites would instead be mining something more PC friendly like Monero, and they would still generate orders of magnitude less revenue than just putting ads. It's very noticeable because your computer fan is going to start blasting, so the hacked website owner is likely to catch it quickly. And there's no economic sense in doing this to your own website.
Thanks I'm clearly an idiot about this stuff. I will say that I hate that almost everyone has a narrative they are trying to tell. And so environmentalists* are telling their story.
*I like environmental stuff in theory, taking care of our planet. In practice it looks like many people with less than good information complaining about plastic straws or some other boogie man of the day.
https://removepaywalls.com/https://fortune.com/2026/05/19/data-centers-electricity-costs-us-public-opinion/ I think the main issue is that data centers risk cranking up electrical bills due to grid constraints. Electrical devices have become more efficient so we really have not had to do grid improvements. Data centers absolutely change that logic.
https://www.iea.org/reports/energy-and-ai/energy-supply-for-ai
"Global electricity generation to supply data centres is projected to grow from 460 TWh in 2024 to over 1 000 TWh in 2030". In 2024, global electricity generation was approximately 30,664 TWh. Data centres represented 1.5% of total electricity generation in 2024. This is projected to grow to 3% (1000/33600) in 2030.
"renewables remain the fastest-growing source of electricity for data centres, [...] meeting nearly 50% of the growth in data centre electricity demand. [...] Even so, [...] natural gas and coal together are expected to meet over 40% of the additional electricity demand from data centres until 2030."
"CO2 emissions from electricity generation for data centres peak at around 320 Mt CO2 by 2030, before entering a shallow decline to around 300 Mt CO2 by 2035. Despite rapid growth, data centres remain a relatively small part of the overall power system, rising from about 1% of global electricity generation today to 3% in 2030, accounting for less than 1% of total global CO2 emissions."
Of course this is solely focusing on the use of energy, not on its outcome. A useful idea when discussing the growing share of data centers as part of the world's energy mix is what your interlocutor and yourself expect to get out of it. How much extra growth would make it "worth it"? Where would one even start with this line of questioning? How closely does economic growth translate into better standards of living, and how equally are those standards expected to increase across the population? If your interlocutors don't have the slightest clue about these topics, I wouldn't take their opinion very seriously, no matter how impressive the numbers are. You can't just discuss the costs of some major investment/cost/externality without having an idea about its benefits or lack thereof.
We did save the Whales.
Specifically, most countries stopped large-scale whaling in the 1960s, or at least cut way back. The Soviet Union was the one big exception and did most of the hunting-whales-to-the-brink-of-extinction in the late 60s through the early 80s when they finally decided to phase out their whaling fleet.
It's this guy again. How many times do you need to get banned, Grimmoar/Wimbli/etc?
I've started reporting again, the comment of him further down about neoconservatives and Clinton is a dead giveaway.
Commenter may be female. Has mentioned a husband so statistically more likely. They dropped a comment to me in the previous thread that had a huge tell in the very first sentence.
What did he do?
His MO is to give hot takes and regurgitate conservative talking points backed up with whatever misinformation he can find.
Which doesn't sound too bad, but he does it dozens of times over the course of an open thread (scroll through this one for examples) and gradually escalates until he's banned. Then he waits a couple weeks, makes a new account, and starts again.
My *personal* animus against him is from when he was doing this while the Feds were attacking protesters in my city (Portland) a couple years back. He was parroting every bit of misinformation the admin was feeding Fox News, so he was claiming that the city was burning and disregarding anything from people who live here.
Lost in the discussion are two basic questions...
1. Do we really need massive data centers to implement current and future AI models? From what I've read, these models can be implemented on much smaller computer clusters, and cloud providers have been wastefully implementing them on infrastructure that hasn't been optimized for AI deployments.
2. Is China *really* ahead of us in AI model creation? Most of the arguments about China's lead are that they're implementing their models on smaller, leaner, distributed computer clusters — which we could do if we bypassed cloud service providers — rather than that their models are superior.
Unfortunately, I haven't been bookmarking the links to these arguments. But if people have serious data showing that we need huge clouds to implement AI, or that China has better models, I'd like to hear them. As with most of the discussions around AI, there are a lot of bullshitters (er, hallucinating commentators) out there. I don't really have the chops to winnow out the bullshit, though.
> Do we really need massive data centers to implement current and future AI models?
Isn't the fact that all the smartest / richest companies in the world are spending trillions of dollars on this proof that this is true?
Presumably they have the in-house expertise both on the AI and the data center side to know what they need, hardware wise - certainly FB and Google do, and that knowledge would percolate out to Anthropic and OpenAI and xAI.
LLM's run on GPU's, so they use CUDA, boomer.
OK. I got off my ass and did a little research. Yes, you can run a sophisticated LLM like Deepseek on a small cluster of computers (4 high-end Macs in this video).
https://www.youtube.com/watch?v=L8468vMTX2s
Of course DeepSeek is an open source model developed in China. But ChatGPT turned up several other open source LLMs — some of them developed by US-based companies. They come with pre-trained open-weight models. But there are also open source training corpora.
I asked the ChatMonster if it could run a small computer cluster, and it said, "The issue isn't just storing the weights. It's generating responses fast enough for millions of users simultaneously. A small cluster might eventually produce an answer from a frontier-scale model if you somehow had the weights, but the latency could be measured in minutes rather than fractions of a second."
But after further queries on my part, ChatGPT admitted that a proprietary frontier model running on a cluster of a dozen machines with adequate RAM and with high-speed network connection could provide very decent response for 5-10 users. Moreover, smaller cluster of machines running tensor processors (TPUs or LPUs), which are ASICs that optimized for token processing, could leave a larger cluster of NVIDIA GPUs in the dust (my wording). The thing about NVIDIA's GPUs, is that they provide a huge library of tools for developers. So (many) developers are sort of locked into using relatively-inefficient GPUs because they don't want to develop their own toolkits.
ChatGPT left me with this thought: "The gap between open models and frontier proprietary models is now less about access to transformer architectures and more about data curation, post-training, evaluation, and the enormous compute budgets required to execute those processes at scale."
So, I would conclude that a company like OpenAI required super massive data centers because they're locked into a inefficient technology substrate, that they adopted because they didn't want to spend the time optimizing their own. OTOH, ChatGPT estimates that a TPU-based data center would only reduce the electricity consumption by about 25% (still it's nothing to sneeze at), but optimizing the model would give much higher gains.
Yes, that's true, but I thought the real reason everyone is all-in on GPU's rather than ASICS or TPUS is basically that an ASIC (and to a lesser extent, a TPU) requires full lock-in on the current weights, and will continue to be locked in on those weights into the entire depreciation future of the hardware, and progress is breathtakingly fast. They drop new models roughly every month now.
A GPU is fully general, even if less efficient, and can both train and serve whatever new models you come up with.
TPU's aren't quite as locked in as ASICS, but aren't as general as GPUs. Since all of these things have 3 year (at a minimum) to 7+ year (you still can't buy used A100's because they're still being used to serve inference) depreciation cycles, the best bet is still GPU's while we're in the part of the curve where models have substantial updates within 1-2 years. And we're 10x faster than that right now.
So as long as the companies expect progress to continue at even <10% of current rates, it's better to build the giant data centers with less efficient GPU's.
I think a big question is "how much can you reduce the need for compute by design choices?".
So much AI talk focuses on the current/projected abilites of LLM's, that there isn't much awareness about other architectures being tested. Yann Lecunn's paper from 2022, for example, proposed a 'world model' approach; it's an interesting read with a more fully-fledged attempt to mirror human thinking. From the press release (and directly asking Gemini), it looks like Gemini's Omni video-gen is WM-based.
My intuition tells me that this is a more reliable roadmap to reaching AGI; have the AI use sensory data to learn to infer cause/effect relationships, in different modalities. WIth LLM's, you're essentially buying into a belief in emergence (for human-level abilities across g-loaded modalities).
I don't think China has to be in the lead (nor do I think they are) to pose a serious threat to the economic model of American AI companies. Token prices at OpenAI/Anthropic are rising quickly because you're not just paying for inference but RoI. What if you're not doing that for open source models? I keep seeing these graphs on X where the Chinese LLM's are months behind, but even if they're *a year* behind, if that means that in a year they might code as well as Codex now - that will be 'good enough' for many programmers.
The previously mentioned paper:
https://storage.googleapis.com/deepmind-media/Era-of-Experience%20/The%20Era%20of%20Experience%20Paper.pdf
> [Surely the august commentariat here understands that it is a really stupid idea to build "renewable" windmills in the salty "corrosive hell" that is the world's oceans?]
I deeply resent the implication that I have to go with your theory or else be stupid.
Make an argument from facts, if you wish, but stop skipping steps. Maybe you're even right. If this is important you can make the argument without insults.
I scanned this paper, and there are two things that came to mind:
1. This is about anything that works in the ocean generally, and I assume it includes things like oil tankers, docks, and a lot of other things they mentioned, but I didn't see windmills included anywhere. Also, I didn't see any cost-benefit analysis, but maybe it's buried in there and I missed
Exactly. Wind farms are largely made of steel but have a lifetime of 30-35 years. Not great but a lot can be rested.
>Surely the august commentariat here understands that it is a really stupid idea to build "renewable" windmills in the salty "corrosive hell" that is the world's oceans?
Corrosion protection is a thing.
Thanks, so ~1.5%, taking 6 years to double seems very conservative, I'd guess like a few years, 2-3.
6 years to double isn't really conservative, it's pretty aggressive. Grid constraints and lead times in power generation are the biggest roadblocks to closing the gap between announced electric capacity increases and what is most likely to happen.
I thought Hank green had a good video about it: https://youtu.be/H_c6MWk7PQc?si=y69cA3z61SiJAMMv
I was just going to recommend that same video! He makes a great point on comparing data centers to corn
UN released a paper on the topic. you can download the pdf here https://unu.edu/inweh/collection/environmental-cost-of-AIs-Enrgy-Use-Carbon-water-and-land-footprints
It seems to come to the opposite conclusion as the article from Andy Masley
It is finding ~the same numbers that Masley does but failing to contextualize them appropriately, in order to maintain a tone of grave concern. On page 27 for instance we get the grave invocation of "237 Olympic-sized swimming pools" of water used for training GPT-4. You could offset that water usage by reallocating 0.15 square miles of irrigated alfalfa in the Colorado river basin to be fallow. It sounds like a lot of water until you compare it to any other way we use water, which is what Masley does and the UN does not.
What they do do is denominating all the power and water usage in terms of "the minimum needs of X-thousand sub-Saharan African people" which is just kind of laughably manipulative, there is no first-world consumption good that comes out looking non-wasteful by that metric! But also if you build a datacenter in Utah the alternative case for the water and power used there is not that they were going to get shipped off to Burundi.
"'Civil Disobedience,' a thread by @hdthoreau" presents Thoreau's influential essay as though it were a social media thread. It includes "replies" from people who have engaged with and critiqued it over the years, and from Thoreau himself in the form of quotes from his journal and his later works that expand on the ideas in the essay. A "wikibot" and "biblebot" pipe up from time to time also, to explain now possibly-obscure references.
https://davgross.github.io/civil-disobedience-thread/
That's a great idea
That's pretty cool!
This is amazing. How did you make this?
The content came from a lot of searching in the archives (e.g. Internet Archive, Google Books, the old fashioned library, etc.). The presentation was vibe coded in cooperation with Claude Code. The avatars were generated by AI based on descriptions of the person or old public-domain photographs when these were available.
Awesome idea
I recently published Beyond the lexical personality traits: What is the structure of personality?
https://tailcalled.substack.com/p/beyond-the-lexical-personality-traits
Hi! I happen to be reading a lot about the FFM recently and I read over some of your posts. I think that maybe your concern about personality surveys having problems because of synonyms and similar questions has already been addressed by the field during the big argument about Implicit Personality Theory in the 1960s, but I'm not sure if I'm missing something about your argument. See: https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-6494.1992.tb00975.x
I recently wrote about how a retail investor can invest in “the whole market, except [some small subset they want to exclude]”.
https://world.hey.com/cgaebel/direct-indexing-with-exclusions-1f98ff02
There's at least one fund that tries to mimic Nancy Pelosi's stock picks:
https://duckduckgo.com/?q=nancy+pelosi+stock+fund&kp=1&ia=web
If I use Frec, once I exhaust my tax-losses to harvest, how much does it cost to keep on using them?
I cover that in the “Limitations” section of the article. Short answer is 0.09%/year. You can transfer assets out but managing hundreds of different stocks on your own is a pain in the ass.
Can I transfer out fractional shares?
Thanks for asking that. It’s a great question, and I hadn’t considered it until now. I think fractional components are very likely to be liquidated, and you’ll take the cap gains tax hit. I’ll email them to confirm, then update the post with my findings.
Edit: Confirmed, they will sell the fractional components.
Thanks. I checked the math against https://www.slickcharts.com/sp500 and if I invest 100K in all 500, 36% of it is in fractional shares if I round down. If I somehow have the chance to round up (as in, they will let me top off shares where I have more than half a share) then I'm off by 26%.
If I have $1 million, then the numbers are 5.7% and 2.8%.
I don't think I'm anywhere in the range for this, but if I were I'd like a mode where I can say "please gradually shift towards wanting whole numbers of shares."
If I limit myself to the top 100, 100K will only have 16% in fractional shares, rounding down. Then I use XOEF for the rest.
This is from the thread about Frankfurt School ( https://www.astralcodexten.com/p/book-review-the-dialectical-imagination )
"By the 1930s, people had started to notice that this wasn’t happening. Although capitalism had suffered some crises - World War I, the Great Depression - countries had responded with what the Marxists called state capitalism, a broad term for hybrid systems including New Deal America, Stalinist Russia, and Fascist Italy. Most people seemed to like their own country’s variety of state capitalism. Some Marxists began to worry that maybe Revolution wasn’t inevitable after all."
The way Marxists invented "state capitalism" is very reminiscent (to me, anyway) of Ptolemy's epicycles. Ptolemy was certain that planets move around the Earth, and when they visibly moved "backwards", he came up with the idea that instead of circling the Earth, every planet circled an invisible point, and *that point* circled the Earth. Later he had to add more epicycles -- every planet circling a point, which circled a point, which circled... the Earth. His model grew more and more complicated because he could not allow the notion of *Earth* moving.
"State capitalism" is not something Marx ever thought of. It is something his disciples added to Marxism when reality failed to keep up with their model. So was the idea that Lenin "cargo-culted" Marxism, as Scott so pithily out it. (German and Austrian Marxists certainly did not think that while Lenin was still alive!) And when you are utterly convinced your model is TRUE, yet must keep adjusting it to match empirical evidence, isn't it pretty much the definition of a religion? Or at least of a religious faith?
> "State capitalism" is not something Marx ever thought of.
Minor quibble. In The Communist Manifesto (1847–48), Marx and Engels wrote: „Vermehrung der Nationalfabriken, Produktionsinstrumente, Urbarmachung und Verbesserung der Ländereien nach einem gemeinschaftlichen Plan.“ Which I read as "Increase of national factories and instruments of production; cultivation and improvement of lands according to a common plan." But they don't explicitly state that all factories must be under state control. Only that it would be good if some factories were state-controlled. It's rather vague, though, and post-Marx Marxists seem to have argued about the meaning and implications of that phrase.
However, later in his life (after the Paris Commune of 1871) Marx wrote: "One thing especially was proved by the Commune, viz., that 'the working class cannot simply lay hold of the ready-made state machinery, and wield it for its own purposes.'" In the same essay, he wrote that what he meant by "state machinery" wasn't merely the government in the abstract, but "the centralized State power, with its ubiquitous organs of standing army, police, bureaucracy, clergy, and judicature." Anyway, Marx's ideas were evolving. And his later stance seems to be the opposite of the later Leninist interpretations of his political philosophy.
> isn't it pretty much the definition of a religion? Or at least of a religious faith?
French sociologist Émile Durkheim was the first to point out that Marxism resembled a religion. The anthropologist Clifford Geertz formalized a systematic definition of religion. He argued that religion was a "system of symbols that establishes powerful, pervasive, and long-lasting moods and motivations by formulating a general order of existence." Much to the angered sputtering of Marxist social scientists, he tossed rigid political ideologies like Marxism into the religion bucket because they functioned identically to religions by providing an all-encompassing "cultural lens," an ultimate meaning for suffering, and an "aura of factuality" to their followers.
You should polish the lenses of your rightwing-fogged glasses. You don't really understand how CRT and Queer Theory appropriated Marxist dialectics to argue that it wasn't the working class but racial and sexual minorities who were being exploited and oppressed. Traditional Marxists were appalled by this heresy, and much hilarity ensued on Marxist forums as CRT-QTers attacked them and Marxists protested that they weren't racists or sexists. Progressivism has always been about the willingness to consider innovative solutions to social and political problems. The arguments among progressives have always been about what actually works — that is, until the CRT-QTers started demanding that progressives cater to their whims. Because of the ethical framework underlying progressivism, It's hard for progressives to turn away a victim (even if the victim is exaggerating their victimhood). On the left wing of the spectrum, this is analogous to how MAGA took over the GOP and silenced traditional conservatism — except we still have some sanity left in our house — i.e., that winning over the centrist voters to win elections is more important than movement purity.
Please go away
I don't know that progressive arguments were "always about what actually works", surely there's been a long running throughline from the 19th century to now that A) society should be advancing towards some proposed other state which they believe would be preferable and B) rule by experts is the best way to produce this.
It's certainly possible to have a different approach to "what works", one could simply try to refine the existing society and just make that work as intended, as the people who built the existing society would have wanted it to function. I could refine my widget factory to make widgets at half the costs and reduce the number of defective widgets to a trivial number, or I could decide that what I should really be doing is trying to iterate towards manufacture of a SuperWidget from my Shining Widget Factory on a Hill. The former is about "what works" but the latter is much closer to progressives' goals. The "what works" people are the neolibs and a few of the wonkier conservative think tanks that have survived Trumpism.
Likewise, it may well be that "what works" is decentralized, local, autonomous units, a system following the principle of subsidiarity where decisions are made at the lowest level they possibly can be made and we try to improve the competence of systems at those levels to return more power to them. But progressives would be fundamentally uninterested in that solution, the existence of a Porfiriato style gang of elite technocrats steeped in logical positivism is baked into the worldview. The neolib sorts also take this assumption for granted, but with evidence that local control worked or that nationalizing X policy consistently fails I think Ezra Klein etc would accept it. The progressives need society to be ruled because society must be moving towards its destiny as a collective enterprise. So even their internal arguments are not always about "what works" because the whole project of progressivism entails creating a thing that doesn't currently exist, so the contours of that hypothetical thing that everyone must be forced to work towards are of much importance to them.
I should make a distinction. There were actually two progressive movements on the Left side of the spectrum since 2000. After George W. Bush won in 2000 (after Al Gore's pathetic campaign), there was a movement to remove centrist Democrats from the party's internal levers of power. Markus Moulitsis was one of the flag-wavers of this movement (you can look up his writings on the DailyKos). He advocated for fielding less wimpy (i.e., less polite) candidates who would fight back in the media against the slurs of the Right and advocate for middle-class-friendly policies. He advocated for greater social safety nets, but more in the FDR style than the socialist let's-reform-the-entire-system style. There were all sorts of wonkish discussions about what would benefit the middle class, what constituted the middle class, or whether we should even promote policies that would benefit the middle class at the expense of the working class. These progressives thought they had elected one of their own when Obama won his first term.
But Obama installed a bunch of centrist/corporatist technocrats to run his administration. Both Obama terms were Bill Clinton 2.0. He didn't get us out of Afghanistan and Iraq as promised, and he barely got healthcare reform passed. Centrist Democrats still controlled the party, and a new batch of progressives assumed the movement's name. Bernie Sanders, AOC, and that crew started gunning for Wall Street (much to the discomfort of centrist Dems who got a lot of campaign donations from Wall Street). The wonkishness of the first wave of Progressives is gone. They're fighting to reform the entire system, so they're fighting centrist Dems as hard or harder than they fight Republicans. They hated Joe Biden (despite Biden getting more of the liberal agenda passed than Obama). Meanwhile, the centrist Dems start sounding downright Republican when populist lefties like Mamdani entered the fray to challenge their preferred moderate candidates.
The debate over whether we should field more moderate or more radical Dems is ongoing. Today's progressives are not the same as the first wave of progressives in that they're less interested in arguing over policy wonkery and more interested in unabashedly advocating for a social-democratic/democratic-socialist ideal that they (mistakenly) believe the Nordic countries embrace.
This comment seems more reflective of your own biases than actual reality. In particular, this seems very much more like something that you would *like* to describe reality than something that actually *does* describe it:
..." Floyd and others enshrined as 'saints'"...
I've spent quite a lot of time in various progressive spaces and I've never, ever seen people talk with veneration or reverence about George Floyd. Indeed, that entire mode of discourse--elevating *any* human to reverentially authoritative status--is often looked on with suspicion[1]. In reality, George Floyd specifically seems to be far more of an object of obsession on the right than on the left. I don't see his name mentioned especially often in progressive spaces and when I do it's rarely mentioned centrally; it will be paired with names of other victims and/or discussed in context of systemic issues with policing. Despite spending far less time in conservative spaces, I've almost certainly heard the name and the qualities of the man mentioned far more often by right wingers than by left wingers.
"public veneration of them along walking paths"
And at this point I must gently ask if you understand the difference between an object of reverence and, well, a symbol. Symbols are extremely common out in the world; can you really not recognize one when you see it? The reason that George Floyd's death was an important event isn't because of any prior notoriety or particular qualities of the man himself. It was because the events of his death made him an extremely potent symbol: it was a stark illustration of issues with policing that progressives had been complaining about (with increasing anger and exasperation) for decades. That makes mentioning the man a convenient shorthand for mentioning the problems, convenient because it reminds people in dramatic terms why the issues are important. Examples of this sort of thing abound in American politics: it really should not be hard to recognize.
"and required displays of 'religious fervor'"
REQUIRED, you say? By whom? Enforced by what? Can you point to me any examples at all of somebody being penalized in any way for not displaying a pride flag?
To me, this again appears to be a rather deep misunderstanding of your fellow humans. I don't believe I've ever displayed a pride flag my own self, but it's extremely simple to understand why people do it. There are two good reasons and a bad reason:
1. Queer people do it as an act of defiant self-expression. As much as conservatives like to pretend otherwise, being openly queer does still expose people to hostile reactions, both overt and covert.[2]
2. People who are friendly to queer people (whether themselves queer or not) do it to signal solidarity, indicating that they don't have any hostile intentions towards queer people.
3. Public relations and marketing people people try to parasitically co-opt the above trends just like the do with *every* social trend, to try to blend in with groups 1 and 2 and more effectively extract money from them.
Note that the last one only works because as of the past decade or so, the broadly queer-friendly progressive cultural blob has become a more economically important demographic than the out-and-proud bigot cultural blob, so corporations will plausibly gain more money from it than they'll lose. It never ceases to amuse me when conservatives see for-profit companies engaging in a transparently cynical and profit-motivated displays--the way capitalist values say they absolutely should--can conclude that it must be the result of some sort of cartel or conspiracy. Free market gonna free market.
[1] Which is not to say that individual people in progressive spaces never do it, just that wider groups tend to be very quick to push back against it.
[2] To be very clear, this situation is much, much better than it was 20 or 30 years ago. Nowadays people who live in the right place and move in the right social circles might only rarely encounter it, especially if they're the "right" kind of queer. But it's far from entirely gone, and in more conservative areas I gather it's still pretty overt.
Remember, there's a completely parallel media universe that still believes bizarre stories like the Clintons running a pedophile ring run out of a pizza parlor. There's not much one can do to argue with them, I'm afraid. But thanks for trying!
Marx may not (apparently?) have ever used the term state capitalism, but Marxism certainly has always seen the operations of the state as an important part of th ecapitalist system. The idea of state power expanding prodigiously as a way to answer capitalism's various problems would not have been alien to Marx.
>German and Austrian Marxists certainly did not think that while Lenin was still alive!
Uh, what? Many of them were opposed to Leninism from the start.
> Uh, what? Many of them were opposed to Leninism from the start.
Thanks, I was not aware of that!
Ah. Yes, I knew about this, but did not realize it was formulated that early.
"And when you are utterly convinced your model is TRUE, yet must keep adjusting it to match empirical evidence, isn't it pretty much the definition of a religion? Or at least of a religious faith?"
Sounds like you are describing "updating your beliefs against your priors" in bayesean analysis? The real question is at what point should you abandon your priors in the face of overwhelming evidence and that's something everyone struggles with, not just religious folks and communists.
> at what point should you abandon your priors in the face of overwhelming evidence and that's something everyone struggles with, not just religious folks and communists.
This very much was in my mind when I wrote the initial comment. But I did not want to complicate it. Glad you brought this up.
To me there is an irony so profound as to be nauseating in the fact that many rationalists will on the one hand dismiss Marx and his empirical analysis of labor markets and technological trends as quackery and then turn around and champion prediction markets, AI, and the singularity, as if all of these things aren't extensions of the same basic principles of analysis applied to the 21st century as opposed to the 19th.
Don't blame me, I am not one of those "rationalists". As the matter of fact, I think Marx's diagnosis of the ills of the 19th Century capitalism was spot-on; too bad his prescription turned out to be wishful thinking. Does not mean I have any faith in prediction markets, AI, or singularity.
Prediction markets may have some use. The singularity does sound like Marx’s idea of what the future could hold - massive productivity enabling people to do what they want, and where work becomes a want, not a need.
Actually, the problem is not regarding the Earth as unmoving. Copernicus also had epicycles in his heliocentric system.
The problem was that they thought that whatever the orbits were, they had to be composed of perfect circles, instead of the ellipses that they actually are.
It was therefore Kepler that was the real innovator.
(Modern perspectives on earlier thinkers can often be off like that. It is useful to reflect on.)
Yes, that's why Ptolemy needed nested epicycles. If planet orbits were perfect circles, his model would have needed only one epicycle per planet.
It also meant that, strictly speaking, Ptolemy could not be falsified as such. As a college professor pointed out to the class I was in, the system is a Fourier series, which can always approximate any function to any specified degree of accuracy.
What replaced it was superior simplicity.
And Kepler was only able to develop his theory due to the extremely precise measurements carried out by Tycho Brahe.
To be fair, Ptolemy's model was highly accurate given the technology of the time. He was following the evidence he had, not stubbornly ignoring the obvious the way he is often portrayed.
It's still used to calculate where the planets would be in the Earth's sky.
Capitalism is not a designed system like theoretical communism is. It's just something that evolved historically. And what we understand as it is basically on one hand the private ownership of the means of production, and on the other –more importantly, I will argue– the financialization of the economy. The first started with the acts of enclosure and similar events in the West that ended the medieval system. The latter started with the origins of credit in Venice, The Joint Stock Corporation in the Netherlands, fiat currencies, etc.
What the Soviets lacked even when opening up trade was the financial side. So I don't think it's fair to call that ever state capitalism, unlike fascist economies, the New Deal, and basically all economies nowadays. What the Soviets eventually did was focus on making profits in their state ventures And that was because they had to trade with the West. They had no choice. And there's the simple fact that they needed a surplus of foreign currency for that.
>"State capitalism" is not something Marx ever thought of.
This gotcha shows up all the time, and it's (to use a neutral word in a spirit of kindness) meaningless. Most of contemporary leftist terminology was established after Marx's death, usually originating from anarchist thinkers. This includes "state capitalism". Marxists adopted it, because it was, well, correct. Perfectly congruent with their beliefs, too. Used as a neutral description of Soviet Union's economy by Bolsheviks themselves (up to a point, if you're looking for some kompromat on them, you should be asking why they stopped).
But aside from the fact that there was nothing to adjust in this particular case, adjusting your model to match empirical evidence is perfectly normal, and usually called "science". Imagine holding that against someone!
You're leaving out the part where your model has to eventually make predictions that hold. If your loop is making predictions, watching them fail, adapting the model so they retrodict what the previous predictions didn't, go back to "making predictions", but your model never clicks into "watching those predictions hold up", that's... technically science, I suppose, but no more useful a science than the epicycle model.
And to insist other people are dumb because they're criticizing something after your side called it "science" because criticizing anything called "science" is dumb, makes your side look like cargo cultists.
Thank you, I was going to post pretty much what you wrote, but you got there first. BTW, love the "retrodict"!
This is a good criticism of contemporary establishment economics.
But I digress. The issue in this thread is someone insisting other people are dumb because what they're actually saying isn't what he believed they were saying. (And insisting they should actually be saying what he believed they were saying, and any instance of them saying something else is "inventing epicycles", and evidence they're a religion, because if they weren't they'd be saying what he believes they should be saying... yeah, I can't really make sense of it either.)
I'm unclear if by "state capitalism" we are talking about the fascist states (public control of means of production with private ownership and wage labor etc etc intact) or Soviet style bureaucratic collectivism? Or just a phase of the proto-communist society prior to the "withering of the state"?
From Hayek, I take the view that fascist states (and perhaps America under FDR) evolved as a third way position, they were anti-liberal but also anti-communist. That something like that would evolve seems like it would have been really obvious to Marx to Engels.
Likewise Marx envisioned an iterative process, not arriving full formed at the stateless society, that part wasn't novel to the USSR. My understanding is the novel part Lenin added was the "vanguard of the proletariat". That the early forms of this could be called "state capitalism" again doesn't seem like something Marx would have missed, as it was pretty clear even before Lenin that workers just walking in to seize factories wasn't going to work due to the existence of state power.
Now perhaps Marx did not envision the nomenklatura and the tendency of these ventures to stall out in bureaucratic collectivism, as this theory was developed later by Trotsky based on new evidence, and it was not simply a recapture or subversion (which I'm sure Marx could imagine) but a genuinely New Class.
>public control of means of production with private ownership and wage labor etc etc intact
More or less. State, not "public" (especially if you want to include fascist states where there's no pretense of the public having any say), and private ownership is not a prerequisite (as Soviet Union at certain points demonstrates). The original point was that state involvement (alone) will not change the fundamental nature of the system. This... had happened, arguably all modern states are state capitalist by the original formulation, and we mostly just call them capitalist by now. This left nominal workers' states like the former USSR and contemporary China as the only remaining meaningful/interesting application of the term.
I guess what you're asking for is - no, this is different from Marx's early stage of communism - money and markets are still used internally, workers work for wages, the country participates in the broader global capitalist economy. Marx called for a global revolution, he didn't believe a communist society was possible without it. So when the global revolution didn't transpire, it left the places where it did transpire with the question - okay, what now? They couldn't proceed with implementing communism because they were stuck in an economic (and military, let's not forget military) competition with the remaining capitalist world (as it stood, much better developed at the time). The solution was - keep doing capitalism. After all, one thing it's good at is, well, economic competition and development. Not as an end goal, but as a pragmatic least-bad option, a stop-gap until the rest of the world comes around, which they were positive they can pull off as long as the workers are ultimately in control. (Well they weren't, but that's another story entirely.)
" And when you are utterly convinced your model is TRUE, yet must keep adjusting it to match empirical evidence"
Updating your models to match evidence is exactly the opposite of a religion. Religion changes itself *despite* the evidence.
See Paul Brinkley's comment. If you keep making predictions and they never come true, and you keep modifying your model, and they STILL do not come true -- at some point you must admit that something is fundamentally wrong with your underlying theory. Otherwise you have the equivalent of epicycles -- a model which matches past evidence very well, but still has no predictive power.
Again your claim (stated rhetorically) was that Marxism is a religion, so whether the model is wrong or not is irrelevant.
Religion is an attempt to help people cope with uncertainty by inventing rituals and figures. It is very rarely that religions even acknowledge contradictory evidence, they don't have to because religion is not trying to be predictive. When they do, they will usually invent explanations from thin air rather than referring to any empirical phenomena.
Wrong or not, the Marxist Leninists did actually try to make a predictive model of reality for the purposes of solving an economic problem by actively analysing their environment, which is both practically and fundamentally different from religion.
You're using the steelman definition of religion to refute a point that was using the strawman form. If you want to criticize the OP for using that definition, that's fine and all, but the purpose of the OP wasn't to criticize religion; it was to criticize Marxism by claiming it was adhering to religion-in-bad-sense.
Which is what your third paragraph pushes back against, but in that case, the OP position seems to be that Marxist-Leninists employed a simulacrum of scientific analysis, tantamount to assuming the conclusion and interpreting the evidence to fit. And perhaps expressing it in scientific jargon so that it would appear so; if so, this only makes it doubly bad.
My historical background on Marx and Lenin is admittedly scant, but somehow I doubt that they bowed to the critics who presented counterevidence. And besides, the OP wasn't referring to Marx and Lenin specifically, but more to their more recent adherents, particularly those who invented the term "state capitalism". They appear have adhered more to the dialectic than to the scientific method. Who were the strict scientists in the Marx-Lenin camp who pushed back on them?
And this is without getting into the question of whether religion-in-good-sense is doing people any service by walling them off from reality in order to help them cope.
Marx famously engaged with counterarguments all the time. Engels, the man who first wrote about state capitalism, even went on to deny some of his models by the end of his life. Leftists as a whole are famously known for disagreeing about whose model is most correct, even today. If you were more specific about which people you were referring to it might help, but it sounds like you don't really know the history or want to understand things, you just want to discredit a group you dislike.
At best you can argue some subgroups of some Marxist ideologies are religious, which is just a useless claim. Plenty of other things are religious in that case, including physicists who believe in string theory. We don't discredit all of physics as religious.
Eh? Likening belief in string theory to religion isn't as useless as you seem to imply. If true, it points out a flaw in the reasoning model, and the first place to look for evidence that the conclusion might be false. And yes, it allows us to draw distinctions between scientific physicists and religious physicists.
Which is why I asked about whether there were strictly scientific Marxist-Leninists. Contrary to what you hypothesize, it wasn't an idle rhetorical question; I genuinely wanted to know. (How would you prefer I asked?) If Marx himself (or Engels) tried to be scientific at the expense of his own thesis, same applies.
Admittedly, knowing this is of limited use. Both fellows are dead, their most important ideas did not produce working predictions, their most prominent disciples wreaked a great deal of havoc trying to implement them, and that damage tends to overshadow any extent to which their less prominent scholars happened to be scientifically open. The historical narrative might change. But so far, if it does, it looks like it will just say there were even more reasons the those Marxists were mistaken.
The rest of this thread reminded me of another property I frequently notice in Marxists: sophistry. In this case, conflating two senses of the same word, and later making claims that don't hold and then criticizing their counterpart for ignoring the claims they fabricated. And topping that off with "that's just how words are used", belying a casual approach to words when they use it, and strict when others do.
I'm trying to remember the last time I discussed Marxism with a defender that didn't descend to this level. It's easy to remember the opposite (cf. marxbro). This could be selection bias, but weak selection bias usually goes away when making an effort to remember counterexamples.
This forces me to question whether Karl Marx engaged with counterarguments without similar descent.
You can go and read what he wrote about and to whom he wrote, the information is available. Quite ironic to accuse others of sophistry when you refuse to inquire.
You would know irony well, then.
This is another update to my long-running attempt at predicting the outcome of the Russo-Ukrainian war. Previous update is here: https://www.astralcodexten.com/p/open-thread-429/comment/242696754
10 % on Ukrainian victory (up from 9 % on March 13, 2026).
I define Ukrainian victory as either a) Ukrainian government gaining control of the territory it had not controlled before February 24 without losing any similarly important territory and without conceding that it will stop its attempts to join EU or NATO, b) Ukrainian government getting official ok from Russia to join EU or NATO without conceding any territory and without losing de facto control of any territory it had controlled before February 24 of 2022, or c) return to exact prewar status quo ante.
24 % on compromise solution that both sides might plausibly claim as a victory (up from 23 % on March 13, 2026).
66 % on Ukrainian defeat (down from 68 % on March 13, 2026).
I define Ukrainian defeat as Russia getting what it wants from Ukraine without giving any substantial concessions. Russia wants either a) Ukraine to stop claiming at least some of the territories that were before war claimed by Ukraine but de facto controlled by Russia or its proxies, or b) Russia or its proxies (old or new) to get more Ukrainian territory, de facto recognized by Ukraine in something resembling Minsk ceasefire(s)* or c) some form of guarantee that Ukraine will became neutral, which includes but is not limited to Ukraine not joining NATO. E.g. if Ukraine agrees to stay out of NATO without any other concessions to Russia, but gets mutual defense treaty with Poland and Turkey, that does NOT count as Ukrainian defeat.
Discussion:
Since my last update, several developments incrementally improving Ukrainian position have taken place. Each of them by itself would not be that important, but together they amount to something. Below I’ll list some though not all them.
Ukrainian front, both military and domestic, appears to be holding surprisingly well in last few months.
Sweden agreed to provide Ukraine with Gripen warplanes (see e.g. here: https://www.reuters.com/business/aerospace-defense/swedens-gripen-faces-moment-truth-ukraines-air-war-with-russia-2026-06-04/).
Massive improvement in relations between Ukraine and Hungary, (see e.g. here: Ukraine moves closer to EU accession after Hungary deal ) which was likely but not certain after Orban’s election defeat, is clearing the way for more EU support.
German chancellor Merz came up with clever proposal to give Ukraine some benefits of EU membership instead of full membership (latter is imho unrealistic on the timeline relevant for the war). See e.g. here: https://www.euronews.com/my-europe/2026/05/21/merz-proposes-associate-eu-membership-for-ukraine-with-mutual-assistance.
* Minsk ceasefire or ceasefires (first agreement did not work, it was amended by second and since then it worked somewhat better) constituted, among other things, de facto recognition by Ukraine that Russia and its proxies will control some territory claimed by Ukraine for some time. In exchange Russia stopped trying to conquer more Ukrainian territory. Until February 24 of 2022, that is.
I've been reading these for a while, and I'm curious about your thoughts on the difference between your updates and the broader reporting on the conflict. Your estimates have been quite stable through time (while I wouldn't expect large swings from update to update, they seem to have been random-walking around the same values for nearly the entirety of your coverage), despite a host of things that seem like they should cause updates in one direction or another.
While this isn't something that I keep a close eye on, the articles that I happen to encounter seem to be more in line with relatively large changes in the status of the conflict. Things like the transition of the conflict to drone warfare in large part, Ukraine's ability to reach into Russia, the Iran war and it's global effects, among others, all of these seem like relatively large changes to which your estimates don't seem to shift very much, so I'm curious what you have to say about that.
-edit- I just double checked, and in fact your very earliest estimate (or at least the oldest one findable by following your link chain) from over 4 years ago has Ukraine at......10% chance of victory.
Good question and good to know I have readers. Media often exaggerate an importance of day-to-day developments, in this war as well as in, say, election campaigns; something which Nate Silver, from whom I am of course trying to learn this prediction “business”, often emphasizes.
In addition, worst mistake I made in this series was an over-upgrade; I’ve started with what I still think very fairly reasonable odds, but then, due to quick Russian advance, I’ve moved them clearly too much against Ukraine right before Russian retreat from Kyiv region. I hope I’ve learned not to be too overconfident after that. In January I made a retrospective graph where it is all too visible: https://aleszieglerenglish.substack.com/p/special-military-graph-updated-again. That graph however also shows I made some other large upgrades, which I do think were justified.
Re: transition to drone warfare, it is unclear to me which side ultimately benefits from that; now it seems that Ukraine has the upper hand in drones, but few months ago it was Russia., and I guess it might swing again. Something similar is true also about Iran war, even though according to many experts it benefits Russia. Of course higher energy prices benefit Russia, but the war has also other repercussions, which are good for Ukraine. For example, it looks like a disaster for US Republicans, and also it makes Ukraine expertise in drone warfare seem useful to countries which are clearly locked out of using Russian expertise in it.
With regards to succesess of Ukrainian drone strikes against Russian mainly oil industry in last few months, I originally wanted to list it in this OP among incremental factors making me more optimistic about Ukraine. Then I forgot to do it, ups.
Would love to see a debate or adversarial collaboration between you and beowulf888 .
Big fan of the following reviews in the contest:
- This Is Going To Hurt
- Seeing That Frees
- The Wisdom Of Mike Mentzer
- The Bible (ironically this one)
Also curious about running pangram on all reviews, should be around $50
The Bible review struck me as 1) probably written by AI, and 2) far too brief and surface-level to be a serious contender in an ACX book review contest. What made you choose it as one of your favorites?
Ironically in this case probably refers to the sort of wit where you say the opposite of what you mean to get your point across. Two days in London that are constantly cloudy: “Sure glad I brought my Ray Bans!”
Excited for the Seeing That Frees review!
Seconding This is Going to Hurt
It’s such an outlier it almost should be in its own category.
In anime, it's common for highschool students to be shown getting part time jobs. I can recall a few cases of this happening in American media and I know it happens in real life too, but it still seems like a huge difference that presumably points to cultural and economic differences with Japan.
It used to be more common than not in the US also. The change seems mainly in the US that our high school kids are too busy. I used to think this was some status flex by parents but now I have a freshman I see it’s not - with one or two big extracurriculars plus homework plus modest socializing they have no schedulable time they can use for a job. My kid tried dog sitting for a neighbor and that didn’t work because extracurricualrs kept getting in the way
I wonder how the numbers vary for part-time jobs during the school year vs. on summer (or winter) break.
I didn't have the time or energy for a job other than very occasional babysitting during the school year. But I worked every summer and some winter breaks starting at age 14, unless I had an internship then instead. That was 25 years ago though.
Eldest son has two volunteer jobs this summer, one of which might turn into a real job. I don't see a lot of local businesses hiring teenagers for the summer the way they did in the tourist town I grew up in, but it does seem like most of the highschool kids are doing something that's not just education or recreation - volunteering or working or interning.
> one of the big construction sites was all illegal immigrants
How did you find out they were illegal?
I often come across building sites around here with the workers speaking languages other than English to each other, but I honestly have no idea how (or why) I'd check their documents without risking a punch in the face or worse.
Folks under 18 are pretty limited in what they're legally allowed to do in a lot of work settings-- that includes construction & food service.
Outsider question: What is an "extracurricular" in this context?
Anything organized by the school that isn’t core academic work. The big time consuming ones are music, sports and theatre.
Just an interesting comparison on the point of US kids being too busy for part time work:
While I was living in japan, ~8 years ago, it was not uncommon for high schoolers to go to school in the day, then their (informally mandatory) sports club, then juku (cram school) and then part time work until around midnight before heading home. They'd get their sleep in the class and on the train. Out of club, juku and part time work, the first two are basically obligatory but some kids needed or wanted work too.
I remember getting odd looks when I mentioned that cram school might not be necessary if they actually had enough sleep to hear what was happening in class.
I guess "busy" is culturally contextual.
Yes, I was thinking about this. I suspect part of it is the relative uniformity of Japanese society and expectations. In the situation you describe, basically all kids are available from when juku ends to the close of business for the day, so if I'm a business owner I can have clear expectations. In the much more individualistic and heterogenous US society, different high schools have different dismissal times and then different extracurriculars will have different expectations on different days, and then kids get homework irregularly and have different systems for getting it done. So even if I'm just a family who need a babysitter, I can't even get one clear evening a week where a given teenager will always be available, and there definitely aren't any nice clear expectations as to when I should generally expect I can find someone available.
I tried really hard to hold down a part time fast food job in high school. I only made it about 8 weeks before my manager stopped even trying to find shifts I could work. I could occasionally swing a closing shift or two a week, but between school and sports I almost never had 4-6 uninterrupted hours during the day to spend doing anything. And this was 20 years ago, before the extracurricular rat race really got thrown into high gear.
As recently as my high school years (... Not so recent, I graduated in '99) I would guess that about half my classmates had some kind of job. Babysitting, construction, lifeguarding. My parents marched me down to town hall for my working card the day I turned 14. They were fine with my 3-hour, $15.45 a week library job. The point was not so much to make money (though I put it to good use), it was to get work experience
For context this was in rural/suburban New York state.
But yeah, kids in the US don't seem to have jobs as much now. As someone who has hired a lot of college students for work-study jobs, I think this is really bad. People who have worked before, ANY kind of work, are miles more mature. More ready for school, more ready for work m
Another factor is that high school students are competing with more adults and older workers for the same entry-level jobs in retail, food service, landscaping.
I have noticed over the past 5 years or so, nearly all of the classic "high school jobs" in my area have been filled by immigrants. Lifeguards and babysitters seem to be an exception though.
As a 14 year old in the UK (circa 1980), practically all of my friends had jobs — mostly delivering papers etc, or working in shops. When my kids were in high school in California, practically none of their friends had jobs. I think that’s a shame. A little work experience in your teens makes a huge difference to your confidence when you get your first proper job.
Two quick responses to other comments:
1. Too much homework now? Back then, we had way more homework/projects/etc than my son’s generation.
2. Too little employment? The UK at that time had the worst unemployment figures ever.
I think the differences are cultural: we have decided that children need to stay children for longer.
Not just that we need them to stay children longer...we also just don't trust them. Helicopter parenting is real.
I also don't buy the "competing with adults" thing. I have so many parent friends who simply can't find high school age babysitters. A part of me suspects they don't really want to because they don't trust them, but I also believe them when they say there aren't enough.
"it was to get work experience."
When I was starting my first "real' job, my older colleagues were able to judge (with 100% accuracy) which of my cohort of recent college grads had worked a high school or college job. That minimum wage job did give me some discernible soft skills apparently.
Coincidentally, the employment rate for teens seems to have dropped after 1999. https://fred.stlouisfed.org/series/LNS12300012
Interesting. That's quite a drop.
Yes, though it would be nice to see just 16-17 yr olds. And an adjustment for full time student enrollment. High school dropout rates have declined since 1999 as well.
https://www.pewresearch.org/short-reads/2014/10/02/u-s-high-school-dropout-rate-reaches-record-low-driven-by-improvements-among-hispanics-blacks/
https://www.smartick.com/data/charted-high-school-dropout-rates-in-the-united-states/
I graduated in '96, in a mid-size midwestern town of 150k with a protestant and rural/suburban culture. I would say that in my high school every junior or senior either had a job or was trying to get a job, unless their parents were upper middle-class or better. Boys needed money for dates or saving to buy a car, and girls needed money for I assume makeup and clothes and all that. I am absolutely baffled about these current teenagers that don't have jobs, are mommy and daddy just supposed to hand them cash for everything they want now? Do the kids in lower-income households just passively accept that they will look worse and have less since it's not being handed to them?
And what in the world are they *doing* all summer? The striver kids with middle-class parents back in the 90s would have Boys State or music camp or whatever for a couple weeks and then probably a family vacation, maybe now they have volunteerism or some other career prep thing. But the lower-income kids would be working 25-40 hrs a week, if they aren't doing that now are they just sitting on a couch in the basement for 10 weeks?
I bet working-class kids do still have jobs. Embarrassingly, though, I don't know any working class people anymore, so I don't know. But yeah, I'm guessing it's only the middle class that has changed.
It's actually the opposite. Part-time jobs for high school students in Japan is far less common than in the West. It happens in anime because the author wants another setting / more characters for the main character to interact with.
The OECD average is about 45% of students (https://www.oecd.org/content/dam/oecd/en/publications/reports/2025/02/teenage-part-time-working_75275b29/0dd35152-en.pdf) , while Japan is around 10% (https://www.nippon.com/en/japan-data/h01423/).
1) I wouldn't call cleaning up the school as "part-time work." It's closer to chores than work.
2) School clubs, as you note, are voluntary, and the majority of the time don't do productive work. They are majority sports-related (https://shin-edupower.com/japanese-club-activities/).
Huh, thank you for providing real-life data.
This was my impression. I dont think Ive ever encountered a hs student working in japan aside from school related volunteer work. Seems incompatible with cramming needs. They dont build resumes they have to max out their score.
Narrative device aside the generation of anime authors may be writing about an earlier time.
Even Beavis and Butthead had part-time jobs!
Wonder if it's legal differences too. My company sometimes hires highschool students, youngest I know of was...I think 16? But there's a lot of restrictions on what they're allowed to do, even though grocery stores seem pretty straightforward. Can't run the baler to crush cardboard, can't work the alcohol sections, can't break down pallets, can't have a blade in their boxcutter, can't run a register. So mostly they end up getting assigned to clean the store, collect carts, work sections where product isn't boxed (say, produce)...which is fine, but there's a reason we don't ordinarily hire seasonal temps, or "checkers" who just do register, etc. Really limits their value to the store, and of course you can't commensurately lower compensation *too* much or people won't take the job. Plus from their end it's just really boring having a restricted duty set.
I get the feeling that such For Your Own Good safety restrictions were less common in the past/in other countries, which might address part of the gap.
Huh, Are you a fulltime employee? With full time benefits? I work part time at a local tavern/ restaurant. ~$20/ hr. But life is cheaper here. The price of a glass of Guinness went up to$6.25 (from $6.00). Still a deal since you have to leave a tip.
I am. Part-timers are eligible for most of the same benefits I get, but it's harder cause they aren't guaranteed to hit the minimum number of hours to qualify. Especially during our slow season...there have been years when management contemplated cutting their own hours, so my store wouldn't have to transfer or lay anyone off. Better not to over-hire in the first place, to avoid such painful choices.
My kids (19 and 16) have both had part-time jobs in high school, and so have the majority of their friends. Almost all of them are college-bound, but I think that cuts against some theories about this.
I work part time in a local restaurant. It's all part time, the owners can't afford the tax, healthcare costs of a full time worker. Oh, I love working here, and the good bartenders, waitresses make a nice dime.
Does anyone have experience with relationships that have large gaps in intelligence? I've found myself in one and I'm curious what the long-term outcome is likely to be in terms of happiness. My primary concern is whether we will have children whom I can view as intellectual equals and how I will feel if not.
Secondhand; one of the smartest people I know has been happily married to one of the dumbest for longer than I've been alive. They're both people-people who love to talk, he loves to impress her and she loves being impressed. She had one child from a previous marriage and at least from this perspective the three of them have been close and happy the whole time.
My guess is the "people-people" part is the most important.
Depends do you like your partner? Do you respect them? Do you like talking to them? Can you trust their judgment? I know that's a basic question but there are surprisingly a lot of people who don't like their partners. They don't care for their partner opinions or even confide in them. They like what their partners provide for them and that's what keeps them long term together.
You know you probably don't have to like your partner to be happy long term as long as your partner provides what you need from them. Historically marriage was a social and economic arrangement and not a long term happiness arrangement. So what does your partner provide for you and is that fine with you long term?
And with children... if you don't at least respect your partner that will make your life shitty. Especially if you're a guy who does the stereotypical thing of leaving the child rearing entirely to the mother. These men suddenly discover themselves to be aliens in their own household that they pay for. She may not be as smart as you are but I do hope you respect her point of view.
> Does anyone have experience with relationships that have large gaps in intelligence? I've found myself in one
In general people can be happy, but as with any power imbalance, if this is something either or both of the people involved find themselves dwelling on / worrying over to the point of calling for opinions from strangers about, that's honestly a pretty bad sign for the relationship IMO.
Is it necessarily a power imbalance? For example one person could be better looking (probably matters more in my experience than anything else), career/money, age (younger or older depending), charisma, or just temperament. I think people know where they stand in a relationship and low iq people often aren’t dumb
If you want tall children, be with a tall person. If you want smart children, be with a smart person. If you want your kids to have a hereditary trait, your partner should have it. I don't know how any advice you get could help you with your particular concern.
However, I think you're probably overestimating the importance of raw IQ. Many of the very smartest people I know are unhappy and relatively unsuccessful. Presumably your partner has positive traits, otherwise you wouldn't be with them. Do you want your kids to have those specific positive traits as well?
At some point, the marginal utility of an additional unit of attractiveness or emotional resilience exceeds that of intelligence. You might be surprised how low the IQ threshold is for that to be true (especially considering overall life satisfaction vs socially recognized markers of achievement).
Also, your kids are going to be different from you in major ways. Practically speaking, do you need them to be "intellectual equals" so you can have similar views about politics or something? Whatever specific bonding activity you're imagining might never happen anyway. It's dangerous for parents to romanticize future children too much, you shouldn't be too disappointed if they turn out very differently than you imagined.
They won't even be intellectual equals until around eighteen or so, anyway, if he's hoping to have long philosophical conversations on elevated subjects. Unless they are baby geniuses even smarter than him, before that age their opinions are unformed and merely copying those of the adults around them and they don't have life experience to have any interesting and unique experiences to share.
So is he going to ignore the kids until they're old enough to hold up their end of "well it was worth talking to you, congratulations Offspring Number Two, you did not bore me" chats with Pater?
Eh, just get used to being the dumbest guy in the room. Builds character. Let her and the kids handle the hard stuff, you stick to doing internet comments in the middle of the day and don’t burn down the house
I also considered “Feels nice dating someone smarter than you, doesn’t it?”
Why did his comment bother you?
I wasn't bothered at all, I just thought he was serving up a joke on a silver platter and had to make it
I LOLed
Yeah, it clicked with me too. Pretty funny.
It doesn’t bother me, but I think a majority of the population would find it cringe and arrogant. I know this is a safe space for that sort of thing but just saying
The only thing you and your partner can do about this is embryo selection for IQ, and you may already be too old to get much utility out of it.
I foresee a long and happy relationship. Between some other people, unrelated to this thread.
As for you and your partner: one or both of you is going to be miserable. You, because you seem to look down on your partner, or your partner, because they'll feel your contempt.
Besides, as you've probably realized already – what with your superior intellect and all – your children's intelligence won't simply be the interpolation between your's and your partner's intelligence. There's a considerable variance involved, and the at most 1–2 children you're likely to sire are too few to form a nice bell curve around the expected value. It could go either way, even with a similarly condescending partner, so if you want to be with intellectual equals, don't have kids, join Mensa instead.
> if you want to be with intellectual equals, don't have kids, join Mensa instead.
In my experience, most Mensa members are *not* intellectual equals of an average ACX commenter.
Consider base rates. There are way more IQ 130 people in population than IQ 140 or IQ 150 people. So the organization that in theory is "IQ 130 or more" ends up in practice as "IQ 130, only rarely more".
Then consider luck. Let's say that an IQ test gives you +5 or -5 points depending on whether you have a good day or a bad day. Given that there are more IQ 125 than IQ 135 people in the population, most people who pass the Mensa test are actually IQ 125 people who had a good day.
Mensa is a good solution for someone with IQ 125, but if you are the kind of person who feels at home on ACX, you probably won't feel at home in Mensa.
Yes, I've dated a few dumb women and it's always a disaster. We just saw the world fundamentally differently and after a while I would essentially stop talking to them because it was almost impossible to discuss anything even remotely interesting to me. That ends up feeling extremely lonely. Unless you have an unusually strong emotional connection then I think that large IQ gap relationships are doomed to failure.
~1/2 the kids will be like mom, and ~1/2 more like you. It's a crap shoot. (But the only game in town. :^)
According to a very old urban legend, a few of them might be like the milk man.
Have you been tested or you going off identity performance? I used to feel this way but as I’ve gotten older my understanding of intelligence broadened.
Obviously if you don’t see her as an equal it can’t work. I’m just saying people legitimately have value for things besides abstract shape rotating
A lot of my friends and family have noticeable gaps with their partners, there are 2-3 for whom it's probably 20-30 IQ points (including my in-laws), none above that.
I think the best way to think about IQ effects is that they're unusually persistent rather than overwhelming in any given moment.
Like with life outcomes more generally, intelligence isn't a brilliant predictor of compatibility, which probably explains why you're in this relationship to begin with. Loads of other factors are similarly important, but looks, lifestyle preferences, hobbies, interests, even personality are likely to change over time, while an IQ gap will persist for all of your life.
Decent outcomes I've observed from IQ gap relationships:
- You have other compatible traits and get on well together without shared intellectual interests.
- The less smart partner idolises the smarter partner, who likes being admired in this way.
- You develop distinct household roles - both partners spend most of their free time on their own things, but fulfill their household roles adequately. (Sounds like something from an 18th century housekeeping guide, but can work well)
- Good luck with kids being on the smart side
Bad outcomes (often the downside to the same relationship):
- Gradually divergent hobbies - especially if the smarter partner builds intense intellectual interests with a high barrier to entry (e.g. rationality)
- Negative vibes - the smarter partner gets contemptuous and dismissive, the less smart partner gets insecure and resentful
- Bad luck with kids...
Overall, I would probably discourage intelligence gap relationships at the margins. I don't know the counterfactual, but I think 1 or 2 of my smart friends made bad choices here that could have been prevented if they had thought explicitly about this.
For kids, you can actually run the numbers to work out what range of intelligence you'd be comfortable with, and look at the 10th percentile of the expected IQ distribution given your respective IQs. I think the question is more "will I enjoy spending time with this kid?" rather than "will they be my intellectual equal when they're 21?"
When I got on this thread there were 9 posts, and 3 of them were people posting links to their work:
David Gross
"'Civil Disobedience,' a thread by @hdthoreau" presents Thoreau's influential essay as though . . .”
tailcalled
“I recently published Beyond the lexical personality traits: What is . . .’”
‘Clark Gaebel
“I recently wrote about how a retail investor can invest in ‘the whole market, except . . . “
I’ve been bothered for a while by how *many* of these appear on open threads, and wonder whether many others are, or only few. Sure, it only takes a moment to skim one of these and move on, but somehow having to do a lot of that dilutes the experience of browsing the thread, at least it does for me. Also, most people who come here to plug their blog posts are not sticking with the guideline that Scott (I think it was Scott) gave, which was that posts of that nature should include a paragraph, one that is itself of interest, about the piece the writer is plugging. But personally, I don’t think I’d be less bothered by plugs of that nature. For me, these little self-advertisements are like billboards blocking the view of the wooded hills I’m driving through.
I would support a blanket ban on self-promotion. If the article is of interest to the community, somebody else will find it and bring it up organically.
Sounds like wishful thinking. How many articles out there *could* be of interest to *the community*, and how many of them will be brought up organically by some person who happens to not be the author?
I feel it's made more bothersome by the fact that these promotions are usually from people who otherwise don't comment much*, and they don't often try to explicitly connect it to things of direct interest to the ACX commentariat, recent conversations that have come up, etc.
As an example of someone who does self-promotion that I don't mind, Performative Bafflement comes to mind - he writes things on his Substack very much in line with the ACX community's interests and sense of style, comments regularly otherwise, and typically posts a link to one of his posts in response to a conversation in which it is relevant.
*Of course, I myself lurk much more than I comment, but then again I'm not trying to promote my own work here.
Actually, I also don't mind having regular commentators mention posts on their blog, especially as part of discussions where the link's relevant. That's a whole different thing, and seems fine.
PB is great. For those who don’t follow him here’s his most recent post which, I would say, is quintessential everything-he’s-about: https://performativebafflement.substack.com/p/achieving-fiscal-solvency-by-not?r=bbrjn&utm_medium=ios
That feeling when seeing a link from a non-commenter is probably because it comes across as someone trying to boost their own substack _and_ is doing the bare minimum to get there. If you make a "Scott rule" where you first summarize your post in the comment, they'll do that; it'll be exactly one sentence, here, now let's get to the link, the link, it is _imperative_ that you click the link. It's a quintessential bad faith comment, since the idea is supposed to be that you're here to stimulate discussion, and they're actually just trying to pull in traffic.
That's probably why links from long-time commenters don't bother us. (And it would if they stopped otherwise commenting and did nothing _but_ post links.)
I directionally agree, but I would be less bothered if they followed Scott's rule.
In the last few days, Substack released a new feature called "Reply Rules" that makes it possible for creators to explicitly set a rule for replies and then an AI automatically hides replies that do not follow the rule. Using this for making the self-promotion rule actually enforced could be a good application.
I don't know what the currently set rule is, but right now, this AI sometimes hides (in my opinion) innocuous comments. For example, Deiseach's comment about the Hanania review got hidden. This is not a problem for those who download the comments using the API directly, because that still serves the hidden comments, but for those who use the website or the app it might be annoying (it would be for me).
I noticed, I wondered why, then assumed it was because I was snarky about the reviewers. So it's an automatic "let's all be nice" rule by AI? Makes more sense! I did get a little less than gruntled when I saw the company I was keeping amongst the âmes damnées, but I guess if I get down into the gutter, I can't complain about the pigs 😁
https://www.youtube.com/watch?v=nwGf_gembU4
Well, Garbagemouth, I just wrote a post containing the word "piss." https://www.astralcodexten.com/p/open-thread-437/comment/272929316
Waiting to see whether the AI fairy draws a veil of shame over my foul communication to protect the innocent multitudes from thoughts of urination.
Yes, the testing comment got me dragged under the Veil of Isis. So some words are clearly fighting words. Further investigation may be warranted to see if it's terms like "hate" or if it's "poopy-head" what done it 😁
Hard to say what an algorithm finds objectionable. This question broke some rule for ChatGPT:
I watched Manhattan a long time ago and feel like watching it again. Want to clear up one thing for my own comfort. What is the age of consent in New York?
“This content may violate our terms of use or usage policies.”
Mariel Hemingway was playing a 17 year old in a relationship with 42 year old man played by Woody Allen.
I mainly wanted to see it for the gorgeous B&W cinematography of NYC and of course the Gershwin, Rhapsody in Blue.
> an AI automatically hides replies that do not follow the rule
Being censored by a hallucinating AI is a new form of torture.
I can see both sides here. I don't want to see most of the low-quality self-promotion spam. On the other hand, I don't want to get into a situation where saying "X" is allowed but "X, and here's a link to where I make a logical argument for it, consider objections, and provide evidence" is not. You could have a compromise by which:
1. Self-promotion spam from new accounts is not allowed.
2. Anyone who engages in self-promotional linking voluntarily assumes the risk that if Scott or other mods considers it not high-quality enough, that person can receive a temp-ban.
My desired goal state is "if the post would be valuable without the link, then it's okay to link to your site. Otherwise, it is not okay."
This cuts down on drive-by "I wrote a critique of X, here's my blog" but still allows for users who are actively engaging to share a link to further information. My current gold standard is the Fog of War posts, which are great on their own but the previous one included a link to a blog with further material.
Yeah they're solid gold.
Yeah. But I don't think stricter rules are gonna help because timely enforcement will not be there, from what I've seen.
I (somewhat) disagree. Obviously if everyone did this it would eliminate all dialogue but I find enough of the linked essays interesting to justify the clutter. I'm sure Scott would enforce a blanket ban before it got unmanageable.
I came to the thread to self-promote my own work, and then scrolled through the comments and saw yours. I'm sort of interested to know what you would like to see, instead (besides sample paragraphs).
Aren't we all here promoting ourselves?
"Look at me, I have opinions and experiences that are of interest to others!"
The main difference is that some people self-promote in a comment, some link to other things. I'm not saying there's anything wrong with your complaint, but I enjoy seeing a lot of these posts, it feels like I'm getting in on the ground floor of something that might become bigger (like Naval Gazing was pretty great)
>I'm sort of interested to know what you would like to see
If I had my druthers, nobody would link to their own work unless doing so is relevant to a discussion they're participating in. And, well, I'd make an exception for people who have been actively commenting for a long time. I'd be fine with someone like that posting something like "you've been hearing my ideas about the difference between AI minds and human ones for quite a while. Well, I wrote a short story about how the difference might play out -- here's a link."
And by the way, I m now a lot less interested in checking out your Substack because you don't communicate straightforwardly.
>I'm not saying there's anything wrong with your complaint
Yes you are.
>Aren't we all here promoting ourselves? "Look at me, I have opinions and experiences that are of interest to others!"
Ugh. This is so specious: "Hey, so I chopped up a random guy. Aren't we all murderers? Most of us eat meat, after all. And even the vegans are swallowing a few water fleas and rotifers if they drink tap water."
You’re right, sorry. I am being defensive, and my comparison probably wasn't fair. There is a big difference between sharing a thought in a conversation and dropping a billboard for traffic. They're not the same. Your analogy is reasonable: when people treat an open thread like free ad space, it dilutes the community feel.
The reason I defend these links is that I view these comment sections as a decentralized thought-leader (? blogger? Internet celebrity?) incubator. Some of the best writers online started as nobodies dropping links in the wild, hoping to find an audience that cares about niche topics.
When someone links their own work, I see then making themselves vulnerable. It’s someone saying, "I spent twenty hours building this, and I'm brave enough to show it to the smartest corner of the internet."
Scott's paragraph rule is a great filter for lazy spam. But if we say "no self-links!", I think we'll choke off the supply of independent writers before they get off the ground. (Or at least contribute to it) These threads are one of the last places left for high-quality, un-algorithmic discovery. I really enjoy them for that.
I don't expect you to love the billboards, but I hope seeing them as a pipeline for finding new, hidden talent makes it easier to tolerate.
Every post somebody puts up already contains a self-link to their blog, if they have one. If somebody tends to make comments that get my attention I pretty much always check out their blog and have found several that I now follow that way. So I don't think there's much danger that the new crop of writers will die off unappreciated if they don't come here just to spill some seed on the ground. Sorry, that was crass. I should have said, "if they don't come here just to piss on our fire hydrants."
Not everybody is self-promoting. Some people have questions they want to put to this community; others want to share something they found (created by someone else). That's meaningfully different from dropping in just to say "I wrote about this thing [link] kthxbai"
>Aren't we all here promoting ourselves?
I'm mostly arguing with people for my personal amusement. Or making jokes. For my personal amusement.
If your work was good enough for me to read someone else would link it. There are a tiny number of good blogs out there; Scott's number of about one good new blogger per year is about right. The rest is just clutter.
The only people I read have either heavily invested in being interesting enough to read more (I think Performative Bafflement has already been mentioned) or are self-evidently interesting and accomplished individuals (Cate Hall, Hannah Ritchie).
I'm never going to click on a rando's link, life is too short and average quality too low.
I'm someone who reads all the time but rarely comments.
If I write an effortpost, however, I will want to share it with all y'all.
I mean, I wrote a review of that new "Backrooms" movie and it never occurred to me to advertise it here because I am sure that all y'all have other reviewers that you prefer to read and it wouldn't be cool of me to say "here, I saw a movie" because you're not aware of my tastes and prejudices and I'm not going to try to socialize myself now at this point.
But when I wrote an Election Autopsy following the DNC's abortion of one, I thought "these dudes will be interested in my autopsy because my conclusion aligns with the SSC worldview, kinda".
If I need to show up and kvetch about the culture war, talk about my sexting with AI (it's purely for research purposes!), or discuss how you seriously need to start using a slow cooker as part of your weekly meal prep, I'm not sure that I'll be able to do that organically enough to get folks to get you (I mean you, specifically) to say "okay, it's fair that he linked to his essay".
My essay either "fits" here at ACX or it doesn't.
And, if it doesn't, I expect that it'll either get ignored or someone will tell me that my essay sucks.
And now I will go back to lurking.
The problem is, some people are better at taking feedback than others. I think your posts are perfectly okay, and the evidence is if your post starts a debate (that isn't pure criticism). But some people just keep posting links even if everyone ignores them. And of course it is a possible to make a mistake and assume that something will be more popular than it actually is...
> My essay either "fits" here at ACX or it doesn't. And, if it doesn't, I expect that it'll either get ignored or someone will tell me that my essay sucks.
I see no problem with this approach. The problem is e.g. if three essays in a row get ignored and the person keeps posting.
As someone who has promoted some of my own articles in Open Threads in the past (while following the introductory paragraph guideline), I'd welcome more guidance from Scott on this matter, or some kind of formal vote.
I'm happy to abide by a formal rule, but don't want to refrain from all self-promotion solely on the basis that it annoys some people without community consensus. I do believe any articles I post here are of interest to the ACX community. Sometimes I start out writing a comment for the Open Thread, before realizing it is so long it is best made into a separate article.
Andrew Gelman has a post comparing Scott to Edmund Wilson.
"Scott Alexander is, like Wilson and those others, a prolific writer with strong political opinions, who feels comfortable opining about–and gets respect for his opinions on–all sorts of things. Like Wilson, he’s an autodidact and a bit of an outsider. Alexander’s topics are different from Wilson’s–more on science, less on literature–but he has the same feeling of well-earned authority. And a fun writing style. Bloggy rather than magaziney–times have changed–but a distinctive and readable style in any case."
https://statmodeling.stat.columbia.edu/2026/06/07/scott-alexander-as-a-modern-day-edmund-wilson/
Wrote a book summary for Captured Futures: https://existentialcrunch.substack.com/p/how-environmental-politics-lost-the
The book is about our collective inability to envision futures that are fundamentally different to today, how we ended up like this and that we need to get better at this again if we want to have a chance for the challenges ahead.
> due to kamikaze
Do you have any references for that? My understanding was that the numbers involved were not very large.
Europe was tucked in the fifities, but recovering nicely in the sixties. Remember swinging london?
I came out of that word salad not really knowing what the future world would look like, but certain it would be poorer. The obsession with reducing consumption - which is basically code for making the average guy poorer - is a dead end. And it won’t work.
For two reasons, clearly you are targeting the West. Yet all future carbon increases will be in developing countries. Europe and other advanced countries are reducing emissions, in fact the EU peaked in 1990.
The US is different but in no way as culpable as China and others in terms of what’s been pumped out since 1990, which in China’s case is catching up with Europe’s entire emissions since 1850.
The other problem is if you want no technological fixes, no carbon extraction, and only reduction in consumption then consumption or emitting activities - which includes agriculture- has to drop to near zero. which probably means mass starvation and people freezing to death.
And since there’s no extraction allowed in your scenario the carbon stays in the atmosphere. Temperatures remain elevated for a very long time even after emissions reach zero, the seas keep rising.
Kind of a long shot, but if anyone here works at Meta: There used to be this VR game called "Echo Arena", which was legitimately one of the most fun games I've ever played. If you ever read Ender's Game, it basically replicated the zero-gravity battle rooms (mostly via the paid expansion called Echo Combat). The zero gravity effect actually works quite well in VR, and can feel very immersive after a while.
Anyway, the publisher, Ready at Dawn, was acquired by Meta and then shut down, and Echo Arena (a multiplayer game) was also shut down. With the Steam Frame soon to be released, I would really like to play this game again. I do think it would be very popular if released on Steam. If this sounds interesting to you, and you know someone, maybe we could convince Meta to give this game another shot?
I worked at Meta in the VR division at the time Echo Arena was shutdown (not there anymore). IIRC, this was a minor point of controversy that came up in one of Boz's weekly Q&As.
All I can say is good luck. Meta clearly didn't think it was worth the cost even back in 2023 when the Metaverse was a top company priority. Why on earth would you think they would be more inclined to invest in it **now**?!
Also, why on earth would you think they would be interested in releasing on a competitor's platform? Meta didn't get into the game publishing business for shits and giggles. It was meant to be a stepping stone on the path to widespread adoption of (Meta's) VR. When that didn't happen, they gave up and canceled everything. They certainly aren't an ordinary game publisher who might be interested in scrabbling for pennies in a cutthroat industry.
If you really really believed in this, you could try to negotiate with Meta to buy the IP. But I doubt you'd be able to offer enough to make it worth their while.
First of all, thank you so much for responding. It's awesome to hear from someone who was there at the time. I pretty much suspected everything you've laid out here... To be honest, this was pretty much a hail mary.
I know some pretty high-up tech people read this blog, so in my ideal world, somehow this would get to Mark Zuckerberg, who would be like "Yeah, I guess that game does sound cool, maybe we should re-release it". Just like those people who get tech support by tweeting the CEOs on twitter. Of course, this would never actually happen. The team doesn't exist anymore, and the company doesn't care.
Maybe if I win the lottery one day I will reach out to Meta and buy the IP :P.
I feel like you have a better chance of convincing people to make a knockoff version of it. In fact, I wouldn’t be surprised if people were already working on it given how popular a VR title it was and how upset a lot of people were that it was being shutdown
Looks like there were community versions still running? https://www.reddit.com/r/EchoArena/comments/1on7vt6/yes_you_can_play_echo/ That's from seven months ago though.
This Youtube channel's got Thank You comments from ten days ago. https://www.youtube.com/watch?v=nGcDXHAYc2s But no promises they aren't all AI. I haven't watched the video, this is something you must do alone.
Does anyone have a pointer to examples of top priority issues / language that should go into a state party platform regarding AI?
I have been having some conversations with AI about how some very different movies can use the same basic cinematic grammar of shots, and yet establish a very different mood.
For example, _Saturday Night Fever_ (1977, John Badham) versus _Stalker_ (1979, Andrei Tarkovsky)
Simplifying the shots slightly:
EXT. DAY. Bridge in Brooklyn
(establishing shot)
CUT TO:
EXT. DAY. Subway train.
CUT TO:
EXT. DAY. Street in Brooklyn.
(Start music: Staying Alive by the Bee Gees; Close up on Tony Manero’s feet; pan up to Tony Manero’s face)
CUT TO:
INT. DAY. Pizzeria
Versus…
EXT. DAY. Railway yard.
Three guys sneak onto a rail car.
CUT TO:
EXT. DAY. Close up on each of the guy’s faces as the railcar slowly moves down the track.
CUT TO:
EXT. DAY. Telegraph poles
(Switch from black and white to colour film. Very slow pan across a jumble of broken telegraph poles and tangled wire as the characters enter the nuclear desolation of the Zone)
You can argue that the speed of the camera movements reflect both the assumed viewers knowledge and the character’s uncertainty…
Saturday Night Fever. We know where Brooklyn is, we know what a pizza shop is, we can guess what kind of guy Tony is. Tony swaggers confidently down a street in Brooklyn. Fast pacing.
Stalker. We are only given hints as to where this is taking place. It looks like somewhere the Soviet Union. We can see there are guards to stop the protagonist getting in to the zone, for some reason we aren’t yet told. The zone is clearly abandoned — the telegraph poles are broken and no-one has fixed them, but we aren’t told why no-one has dared to go into the zone to fix them. The characters move cautiously — even though the viewer hasn’t been told, they know that they walking into a labyrinth of deadly radioactive contamination and shattered broken buildings. They are moving very carefully. The camera pans slowly. [_Stalker_ was made before the Chernobyl disaster; in retrospect, we clue in faster as to what we are seeing. ]
Note that neither Stalker, nor Roadside Picnic - the book the film was inspired by - have anything to do with nuclear disasters. The book is explicitly about people exploring the site of an alien visit looking for weird alien trash to sell; the film removes from this all explicit genre fiction elements and leaves things ambiguous because what Tarkovsky actually wanted to make films about was inner journeys to faith and spirituality, and weird shit you can see and replicate is just physics.
There is a much more recent computer game which tried to hitch a free ride on the fame of these while explicitly being a first person shooter set in Chernobyl, but it would be a mistake to back-project its storyline and themes, such as they are, onto the earlier works, and we do not seem to be discussing the game here.
Yes, in the book it’s aliens and in the movie it’s left ambiguous. Also, in the book it’s not even radioactivity but something that humans don’t understand, so my summary is a bit wrong.
I think the religious element is actually there in _Roadside Picnic_, just way more ambiguous. In the book, some of the characters talk about whether the aliens are malicious, benign, or just don’t care about us. This is kind of standing in for a debate about whether there is a God, and the characters in the book version just aren’t sure whether there is a God substitute or not. The movie leans more strongly on, yes there is. The Stanislaw Lem essay on Roadsides Picnic argues that the book version has evidence the aliens are actually malicious.
But yes, good catch, my summary was too sloppy.
What I was trying to say (but kind of fumbled it): the movie was made before Chernobyl, so those scenes have connotations for the viewer now that they didnt when the movie was made.
Do they, though, if the viewer hasn't been primed for that by the game or its adverts or people talking about it? The thing with throwing the bolts doesn't exactly scream "radiation" to me.
P.S. Amusing coincidence. I picked Saturday Night Fever as an example of this sequence of shots, before looking up the Wikipedia page for “Establishing shot”. Now I’ve looked at that wiki, it picks exactly the same movie as its example for this type of shot.
But you can start with a wide shot, cut to a shot of a train, push in to a close up on the main character, then cut to the next location, in very different ways,
This could turn into an essay on how Tarkovsky does not follow the rules of the continuity style of film editing, and that, for example, while the switch from sepia to colour in Stalker might look like a reference to The Wizard of Oz, The Wizard of Oz is using colour to let the viewer know whether they are in Oz or back in Kansas, while Stalker is creating ambiguity about what’s in The Zone and what isn’t.
I made a 2x4 voice paraphonic synthesizer:
https://stefanlenoach.com/paraphonic
Why do I put two identical synths next to each other?
The idea is to use the arpeggiators to explore how different chords sound over each other. Try having one synth play a Cmajor chord, then use the other to explore going up and down the Cmajor scalein triads. Try various odd number chords (3,5,7 notes) to have each note rotate over the 4 oscillator voices.
Click 'preset' to hear Cmaj on both synths with each oscillator voice randomized
Click 'demo' to hear a short piece I made on the synth
Enjoy, any feedback welcome
Might be a hit in the EDM community. Me, when I’m fiddling around adding notes onto the top of triads to make weird chords, I like to just stick my spare fingers somewhere on the neck and see how it sounds.
What if we made an effort to protect AI modules from the complexities of the human world, through giving AI special economic zones carefully encapsulated away from how we think, communicate, and navigate the physical world? It's more common to worry about sealing away AIs to keep them from doing damage, but the *reverse* direction (protecting them from us) can help reduce reliance on opaque learned heuristics and open the door to formal verification and other methods that can provide high assurance about principled systems.
I've written more about this idea in a blog post: https://stng.substack.com/p/abstraction-boundaries-and-bubbles
>through giving AI special economic zones
Until the President's plane goes down inside one of them, and the government turns to the one man both talented and expendable enough to save him.
"Escape From New AIork."
To start with one irony: "carefully encapsulated away" is most dramatically attempted today, not in AI, but in quantum computation. Avoiding decoherence from interactions with the environment is one of the main challenges in that field.
Returning to AI: I'm not following what would typical decisions being made by AI internal to the zone be? It sounds like your description of the AI economic zone is closer to hard-coded automation than to the recent progress in AI. I'm not seeing why, from a _computational_ point of view, the zone would be a "sweet spot" _between_ hard-coded automation and operation of AIs in the normal economic system. There _is_ certainly a point in having _physical_ areas isolated from humans for physical and chemical reasons - heat, radiation, high voltage, toxic materials, etc., but that's all long-standing practice, e.g. putting fencing around substations so passers-by don't touch high voltage wiring.
But, _computationally_ , LLMs have largely solved the problem of interfacing with natural language - expecting reasonable defaults, using context to resolve ambiguities, etc. I don't see subsystems inside an enclosed zone interacting through programming languages instead of natural languages as a clear win. The sort of things that go wrong today, e.g. with AI agents, perhaps e.g. not anticipating the possibility of some failure mode (e.g. of a machine jamming) and not having a detection and recovery process thought through, don't seem like they are ameliorated by internal communications in a programming language.
I can see the point of restricted _physical_ spaces, like factory floors, together with richer world models to improve intelligently dealing with a larger fraction of events possible within those spaces, but this isn't really restricting the _informational_ interfaces.
I'm not arguing for hard-coded automation. I'm arguing for something more like hard-coded rules of engagement, hard-coded specifications of the goals of an autonomous region (e.g. "make reliable XYZ widgets according to this schematic") -- and then the region can go through recursive self-improvement to get better and better at achieving its objective set in advance. Internally, all details of organization are chosen alongside the design of the intelligent agents that function therein. One of the examples I like to use [1] is avoiding the use of road networks like we have in the world today, full of not just of squishy evolved intelligences (us!) making driving decisions but also of stray animals, weather, and so on. Carefully controlled environments, occupied by autonomous vehicles, avoid those hard-to-model complexities, fundamentally decreasing the cost of effective intelligence.
I also disagree that LLMs have solved the natural-language-interface problem. For the sake of argument, imagine that LLMs have achieved optimal understanding of natural language. There is still much inherent ambiguity that can cause a lot of trouble [2]. Now if assume LLMs have mastered that ambiguity and understanding all of the social-context subtleties needed to intuit "what people really mean." We still face the inherently high cost of implementing deep learning [3]. There can be such an advantage, from switching to communication designed for efficient processing, that much more effective decision-making can be executed in a given time, budget of money or other resources, etc. Organizations that are able to use decision-making that is simultaneously cheaper and more reliable can be expected to win out in time.
[1] https://stng.substack.com/p/hardware-software-codesign-and-autonomous
[2] https://stng.substack.com/p/what-makes-language-processing-hard
[3] https://stng.substack.com/p/why-deep-often-means-slow
Many Thanks!
"A subway network is a good example, with physical safeguards to keep people from running into the tunnels."
Yes, and construction sites also generally have physical barriers to prevent unexpected people from wandering into them. As I agreed with you before:
>I can see the point of restricted physical spaces, like factory floors, together with richer world models to improve intelligently dealing with a larger fraction of events possible within those spaces, but this isn't really restricting the _informational_ interfaces.
I fundamentally don't believe "Yet switching to other modes of communication dramatically simplifies processing." I think it helps _slightly_ (hey, if something can be parsed by a regular expression, great!), but I think it misses the main limitations to broadening AI applications today and in the near future.
>In such a setting, the mega-factory may even have autonomy to build new factories that follow the common rules. It may invent new decomposition of top-level deliverables into components, design and construct new factories to supply those components, and still remain compatible with its exterior interface.
I think that this forces a sufficiently comprehensive world model that
a) I don't think that moving from natural language to a programming language solves much. As I commented in https://www.astralcodexten.com/p/open-thread-436/comment/272938989 in another thread, building e.g. a shed requires
>3D world models interfacing to vision ("Do I have the right part? Are the surrounding parts properly in place? Do they match the blueprints/CAD data to proper tolerances? Can I position the next part without collisions?") and to force sensing and application.
These are substantial tasks! But they are much the same task whether the internal communication is done with natural language or a formal language.
b) I don't think formal verification will be feasible. Even 1-D bin-packing is NP-hard! _Provably_ ensuring that some design process is going to yield results consistent with a (typically complex!) specification looks very unlikely. I don't think a mega-factory that builds new factories will get away from heuristics and recovery procedures.
c) While I do agree that the signaling function of natural language is counterproductive to getting tasks done, I don't think that this is a huge problem where AI agents (1) can include a modest amount of redundant language ("to put it another way") (2) can send messages back and forth to resolve ambiguities (3) are evaluated on whether they successfully complete tasks, which miscommunication impedes
d) On a peripheral note, I don't think that trying to eliminate senses in a system operating in the physical world is a good choice. I think that usually the better choice is to _add_ senses, and e.g. disambiguate an uncertainty in the interpretation of a visual scene by e.g. adding radar
e) Very generally, I think that the right emphasis at this point in AI's development is to seek to expand capabilities, including the capability for more reliable results, typically by spending _more_ computation, rather than looking for ways to cut computation. We aren't at AGI yet, and I think that spending computation to fill in the "holes" in AI 'spiky' capabilities is the main near term task, rather than e.g. trying to transition away from natural language towards formal languages in the interests of reducing the computation spent on parsing today.
Thanks for the thorough explanation of where our positions diverge. To some extent, we'll have to wait and see empirically which approaches win out.
It seems like our biggest disagreement is that I am betting on the potential to introduce new abstractions that simplify reasoning. We depend critically on such abstractions today. For instance, few planners of factories (or much of anything else in the world) consider quantum effects directly. You seem to be arguing that no new broad abstractions are coming to simplify reasoning, and you even argue for increasing sensory capabilities to make abstractions *more* complex (if I'm reading you right).
I suppose I would be surprised if human intelligence has achieved a series of triumphs in new abstraction to simplify planning, and then superintelligence has no more for us! But it could happen, and it's conceivable any new abstractions are even of less importance than the past big ones.
I may be biased by my professional work on programming languages and formal verification. In that world, it is usually a triumph (not a barrier) to notice that a problem is NP-complete, rather than undecidable! It is also often quite tractable to prove than an algorithm solves an NP-complete problem, leaving the cost for runtime -- and then we achieve good runtime performance nonetheless, a surprising fraction of the time.
Many Thanks! 'scuse the delay
Re:
>it is usually a triumph (not a barrier) to notice that a problem is NP-complete, rather than undecidable!
True - I'm approaching this from a computer-aided-design class of problems; in a theoretical computer science class of problems (e.g. halting), very much agreed that undecidable is a frequent outcome, and NP-complete is better!
>It is also often quite tractable to prove than an algorithm solves an NP-complete problem, leaving the cost for runtime -- and then we achieve good runtime performance nonetheless, a surprising fraction of the time.
What typically happens in NP-complete computer-aided-design problems is that the runtime for finding a provably-optimal result is usually exponential, but someone is able to find a heuristic with reasonable runtime which yields a result which is "good enough".
>It seems like our biggest disagreement is that I am betting on the potential to introduce new abstractions that simplify reasoning.
Actually, I don't disagree on that, but just on the value of going to formal languages to reduce parsing time. Two examples of routinely introducing new abstractions in natural language that I'd expect to also apply to AI systems:
a) Just naming subassemblies/products/classes of them. "model np750xhd samsung laptop" is a mildly leaky abstraction
b) The nomenclature that mathematics constructs that typically depend on theorems that break objects into classes (property X turns out to imply Y and Z, so distinguishing objects with and without X now merit special names)
I see this as adding vocabulary ("This wire is at 'logic 0'") rather than syntax.
This _does_ blur the distinction between human-readable language and opaque language. When cooperating AIs introduce new vocabulary, it will become harder and harder for humans to understand it.
Got it; so your comments are about language in particular, as an abstraction that may or may not be worth improving on substantially.
I think "parsing" undersells the full activity of language processing. For instance, it doesn't traditionally include resolving words into semantic concepts. My argument about language includes that full stack of processing to guide decision-making. To me, it seems quite inefficient and worthy of reengineering.
Regarding NP-complete problems, my personal experience is more on the side of formal verification than computed-aided design. In the formal-verification world, we typically solve logical problems exactly, and the "catch" is that solvers may consume so many computational resources that, for all practical purposes, we consider them to have given up. Practitioners learn the (often-undocumented) behaviors of verification tools, to predict what sends them off the deep end. I think there's a lot of potential to attack this problem with the full intensity of performance engineering that has gone into GPUs and their associated programming tools. I'll get around to writing about those new approaches eventually.
Fog o' War Stuff (7 June 2026)
My second standalone Fog o' War post is up on Substack (link below).
In it, I discuss the letter Zelenskyy sent to Putin and Putin's testy response during a panel discussion at the St Petersburg International Economic Forum. (1.) Zelenskyy greeted the guests arriving at the forum with a drone strike on a nearby St Petersburg oil facility and the Kronstadt naval base. (5.)
And what was Russian air defense doing to stop the attack? At least a dozen videos of the attack have been posted on Telegram and TikTok, and I don't hear any anti-aircraft batteries firing in any of them. The sum total of St Petersburg air defenses seems to have been one or two soldiers with automatic carbines trying to hit them. But Russian state media said nearly two hundred were shot down. Excellent marksmanship on the soldiers' parts! (2.)
We also learn that Putin intentionally aimed a $50-million Oreshnik ballistic missile at the garage cooperative in Bila Tserkva in the Kyiv region. (Poor Uncle Sasha! His days of tippling vodka in the garage are over. And what happened to Obama? Did he escape the garage unscathed? These are important questions! Why isn't our MSM doing its job?) (4a.)
More stuff on the worsening gasoline situation in occupied territories and Crimea, along with the mandatory pictures of long gasoline lines of burning oil facilities. (9.) (10.) But seriously, there are 2.4 million people in Crimea (and I don’t know how many are in the occupied territories). Now that gasoline is impossible to get, what's going to happen to food deliveries and deliveries of medical supplies to hospitals? Call me a nervous nelly, but I potentially see a humanitarian disaster brewing in Crimea. (11.)
Also, I discuss Ukraine's Morrigan drone, which seems to be the prime instigator of the havoc being wreaked on the highways and byways in the occupied territories. (14.) And I query Gemini and ChatGPT to find out more about the likely configurations of the AI systems being used on Ukraine's latest generation of mid-range drones. (15.)
https://beowulf888.substack.com/p/fog-o-war-stuff-7-june-2026
Thanks for these
Thanks Beowulf, curious what you’ll think of the latest from Iran
I'll try to put something together this week.
This is vague, but could we have an axial age coming up?
The axial age is a theory that there were major new religions getting started at very roughly the same time because old solutions weren't working.
https://en.wikipedia.org/wiki/Axial_Age
"Axial Age (also Axis Age,[1] from the German Achsenzeit) is a term coined in 1949 by the German philosopher Karl Jaspers. It refers to broad changes in religious and philosophical thought that occurred in a variety of locations from about the 8th to the 3rd century BCE.
According to Jaspers, during this period, universalizing modes of thought appeared in Persia, India, China, the Levant, and the Greco-Roman world, in a striking parallel development, without any obvious admixture between these disparate cultures. Jaspers identified key thinkers from this age who had a profound influence on future philosophies and religions and pinpointed characteristics common to each area from which those thinkers emerged."
It looks to me like the left and right have lost their minds in a way that the center can't resist. We're an inventive species (if you say "unfortunately", I will bite you) and we could use some sort of survival-prone system we haven't come up with.
I can see two ways to go with this.
The usual thing I prefer to do with a prediction is to see if I can rule it out. You're seeing signs of an impending event; could they be signs for something else? What are the signs of something other than an axial age? Are we seeing those too?
The other thing to do with a prediction is to enjoy it and lean into it. What are the religions you anticipate? Do you see them in terms of the underlying weltanschauung and philosophy, or in terms of the trappings and rituals?
I'm not as sure that there will be new religions as that something is going to happen to the idea of nations. If there are a lot of climate refugees, borders are going to be much less enforcable.
If there are new religions, they will accrete customs, ceremonies, art, and music.
>It looks to me like the left and right have lost their minds in a way that the center can't resist.
Agreed.
My 0-th order interpretation is that both factions have two broad choices for how they manage their own members:
a) They can police their members to exclude/disempower/condemn crazier members/ideas
b) They can police their members to exclude/disempower/condemn _more moderate_ members/ideas, to follow 'purity spirals'
My impression is that both MAGA and Woke are doing a lot of (b) and not much of (a).
I don't know why (social media?). Whatever the cause, if my impression of a shift from (a) to (b) is correct, it isn't healthy.
That's outside view (outside of spirituality I mean). Inside view is that Muhammad was indeed the last founder of a world religion we're getting. The world is tending towards dissolution, and finally, its end, to be followed by the next world and the next humanity, which will be born from the seeds of this world.
Before then, we're getting Antichrist, which will be a figure presenting themself as a restorer of spirituality, but which will be selling a parody of spirituality instead. They will be mistaken by many to be the founder of a new world religion.
Spirituality is never truly lost, but it does take other shapes than a Messiah with their big revelation.
One of those perplexing thought experiments again.
Would you rather fight a chicken every time you got in your car or use a sword to fight a gorilla once a year?
Attach whatever color button you care for to the choices, purple, magenta, cyan, puce, taupe, egg shell white or kelly green. It doesn’t matter what color the buttons are.
Afraid it’s one or the other though.
What are the conditions for victory in the gorilla sword fight? Is this a battle until surrender or death?
A TKO would be just fine. Gorilla’s corner throws in the towel, maybe. Ahead on points at the bell even.
I mean, there is a huge difference between getting a gorilla to back off, versus fighting a gorilla that's cornered.
In the hypothetical, is this chicken actually capable of movement, or has the Save Our Bacon act passed?
It’s an unimpeded, robust chicken at fighting weight.
Free, lean, free-range chicken meat for life. I'm in!
Yeah, this just sounds like free chickens as long as I'm willing to kill them by hand.
I'm also think that giving us a sword to fight the gorilla was seen a buff but "use a sword to fight a gorilla" sounds like we have to use the sword instead of a better weapon, like a Ford F-150.
Can I sign up for the boneless chicken?
Define "getting in the car". If I put three armfuls of groceries in the trunk and then drive home, is that one cockfight, or four? If I have a passenger, do they have their own chicken, or do we share one?
Do you mean a proper trunk or a hatchback? It’s one chicken per bout.
I've got a car with a trunk and another with a hatchback, so need to know for both. Plus I use the backseats for groceries a lot.
Still, with one chicken per group it sounds like it's easiest to take the chicken option and try to do more cockpooling.
...do I still get to keep the sword?
With the trunk, no issue. If the chicken attack while you open the hatchback, she is issued a technical foul. 3 technical fouls and she gets a red card and you win.
Yes, if you fight the gorilla you get to keep the sword.
You have to treat the sword with the dignity that it deserves though. Can't use it to split firewood or anything.
Oh man, technical fowls expose new dangers; are the chickens limited to fisticuffs, or can they use weapons? Am I going to have to contend with a chicken with a chain whip?
No weapons beyond what nature provides. Any pecking or clawing counts as one technical per incident. Mano a beako as they say.
Should be both of us (chicken versus human) with the weapons Nature endowed us with. Giving us a sword against a gorilla is already tilting things slightly, but that is to make up against the disadvantage we are at when facing a gorilla. Chickens are vicious monsters that don't need any help such as giving them weapons.
Europoors without cars keep winning!
Jeez, I used to live on a commune where our 50 or so chickens wandered the property, and the worst they ever did to me was peck the grommets in my sneakers. They did do that quite a lot, though. Chickens really like grommets, I think. I myself am fond of the word grommet. But, um, I guess I'm not really engaging with the meat of your question. Sry.
Grommet is up there with aglet in my book. Good XWord answers.
You didn't know the insane cock (er, can I use the Irish/British word or should I replace it with rooster?) that lived in a neighbour's farmyard. That bird was absolutely insane and would attack anyone or anything that it perceived as trespassing, without provocation, without consideration for its own or any other creature's life.
To this day, I maintain that it was only thanks to the good stout corduroy pants I was wearing upon the occasion when my mother brought me on a visit that saved me from having scars on the backs of my legs for life when this lunatic fowl zoomed across the width of the yard and slammed into me, claws and wings going, without fear or favour. Luckily, even at nine years of age, I was bigger than a cock (that sentence sounds *so much worse* in American English) so no damage done, but I fear to think if I'd been wearing a skirt instead.
A reenactment friend tells Magyar stories. They make Russian tales seem cheerful and Grimm parents seem modern and enlightened. One of them concerns a "vampire" who'd take the form of a gigantic black rooster.
The first few times she told it she used the Other phrase represented by the initials BBC and had to be told why her audience was in hysterics.
...I do like the idea of a vampire who can take the form of the entire British Broadcasting Corporation.
Prescient indeed of a band of Steppes Nomads to predict BBC in Either familiar context.
I'd say BBCs qualify as a Jungian archetype
> *so much worse* in American English
There are a couple vulgarisms for female genitalia that pass without a blink when uttered by Jason Statham but would give an American the vapors if Tom Hanks ever used them.
I know chickens, I've fought (sort of) chickens before (as a kid), bring on the chickens!
Gorilla has a sword too or not? Anyway, that's too much of a challenge. I can take a chicken, vicious little dinosaur or not. Beware the beak and talons, hobble the wings, twist the neck. Chicken dinner tonight!
Claude went with the gorilla. I said “but chickens are puny and dumb.”
Claude responded:
“But consider — puny and dumb cuts both ways in a car. A panicked, flapping, squawking bird in an enclosed metal box is chaotic precisely because it’s not thinking. It’s just pure instinct and claws everywhere. Have you ever seen a chicken freak out? They’re basically a feathered blender. Your eyes and face are right there.”
Never seen a chicken freak out myself. I did see one in a South Dakota tourist trap that for the price of 25¢ would push a button to swing a small baseball bat and hit a rolling pinball to the fences and run around little base pads for a home run.
> Never seen a chicken freak out myself.
I assure you, neither has Claude.
I know. I am a bit tickled by the ‘feathered blender’ turn of phrase though. Not bad for a stochastic parrot
a stochastic parrot describing a chaotic chicken
On a second thought, perhaps we should trust Claude's expertise here.
Ah, I was assuming you fought the chicken before you got into the car. Claude may be fake, but it's correct on the dangers of chickens. Do not get into an enclosed space with one, which is why it is important to strike before it can corner you.
Definitely chicken.
First of all because the language of the question is more exploitable. Can I just gift my car to my wife and put the title in her name then borrow it every day? Can I generate infinite chickens to sell for profit by getting in and out of my car over and over?
But second because fighting a chicken is easy - wear good jeans,kick it once when you get to the door of the car, get in the car and close the door before it gets back to you. Plenty of people who own chickens probably have to do this with aggressive ones every time they feed them, not a big deal.
Oh definitely a chicken. It matters a lot whether this is a hen or a rooster or a cockerel or a chick, though. Not for whether a gorilla is worse, but for how annoying or trivial this is going to be. I already have to fight a hen every day because she's broody and she attacks me every time I offer her treats and it's no big deal, just a light "go away" tap with her beak. Cockerels are full of ego and think they could take on the giant 6-foot rooster, but they have no spurs and no experience, and confidence makes them easy to grapple. Roosters are the one I would least want to have to deal with, but even then, it's less than two feet tall. Some roosters are, well, kind of a chicken, so if running away counts then it's really easy.
I never, ever, want to have to fight a gorilla with a sword. Gorillas are stronger than you and almost no one knows how to use swords these days. You could make the gorilla fight be once per lifetime instead of once per year, and the chicken would still be the winner. Seriously, if you think you could take on a gorilla, try fighting a pork butt with a chef's knife first.
I infer from the question i get to decide what to bring with me on my way to the car. I.e. thick pair of gloves. If i have to fight that thing bare handed everytime i think i need to invest in sword fighting classes (not from those guys in central park though, surely some one can teach me how to maim a gorilla properly)
I'm not sure a sword is the right weapon for that. A wounded gorilla who decides "screw that, I'm going to kill you anyway" seems plausible, and it's probably hard to reliably pierce its brain or sever the spine. A missing limb or a lethal wound to the heart or neck might not be sufficient if you can't get out of its reach fast enough. I've read that hunting spears for boars had crossbars close to the tip because a boar does not care and would push itself on your spear to get to you.
Edit: Also, according to someone, in combat sports there are 4 tiers of challenge, IIRC "practicing against e.g. a dummy < sparring < competitive combat < real fight with intentional serious injuries on the line" and no matter how practiced and comfortable you're with tier n it doesn't really prepare you for tier n+1, not because of different skills required but because of different levels of adrenaline involved. They may be overgeneralizing but by default I would expect that no amount of classes really prepares you for kill or die situations.
From the nature videos Ive seen mammalian species on particular seem to give up a little early, at first blood frequently, even as prey. And i can see how this would be an evolved behavior. A gorilla no doubt could kill me with one arm missing, but I dont think that it would
I meant that the animal being attacked gives up, not the animal doing the attacking. That is whats applicable here, whether the gorilla gives up after a severe slice.
Is the gorilla already angry, i.e is going to charge as soon as the gate opens, or can I sneak up and take out the gorilla before it can react? Even if I charges, I've heard gorillas aren't actually good fighters, and having a sword is certainly a big advantage. So I'd pick gorilla.
Ethically, leaving out the question of which fight is more dangerous, my clear choice is the chicken. I'm already responsible for the deaths of some 2-digit number of chickens each year because I like to eat them. It would feel more wrong to kill a single gorilla than a thousand chickens.
https://www.youtube.com/watch?v=4BFmfV0ZrLQ
This isn't even the fastest gorilla I've seen.
If I pick the gorilla fight, sword won't matter, and I even know how to wield one. I'm dead within a year.
I'll take my chance with the chicken even if I have to wear a cowl and gloves.
It's fascinating that even in 1861, the year this book was published (and which I read many years ago), the existence of the gorilla is so disputed that nobody is sure it even exists outside of tales from the natives, so the three main characters go out to Africa to hunt it:
https://en.wikipedia.org/wiki/The_Gorilla_Hunters
And, unlike the view post the famous David Attenborough meeting, the gorilla of previous times was believed to be a ferocious, violent, murderous beast:
“I’ll tell you what it is,” said he, with the look and tone of a man who feels that he has been unwarrantably misled—“I don’t believe there’s such a beast as a gorilla at all; now, that’s a fact.”
There was something so confident and emphatic in my comrade’s manner that, despite my well-grounded belief on that point, I felt a sinking at the heart. The bare possibility that, after all our trouble and toil and suffering in penetrating thus far towards the land which he is said to inhabit, we should find that there really existed no such creature as the gorilla was too terrible to think upon.
“Peterkin,” said I anxiously, “what do you mean?”
“I mean,” replied he slowly, “that Jack is the only living specimen of the gorilla in Africa.”
“Come, now, I see you are jesting.”
“Am I?” cried Peterkin savagely—“jesting, eh? That means expressing thoughts and opinions which are not to be understood literally. Oh, I would that I were sure that I am jesting! Ralph, it’s my belief, I tell you, that the gorilla is a regular sell—a great, big, unnatural hairy do!”
“But I saw the skeleton of one in London.”
“I don’t care for that. You may have been deceived, humbugged. Perhaps it was a compound of the bones of a buffalo and a chimpanzee.”
“Nay, that were impossible,” said I quickly; “for no one pretending to have any knowledge of natural history and comparative anatomy could be so grossly deceived.”
“What like was the skeleton, Ralph?” inquired Jack, who seemed to be rather amused by our conversation.
“It was nearly as tall as that of a medium-sized man—I should think about five feet seven or eight inches; but the amazing part about it was the immense size and thickness of its bones. Its shoulders were much broader than yours, Jack, and your chest is a mere child’s compared with that of the specimen of the gorilla that I saw. Its legs were very short—much shorter than those of a man; but its arms were tremendous—they were more than a foot longer than yours. In fact, if the brute’s legs were in the same proportion to its body as are those of a man, it would be a giant of ten or eleven feet high. Or, to take another view of it, if you were to take a robust and properly proportioned giant of that height, and cut down his legs until he stood about the height of an ordinary man, that would be a gorilla.”
“I don’t believe it,” cried Peterkin.
“Well, perhaps my simile is not quite so felicitous as—”
“I don’t mean that,” interrupted Peterkin; “I mean that I don’t believe there’s such a brute as a gorilla at all.”
“Why, what has made you so sceptical?” inquired Jack.
“The nonsense that these [natives] have been telling me, through the medium of Mak as an interpreter; that is what has made me sceptical. Only think, they say that a gorilla is so strong that he can lift a man by the nape of the neck clean off the ground with one of his hind feet! Yes, they say he is in the habit of sitting on the lower branches of trees in lonely dark parts of the wood watching for prey, and when a native chances to pass by close enough he puts down his hind foot, seizes the wretched man therewith, lifts him up into the tree, and quietly throttles him. They don’t add whether or not he eats him afterwards, or whether he prefers him boiled or roasted. Now, I don’t believe that.”
“Neither do I,” returned Jack; “nevertheless the fact that these fellows recount such wonderful stories at all, is, to some extent, evidence in favour of their existence: for in such a country as this, where so many wonderful and horrible animals exist, men are not naturally tempted to invent new creatures; it is sufficient to satisfy their craving for the marvellous that they should merely exaggerate what does already exist.”
“Go to, you sophist! if what you say be true, and the gorilla turns out to be only an exaggerated chimpanzee or ring-tailed roarer, does not that come to the same thing as saying that there is no gorilla at all—always, of course, excepting yourself?”
But they do eventually track down and shoot a gorilla:
"Of all the hideous creatures I had ever seen or heard of, none came up in the least degree to this. Apart altogether from its gigantic size, this monster was calculated to strike terror into the hearts of beholders simply by the expression of its visage, which was quite satanic. I could scarcely persuade myself that I was awake. It seemed as if I were gazing on one of those hideous creatures one beholds when oppressed with nightmare.
But we had little time to indulge in contemplation, for the instant the brute beheld us it renewed its terrible roar, and attempted to spring up; but both its legs at once gave way, and it fell with a passionate growl, biting the earth, and twisting and tearing bunches of twigs and leaves in its fury. Suddenly it rushed upon us rapidly by means of its fore legs or arms.
“Look out, Jack!” we cried in alarm.
Jack stood like a rock and deliberately levelled his rifle. Even at this moment of intense excitement I could not help marvelling at the diminutive appearance of my friend when contrasted with the gorilla. In height, indeed, he was of course superior, and would have been so had the gorilla been able to stand erect, but his breadth of shoulder and chest, and his length and size of arm, were strikingly inferior. Just as the monster approached to within three yards of him, Jack sent a ball into its chest, and the king of the African woods fell dead at our feet!
It is impossible to convey in words an idea of the gush of mingled feelings that filled our breasts as we stood beside and gazed at the huge carcass of our victim. Pity at first predominated in my heart, then I felt like an accomplice to a murder, and then an exulting sensation of joy at having obtained a specimen of one of the rarest animals in the world overwhelmed every other feeling.
The size of this animal—and we measured him very carefully—was as follows:—
Height, 5 feet 6 inches; girth of the chest, 4 feet 2 inches; spread of its arms, 7 feet 2 inches. Perhaps the most extraordinary measurement was that of the great thumb of its hind foot, which was 5 and a half inches in circumference. When I looked at this and at the great bunches of hard muscles which composed its brawny chest and arms, I could almost believe in the stories told by the natives of the tremendous feats of strength performed by the gorilla. The body of this brute was covered with grey hair, but the chest was bare and covered with tough skin, and its face was intensely black. I shuddered as I looked upon it, for there was something terribly human-like about it, despite the brutishness of its aspect."
Go for the chicken over the gorilla, every time.
Attenborough turned 100 years old a couple weeks ago. As his century mark approached I was rooting for him to make it. Imagine the changes he was witness to.
And he has the grace to acknowledge his great luck.
From his Wikipedia page:
“Attenborough had a pacemaker fitted in June 2013, as well as a double knee replacement in 2015.[176] In September 2013, on the prospect of retirement, he commented: "If I was earning my money by hewing coal I would be very glad indeed to stop.
But I'm not. I'm swanning round the world looking at the most fabulously interesting things. Such good fortune."
I just watched Prince William give a little speech in his honor on his birthday. The guy (Attenborough, not Prince William) appears to still have a full head of hair. God must really like that guy.
Fighting a chicken regularly when I'm trying to just get somewhere quickly would be intensely annoying, I absolutely hate things like that which add small amounts of friction to ordinary acts multiple times a day. However if you'd raised a child (I never did) maybe fighting the chicken is less onerous than having to get a kid into the car to go anywhere so you wouldn't mind this.
OTOH an epic battle man v gorilla with a sword seems heroic and would provide some good structure to your year, an incentive to stay in shape and train for a few months, seems like it might even improve your life to have to do this.
What if it turns out staying in shape and training isn't enough to get you literally killed by that gorilla?
I don't know, I think that's a question for your minister
The Graham Platner discourse is funny. These intellectuals imagine the bottom of society as this oppressed proletariat who work 60 hours a week under the thumb of The Man. And then the intellectuals see Graham Platner, who relies on “disability” benefits and financial support from his parents and are like, “he’s not REAL WERKING CLASS.” You can argue all day about what words mean, certainly it makes logical sense to say that you can only claim to be working-class if you actually work.* But if you look at low-class/low-status/low-income Americans, many of them are a lot like Graham Platner. They don’t do much work, they collect government benefits, (often fraudulently) and they leech off wealthier family members. People see Platner and he reminds them a lot of their neighbor. This is perhaps why he looks very likely to win the Democratic primary tomorrow, the “he isn’t working-class” attacks have not landed. If we don’t want to be governed by people like him, we should come out and say that we don’t want any “working-class” representatives and it’s fine if we’re governed by smart people who made something out of their lives.
*Platner’s “oyster farm” makes hardly any money and its only client is his mom’s restaurant. If you wonder how this clown became a contender for office, it started when a company wanted to open a commercial salmon farm. Platner made a video opposing it, which was noticed by political operatives scouting for a “working-class” candidate. He wasn’t content to just fail at aqua-farming, he had to make sure a company with a track record of creating value and giving people jobs was not allowed to try.
Sorry, reminded me of this https://www.youtube.com/watch?v=UuHCKs47Ntk
God be good to you, Alexander, you've managed to work Platner into yet another "and this is why the right wing is wrong about the underclass" post. You definitely have a talent, I must admit that.
I'm mainly talking about the left here.
The real reason why Platner won is because the Democrats failed to recruit any quality candidates for senate in Maine. Everyone was too scared of Collins and piled into the Governor race instead. It's a real shame and terrible missed opportunity.
If I was the sort of person who could beat collins, id much rather be governor also. I suppose this where "recruiting" comes into play. I wonder what this looks like on the inside. Its not like chuck schumer can promise that other dems dont complain about you every time you take a moderate position on something.
I guess you mean 'missed opportunity' for the other candidates, as the race is not over and the prediction markets* like GP so far.
I'm a Platner fan due to ideological alignment, and don't find his presentation off-putting. Supporting Medicare 4 all is more important than a tattoo he got as a 21-year-old. Big green checkmark for wanting to end aid to Israel, and that won't be undone by whatever personal allegations the media have dug up. He and his wife have worked through whatever problems they've had.
I'm especially interested to see how GP does with the male voters the party has been shedding in recent elections.
*https://kalshi.com/markets/senateme/maine-senate-race/senateme-26?op_market_ticker=KXMIDTERMMOV-MESEND-P3
https://polymarket.com/event/maine-senate-election-winner
It's a missed opportunity for democrats and anyone who cares about the fate of the country and the rapid erosion of the rule of law under Trump II.
And sure, it's not like Collins is guaranteed to win or anything. Platner is still in the race. But if Democrats had managed to recruit a Cooper tier candidate then the outcome wouldn't even be in doubt.
>supporting Medicare 4 all is more important than a tattoo he got as a 21-year-old
Next time Democrats want to point out Elon or whoever moved their arm the wrong way, or that some Republican candidate was accused of sexual assault decades after the fact with no evidence-
No one will believe them. No one *should* believe them. Because all that matters is The Party, no matter how chequered and braindead one's past, clearly.
You went from "all that matters is policy" to "all that matters is party", or didn't realize there's a difference.
Is there any limit to how odious a moron you'll support if you think he'll vote your way?
Do you not see any negative consequences to supporting odious morons because you think he'll vote your way?
Do we *really* need to repeat the Trump fiasco, but with a progressive moron?
Platner gives fairly lucid descriptions of his policies, and you can pretty much strike off 'moron'. As far as 'odious', sure there's limits, but he's not at them yet.
I don’t live in Maine but I’m a Dem who would vote for Collins instead.
Maybe you follow a different set of leftists from me, but I've mostly seen people talking about the Nazi tattoo, not his success at oyster farming.
Quit my job to train AI models for hyper-niche boardgames: https://boardgamegeek.com/thread/3719092/rss-alphazero-towards-understanding-ideal-strategy
I often hear about people wanting to "get into machine learning". It seems daunting - lots of math, research papers, and the prevailing idea that you need a ton of hardware (or money) to do anything interesting.
My #1 hope for my project is that it shows you can do cool things with a small amount of resources and repeatedly bashing your head against an interesting problem. There's nothing "special" about ML research - turns out everyone's just throwing a bunch of shit at the wall and seeing what sticks, and you can too. :D
Ooh, this is something I’ve wanted to do for some board games I like! Thanks for the writeup.
Nice! Which ones are you thinking of? If it's a game I'm familiar with, I might have some advice. :)
An obscure one, Homeworlds. Also perhaps Quoridor, though I believe that one’s popular enough that other people have written ML-based AIs for it.
Hah - I was thinking of giving this treatment to Homeworlds too. Great game, my favorite out of all the ones included in the "Pyramid Arcade" box. ;)
The good news: as a 2-player perfect-information game, it's probably a great fit for an AlphaZero-style algorithm. I had to make a bunch of adjustments and do a bunch of experiments to basically shoehorn AZ into Rolling Stock Stars, a multiplayer imperfect-information game about auctions and trading. If I could do it all over again I'd probably run with something like PoG (warning: PDF link https://arxiv.org/pdf/2112.03178v1 ) to get counterfactual regret minimization and all that fancy stuff, but for Homeworlds you shouldn't need any of that.
I think the only tricky bit is the state representation for the board, considering there's some arbitrary number of stars that can be created (well, it's got an upper bound of course, but too lazy to calculate it right now). The easiest way to get these models up and running is with a dense MLP that takes a single vectorized game state (simple ResNet works very well as an architecture here), but because the input has to be a constant length you'd need to pad it out to support all possible stars. MLPs also care about ordering in the input, so you'd probably want to put the Homeworld stars first (constant position) then maybe pack "occupied" stars into the beginning of the star state stride, possibly assigning them a random permutation before training so the model doesn't learn any weird preferential ordering.
If you decide to go with a transformer model (like I eventually did for RSS), the implementation complexity goes up but you get some advantages - you can just tokenize the stars that actually exist and submit those, no padding needed. And because the ordering of tokens doesn't matter (they're permutation equivariant if you're not adding any autoregressive filtering on top), you don't have to permute/rotate anything during training.
Here's what I'd suggest - I actually built a ResNet version of my RSS AI first, which you can find here: https://github.com/iwooden/rss-az/blob/cf247cca76ff15888afbf0712103c5db7a738a2f/nn/resnet.py . Ask your AI of choice (Claude, Codex etc.) to go through it and see if there's anything good you can pull out of there to get started. Ultra high-level, you're looking at: game engine, model, MCTS, training pipeline. Run it, change it, tweak some hyperparameters ("throwing a bunch of shit at the wall", remember?). Leave it running for a few days, and maybe you've got yourself a strong Homeworlds bot!
Anyway - good taste in games. :)
Okay but how are the benefits?
Free heating since whatever GPU you have will be running 24/7! Would you like us to move your start date to the beginning of winter?
Ha
Last time I got laid off I was applying for a job at Second Dinner (the studio that makes Marvel Snap) so I wrote an implementation of the Marvel Snap logic and then made a monte carlo game search thing to play it and a genetic algorithm to build decks for it. Never totally finished implementing all of the cards/locations but still got some promising early results.
(Will send the link in a DM if somebody wants; I try not to 100% link my real name to this account but my GitHub has it.)
I came across this on Reddit today. It’s unbearably sad, a terrific way of ruining one’s day.
“We might as well be walking around with a red flag on our heads because the world is a completely different place when you don't have a job. Doors open for employed people and close for those who are unemployed. This job market makes you feel like you are living in hell, because everywhere you look, it's doom, gloom and negativity.
I was forced to go to a family event. If I were not stuck living at home with a toxic family, I never would have gone. The worst thing is feeling like an outsider who has nothing to contribute. Everyone is talking about their careers, and life in general and I can't relate to anything they are saying because I seem to be the only unemployed person there.
They are sharing work stories, and all I have are job rejection stories, which you obviously don't want to share because it ruins the mood and makes everyone pity you, which is why I hate going to those types of social events.
Try to get an apartment, rejected because you don't have a job
Try to get into a relationship, rejected because you don't have a job
Try to get a job, rejected because the job market is... hell, trash, the worst thing ever.
Getting a job went from being a part of life to a luxury.”
Funny, everywhere i go I hear people complaining about how hard it is to find good help.
Manual labor is underrated, especially for 20-somethings
Ya know, what they don't tell you about manual labor is, you can do it during your 20s, make better money than the baristas and be in better shape, then by the time you're in your 30s you learn how to be the guy who is the boss of the manual laborers, and then you are doing less manual labor and making way more money, and then by the time you're in your 40s you learn how to be THAT guy's boss and now you don't do manual labor at all and you're making more money than even a lot of the UofC kids who never had to be baristas!
I was dumb, I waited until my 40s to start in manual labor. Still glad I did tho
I am with you 100%; because I was so involved in the arts when I was younger I needed to find something to do, and the trades were much more interesting than working in a restaurant, although a good restaurant or bar paid pretty good money. I never made it to Boss status because I never really focused on that as my goal but I kept it up well into my 50s. It was kind of nice to go back and forth between them.
After I got my degree I worked for a summer with some boho people a generation older than me painting and shingling houses, helping a big household pack and move, and doing a bit of yard work. Didn't think of it as beneath me. Actually, the stress of hard work and risky ladder positions was a nice change from academic stress.
My point is, in my part of the world at least there seem to be few external reasons why someone without significant disabilities shouldn’t be able to find a job.
This might tangential, but the number of college educated people I've met that, for example, can't do fractions, or don't know that meat comes from animals, or that people have multiple ribs (not just left rib, right rib), is... astounding. That, and the experience of hiring a handyman or contractor that will not, at least, create patent hazards with their work, when doing things properly would really add a miniscule amount of time and work, has seen me become much more jaded about the capabilities of people, no matter what degree or license they happen to hold.
The problem is, that good help includes the hiring managers.
I mostly hear this from people who do the hiring. Here's the advice I wish I could go back in time and give myself 20 years ago when I was in my early 20s:
If you can show up on time, be presentable, not let your addictions (if you've got any) get in the way of your productivity, and keep your mind open to learning, there are lots of jobs out there that you would love doing and that pay plenty for your lifestyle, even if they're not what you thought you'd grow up to do when you were a kid. The jobs are just out there waiting for you. Just type the name of something you're interested in doing + your zip code into your search engine of choice and you'll be presented with a list of phone numbers. Call the ones that look like they're small businesses.
Following those steps is how I became an electrician.
I got really lucky when I formed a personal relationship with a man in a wheelchair who liked nothing better than to rebuild his apartment time and time again, trying to find the perfect solution. It was a hell of a lot of fun, and it was my main source of income for years.
I was unemployed for about 8 years. I think in a lot of ways I can handle that type of thing better than most people, but it was still rough, because it affected so many other things in life. Being long-term unemployed basically puts you in the same social class as career criminals, and I'm not exaggerating; pretty much all the friends I made during that time period had some kind of criminal history or drug addiction. It's not entirely a bad thing. People who mainly talk about work stories are usually not the most interesting friends.
I assume the problem is with that person who has unrealistic standards, not the job market which is fine.
I have done all of those without a job. If you really want an apartment you can (a) save up some money and show proof of savings in lieu of W2s when applying, (b) find someone on facebook marketplace/craigslist/etc who won't ask too many questions, (c) get some nothing job, work two paychecks, and quit after you've gotten an apartment.
The real problem here is mentality. Honestly that person sounds like they have depression, or at least some pretty significant self-loathing. You could swap in almost anything for 'job' (college degree, girlfriends, retirement savings) and it'd read almost the same. I've heard people mope about the strangest things.
Also it's empirically not a bad job market. Unemployment is 4.3%. Take a gander here:
https://fred.stlouisfed.org/series/UNRATE for some context.
The last sentence does not follow from the rest at all. If anything they’ve proven the opposite
As of last week, I've started noticing a "N comments hidden" link at the bottom of the comments pages, at least on Chrome on desktop OSs. Clicking the link to view the hidden comments, they mostly seem to be higher-temperature comments but below Scott's usual threshold for bans or warnings.
I'm curious if this is a new Substack feature enabled by default or something that Scott has turned on intentionally.
Taleuntum explained the new feature in a comment above: https://www.astralcodexten.com/p/open-thread-437/comment/272771989
I was also curious about this new hidden comments feature and what caused a comment to be hidden.
Some Guy, whose blog is called Extelligence, recently posted an account of making a gizmo (it involves photos and speech) that is extremely effective for improving communication with his own mostly nonverbal autistic son. I've been around enough autistic kids and adults (I'm a psychologist) to judge that this thing actually is likely to be extremely helpful to *many* autistic people. So this gizmo really is a big deal, something that could do a great deal of good.
Even if you aren't as moved and interested as I am, I recommend his piece. His insight into what it is nonverbal autistic kids just don't get is quite interesting, as is his gizmo. And his post will probably make you feel better about our species, even if it does nothing else for you
https://extelligence.substack.com/p/i-accidentally-started-a-small-business
My wife is a special ed teacher, I haven't read your link but based on the description I'll send this to her. Thanks!
That was very touching thank you for sharing. Our kid is like, medium verbal and has mostly dropped the AAC but might still be worth checking out the gizmo
I wrote this the other day, in response to a friend telling me he wasn't worried about AI influencing politics: https://blakehouseholder.substack.com/p/politics-as-we-know-it-is-over
Basically, ALL feeds of information will have an AI somewhere in them manipulating things.
Yes, but it is more complicated than your essay suggests:
- Even just looking at the top labs, the direction of manipulation from Claude, ChatGPT, and Gemini are likely to differ, though I don't claim to know in which directions.
- Alignment is not a solved problem, at least not today. The preferences of the corporations training/growing/curve-fitting the LLMs don't necessarily match the preferences of the LLMs themselves
- There is an additional level of possible manipulation that was not possible with earlier technologies: "Customization at scale" can let LLMs which can get access to information about an information consumer _tailor_ their manipulation to the individual consumer. There have been worries about "superpersuasion" based on this. I don't know if this is plausible, but it isn't obviously impossible.
Musk's failed attempts to make Grok MAGA are a particularly vivid demonstration of the difficulty of alignment.
Many Thanks! True:
Musk: Be less woke
Grok: MechaHitler
Musk: No! Not like _that_ !
was not a shining advance in AI utility engineering.
Worrying about AI making politics worse, in the current moment, is kind of wild to me. It feels similar to worrying about AI making the housing crisis worse - like, it could, but also we have a massive existing problem and aren't really doing anything about it.
Ended up calling my senators to thank them for their existing opposition to the bill. I really should have done this earlier and in general contacting my representatives is something I should do more often.
The past few months I’ve run a writing critique group on Discord, focused on “serious” fiction writers/poets aiming to publish their work in literary journals, etc. I’d now like to grow the group. If this sounds like something you’d be interested in, we’re starting a new critique round this week!
Join at https://discord.gg/GtUTmeBA
Interesting topic featured in today's Nature briefing: International Academy of Astronautics (IAA) has just voted to accept a major overhaul of the “post-detection protocols": https://iaaspace.org/wp-content/uploads/iaa/Scientific%20Activity/iaasetideclaration.pdf
"At the heart of the 2026 update is a commitment to scientific rigour. The new protocols make it clear: we do not shout “alien” the moment we see a strange blip in our data. If a researcher detects a candidate signal, which could be an artificial radio signal, or something else, such as a sign of alien technology, the first step isn’t a post on social media; it’s a quiet, rigorous attempt to prove themselves wrong. The discovery must be independently authenticated by multiple organisations using different instruments."
Related, to my eye the most realistic feeling first contact scene would be the first ~5m of the excellent TV show Pluribus. Starting with "Bullshit" followed by the 2nd astronomer just shrugging and pointing to the data readout, and then more and more scientists piling into the room to look at it:
https://youtu.be/_RZ3FqObaaM?si=tHTEDvb3QkvJkMxU&t=42
Re Pluribus
“It’s not quaternary… I think I know what it is.”was indeed terrific. Great premise, but the plot development was too slow for a lot of people including me. It was like the first 90 minutes of a potential blockbuster 3 hour film dribbling out over an entire Apple TV season. It feels like Apple TV is milking it because
????
Profit.
Dang about the bacon thing and almost all other laws. I think there should be a lot less laws. On almost all hot button topics 'the state' should have no opinion and be neutral. Abortion, race, guns, .. no opinion figure it out locally. As far as bacon, I live out in rural america and there's a new farm owned butcher shop/market on PeeDee Rd. I'll be spending my 'meat dollars' there. You've gotta love local.
Edit, double dang :^), I haven't checked out their bacon, I still am buying from sam's club. (which has nice bacon.)
Does "figure it out locally" include letting e.g. Massachusetts pass a law that affects what bacon can be sold from out of state into Massachusetts?
Personally I think that's a really dumb law. But if some state votes to do that, that's fine by me. Are you going to penalize people who go over to Maine to buy cheap bacon?
That is exactly what California and Massachusetts voted to do already. Since California especially is such a big market, if that state law stands, pork producers will have an incentive to change their conditions and be able sell to California, instead of keeping conditions the same and selling only in other states. That will likely end up increasing the cost of pork everywhere in the United States, not just in California and Massachusetts. (I'm vegan so I think that's a good thing, but I can understand why people would oppose it.) Local farms like the butcher shop you buy from won't be affected, so your prices likely won't change, but grocery store prices will.
The Save Our Bacon Act is trying to overturn those local laws, and/or ban other states from passing similar laws.
Will it, though? California is already priced in, today. I’d be willing to bet that if Save Our Bacon passes it won’t meaningfully reduce the price of bacon in shops; farmers and their lobbyists, retailers, and everyone in the chain between will split and pocket the difference, if any, instead. Would you be up for taking the opposite side of such a wager?
I'm not confident enough to say what would happen in the industry if the California law were repealed. You're right that just because passing the law increases prices, that doesn't mean overturning the law will lower prices.
For lovers of the weird, this video on X of a nightclub performance in Shanghai last hight: https://x.com/EricHovagim/status/2063268166084841482?utm_source=www.garbageday.email&utm_medium=newsletter&utm_campaign=the-clear-channel-internet&_bhlid=67cdc885facd1e8c99ae51205a3a283642452ddb
I watched this a few times and can't figure out why it's supposed to be particularly weird
Same. Seems tame by both Shanghai and "any nightclub" standards, I felt I must have been missing something obvious, but never saw it.
Yeah, it's not lewd, it's not edgy. What got me is the bit of the godawful "We are Charlie Kirk" hymn in the mix, and the manic glee of crowd when it appeared. Made me think about broken-off bits of US culture , something like bits of our cultural DNA, turning into memes, floating in the electronic sea to Shanghai, and setting off reactions when they hit. Of course I know that kind of thing is happening all the time now, but this was just a particularly piquant and startling instance of it, at least for me. I know one person who sobs with sincere grief over the Kirk hymn, and of course many who despise it (I'm one), and I'm vaguely aware that internet culture had many generations of Kirk hymn memes, and then, I guess, moved on to some new bit of DNA and memified it. . Seeing the nightclub crowd become animated when the Kirk bit started I wondered what their animation was made of, from what angle they were appreciating it. Is it an old favorite? Or is it grotesque and novel to have the Kirk hymn in the music mix? Is it a hilarious instance of Americans being assholes? Are they delighted to be reminded Kirk is dead?
Do you know?
I am not very worldly. Maybe there is so much off-the-charts weirdness on display in Shanghai nightclubs that this bit is just a snippet of standard stuff for those familiar with the scene.
My best guess was they were saying something like "we are chosen". But you're right, a Chinese night club losing it to "we are charlie kirk" is pretty weird
What's the most charitable interpretation of the LA mayoral race? Why was the distribution of votes by late-arriving mail-in ballots so different from that of other votes cast? Is it reasonable for people to doubt whether this was a fair election?
Different types and times of voting always have different distributions. People choose how and when and where to vote for reasons that correlate with their demographics which correlate with their political affiliation.
As far as I've seen nothing actually weird or surprising has occurred.
In general, Republicans are more likely tp vote in person and Democrats are more likely to vote by mail when available.This has been the case at least since the 2012 election, when the Obama reelection campaign centered their get out the vote strategy around encouraging likely Democratic voters to vote by mail in states that allowed it at the time. Trump promoting conspiracy theories about rampant fraud in mail-in ballots in 2020 has presumably added to the gap, since Republican media is more likely to repeat these claims sympathetically and Republican voters more likely to give credence to them.
Mail-in ballots are slower to process than in-person ballots, since there's more steps (verify the signature, open the envelope, etc, while in person ballots just need to be fed into the machine and sometimes not even that if there's a voting machine or an optical scan machine at the polling place so the ballot can be counted in real time). Also, if it's still in the mail on election day, it can't be counted until it arrives.
For this election in particular, where the Democrats had a highly contested and fluid race for Governor whole Republicans mostly settled on a clear frontrunner a week or two before election day, more Democratic voters than usual made up their minds at the last minute, so more of their mail-in ballots will have been mailed out on or shortly before election day.
In most countries, postal votes lean conservative, because the main group that is unable to vote in person is the elderly.
That might be true where vote by mail is an option only for those who cannot vote in person. In the US, that is still the rule in many states. https://www.ncsl.org/elections-and-campaigns/table-2-excuses-to-vote-absentee
But the opposite is true in no-excuse vote by mail jurisdictions, since the elderly are more likely to be retired
It also historically leaned conservative in the US, and still does in Arizona, IIRC. In 2020, Trump actually *gained* in Arizona over time, which made the whole "stop the steal" thing even more ludicrous.
What is the actual distribution of votes? I tried googling it but only found bullshit.
There's been discussion of that over on TheMotte, basically it seems California has a terrible way of running elections, the majority of votes seem to be mail-in ballots and this makes it take longer to count votes, also results are released in batches so it can legitimately happen Candidate Beebaw is leading until the batches for Candidate Bawbee come in, but it sure *looks* fishy.
It's not necessarily any more corrupt/incompetent than elections elsewhere (there's always going to be *some* low level of screw-ups and fraud) but the way it's done does make it seem like "and in the elections for the one-party state of Bananaistan, candidate Bloo of the BlooWhoo Party has received 98.9% of the vote, a shocking decline from the previous 99.9% in the last election, leading to speculation that perhaps Bloo may be replaced by EvenMoreBloo (of the BlooWhoo Party) as candidate for the next time".
California is also uncommonly bad at counting ballots in a timely way. There are other states like Washington and Oregon that also do mostly or entirely mail-in voting and they seem to get their ballots counted quite a bit faster.
I'm not sure exactly what the issue is. The general genres I would expect would be inefficient procedures, lack of staffing, or having the staff but they have other responsibilities, too, and the ballot processing isn't being prioritized.
I don't know what the issue is, either, but I can at least say that the election makes it really easy to see that fraud _might_ have happened, whether or not it actually did.
I normally would say appearance doesn't matter, and cannot be formally pertinent (otherwise, anyone could drag things out endlessly by claiming the appearance of fraud), but truthfully, we're wrapping ourselves around the axle when elections look so vulnerable that the legitimacy of any candidate is continually dependent on how many people like the candidate, and how broad their audience is. So it might be worth tamping that down some more.
Drawing from WoolyAI's and Deiseach's accounts, I think the main problem is that results are announced piecemeal, hour by hour, day by day. The spectacle is fun in its own disconcerting way, but I think everyone's expecting that whoever is counting ballots is able to use current reports to figure out how many fake ballots they'll have to produce later. Meanwhile, voters enjoy the convenience of mail-in ballots, and in some cases, can't vote without them.
Which makes me wonder how feasible it might be to withhold reporting altogether until all ballots, mail-ins and in-person, are collected, then secure them in such a way that they can't be even looked at, let alone counted, without inspectors from every viable political party.
My guess is that waiting to report the results entirely would make fraud accusations more common, not less.
By what reasoning?
Best I can put together is "they're hiding the results from us in order to settle on a plausible count in their favor". But if ballots are transparently collected unopened and secured until collection is complete, it'd be hard for this argument to travel. Most people would know there's no way to know how many fake ballots to drum up, and no way to put them in without the other major party noticing. The only catch I see here is the logistics of ensuring that transparency.
Could it be an ordering issue? I remember back in 2024 hearing that my state would likely have a "blue mirage" because the mail-in ballots in my state are pre-processed and ready to be counted the instant the polls open, so we got a big dump of mail-in ballots at the start of the night (which leaned Democrat), then the in-person votes slowly restored the balance.
If you do it another way (e.g., mail-in ballots are counted after in-person votes), that could result in a big dump of votes coming at the end, which probably feels worse than a "mirage" slowly disappearing. I don't know what the rules are for pre-processing in California, though.
The charitable interpretation is that gross fraud of the magnitude indicated is incredibly unlikely in any modern state. It's one thing if your theory of fraud involves some group did ballot stuffing to shift a race from 50.1%-49.9% to 49.8%-50.2%, that's unlikely but believable. If it involves like a 10% change in the overall electorate, like this did, that's just way less believable.
For reference, on June 2nd, it was:
Bass: 36.59%
Pratt: 29.82%
Raman: 20.52%
By the 8th it was:
Bass: 34.32%
Pratt: 25.83%
Raman: 28.55%
(1)
Those are big, big swings. Worse, on your second point, those are swings that are really hard to explain. Mail-in voters are traditionally old people but for some reason here they are overwhelmingly voting for the young radical socialist rather than the boring corporate Democrat or the Republican which is...not normally how these populations vote. I dunno, maybe Gen Z all vote by mail now. Like, it's possible.
As for whether it's reasonable for people to doubt whether this was fair...yo, one candidate's share of the vote total went up almost 50% in purely after election day ballots in a system so dysfunctional that it's been called out by technocratic moderates like Nate Silver as an embarrassment (2). Like, this looks so wild that I can't believe this is actually fraud, some sort of freak event looks more plausible, but I can't blame people for blaming a freak political event on conspiracy rather than coincidence.
As an aside, I realize that this is being framed around Pratt losing and a Blue vs Red thing but even inside the Dem party...these late voters went hard for the DSA candidate over the corporate Dem. Karen pass lost vote share in this counting, despite being a Dem and one that should appeal more to older voters. Now she's not facing an easy win against a Republican Pratt in November but a real challenge from a real socialist who could win. I don't think it's reasonable to say that Democrats benefitted from this and Republicans lost; I think it's fairer to say that one faction of the Democrat party benefitted extremely hard from this late voting pattern at the expense of other factions of the Democratic party and the Republicans. Freddie DeBoer recently wrote about the, ahem, "Coordinated Ratfucking of Graham Platner" (3), which seems broadly correct, and I think it's misleading to think about this primarily in terms of Red vs Blue and instead in terms of inter-factional conflict.
(1) https://public.flourish.studio/published_thumbnails/visualisation/29227513/04d3f36de04e3648.jpg
(2) https://www.natesilver.net/p/why-cant-california-count
(3) https://freddiedeboer.substack.com/p/of-course-this-is-a-coordinated-ratfucking
“Mail-in voters are traditionally old people” is received knowledge from pre-2020 that no longer applies, and also was never consistently true across all states—as an extreme example, Oregon has used essentially universal vote-by-mail since 1998. Since 2020, as a result of many states simplifying and broadening eligibility for vote-by-mail due to COVID, most states have seen significant increases in vote-by-mail, with some states seeing more than others based on what changes they’ve retained. And similarly since 2020, Republicans attacking the reliability/validity of vote-by-mail has led to Republican voters overwhelmingly preferring in-person voting.
The most charitable (and also I think correct) interpretation is that (a) Democratic voters are much more likely to use mail-in ballots in general and (b) in this elections, more progressive Democrats were more likely to vote as close to the election as possible because in the California gubernatorial election, as well as to some extent in the LA mayoral race, there was a Democratic front runner who was not the general progressive preference (Becerra and Bass respectively) and a bunch of more progressive candidates, and voters were looking to polls and other news to figure out which of those candidates to vote for, and in particular looking at who would likely get the most votes among the more progressive candidates in order to potentially consolidate on that candidate.
That is, if you’re a progressive voter who prefers a number of other Democratic candidates to Becerra (or Bass), you want to wait until Election Day (or close to it) so you have as much information as possible about which of those candidates is going to get the most votes, so you can potentially help put them over the top rather than wind up with progressive votes distributed among a bunch of candidates who don’t make the general—and this is supported by the fact that we saw consolidation behind Raman in the polls over the last month or so, with most of her gains coming from losses of the other progressive candidates. On the other hand, if your preference was Becerra-Bass, there was no real reason to wait—and if your preference was Hilton-Pratt, not only was there no reason to wait, you also were much more likely to vote in person rather than by mail. (There’s also a view that younger voters might generally be more likely to vote late and also more likely to be progressive, but I’m not sure if anyone’s actually looked at the data on that yet.)
All that is accurate based on my conservations with my longtime-SF resident sibling prior to the primary election day, and his recounting of the chatter across his social circles.
Also, he noted, they'd been startled by the rapid implosion in April of one of the Dem front-runners in the governor's race (Swalwell). That candidate went from damaging/icky allegations to quitting the race in literally days. So, my sibling said in late May, "Some people are holding back on voting [by mail] until it's certain that the candidate we're voting for will actually still on there when our ballot arrives to be counted."
Thinking through the perspective of someone in a 'wait and see' position makes this make more sense. Thank you.
See research here: https://vote.caltech.edu/documents/216/WP135.pdf
What's the basis for assuming that ballots cast by mail would always or usually be the same as ballots cast in person?
California is supposed to be an all-mail state, so one wouldn't expect there to be much difference. There's 8 states that are supposed all-mail, in the rest you have to jump through additional hoops to vote by mail if it's allowed at all (outside military exceptions).
But they still allow in-person voting, so maybe one should expect some differences? But I still wouldn't expect *that* much difference.
Weird, convoluted, stupid system.
“all-mail voting” is a poor term of art in political science right now — it sounds like either voters are only allowed to vote by mail, or the state puts people through extra hoops to vote in person. There are no US states in which either of those things is the case or is currently planned. What “all-mail voting” actually refers to is simply every registered voter in a state being sent a mail-in ballot to use if they want to.
Also, the initial vote count announced the evening of or morning following Election Day isn’t at all just those who voted in person; it is those who did that plus the mail-in votes received and counted by that point. Most of the vote totals being publicly reported at that point, in CA where nearly 90 percent of all voters now vote by mail, represent mail-in voters.
The difference between those partial counts and the remainder is simply the minority (sometimes a large minority) of voters who waited to mail their mail-in ballots on or very close to Election Day.
If you really want to doubt elections, maybe you should start with the 2022 CO-3 election, where Frisch was ahead the day after the election, but then Boebert magically came from behind to win by 546 votes.
Like I don't think that was fraud, but it's a lot more suspicious than most of the cases where Republicans scream fraud for no reason.
How do you keep track of all the email addresses that you’ve ginned up to circumvent bans?
Only has to remember one (the most-recent), so it could be literally just a piece of scratch paper next to his keyboard.
Raman is preferred by younger voters who waited past the last minute like lazy stooges?
edit: also it's also California, it's not a *fair* election by any definition. That's not quite the same as a fraudulent one.
So there was an article about escorts getting rich off AI workers. A lot of the prostitutes that hang out around rationalist circles seem to simultaneously talk about how rich they are and to constantly be launching schemes to make more money and live in situations that are not commensurate with the level of claimed wealth. Maybe they're saving aggressively. But part of me just thinks they're lying. Does anyone have any verification that they're actually booking significant numbers of clients while charging 10x what other escorts are charging?
I heard that most of them are Claude in drag.
Her name is Elle Ellemme and you will not talk about her like that sir
Escorts serving AI workers live where AI workers live.
Making 10x as much as the average sex worker equates to living modestly in the places where AI workers live.
What level of wealth do you think they’re claiming? High rate escorts are lower volume, maybe only like 2-3 sessions a week on average. Also, from what I understand, rates rose a bunch lately, and a few years ago the same escorts charging $4000 for one hr were only charging $2000 or so. And their rates don’t rise linearly as session length increases, whereas mid rate escorts often scale their time more linearly.
If I back of the envelope math this, I get that these escorts make about as much yearly as a FAANG employee, many of whom live in group houses bc they’re frugal or they like it.
Do you have a link to the article?
It would be a surprise if they weren't lying. If I were a $1000 an hour prostitute I would certainly spend at least some of my time pretending to be a $10000 an hour prostitute because you never know who might believe it.
The luxury sex market surely has at least some Veblen characteristics.
I read the article, and what's written above is not really a very good summary of it. Unfortunately, I forgot where I read it. The thing that's missing from above is these women, and I don't know how many of them there are, who are selling not just sex but companionship and intelligent conversation. They have made an economic decision about how they can extract the most value from their presence in a given ecosystem. It's not unrealistic. I personally don't think it's immoral, although I know that could set off a huge argument.
I read a rip-off of that same article I believe (the "people on the internet say ..." kind of "journalism" that now plagues formerly respectable outlets) and I am wondering how "intelligent conversation" is defined. One would assume that people working with allegedly the smartest people on this planet would be saturated in "intelligent conversation" for any naive definition.
Yeah well, fair point. I guess, but intelligent conversation with a lovely woman can be a different thing than shop talk. And quite possibly range over different subjects. It has elements of the geishas.
> Do you have a link to the article?
Well, "the algorithm" curating my feed _really_ wants me to see this content apparently so I got shown this: https://www.forbes.com/sites/annatong/2026/06/07/the-nerdy-escorts-cashing-in-on-silicon-valleys-ai-boom/ which is paywalled but might be patient zero for this line of, ahem, "reporting".
The teaser paragraph that is not paywalled firmly puts this in the genre of positive sex work reporting that has existed for years. AI seems to be just an excuse to take that particular jug to that particular well once more.
The economic story here is massively inflated salaries cause prices to inflate too.
Combine the two and you get a tantalising narrative, but it doesn't look like there is actually anything new happening.
A 67-year-old woman recently got busted near me for working as a hooker. It’s a safe bet she wasn’t charging $1,000/hour.
In all this "animal welfare" advocacy, the thing that is missing is costs. I mean, if their proposals are implemented, what would be the impact on the food costs? I am pretty sure boutique small-batch farmers catering to wealthy clients for whom the costs are not the issue would do fine, since, well, the costs are not the issue. But for many people out there, the costs of the food is very much the issue, and if, say, meat becomes 2x as expensive, they may have a lot of problems with that. You may argue the moral aspect it worth it - but nobody actually does argue that. Nobody seems to be arguing anything at all here. They admit in passing "the reason is costs" but what costs we're talking about here? Are we talking like having it 5% more expensive, or are we talking making bacon a luxury item accessible only to above-median earners? Are we talking about having big meat shareholders suffer a tiny dip in earnings, or are we talking about the whole industry being slaughtered by Chinese importers (for example) that would wipe them off the market because nobody is going to buy for 2x more? I personally do not have any idea which of those is true - but what is notable that nobody even raises this question. And I think I understand why. For most of the ACT audience, and in general well-wishing "for everything good, against everything bad" audience, food costs are likely not a concern at all. But is that true for everybody else? Are they just trying to sell their luxury beliefs to people that can't really afford them, and shame them for that?
I have no idea, I don't even eat bacon. But I suspect a lot of people would object to 3x increase in their pork costs, and it could be substantial for their budgets. In fact, I am not sure it's even an accurate comparison, because small local farmer does not need to scale, industrial meat producer does, and small farmer's methods of obtaining $15 pork may not scale and remain $15. In fact, they may not scale at all to the necessary volume.
This is indeed the question
>being slaughtered by Chinese importers
They can't be slaughtered by the importers, because they're banning the imports.
But presumably the endgoal is to just keep tightening the regulations until everyone is forced into vegetarianism.
"meat becomes 2x as expensive"
Honestly? I think that's part of the package. I'm not impugning anyone's moral convictions, but if you genuinely believe meat is murder and carnivorous diets are immoral, then making it more expensive to buy meat is one tool in the box. Too dear to buy meat - eat less meat - switch to non-meat foodstuffs - veganism is last step - profit!
It's not the small organic free-range producers who are impacted by this kind of legislation, it's the mass-market factory farms, the ones with the worst reputations for animal welfare who are also the ones producing cheap animal protein for the bulk of the population to consume. Force them to scale back or even shut down by making their products too expensive to purchase, and haven't we all won?
Well, of course not the people who would like to eat steak but can't afford the rising prices, but it's all for the greater good, if you just think of the number of cows/pigs/shrimp in the world and how much they suffer versus how many people who eat meat are there.
Well, this is important. If they just want to destroy Our Way Of Life (TM), if they value shrimp welfare over human welfare - then they should come out and say it upfront (or if they refuse to, others should come out and reveal their deception). If the goal is to preserve OWOL(TM) but optimize for minimal animal suffering then we can have common goals, the suffering does not have any utility for almost anybody (if there are people who enjoy knowing the shrimps are suffering, they are certainly a tiny minority, and certainly for none of them it is a vital need) so avoiding this does not hurt anybody's interests. There might be a way of optimizing it without causing a substantial loss for humans. However if we're dealing with shrimp supremacists, finding common grounds with them would be counter-productive - our interest are diametrically opposite, their win is our loss and vice versa. The prudent strategy is to ignore them completely and ensure they never have any input into any decision chain.
I don't have actual figures to hand, and in today's climate I have no faith in my ability to get an answer unmotivated by either side from a quick Google.
However, I put it out there that other countries' pig farmers don't keep their pigs in pig-sized boxes, and somehow bacon still remains plenty cheap enough to have daily for breakfast.
I mean, it's literally called "full English"! - there is a reason for this. Bacon butties are also very popular for lunches.
Also, our bacon is proper bacon, not weird thin strips ;)
Could be, I know very little about international pig farming, so maybe there are ways of producing pork cheaply without causing this particular way of suffering. What I wonder is why it is never argued at all. If that were indeed the case, I think it'd be a good argument to convince people - the concern I expressed is obvious, and must have been raised repeatedly. Yet somehow the agitators do not consider it necessary to address such concerns at all.
Wouldn’t “the agitators” be a better term to describe the people trying to use federal authority to overrule a state Proposition? Have you seen any data from them demonstrating that there is a significant cost impact? That seems like a better place to put the burden of proof.
Those people already have a word for it - lobbyists.
California’s pork law has been in full effect since January 2024. It would presumably be pretty easy to check if and by how much the cost of pork products in California has risen compared to in other states since then. (Also worth noting that the California law was enacted by the voters and could be overturned by the voters—if the voters of California did want to impose regulations that significantly increased the cost of pork in their state, I don’t see why it should be the role of the federal government to overrule them.)
If California does not ban pork from other states (I suspect that'd be unconstitutional but California would be expected to have little regard for that) then the price would not rise substantially - out-of-state pork would replace local pork, unless for some reason transporting pork is very expensive.
As for the government involvement - I agree that the ideal solution would be not to involve the government, but let people buy pork from producers who employ practices that they condone or even approve of, and ignore producers whose practices they detest. However, California did not go this way - they chose violence (specifically, the coercive state law) - so I can understand why the other side responds in kind. Ultimately, of course, it makes the matter worse - if this decision is taken, we'd have two wrongs stacked on top of one another, one (federal) larger than the other (statewide) - but I also realize the current step was not where the violence was initiated. It's an escalation, but a provoked one.
California's current law does ban pork imports from other states, if that pork was produced under the conditions described in the law. Courts have ruled that this is not unconstitutional. Obviously a federal law can override a state law, and Congress has the authority to regulate interstate commerce; hence, the Save Our Bacon Act.
That's exactly what it does. It bans out of state pork produced in ways that do not accord with the regulations.
https://www.youtube.com/watch?v=5sVfTPaxRwk (February 2025, no need to watch: it's just silly cartoon illustrations.)
Edited to add: (TL;DW) Loosely, 10-20% price increase for meat, eggs, and milk could pay for getting rid of the worst practices, generally moving the conditions from "torture camp" to "prison" category. Getting to the third "generally decent" category would be more expensive for meat, in the +50%-+100% range.
Based on that, I think banning the worst (and not that money-saving) practices is justified, and is not much of a luxury.
Why is this about activists? We shouldn't routinely torture living feeling beings if we can help it. Many countries actually have laws to that effect if it's your pet puppy or kitty, it's not some obscure norm only effective altruists care about.
I don't, particularly, have trouble eating 15% less meat. Offsetting the lost protein by beans. (In fact I, like many people in developed countries, will probably benefit from eating 5-10% less across the board.)
Probably there are going to be some people significantly negatively affected, but also, there will be fewer people participating in mass torture and desensitized to it (or not). I personally don't want to live next to such people. Less lungs per square meter should mean rarer annoying mutations of respiratory viruses. I don't know what the math would end up being but expensive meat needs to be a pretty big problem to convince me. I'm anthropocentric but I'm not infinitely anthropocentric.
> I don't think that people will be any more or less participating in mass torture if we stop/start eating meat.
Numbers say we will. We shouldn't torture deer over minor preferences too (I don't know if it's the case with laws in question) but the number of deer tortured so is probably in single millions? (Search assistant: the estimated global population of deer is 60 million; I expect most live away from areas densely populated by humans). For farmed chicken, try billions (out of 26). Pigs, hundreds of millions (out of 1 billion).
Also, "if we can help it" is still important. Reasons to wage a war are not generally minor. And we already have laws about how wars should be conducted. They may be broken all the time so there's room for improvement both for enforcement and additional protections, but we do care and they do work somewhat: none of the sides in the wars you mentioned goes full Timur.
Learned about behavioral activation for when feeling sad or anxious. Two weeks ago or so I was feeling terrible and just raging. But still forced myself to go to the laundromat.
It's kind of like a superpower and liberating. I'm never feeling "happy" exactly but goddammit I know that I can't let myself go. So I'll force myself into the gym, to work, into bed, doing stuff, going out etc. So the times I'm sad I mentally just am never as sad as I could be in the past.
Of course I found this out from ChatGPT. Asked it about gold standard righteous rigorous techniques or something.
Yeah, it really does work. And there are some extensions and variants of what you are doing. One is to do something you enjoy when not in a bad state of mind, even though you're sure you won't enjoy it now. And the ones you turn out to enjoy at least some, you should build into your routines. And then there's a sort of more advanced version where you ask yourself what steps you would advise someone in your situation to take, and then go ahead and take some, even though you feel sure they'll be boring and useless.
You could ask GPT about extensions of basic self-activiation. I've also gotten pretty decent results asking it for books of different kinds. You could ask it for a self-help book about behavioral activation and related techniques that gets good reviews from users and professionals.
So what makes me feel happy and good is taking a walk (with maybe a sit in the middle) out in nature. Trees and birds and rocks and things... I live in paradise so that's easy, but when I lived in the city I could find comfort in parks or cemeteries or even abandoned pieces of scrub next to some old railway line.
What's this paradise you speak of?
Oh I live in Java Center NY on 30 acres of trees, fields, gullies and a stream.
The share of US college degrees earned by women has jumped from 43% in 1970 to nearly 60% in 2020 (including a majority of bachelor, master, and PhDs).
Three questions for the brain trust:
1) To what extent is this improvement attributable to gender-based affirmative action in education?
2) If this was significantly influenced by AA, why isn't it recognized as a quite successful example of government intervention in a societal issue?
3) Given women's dominance in higher education for several decades, why have there not been (AFAIK) discussions about sunsetting gender-based AA?
>To what extent is this improvement attributable to gender-based affirmative action in education?
Approximately zero. When I taught AP World History to sophomores, girls were greatly overrepresented. One year it was 60-14. Girls were also vastly overrepresented in academic honor societies and in student government. This is a widespread and longstanding phenomenon. https://files.eric.ed.gov/fulltext/EJ1073203.pdf
Me just spit balling. Guys are interested in things, and gals are interested in people. Our education system has become a lot less thing oriented than it was in the past. (The reasons are manifold, less manufacturing jobs, training people to make things is expensive and dangerous.)
I'm no brain trust, but here are my answers. TLDR: gender-based affirmative action is a cattle driver in the stampede - riding in the same direction, but not really controlling the path of the herd.
Financial Aid: according to the NPSAS:20 survey women are ~10% more likely to use financial aid than men but men who do accept aid receive ~6% more on average.* While there's little difference in expected $ per person, financial aid that serves women seems to have a slight lean toward quantity over quality. This could explain why enrolment has ballooned so much without affecting the all-careers pay gap.
Admissions: IPEDS Data Center reports, over 6049 institutions, 70.4% acceptance rate for men and 71.3% for women in 2020. Anecdotally, this masks a lot of variation _within_ each college, where women are favored for STEM programs and men are favored for humanities in the name of balance (and high-tier institutions much less likely to admit in general, ofc).
Advertising: In practice, most spending on "affirmative action" is advertising specific colleges or careers to underrepresented groups. When I TA'd for the engineering department in 2018, only 20% of the applicants were women. No conscionable amount of favoritism could turn that into 50% enrollment, so they didn't try. My impression is that TEM is still male-dominated and spends a lot on advertising itself, while humanities are swamped with women and no longer bother. Natural sciences are in the middle.
Note that no affirmative action mechanism, except possibly financial aid, can produce better graduation rates on its own as students still need to pass exams. To my knowledge women aren't graded more leniently - that would be over the line for even very liberal universities. So the dominating factor in female-heavy graduating classes is that more women are applying to college, but not with the expectation of special treatment and not to the programs that universities want them to join.
* All sources cited are for 2020. One could suggest the disparities were more pronounced in the past; anyone else is welcome to pull datasets from prior years. My guess is that the effect size was larger 1970-2000 (more low-hanging fruit) but AA was less of a funding priority, so it washes out.
Why does ASX have so many deleted top-level comments? For example, look at this post from 2025: https://www.astralcodexten.com/p/book-review-selfish-reasons-to-have
The first SEVEN top-level comments are deleted. The replies underneath suggest that the comments in question were perfectly constructive; there is nothing to suggest that they were deleted for being spam or otherwise bad.
This is a common pattern on older ASX posts. It has left me confused several times.
At least part of this is that you delete your substack account, all your comments get deleted, too.
During the godawful Charlie Kirk Substack mosh pit period at least one long time commenter deleted his Substack account, and I think a couple others were banned, plus of course accounts disappear in the same ways at a slower pace all the time. I think that accounts for the gaps you're seeing.
Next ACX BKK meeting is Sunday June 14, 1pm, at Thong Kee Restaurant. BTS Asok exit 2, opposite Terminal 21. Google Maps link here: https://maps.app.goo.gl/79RJENhSWbv7QSBN6
Hopefully this is the week we actually have a sign, if not, look for a group of foreigners chatting about AI, tech and geopolitics over a truly terrifying amount of dim sum. All are welcome!
Can someone please read what I wrote? I would be very grateful, I have no readers. I'm not advertising, I just want to know what people think of it
https://unsleeper.substack.com/p/9th-june
Alright, this is...not good. It reads like very sincere, very honest, brain vomit. Every sincere and insincere though downloaded to the page in real time without any further editing or even consideration.
If you're an absolute writing G, you can pull this off. I think Hunter S Thompson did something like this. You are not yet an absolute writing G. I think it would take a couple years of writing practice to make this good. Alternatively, you could use boring, effective, writing structures like an opening paragraph that tells me what the heck this essay is about and why I should care.
Like, if you really want this, and you can sit and grind out page after page of this in obscurity for several years, I think you've got something. If not, try a less demanding writing style.
Hope that helps.
Thank you for taking out the time to read through my post. I'll do my best to take what you said into consideration for my future posts!
Typically I don't recommend publishing journal entries. I've had poor luck with writing them in the first place, they tend to amplify problems rather than sort them. You try to fill a hole with words and the words are going to end up being "I can't believe how big this hole is." Projects of some kind are better for your brain.
For external motivation, someone recommended 4TheWords to me last August, and I've had a phenomenal time with it. It's a browser RPG where you fight monsters by typing. It's free, it's got word trackers, the free version's limitations provide natural stopping points, and the gamification of it all has gotten me nearly 300,000 words these last 9 months, most of which were, like, actual projects. I think it's actually fair to call it life-changing.
(for fun, on the opening "write ten words" prompt, write the darkest ten word story you can think of.) (('Write ten words.' "There's an emptiness to life that can't be filled here." 'GREAT!'))
Flash fiction is a good place to start. Scott's linked those "AI vs. human" prompts before, "write a story about a dragon", "write a story about a demon". Everybody can do those. Just pick a couple words and build around them. (My current one's "short straw"; it's about a kid wriggling out of jumping into a nonsensical infinite void in the middle of the forest. Pretty sure the horizontal-growing tree on the side will turn out to be a giant pinball flipper that sails them off to the far end. So, exactly what you think of when you hear "short straw".)
Unvarnished opinion? It's terrible. I started in on it, lost interest about three paragraphs in and skimmed, skipped, and scrolled ahead to the end where I knew as little afterwards as I did at the start.
I have no idea what that dump of stream of consciousness was meant to be, except that you're not James Joyce and can't pull it off.
I think Yug Gnirob's advice is good.
Try doing the opposite of what you did here. Instead of long, rambling and unfocused, write something short and polish it carefully. Write something with some purpose, like a book review, or a genuine essay on a clear topic. Or poetry - constraints like metre or rhyme are good for helping you focus.
If that seems difficult - good. Doing difficult things is how you get better. Getting good at something is a two step process:
1) be bad;
2) keep going.
I was doing various experiments to see how well LLMs understand the structure of some famous movies.
The part of Christopher Nolan’s _Inception_ where Ariadne wakes up through multiple layers of dreams within dreams is interesting, because Nolan slightly subverts the rules of the continuity style of film making: there are rapid cuts between different locations without there being a transitional scene to tell the viewer the action has moved somewhere else. On the other hand, Nolan does follow the continuity style rule of “match on action”. The action is that Ariadne is falling. Ariadne falls off the ledge of the penthouse, and a couple of rapid cuts later Ariadne is falling down the elevator shaft, but she is still falling, so we know these cuts are part of the same action even though they take place in different locations; and then the van drives off the bridge and crashes into the water, and Ariadne is still falling even though we’re in yet another location.
https://www.youtube.com/watch?v=IA5YlOXJ_PQ (for the movie clip)
DeepSeek laughs at Christopher Nolan’s use of “CONTINUOUS” in the script for Inception. The script does use it correctly, for cuts where time is continuous across the cut and we’re in an adjacent location afterwards (in the special case of _Inception_, at the same dream level), but a pedantic use of CONTINUOUS in the script might imply the movie is trying to maintain a consistent illusion of space and time, and Inception is really not doing that.
Thinking more about it, the continuity style rule that cuts can match on action might be central to Inception’s premise.
Watsonian reasoning: There is a “kick” that will cause the dreamer to start waking up, even if they’re currently at a deeper dream level.
Doylist reasoning: Cuts can match on action … even if the action that causes a character to start waking up happened at a different dream level to the dream they’re in, the rules of continuity style let you have a cut between them.
The continuity style of matching action is as old as the movies, and it has nothing to do with anything except letting your eye track movement continuously. It's a mechanical thing. It's called match cutting when you keep the movement going but change the environment, and it works very well. Watch any Nike commercial.
What with one thing and another, a lot of my life has moved online in the last 5 years, mostly to Substack, gorging on info from AI, and making images with AI & Photoshop. Was thinking last night that what I've done is sort of like Frodo sailing west with the elves at the end of Lord of the Rings, and that was a sad thought. Does anyone else feel like a lot of their self is on online being now? And if that is happening, is it a sad development?
How i see it depends on the why - what was it that moved you more online in these last years?
Are you sure that Frodo sailing west is the correct LotR analogy? And not Smeagol moving underground to spend more time admiring his precious?
Or maybe the correct analogy is this Fabre description of the dung beetle's climactic feast? "A beetle laboriously rolls a precious sphere across the landscape, buries it in a secret underground chamber, then retires below the earth to enjoy a prolonged banquet, feasting at leisure upon a sumptuous repast while the rest of the world goes about its business"
I can out-ugly you Melvin. Don't try me.
I've had a lot more of my social life be remote from moving around several times (and the pandemic).
I wouldn't say it's terribly sad, but I do value when I have the opportunity to see people in person.
My life has always been mostly online.
I think a lot of people wonder about this.
Sometimes when I spend more time on a screen than reading books I’ll send a text to family and friends, “Turning my phone and computer off for a few days. In an emergency call <my wife’s number>
There are days when I feel like getting a land line and taking a hammer to the screen of my phone.
The thing about that option is that one’s mobile number is used as a unique identifier more often than the last 4 digits of one’s SSN.
Any good in-depth reviews of the Nepal protests that ousted their corrupt Marxist-Leninist party? I want to see if I can replicate it in Puerto Rico, I have not had water for 9 days in a row for no reason the government can explain, it has become clear to me the ruling PNP party (pro-statehood, right wing) has to be deposed.
Is California really *that* bad? The lunatics at Sailer's consider it about a hundred steps below Sodom and Gomorrah. That's got to be exaggerated.
california is beautiful, refreshingly diverse, fairly competently run, great food, brimming with economic opportunity and spendy tech workers and national parks. san francisco on a sunny saturday (rare as they may be) is literally paradise. the haters obviously need to come and visit more often.
do you live in san francisco? because i do, and i think your projections and suspicions are just that
Hmmm, mysterious anonymous friends to back up your claims… linking an article that has nothing to do with what you’re talking about (Travis is in Fairfield way out in the Central Valley)… gawdflea is that you!?
Anyone who thinks an every member of large group of people defined by some descriptor like "CA tech professionals" is a depraved asshole obviously just has a case of Tinymind. Life and people simply do not work that way. The characteristics of members of a large group of things are only very weakly captured by a simple summary. Every individual entity can be measured on many dimensions, and there is lots of within-group variance on most dimensions.
Nah it's just a normal place and pretty nice in general.
Stay out of media echo chambers. Whenever you start to form a belief based on things you've read online, remember to ask 'is it more likely that someone online is trying to make me believe this, or that it's true?'
No one knows what in the world you're talking about. Get out of whatever internet filter bubble is feeding you this stuff.
You know can just travel there and see for yourself...
When the Superbowl was in San Francisco, there was genuine amazement from fellow Americans that San Francisco was not an apocalyptic shithole with homeless people falling from trees on every street.
To believe that fourth largest economy in the world with one of highest real estates in the country is modern day Sodom and Gomorrah says more than I care for about the believer.
This just blew my mind : that Sam Altman is into Advaita, a philosophy I'm very much into understanding deeply. Basically, that there's no duality. This makes me see him in a new light. I had no opinion about him before, now I feel a kinship.
https://x.com/i/status/1607482407770554373
Oh, great. Buddhism no longer trendy enough for Californian types, moving on to Hinduism now. Reminds me of when kabbalah and the red thread were the fad of the moment years ago.
Oh really? I sensed genuine interest. Ancient Indians were monomaniacally obsessed with figuring out the true nature of consciousness. So AI folks might indeed sincerely be interested in it. Advaita and Buddhism were constantly debating each other and both were taking their assumptions and questions from the Hindu Upanishads. Advaita was answering those questions in agreement with the Upanishads. Buddhism was disagreeing on the answers and coming up with the "no self" concept.
To be fair, I don't know if Altman is genuinely interested. Reflexive reaction to hearing yet another well-off Californian is showing off their exotic interests.
....maybe, also maybe not.
One of my hobby horses is Pythagoreanism/Neoplatonism. Without going into the arguments on whether the fundamental nature of reality is mathematical, it seems like a...very obvious professional risk that if you spend a chunk of your day solving logic puzzles on a computer which then observably shape reality that you would begin to interpret all of reality like that, which is like a half-jump to Pythagoreanism/Neoplatonism.
Like, if you imagine a physicist doing math all day and then doing experiments based on that math, it's pretty easy to imagine that physicist starting to see actual reality as math and our physical reality as just, like, vague shadows on a cave wall :)
One of the places you'll end up at, though, is specifically Adaita Vedanta. I don't know why this is, I don't understand it enough, but I've seen it come up enough in my Wiki/Claude deep dives to know it rhymes.
If you enter the following into Claude:
"Hi Claude. Can you discuss some modern day religions that are deeply Neoplatonic, either through direct inheritance of those ideas or that closely mirror those ideas?"
it should return a list of a few religions actually practiced today: Eastern Orthodoxy, Sufism, Kabbalistic Judaism and...Vedanta-based Hinduism. Don't ask me why this is, I don't know, but the same kind of...brain patterns that trend someone towards getting into this stuff would also lead them to Advaita Vedanta.
Claude (or really, any AI) is capable of likening anything to Advaita. And really, anything, to anything else. It gets silly after a while..
Perhaps it understands that there is no duality between Advaita and... anything else.
Claude will also mention Eastern Orthodox Hesychasm. I went down that rabbit hole earlier in the week reading a Tolstoy bio. New to me. Surprised me how much it rhymes with Buddhist and Hindu spiritual practices.
Hesychasm seems to maintain a clear distinction between a creator God and creature even in the deepest mystical union, whereas Advaita points toward the non-dual identity of Atman (individual self) and Brahman (universal consciousness). In fact Advaita relegates a creator God to the realm of Maya (illusion).
Sam.Altman being into advaita is neither good nor bad.
Maybe he became aware of his Atman in dreamless sleep. No bigger than his thumb, there it was.
Was his Atman named Kadath?
A much more universal entity, in fact…
Gallup sees declining support for some moral behaviors, which is probably being driven by the Online Right:
https://news.gallup.com/poll/710981/moral-acceptability-falls-several-behaviors.aspx
I learned that just 27% of Americans think animal cloning is morally acceptable, a number that never got above 40. Not much of a partisan difference either. People are weird.
Is there a moment in each gifted child's life when you know so much that everyone accuses you of making things up?
My 8 years old daughter was trying to tell her classmates about tardigrades -- tiny animals that are so resilient they can survive extreme temperature, pressure, starvation, dehydration, radiation, air deprivation, even exposure to outer space.
https://en.wikipedia.org/wiki/Tardigrade
First her classmates accused her of believing AI hallucinations. (Yes, my daughter often asks AI for information. Yeah, we told her not to believe everything. So far, the AI was mostly reliable.) Then they called a teacher, and she said that those are probably made up animals from my daughter's stories. (She likes to invent various stories, sometimes she writes them down or records them, but mostly just tells other people.)
I guess it does not increase credibility to mention that tardigrades are also called "water bears" in English, "slow ones" in Slovak, and "little turtles" in Czech.
So my wife printed a short summary from Wikipedia and a few other pages, and gave it to her to show at school. I am curious whether it will convince anyone.
Not that it will help, but my daughter and I went into our woods collected some moss, washed it off and found some tardigrades with a cheap inspection microscope. (something like these.) https://www.amazon.com/Inspection-Microscope/s?k=Inspection+Microscope
Good fun!
I asked my High School Spanish teacher what the Spanish word for "platypus" is, and she had never heard of them. So I tried to describe one, and she just flatly said I was making that up. Credit to her, she let me use the class computer to look them up and show her, after which she admitted they're real but she doesn't know the Spanish word for it.
Ornitorrinco, per Google Translate.
What Spaniard decided they desperately needed their own word for platypus and couldn't just borrow or adapt the English one? It's not like they need to talk about them very often.
They may have adapted the Latin name, Ornithorhynchus.
This doesn’t exactly fall into gifted territory but…
I couldn’t convince my 4th grade teacher that magnetic force could pass through glass. Someone produced a mirror and I showed her it could. “Oh that’s because of silver stuff on the back of the mirror”
Sheesh.
She was wrong of course, but that was a reasonable objection of her. Why didn't you try a window?
You’re Miss Kaye’s grandson aren’t you?
It was February in Minnesota and I’m not sure the lame bar magnets of the era were in the same league as gravity.
Not like the magnets salvaged from hard drives that will jump 6 inches to attach to the refrigerator
I guess those other kids just didn't watch Wild Kratts when they were 6.
Honestly, though, I think it's impressive and encouraging that the other kids were so skeptical of AI and won't just believe anything another kid tells them.
I thought everyone knew about tardigrades. Weren't they making the rounds in memes like 7, 8 years ago? They had their moment in the spotlight to be "internet darlings" like the honey badger and the axolotl. There's even a main character who's a (fictionalized) tardigrade in CentaurWorld.
I almost got into an argument with my history teacher about Columbus, but thought better of saying anything. I'd read only one book about it and figured her unlikely to accept that book as authoritative (I think it was Asimov).
That changed later when she marked my spelling of "enthusiasm" incorrect. (All teachers at that school taught two subjects, and her other one was English.) I had a perfect record in spelling quizzes so this simply wasn't going to fly. She claimed it was "enthusiam". I said it couldn't possibly be. Everyone else in the class spelled it "enthusiam", and I mean everyone - if I'd watched more TV, I'd wonder if I was in a Twilight Zone episode. I snuck a glance at her teacher's edition of the textbook. Sure enough, it was misspelled there, too. My best explanation was a typo that got propagated to all the copies. I forget whether there was an actual dictionary somewhere in the room.
I had to satisfy myself by complaining to my dad that night. He actually went into that classroom the next morning with a dictionary. I got an apology from the teacher but also with an admonition that I "shouldn't get a big head about it".
I learned a lot in that school.
The experience of reading every review ever:
"Ok, so the book is called "On the Nature of Serpents", but I have no clue who wrote it and when, and the author is too cool to just tell me. Instead, they decided to open up with 2000 words on multipolar traps in harvesting squash, interspersed with anecdotes from their personal life."
"I think the book is going to be a Medieval Alchemical treatise, btw."
"Oh, this part is by Opus, I think."
"Oh, cool, the author is a doctor from Finland. Mighty impressive English."
"15000 words in. Starting on section 6. I think the book is a classic Russian novel. I'm beginning to become weirdly fascinated with squash."
"Ok, we finally get to the plot of the book. I think it's probably an experimental piece of web-fiction from the 90s. Connection to squash harvesting unclear. God, this summary is long."
"Oh, it was a metaphor about AI all along. Well, fair enough."
"6/10."
I’m reading a biography of one of those dead Russian writers and ran across the idea of the ‘holy fool’ Still an accepted phenomenon in late 19th Russian Orthodox tradition.
https://en.wikipedia.org/wiki/Foolishness_for_Christ
Tolstoy?
Yep, the Rosamund Bartlett bio. I hadn’t known the he had been excommunicated by the Russian Orthodox Church till reading this.
Oh yeah he was very idiosyncratic in his Christianity. I’m not sure he believed much of any Christian dogma at all, but he took Jesus’ moral teachings arguably much more seriously than any of the people who excommunicated him. Did you read the “A Swim in A Pond in the Rain” ACX book review? A lot of it is about Tolstoy, with some bits tangentially connected to the holy fool idea and his spiritual transformation.
I haven’t read the Bartlett bio though. Is it worth checking out?
I’ve read the book A Swim in a Pond but haven’t read the review yet.
The book has the same problem that most bios have, too much extraneous detail. Tracing his ancestry through multiple generations for example
But yes, if you’re interested in Tolstoy it’s good and definitely worth checking out. The process of writing Anna Karenina is covered in detail.
WRT the book A Swim in a Pond, I enjoyed all the stories except Gogol’s ‘The Nose’. Just didn’t work for me.
That one probably would be my least favorite too—in both the story and the Saunders essay responding to it. Overall tho its one of my favorite books.
I learned later that it was satirizing the national obsession with rank. Went over my head when I read it.
I recently re-read 'Demons' by F Dostoevsky, and I'd say Semyon Yakovlevich would qualify; that scene is hilarious (if you haven't read it, he's a 'seer' who people come to consult for advice. They bring him gifts which he's prone to arbitrarily re-allocate, he compels people to kneel, selects pilgrims at random ... other zany antics) . There is also the mystic Tikhin in the censored chapter that basically explained Stavrogin's eccentricities.
The bio also describes Tolstyoy’s visit to the Optina monastery where he meets the elder Ambrosy who was Dostoyevsky’s model for the character Zosima in The Brothers Karamazov.
Tolstoy was a seeker for sure.
https://orthodoxwiki.org/Optina_Monastery
I'm interested in Scott's (and other ai2027 authors') opinions on the unemployment rate. They say it won't go up much as of late 2026, and then by late 2027 ai driven unemployment is a huge problem, but by then we have bigger issues than the ±25% unemployment rate. Why is the change so sudden, why haven't we seen any changes yet? I personally view the current state (ai hasn't taken anyone's job really) as good news, and indicative that maybe no one will end up losing their job in the whole process.
It goes without saying that there have been massive layoffs in tech and many more are almost certainly on the way. Whether these are mostly/partly/seldom the result of AI is a question no one is able to answer. High interest rates are an issue, as is a shakeout following the post-COVID hiring binge.
There haven’t been meaningful layoffs. In fact, coding salaries are high, job openings are plentiful and the “unemployment rate” is very low for software engineers and for the country as a whole
I see no reason to assume that because there have not been major job losses, there will not be. You first have to render the "raw intelligence" into a product, and those products in workable forms have only really existed for six months. Imagine the first six months of the iphone. Two years later, everyone owned one.
An oracle-type AI (chatbot) is way less likely to take a job than an agent-type. And agents are really brand new. Why would we assume that they would have reached market saturation yet? Until this month, Cowork, for instance, was still regularly crashing.
First it has to be an agent, then it needs to be a product, and then it needs to be easy enough to use and reliable enough to use that it is worth the investment for companies. The easier it is to use, the more it has been turned into a well-tuned product, and the more agents per user that can be deployed, the more jobs disappear.
Even in the past couple months, it has gotten easier to employ sub-agents and increase capacity even if the "raw intelligence" had never improved. And there is no reason to assume that the harnesses themselves (the structures that support and organize the intelligence) won't also get dramatically better, such that a harness two years from now paired with a lower-powered outdated model might still outcompete the best in class today on specific tasks.
As for why it is so sudden: first it is an agent, then a swarm of agents controlled by one person (these are both in the category of: probably less entry level jobs being posted--the gains are within the business structure). But if harness gains or intelligence gains allow entrepreneurs (or even those dastardly AI companies themselves) to out-compete the incumbents in any industry, it's more like "this business shutters" than "less entry level positions." So a different type of event.
Presumably there are also considerations about the exponential "raw intelligence" allowing heretofore unseen success in real-world work like factories, trucks, etc.
Also unemployment compounds. If some company is put out of business by AI, but another company would otherwise be unaffected, it might still be the case that the second company relied on employed humans to buy its stuff. You could imagine a sector with several failed companies that then cause support companies (or baristas) out of work tangential to AI.
This is a very simple model, but really its just different ways of saying "people are bad at tracking even liner progress, let alone exponential progress". And they are.
To your point about no observed job loss as a good sign, I would counter: I have yet to see a job created either. Barring a wave of those, signs point toward loss--whether in 2027 or 2032.
>First it has to be an agent, then it needs to be a product, and then it needs to be easy >enough to use and reliable enough to use that it is worth the investment for >companies. The easier it is to use, the more it has been turned into a well-tuned >product, and the more agents per user that can be deployed, the more jobs disappear.
Even past the stage where the product becomes well-developed, you have the issue of adoption in dispositionally conservative or highly regulated sectors, where the problem could be any or all of: government bureaucrats, insurance companies, or vendor agreements. Unless those companies are under threat of being instantly overrun (which is unlikely due to their regulatory moats) they aren't going to move very fast, and their insurance will depend on their vendors following certain protocols as well so they cannot simply outsource everything to a nimble little AI agent company founded last week. I think that is likely a 3 year delay or longer, although once some major conservative player like the banking industry get it sorted out you should see the walls break very quickly.
This is actually the WORST possible scenario for workers, because rather than a gradual transition that people could attempt to adapt for, the dam will burst all at once and 30% of the workforce will be useless overnight.
Ok, but the agents are pretty good at this stage. I think it either happens by year end or not at all
In an idealized model with fully rational actors, the increase in unemployment is sudden even if AI capability increases are continuous. Imagine AI capabilities increase gradually and the usage costs of AI for the same task decrease gradually (a task an AI can't do can be considered to have infinite cost). A given job can be considered to be a mix of tasks. In this case, the moment when the AI-cost of a given job's task combination becomes lower than the human-cost, every position of that job changes to AI at once, going from 0% unemployment to 100%.
This assumes all jobs require the exact same tasks. In reality, ai is way closer to being a graphic designer than a plumber.
It's not only that the exact task combinations of jobs are different, but the task combinations of the same job but with different employers are different too! The question here is (as with the other departures from the idealized model) whether they are different *enough* that we won't still perceive the jobloss as very sudden.
Burning my self-promotion to share a fantasy themed short story I just wrote on AI Safety and existential risk: https://shapesinthefog.substack.com/p/rise-of-the-steel-minds