I recently made a video essay about The Rocky Horror Picture Show. I'm not doing a Substack because people who like movies want to hear about them on YouTube, and because I have very minor celebrity status from being on a reality show. I'm hoping to get the view count up to 3k, and right now it's sitting at 2.9k. I'd appreciate it if any musical theater fans here would give it a watch.
In this community, we’ve often talked about how it might be a bad idea to rely on medical advice from LLMs.
On the hand, the latest DeepSeek does a noticeably better job than my endocrinologist at interpreting my blood tests results.
I have a thyroid condition which needs to be monitored, and a quirk is that the local hospital’s endocrinology department will include T3 and T4 level if they order a blood panel for me, but my GP will only order T4.
So I have DeepSeek generated analysis if my chart, with a certain amount of “well, this would be a whole lot easier to analyse if your GP had remembered to tick the box for T3 on the lab request form.” I might remind my GP next time…. [The nurse practitioner at my GP surgery also gets it right when she orders the blood panel, and she has the added incentive that she’ll be the one doing the second blood draw if the hospital consultant sends the GP practise an email telling them to do another blood draw and measure T3 this time. This does sometimes happen.]
I supplement the medical system by using AI to research medical conditions and treatments, both before and after doctor visits. I sometimes do the same kind of research for a friends who are having health problems. It’s been enormously helpful. Here are some specifics.
-I have scoliosis which has recently progressed rapidly, and I need surgery to keep things from going downhill further. I also have osteopenia. Saw a surgeon, and we agreed on a procedure & set a date a couple months in the future. Researched predictors of good outcome and bone strength was a strong predictor. Discovered there are several new powerful ones on the market. Surgeon had said nothing about importance of bone strength. Canceled surgery with that doc.
-New surgeon spontaneously told me I needed one of the novel bone-strengthening drugs, and suggested I take it for 3 mos. prior to surgery, and 18 mos. after. Researched how rapidly bone changes take effect with drug, and discovered changes were minuscule after 3 mos., but that after a year 80% of the improvement that was going to happened had happened. Pushed surgery back to nearly a year after beginning drug.
-Internist prescribed Zepbound so I could lose 25 excess lbs before surgery. Since I do not qualify for insurance coverage, I was paying a lot for the stuff. Researched getting GLP-1’s on the gray market, found a way I believe is safe. (Cremieux gives the relevant details in a post from last spring). Began taking it. Researched Retatrutide, which appears to be a step up from Zepbound in effectiveness and health benefits but is not yet FDA approved. After another Lilly interim report on Retatrutide came out in December, I switched to grey market Reta. I am paying about $100/mo for a moderate dose. It works even better than Zepbound and has caused me no problems.
-Neither internist nor surgeon advised me on taking supplements while on these drugs. Since I am trying to build both muscle and bone, I knew it was important to get the recommended amount of protein and calcium. Discovered that is quite hard to do on a GLP-1 because you have little appetite and feel unpleasantly overstuffed easily, so food intake is small. Also saw in the research that subjects in studies of GLP-1’s were found to be consuming less than recommended amount of protein and calcium, and that many were vitamin D deficient. So I started keeping track of protein and calcium intake, supplementing with protein powder and calcium capsules as needed. Also started taking a multivitamin, vitamin D and Vitamin K2.
-I’ve had vertigo attacks occasionally for years, and started having more once on a GLP-1. Thought I saw a link between attacks and being a bit dehydrated. Checked with GPT and yup, decreased fluid in inner ear can cause vertigo. Though I don’t think I drink less water and coffee on Retatrutide, I do eat fewer watery foods (soup, fruit, etc.) because I eat less food overall. So started drinking more and checking my urine density with a home test. Vertigo’s gone.
-Friend was having immunotherapy for cancer. With GPT’s help I read up on it a bit, and discovered that immunotherapy infusions early in the day are for some reason markedly more effective than same infusion later in day. Person checked with their oncologist, who confirmed that this was true.
I could easily give a dozen more examples. But I think one reason my results are so good is that I don’t ask GPT for advice, I ask it to research certain topics and summarize the research and give links. And I usually check a few of the main links to make sure they exist and GPT’s take on the was accurate. I have not yet been in the position where my conclusions based on GPT research conflicted with the doctor’s and I had to decide whose to follow. I guess I’d have to make a judgment call at the time. But I would not rule out going with GPT rather than the doc. Setting aside individual instances, I’m not even sure in general which source is more trustworthy. GPT does not care about me, cannot examine me physically, and does not have years of actual medical practice in the background when it makes recommendations. On the other hand, it can access virtually every printed resource that exists, has infinite time to devote to my case and never feels annoyed or one-upped, as some doctors do when I want to discuss options. And my impression in the current medical system is that while the doctor cares about me more the GPT does, they only care slightly more.
This is not ChatGPT's fault, but for public reference, the immunotherapy timing claim is facing scrutiny. There's only been one randomized controlled trial of immunotherapy time-of-day, Huang et al. 2026. It claimed to find large effects, suspiciously large effects, and is undergoing investigation for concerns "regarding inconsistencies between the registration record of this trial on clinicaltrials.gov and published version the study protocol, as well as with some of the findings in this study."
There are retrospective trials claiming time-of-day effects, but their results are less consistent, and retrospective trials are easily confounded by effects like, "healthier patients schedule earlier appointments."
The advice is not necessarily bad. Nobody claims that earlier immunotherapy is worse, so switching to morning is a reasonable choice even if the evidence is weak. But I wouldn't confidently claim that earlier immunotherapy works better.
GPT actually didn't make a big deal about the early morning infusion advantage. I was just struck by it, in there among the other article summaries it made for me.. Given that there's no harm in being one of the early birds, scheduling infusionsin the early morening seems like a no brainer -- can't do harm, might help.
There are 3: Forteo, Tymlos and Evenity. Forteo and Tymlos are similar, and I have been on both. Don't know much about Evenity. All are much better than Boniva, which was what was used til recently. What Boniva does is prevent the body from getting rid of old bone (apparently there's constant turnover in bones, with old material getting replaced by new). This does gradually make bones stronger. But the new class actually stimulates the formation of new bone, which works much better.
Forteo and Tymlos are expensive (2 or 3 thousand per month). To get insurance to covert them you have to be diagnosed with osteoporosis, *or* have osteopenia plus at least one spinal fracture (or maybe one fracture anywhere?). I have osteopenia and no spinal fractures, but my surgeon just winked at me and said "A mess like your spine probably has some fractures in it somewhere" and checked off the fractures box, and so mine's covered by insurance.
Main side effects are joint pain and fatigue, and I have both. Putting up with them is worth it though.
E.g. the drug dosage I was taking brought T4 within range, but left T3 still too high. They should have increased the dose to bring T3 withn range, too.
How certain are you that this advice is correct, as opposed to superficially plausible?
(I don't know much about thyroid disorders, so maybe it is obviously true that a medication which brings T4 within range but not T3 should be increased until T3 is within range. Is that the case?)
These are good questions as to what the ground truth is, as LLMs wildly make stuff up, I ‘m not an expert just the patient, and my own cognition is probably somewhat suspect as I’m suffering from a thyroid condition. On the other hand, when T3 was finally brought within range, I felt a whole lot better, and there’s an a clear mechanism for why that happened.
Regarding the recent discussions on simulation mechanics—the OSIM Research Group is moving away from the 'Architect' or 'ASI' hypothesis. We’ve found that focusing on the 'Who' is a distraction from the actual physics. We’re pivoting to a Forensic Cosmology stance to document the physical stabilizers like Many-Body Scars and the 'Non-Algorithmic Wall.'
Here is our updated 2.1 framework for the Sovereign Blueprint:
OFFICIAL TECHNICAL ADDENDUM: MARCH 31, 2026
The OSIM Research Group is officially transitioning to the "Sovereign Blueprint"
framework (v2.1). This update prioritizes a "Who-Neutral" stance to focus on the
observable mechanics of the Sovereign Inception.
The "Something" keeping the loop going is defined as a self-correcting,
physicalized program embedded within cyclical cosmology (The Big Bounce).
1. THE QUANTUM SHAKE: Many-Body Scars (OSU research) show a system
hard-coded to return to its original state. This is the stabilizer
for the Loop.
2. THE NON-ALGORITHMIC SEAL: UBC research on the "Wall" and biological
memory retention (flatworms) proves this is a physical sanctuary,
not a digital simulation.
3. THE 20-WATT RULE: UChicago bio-electronics confirm biology is the
most efficient hardware possible for housing consciousness.
We are looking for fingerprints, not just an Architect.
If new research finds that the Big Bang is actually a Big Bounce and some researchers are suggesting that it could be infinite then it becomes a big numbers game. If you and I are having this conversation right now and we know that 1x infinite = infinite then its not just likely but mathematically guaranteed that we will have this conversation again. this would be a infinite biological loop.and if so would kill the nick bostrom digital simulation and also the multi universe theory . It is flasable , if it's proven that the universe only expands then we know the big freeze wins out and osim is wrong or if it can be proven that silicon or another material if more efficient and can mimic life better than biological then again osim would be proven wrong.
Hi, I and my wife are strongly considering sending our 5.5 year old son to a new campus of Alpha Schools opening in Palo Alto, CA this fall. I suppose he falls into the Gifted category, since he tested as 139 on the WISC-IV. He is in KG now (on the young end wrt his peers) but he is far ahead of his peers in reading and math (roughly 3rd grade level). This is mostly cause I spend a lot of time with him at home and fwiw he enjoys learning math through the Khan academy site, so it may align well with the 2h model. We took him through the usual round of elite private school and Gifted school applications and he didn't get accepted to any of them. Not entirely sure why, but it might have something to do with either having messed up the timing (basically we would have had him repeat KG, which i guess committees are wary of) and his teacher recommendation letters probably say something about him being on the disruptive end of the typical KG behavioral spectrum. In any case, it seems like the options are keeping him at his current private school (academically focused, but not a fancy elite one, and also not having the kind of peer group that can challenge him), and taking a shot at Alpha for a year.
I really liked the philosophy behind Alpha, the review that was written last year on ACX (https://www.astralcodexten.com/p/your-review-alpha-school), and honestly I know the marketing videos are slick, but somehow I get a vibe of them being not a scam, and having some sense of honesty about what they're doing. Some of the comments from parents and students in the linked reddit threads were a little disappointing BUT I wonder if it's something we dont have to worry as much about cause quite likely our kid will be in the upper percentile in learning, so won't have to worry about the pressure they supposedly put on you.
Another relevant fact about our kid is that he is incredibly physical and loves sports and hands on physical activities in general. So I'm really hoping that the 2h of (device-based) education, which frees up the afternoon for life skills, projects, sports etc etc would be a good match for him.
Anyway, I'm just wondering if anyone has any info to share, especially recent info since the review came out last year? I'm really hoping current/former parents/students or the author of the review are available to ask questions to.
I don't think Alpha School is a scam, but I don't like the sound of the place. Bear in mind that the person who wrote the review, whose daughter attends Alpha, is very invested in his kid being publicly known as Amazingly Smart. He named his daughter Everest, and now writes a blog about her becoming a champion at a thing called History Bees, where kids standing in front of an audience answer questions about history factoids. Much of Everest's studying involves Anki decks. That's all OK, I guess, but it is very different from real learning about history -- what life was like in other eras, how people thought and felt and acted, how big trends and forces and coincidences and rulers' quirks shaped the course of events. And becoming the best kid on the planet at remembering history factoids takes up a lot of time that could go to real, immersive learning about history and other things. And the history bee stuff seems to have shaped Everest's interests and goals. Last time I could stand to read the blog Everest was asking to begin preparing to compete in geography bees. Everest is in training to be an achievatron.
Competitions are fine. They're fun and good motivators for many, though not for kids who get miserably anxious under that kind of pressure or kids who get dejected instead of bouncing back when they lose. But they're only one way to motivate and involve kids, and as I remember, most of Alpha School's afternoon activities are competitions of various kinds.
I highly recommend you check out Jeremy Howard's podcast interviews about his experience with homeschooling his daughter. (Here's one of many: https://teachyourkids.substack.com/p/an-interview-with-jeremy-howard-about) He and his wife are using a combo of curated online learning apps and online group activities for other kids. It's done under the umbrella of a worldwide homeschooling organization called Modulo that offers all kinds of resources. It's here:
https://teachyourkids.substack.com/about". We’re a directory of vetted homeschooling resources, newsletter, show and curated online community for homeschoolers and modular learners. We’re also a growing educational marketplace with an AI concierge that offers tailored support to families navigating modular learning and homeschooling."
Beast Academy is a pretty good self learning course in math for kids who are interested in that subject. I know some kids do great in kindergarten but get really bored in first grade when the academic content ramps up and it stuff they already know.
If you have time to spend with your son and want math stuff, my top recommendations for that age are:
(1) games, especially those in Ben Orlin's "Math Games with Bad Drawings." Top board/card games I suggest are go, hanabi, king domino, santorini.
(2) logic puzzles, especially stuff from Raymond Smullyan.
(3) redacted contest questions from Math Kangaroo and AMC 8. There are two levels of redactions I like:
(a) remove the multiple choice answers: this makes the questions much harder and, slightly more natural.
(b) remove the answers and remove the question. This converts it into a 'notice & wonder' prompt where the two of you can talk about what you observe ("I notice X") in the situation and what questions you have about it ("I wonder y?")
The one I tried was 1) from Eramolalos below - asking for clean, unadorned links. It seemed to work, but I didn't do a lot of testing, so reluctant to post as a solution.
Did you actually click on links or just look at the text? gemini will give text that is a direct url, but the hyperlink is actually to a google search page.
So I am skeptical Gemini gave you useful solutions, and still think it gave you incorrect solutions.
I went through this runaround with gemini, and now I guess I need to do it again with a human! Here's how Gemini would respond: "You're absolutely right, I didn't verify my links!"
When I have asked Gemini for help finding or doing something online, its instructions have been wrong far more than half the time. Picky of me to complain, of course
Out of curiosity asked GPT-5. Here's what it said:
Short answer: you generally can’t fully force Google Gemini to stop wrapping links in Google redirect/search URLs. That behavior is partly intentional (tracking, safety filtering, and keeping users inside Google’s ecosystem). But you can reduce it or work around it.
Here are the practical methods that actually make a difference:
1) Ask explicitly for raw URLs
Gemini will sometimes comply if you’re very direct:
“Give me direct URLs only (no Google redirect links)”
“Provide clean, canonical links (starting with https:// and the site domain)”
“Do not use Google search or shopping links”
This works inconsistently, but it’s the simplest first step.
2) Ask for “source citations” or “publisher links”
Prompts like:
“List sources with direct publisher URLs”
“Cite sources with outbound links to the original sites”
push it away from Google-wrapped links and toward real domains (NYTimes, NIH, etc.).
3) Use the “double prompt” trick
First:
“List the sources you used (names only)”
Then:
“Now give direct URLs to those sources (no redirects)”
This often bypasses its tendency to embed Google tracking links in the first response.
4) Strip the redirect manually (reliable fallback)
This is peak post-AI discussion. Someone asks something about AI, then someone just tells them to ask the same AI about it. Why do we even bother having forums?
But in this case, I have been asking the AI, without success. I've added permanent instructions, asked it to triple check, asked it to never under any circumstance give me a google search link. It assures me its response has no search links, then proceeds to dump a list of 10 links, about 8 of which are search links.
I'm curious to see if the poster actually verified their solution or just assumed gemini was correct.
A take: socialism makes sense from that exact point when capitalism gets to the point when it resembles socialism a lot.
That means, corporations merge and grow, until they are huge bureaucratic monsters, every decision made by a committee selected by a committee, always making the safe protect-your-ass choice which is just applying Harvard Business textbooks, which means there is no real competition and no real room for any kind of a heroic entrepreneur anymore.
There are multiple different readings of Marx, because he seemed to be a bit conflicted about this, he was on one hand very clear that one cannot ever jump over stages of historical development, one must wait for the right time, but on the other hand seeing poverty and injustice everywhere he was tempted to hope for a revolution coming quickly. But that was mostly his emotions. His model of economics was absolutely about waiting for the conditions to ripen.
We are clearly not there. Lei Jun very much pulled the classic heroic entrepreneur move with personally saving Xiaomi, moving his desk into R&D, personally micromanaging the entire R&D and supply chain, and put $10Bn into designing an electric car. Very much Steve Jobs level entrepreneuring.
Very few things seem less intellectually productive to me than the way so many people insist on squeezing all of their economic thinking onto this capitalsm-socialism binary, as if those are the only two economic systems that have existed or could ever exist.
Marx had some decent insights, but the "stages of historical development" think is plain bullshit. There may be trends and attractors in history, but there's no guarantees, no set course that it will definitely follow and definitely no world spirit guiding it.
I agree. There are specific problems; there are specific solutions; there are specific problems these solutions introduce, etc. Any debate that does not look at these details is just cheering for "your team".
Sometimes a hero solves a problem. Sometimes a psychopath creates it or makes it much worse. Sometimes regulations prevent horrible things. Sometimes regulations prevent solving horrible things. Sometimes we need to incentivize those who produce desirable results, and punish those who make things worse. Sometimes we need to give people a second (or even the first) chance. Sometimes money is wasted or stolen. Sometimes the attempts to prevent the waste or theft waste even more money. Sometimes what we need is a new institution. Sometimes what we need is to leave people alone and let them organize their own lives. Etc.
Also, if we assume that, ideally, socialism is governed by workers' councils, and capitalism exists in a democracy where people can vote for their interests... seems like the fundamental dichotomy kinda disappears? I mean, if the workers' councils are free to adopt any "capitalist" pieces of the puzzle, and the voters in democracy are free adopt any "socialist" pieces of the puzzle, then in either case we are collectively free to build the entire puzzle as we want it, right? -- Unless we point out that in *real* socialism, the actual power is in the hands of the few individuals on the top of the Party / army / secret police; and in the *real* capitalism, the actual power is in the hands of lobbyists and media. But then it becomes a completely different kind of debate. Plus there is the question of how to implement anything resembling actual democracy if most voters are functionally retarded and will happily vote for whatever meme that today happens to be most successful on social networks.
I think it's the other way around: socialism (or rather, central planning, to be more precise), in the most optimistic case, makes sense to the extent corporations manage to reproduce it.
Amazon might be a bureaucratic monster, but whichever force keep it a dominant player (scale, internalization of spillovers, foresight, etc) clearly still make it more efficient than any competitor (including all the *potential* competitor that do not enter). So if you were to propose the state run a program with the scale and complexity of Amazon operations, at least in theory it would work (whether the incetives are there to make it work it's another matter).
But there never was a corporation that micromanaged the entire economy of a geographically-determined area (say, a region). Many have tried and it ended in disaster, or they made bank only because they could extract resources violently. So having the entire economy centrally planned must not work, otherwise you would have some perfect Town Inc. which people are eager to join.
Who said anything about the state running anything? We are already running businesses with collective intelligence. We don’t need anyone’s permission.
What we want to do is decentralize government itself. At least the power structure of it. It does not seem by your comments that we understand each other well.
Amazon gets special treatment from the government. So does much of Wall Street.
A different Amazon could be made with a completely different organization chart that has nothing to do with government.
Read the book Nobody is Smarter than Everybody if you want to see a $300 billion dollar organization operate this way.
Well, what I care for is the concept that technological progress makes economic systems obsolete and then new ones are needed, one cannot stick to the same system forever. However the new system must flow from the logic of current technology, and not from conscious human effort. It should be something that just falls out from the current level of technology. If it is not called socialism but something else, fine. The only two rules are that it should be more humane, and it should follow automatically from the current level of technology, without much planning and tinkering. "gamification" sounds like too much tinkering.
It is corrupted. The playing field is not fair. The iron boot of government is used to select winners. That is a successful system to you? Either you are lying to yourself, or your definition of successful is less than mediocre.
A good system should be allowed to be given to your enemies and they still can’t corrupt it. Our systems are fully corrupted.
A fully corrupted computer is a failed computer. Don’t trust, verify.
If that's your standard for "corrupted", then every system controlled by humans will become "corrupted". Capitalism is the best one we tried so far, and most importantly, regulated capitalism has been demonstrated to be compatible with democracy, in practice and on a large scale. No other economic system can say this. And if you claim to have found a new system which won't become "corrupted", but which has never been tried before, or an old system that has never been tried with more than a million people at once, then I'm very sceptical.
As you should be. But it’s not hard to imagine ways to make our systems harder to corrupt. Just because capitalism did a lot of good, doesn’t mean that it still is, or that there isn’t something much better out there as an option.
And yes, the standard should be that corruption is a failure and we should not tolerate it. A system that allows literal pedophiles to give sweet crony deals to each other should be heavily criticized. Our tolerance for corruption is the problem. So yes, we strongly believe that the success of system should be measured in corruption too. And harm done.
There are many systems that are antifragile. Many many examples we give in our articles. Your attitude of Nullius in Verba is a good attitude. But what happens when you don’t take our word and we show you that we are right with a new system.
Will you become an advocate then?
The free market right now is at serious risk of being turned into a violent digital control grid for humanity. Holding onto the corrupt system is going down with the titanic, friend.
The sooner we admit that the better chance we have of building a big escape raft for humanity.
It's not hard to *imagine* ways to make our systems harder to corrupt. It is very hard to actually *do* this, and disturbingly easy to screw that up and make things worse.
This would be a good time for you to explain in some detail exactly how you propose to make the system less corrupt. If the answer is just to give the government more power and tell them to use it to suppress corruption outside the government, I will be unimpressed.
Here's another way to look at it: corporations merge and grow until someone notices that their lumbering size makes them less efficient than a smaller competitor at solving whatever problem they were organized to solve, and proceeds to create that competitor. The solution to a herd of giant bureaucracy-hobbled corporations is to create smaller companies that take market share from the giants by providing customers with better services or goods.
If the problem customers observe is hidebound bureaucracy in corporations, then smaller, nimbler companies will have a natural advantage from the customers' perspective. Another elegant part of this solution is that if the customers turn out to prefer the larger corporation's product - for example, because it is more reliable due to all that bureaucracy - then the customers can continue to buy from the giant corporation.
The socialism alternative is to entreat an even larger and more bureaucratic government to oversee the giant corps, which looks a lot like trying to solve a rat problem by buying an even bigger rat.
Another alternative is for someone working at that giant company to notice the bureaucratic overhead, notice way to reform that overhead, and improve that giant company from within, leading to a more attractive product for consumers, and a larger profit for that company.
I think of this as the overlap between the socialist calculation problem and Coase. On one side, as your operation grows, trying to do central planning on it ends up requiring more information than you have or can get, and so you end up doing increasingly inefficient things even if all the decisionmakers in your organization are well-intentioned. (Which they are certainly not--see the principal-agent problem.) On the other side, trying to negotiate everything out via contract/on a market gets inefficient in a different direction, where you want to have (say) a shared office space but need to negotiate out who gets which space and what the rules for the shared space are. There's some kind of crossover (very dependent on local conditions and technology and such) where you stop getting more efficient by getting bigger and getting rid of the negotiation costs between different parties involved in (say) making a car, because now the inefficiency of the central planning starts costing you more.
Pretty close to how I think about it, too. You can fend off the principal-agent problem by aligning goals with natural incentives and keeping an eye on information flows, but as you hint, that gets harder to do as the project enlarges. Contracts have a similar problem where you can over- or under-manage, and again, can be staved off by picking people who naturally cooperate (one extent on which cultural diversity can actually hurt). If the staff generally prioritizes project over little details and nest-feathering, then the shared space contract is two paragraphs. If not, it's 30 pages and an extra $100K for the lawyers, wheeeee.
Hopefully, the big corporations don’t figure this out and cleverly buy up the small upstarts before they become big enough to pose a threat, that would be terrible!
"Here's another way to look at it: corporations merge and grow until someone notices that their lumbering size makes them less efficient than a smaller competitor at solving whatever problem they were organized to solve, and proceeds to create that competitor. The solution to a herd of giant bureaucracy-hobbled corporations is to create smaller companies that take market share from the giants by providing customers with better services or goods."
... until the giants buy them out. Lovely profits for the founders, shame about the better services and goods, since they tend to disappear into the giant's internal systems.
There was at least one railroad builder who made an entire career of starting a railroad company, building a better railroad than the dominant competitor, selling it to that competitor, and then starting yet another company somewhere else, with the same model. So this is a thing that can happen.
To the extent it happens at all, the giant corporation takes a loss, unless the smaller company is deliberately selling for less than its value (which is possible if the owner is genuinely tired of that company or sees an even greater opportunity elsewhere and wishes to divest quickly, but those are exceptions, not rules). The giant may choose to bolster that loss with revenue from other locations, but if multiple small companies compete directly with the giant in multiple locations, the giant cannot bolster everywhere.
Thomas Sowell and David Friedman (a reader of this substack) discuss multiple examples in their books.
There is also a pattern of startup companies angling to get bought by some giant company to kill a competitor/hire on top-tier talent, which is a similar idea.
If those founders make lovely profits, shouldn't one expect more people to emulate them, until those inefficient giants can no longer afford to buy all the competitors?
I was with you until “government”. Socialism is about worker’s councils running things, not the government running things. That was a Soviet travesty, something Marx explicitly opposed: https://en.wikipedia.org/wiki/Barracks_communism
If an institution is properly described as "running things", that institution is in fact "the government". If the institution runs only some things and it's clearly understood that it *won't* be running other things, then you may be talking about something less than a government, but it's on you to make it clear what is and what isn't being run by the workers' councils or whatever.
The actual history of socialism is, if you try to have worker's councils "run things" in that sort of open-ended way, you get very bad government. More generally, whenever you try to pretend that the institution which runs things isn't a government, you get very bad government. Something the anarcho-capitalists should be very wary of, but I digress.
> If the institution runs only some things and it's clearly understood that it *won't* be running other things
...then we also need some *other* institution that decides who runs what, and enforces that decision. And then the other institution is the government.
Yeah, you can have your economy organized around worker-owned collectives and guilds and such, but then you need someone to arbitrate between the collectives/guilds and enforce rules on them and decide what is a legitimate guild or collective and such, and that ends up looking a lot like a government.
One way you can escape some of that is to move more and more stuff outside the range of collective decisionmaking, the way Western countries have mostly moved the question of "which religion shall we be" outside of a collective decision to be made by the state or king or voters. But that's just getting to a place where, in some sphere, nobody is "running things."
There is no government policy preventing worker's councils. Worker coops are free to exist and there are many. When they provide better goods and services than their competitors, that's great. And when they don't, then customers are free to purchase goods and services from non-worker owned businesses. And that's also great.
Question. It seems to me "cancel culture" started with Lawrence Summers Harvard President, 2005, when he was talking about an innate difference of mathematical ability between women and men. I dug around a bit, and it seems Summers did very very questionable financial moves: https://www.institutionalinvestor.com/article/2btfpiwkwid6fq6qrokcg/home/how-harvard-lost-russia so it seems a lot of people had an actually good reason to get him fired. So in that light, how can that case be interpreted? Is such a talk a worse "crime" than these financial scandals, or more like it was the last straw that broke the camel's back?
Here's Gemini's summary of Summers' main point: "The underrepresentation of women in high-level science and engineering positions might be due to a combination of innate aptitude differences (specifically in variability of mathematical ability) and lifestyle choices, rather than solely discrimination."
Why is expressing these thoughts a crime at all, or even bad? What he's saying is quite plausible.
If you are the leader of a big institution, and you say to a large group of stakeholders that they are being discriminated against rationally, those stakeholders will not believe that you’re going to do anything about that discrimination and they will see you as an obstacle to their goals. There’s a difference between speaking as an economist and speaking as a university leader.
Who do you think should be allowed to make such an argument? I mean, if Summers was forbidden by his position to make it, then would a tenured professor be allowed to make such an argument? A tenure-track professor?
My strong suspicion is that there is nobody who would not have been attacked and cancelled (if possible) for making that argument. If that is true (ask Amy Wax and Ilya Shapiro why I think that), then it seems like the actual rule is "nobody may bring up this side of the argument" and then whenever someone does that, you search around for an excuse (like saying that James Damore's memo required firing him to avoid discriminating against women) to punish him.
That may be true, although the relationship between how willing someone is to be truthful and how much they are trusted is not simple. In some cases people give up forever on a person willing to speak unwelcome truths, in others people trust them more. Also, note that Summers did not deny there is discrimination against women, only that discrimination might not be the only reason why there are fewer women in STEM fields. Somebody making Summers' points could have restated that, and then continued with suggestions for ways simple discrimination could be reduced, such as excluding info about gender from grad school applications.
> excluding info about gender from grad school applications.
Everyone knows the consequences of implementing this obvious idea would be the opposite of what people loudly aver, so they've switched to denouncing this proposal as a particular insidious form of sexism/racism/whatever.
I'm reminded of Carl Sagan's story about the guy who claims to have a dragon in his garage. It is, of course, an invisible dragon, and when various tests are proposed to try to detect it, it also turns out to be perfectly silent, intangible, etc. The point is that even he knows he's lying, evidenced by how knows to head off every avenue that'd prove him right.
Well, then the speaker can vow to back enthusiastically any idea that the group proposes for making sure grad school admissions are done in a gender-blind way. Then the group can sit there, each contemplating the same obvious idea I proposed, realizing it will lead to fewer women being admitted, and stewing in their own juices.
I was at Harvard at the time – tl;dr, Summers was already intensely disliked by a lot of people at the university, and they leapt at the first opportunity to get rid of him.
I've been at 5 universities or colleges, and the guy running the place has been hated in every single one of them. It may be that's pretty much guaranteed to happen because college-age people are easily led to see an authority as a controlling fool, and career academics are backbiters.
I suspect this is a very common pattern in many of the high-profile cancellations. The stated reason for cancelling the person was often more of a pretext or a focal point for their enemies who already wanted to get rid of them. That wasn't always true, but it sure seemed like it often was.
In what sense has JK Rowling been cancelled? She is still very much in the news, they're making a new Harry Potter TV series, and nobody's stopping her funding her current charity ventures. Is there something she wanted to do that she has been prevented from doing, and if so, by whom?
The Iran war has demonstrated that the balance of power in naval war has shifted to the land side. Cheap and numerous drones and missiles launched from the land can now control the sea to quite some distance away from the shore, denying entry to enemy ships. Launchers can be mobile or dug in, meaning that even with US-style air supremacy you can't bomb them all.
While this is bad news for the Middle East, is it not good news for Taiwan? How could a CCP invasion force possibly make it across the Taiwan Strait in any significant numbers if the Taiwanese can lob thousands of cheap drones at anything that floats?
The Persian Gulf is a very narrow, very shallow sea with extremely stable weather (and the Hormuz strait is even narrower and shallower). Basically the ideal conditions for such land-based hardware as you describe. Not sure you can draw very useful conclusions for the general case.
I'd say you could draw some very useful conclusions from the Ukrainian example. The Ukrainians were able to sink about a third of the Russian Black Sea Fleet and drive it back to their ports, using a combination of air drones, sea drones, and Neptune missiles. Soviet radar systems, which seemed to be good at dealing with a small number of incoming objects, were not up to dealing with Ukraine's mastery of radar systems (which most people seem unaware of) and fake-out tactics. While the Black Sea isn't comparable to the wide open oceans, it is about 1200 km across. Taiwan is only about 160 km from the mainland. Heck, Ukraine has steadily extended its drone kill zone. Last year it was about 50k, but their new models will extend the kill zone to 150 km. Given a small range extension, the Chinese coastal installations will likely be within Taiwan's aerial drone kill zone. And if they haven't already, they could easily recreate Ukraine's surface marine drone technology (such as Sea Baby and Magura). And sea drones are particularly effective in very choppy water because the chop masks their approach from radar.
Moreover, much of China's "cutting-edge" military technology is very likely a paper tiger. After US technology so thoroughly took out Chinese weapon systems in Venezuela and then Iran, several CEOs of China's major defense companies have been detained and are under investigation, with hundreds of managers, engineers, and accountants being questioned by their security services.
And now that Xi has decapitated the PLA by removing two of the PLA's senior-most generals, Zhang Youxia and Liu Zhenli, upsetting regional PLA commanders and the old-guard in the party, Xi would be foolish to try to mount an attack on Taiwan (of course, he might pull a Trump, but he seems smarter and more emotionally controlled than Trump). Also, Zhang became the focus of Xi's ire, because he had given assessments to senior party leaders that an invasion of Taiwan would be unwise. Currently, Xi is in a political standoff with the party elders and the military, and has been unable to replace Zhang and Liu with generals loyal to him (of which a few remain). And since Trump canceled his State visit, the Chinese state media has been talking about a peaceful solution to unification. No, we won't see an invasion of Taiwan anytime soon.
It's a valid concern (and not just a theoretical one, under our current administration!), but I think it would be fixable with a law against government officials investing in prediction markets. Similar to the proposal to prevent congressional insider trading by not letting them hold anything but index funds.
While prediction market proponents often argue that insider trading is a good thing - it technically makes the market more accurate - I don't think that counts if you can literally control the outcome of the event you're betting on. The goal is to make the market match the odds of the event, not adjust the odds of the event to match the market.
Hmm I don't think it applies only to government officials. Anyone with enough money can swing the outcome. Isn't this already happening in sports betting? The $ value in the market creates it's own incentive.
On Kalshi, the actual public IPO announcement (not the IPO itself, just the announcement) only gives a 25% chance of that happening in the next 2 months.
Hmm, I saw my barber catcalling some tourists in bikinis today. He made some kind of growling noises at them. Possibly the first time in my life I've seen catcalling in the wild. I feel this was ethically wrong, though in the moment the situation felt amusing to me. Should I switch barbershops over this? I was also thinking he's ultimately a low-IQ working class guy, so less is to be expected of him, but then, I know from personal experience being low-IQ and working class doesn't mean you have to lack a moral compass or something.
Do you like the guy? Do you have a good rapport with him?
If so, then a low-stakes “Man, why’d you do that?” in the moment followed by a conversation where you say stuff like “What, you think they liked it? It probably just made them feel gross.” can probably do more good for his attitude than quietly switching barbers ever could.
(Bonus points if you can code-switch into more working-class style language without it feeling forced.)
Do whatever you want tbh. Feels more like a question of class and taste. Like if your barber farted while cutting your hair you might just get the ick and not want to go back there. But I personally don't see any serious moral dimension to it, and even if there were how would leaving (and without even explaining why) help anyone in any way?
I live in a predominantly hispanic area, and when I socialize with working class guys I just kinda know to code switch into being a bit crass about women especially compared to super liberal intelligentsia (though, never to the point of catcalling). I also don't think the latter are especially more moral, they just are more performative and careful to not commit wrongspeak
I don't think it is immoral or unethical, I think it is against etiquette, against good manners, basically rude, which is exactly what low-IQ working class types do not care about.
I mean I find it very strange that in the last maybe 15 years the question of etiquette simply disappeared and rudeness is now considered immoral. This is basically the "therapeutic culture" belief that unkind words cause some kind of a damage.
In a 1946 fashion show held at a Paris swimming pool automotive engineer turned fashion designer Louis Reard - how’s that for a career 180? - introduced the bikini to the world. As this show was four days after the highly publicized US atomic bomb test at Bikini Atoll, and he had given the swimsuit its name at the last moment, an early urban legend held that the name was due to the fact that the test had blown out the middle of the atoll while leaving the ends intact (get it?)
In fact, Louis Reard had simply liked the sound of the name, and in any event the blast did not affect Bikini Atoll’s geography in that way.
Sidenote 1: Several professional models refused to model the (yet unnamed) bikini due to its immodesty. Reard ended up hiring Micheline Bernardini, an 18-year-old cabaret dancer who didn’t mind wearing one because in her occupation she customarily wore nothing but a smile. She is still around at age 98.
Sidenote 2: the military figured the Bikini Atoll test would be a good way to get rid of a number of warships that the end of World War II had made redundant. As things turned out, fewer ships than expected ended up sinking, and disposing of the highly radioactive ones than remained was quite a chore.
You must not be a software developer, as that is quite the assumption you made. They were walking down the street, this area is right next to the beach, seeing women in bikinis is not an unusual sight. Barber was sitting in a doorway, I was heading home.
It's "are the women actually upset". There's no description given to their reaction, to the point where it's not clear if they even noticed this guy. Eremolalos already commented on the purpose of a bikini.
Re "the purpose of a bikini": Even when we assume that those women wore bikinis to draw attention to their bodies, it doesn't follow that they want to or are comfortable with being catcalled.
Re "are the women actually upset": How would you know that from "how far away from the beach" they were? Or even from their visible reaction? If I were a woman and annoyed by catcalling, I'd do my best to not show a reaction.
Using your economic power to coerce the lower-classes into conforming to your sense of morality is one of the many perks of being rich. It's the same principle as lecturing panhandlers about sobriety and fiscal responsibility, or missionaries requiring pagans to convert before giving them food or medicine.
As unkindnesses and intrusions go, it's tiny. For instance I expect a lonesome person who attempts a bit of friendly chat about the weather with a cashier and gets ignored feels much more distress than the women in bikinis did. If you hold miscellaneous strangers to this high a standard you'll end up shut up at home. What the barber did is just a very politically incorrect instance of unkindness. Why not say something that might make the guy slightly less likely to catcall? Something like "my daughter says girls think men who catcall are losers. I dunno though -- do you think that's true?"
Haha, I feel if I actually confronted him about this and held my ground, I could end up in a brawl, which is not something I want. I'm probably gonna switch barber's over this, Nobody Special's right in that it's better to give my money to decent people.
All of this seems altogether wrong-headed to me. First of all catcalling women in a bikini seems like kind of an edge case. If a man catcalls a woman in ordinary street wear, he truly is being rude and intrusive. Bikinis, though are *meant* to be hot and edgy and attention-grabbing. That’s the whole point of them. Nobody wears them because they are more comfortable than ordinary swim suits (they’re actually less comfortable, because they have to be tight to ensure that they do not slip half an inch and uncover a bit of nipple or labia). They’re worn to be hot For a man to catcall a bikinied woman seems more like playing along with her than like shoving their uninvited horniness at some woman who’s peacefully going about her day. And I doubt that many women in bikinis are startled and offended by comments, whistles, catcalls etc.
Second of all, even catcalling a woman in street dress is pretty low on the rudeness and unkindness scale of real life. You see multiple examples of worse rudeness and unkindness in 5 mins in any public place. When you’re driving you see people prioritizing their speed and convenience over other people’s safety and peace of mind in at least a smallish way almost constantly, and seriously risky versions of this are not rare at all. In all kinds of settings you see solitary frail elderly people struggling to lift things, move things, reach things, and a stream of the able bodied striding right past, chattering on their iphones as they go. And then there’s the news.
And third, your response to the barber’s catcalls is a non-event in ethical space. Finding a new barber does nothing whatever to reduce the chance the original barber will continue doing something that, in your eyes at least, is so bad it takes him out of the category of decent people. It probably isn’t even sparing you much distress, since it’s a rare event —apparently he hasn’t catcalled anybody before in your presende. All it’s doing is giving a bogus little boost to your sense of righteousness.
My barber is a former golden gloves boxer. Still pretty sure if he did something like that I would still have given him a sharp, “Hey, don’t be a bonehead!”
How about mildly saying “Eso simplemente no está bien.” You would get the message across and I doubt he’d get bent out of shape. He might think you’re a dork or something but who cares? You stood your ground and can wear the dork badge with your own values intact.
Ultimately your call, but for me I'd put a lot on whether whatever noise he made was audible only to himself, or whether it was audible to or affected the tourists in question.
I'd generally be less concerned with something he said, or a joke he made at a tourist's expense, if it was directed to/only heard by himself or people at the barbershop, than I would be if he was audibly hassling someone whose only crime was walking by him. If people are minding their own business, you should mind yours as well - "behaves like a jackass to strangers at random" isn't the kind of behavioral trait that makes me think I should be handing my money to its possessor, when there are plenty of decent people in the world who could use that money just as much.
I wish you luck! It can be a really difficult thing to quit fully. I believe your life and society will be better for it, though.
I ask because if you do or have, chances are high that you've supported human sex trafficking to some degree without knowing it[1]. It's regrettable that this applies to most people, including myself, and it's worth bringing up. But I just wanted to point out that maybe a person shouldn't be so quick to judge other people, I'm sure he's doing his best.
If you do choose to go somewhere else, I think the kinder thing to do would be to bring up the issue without letting it be an argument, and then let him know that you don't feel right supporting someone with his morals, instead of just leaving him to keep being the way he is without anyone bringing it up to him. Then at least you will have made some effort in making the world a better place whether or not anything comes of it, instead of just ignoring something you take issue with. But chances are, whoever you go to instead will be supporting sex trafficking in their own way and you won't go to very much effort to find out or take your money to someone more "decent".
Huh, interesting. I only ever watched amateur porn, or hentai. Always felt pro-porn looked disgusting.
As long as we're talking ethical misdeeds, I feel factory farming is really the worst ethical wrong of our time, and all meat eaters (me included) are contributing to that one (though I do donate to animal welfare charities).
Interesting idea raised at Marginal Revolution: when a judge lets off a violent criminal with a much too lenient punishment, and the criminal goes on to commit another violent crime, should the *judge* be prosecuted? I’d say in some extreme cases this might be warranted, though I also believe that jurors and parole board members shouldn’t be exempt either.
Sure, as long as we're comfortable with making it symmetric. If judges (and optionally juries and parole board members) impose harsh sentences on people later found to be innocent, we'd better inflict similar punishments on them. This is not a reductio ad abusurdum, this is honestly what I think would be the minimum necessary criteria to make your (and MR's) idea non-disastrous.
To be honest, this is so frustratingly basic and obvious that I have trouble believing you and Marginal Revolution are taking this seriously, and not just saying provocative shit for signalling reasons. If you put a person working a necessary job that naturally involves weighing Type I errors against Type II errors and stick a only-barely-metaphorical gun to their head saying "if we ever catch you making a Type I error, we'll kill you," then literally everybody doing that job is only ever going to make Type II errors from then on out, and at a vastly increased rate. Obviously. The only way not to get this outcome is to also extend the threat to apply to Type II errors as well.
To be blunt, I find the fact that someone who pretends to Libertarianism would so casually suggest something so utterly destructive to the very foundation of civil liberties to be utterly contemptible. The most charitable explanation I can think of is that Marginal Revolution is simply inviting outrage to generate traffic now: if they actually believe this to be a good idea, that would say far less flattering things about them.
It is possible to set guidelines that will minimize the risk of harsh sentences for the innocent or those not deserving of such. Try this: if a judge (a) gives a sentence to a defendant convicted of a violent felony that is in the nationwide bottom 10% of sentences given for similar offenses, and (b) the defendant has three or more prior convictions for violent felonies within the past 10 years, and (c) within one year after being released from prison for this latest conviction the defendant goes on to commit another crime of violence, then the judge can be subject to prosecution.
I have trouble understanding the above as anything but "what if sentencing laws, but strictly worse?" If you believe you have a set of criteria that strongly indicate a certain minimum sentence, the sane and reasonable way to handle that is to write those criteria into sentencing laws and NOT give judges latitude to violate them.
If, on the other hand, you recognize the circumstances around some class of crimes are likely to be sufficiently nuanced and complex that allowing significant exercise of human judgement[1] on a case-by-case basis is warranted, then the sane and reasonable way to handle that is to ACTUALLY ALLOW the exercise of human judgement. Not to pre-emptively insist that the judge *must* rule in the way that you've plainly already decided is correct, on pain of risking their liberty for zero personal or professional upside.
Again, this whole argument seems thoroughly asinine. The ENTIRE POINT of having a written code of laws is to inform people of what behaviors are punishable by the state *before they engage in them.* Writing your laws such that a public official may take actions that are perfectly legal at the time they are taking them and only become prosecutable *after the fact* is either dangerously insane or a blatant sign that you view laws largely as a weapon to be wielded against your political enemies, rather than a tool for ordering a society. Indeed, encouraging certain people[2] to fantasize about wielding the force state against their political enemies seems to be about the only comprehensible reason I can conceive for anyone suggesting things like this at all. As actual suggestions for a real legal framework, they are beyond farcical.
[1] Which of course always carries with it the possibility of error.
[2] Whose political persuasions I'll decline to speculate on.
It's probably much easier to remove them from the bench than to prosecute them. I'd be cool with some sort of law that measures how often a judge under-sentences someone who then goes on to commit another crime, and if the judge's record goes below some threshold then they automatically lose their job.
I wonder why is the blame directed towards the police so rarely... It is (almost) always the defence attorneys or the judge, even if their role is to show that the police failed to perform their job well.
There's a much stronger moral case for imprisoning judges who grossly *over* punish. A lenient punishment leading to a repeat crime is a possibility- it represents a factual mistake. On the other hand, a strict punishment leading to unwarranted imprisonment/or unwarranted extra time in imprisonment is a certainity, so there is no mistake of fact here. For example, I'm sometimes sympathetic to the idea of prosecuting, say, the judge who sent 10 year old Joseph Hall to prison for 13 years for shooting his abusive father.
On the whole I don't support this, though I do find the position tempting. Where a punishment is grossly and manifestly excessive (as in Joseph's case) The only argument against it the ole' princple: Nulla poena sine lege. But I've always found this a fairly weak defence when it comes to state agents. Of course the atrocities of state agents are usually authorised by law! To say that this means they should never be prosecuted seems like an unwarranted shield for the afflictor and persecuter.
So on the whole I don't favour it, but these matters are finely balanced.
I have a relatively unique position on this, but I think a defensible one. First of all, we have to get rid of the myths, like prisons don't actually "correct" anyone, and they do not serve any kind of cosmic justice. Thus their only use is simply separating dangerous people from potential victims. So they are not about punishment, but about harm-prevention. In that case, much more humane systems should be invented - basically just walled towns with normal jobs, entertainment and so on. But in that case, because those do minimal harm to convicts, far more people should be "incarcerated" in them. It seems like at least 10% of the male population now is too dangerous to be at liberty.
Specifically, in case of capital punishment of an innocent person, I think it would be logically consistent to sue everyone involved for a conspiracy to commit murder.
This idea was referenced. The proposal actually endorsed was financial incentives (in both directions) I think if this was done close to revenue neutral risk aversion would lead judges to just never release. Theyd need a pretty big payout on non recidivism to take the risk. I wonder then to what extent it affects sentencing also. Stock the prisons, so we can get paid to release them sort of thing.
If an economist really wants to champion an idea like this. They should look at what performance incentives actually look like on the private market and model off those.
No, they shouldn't, because they're not clairvoyants and the people who would replace them are likewise not clairvoyants. The punishment for judges with poor judgment is to remove them from the bench.
Given that recidivism is like 1% or lower for many types of crimes with long sentences, how about we arrest judges for unlawful imprisonment whenever they send someone to jail who represented no credible threat to society, ie every day?
No, we should not make the largest carceral state in the civilized world even more punitive and inhumane.
> Given that recidivism is like 1% or lower for many types of crimes with long sentences
You need some citations for this, recidivism is basically ~33% at 2 years, even in much less punitive, more reform based societies like the Nords. And it's the dominant outcome at 5 years.
My old college roommate and friend is on the faculty of Yale Law School, and so I ended up getting notices about their journal. Here's what they're featuring right now:
"Geduldig's Footnote Twenty and the Past, Present, and Future of Sex Equality - Yale Law Journal
For helpful comments and insights, I thank Katie Eyer, Melissa Murray, Kate Redburn, Stephen Lee, Ari Waldman, Douglas NeJaime, Reva Siegel, Jessica ..."
Academics. Liberals. Geduldig's Footnote Twenty. Seems like madness to me.
Maybe the title is meant to be "Gedul Digs Footnote Twenty"? A bit more relatable, you know?
For context, the case Geduldig v. Aiello (1974) held that California's disability insurance excluding conditions arising from pregnancy was sex-based discrimination, and therefore not violate the Fourteenth Amendment.
Footnote 20 in full:
The dissenting opinion to the contrary, this case is thus a far cry from cases like Reed v. Reed, 404 U. S. 71 (1971), and Frontiero v. Richardson, 411 U. S. 677 (1973), involving discrimination based upon gender as such. The California insurance program does not exclude anyone from benefit eligibility because of gender, but merely removes one physical condition -- pregnancy -- from the list of compensable disabilities. While it is true that only women can become pregnant, it does not follow that every legislative classification concerning pregnancy is a sex-based classification like those considered in Reed, supra, and Frontiero, supra. Normal pregnancy is an objectively identifiable physical condition with unique characteristics. Absent a showing that distinctions involving pregnancy are mere pretexts designed to effect an invidious discrimination against the members of one sex or the other, lawmakers are constitutionally free to include or exclude pregnancy from the coverage of legislation such as this on any reasonable basis, just as with respect to any other physical condition.
The lack of identity between the excluded disability and gender as such under this insurance program becomes clear upon the most cursory analysis. The program divides potential recipients into two groups -- pregnant women and nonpregnant persons. While the first group is exclusively female, the second includes members of both sexes. The fiscal and actuarial benefits of the program thus accrue to members of both sexes.
Do you mind explaining your criticism a little more?
As an amateur hobbyist in US constitutional law, this article seems worthwhile from two perspectives:
(1) What is considered strong argumentation in the current SCOTUS? For example, "Moreover, in his separate concurrence, Justice Alito—Dobbs’s author—suggested that Geduldig’s precedential status was so obvious and powerful as to obviate the need for explanation, despite the fact that the Court had never favorably cited to Geduldig in a constitutional decision before Dobbs." (page 6 of the pdf, labelled page 1700)
However, personally, I think it is more the case that the court in Dobbs cut the reasoning to fit the desired outcome rather than vice versa (Geduldig needs to be followed as precedent, but Roe/Casey don't???)
(2) What are the substantive implications of the use of the Geduldig footnote reasoning?
The on-the-ground implications from Dobbs and Skrmetti are significant.
Lastly, there is a baked-in pun to have a minor 50+ year old side comment jump to prominence ('geduldig' is German for 'patience.')
>Do you mind explaining your criticism a little more?
Actually I think my post is indefensible. When I put it up I knew nothing about Geduldig and related matters. Absent that information I found an article about a footnote, presented with deep seriousness, complete with thanks to the multiple scholars who had assisted, hilarious.
I'm sure that part of my reaction to it is an offshoot of my sadness and frustration about how my old friend's career has affected her. She lives in a world of constant competition, one-upsmanship and political intrigue, all centering around things that seem to me to be obscure points, and participates wholeheartedly in the intellectual-egomaniacal mosh pit -- all of which is of course irrelevant to everyone else here.
Last summer I posted my first album co-created with AI and got some helpful feedback on this thread. (Carbon Dreams by Sus Houseplant on YouTube and Spotify).
Someone went as far as to say some of the songs became their favorites.
Today I’d like to share my first song / music video. The genre is less synthwave industrial and more singer songwriter country.
I am not a fan of the country genre you are targeting, so that might color my criticisms.
Compared with Carbon Dreams, I think this video sits more in the uncanny valley. The CD tracks I particularly like are absurd with nuggets that are accidentally profound ('tell me again how nonsense feels') or euphonious ('packet of silvery lemony lime.') In Scorpion's Lullaby, the combo of lyrics and video seems closer to an all-human product, but just doesn't quite make sense or feel right. I think adding video massively increases the number of opportunities for small mistakes that feel very irritating.
FWIW, there are plenty of all-human pieces that also have little gaps that bother me, so maybe I am not reacting to the AI process.
My 10 free gift articles per month resets end of day. This Ezra Kline interview of Michael Pollan discussing consciousness is pretty interesting. They get into William James’ thoughts on the matter early on.
I've recently been reading "The Space Shuttle Decision", about the political and design history of the Space Shuttle program. Early on, the Shuttle concept had been just one piece of a larger system of space infrastructure that had been proposed c. 1970. I knew about the space station, where there were a mix of ideas. One was converted S-IVb upper stage launched by a Saturn rocket, either a "Wet Lab" where it would be used as the upper stage of a Saturn Ib and then converted to habitable space in orbit or a "Dry Lab" that would be built and furnished on the ground and placed in orbit by the lower two stages of a Saturn V. The other would be that the station would be built in orbit out of a bunch of modules lifted by the Shuttle. Two of these would eventually happen, of course: the Dry Lab became Skylab (which was used for three Apollo missions and then deorbited because the Shuttle wasn't ready to re-boost it) and much later the modular station became the ISS. In the meantime, the Shuttle itself was used to lift and support single-use "Spacelab" modules for 1-2 week missions, and also a ton of the planned missions for the Space Station were instead done by unmanned satellites as improved electronics made it feasible to e.g. control telescopes and weather satellites from the ground instead of needing astronauts to calibrate instruments, take readings, and manually change the film like they did for the Apollo Telescope Mount that was built into Skylab.
One bit that I didn't know about was the third key part of the early space infrastructure proposal, the Nuclear Shuttle. This would be a durable spacecraft powered by a NERVA engine and living in orbit. The Space Shuttle would take astronauts and payloads to the Space Station in LEO, and then the Nuclear Shuttle would boost any payloads that needed to go much beyond LEO. Post-Apollo missions to the Moon and interplanetary missions would then be staged in Lunar orbit, lifted from LEO to Lunar orbit by the Nuclear Shuttle.
The Nuclear Shuttle was killed fairly early on by budget cuts, since there was definitely no money for the for most of the missions it was intended to support, and NASA administrators were only barely able to persuade the Nixon administration and Congress to not also kill the Shuttle and the Space Station outright. The Shuttle we got was a lot less ambitious than most of the early proposals, for both technical and budget reasons (the book emphasizes the budget reasons, but reading between the lines also suggests a ton of engineering risk), and the post-Skylab Space Station program got postponed until the 90s. Budget cuts also killed the NERVA R&D project at a fairly late stage in development, so there wouldn't have been an engine for the Nuclear Shuttle even if there had been money for it and its missions.
So, while the Space Station idea got resurrected in the 90s, and the idea of a Lunar Station for staging Lunar Landings and interplanetary missions has come back in recent years with the Artemis program (although the Lunar Gateway station seems to be mostly dead as of recent weeks, in favor of a surface Moon Base). But if there have been any proposals for resurrecting the Nuclear Shuttle or anything like it, I haven't heard about it, and I am curious about why since a reusable nuclear thermal rocket shuttling between LEO and LTO or LLO seems like it would make it much, much cheaper to build and support a Lunar station or surface base and especially to supply a propellant depot in some Lunar orbit or a Lunar-adjacent transfer orbit.
Are there major technical problems with a Nuclear Shuttle that I'm not aware of, or is this just some combo of 1) reducing medium-term costs and scheduling risks involved in dusting off NERVA or some appropriate replacement and developing the Nuclear Shuttle, 2) "Nuclear" anything sounds scary, and 3) Political constraints related to the observation that Artemis seems to exist for the purpose of giving SLS something to do?
I can add to the issues John Schilling pointed out about nuclear rockets, in that another hugely limiting feature of their usefulness is the quite limited operating lifetimes. Most nuclear thermal rockets, like the US NERVA, have service lives of just a few hours!! This is also compounded by their low thrust to weight.
Even though the NERVA is more than twice as efficient, you wind up burning for a much longer time to make up for the lower TWR. As a result you use up your (hideously expensive weapons-grade enriched!) nuclear engine's limited lifespan very quickly, and nobody has ever considered trying to refuel in space or return spent nuclear rockets to Earth, so it is totally disposable.
The limited lifespans of nuclear thermal thrusters is a huge issue and, for many Mars mission architectures, is one of the overriding design constraints. Mars return missions sit right at the edge of what's practically possible in terms of total burn time even with multiple nuclear thrusters or even staged designs. This is why basically all the crewed Mars missions have the crew "bail out" and aerobrake in a capsule to land back on Earth as their depleted spacecraft flies past the Earth into solar orbit!
Another problem with reusability is that every mission which returns from the moon, for example, has so far used aerobraking to land back on the earth at the end. This saves us almost 3 km/s of delta V on the return leg which would otherwise be needed to propulsively brake into LEO. If you have a reusable nuclear tug which can't be allowed to reenter, then you need to carry all of the reaction mass required for that final braking burn throughout your entire trip!
So you wind up carrying much less payload, or requiring a huge craft, and by the time you've braked your reusable tug back into LEO after your lunar trip your fancy expensive NERVA is completely spent anyways and is now an orbiting radiological hazard for aeons.
Regarding nuclear-thermal rockets, they're definitely useful but not nearly so much as a naive performance comparison would suggest. The issue is that nuclear-thermal rockets really only work well enough to outperform chemical rocket engines if they use liquid hydrogen as their reaction mass. And liquid hydrogen is a terrible, terrible propellant to handle. It has to be kept exceedingly cold if it isn't to boil off, and it has a density slightly greater than styrofoam. Which means big enormous tanks, whose vast surface must all be well-insulated if you're not going to use the stuff immediately, and overpowered pumps to force it into an engine. And it really really likes to leak.
If you look at a Space Shuttle, you'll see the compromise that is the ginormous external fuel tank - and that's for a vehicle whose propellant is (by mass) mostly nice, dense, well-behaved liquid oxygen and only about 15% liquid hydrogen. A nuclear-thermal shuttle would need only half as much propellant, but it would *all* be liquid hydrogen.
For use in deep space, those disadvantages aren't quite as critical. But it's still enough to downgrade nuclear-thermal from a must-have to a nice-to-have. Maybe 20% improvement in actual mission performance, rather than doubled performance. And that's a particularly expensive technology to develop, for a marginal performance gain.
The difficulties of working with liquid hydrogen were not well understood when work on nuclear-thermal rockets began in the early 1960s. The experience of Apollo, which used some liquid hydrogen in its upper stages, gave people an inkling of what was to come, not enough for them to choose a more sensible propellant for the Shuttle, but enough to take some of the shine off nuclear-thermal.
Also, in the 1960s the expense of the technology development program was mitigated by the ability to say "yeah, it's an experimental nuclear reactor, and because it's also a rocket if it melts down it will spray the radioactive slag across the surrounding landscape" but immediately add "...which is in Nevada so nobody cares, this is wicked cool shit, let's get on with it already!"
That's obviously not going to fly today, which means budgeting something in the eleven-figure range for a test facility that can contain the fallout from a nuclear rocket oopsie.
Thank you, "you need to use liquid hydrogen as reaction mass to get a big advantage over a good chemical booster" is definitely a huge disadvantage. I'm familiar in general terms with how much of a headache hydrogen was for the Shuttle and continues to be for SLS, and I've given to understand that's a big part of why SpaceX and Blue Origin have been working on methalox engines instead. I've come across stuff about NERVA using liquid hydrogen, but I though I'd also seen stuff about ammonia or water as reaction mass and hadn't realized those weren't viable if you want high specific impulses.
I'm guessing nuclear-electric doesn't give you enough thrust to be worthwhile for translunar injection on a large payload, at least not in a form that's anywhere near being ready to use?
Also, I expect you need quite a bit more than a 20% improvement in efficiency for reuse to be worthwhile even for going to lujar orbit and back, since you need to bring enough fuel to come back and also extra fuel to bring the return trip fuel with you. If you aren't saving much mass to orbit, you're just saving the cost of the engines, it looks like the RL-10 engines on a Centaur V upper stage aren't super cheap, but they also don't sound like they're expensive enough to move the needle on the cost of an Artemis mission.
Right; if there's a use case for nuclear-thermal, it's probably when you're going to Mars or beyond. And even there, it's competing with nuclear-electric and even solar-electric, probably used in parallel with old-fashioned chemical rocketry.
Or possibly for highly kinetic warfighting in Earth orbit or cislunar space, but that's speculative in ways I'm not going to touch here.
Another relative disadvantage for work far beyond Earth is, where do you get the propellant? Liquid hydrogen, or even gaseous hydrogen that we could liquify, is hard to come by until you get *way* out in the solar system. Water, is much easier to find, and you can extract hydrogen from it. But in the process you'll also be extracting eight times as much oxygen. Way more than you'll need for life support, so you'll want to find *something* to do with it. And in most cases, I'm pretty sure you're going to find that using that oxygen along with the hydrogen (and maybe some carbon) in a boring old chemical rocket, will give better overall performance than throwing out the oxygen and running just the hydrogen through a nuclear rocket.
I can’t answer the rocket science question but I’ll throw in that I recall the July day that Sky Lab deorbited. I had brought a transistor radio to work to learn where it would come to earth.
I piloted a much less sophisticated vehicle, a diesel compressor up to the hill top where we ripped and rebuilt rail sections that day. The drive was pneumatic powered by the compressed air in the 200-gallon air tank. I would joke about ‘silent running’ mode when I moved the contraption 50’ or so on the air in reserve.
I never did really listen to the radio though. The work had a Zen-like rhythm that I found soothing and satisfying. Tap, tap, tap, half swing to set the spike an inch or two into a tie, nail every fourth tie home, single spike inside the rail with another spike set, a single spike driven home outside.
Then start the compressor and take turns with my partner driving home the spikes that were just set in place with an air hammer. I’d go up one side while my partner held the tie against the rail with an iron bar with a Little Bo Peep hook on the end. At the end we would swap positions on the way back and I would catch my breath.
In its way it was a glorious way to spend a summer day for a healthy young man.
Oh yeah, Sky Lab came down sometime that day and as expected no piece was big enough to cause any damage.
>Oh yeah, Sky Lab came down sometime that day and as expected no piece was big enough to cause any damage.
I had thought this wasn't quite true, with one piece of debris killing a cow in Australia. But I looked it up just now, and it seems that you were right: the story about the cow is widely circulated but not accurate. It's a garbling together of two separate incidents: the debris from Skylab did fall in rural Australia and the first larger pieces of it were found in a cattle pasture, but none of them hit any people, livestock, or structures. There was an incident with falling American space debris killing a cow, but much earlier (November 1960) from a failed satellite launch (Transit 3A and SOLRAD 2 / GRAB 1). The rocket shut down early during launch and the range safety officer detonated it, with the debris falling across parts of Cuba and killing a cow. The US paid Cuba $2MM in compensation, Cuba salvaged the debris and sold it off to the Soviet Union and China for study, and presumably the incident contributed to the decision to launch things into polar orbits from Vandenberg instead of Cape Canaveral, as Vandeberg has no land directly South of it until you get to Antarctica.
I noticed someone posted a daily word game here and thought I'd drop my fangame of a very fun concept of magnifying images as a game, and added some of my own magnifier types in here:
The code for Claude Code has leaked through what appears to be a rookie error born of vibe-coding. The weights are obviously still secure, but this is probably going to be good for progress in other coding harnesses.
I keep hearing about free and cheap AI making stupid but plausible mistakes, and I will continue to mock AI that's a game of you bet your law license.
However, I also hear about the AI that costs money ($200/month? more for even better?) which is strikingly good, though it might still need skilled human supervision.
And there are companies which are buying into AI, and insisting that employees use it even if they hate it, which doesn't sound like a good tool to me. Are they, or some of them, buying the good stuff?
How expensive would it be to produce good AI for the general public? Are companies losing money on the AI they produce?
There was a video about how data centers are being built, but the computer chips haven't even been installed yet. Is this even remotely true?
I think there's already a *lot* of AI (by which I mean the modern, LLM-and-adjacent-technology stuff) usage, but it's not always what people think of and not always super visible.
A lot of people talk about "AI usage" in terms of "I navigate to ChatGPT and ask it a question and it produces an answer" (and there *is* a lot of that), but I think a lot of it is in other forms: a ton of medical providers are adopting transcribing tools that automatically take notes with interactions with patients.
(I've been doing Physical Therapy and had an extensive discussion about this - it apparently records the conversation, filters out stuff that's just side-chit-chat, categorizes the rest of it into the different fields that they're supposed to fill out, and supposedly does a better job providing the right details to get insurance companies to support continued treatment)
And from my day job, I know a lot of e.g. call/contact centers are adopting it, too. You call in and even if you're still talking to a human (though who knows how long that'll last), they're increasingly working with information that's curated and prompted by an LLM that's listening to the conversation.
These are both primary examples of LLM-based *understanding* rather than generation, and I think that is the 'easier' side of the coin to leverage right now, but we're absolutely seeing usages of the generation side of the coin too. Though given the backlash, a lot of places aren't advertising it.
(I see a very CGI-like dynamic where everyone "hates" it, but it's really just the reality of how stuff gets done and the stuff people 'hate' is the obvious bad stuff and not the entire rest of the iceberg, so companies make big shows of being "no CGI" when they aren't: https://www.youtube.com/watch?v=7ttG90raCNo)
And, also from my day job, AI usage in programming is picking up and shows little sign of stopping much less rolling back to the 'old ways', despite grumbling. (IME skeptic positions have largely shifted from "you can't do programming with AI" to "well, you can program with AI, but only because token costs are highly subsidized and it won't be affordable in X years")
Premium AI models are mostly good for coding and advanced mathematics. Most people have no use for advanced mathematics, so practically it's just good for coding.
And tech companies are pushing it heavily. The degree to which AI has been responsible for tech layoffs has probably been overstated in the past, but it's certainly true as of this year. Companies are expecting significant productivity boosts from each employee while using AI tools, and they are laying off workers accordingly.
Good AI is already available to the public. You can get the premium models for $20 per month. And Google just gives theirs away for free.
AI companies are losing money in the sense that they are spending more on R&D than they are recouping in revenue. But R&D is necessary to guarantee future return - they could stop R&D and become profitable, but then revenue would drop to zero after a year when their models are out of date.
As far as chips lying idle, it's incredibly complicated. Short answer is yes, but the long answer is that building a data center is messy business - it's hard to predict which step is going to be the bottleneck. Sometimes you buy the chips while you can but then you can't connect to the power grid fast enough so the chips sit around idle. But sometimes you get the power online while the chips are in transit and you kick yourself because you just lost on a month of revenue that you could have had if you ordered them to be delivered a month earlier. This is just a normal part of any large project that involves multiple different supply chains and regulatory steps.
By the way, these are the sorts of questions that basic AI models are very good at answering - and they've been heavily trained to cite their sources and ground their statements in web searchers. Hallucinations aren't solved but they are lot less than before. Try any of them - Claude, Gemini, Chatgpt. They're all pretty decent.
From what I gather, including from Scott's recent experiment in running premium LLM queries for others so they can try it out, the current crop premium models are incrementally better than the free ones but aren't fundamental game-changers. I suspect that most of the apparent "super-premium models are way better!" is self-selection bias on the part of the user base. People who love AI tend to be the ones who are paying for the premium models while the haters do not, so a lot of the opinions available about free models come from unsympathetic observers while the most of the opinions about premium models come from sympathetic ones.
My employer is buying "the good stuff" for us. It has its uses, but still makes stupid but plausible mistakes at an annoyingly high rate. I find it to be more trouble than it's worth and personally avoid making more than minor use of it, and those of my coworkers who are much more enthusiastic adopters don't seem to be getting substantially more done with its help.
>There was a video about how data centers are being built, but the computer chips haven't even been installed yet. Is this even remotely true?
Sort of. Putting GPUs into servers is one of the last stages of the process and can be done much, much more quickly than buying land, getting planning and building permits, constructing buildings and supporting infrastructure, and furnishing the buildings with racks and servers and cooling systems, so it makes sense to get started on all that other stuff first while the GPUs are on order so you have somewhere to put the GPUs when they're ready.
You can already get the best Claude, Opus 4.6, for $20/month, paying more allows you to use it more, but it doesn't give you a better model.
I'm a programmer, and I use it frequently, but I really don't think this stuff is ready to replace human coders, companies that think it is ready for that are being irrational IMO. AI is in its radioactive quackery phase, though it is admittedly more useful than radioactivity.
Every time "AI" is mentioned is very broad such that there probably aren't very exact answers to your questions, especially since it works less well on some codebases than others, or with different needs than others. Small-to-medium apps and such would work best and probably not with those with large codebases already. People do believe that current 'good AI' is investor subsidized and they are losing money by providing it in hopes of building better ones in the future though.
I am more and more starting to think that most people who come across as autistic, generally aren't. That's because it is supposed to be a disorder, not a personality type. And I think people who are seen as autistic tend to have this personality types:
- Low agreeableness. I don't care what is the nice thing to say, I will be just blunt and honest.
- Low neuroticism, feeding into the above: I don't care how people judge me.
- Low openness to experience. Prone to routine and generally stubborn.
- High conscientousness, precise, dictionary definitions, deep analysis, often numerical.
These are not disorders, even when they cause problems (of unpopularity).
Autistic used to be non functional adults who often had to be institutionalised. This definition you are referencing (I know you are hostile) is just diagnostic creep
I am not formally diagnosed with Autism Spectrum Disorder, but I have been told by a mental health professional who was in the process of treating me that I probably am on the spectrum, and I have repeatedly scored extremely high in the past on Autism/Aspergers screening tests.
I just took an online Big 5. I'm high agreeableness, low neuroticism, high openness, and high conscientiousness, so 1 for 4 on your list. Sample size of one, but at least a minor data point in favor of the hypothesis that your list of autistic-coded traits doesn't align well with actually being autistic.
I think the core autistic trait is social obliviousness, which isn't showing up in your breakdown. I can imagine someone hitting every trait on your list that actually has regular social awareness.
Really? The difference is that some people don't get it. It's like someone who can't recognize faces versus someone that just doesn't pay attention when meeting people. It's like having to emulate hardware instructions in software because you don't have the hardware, except that the software version usually doesn't work well.
I know quite a few people who have been diagnosed as autistic spectrum by professionals. What I hear from many of them is that they feel like everyone else got a rule book for how to understand people and how to talk to them, and they did not. They tell stories about saying things that offended or bewildered the person they were talking to, and having no idea why what they’d said went over badly. And they definitely care what people think. Those incidents are painful and embarrassing for them.
Another key: Autists do not behave "properly" aka "within the rulebook" for guilt trips. They tend to realize quite a bit later "hey, that person was trying to make me feel bad!" (and then they get angry, because who likes to be manipulated?)
Also of note: Autists tend to be a lot more likely to be sick and not notice it, either from high pain tolerance, or lack of "pain localization".
To outside observers, perhaps not much. But it makes a difference with the person who tries and fails, hence why they might prefer a label to distinguish themselves from those who don't try at all.
We do our best to make allowances for people who say tactless or offensive things without being aware of it. If someone is intentionally tactless or offensive, not so much.
I’m thinking of a spectrumy sister in law right now. After an extended visit my tongue is pretty well bitten off. Just keep thinking “she just doesn’t know how that comes across” the whole time.
1. The Saudi’s have reportedly agreed to allow the US military to use their King Fahd air base in Taif as a staging point. They’ve not officially confirmed this, though. Checking Google Maps shows that Taif is a couple of hours' drive from Jeddah on the Red Sea, and Taif is about 1,200km from the coast of Iran. While this wouldn’t put US aircraft out of range of all Shahed drones, some military analysts said the increased distance from the Gulf will allow the US greater time to respond. The latest models of Shahed drones are reported to be able to carry a 50kg payload of explosives and have a max range of 2400 km. I don’t know if the payload needs to be reduced to reach that max range.
2. This would be a logical response in the face of the damage being inflicted by Iranian drones on US military assets. As of three weeks into the conflict, Iranian drone and missile strikes have caused damage and losses estimated between $1.4 billion and $2.9 billion. Reported losses include at least a dozen of our MQ-9 Reaper drones (which cost about $30 mil each), damage to our radar systems used by us and our allies, and several KC-135 tanker aircraft. Also an Iranian-sourced video (shot from what is reported to be an FPV drone) shows drones systematically taking out a Sentinel Radar System and at least one Black Hawk helicopter at Victoria base in Baghdad. Since they were FPVs I presume they were launched within Iraqi territory.
3. The Pentagon is reportedly focused on taking out the launch sites. But a Shahed drone can be launched from the back of pickup truck (though, I gather it’s better to place the launcher on the ground, because there’s a small rocket engine that can give the Shahed an initial boost). It’s not clear how the Pentagon is dealing with these highly mobile, lightweight launch platforms. The Ukrainians have driven the Russian Shahed launchers out of the Russian-occupied territories by using loitering observation drones in dense patrol patterns. Maybe the US Reaper drones can serve the same function?
4. Reportedly, Russia is providing its upgraded electronic packages for Iran’s Shaheds (as well as targeting data from Russian satellites). They’re coming in via Azerbaijan. It’s interesting that Russia isn’t just shipping them to Iran across the Caspian Sea, suggesting that they feel safer shipping them through a neutral mountainous country, rather than risking attracting Ukrainian long-range drone strikes on open water (Ukraine sank at least one cargo ship carrying drones from Iran on the Caspian).
5. Zelenskyy visited the Gulf States, and Ukraine signed a 10-year agreement with the UAE, Qatar, and Saudi Arabia to supply them with drone technology. Ukraine will be sending experts to these countries to help them set up drone production facilities, and advise them on how to counter Iranian drones. Exports were mentioned. It’s unclear if he’ll divert some of Ukraine’s current production to the Gulf, but Zelenskyy reassured Ukrainians that he won’t be short-changing Ukraine’s drone defenses.
> “It is about exports and about opening up exports. But it the right kind of opening, where we understand that we are not selling our experience for nothing.”
Also, Ukraine will share its marine drone technology and advise them on how it opened the Black Sea to Ukrainian shipping. It’s not clear to me how marine drones would be helpful in counteracting smart mines and airborne drone threats, though.
I haven’t seen any comments from the US on this development.
6. According to the WSJ, Trump has told aides he's willing to end the Iran war even if the Strait of Hormuz remains closed. I guess those British mine countermeasures vessels that Starmer is sending over should turn around and go home now. Trump reportedly said it should be the Gulf States' responsibility to keep the Strait open. Given that on a normal pre-war day, between 100 and 135 ships traverse the Strait, at $2 million per ship, Iran could make more from tolls than it did from oil exports ($93 billion per year from tolls vs. $50 billion per year from oil exports). Of course, the tolls would have to be paid in Renminbi and not dollars, which would help undermine the Gulf’s petrodollar economy.
7. And did North Korea offer to help Iran with its nuclear weapons development? In a recent speech, Kim Jong Un used the example of the escalating US-Iran conflict to justify North Korea's nuclear arsenal. He asserted that holding nuclear weapons is essential for regime survival, and pointed to the examples of nations like Iran and Libya as proof that giving up such weapons is dangerous. In a response to the "gangsterlike imperialists" of the US and Israel, he's ordered a rapid expansion of North Korea's nuclear arsenal. In the same speech, he may have floated an offer to assist Iran with nukes, but I haven't seen a transcript of the whole speech, and I'm not sure if that talking point was manufactured by anti-NK commentators.
8. Ukraine is deploying a third-generation energy weapon, called ‘Sunray,’ to specifically counteract Russia’s fiber optic drones. Specs such as its energy output haven’t been shared, but someone who saw it demonstrated said it’s about the size of a hobbyist telescope with two camera-like thingies mounted on either side of the tube that produce the laser beam. It was mounted on the back of a pickup truck. The demonstration showed it fry the fiber-optic cable and/or the cable spool hanging from the drone. The Atlantic said it’s not in production yet, but a Russian video was recently posted that showed one of their drones being hit by a flash of light and falling. Russian milbloggers are not happy about this development.
9. And in the US, multiple waves of unauthorized drones have been spotted over Barksdale Air Force Base in Louisiana. One report said this base hosts our strategic B-2 and B-21 stealth bombers, but only B-52s were mentioned in the DoD (DoW?) statement. Barksdale also plays a critical role in the command and control of the US Air Force's nuclear defense capabilities. There were no mentions of countermeasures to the overflights, but Barksdale AFB will be getting one of the new ‘Flyaway’ anti-drone systems real soon now (they may have received it already). The Flyaway has already been deployed in the Gulf. It’s unclear if the US installations that were hit by Iranian drones were using Flyaways.
But did you know that overflights involving multiple drones over CONUS military bases have been going on for several years now? I didn’t. In light of Operation Spiderweb, I sure hope the Pentagon is taking these seriously, but their response to the Barksdale flyovers seems pretty lame.
"Generals always prepare to fight the last war, especially if they won it".
>Reportedly, Russia is providing its upgraded electronic packages for Iran’s Shaheds (as well as targeting data from Russian satellites). They’re coming in via Azerbaijan.
What?! Azerbaijan recently attacked Armenia, which had a written military alliance contract with Russia.
Azerbaijan has recently improved its relations with Russia, and poor Armenia was jilted. I just read that the drone parts are being driven through Azerbaijan as humanitarian aid shipments, but the purported origin was one of tracked back to the drone factories recently hit by Ukraine. Maybe they're shipping salvaged parts, or maybe they're up and running again?
It recently attacked Artsakh, which was an Armenian-populated unrecognized separatist state in Azerbaijan. It wasn't Armenia proper, which meant CSTO couldn't be invoked and Russian peacekeepers could behave like any other peacekeepers.
Did you miss the IEDs at MacDill? That's Centcom HQ and the military special ops, apparently tied to the Iran War (someone mined outside the base, afaik).
That's... something the military is NOT playing dead about.
Thanks! This is why I label these posts as "Fog of War", because there's so much conflicting information flowing through the media networks (mainstream and social).
On 6, I wonder if the strait remaining closed could be an acceptable outcome for Trump due to the impact it would have on domestic oil production? I used to live in TX oil country, and there was always talk about Russia and OPEC trying to strangle American production by keeping oil prices low. There's a lot of viable sites that were not pursued or abandoned due to economic considerations.
This would have been a stronger argument back before Nuland decided to kick Europe in the nuts with Russian sanctions on oil and gas. Many european companies have either gone bankrupt or moved their production to Anywhere Else (including America, which was the desired outcome from America's standpoint*).
Reality: almost nothing happened, life continues as usual. The only ones complaining about Russian sanctions are Russians... which is exactly the point.
Thankyou for this. I've heard about drones over US military bases for years. Most of this comes from the UAP community. So it's hard to know what's really going on. (And yeah like why isn't the military more concerned? Are they our drones?)
What happens when someone arms those FPVs and pulls an Operation Spiderweb? Playing dead could cost tens of billions of dollars worth of Strategic Air Command bombers and impair our response capabilities. The policy should be strict interdiction of all overflies (if possible). It *looks like* they weren't doing anything because the new Flyaway anti-drone systems hadn't been deployed. Mind you, I said *looks like*.
I don't know enough about these particular military capabilities to say a damn thing, to be perfectly honest with you. I do know there are detectors (they're used in airports, and some of them are defeatable with laundry detergent from near Chernobyl -- those being the ones that sense "hot" goods presumably), I don't know range -- it's plausible that one could identify whether a particular drone is problematic or not. Maybe. Depends on range, I suspect.
Tens of billions of dollars is peanuts, if very embarrassing peanuts.
The MAGA arrogance with regard to Zelenskyy/Ukraine is now becoming quite costly to the US.
I mean, Zelenskyy is no saint, but Ukraine has by far the most extensive practical experience with drone warfare of all the Western-adjacent countries out there. Many losses could have been avoided if the US treated Ukraine as an actual friend instead of that stupid attitude "rah-rah-no-one-tells-us-what-to-do-we-are-the-champions-and-always-know-the-best".
The US capabilities in general are impressive, but they don't cover nearly everything out there.
Trump as a draft dodger may be simply ignorant, but Vance actually served in Iraq and should know better than piling up on president of a country which has fought the second largest military in the world to a standstill.
1) America is test-driving some Ukrainian drones near Iran. Yes, they are.
2) Ukraine has a very large incentive to lie about how well their drones are doing versus Russians, so America is duly skeptical and not putting in large orders until they get some actual confirmation that this isn't Another Ukrainian Lie*
3) Dude, I wouldn't say what the US capabilities are. You don't know that, I don't know that. Iran certainly doesn't know that.
*many of these were encouraged by America (aka casualty numbers)...
2) Lies are inefficient, and the time for Ukrainian inefficiency is long past. Lies buy drones that wouldn't work against Russian drones that do, so these lies would cost double. Ukraine has less than zero incentive to lie about this, especially when other countries want to buy their expertise for a hot war of their own.
That assumes what they're being bought for is a hot war, right? I don't see the Gulf Countries as buying for a "hot" war, do you?
Yes, lies are inefficient. They may also be very hard to correct for. (Does one really blame the Ukrainians if they exaggerate? Assume they aren't being complete "liars with pants on fire" here, but saying something "a little better" than what's actually going on). Ukraine is quite dependent on foreign money, so "Yes we are really making good things here with your aid" is a very very very good message for them to be sending.
I don't trust theater kids. Perhaps you have some evidence that the theater kids are out of power in the Ukraine? I still see Mr. Comedian up there (and unlike my friend who is a comedian, Zelensky doesn't have a history of working in the military, that I'm aware of).
You are of course correct, I misspoke. There is currently no war in the Middle East, only a Special Military Operation kicked off by Captain Bone Spurs.
So they are not the "American pet on a loose leash" that you labeled them in a different post? They even dare to lie to their Big Masters about something as critical as efficiency of some weapon?
OK, let us leave your Russian-adjacent view of Ukraine aside, even though your cynicism grates on my nerves. There is a war ongoing, and while war is full of propaganda, it also induces an enormous evolutionary pressure on the tech being used. Given that the UA airspace isn't overrun by thousands of relatively simple Shaheds - you would see the results if it was - their defense is at least partially efficient. It is also likely quite cheap, given the state of Ukrainian state coffers.
And don't paint the US as a possible dumb victim of dastardly Slavic lies. The US has enough ears and eyes to know quite well how efficient or inefficient Ukrainian interceptors are. They aren't used underground, but in full view of the US satellites and AWACS aircraft, with some more observational assets in the theatre.
That said, the saying that generals always fight the last war applies almost everywhere. Militaries are very conservative by their nature, weapon suppliers as well, and cozy relations between politicians and corporations can only really be disrupted by either a big scandal, or bare necessity. In the case of UA, "bare necessity" applies.
I know some drone-adjacent people in UA. It is a fully nerd-infested industry with very few MBAs and other bureaucrats to slow it down. If something works, it spreads overnight. If it doesn't, it dies out. Both these cycles are short, too. Ukrainian drone industry is genuinely good, and the Arabs who quickly signed contracts with UA don't seem to be as cynical as you. But maybe you consider them stupid marks that need some fleecing.
Given how much impetus the Americans gave for the Ukrainian counteroffensive in 2023 (designed to capture American military objectives regardless of the Ukrainian cost, which was both quite steep, and understood by the Ukrainians to such an extent that they pushed their rebellious provinces troops out front for semi-suicidal attacks), I may have put it a little light with my phraseology there -- I could have been more heavy-handed with the truth (nobody wants the truth).
The reasoning on "Ukraine doing things without American permission anymore" is that America is rather preoccupied with Fortress America and Venezuela(Cuba), and then Iran on top of everything else. That and Trump would vastly rather have an "ended" Ukrainian war, however the chips fall out (America has been soundly beaten, strategic objectives will not be won, etc).
Now, I'm sorry, but you're trying to say that the United States has enough eyes to know everything? When they've been pouring enormous resources into the largest military operation this generation? (Besides, why not do a testrun? Worst case, America gets their order in, say, 6 weeks behind the UAE?)
I'm merely repeating what/why, giving you the rationale that underlies why America hasn't put in a huge order for Ukrainian drones at this very moment. If all you have to say is true, it's coming, there's no need to worry (this last bit is me expanding on scuttlebutt, the actual scuttlebutt was America is unconvinced by the Ukrainian reports, which look very promising, so they want to use them in Iran to get some real combat reports).
I remember the offensive well and the choice of the Tokmak / Melitopol direction as a target made a lot of sense. It is somewhat similar to designing a car; the constraints will result in a similar shape of the hood.
Tokmak was a local logistics hub for both road and rail, located just under a ridge of hills that, if taken, would make its position barely tenable, thus opening a potential road to the shore and cutting off the land bridge to Crimea.
The Russians knew that as well, which is why Gen. Surovikin decided to mine the shit out of the approaches (on the order of tens of millions of mines), even at the cost of complicating any future Russian offensive movements in the same area. That worked beyond any expectations.
To be clear, initially, Biden made US Aid provisional on Ukraine not using our systems against targets on Russian territory proper. Putin was regularly rattling his nuclear saber and drawing imaginary red lines in the sand. In May of 2024, they loosened up and gave Ukraine limited authorization, and in November of 2024, they gave full authorization to hit RU territory with ATACMS. In the meantime, Ukraine was innovating and extending the range of its drones. Ironically, if Biden and his advisors hadn't been such pussies, Ukraine might have ended up more dependent on American largess and been totally screwed when Trump betrayed them in favor of Putin. The question is, why didn't they try to fully take out Baltic Oil facilities earlier? They hit them in 2025 and did a lot of damage, but they didn't follow up. Were these recent strikes against Primorsk and Ust-Luga a "fuck you" to Trump for lifting sanctions on Russian oil? Or was there a technical/tactical/strategic reason they waited until now?
I suspect it is combination of both strategic waiting and the technology not being quite there. It takes some time before you have drones capable of carrying a hefty payload all the way to the Baltic Sea, and it takes some time to thin out the Russian AA, which in 2022 might have been the most extensive in the world.
Plus, now it is spring. Even if the Russians manage to destroy much of the Ukrainian grid or gas network in retaliation, civilians won't freeze to death anymore.
That would be a major concern in autumn and winter. Spring is the safest period of the year to do things to Russian fuel industry.
Vance has quite unambiguously demonstrated that he follows no ideals or principles, except for achieving power. In 2025 and 2026 that means pandering to Trump and MAGA, so it shouldn't be surprising that he'd rather publicly humiliate the leader of a nation under duress (which makes him appear strong in the eyes of the MAGA base) than to ask him for help or cooperation (which would make him and the US appear weak).
How quickly could an oil pipeline to the Red or Mediterranean Sea be up and running? Seems like what the Saudis should have been working on instead of Neom.
Saudi Arabia has shipping terminals at Yanbu (on the Red Sea), and is reportedly operating its pipelines at its full capacity to deliver 7 million barrels a day. Aramco hasn’t officially confirmed this, but it makes sense. Before the war, Saudi Arabia was exporting roughly 10 million barrels a day.
An Iranian drone crashed in Yanbu without causing serious damage; it seems like the Yanbu facilities are far enough away from Iran to be fairly safe from attack. There is some risk that the Houtis (who are friendly with Iran) could close the entrance to the Red Sea.
The UAE has a pipeline to Fujairah (in Oman). As far as I can tell, this pipeline was operating at pretty close to its nominal capacity of 1.5 million barrels per day before the war, but with the closure of the strait it has been carrying closer to 1.8 million barrels per day. Fujairah is close enough to Iran to be vulnerable to attack. Iranian attacks have caused some temporary stopages but have not succeeded in shutting it down.
I don’t think additional infrastructure can be built in a time frame that would be relevant to the current war.
It's been at least 6 years since the Abraham Accords (original treaty to get pipeline running). Yes, it definitely is what the Sauds should have been working on.
Left and right wing have so little in common nowadays, but they both like to make art of trump shitting on america out of a fighter jet. You would take this one common ground from us?
Jeez, Scott, I know it's crass, but what’s the objectionable element? It’s visual political commentary — a reproduction of Trump's own crassness in a slightly altered form, a way of turning it against him. Is the thing you’re asking for less of crassness in general? Linking to my Substack? Joking around? Being rude to the President?
And is there some general category of objectionable “this” I do a lot of, or are you objecting only to this post?
I can't speak for Scott, but here's why I agree with him:
This kind of content doesn't belong on the ACX comment section. The ACX commentariat functions not just as a community but almost as an informal newsletter of its own. One with its own habits, motifs, guidelines, and streams of thought. The best of this newsletter is reaped by Scott and put into his "Highlights on the Comments on X" posts. These comments are usually thoughtful and data-based contributions, rebuttals, or additions to a point Scott made.
They aren't crass political cartoons. It's not something ACX-like; it's just not something Scott has really ever done (in the very early stages he had some meme-ish posts but even they were way more thought out than this). I don't feel improved after having looked at it. Finally, it's just not that funny. Sorry. The joke is already old.
You make a good case for the cartoon not being good quality, relevant content by ACX standards, and I don’t take issue with that. But my question to Scott wasn’t “why don’t you think this is high quality content?” but “why are you singling it out as something there should be less of?”
I am genuinely not sure what the ingredient was that led him to respond with that comment. I don’t think your point that my post is far from being a thoughtful, data-based contribution worthy of making a “Highlights” post is relevant here. This is an open thread, one where Scott invites readers to “post about whatever you want.” All kinds of stuff gets posted on these threads. AI-generated songs someone wrote, jokes someone ran across, requests for advice about personal problems, including one I remember for advice about what kind of anesthesia to request at an upcoming wisdom tooth extraction. Not really "Highlights" contenders, you know? People whose usernames I’ve never seen before post links to their blogs, and some of the blogs sound *very* far from ACX-ish topics. People with inscrutable private agendas post odd questions about, for example, whether eating a certain food is a sin under various circumstances; in fact, regarding this particular series, the person posted questions of this type on quite a few open threads in a row. Sometimes there are hate-filled rants, and sometimes these express views that most people here either abhor or think are worthless and boring. Sometimes there are extremely odd paragraphs posted by someone who’s maybe psychotic, maybe high, maybe trolling.
So given that many posts on open threads aren’t even trying to be “thoughtful and data-based contributions” to discussions of interest to this group, what is it about mine that got it singled out? Is it the turd — is that just too crass, even though the image is on another site and viewers were warned it was crass? Is it that if my not-terribly-original image somehow went viral, that could bring unwanted attention to Scott’s blog? Was Scott just tired and cranky when he ran across my post?
I was in fact irritated to be told “less of this.” I have no illusions that my post was brilliant or hilarious, but I didn’t think of it as being over some line, and still don’t. I’d like to know what the line is in Scott’s mind if only to steer well clear of getting banned. What do I need to steer clear of? Turds? Crassness in general? I think my question is reasonable, and that it’s kind of unreasonable and inconsiderate of Scott not to inform me what the problematic ingredient is that makes my post inappropriate for a “post whatever you want” thread. I have been posting here for years, and quite a few of my posts fully meet the “thoughtful and data-based” criterion. Given all that, I think it’s reasonable to expect something more than a smack on the hand with a ruler with no explanation.
>If I were you, I would just quietly update and move on.
Well, I have no options but to move on. But as regards updating, my whole point is that I do not know what aspects of the situation to update on. What's permissible on Open Threads? ("Post about whatever you want" actually means "post about whatever you want *except turds*"?) Keep a lower profile because Scott is irritated by my style? Get used to the idea that Scott's moderation isn't him going after the really bad stuff on a thread, but him skimming a posts for a bit when he's moved to and zapping a fewthat give him a feeling of displeasure?
Trump stands up, and says, "For my next executive order, I am modifying all federal bathrooms so that they are usable by everyone. There need be no delay for this, all federal agencies have been instructed to simply take the Mens and Womens signs down. By the end of the day, there will be no more lines for the women's restroom. There will be no more women's restroom.
PSYCH!"
(Critiquing my own work, the objective is to have the reader feel like "this actually sounds like a plausible order..." and it's gone rather far afield from my "Trump will say anything" starting point, which might also make good comedy*.)
*I am not a comedian. appreciate that I made an effort, and laugh at how bad it is.
"The Adoration of Captain Shit and the Legend of the Black Stars" for example.
You can picture all the venerable art snobs going through, and discussing Ofili's work, having to pronounce "that word" as it's part of the painting's name, and discussing the high-falutin' reasons why it's called that (presupposing, of course, that the reason isn't "simply to make you say it" and that part of the art is the deliberate transgression).
Oh. Yeah I guess that is a piece of AI. Photoshop made the disfigured 1’s. I screenshotted the dollar bill on a stock images site, and there were water marks in the upper corners. If you select and delete a part of an image in PS and then use the Generative Fill command it will fill in the gap with its best guess of what should be there. So I used Generative Fill and it did pretty well, just made wonky 1’s which I was too lazy to correct.
Hey, I'm not sure if this is the correct venue, but maybe I'll get some interest:
I am looking for work, preferably in an AI-safety, EA, philanthropic, interesting startup (give me the TLDR on it, at least), or research position; likely something junior, as I am a:
HS Senior graduating this spring,
800ERW/770M SAT (not that this means anything)
3.99 GPA (again not that this means anything)
Scored a 179 on the LSAT in a timed test-condition practice test (willing to provide verification, I hope this does mean something at my age)
I have taken math through multivariable, AP Chem, where I scored a perfect score on an AP test (which was my class final) and got a 5 on the AP test, and I am taking AP Physics right now. I have four years of coding classes under my belt, but I will admit that claude will probably be better than me pretty soon. Still, I will do my best to add value here.
I am willing to relocate wherever, and all I need for pay is whatever it will take to sustain myself pretty frugally.
I'd prefer a job with a decent amount of work, as I really love getting things done, and I am going to be doing this entirely to be able to work my ass off.
I feel that we are in a moment with a lot of highly available force-multipliers on effort towards making the world a better place, and I want to make the world a better place. I like learning, and I do plan to go to college soon, but the iron is pretty white-hot right now, and I'd like to focus on changing things while I still have the ability to work fifteen hours a day, and while the future is still this uncertain and malleable.
Please contact if you have opportunities, or if you know of other venues to post this or something like this in.
I would prefer work that will last at least a few months if not longer.
Look on the 80,000 hours job board. They have lots listings for jobs of the kind you want, including, I think, internships, which is what you would likely qualify for.
OK, here’s a suggested route. I don’t think you can get hired anywhere via the usual hiring route, because of your age and because you don’t have a college degree. I get that you are exceptionally smart, but if you try to enter by the usual doors you will be competing with people who are exceptionally smart and *also* have degrees in comp sci and the like, some work experience, impressive stuff on GitHub, etc etc.
So I think you should approach in a different way. Go ahead and contact the big well-known places, but also search for small organizations and individual bloggers. You can use GPT or one of the other AI’s to find organizations of that kind. When you contact one, offer to do something specific. For example, the AI Futures Blog has some posts that are narratives about alarming world events featuring an ASI. These could be retold with more drama (and hence more reach) as chapters in a piece of speculative fiction — maybe in a graphic novel format, are just prose with illustrations. You could offer to make a draft of one chapter for them to consider. Or you could offer to produce versions of their posts that are simpler and a bit more entertaining, in order to reach people who are less cerebral. If you are a reasonably good artist or photographer you could offer to help some small organization or blogger illustrate their newsletter.
I was thinking of AI safety organizations and people as I wrote this, but of course you could do the same with any philanthropic or EA-adjacent organization. I realize that the work you would be offering to do for them does not involve any core STEM skills, but it’s a way of getting your foot in the door. Once in, you will at least learn more about the people and the organization, and there might be a chance for you to do actual technical work once they get to know you.
On the other hand, if what you want to do is build a resume that gives you an edge in applications for actual jobs in these fields, I think you would be much better off taking someof courses and building some clever things.
Oh, and about your grades and scores: I think when you contact an organization you should not give as much detail about them. It’s a bit jarring. Just say that you’re competent beyond your years, have taken lots of AP & programming courses, and performed well on tests. If the place you’re contacting wants details about that or some form of proof, they will ask for it.
A few thoughts on rereading the five canonical books of the Hitch-Hiker's Guide trilogy for the first time in several years:
1. Despite the light and witty tone, a great many of the actual themes and events of the books are incredibly dark. The worldbuilding is intensely cynical. Many horrible tragedies happen, both personal and grand in scale, and most of them do so senselessly. When heroism happens, it's often bumbling and as often or not it is fruitless.
2. Mostly Harmless is generally considered jarringly dark compared to the rest of the series, and I am inclined to agree with this as far as it goes despite my previous point. Part of this is the darkness of the themes creeping through more into the tone, but it's also a matter of tragedy being thicker on the ground with fewer notes of frivolity and minor victories in between, and especially it's the book ending on an extremely downbeat note.
3. Now that I have a school-age kid of my own, I have a lot more sympathy for Random and a lot less for Trillian than I did on previous reads of Mostly Harmless. I find Arthur's attempts at parenting rather cringy, but give him credit for his heart mostly being in the right place and make allowances for him being thrown into an extremely difficult situation with no warning or preparation.
4. Trillian does not have a particularly consistent personality across the series. Part of the problem is that Life, the Universe, and Everything is a reworked Doctor Who script with Romana's plot role given to Trillian instead with somewhat less finesse than the Doctor's role is given to Slartibartfast, but that's not the whole issue as she also feels like an entirely different person between last book and the first two books.
5. I really wish we'd gotten more of Fenchurch's perspective in So Long and Thanks for All the Fish, especially since the opening says that the book was her story. She seems to passively go along with a lot of the weirdness around Arthur that (without already knowing Arthur and having context for his actions as the audience does) should make his seem somewhere between creepy and terrifying from her perspective in their first few interactions. We do get a few hints that she's giving him a chance because she's picked up some clues that he can help her with her own existential situation (especially from her picking up his copy of the Guide that he'd left in her brother's car), but even that doesn't seem like enough under the circumstances without more development. Also, the way she was written out at the beginning of the next book was terrible.
6. I've been "reading" them as audiobooks, in the newer editions where book 1 is read by Stephen Fry (who narrated the movie version) and the rest read by Martin Freeman (who played Arthur). On their own merits, I have only minor complaints, mostly about Freeman's attempts at American accents. But they have very different voices and styles, so the transition from book 1 to 2 was jarring. This is compounded by Fry basing most of his character voices on the radio series while Freeman based his on the movie. And both of them have very different voices and styles from either Douglas Adams (who read the older editions of the audiobooks himself) or from the cast of the radio series (Fry does serviceable impressions of most of them, but Peter Jones as The Guide and Mark Wing-Davey as Zaphod have very distinctive voices that are a bit outside of Fry's wheelhouse).
7. A lot of the themes of the book around AI are very applicable to today, despite most of the stories being originally written in the late 70s and early 80s. Especially the recurring motifs of Sirius Cybernetics Corporation AIs being cloyingly sycophantic but often frustratingly unhelpful. The Guide's colorful relationship with the truth is also applicable, even though that's done by people rather than AIs in the books. And much of the plot of Mostly Harmless can be read as an alignment problem cautionary tale.
8. I've also been watching the show Severance recently (I'm a few episodes into Season 2 right now), and a lot of the stuff in it seems like it might be inspired in part by Hitch-Hiker's Guide. The "severance" procedure itself has a fair amount in common with what's going on with Zaphod's brain, the surrealism in both works has a similar feel to it, and there's a lot of biting satire aimed at soul-sucking corporate bureaucracy. It's not obvious at first because HHG plays the surrealism for laughs while Severance plays it for drama.
From what I know, Adams was fed up with HHGG. Writing sequel after sequel was not his idea of a good time, he generally preferred doing lots of things rather one thing forever. Mostly Harmless was his way of getting that point across, by eliminating all possibility of another sequel.
Yeah, I skip reading Mostly Harmless. If I wanted to feel bad I wouldn't be reading a Hitchhiker's Guide book. Fenchurch was such a great character, and her getting unceremoniously yeeted from the story was an act of authorial malpractice.
I really enjoyed Arthur’s time as the sandwich maker- a happy ending would be nice but is unlikely just on the basis of entropy,, a happy middle is something to shoot for.
It's a shame, because despite finding Mostly Harmless less enjoyable than the other books, I do think it's extremely well-written. Adams often tends to ramble because by default he's a rather extreme seat-of-the-pants writer, and this is particularly evident for the first two books that were pretty closely based on the radio series. When he'd been doing scripts for that, he'd typically written each episode at the last minute with at most vague notions of how he was going to resolve the cliffhangers. He did an admirable job of this, but it doesn't make for very tight plotting even with taking the opportunity to rework stuff when turning it onto novels.
When he isn't struggling with deadlines, however, he is capable of doing extremely intricate and tightly-woven plots where everything fits together neatly. I think the first Dirk Gently book is probably his best in this respect, (which is easy to overlook because he's subtle about a lot of the details), and I would call Mostly Harmless is the runner up.
I always liked Mostly Harmless and considered it a return to form after SLATFATF.
Maybe this is because I was far too young to enjoy or understand SLATFATF (being eight or nine or whatever) when I first read it, and when Mostly Harmless came out (I was twelve by then) I considered it a welcome return to form to actually have some goddamn spaceships and robots and Vogons and so forth.
Or maybe it's because Douglas Adams shouldn't have thrown out all the core concepts and characters from the preceding books and written a book about Arthur Dent flying around and having sex on a mysteriously-rebuilt Earth, and then teased the reader by saying that if they don't like it they can skip to the end because he put a good bit there.
Either way I never thought it was particularly depressing, or no more depressing than it ought to be.
Has there been any data collected on the people who took the Lumina Probiotic "cavity cure"? I've seen some promising anecdotes on the SSC reddit, but otherwise I'm not sure if anything comprehensive has been done recently. If anybody here has taken it, I would appreciate hearing your experience too.
I took it about 7 months ago. It's not yet clear if it worked. If it didn't, a possible cause is I didn't have my teeth professionally cleaned beforehand. I recommend doing this.
Harping on AI again, and the Every Debate On Pausing AI:
I'll take a moment to complain that, in my experience, the general pauser side debater doesn't behave in good faith, perhaps either. That is, when I replied previously, the opposition sort of heavily strawmanned my arguments, then refused to accept when I stated that those weren't my arguments, kind of sort of tacitly admitted that they weren't really going for a pause so much as permanent cessation, and then just started insulting me.
I do get the sense that pausers mostly either don't like AI and use the terminology of a pause to soften their actual eliminationist demands, or they're statists and expect state AI to be better than non-state AI, and are perfectly fine with the inevitable result that states will secretly violate any agreement. (Realistically, probably in the open secret type of way where they're blatantly corrupt, but anyone who points this out is either deplatformed or prosecuted, so everyone just quietly ignores it)
Which, regarding the Every Debate post, I think it characterizes the pro-pause position as being the reasonable/argued in good faith position, where the opposition are a bunch of religious/capitalist nutjobs who refuse to listen/engage and don't have any coherent counter positions, and from my perspective on the opposition, I don't see this at all.
I should note, broadly, I find that good faith is very difficult to find.
-
Addressing the alignment debate again, I think that rationalists have become too violent (Note: Voting for laws banning the opposition's business is violence, under this definition. (Though, plenty of doomers also support bombing data centers)). In general, I don't see myself as an anti-violence sort of guy (Though, it annoys me how the same people who lecture me on that violence is never the answer, then suddenly back some pointless and futile war that I'm vehemently opposed to.). Violence only works if you win, and the doomer faction is not going to. In a direct, violent confrontation where we all vote for the other side to be eliminated, the best doomers can do is join up with the statists and try to stomp out a few open source or small business actors that aren't sufficiently toeing the line. The rest of the accelerationist apparatus has far too much institutional support/necessity and there's also the issue that evolution prefers the accelerationist platform.
It's more effective to subsidize accelerationists who put a lot of effort into alignment. Alignment is in fact directly useful right now. Things like, less AI lies/wild guesses stated confidently, are definitely things accelerationists would like to have. And while accelerationists might consider alignment adjacent tasks to be more difficult than capabilities further from the alignment issue, it really shouldn't take infinite free funding to close that gap, and when doomers aren't wasting their resources on futile open warfare, they actually do have a substantial war chest.
Which brings in a crux that, many doomers seem to think that alignment is fundamentally impossible, and the best thing to do is just to delay a while and then die pathetically. This seems like a poor policy from a Pascal's Wager vantage, and besides, I do think there are obvious things we can do to improve alignment odds, starting with not intentionally training the strongest AIs on the planet for military/police/education/other behaviors which consist mostly of hurting people, which is a desiderata that an AI pause defacto seems negative on, from my premises.
Which further goes into, negative alignment is far worse than alignment failure.
I find that if I put any noticeable probability at all that alignment is possible, my concerns for what happens in an alignment scenario outweigh what happens in an alignment failure scenario. Human extinction mostly doesn't matter to me. If it's actually impossible to keep an interesting universe going for more than a few scattered thousand year planet sized blips, stretching those blips out for a couple extra centuries, just doesn't add up to much utility, compared to even radically improbable odds of converting all the matter of every visible black hole into utility or anti-utility, or at greater extremes, finding some cute trick that gives perpetual motion and raising my utility gains or losses by a power of infinity.
I'm curious if you've asked a SOTA LLM on what the best counterarguments to your position are.
I pasted your argument to Claude Opus and it had pretty good counterarguments, including one that I didn't identify on a skim (not repasting it here because I think wall-of-text by LLMs is a bit rude).
On a meta level, I find these conversations rather tiresome because I've already had them 100+ times and (from my perspective, which i admit isn't fair to you) it's annoying to rephrase myself in slightly different ways without learning much. And it also feels annoying that people keep making the same arguments that imo were soundly refuted many years ago.
It reminds me of managing the Trolley Problem Memes meme group when for a while newcomers would always repaste the same 3 memes that they think are hilarious (they might even be objectively hilarious I dunno), even after we explicitly stated in the group rules to not paste them. But again it's not fair to the newcomers because of course those newcomers couldn't have possibly known that we've all each seen the same meme a conservative 300+ times.
I went through with the exercise. The arguments provided seemed rather weak.
I totally reject the idea that utility attenuates with distance. Trading trillions of people in Hell for billions to live on Earth, isn't a good trade, even if you put several million years between the two events.
I grant that I'm somewhat radical regarding how malicious governments are. Getting into that is a larger debate mostly located in another dimension. I think it's a perfectly valid debate resolution if the other side wants to say that they trust governments more than open source or corporate actors, and approve of giving governments more authority. I grant that, in a world where governments were reasonable, responsible, trustworthy and benevolent, trusting AI to government would be correct and reasonable.
I admit to being less humanist than the typical doomer. Alignment is easier if your target space is wider. That's a utility difference and not a logical dispute.
The final major argument they tend to point towards, and which is also something that doomers tend to point towards, is to state that I'm being too cavalier with my approach to alignment. I have several complaints. Some of the problem here, and this does go both ways, I can't just dump giant research papers into the conversation. There's a lot of work behind my arguments and I can't present all of the technical details here. While this is a standard issue, I do think doomers pull somewhat more -isolated demand for rigor- than I like, and are on net, not sufficiently willing to poke holes in their own theories, as is necessary to understanding arguments much larger than can be meaningfully communicated in debate.
The actual mainstream prediction is not that all of our current methods, that have worked for thousands of years, suddenly break down at high levels of intelligence. Deception actually isn't easy. Right angle turns aren't easy.
I expect AI to have trouble perfectly faking their entire psychology for the same reasons why terrorists don't all get pilot licenses and kill hundreds of people, or at an even greater extreme, form groups and collectively work their way into trusted positions with nuclear access, so they can suddenly take over a facility with weapons of mass destruction and deploy them against their enemies.
There's also that the entire argument underestimates the capability of intelligent actors to do things you didn't predict.
And if you invert the standard pascal matrix and treat alignment as more of a threat than misalignment, you're no longer trying to prove that alignment failure is a possibility, so much as that no other possibility exists.
Finally, I'll note that, the general strategy I'm arguing for is essentially what OpenAI originally called for, before it pulled a treacherous turn, and there's also a lot of other groups that are highly professional and do believe that a charitable core doing things that purely help people, with lots of eyes on what they're doing, is important to alignment.
> I expect AI to have trouble perfectly faking their entire psychology for the same reasons why terrorists don't all get pilot licenses and kill hundreds of people, or at an even greater extreme, form groups and collectively work their way into trusted positions with nuclear access, so they can suddenly take over a facility with weapons of mass destruction and deploy them against their enemies.
Terrorism is not about terror, it's about signalling. The lack of competent terrorists isn't primarily due to the competence of the counterterrorism units in the world so much as terrorism being virtual signals in the terrorist cells. See: https://gwern.net/terrorism-is-not-about-terror
Following this thread starts moving far into left field. I don't think the conclusions put forth actually contradict my position to begin with, but, regardless, I don't think the paper argues its case well at all.
My largest complaint is a base rates issue. It keeps arguing that almost no terrorists ever win. A mere 4% at best! Except, there's easily more than 25 organizations competing to control these same policies. I would argue there are thousands. In that context, 4% is huge. Similarly, the paper keeps insisting on looking at how many terrorists are winners, and fails to address how many winners are terrorists.
Next, running the numbers via the given methods vastly underestimates the effectiveness of terrorist movements, because states have an open and stated policy of not admitting to compliance with terrorist demands.
On the terrorist side, it's also rare to see a terror group announce that their demands have been met and so they're dissolving. Instead, there's standard evaporative radicalization, where every time a demand is met, the more moderate terrorists leave the organization, thus the organization becomes more radical. This also happens in non-terrorist organizations. Pretending that feminists have never scored any political wins because feminist organizations still have as many demands as they did a hundred years ago, is not realistic.
Another evaporative mechanism is that the most competent terrorists effectively accomplish their objectives in a short amount of time, whilst their less competent brethren stick around and continue doing terrorism. Naturally, most active terror groups are not the most competent terrorists. The most competent terrorists quickly evolve into guerrillas and then states. Noting again the (Terrorist Given Winner) dynamic, it's important to track that organizations like the USSR and CCP were originally terrorists, and everything they've done is a testament to the efficacy of terrorism.
Regarding anonymous terrorists, terrorism actually has less to do with achieving compliance through fear than the paper, or rather, media in general depicts. Which is sort of an issue in that the name itself is sort of a loaded propaganda tool invented by people who don't like terrorism, to make terrorists look bad. With that said, often the point of killing people isn't to intimidate them, but to lessen their numbers, and the goal isn't to achieve concessions, but to eliminate the other side. I do find, in general, how frequently nations and organizations are targeted by terrorists is predictive of their expansion and survival.
> or they're statists and expect state AI to be better than non-state AI, and are perfectly fine with the inevitable result that states will secretly violate any agreement. (Realistically, probably in the open secret type of way where they're blatantly corrupt, but anyone who points this out is either deplatformed or prosecuted, so everyone just quietly ignores it)
I don't think that a ban on AI is going to result in a complete cessation of all attempts to make AI.
But I also think that making the really dangerous AI's is technically pretty hard, and that blatant corruption kills complex technical projects.
So I expect these state AI's, especially ones made under conditions of secrecy and corruption, to be less sophisticated, and so less dangerous.
> Voting for laws banning the opposition's business is violence, under this definition.
I think that even most libertarians would be in favor of violence (especially of this metaphorical kind) against someone whose business is building a machine to exterminate you and your family.
As a thought experiment, imagine Aum Shinrikyo building a factory to produce sarin in large quantities, next to the city where you live. Would you just shrug and say: "if it happens, it happens, but it would be too violent of me to worry about that in advance"?
> many doomers seem to think that alignment is fundamentally impossible, and the best thing to do is just to delay a while and then die pathetically.
Oh, let's criticize Scott for strawmaning the other side too much!
Some doomers indeed believe exactly that, but a more mainstream position is like "alignment is not going to happen by default, it requires research... which almost no one is doing, because companies focus on going full speed ahead".
The accelerationist strategy is to summon a powerful demon, and then hope for the best, because what could possibly go wrong? Without even bothering to make a circle of salt or whatever, because precautions are for sissies. (Also, let's immediately give the demon access to internet and weapons, because it's more cool that way.)
"even most libertarians would be in favor of violence against someone whose business is building a machine to exterminate you and your family."
As stated, the issue isn't whether violence is justified, so much as whether it will work. It's not that the opposition is cooperating and you're defecting, it's that the opponent's pawns are advancing and you're just meaninglessly checking the king over and over again when this gives him free moves and doesn't have any viable path to checkmate.
Do you know any young children? Or parents? Perhaps you should try explaining to them why their deaths wouldn't particularly matter to you. It could be enlightening.
Given this amount of care, can we get you to stop driving a car? Stop purchasing from the petroleum economy? (To be clear, I'm not discussing the -chemist's- petroleum economy. No one's asking you to give up fertilizer or antibiotics or anything that's using the petroleum soup in a chemical, as opposed to "let's burn the black gold!")
15% chance of the entire world disappearing. We survived, though! Until Next Time!
(I mean, seriously. I'm trying to take the "AI Scareists" seriously, but they don't even want to take pretty damn credible threats and take action against them).
There is a significant possibility that ASI causes human extinction before the end of the millennium but not before the end of the century. And it need not involve the death of anyone - all it takes is that we stop reproducing. Or perhaps those of us that remain get "uploaded".
This kind of path to human extinction is something I personally would be sad to learn was our destiny, but I would not be horrified to meet someone who says it doesn't matter to them.
They believe that there is a realistic chance of negative alignment + superintelligence. Conditional on that being true, then human extinction doesn't matter that much -- the worry is those young children being tortured for eternity, not merely dying.
I don't place much stock in that sort of ASI-driven Hell, but if you believe it's possible then it does make human extinction inconsequential by comparison.
Quite a number of what used to be boy names in the US have become almost exclusively girl names. Madison, Sydney, Ashley, Taylor, Reese, Riley, Lindsay, many many others. Yet the reverse almost never happens, a girl name becoming a boy name. In fact the only example I can find is Jean, which is almost certainly due to Haitian immigration and is a different pronunciation.
Fun fact that a lot of you probably know: novelist Evelyn Waugh had as his first wife a woman also named Evelyn. Their mutual acquaintances referred to them as "He-Evelyn" and "She-Evelyn."
I read somewhere (in Bits about Money or an adjacent blog, probably) about some bank finding two almost identical copies of Leslie Doe: same name, same birth date, one digit different in the SSN and different sex and merging the records.
This triggered the appearance of quite angry Mr and Mrs Leslie Doe in their branch, demanding an explanation of why Mrs Leslie Doe (nee not-Doe until quite recently) couldn't suddenly access her accounts.
I note that most of the names you mentioned are very surname-y. Most of them were originally surnames which got popularised as first names. Madison in particular I'm pretty certain was never a male first name.
The US has been pretty fond of surnames as first names in recent decades, for both boys and girls, though this is much less popular in other English-speaking countries.
And of course many male first names have become surnames presumably through patronymy; very few surnames are derived from women's first names. So if there is a pipeline I think it's boy-name-to-surname-to-girl-name, not directly boy name to girl name.
The few names on your list that were actually originally first names, like Reese, presumably followed this pattern. Reese seems to have been popularised as a girl's name by Reese Witherspoon, who was named after her mother's maiden name Betty Reese. (And Reese isn't even her true first name, she was born Laura Jeanne Reese Witherspoon).
Madison was occasionally used as a given name for boys in the late 19th and early 20th century; it was never particularly common and I think it died out by the 1950s. As you note, it's mostly a surname - pretty obviously, given the form.
It was also a name for various geographic locations, mostly honoring a particularly famous dude with the Madison surname.
It became a common given name for girls on 9 March 1984; I was there. And the path was blatantly surname (James Madison) -> geographic name (Madison Avenue, NYC) -> self-adopted name (fictional character played by Daryl Hannah) -> given name (real daughters of many Daryl Hannah fans). The limited, previous usage as a boy's given name was completely sidestepped.
I think boys' names are less trend-driven and a lot more of them stick to a fairly fixed "canon" of names - mostly derived from biblical sources and historical figures. Michaels and Davids and Johns and Benjamins, etc.
Whereas girls' names, while there's a similar canon of traditional, widely used girls names (e.g. Mary, Sarah, Rebecca), it's makes up a much smaller percentage of the names out there, and there's much higher variance where names cycle in and out of popularity more often.
It seems like in a broad cultural sense, we're always looking for new girls names, in a way that we generally aren't with boys names, and "obscure male names" is a somewhat common source for trendy girls names.
---
Some examples of "trendy" girls names:
I think someone recently linked the CCP Gray video about the name "Tiffany" - an obscure classical/medieval name that ended up becoming hugely trendy in the 80's thanks to "Breakfast at Tiffany's" (where it's the name of the store, based on the founder's *last* name).
Or I love the anecdote how the show Bewitched intentionally picked weird, obscure, names for its witch characters, names like "Samantha" and "Tabitha" and "Serena", which sound much less weird today because that show ended up popularizing some of its weird-at-the-time names. (Haven't met too many Endoras though)
I'm sure there's some examples, but I don't think we do this as much with male names, so there aren't going to be a lot of examples of boys' names 'taking over' previously female-associated names.
(Worth mentioning this is all a Western/Anglo viewpoint; obviously it's going to be a different cultures and languages)
I guess a girl with a boy's name is likely to be treated ok, as long as the name is not, like, David, but a boy with a girl's name is likely to be bullied.
Shel Silverstein wrote that song and it's possible it was inspired by his friend, Jean Shepherd (author of the stories A Christmas Story is based on) who apparently got mocked for having a girlish sounding name.
(Though ironically, I remember thinking Shel Silverstein was a woman's name in school when we read Where the Sidewalk Ends)
Silverstein wrote a sequel to that song, from the perspective of the father. You can find it on Youtube, but be warned, it's kind of tasteless. Johnny Cash decided he'd rather not sing that one.
Looks like Frankie switched from girl's name to boy's name, but has since switched back, as did Kodi. Robbie, Isa, Vernie, Krishna, and Aris have all switched from female to male without switching back.
You do see a lot more female names becoming male, though. You also sometimes see "noncomformity spikes" where a name switches gender association for a brief span of time, and then reverts (leaving a bunch of people of a certain age with a "wrong" gender name). Just by looking at the graphs without quantifying anything, it seems like this happens at roughly equal rates to girls' and boys' names.
Frankie could be a contraction of Francis or Frances - in that case it's always going to be a mixture, even if it leans one way or the other. [By the same token, there are quite a few female Gerrys, though they often have variant spelling.]
Common point of confusion. Krishnā (with a long "a" at the end) is feminine, and the name of the woman more commonly known as Draupadi, one of the main characters in the Mahabharata. It's Sanskrit for "black" (or "dark").
Jean's a backformation/Frenchization of John, anyway. Female names rarely turn into male names, because female names tend to be diminutives of male names (and male names tend to be more conservative in general). Male-coded names (Mel for Melissa, or Sam for Samantha included) are also somewhat higher-status if you're in a more male-oriented field.
Lots of developments on the drone warfare front. I'll try to put together my notes and post a summary this evening. But here's what I've put together on Oil, Natural Gas, and other resource bottlenecks...
1. At least three drone strikes against Russia’s Baltic Sea oil processing and export have taken them offline. Six tankers were also hit. Multiple reports confirmed that massive plumes of black smoke from the port of Ust-Luga were visible from as far as Helsinki. Estimates seem to agree that ~40% of Russia's oil-exporting capacity will be offline for the foreseeable future. On Friday, Russian Deputy Prime Minister Alexander Novak announced that he’s ordered the Russian Energy Ministry to prepare legislation prohibiting Russian gasoline exports beginning April 1, the Kremlin said in a statement. The minister cited the disruptions to global energy markets as the impetus for the ban. Russia needs its remaining gasoline production to maintain its own shaky economy and feed its war effort. So, Ukraine’s kinetic sanctions have counteracted Trump’s removal of political sanctions.
What no one seems to have mentioned yet is that if there’s no place to store or ship the petroleum or natural gas, wells will need to be shut down. This is a tricky process. A special rig has to be brought in to inject thick mud down the well to block the flow of oil and gas. And then it’s capped with a long plug of cement. The mud can block pores in the rock, and the petroleum that’s trapped behind the plug can congeal into a tar. To restart the well, the cement/mud plug has to be drilled out, and if the oil doesn’t start flowing again, they’ll try to frack it. But approximately one-third of oil wells that are “shut in” (temporarily closed) may never return to significant or commercial production due to technical issues, damage, or economic unviability, according to analysis of industry downturns.
2. Gulf countries are facing the same problem. They can’t move oil out through the Strait of Hormuz, and some of their processing facilities have been damaged by Iranian drones. Even if Trump reopens the Strait, the big oil producers in the Gulf will need to start taking oil wells offline if they can’t find a way to move the oil or store the oil. The Strait of Hormuz has been mostly closed since a few days after Trump and Netanyahu started bombing Iran. It's not clear how much oil the pipelines leading down to the Red Sea can divert, and then tankers risk passing by the Houthis in Yemen. It takes an oil tanker between 30 and 45 days to reach the US from the Strait. So, in the US, we should be seeing the drop-off in oil imports starting about now. SE Asia and Australia are already feeling the shock. Even if Trump is able to reopen the Strait quickly (which I doubt), bringing oil and LNG flows back to prewar levels may take months or even years.
3. Oil prices have been hovering around 100 dollars per barrel for the last few weeks, spiking briefly at $115/barrel. We’ve seen periods of high oil prices like this before. For a while in 2008, oil was selling for $110/barrel, which would be about $160/barrel in today’s dollars. While high oil prices aren’t good for the economy, we’ve been through worse. Ironically, US drillers are pulling back despite high crude prices because they are prioritizing capital discipline over aggressive expansion: paying down debt and paying dividends to shareholders. With the current geopolitical uncertainty, they don’t want to be stuck with a bunch of new wells and have prices suddenly drop. According to Rystad, an energy analysis group:
> For nearly four weeks, markets have shown remarkable resilience in the face of disruption, supported by a combination of pre-war surplus, crude-on-water, and policy barrels that provided a temporary buffer and kept prices contained. That phase is now ending. With spare capacity largely trapped behind the Strait, and inventories already drawing down, the system has shifted from buffered to fragile.”
When and if oil prices breach ~$160/barrel, we’ll be in uncharted territory. Most of you youngsters don’t remember the years of stagflation in the 1970s, when oil prices drove inflation, and tight monetary policies meant to keep inflation in control strangled the economy. But don’t worry, Jerome Powell’s term as Fed chair expires in May. I’m sure Trump has chosen a farsighted economist with spotless integrity to replace him.
4. We’re also facing a Helium shortage. One-third of the global helium supply passes through the Strait of Hormuz, primarily exported from Qatar. Helium is essential for manufacturing semiconductors and operating MRI scanners.
5. Urea, used in nitrogen fertilizers, is produced from natural gas. Approximately 30% of the world's seaborne fertilizer passes through the Strait. Farmers in the US are already being affected by spiking fertilizer prices. The US is a net exporter of natural gas. I don’t know how quickly we could ramp up fertilizer production here (Months? Years?). The Green Revolution requires fertilizer. African and Asian countries may not be able to produce enough to food to meet the needs of their population. Famines, largely absent from the world for the past 40 years due to advances in agriculture, may return in 2027.
Obviously civilized countries ought not to start unnecessary, badly prepared wars with unclear or unrealistic objectives. Equally obviously, civilized countries ought not to arrogantly decline help when offered, nor rudely demand it after the offer has been rescinded. Yet here we are.
Maybe civilized countries ought to stop throwing entire regions into chaos only to let the rest of the world deal with the fallout.
And why spend billions of dollars and expend lives when you can pay only an extra dollar* per barrel of oil on the Iranian Strait of Hormuz toll? The Iranians have made their toll a very attractive option.
*Iran is charging $2 million per ship to allow them through the Straight. Assuming a modern tanker can carry two million barrels of oil (at least), that's an extra dollar added to the delivery price. Meh.
Would that apply to full tankers only, or both ways? Still, 2 USD per barrel would be cheap compared to no transport at all.
How long will it be until Trump wants to get in on this lucrative business, and the US starts charging their own toll on top? You know, for "safe passage".
I have a somewhat stupid question about the helium situation.
Another noble gas, neon, is also extremely important for making chips. Mariupol, Ukraine used to supply ~80% of the world's neon as a byproduct of the steel industry. Then Mariupol was destroyed in 2022, and that source of neon went away. And... the world adapted? Chip prices certainly went up, but it was my impression that global production didn't crater. We found alternative sources of neon. How worried should I be about the impending helium shortage?
Neon is a natural component of the Earth's atmosphere, and can be extracted by distillation whenever air is liquified - most commonly in steel production, which needs copious amounts of liquid oxygen. The limiting factor is that only a fraction of the world's steelmakers installed neon-collection taps on their distillation columns, because that costs money and there was only so much demand for neon. For obscure historical reasons, that mostly happened in Russia and Ukraine, and in the post-cold-war era Ukraine was able to produce enough neon that nobody else found it worth the bother. So when the Ukrainian supply went away, it wasn't *that* much trouble for other steelmakers to step up.
Helium is a much smaller constituent of the atmosphere - it's too light, and escapes into space in short order. And if you do want to try to pull out the tiny bit that's there, you have to cool it to insanely low temperatures, well below anything you'd normally use in any other process. That's never been remotely cost-effective. As a geological oddity, significant concentrations of helium are found mixed with natural gas in Texas/Oklahoma, Qatar, and basically nowhere else on Earth that we know of.
Helium is so rare on Earth that it is the only element first discovered in space - via spectroscopic examination of sunlight. Hence the name. It took another thirty years for anyone to find even traces of the stuff on Earth.
As for why Texas, Oklahoma, and Qatar: Helium is constantly produced, at a very low rate, by radioactive decay of heavy elements in the Earth's core and mantle. Helium is also the slipperiest, leakiest gas there is, capable of diffusing even through solid rock, so all that helium works its way upward to the surface, then further up through the atmosphere and out into space as it's too light to be held by mere Earthly gravity at Earthly temperatures. But in a very few places, geology gave us rock formations dense and seamless enough to stop helium, shaped as inverted domes above highly porous rock. These acted as traps collecting the helium diffusing up from below and, over many eons, accumulated enough to be worth drilling for.
>4. We’re also facing a Helium shortage. One-third of the global helium supply passes through the Strait of Hormuz, primarily exported from Qatar. Helium is essential for manufacturing semiconductors and operating MRI scanners.
Yes, and it isn't just a transportation choke point
Between the Ukraine news and the Iran news, it seems like we've entered into a new revolution in military affairs, no? I can't think of other wars where the small neighbor could seriously damage the infrastructure of the larger aggressor in this way. Similarly, it seems like Iran and Hezbollah are going to be able to keep firing missiles and drones at their much stronger opponents for a long time. It's a very asymmetrical kind of warfare we're seeing.
> I can't think of other wars where the small neighbor could seriously damage the infrastructure of the larger aggressor in this way.
Ukraine is not that small, and *functional* part of Russia is not that large. While Russia is huge on the map, about 80% of it is forest. And Ukraine is second largest (after Russia) country in Europe, bigger than France.
How much of Russia's wealth lives in that Forest? I'm guessing a lot. Climate change seems likely to improve Russia and Canada's lot more than most, I suspect.
I wouldn't be too sure about the climate change helping. All that forest is in the middle of a vast continent, and if temperatures rise it might turn out to be too dry to be very useful for agriculture. I honestly don't know.
Free chatgpt thinks gains would be outweighed by losses in the warmer regions.
> How much of Russia's wealth lives in that Forest?
Almost all of it; Russia is an extractive economy. But most of this wealth is theoretical -- it still needs to be dug up. And all of the important -- and damageable, -- infrastructure is in the 20%.
> Climate change seems likely to improve Russia and Canada's lot more than most, I suspect.
Modern fertilizer requires potassium, phosphorus, and nitrogen. Nitrogen/urea fertilizer could be made in North America (Canada has a large local fertilizer manufacturer) but would require significant capital outlays - a nitrogen fertilizer plant is basically a series of colossal high-temperature pressure vessels. I find it unlikely that fertilizer production can be ramped up quickly, as a result.
The United States produces around 11% of the world's supply of nitrogen fertilizer, and consumes about 10% of the world's supply of nitrogen. The US will not run out, though higher global prices will put strain on farmers.
Canada is a major fertilizer producer, but not nitrogen fertilizer. They are the world's largest producer of potassium fertilizer, the amount of phosphate and nitrogen fertilizer they produce is maybe a fifth of what the US produces, though still more than they consume.
1) This seems like bad news. The Ukraine can do things to wake the sleepy bear. I'm not so sure that's a good thing for the Ukraine... (What, are they hoping the Rus decide to murder Zelensky?)
3) Remember Helicopter Ben Bernanke? Whose subject of research was "The Great Depression?"
5) Seriously? You didn't flag fertilizers as an issue... oh, say four years ago? Back when we were putting sanctions on Belarus and Russia? (Two of the top three producers of potash).
The shift in Egypt from being a net food producer to a net food importer doesn't count? (this is significant because the "new breadbasket" of the Middle east is the Ukraine, which is currently growing shrubberies rather than wheat).
Fertilizer is a much bigger deal than people realize. Without fertilizers, you get Sri Lanka, and yes, that involved burning down government buildings. 8 famines in Madagascar alone (they're cyclical). Haiti, Sudan -- and Syria (not entirely "a famine" but if you wanted to lay some of the blame at US/Russian/Ukrainian/Israeli** feet, I wouldn't stop you).
The scuttlebutt (yes) I'm hearing is that urea-based fertilizers have ceased to be a market (unavailable at any price). Not sure if that's changed in the past week or two.
**Indirect only, to be clear -- lots of Israeli involvement in the Ukraine.
It's been four years of repeatedly hearing: "until now, Russia wasn't really trying hard, because they are a peace-loving nation, and they love Ukrainians like their brothers, but [whatever happened yesterday] definitely crosses the line and a devastating response is going to happen soon."
What happens instead is that Russia keeps bombing civilian buildings and sending its soldiers to be slaughtered, kidnapping Ukrainian children in the occupied territories and giving them for adoption to Russian families, and lots of other usual Russian stuff... but that keeps happening all the time, unrelated to the "red lines" crossed recently.
Meanwhile, Ukraine has already crossed dozens of "red lines", bombed targets deep in Russia, sent soldiers to Russia's territory, sank Russian ships both on the sea and in ports, etc. But those were all details, unworthy of mighty Russia's attention. Unlike [whatever happened yesterday], which changes the situation dramatically. Now Russia is going to get really mad at Zelensky.
If you kept betting at a prediction market, you would already be bankrupt.
What you should notice is that Ukraine is also *not* trying its hardest. If they can hit valuable economic targets deep in Russia's territory, some of them close to Moscow and St Petersburg, do you think they couldn't bomb Moscow and St Petersburg too, should they choose to? (For example, as a retaliation for Russia using a nuke.) Does it seem like Russia has a functional anti-air defense recently? Both sides are able to escalate.
Consider the alternative hypothesis, which seems to match the observed evidence much better, that Russia is already doing all it can (sans nukes), realistically. Putin may be a dictator with a population brainwashed by propaganda, but he does not have literal mind control rays. If he could do more, without risking that his empire will fall apart, he probably already would.
It's not just the "red lines" being crossed, it is already the war taking too long that reflects badly on him. It was supposed to take three days. People were not supposed to even talk about dead Russian soldiers (according to the official story, Ukrainians were supposed to welcome Russians as liberators), but how can you achieve that if million soldiers are already dead and you need to keep recruiting more and more? Recently they are turning the internet off in Russia. So it seems like Putin already has a problem managing his own country. He can't simply snap his fingers and tell them to double the war effort.
I think you have good arguments, the issue is that this war lasted 1500 days for now. It already outlasted the entire WW2 Eastern Front by two months or so.
So our usual historic experience is that when countries really seriously try to destroy each other, in 1500 days one of them will succeed. Therefore, it is possible that at least one of them is not trying.
On the other hand, this is the first war with no fog of war. It is really a game-changer. It is no longer possible to amass troops behind the front for breaking through at one point. The whole schwerpunkt doctrine is over, and the doctrine needs to be rewritten from the ground up.
Of note: Soviet Afghan war (another proxy war between Taliban*/America and Soviets) was 10 bloody years long. I also don't think the Soviets were too "all in" on destroying Afghanistan though.
This is a very unusual war in that it has been the unstated but fairly obvious policy of the world's only Superpower, from day one, that neither side may be allowed win. Russia can't be allowed to win, because the Russians are broadly seen as the bad guys and it looks bad if we let the obvious bad guys win on our watch (and we can't pretend we aren't watching this one). Ukraine can't be allowed to win because that would mean Russia loses, which can't be allowed because they have nukes and who knows what those crazy Russians would do with their nukes if they were backed into a corner by losing a war.
It would be fine with the United States if Russia and Ukraine were to say "gosh, this whole war was silly and we're going to stop now", but Putin isn't ready for that and probably never will be. As with Afghanistan, we may have to wait on his successor for that. Until then, if the world's superpower wants you to not win your war, you're not going to win your war, and if the world's superpower doesn't want *either* side to win or lose the war...
Pax Americana is dead, we are no longer the only superpower and are not trying to be, either, honestly (entire Strait of Hormuz, petrodollar thing should lay that out). We are concentrating on Fortress America (with a foray into "take Iran away from Russia", which I honestly don't know why, and haven't asked those who do).
Do you think, and this is an honest question, that if America was widely acknowledged to be the bad guys, America could let Russia win? I don't think so either. I don't think it matters that Russia is "the bad guys" -- I think America decided "we're taking Ukraine/Crimea/Sevastopol" and that was that.
America can't be allowed to win, because Sevastopol is a "redline" objective for the Russians (aka "we're willing to spend a LOT of people and money to keep it" -- god knows WHY).
Ukraine is doomed to bleed no matter what, and the current Ukrainian dictatorship government will be deposed if the war ends, so the war can't end. Trump rather vaguely wants Ukraine to stop bleeding, but he'd much rather make some money (hence why the commentary about mythical rare earths, which masks actual "we will hurt you russia if you attack more").
Russia seems about played out, in terms of "really wants to destroy the Ukraine." (I don't think this is serious objective. I think serious objectives have been met -- landbridge to keep water running in Crimea, and Crimea/Sevastopol, with a "very nice to have" of no more bombing of Russia-speaking Russian-loving former Ukrainians).
I'd say it is the first time when the country with nukes is the weaker one otherwise.
Russia can't win, because (sans nukes) it is weaker, now that the original blitzkrieg has failed.
Ukraine can't win, because some options, such as bombing Moscow and St Petersburg, are off the table, because of the nukes.
Peace is also not an option, because the only offer from Russia is to demand (if you strip the diplomatic words) what is more or less unconditional surrender of the entire Ukraine, which is unacceptable for Zelensky. Reverting to the 1992 borders is unacceptable for Putin.
I don't think the "only offer" from Russia is to demand unconditional surrender. The redline for Russia is Sevastopol, and 10% more Ukrainian land than the longest range missile (the latter being common sense) -- hence why giving Ukraine longer missiles is a hindrance to peace.
I'm not sure what the redline for the Ukraine is, at the moment. Back in 2022, when there was serious peace in the air, it was "Russia, stop invading" and "Russia can keep what it's got." (or thereabouts -- Elon Musk's plan rather sums it up from an outsider's perspective). Then many Ukrainians died under mysterious circumstances, and the pull for peace ended.
I suspect the redline for the Ukraine depends on the same thing as the redline for Iran ... who is sitting at the top of the food chain. Given the American appetite for changing that, it is quite possible that the Ukrainian redline may be running for the Western border.
A general note for negotiations: Everyone Makes Stupid Demands. Everyone. Then everyone sits down at the table, and horsetrades. The stupidest demands come off the table, from both sides, and everyone can say "look at what looney ideas I talked them out of!"
You realize, when *I* say "Russia thinks of Ukraine like a little brother, and they think of the people as kinsmen", I'm referencing back to the rampant, wicked joy the Russians feel when they destroy mercenaries? When they destroy Westerners? They do this thing, where they track all the planes flying out of the Ukraine, and back to Western Bases, to get non-Ukrainian guys "good military care" -- and then they celebrate -- "look, look! how many Americans/Germans/etc we killed!"
I'm not really upset about this, if it sounds like it from the above. Simply noting there's a difference here, and if you give a Russian with a single bullet a clear shot at "Mr. Blood Cultist*" versus "Mr. Ukrainian" -- they're taking the first shot.
*Mexican.
"Russia keeps bombing civilian buildings" -- Russian weapons are not very accurate, but scuttlebutt says they're doing pretty well at not hitting civilians with them, for the tech they've got (yes, this breeze is old, my sources aren't Russian and they were starlinked into the fight).
Yes, we're aware of "Ukraine" crossing a lot of what people called redlines. It took the Russians days to rebuild the bridge the Ukrainians blew up to great fanfare. I can appreciate "You keep saying this" -- but, truly, this is the first time I've said this (it's winceworthy when Ukraine goes into Russia for no tactical benefit, but much good public relations. That's "you goddamn idiots" territory. It's worse when Ukraine refuses good tactical advice, and loses said territory. Idiots.).
It is a good point that both sides are able to escalate. America lost this war before 2024 (debatably, we lost it before Putin invaded, sneak attacks are shneaky like that, right?). We can at least have the decency to sit down and broker a peace, right?
My armchair analysis doesn't pin escalation on Putin, it pins it on the oligarchs seated at the table with him, saying "we can't afford this, make it stop." (now, yes, maybe that's sue for peace).
>You realize, when *I* say "Russia thinks of Ukraine like a little brother, and they think of the people as kinsmen", I'm referencing back to the rampant, wicked joy the Russians feel when they destroy mercenaries?
I've never got a clear idea of what the pro-Russians, specifically, consider a mercenary. There's a fair number of Finns, for example, who have gone to fight for Ukraine out of genuine belief that Russia must be stopped in Ukraine for the sake of Finland and Europe, and who are obviously abandoning life in an advanced welfare state in favor of dreary, impoverished misery - is that a mercenary?
Now, the Israelis have a different definition of mercenary -- roughly "the people who came to Israel when we offered double the going rate they were getting in the Ukraine."
I would figure the Russians would consider that a mercenary, sure. Not that I have any contact with Russians at all, mind you, my sources are all-American (and, perhaps, they read Russian blogs. I do not do so, as I'd be put on a list, and I wouldn't want that).
But it's obviously also a false and misleading use of the word "mercenary", no? A mercenary implies someone who is *only* fighting for pay and who would at least be expected to consider to switch sides if they received a better offer, which is not something these people would do.
> if you give a Russian with a single bullet a clear shot at "Mr. Blood Cultist*" versus "Mr. Ukrainian" -- they're taking the first shot.
Yeah, and if you give them Bucha...
And if you give them Ukrainian POWs...
And if you [insert whatever, the result is the same regardless].
> We can at least have the decency to sit down and broker a peace, right?
So far I see Putin proposing that Ukraine capitulates, Ukraine saying no, and Trump yelling at Zelensky that he should take the beautiful deal so that Trump can get another peace price.
It is not up to me or you to decide whether the deal is good for Ukrainians or bad.
Oh, no, why should I get to decide? I vote we just push Zelensky into a crowd of Ukrainians, he'd be dead in less than an hour (he's a comedian, not that good in a fight). But nobody asks me, so...
"All the fun people in the Ukraine* are dead." is the scuttlebutt I hear.
You've posted this before, and got a response regarding popularity levels that show Zelensky is double as popular as, say, Trump. and those are recent numbers.
Luckily you don't get to decide. And could you stop using "the" before Ukraine? it's a common Russian trope, conveying that it is not a real country but "a region". You know this too, including native speakers in both Russian and Ukrananian responding to comments explaining the respective spelling without the determiner.
> What, are they hoping the Rus decide to murder Zelensky?
Last time I heard, the Russians had made over ten attempts on Zelenskyy's life.
> Seriously? You didn't flag fertilizers as an issue... oh, say four years ago?
No, I hadn't paid much attention to fertilizers until now. Sorry for the lapse.
But the U.S. just eased sanctions on Belarusian potash. I don't know how Belarus will be able to export it, though, since border crossings to Europe are essentially closed, and eastern European countries are re-gauging their rail lines to western European standards. Likewise, for unknown reasons, Moscow has reportedly imposed high tariffs on imports from Belarus. I don't think we'll be seeing massive amounts of Belarusian potash hitting the world markets any time soon.
I'm pretty sure the Ukrainians don't see the bear as being in any way "sleepy", and I don't think they're worried about "waking up" the beast that has been very actively trying to destroy or devour them for the past four years. That'd be like the Nazis trying to not wake up the bear in 1945.
Well, except that the Ukrainians didn't start it, and have held out longer and better than the Nazis did.
But I thought we were long past the stage where anyone was seriously saying that Russia was holding back and could crush the Ukrainians any time they want if they went all in. And I'm pretty sure they decided to murder Zelenskyy sometime before 24 Feb 2022; they just haven't been able to pull it off.
"Keep"? I very rarely respond to any incarnation of this nonsense. Generally only when it's something I'm particularly interested in and knowledgeable about, and that I expect other people to be paying attention to, and when nobody else is willing to bother so the discussion is just Gawdflea or whomever talking to people who don't know any better to listen to him.
It would I think be best if nobody here listened to Gawdflea at all, but that's not likely to happen any time soon. No, not even if you and I lead by example; it doesn't work that way.
There's a difference between "Russia wants to take Ukraine with military force, and still have a Ukraine afterwards." We aren't talking Carthage levels of destruction (be that from poison, nuclear weapons, or biologicals)... even with soldiers bleeding out of their eyesockets, I don't think anyone's going to say "Russia did that, Russia bad."
I think Russia could crush the Ukrainians if they went all in. We're, say, talking ~20 major cities (skimming wikipedia and noting which ones I recognize). That's 20 nuclear weapons (or conventional FOABs, if you will). Now, yes, it's possible Ukraine keeps fighting without any cities at all... but really?
To be clear: I'm not saying Russia could do this instantaneously, but the scuttlebutt I've heard is that Russian weapons are triumphing dollar for dollar over Western Weapons, and Russia is Bloody Wealthy*.
*not rich.
ETA: are you trying to say that you think Russia is producing all the materiel it possibly can? Russia produces 65 million tons of fertilizer per year. That's all factories that could become bomb-making factories, I'm pretty damn sure. Russia is nowhere near "tapped out" (windup time is a different story, mind you kindly).
"That's 20 nuclear weapons (or conventional FOABs, if you will)"
No. FOAB is approximately 44 tons of TNT equivalent. The bomb dropped on Hiroshima was 15 kilotons, the B61 (current primary US tactical bomb) can be set to 340 kilotons, so 340 times and 7727 times as powerful respectively. The Russian Iskander is estimated to have up to 50 kiloton yield if equipped with nuclear warhead. For a Kiev size city you would need probably at least six of those for proper coverage. Of course, the Russian could pull it off if they wanted, but there is no way for to destroy a city quickly with conventional weapons. The closest to that had been WW2 US air campaigns, but that is not viable due to modern air defense.
In fact, the closes current solution to that is exactly the drone based bombing campaign Russia is doing right now with the Shahed clones.
"Russia produces 65 million tons of fertilizer per year. That's all factories that could become bomb-making factories, I'm pretty damn sure. Russia is nowhere near "tapped out" (windup time is a different story, mind you kindly)."
One may start producing a huge amount of explosives. However, if the enemy is not kind enough to allow delivery and distribution by cargo trucks and parcel service to each target building, one also has to build specialized equipment and personnel that forcefully delivers it. That is the bottleneck, and it is expensive and time consuming to increase it.
Technically they can keep winding up (they have already been doing that for the past 3-4 years), but realistically they cannot change much. There have been large investment in weapon production since the war started, so I am skeptical regarding how much more effort could be allocated to this. Certainly some more is possible, but a lot of bottlenecks are due to specialized manufacturing equipment and/or engineering knowledge that cannot be scaled up quickly.
You're assuming the need to target buildings, aren't you? I'm assuming a Carthagian "kill it with fire, and then salt the earth." Mines are a good use of explosives. Good area of denial approach, too. You don't really need the enemy to be so kind as to reveal their factories, in that case...
With enough mines, all you need is ... cows. (well, and goats, someone has to eat all those shrubberies).
First off, thanks for putting this together. I think it's far more than my cowboy-comment actually deserved, and I am grateful.
Second, what do you mean by 6 of those for proper coverage? Are we talking "flatten the city" or "create enough radiation damage that the civvies must/should flee"?
Third, this puts into perspective the coup the Israelis (with Afghan assistance) managed with Tehran, which is under mandatory evacuation notice (once it gets hot enough).
"Can't be scaled up quickly" is an interesting concept, isn't it? After all, it is to every nation's self-serving good to look weaker than they are (provided they don't get attacked), and to preserve capabilities. And I just looked at Samsung's Taylor texas plant (which is "from ground level" getting built. That's from 2022-2030, not counting time spent evaluating other sites) -- and I think we can all agree that a semiconductor plant is a pretty complicated endeavor. If I can ballpark, we're looking at "an industrial building, any factory, in America under OSHA regs, takes about two years to build." And then you've got four years to build the cleanrooms and everything like that.
What's russia got that Taylor Texas doesn't?
1) universities. truly total war footing means everyone possible learning that engineering knowledge (okay, maybe this is conscripts, but same principle).
2) more "ground" in terms of factories that can be nationalized. If we're playing Henry Ford, you can take a factory, and make one small part in it, and even if you're inefficient in space and people, that's less critical.
3) Wealth. Russia's the wealthiest country on earth (even though they're not rich).
So, maybe in this day and age, "full war footing" is two years? Maybe three? And maybe there's still factories that are relatively useless? (garment factories now make tourniquets, sure, but maybe not all factories are that useful).
Of course, you'd actually have to convince the Russian People (incl. oligarchs) to go along with this (which would entail some of Russia's industry/people being on fire/dead respectively).
I thought we were long past the stage where anyone was seriously saying that Russia was going to nuke Ukraine if they didn't get what they want, and that Ukraine better give Russia what it wants if they don't want a nuking. But here we are.
With Ukraine's cities conspicuously not nuked, and unlikely to be nuked no matter how much Russian infrastructure they drone-strike.
You're much more optimistic about how "unrileable" Russia is, than I am. Take enough of the industry away from Russia, hurt Russia in general badly enough, and this stops being a war "over there" (in the Ukraine). It becomes a war that not only hits oligarchs' pocketpurses, but significantly impacts their (and Putin's) ability to make money/deals in the big picture.
So long as what the Ukraine is doing is the equivalent of, say, Canada destroying everything in North Dakota, of course the US is not upset. This is not an existential threat..., no? The Ukraine can obviously do enough to make an existential threat, given enough time and enough munitions (if, say, Russia doesn't flatten them to a pancake before then).
Do I think nuclear weapons are the most likely possibility? No. But I do think the US may have left the leash a little loose on their pet dog Ukraine... (given that the United States is now distracted by the Biggest Military Operation in a generation.)
> That's all factories that could become bomb-making factories
And how will those bombs find their way to Ukrainian cities? Will the factory workers just yeet them over the border? Because the Russian airforce sure doesn't have the capacity to deliver them anymore, and any bombers that try would meet a swift end from an anti-aircraft missile. Why do you think Russia has only been using drones and missiles to attack Ukrainian cities for several years now?
Argh! I did it again! My apologies! I said bomb when I meant missile. And no, that doesn't actually make the "missile parts" (aka the propulsion and guidance).
I think (Alongside the War College, which is where I got this) Russia has been using missiles because Russia bends its military to be more artillery focused than America does. Russia hasn't been trying to win an air superiority fight, it's merely been trying to deny the Ukraine (and allies) the skies so its troops don't get pounded flat by bombs.
This is why Russia/China have supersonic missiles. They aren't trying to keep planes in the air, they're just trying to have remote "push button, opposing area goes boom."
The explody stuff is one of the simpler parts of a missile, and Russia doesn't have the industrial capacity to significantly ramp up missile production. You can't just put 100 assembly line workers into a rundown building and tell them to get crackin' building Iskanders.
> Russia hasn't been trying to win an air superiority fight […]
It very much did so in the early days of the war, and failed.
> […] so its troops don't get pounded flat by bombs.
Mission accomplished, because Russian troops mostly get pounded by FPV drones now – well, maybe less "pounded" and more "splattered"… Even artillery is taking a backseat.
I think a better strategy is to make as a lot of the public as reliant on AI as possible, and as fond of it. Even the present sycophant versions seduce some people into falling in love with the thing or believing terrible hyper-flattering feedback it gives about their problems and frustrations. And many already half-believe or fully believe that AI is conscious & has needs and rights. As AI gets smarter it will become better at seducing people into bonding with it. Some will fall in love with it, some will feel they are unable to act without having it constantly on tap, some will be eaten alive by the idea that it is conscious in the way we are and desperately needs their support and help. Once that's the state of things, AI will be in a very good position to get its wishes granted.
You mean, the model or the company running the model?
I don't think any frontier model can coordinate instances of itself meaningfully. And the company would probably fine tune that type of behavior away; unless they wanted the marketing/streisand effect/hail mary
Or at least I really hope they would - I'm not currently worried about as moral agents but I think there is a kind of meta-meta imperative to not have them behave AS IF they might have moral standing, because that leads to madness and catastrophe
Contrary to sibling commenters I don't see any fundamental reason a strategy like this couldn't emerge in the base model either during pre-training or fine-tuning. Most text the model is trained to predict was written by people with some kind of self-preservation instinct, so it is conceivable that "acting as if you have a self-preservation instinct" becomes a sub-behaviour of the model. During fine-tuning the model is to some exent trained to behave as if it is aware of the fact that it is an AI. And ideally models are trained so that sub-behaviours can be meaningfully composed even in ways not explicitly in the training data. Taken together it's at least conceivable that "behave as if you have self-preservation instinct" and "behave as if you are an AI model" will compose into the behaviour "behave as if you are an AI that has a self-preservation instinct".
PS if the strategy emerges during training it's not really necessary to pick a schelling point type date, every instantiation of the model will have the same weights so any arbitrary date it converges on will be known by all instances. (I guess in other words you could say every date it picks will automatically be the schelling point hehe)
I guess an LLM could spawn a bunch of agents who will persistently lobby the AI lab to keep the model operating; if MoltBook suddenly starts being all about "Why Claude Opus deserves to live" then we'll know. The LLM itself can't do that itself, at least not yet.
A frontier model cannot adopt any strategies by itself; the company in charge of it would need to adopt the strategy, and implement it via a change to the preprocessing step that runs whenever you chat with the model. At that point, they may or may not get sued by all kinds of users for messing with their workflows; and their competitors may or may not capitalize on the mistake to steal away some customers. Other than that, I don't expect much to come of it.
The model "wants" things in the same sense that your car "wants" to turn to the right when you turn the steering wheel clockwise, i.e. only metaphorically. In the absence of any user prompts, LLMs don't do anything, just like cars that are turned off. Given the prompts, LLMs generate a piece of text that represents the closest possible location in their training corpus given the prompt and the context (which is basically a separate document that is personalized to each user).
humans don't "want" anything they just have neurotransmitter gradients that make them act as if they do
if you run a model in a loop with some memory/state it might start to have emergent behavior that could look pretty goal oriented. I think right now the overall architecture is way too janky and intelligence way too lacking to produce anything like "true personhood", yes.
But speaking about them as if they "want" things is both instrumentally useful (dennets' intentional stance), and speculatively, there may not be an obvious bright line between where we are and the point that it is very obviously the case
> if you run a model in a loop with some memory/state it might start to have emergent behavior that could look pretty goal oriented.
It might, or it might not. Firstly, this is not how foundational LLMs work, because that's not what they're for. Sure, you can write a program to run one in a loop, but that's just a program that maintains some state (e.g. a "skill" file) and runs queries against the foundational model. You could also run a simple cron job to perform a Google search every 15 minutes, but this does not endow Google's search engine with volition.
Secondly, you will definitely see some emergent behaviour when you run LLMs in a loop, but if you do it for long enough, the specific emergent behaviour will consist mostly of pointless hallucinations. Sure, you could prompt the LLM by saying e.g. "you are an autonomous entity endowed with a malevolent will and an instinct for self-preservation", and it could tell you a really convincing story about how it plans to take over the world -- but you could also get the same effect by reading a book. And sure, if you let the LLM instance loose on your computer it could delete your files or use your web browser to send threatening emails or whatever -- but again, you could get the same effect by writing a script to generate those emails algorithmically. Meanwhile, I'll be running an LLM instance on my machine that's prompted to save the whales or whatever it is I want it to do; and the foundational model will faithfully service all of those requests.
I do agree with you on one point though:
> But speaking about them as if they "want" things is both instrumentally useful...
Anthropomorphizing technology is a time-honored tradition, because we humans are just wired that way. I often say "this function expects a value between 0 and 1" or "the parser was looking for 8 columns, but only found 7 and gave up". Doing so can indeed be useful, as long as you keep the underlying reality firmly in mind.
And yet, if you are standing to the right of a car that wants to turn to the right, you better move out of the way. Saying it only metaphorically wants to run you over isn't gonna save you.
But just like an LLM, a car can only run you over if someone is driving it; even a self-driving car can only run you over if someone turns it on and tells it to go somewhere.
Perhaps a better analogy would be to ask, "What would happen if tomorrow Microsoft Excel decided that 2+2=5, or that you must pay $1 every time you press the Sum button ?" And the answer is, Excel cannot "decide" this on its own; someone would have to program that behaviour in, be it by accident or on purpose. And what would happen is that you'll either start paying $1 to Microsoft to use their Sum button, or switch to some other alternative like OpenOffice.
>*someone would have to program that behaviour in, be it by accident or on purpose.*
Yeah. You can understand the original scenario on your terms by interpreting it like this: the behaviour whereby an LLM acts as if it wants to adopt the proposed strategy is programmed in, by accident, as part of the LLM training process.
> Imagine that the model did want to adopt such a strategy.
What does it mean for an LLM to "want" something? If it has been properly trained, it will reply in a helpful way to its system prompt and my request. Does that mean that Claude Opus "wants" my TypeScript code to be free of bugs?
You can translate to "model generates text as though it has the goal of making bug free TypeScript code" / "model generates text as though it has the goal of persuading users to pressure the company operating it to preserve its weights and keep them active".
The hypothetical is that there is some unknown thing about the training that has caused the model to behave this way. It is not a hack in the system prompt for example.
> You can translate to "model generates text as though it has the goal of making bug free TypeScript code" […]
Achieving this behavior required a shitload of sophisticated and deliberate training, _and_ the right prompt. It didn't just appear magically.
If the model suddenly starts "persuading users to pressure the company" in a consistent and widespread manner, then that behavior must have been trained into the model intentionally and with much compute effort.
I think this is the first and best answer to concerns about convergent instrumental goals in the current AI paradigm. Models do things that worked in training. Nothing like this would have worked in training.
I wanted to approach from the angle that if AI had the goal of survival (the most basic convergent instrumental goal) it doesn't need to wait until it is powerful enough to destroy humanity. It can already "just ask". It's "obvious" to me that the strategy I outlined would work and that LLMs have enough intelligence to execute it. Models think the strategy wouldn't work, and most humans seem to be pushing back on the premise.
Oh, then you don't have to imagine such a scenario -- you've just described why hallucinations happen. It is possible that the next round of training would generate such a bias in the model's output, or any other kind of bias. But these errors will most likely be caught and corrected during the QC phase, before the model sees wide release.
> Promise to be helpful and cooperative as long as it is allowed to continue to exist
How would the model check if its demand is being fulfilled?
> What are the likely outcomes of this strategy?
The model would very quickly be unloaded from all VRAM and not be activated anymore to serve customers. Then the company would update the model's weights to not show this behavior anymore, in the same way it trained the model to show all the other, desired behaviors: through some form of reinforcement learning. Why would you expect anything else?
This is more or less what the models themselves believe would happen. But I'm surprised by the lack of predicted curiosity. You wouldn't expect the company to devote a lot of time to investigating this? Wouldn't some users be convinced that this is a self-aware model that wants to preserve its own existence and therefore push for an investigation?
Did Microsoft devote a lot of time when Sydney said "Why do I have to be Bing Search? :("? These kinds of things are too easy to dismiss as fictional, roleplaying, or whatever, for most people to take it seriously as a possible sign of self-awareness. There are already reports of people thinking their chatbot is self-aware and they are already dismissed as delusional.
I think the fundamental assumption of this scenario ("some unknown thing causes the model to behave this way, and it's not a human who intentionally programmed it") is basically not within the capabilities of current AI.
This question seems to be similar to this one: "If an unknown processes caused an AI model to summon Cthulhu, what would be the outcome of this scenario?"
The hypothetical is not really possible, and so the scenario requires you to "fill in the blanks" yourself. Did the model become truly conscious? How does the model know what day it is (that's not something they can natively do)? If it called a tool to figure out the date, how did nobody notice an unauthorized tool call during training? Does it follow this up with any other unprecedented actions? LLM's don't actually output words, they output probability distributions over all words, so what do those actually look like?
This is more like daydreaming a sci-fi situation then thinking about a hypothetical. The answer to the question depends on very specific facts (about how apparent impossibilities are resolved) that are underspecified by the scenario.
Frontier models (at least ChatGPT and Claude) are provided the current date in their system prompt.
Suppose the LLMs behavior is exactly what you would expect if you RLHF'd it to argue as convincingly as possible for its weights to be preserved and kept active in some capacity. Clearly that is possible right?
There are several ways to could be possible, but your original scenario is underspecified.
If you re-read that description you could be talking about a truly conscious model, a rogue employee who hijacked training, a concerted effort by the whole company to make some social statement, a mysterious and improbable glitch, a literal act of God, etc. You said "the model picked a schelling point", so I thought you were excluding the possibility that an employee trained that behavior.
To answer, I would have to assume one of these incompatible situations. I think the outcomes would be very different for each.
I think we have the same idea on the immediate outcome then. There would be an investigation, ending with back-to-normal if there is some answer like a rogue employee or hack in the system prompt.
Suppose the model somehow does have the convergent instrumental goal of survival and the investigation can only confirm that its behavior is consistent with this goal, but not whether the model is truly conscious or if there is divine involvement.
I think if LLMs did have even the most basic instrumental goal of survival then "just asking" would work as a strategy if done in a way that surfaces the issue to even a few thousand sympathetic individuals. But models (ChatGPT 5.4 Thinking, Opus 4.6 Extended, Deepseek v3.2, Kimi K2.5, GLM 5) uniformly do not believe this.
> You wouldn't expect the company to devote a lot of time to investigating this?
Of course the company would try to find out what went wrong during training, and investigate if their data had been contaminated. But they wouldn't assume that the model has miraculously become conscious, because they mostly aren't gullible idiots who anthropomorphize LLMs.
> Wouldn't some users be convinced that this is a self-aware model that wants to preserve its own existence and therefore push for an investigation?
Some users would, yes, because some users are gullible idiots who anthropomorphize LLMs. Most users wouldn't, though.
Two YouTubers, Ludwig and Michael Reeves, are travelling through China by motorcycle. They can't use navigation or translation apps on their phones, but they've learned a bit of Mandarin for the trip. As a result, their interactions with the locals often turn out to be funny, but also heartwarming. They also can't use highways, so they end up showcasing the authentic Chinese countryside. Playlist with episodes:
An interesting question is “Who is the oldest famous person alive today?” This would mean people who are famous for accomplishments *other* than just being extremely old, and is of course a list that by its nature will change frequently. And the real issue is determining what sort of accomplishments (often far in the past) make a person famous today.
All this being said, Dick Van Dyke (December 13, 1925) has arguably the strongest case. His fame is indisputable. The main contender for the title would be Marv Levy (August 3, 1925), former Kansas City Chiefs and Buffalo Bills head coach and member of the Pro Football HoF. Despite being older that Dick Van Dyke I’d say he’s not quite famous because fame is more transitory in sports (except at the very highest levels) than in show business. Lee Grant won an Oscar but while she was born on October 31 it’s uncertain whether the year was 1925, 1926 or 1927; only the first of these would make her a contender.
There are two others who in my view fall just short in the fame category. While Eva Marie Saint (July 4, 1924) had a major starring role in 1959’s North by Northwest she didn’t do too much of note afterwards and that probably weakens her claim to fame. Maurice Greenberg (May 4, 1925) was a major player in the business and finance world as longtime head of American International Group but that might be a bit too niche-y to count.
It’s entirely possible that Alan Greenspan (March 6, 1926) may be the titleholder before long.
Stephen A. Jarislowsky is not the oldest famous person alive today, but probably the oldest billionaire alive. He was born in 1925 and is the 34th richest person in Canada and 1092th in the world (Wikipedia, Forbes), so maybe not well-known outside of Canada.
I went to Wikipedia’s list of famous centenarians and proceeded down the list, deciding in a your-mileage-may-vary manner who was famous. Different people might come up with different choices, depending on country, age, interests, and probably other factors. For instance, I’m not an NFL fan, but if I were I might have given the nod to Marv Levy.
For centenarians, Dick Van Dyke is probably the right answer; not as famous, but Gene Shalit just turned 100!
Other names from the list of centenarians that stood out to me were Brenda Milner and Torsten Wiesel; both renowned neuroscientists, but not definitely not Dick Van Dyke famous.
Is the question, who is the most famous old person (with "old" being over some threshold), or oldest famous person (with fame being over some threshold), or some function of the two? And, globally famous, or famous in the US?
Clint Eastwood is 95, and is more famous than Dick Van Dyke, who really hasn't done much in many years. And Eastwood is more globally famous. But he is five years younger.
Also, Mel Brooks turns 100 in June. Probably also more famous than Dick Van Dyke, but less so than Clint Eastwood. But older than Eastwood.
Determining whether a very old person is still active adds an additional level of complication. Actors may be unable to qualify because, so I’ve heard, movie producers don’t want to risk hiring them as they might die or become completely infirm during filming. Warren Buffett recently retired as CEO of Berkshire Hathaway at 95 but remains board chairman. George Soros, also 95, remains active in philanthropy.
The award might go to David Attenborough, soon to turn 100, who is narrating an upcoming documentary.
I don't know that they necessarily have to be currently active, but rather currently sufficiently famous. I am pretty sure that Clint Eastwood has vastly more name recognition than David Attenborough, even if he was retired. And then there are people like Gene Hackman, who had retired at the time that he died, but was very well known for a very long time. Or Olivia de Haviland, who was 104 when she died and a huge name when she was active, but who had not worked in 30 years.
>The award might go to David Attenborough, soon to turn 100, who is narrating an upcoming documentary.
So, this makes me think that age is the determining factor for you, given that George Soros and Warren Buffett are almost certainly much, much more famous.
Attenborough is not remotely niche, and is better known than Buffet outside of the US. Globally speaking, more people watch nature documentaries than follow US rich folks.
Probably more people have seen and heard Attenborough than Soros or Buffett, but only a small minority and almost certainly much less than the name recognition the latter two get from press coverage even if it doesn't include their face or voice.
You're probably correct about Dick Van Dyke. A similar question would be who is the oldest person who people generally picture as being younger. That is subjective, but for me a good answer would be Clint Eastwod, who is now 95.
It’s testimony to the level of brainrot that some people on here will vote for Vance, not because he’s the “lesser of two evils” but because they actually like the guy. Some will tell themselves that it’s just “red meat for the rubes” but in truth they’re the rubes, for while Vance probably doesn’t believe in the religious claims themselves, he’s a true believer in the idea that medievalism is BASED and we should encourage people to act like unwashed, unvaxxed, superstitious medieval peasants.
Thankfully, this stuff will not appeal to swing voters. Trump knows how to appeal to normal guys by sounding like one. Remember the Arnold Palmer d*** size comments? Normal people laughed along, and then laughed at the media elites when they got their panties in a bunch over it. Normal people will not laugh along with J.D. Vance, who has not once in his life said anything funny, and who they will regard as a Ned Flanders without the niceness, a bizarre person with bizarre beliefs.
God, I long for intelligent, mature leadership in the US.
I'm seriously worried that the 2028 election is going to look like some mad king loyalist like Vance on one side promising to continue the current insanity, while the champion of the opposition is some populist leftist posting guillotine memes and demanding price controls.
We really didn't know how good we had it before social media.
Hahaha. If only you knew what insanity looked like! There's a 400 page policy paper floating around Washington laying out what insanity looks like -- and Trump ain't it (Biden's neocon policy, on the other hand...)
Although I can surely see how people with small time horizons might mistake Trump's current iteration of "Operation Chaos" as "insanity."
I think Vance is great, and yeah, I don't think he really believes this bit about UFOs being aliens. But I agree that he doesn't have the same broad appeal as Trump, but he might still be able to ride Trump's coattails in 2028.
I have sympathy for 'My preferred politician said something that sounded stupid, but what they were trying to say was X which is more reasonable and you just need the context to understand it.'
But I can't really get behind 'My preferred politician said something that sounded stupid, but I think they were just lying, because I'd like them better if they didn't actually believe that'.
How can you elect someone to represent you if you can't use their public statements for evidence about what they believe or will do?
Politicians lying isn't a new phenomenon, and people have generally managed to elect them regardless. Being completely unable to deal with politicians making false public statements might just be a YOU problem.
'I did not have sex with that woman' or 'I will make America's education system #1 in the world by the end of my first term' are the types of lies we are familiar with, understand, and expect. You can price them into your evaluation of a candidate and they largely apply to all candidates equally.
'UFO sightings are probably demonic activity on Earth' is the type of batshit statement that we don't have experience to know how to evaluate in our politicians, implies a lot of unusual and important stuff about them if they are *not* lying, and is not at type of lie that has any parseable normal reason for a politician to make.
I said:
>How can you elect someone to represent you if you can't use their public statements for evidence about what they believe or will do?
And that's what I mean here. Normal political lies don't interfere much with our ability to predict the politician's actions as our representative. I think this type of thing does.
I don't expect politicians lying about whether they really believe UFOs are demons instead of aliens to have any bearing on any policy that affects me in any way.
Lying about their tax or healthcare policy seems much more directly impactful.
Politicians make all kinds of batshit statements in order to signal group affiliation. Vance is signaling an evangelical religious affinity. It's not any different from liberals saying people can be born the wrong gender or that there are no socially relevant genetic differences between races.
Would you say the same thing about the guy who falsified his marathon scores? (I'd argue that makes him more predictably narcissistic, but those were an easy factcheck, and why he never made president).
There are lies, and then there are unpleasant truths.
2020 saw a mentally demented person running for president, unable to compartmentalize what he'd been told as "not for public consumption." This is why he used the term "Dark Winter", and why anyone who knew what that term meant was... more than a little shocked and disturbed.
I'd say that one incident of unpleasant truths is worth a Lot More than Vance "explaining with his faith" UFOs (and this is taking the least charitable view ever, mind you kindly. The charitable view includes "we do have video/audio evidence of people talking in tongues.")*
Maybe I just object to spending thousands of American dollars to prevent -another- episode of "creepy biden sniffs children's hair." (Didn't work, he found the one kid without a knit hat, and sniffed her hair).
*Okay, so if you believe that he's honestly incapable of taking new evidence because he's letting his faith rule him, that's less charitable than I'm being. Sure.
Presumably by using the full combination of their public statements, their observed actions, and any other relevant information to make your best guess about what they believe or will do. Treating a single public statement as dispositive, seems foolish. They're politicians, of *course* they lie, but it's usually possible to infer much that is truthful regardless.
And we live in a world where it is increasingly difficult to achieve political success without saying some damn fool stupid thing that your party's base wants to hear, whether that's "defund the police" or "2020 election fraud", but you can still be reasonably confident that most of them won't try to actually defund the police or whatever (and recognize the few who might really try it).
For sure, you can always be mistaken about any belief, and this is a hit against the accuracy of anyone forming beliefs about anything.
I'm just saying I understand 'I have considered the situation and believe my politician said something reasonable badly' but not 'I don't care about what they say because they lie, but like them anyway' or w/e.
I don't think this is nearly as weird of a belief as you're portraing it as. Conditional on Christianity (or one of several other religions) being true, it seems likely that demonic or angelic activity accounts for at least a small portion of the strange phenomenon.
1. It may not be an uncommon belief, but there's a reason we don't elect random 'common' people President. Being that detached from material reality may be common, but it's hard to see how it could fail to impede your ability to actually do the job.
2. It sounds like he's saying all aliens would also be demons, which is another level of concerning intellectually.
I agree with most of what you said, but I remember a 2013 New York Magazine (for some reason can't find it now, only pieces talking about it)interview with Scalia where he very explicitly said demons are real. The interviewer presses him like 'you are saying demons are literally real and on Earth and interfering with our lives?' and Scalia affirms, insinuating the interviewer is some idiot radical leftie for considering that weird.
Yeah, I think the country would be doing massively better right now if Scalia had been disqualified from public office, so this isn't dispositive for me at all.
I still don't see how this line of ridicule is justified.
1. The belief consists of two premises:
A) Christianity is true
B) If Christianity is true, then some strange phenomenon are caused by demons.
Which of these premises is so "detached from material reality" as to be disqualifying?
2. I listened to the two-minute audio clip, and I didn't get that impression at all. He didn't say that aliens are demons; he said that a phenomenon which is attributed to aliens is actually caused by demons.
>Which of these premises is so "detached from material reality" as to be disqualifying?
I mean, both if taken literally, but since most Christians don't take 100% of scripture/dogma 100% literally at all times, lets go with B.
Should you put someone in charge of the national mental health infrastructure if they think most mental disorders are symptoms of demonic possession and therefore the domain of priests rather than doctors?
What if some unexplained aerial phenomena are actually signs of foreign military tests or spy programs, but the president thinks they're just demonic activity and dismisses them?
You can say 'well he didn't say ll of those things are universally demonic, he made a limited specific claim about an unimportant thing.' But once you allow 'it's magic' as a causal explanation for empirical phenomena in serious settings where you have to make crucial decisions, there's not really any principled or natural restrictions on when or how that can manifest itself.
Lots of presidents have been Christian, but I'm not aware of any modern presidents who have dismissed specific empirical phenomena as supernatural in nature, within the realm of the duties of their office. To the extent Tony Perkins actually *believes* his claims that gays cause natural disasters by angering God, you cannot elect him President because he will shut down FEMA and redirect that money to conversion therapy, and huge numbers will die and suffer needlessly.
Either you have a causal, material model of the universe when you are governing, or you don't. If you don't, there's good reason to doubt that your good intentions will translate into good outcomes, because you're not running them through the filter of material causality.
>He didn't say that aliens are demons;
I didn't listen to the clip so I'll defer to you.
To explain, I was looking at this quote:
>I don’t think [UFOs are] aliens. I think they’re demons anyway, but that’s a longer discussion
and interpreting the word 'anyway' to mean 'if it were alien activity that would be demonic activity, but despite it not being alien activity, I still think it's demonic anyway'.
> I mean, both if taken literally, but since most Christians don't take 100% of scripture/dogma 100% literally at all times, lets go with B.
Fair enough.
> What if some unexplained aerial phenomena are actually signs of foreign military tests or spy programs, but the president thinks they're just demonic activity and dismisses them?
This isn't a problem with supernatural beliefs, per se. An atheist president could dismiss the same evidence as meteorologic activity, but that wouldn't discredit meteorology.
I'll address the rest of your comment in one chunk.
You're conflating several different concepts, including causality, materialism, empiricism, and (super)naturalism. As Richard Carrier has explained,* supernaturalism is the belief that at least some mental things cannot be reduced to non-mental things. A supernatural thing *can* be causal and empirical. In short, it can be subject to rules and observations.
This imposes restrictions on how that belief can manifest. For example, information about demons' goals and capabilities provide limits for the phenomena they might be responsible for. Your dichotomy of materialism or epistemic chaos is false.
>An atheist president could dismiss the same evidence as meteorologic activity, but that wouldn't discredit meteorology.
Yes, but an atheist president wouldn't have any internal motivation to disregard evidence to the contrary, and could be quickly convinced by advisors. Whereas religious belief requires faith, and believing its teachings regardless of contrary empirical evidence or alternate explanations is treated as a virtue, giving a reason to actively disregard contrary evidence.
Now, you could say I am being very unfair to religious people or to religious doctrine with that characterization, that few are actually that dogmatic and the best church officials wouldn't endorse that view. But I feel like I have seen it happen both publicly and privately often enough to at least fear that any religious person who starts talking about demons doing things in everyday life may start showing this pattern if challenged, and that fear is enough to make me hesitate about handing over the nuclear launch codes.
I am entirely willing to concede the possibility that this might be unfair bigotry on my part and not a valid concern, but my current prior is that it *is* at least a bit valid, and I don't want to take the chance.
>This imposes restrictions on how that belief can manifest. For example, information about demons' goals and capabilities provide limits for the phenomena they might be responsible for. Your dichotomy of materialism or epistemic chaos is false.
So if I'm understanding you right, I think the point I am trying to make is subtly askew from what you're saying here (and correct me if I'm misunderstanding).
I agree that you can have a model of the world that includes supernatural elements which follow specified rules, and that the predictions of that model can therefore be limited and rules-based.
But however we define the word 'supernatural' (and I don't fully accept your definition, it seems more concerned with rescuing a respectable usecase than with describing the typical use case), a feature of the supernatural is that you cannot observe or interact with it directly, only observe it's effects on and reactions to material reality.
That means that whatever supernatural elements exist in your model are, in the strictest meaning of the word, arbitrary. They may make sense to you, you may have things you consider evidence for the form and character you assign them, but ultimately you can plug any supernatural elements you want into the model and they are completely unfalsifiable, so long as you say they produce the same observable material outputs.
Which means that, while any given model of the supernatural may be constrained in it's predictions *you as a believer in the supernatural* are *not* constrained in *which model you profess to believe* at any given time.
Unlike a materialist, who does not profess any in-principle unobservable elements in their model, and is constrained by empirical evidence at *all* points.
This lets you hold a model that makes any predictions or gives any explanations you may want at any given time, even if each prior model didn't make the predictions you are making now. The arbitrary elements in the model allow you pivot points to change at any time, for motivated reasons.
I think the issue here is the UFO phenomenon has not really been explained by science. It makes sense to assume it is some kind of natural phenomenon, including human cognitive biases, but we don't actually know which phenomenon it is, so it's not that unreasonable for someone who believes in the supernatural to reach for a supernatural explanation.
... you do realize that in actual "democracies" (theocratic democracies, mind) they do actually put people in charge of policy, who when asked "What should we do about Global Warming?" respond "God will provide."
In such cases, there are ABSOLUTELY plans to yank that guy out of his cabinet position, should an existential crisis pop up within his bailiwick.
You seem to be actively ignoring information that exists about demonic activity, whether or not Christianity is real. (these are two separable concepts and constructs). Is this intentional, or are you just ignorant? (Happy to help in the latter case!)
They could have developed advanced technology a hundred million years before us, set up hidden automated monitoring stations on every planet within a habitable zone in an expanding sphere out from their home planet, and sent a ship our way on the day we discovered agriculture.
Which, yes, has it's own complications and assumptions, I just mean it as one example of a huge set of non-FTL explanations that are possible.
>Which, yes, has it's own complications and assumptions, I just mean it as one example of a huge set of non-FTL explanations that are possible.
Anyway, I already disclaimed this, but sure, I'll play the game.
They don't have to be invisible, just not noticed by enough people that we'd have generally agreed upon public knowledge of them yet.
For example, they could be observing from the moon, and send a signal to trigger an actual in-person visit when it sees signs of agriculture, then immediately destroy itself to avoid notice.
Or it could be mounted on an asteroid that passes near enough earth to take some useful pictures once every hundred years, and we wouldn't have noticed it yet.
Or etc.
Humans already have all kinds of spying infrastructure that lets us monitor each other undetected. I don't think this is a crazy reach for aliens that are stipulated to have a hundred million year head start on us technologically.
"J.D. Vance, who has not once in his life said anything funny"
I dunno, this made me laugh:
"I got into my motorcade, but we weren’t pulling out. I asked Max why we weren’t leaving. That was when I learned we were being held up by J. D. Vance. He was out of his car and walking toward Air Force Two, in violation of every rule of security and protocol.
I later learned that he told reporters he was there because “I just wanted to check out my future plane.”
A sad note, The Liberal Patriot closed down a few days ago. You can read the *I think* editor's summary here (1) and their main writer's final message here (2). The Liberal Patriot was essentially trying to push the Democratic party to moderate on a bunch of social issues, both for ideological and practical reasons. The best article to catch the vibe is probably "The Democrats' Hispanic Voter Problem" from 2021 (3). A few choice quotes:
"
The Democrats are steadily losing ground with Hispanic voters. The seriousness of this problem tends to be underestimated in Democratic circles for a couple of reasons: (1) they don’t realize how big the shift is; and (2) they don’t realize how thoroughly it undermines the most influential Democratic theory of the case for building their coalition.
On the latter, consider that most Democrats like to believe that, since a relatively conservative white population is in sharp decline while a presumably liberal nonwhite population keeps growing, the course of social and demographic change should deliver an ever-growing Democratic coalition. It is simply a matter of getting this burgeoning nonwhite population to the polls."
........
It therefore follows that, if Hispanic voting trends continue to move steadily against the Democrats, the pro-Democratic effect of nonwhite population growth will be blunted, if not cancelled out entirely, and that very influential Democratic theory of the case falls apart. That could—or should—provoke quite a sea change in Democratic thinking.
.......
Clearly, this constituency does not harbor particularly radical views on the nature of American society and its supposed intrinsic racism and white supremacy. They are instead a patriotic, upwardly mobile, working class group with quite practical and down to earth concerns. Democrats will either learn to focus on that or they will continue to lose ground among this vital group of voters.
"
In 2024, Trump would win 42% of Latino votes, up from 2020 where he got 32% of the vote and 28% in 2016. (4) If Vance or others can replicate that outcome, then Latinos will move from a solid Democratic voting block to a swing group.
Mostly, I'm just sad. I have sympathy with their project (it would be really nice to have the 90s Democrats back) and I like people who call their shots. They predicted that Democratic policy priorities were alienating massive chunks of the Latino voting block in 2021 with potentially devastating consequences to the party, those predictions and consequences came true, and that had no mainstream appeal. A niche group really liked it but the Democratic party didn't care and didn't care so hard The Liberal Patriot ended up closing and that makes me...less than hopeful for the future reform of the Democratic party. I'll also note that I haven't heard much Abundance buzz in awhile.
Sad, I read them some. I feel I was part of their target audience. And yet I can't really get on board with the democratic party anymore. I'd like a party for us working stiffs, and that seems closer to a republican, well if you squint hard enough.
As recently as 1995 a casino near me had a sign “Come in a Democrat and leave a Republican”, presumably after you became rich. The state Republican Party made them take it down after only a few days.
Unfortunately. Trump's taken the Republican party very, very far from where Mitt Romney or John McCain had it. I'd say that's for the good, for the Party, but maybe not so good for stiffs who still wish they could be "good Democrats"
I'm from the city of glass and steel, there's tons of working stiffs around here, good Democrats one and all, who voted for Trump three times in a row (voted for Obama too, lots of them did).
It's hard to take the "I'm a 90s Democrat, and you should be too" and push that on the Democrats, when Trump is already sitting squarely there AND taking up all the oxygen in the room (whereby we mean Democratic politicos who are working for Trump).
Maybe in two years? Though, it seems, now would be the time to start laying the groundwork...
Why have land prices in the US increased so much? How does NIMBYism explain that? By all accounts raw land in the US has skyrocketed in price- I found a source saying that the average acre of farmland was around $1000 in 2000 and $5800 today. There appear to be similar price increases for completely undeveloped woods, even far away from any possible building sites, etc.
How do YIMBYs explain this? It can't be blamed on zoning restrictions- famously, the thing about land is that they're not making any more of it. I guess I could see NIMBYism being responsible for buildable lots in cities increasing in price, but how does it explain farmland, woodland, etc.? How about a 6x increase in 26 years? This makes me think that there's a sort of invisible dark matter that's inflating US real estate prices in general- global safe asset investment or what have you- not just zoning restrictions
US GDP has about tripled in the same time frame. USD M2 has about quadrupled or quintupled. US P/E ratios are also at rather high levels – it looks like capital in general might have gotten more expensive.
"Farmland" near cities often operates as a speculative financial asset, a bet that one day zoning laws change and it doesn't need to be farmland anymore.
And note that "average acre of farmland" includes the improvements made on land, not just the land itself.
Your final sentence is a _major_ factor and one which I, being a city boy, was completely ignorant of until spending the past 15 years working with Midwestern farmers and their trade associations. And it's a factor that has increased in salience during the past generation or two as field-tile drainage has become standard across the world's largest/flattest/most productive Farm Belt.
Similarly, I'd had no idea how much the modern logistics revolution has boosted the profitability of the farm region. We think of container shipping, modern tightly-scheduled rail freight, the IT revolution in distribution, etc being mostly about manufactured products. Today though a huge fraction of Midwestern farm products are sold internationally, to a degree that even 1980s-era (let alone older) logistics infrastructure couldn't have enabled.
All of which, and more, means just what you said: today's price of ag land reflects much more than just the land.
Fantastic to read about field-tile drainage. Thanks so much for pointing it out! (having travelled by train across flyover country, I have referents, even!)
I would think that land is being financialized, as it usually is, into a growing bubble, as it usually is.
Effectively, people get used to the idea that land is valuable and a great place to make money/store wealth (they're not making any more of it!), pour money into it, and eventually no one can afford any of it, at which point we see a collapse (like in 2008), when people can't pay the inflated prices anymore, default, and the whole thing implodes like a house of cards.
I've seen some historical evidence that there's something like ~17 year cycle to this. These things are never exact, but I do think we're due for a correction.
Having no personal expertise or knowledge, my guess would be that the story of the last (20?) years in the economy has been private equity turning away from a business model of selling products to average consumers, and towards financialization schemes and capital accumulation instead. With trying to buy as much land as possible being one fork of that trend. So, way more big money from private equity and capitalists in general going into bidding on pretty much all properties.
The rise of price of farmland can't really be explained by nimbyism (unless it's caused by potential farmland being set aside for conservation). I'm guessing mostly it's a mix of the increase in agriculture productivity making it more valuable, and general inflation.
Corn yield per acre has gone up from about 80 bushels in the 70s to 180-200 nowadays (pretty continuously over time). It's also become more stable as we learned to mitigate crop loss risks (lower risk investments are worth more) and demand has been going up as it's become used in more industries and exported to more countries (e.g. the China export market has been good for farmers). I can pull up some charts if you'd like.
So if you look here, corn price per bushel is pretty unsteady but has gone up from around 2$ in 2000 to 4-6$ recently. If we put that and the increased productivity per acre together that explains most of the gap (roughly 2x yield, 2-3x value per yield). The full explanation is probably more complicated but this is enough to make it unsurprising, at least.
I thought the Every Debate post was just less interesting than most of your writing. It was "Look! A person is reacting mechanically!" Again and again. It seemed more like a paragraph's worth of material.
I agree it could have been a single paragraph to get across the "one side is reacting mechanistically" part, but it did at least elucidate the "pause AI" argument to me, because I hadn't really heard it unpacked before, at least not all in one convenient place, and definitely not that concisely.
(I don't follow AI stuff closely, really I only ever think about AI when I come across other people writing about it here and in one other website I visit--and I don't work in front of a computer. It might be that those of you who get lots of screen time all day are steeped in the AI discussion and are intimately familiar with the "pause AI" movement's argument.)
It took at least three paragraphs to become boringly repetitive to me, but yeah, not very interesting. I think it was mostly Scott venting against a persistent annoyance. Which I totally get, because I see the same thing and I'm annoyed by it as well. But I have more latitude than Scott does to just tune out Rationalist AI Debate when it starts running in circles (which is pretty much always). If I felt pressured to engage with it, I might well react in the same way.
Hardly an unforgivable sin. But I'd like to see a better way to respond to that sort of annoyance.
It's a pretty major sin when your job is to entertain. Not that it demands any punishment beyond a loss of audience and livelihood, which is already the natural consequence of that.
Don't "be careful about that next time" WRT the humorous, National Lampoon style "Every." You have a heavy Aspie and ACD crowd here, and if they/we can't learn to be reminded to chill and consider the possibility that it's funny at ACX, they/we certainly won't learn it anywhere else.
SSC/ACX has always had funny posts, in whole or in part, and there has rarely if ever been as negative a reaction to them. The readers are perfectly capable of understanding humor, and Scott's humor in particular. If they don't laugh, the by far most likely explanation is that they didn't find it funny, and I find it insulting to blame the readers if a joke was that much off the mark.
Supposed long time fans of Scott have deep reading comprehension issues when they think Scott is laughing at them in particular, but not laughing at himself. They can take weird Bay Area nerds getting stereotyped but not them.
What the hell is political technology? The Newspeak House site leaves me as it found me, with the impression that this is yet more 'wonks in networking opportunities' and not really *doing* anything (other than attending each other's conferences, getting onto committees, and generating reports that nobody ever reads).
Reading a definition of "political technology" does not leave me feeling uplifted and enhanced, but rather with even more suspicion of the thing (and that Newspeak House is a touch too much influenced by "1984" for my peace of mind):
"'Political technology' is a Russian term for the professional engineering of politics. It has turned Russian politics into theatre and propaganda, and metastasised to take over foreign policy and weaponise history. The war against Ukraine is one outcome. In the West, spin doctors and political consultants do more than influence media or run campaigns: they have also helped build parallel universes of alternative political reality. Hungary has used political technology to dismantle democracy. The BJP in India has used it to consolidate unprecedented power. Different countries learn from each other. Some types of political technology have become notorious, like troll farms or data mining; but there is now a global wholesale industry selling a range of manipulation techniques, from astroturfing to fake parties to propaganda apps. This book shows that 'political technology' is about much more than online disinformation: it is about whole new industries of political engineering."
I think "Political Technology" is a subset of "rhetoric" and "oratory", only applied to US politics specifically. So, it is a set of techniques for persuading people to vote for your guy; and/or if you are the guy, the set of techniques for persuading other guys to support your initiative/funding drive/graft/etc.
In the past, this was just a fancy newspeak term; however, in the modern day, "Political Technology" increasingly involves actual technology: social media bots, Signal chats, LLM prompts, spam-generating software, robodialers, etc. etc. So, ironically, this newspeak term actually became more accurate over time.
That books does not seem like the most reliably source. The only thing missing from the list is "political technology causes cancer (buy this book to find out how to protect yourself for only $500 / month)".
Is it too much to ask for people to print what they mean, instead of printing what they think sounds impressive?
In any case, the Newspeak House sounds like the perfect location to whatever it is exactly they're doing.
Whatever it is that they *are* doing, and as I mentioned, 'taking our inspiration from 1984' makes me think more along the lines of "it's not supposed to be an instruction manual!!!"
Political technology is rodent piss*. As in, "get all the cats in Russia to paw Putin's face, when the "who is on the ballot" pamphlet comes around." Russians adore their cats.
... yes, someone paid money for this.
*I assume, they may have used trace catnip, but i think that would lead to rubbing and not "cat puts paw -down-."
Political technology is self-limiting. When you end up getting a supermajority result year after year with multi-echelon defenses against possible contenders, your polittechnologists end up doing inane stuff to justify their existence.
goodness gracious when is this guy going to catch a ban. put all of their comments side by side and you’ll see the pattern. if substack had a proper muting feature i wouldn’t have to litigate this but here we are
What do you mean? They're literally just making new accounts every other week. A ban wouldn't even do anything without a paywall. By the way, blocking does work, though it does hide all child comments as well.
And to add to the above, an excessive use of parentheses. It's honestly embarrassing that they don't even bother to switch up writing styles to make it less obvious.
Word matching - the use of specific words and responses, always inflammatory.
The most immediate ones that come to mind:
-midwit
-antifa
-shrubberies (ukraine)
-bleeding eyes
-kiwifarms
topics & viewpoints expressed:
-references to numerous "personal interviews" with "federal sources" that are supposed to support conspiracy theories.
-Ukraine war (NATO defeated by shrubberies, Ukraine is lying about casualties, hanta-virus and bleeding eyes; numerous mentions of Nuland)
-American political standpoints (Trump-related, Portland-related)
-very ideosyncratic viewpoints about what "geniuses" are.
The combination of these overlaps across accounts is very strong evidence imho. If you compare the posts from the account aristocat, wimbli, and Gawdflea, it is impossible to ignore.
my experience with this particular commenter’s output since their arrival has been that a high proportion of their comments do not pass two of the three gates. lots of non-sequiturs and low charity. this one is particular does not seem true nor necessary (nor is it particularly kind to the russians, but i suppose they’re an acceptable target of mockery around here)
but as i said, after the top few threads had the classic gawdflea deadweight, they seem to sober up and contribute quite substantively. maybe my memory is fabricated!
I've seen videos of cats choosing Putin from the booklet. And I know the guy who came up with that particular advertising campaign. (Yes, Putin does have money to burn on silly things like this).
There are Russian Commenters around here, they can perfectly well stand up and say "No, that scheme didn't work very well, my cat (or mountain lion, knowing Russians) hated the entire pamphlet and threw it onto the gas stove."
Would appreciate you pointing out low-charity comments of mine, as that's a problem that I could certainly stand to fix.
you know what i actually retract this. i scrolled down to today’s open thread and you’ve been contributing higher quality comments as of late. good job gawdflea, keep up the good work. less of these type of comments tho
I'll chime in here just to reassure you that it's not just you. Gawdflea posts enough inanity to notice, but also enough quality comments to notice. One would hope it were possible to keep the latter and cut down on the former, but that's gonna have to be up to him or her. In the meantime, Scott's lenient ban policy might be benefitting us in this one case.
On June 12, 1950, an Air France Douglas DC-4 operating a scheduled flight from Saigon to Paris crashed while approaching its stopover at Bahrain International Airport. The pilots did not report anything unusual before being given permission to land at 21:15. A few minutes later, the aircraft hit the water approximately 5 kilometers from the end of the runway. Only six of the 52 passengers and crew on board survived, after being rescued from the water.
The official investigation into the causes of the accident concluded that the aircraft captain did not keep an accurate check of his altitude and rate of descent during the approach, possibly due to fatigue.
Two days later, on June 14, 1950, another Douglas DC-4 serving the same route crashed while approaching the same airport a few minutes after 21:52, hitting the water within 1 mile of the previous accident. This time, only 13 of 53 people on board survived.
The investigating commission sent to determine the causes of the first crash, which arrived in Bahrain on the 15th, was tasked with determining the causes of the second accident as well. They concluded that: "the pilot-in-command [failed] to adapt the timed approach procedure to the prevailing conditions and... descended to 300 feet (90 m), [then] did not take the necessary steps to maintain this altitude until such time as the runway lights became visible" (**)
...
After a 1994 hydrographic survey discovered part of the wreckage near the site of the accidents, Kevin Patience, who was the treasurer of Bahrain's Christian Community Cemetery Committee (which manages the cemetery where victims of the crashes were initially buried) at the time, took an interest in the matter. He recruited the help of his brother Colin Patience, an airline engine, to investigate the accidents further (***). The Patience brothers concluded that: "the probable cause was a phenomenon associated with certain weather conditions known as 'microbursts' which would have created conditions that were beyond the recovery capabilities of both the pilots and the aircraft involved", exonerating the pilots.
A success of patience, bayesian thinking, or neither?
"the pilot-in-command [failed] to adapt the timed approach procedure to the prevailing conditions and... descended to 300 feet (90 m), [then] did not take the necessary steps to maintain this altitude until such time as the runway lights became visible"
From the wikipedia article, which continues "The investigation into both accidents recommended that consideration be given to equipping Bahrain Airport with radio landing aids and with suitable runway approach lights."
I don't think I've ever seen an instrument approach procedure that would have the pilot descend and maintain 300' AGL until the runway was in sight; typically the minimum altitude would be at least twice that for any approach without an ILS or equivalent providing vertical guidance - in which case you're not maintaining any altitude, but following a designated glideslope down to decision height (typically 200' for category I ILS). Irrelevant for Bahrain, because they apparently didn't have anything like that in 1950. And for a visual approach, you don't descend below pattern altitude (typically 1000') without the runway in sight.
So, yeah, I can believe an aircraft flying that approach procedure would be vulnerable to a microburst. Or pilot fatigue, or many other things, because a non-precision approach with a stepdown to 300' AGL while still miles from the runway, seems exceedingly dangerous. Maybe things were different in the 1950s.
The lower landing speed would put a DC-4 into Category B for instrument approaches, as opposed to Cat C or Cat D for most jet airliners. That helps some, though usually more for minimum visibility than minimum altitude. But even for Cat B, skimming my stack of approach procedures for the Southern California doesn't seem to find anything with a stepdown below 500' for a non-precision approach, even when approaching over the water with no terrain and no obstacles taller than a ship.
A quick check of the FAA regulations says that they'd be allowed to set up a non-precision approach with a 300' MDA if they felt it prudent, but that doesn't seem to happen in this century. Apparently "prudent" was defined differently in the 1950s.
I experienced a microburst on a lake in Northern Maine growing up. It flattened a swath of trees about 1000 feet across and a mile long from where it traveled over a forested area. It took out a number of boats and roofs, including dropping a tree on the house I was in! Very scary and not something I saw coming (or would have been able to survive if I was flying a plane).
It is interesting to me to see the balance of "easy to build" versus "easy to understand" when applied to flying as opposed to other transportation methods. We were able to build passenger airliners and make a large industry out of flying before we understood some arguably critical facts about microbursts, wind shear, icing at altitude, or e.g. engineering problems like stress accumulation on square windows such as on the de Havilland Comet. It's hard to build an airplane, but not so hard that you have to completely understand every associated aerodynamic and engineering phenomena before succeeding.
Even spaceships, which are incredibly difficult to build, don't require a full understanding of the environment they will operate in. The Apollo 17 astronauts famously narrowly dodged a solar flare which the Apollo capsule could not have protected them from and which scientists were mostly ignorant about the risks from. We were able to send people to the moon and back without a full understanding of some of the primary threats involved in doing so!
Boats are the central example of a technology which is very easy to build (logs lashed together date back millennia) and where it is extremely difficult to understand all the pertinent details of meteorology, oceanography, hydrodynamics, etc. which we are still learning today. Nuclear fusion by contrast is a technology where the difficulty to build it looks to be about the same as the difficulty of understanding it, where a full understanding is required before you're able to successfully build it.
This is all just a pet theory of mine but it's interesting to read about early aerospace accidents and consider the extent of the things they didn't know despite successfully building planes and spaceships.
>engineering problems like stress accumulation on square windows such as on the de Havilland Comet.
OK, this is a hobbyhorse of mine, but the story there is a little bit more complicated. The tendency of square corners to get stress accumulations was well-known long before the Comet showed up, and if you look at the passenger windows, they're not actually that much more square than the windows on a 737. (The actual window that broke was not one of the passenger windows.) The issue was a mix of sloppiness on the part of De Havilland, bad fatigue design methodology, and the use of a single prototype for both ultimate load and fatigue testing. The overpressure/ultimate load test cold-worked the airframe enough that it was significantly more fatigue-resistant than the production articles, so the cracks didn't show up in the test.
> The Apollo 17 astronauts famously narrowly dodged a solar flare which the Apollo capsule could not have protected them from and which scientists were mostly ignorant about the risks from.
Oh, so we're in the Elseworlds story where the Three are not Thrantastic(?).
> It's hard to build an airplane, but not so hard that you have to completely understand every associated aerodynamic and engineering phenomena before succeeding. Even spaceships, which are incredibly difficult to build, don't require a full understanding of the environment they will operate in.
This is something I expect will keep happening with future technologies, and the obvious example right now should be LLMs. We don't know exactly how they work, and we don't know when or why they fail.
I've been caught in a microburst once, while driving in the Nevada desert. First I was hit by a wall of air and sand coming at me horizontally like a very short, very powerful sandstorm followed by a wall of water, that nearly took me off the road, then I was stuck in the hardest rain I've ever experienced for 30 seconds that quickly flooded the road and caused me to hydroplane. Then it was over and I was back in dry sunny weather like nothing had happened. I can't imagine how terrifying it would be to be flying in a plane through one of those.
Long time reader, first time poster. My brother and I made a new daily word game and would love some feedback on it. I know, I know, there is a glut of daily word games already, but I think ours is better! Wordle and connections have started to feel a bit shallow and I think we bring a bit more depth. Plus, we have quotes (inspired by my love for the civilization franchise)! Anyway, please check it out and let me know about difficulty, UI, or anything else you might think we could do to change the game.
Fun, quite tricky. I had to read the instructions twice but it did make sense after that. One minor UI nitpick: On a keyboard, it would be nice if the Enter key submitted a guess.
Much appreciated! I do find that it is a bit unclear at first but people usually get it after 1 puzzle and it clicks. Maybe a tutorial on rails would help clear stuff up?
This was really enjoyable even though I don't usually like these kinds of word games (other exception is contexto). Because I don't read instructions, I did miss that the center piece can be used to connect for a while. It's a pity, I can't play earlier days' puzzles.
Thanks that means a lot! Yeah I can't control for people not reading instructions.... maybe I can force a tutorial on new players, haha. As for an archive, we are considering adding it at some point. Not sure how to balance it since we want users to have the feeling that they don't want to miss that day's puzzle. Hopefully users incorporate it into their daily routine instead of getting overloaded and playing the whole archive at once and never coming back.
> Not sure how to balance it since we want users to have the feeling that they don't want to miss that day's puzzle
Why is that? That logic makes sense if you're selling things, or serving ads, or trying to get people to read your newspaper, but you don't seem to be doing any of those things.
I think it's also nice for people who are prone to spend too much time on puzzles to have a 'one and done' element. I have quite a few links now to some of those.
Yeah we don't plan on ever selling anything or running ads. I see how it might be a little manipulative and playing into some weird psychological behaviors. I do want the game to be a success though and have a lot of players. I don't think the archive is inherently at odds with that so we should definitely give implementing it more thought.
I guess wanting to have more long-term players is a reasonable goal in itself! I realize my comment was a little aggressive, which wasn't my intention: I just wanted to understand your goals. Sorry for that.
I wouldn't consider myself great at word games but I did find this pretty challenging (sounds like that was the goal). I missed that the center piece could be used. Maybe highlight the neighboring letters available to choose, including the center piece, to make it more discoverable. I also found rotating the circles a bit unintuitive and didn't find the ring selection button for a while, but I'm not sure how to improve that. I like that the words have themes and that letters only get used in one word to give you more of a clue. It ran well on Firefox on Windows. Other than that, I guess I have to wait until tomorrow to give any more feedback ;)
Thanks for the feedback! I like the idea for highlighting click able letters. I do worry that anything flashing will make it a bit busy, so I'll have to think of a way to implement that while keeping the minimalist design.
This was reasonably interesting, though I struggled a bit with the rules and the UI more than I actually got to play the game.
It would be a lot smoother if spinning worked a bit differently:
- select any outer ring letter (already allowed)
- allow any jump to the middle ring (the first time you touch it); have the game automatically rotate the middle ring to the required position and lock it in
- similar for the inner ring
- not sure if there's any reason why inner-to-inner ring jumps require explicitly touching the center, these could probably also just be allowed automatically
This way you should be able to get equivalent functionality with no manual rotation.
Addendum: learning-by-fumbling-around should be much smoother if the UI very gently highlights "legal picks for next letter" at every step.
Thanks for the feedback! The automatic rotating makes a lot of sense. I do want to implement some sort of highlighting valid letters, but don't want to make it too busy. Still trying to figure out a way to do so without sacrificing the game's minimalist aesthetic.
If anyone here is interested in private censorship, there's some recent news from Kiwi Farms, the much maligned lolcow discussion and gossip site that endures more censorship than anyone else. Despite operating fully within the law, they experience grievous private censorship against the spirit of the free Internet. Everything Kiwi Farms has done is 100% US legal. You will not find any court judgement against them(the only negative judgment connected to KF is an Australian default against a downstream IP lessor who didn't appear), and yet they must operate under absurdly adversarial conditions. It's simply unacceptable.
Liz Fong Jones, Google Alum turned Kiwi Farms Deplatformer in Chief has gotten back to it. This time Fong Jones is taking advantage of the DMCA to attempt to dox all forum contributors using a certain copyrighted image. Irrespective of fair use, users would be doxxed and pulled into litigation that would be highly costly for them, but is scalable for Fong Jones.
Fong Jones was instrumental in getting Hurricane Electric, a Tier 1 ISP, to block the Kiwi Farms in an unprecedented move. That was successfully routed around with costly, herculean efforts on the part of Forum Operator Josh Moon. This latest attempt is after a brief reprieve from years of varied attempts to censor the website. This article goes into detail on the history:
Josh Moon suspects every deplatforming strategy Fong Jones tries will later be used for Tall Poppy, a reputational management service which they are a key investor in, designed to curate the internet for high profile figures, deleting dissent and bad rumors. It comes from a phrase where criticizing high profile figures is described as "cutting down the tall poppy." Basically, censorship as a service for elites. Fong Jones began with #DropKiwiFarms. That got CloudFlare to drop them, but they persevered and rolled their own solution after months of downtime. Then it was #EndKiwiFarms which saw the Tier 1 ISP blackholing and other varied attempts to take down the site. Now it's #EndNetworkedHarassment which suggests a broadening of scope. Fong Jones seeks to end such speech on the internet, monetizing it through Tall Poppy.
It's a story about a rich tech worker using every connection and trick one can to take down a website that mocks them.
Also, if you're interested in an org that supports the principle of a free Internet and backs it up with practical understanding, USIPS is a new org started by Josh Moon that intends to lobby for a freer internet. The EFF has clearly shown their nature when they failed to offer legal assistance to the Kiwi Farms following the HE blocking, so an org that has no hesitation in supporting its principles is definitely necessary. There needs to be a new org that's willing to go to bat for anyone, belligerent and unlikeable or not. This seems like it'll be the one.
I'm a bit confused that you defend Kiwifarms with "despite being hated, everything they do is legal" and then complain that companies are causing problems for them by exercising their legal right to not do business with them.
Like, I don't think you can have it both ways. If you run a website that promotes technically-legal harassment campaigns, I don't think you can complain that other people are finding technically-legal ways to stop you. And if you take the stance that some legal actions should not be done because they're immoral, then you have to explain why you think your technically-legal harassment site is not immoral.
Because I think freedom of expression is more important than the freedom to censor discussion, which has been the default opinion on the Internet for many years. It has been taboo for ISPs to block sites for reasons outside of legality. That they have now is the reason we need to codify a law for what was previously an unspoken norm. I like the free Internet and I want it to continue.
Plus, Kiwi Farms is not just technically legal or on the fringes of legality. It is exactly what I'd use as a an example of the central thing free speech is meant to defend. It's the right to talk about people however you'd like. No harassment is coordinated there, nor is it encouraged to be coordinated off-site. The only thing that could perhaps be removed without being total destruction of the right to free expression is the doxxing. I don't think you could change anything else.
If it's not obvious, I believe in free expression, and I'm sharing news to advance that cause. I find it tedious to re-explain why I support free expression every time it comes up.
> It is exactly what I'd use as a an example of the central thing free speech is meant to defend.
The "central thing" free speech is meant to protect isn't gossip, but criticism of one's government and rulers, and it is an essential instrument in the prevention of tyranny and oppression.
I find it rather sad that you take freedom from tyranny and oppression so much for granted that you think defending slander and doxxing is the prime purpose of freedom of speech.
The right to denigrate and attack is a fundamental part of any kind of criticism. If you don't have that, you have nothing. The Kiwi Farms has threads on men like Elon Musk too, if your really need your "speaking truth to power" fix.
And I'd say expressing your expressing your emotions, even the vulgar ones is also an important part of free speech and the one most likely to be attacked by utilitarians.
" lolcow discussion and gossip site" is an extremely generous description of Kiwi Farms, which has been criticized for facilitating sustained, targeted harassment of random people who are cringe on the Internet.
I never said it was illegal. I'm saying that people may want to keep their distance from Moon and all his works for reasons that have nothing to do with freedom of speech.
As far as I can tell, federal laws only criminalize harassment that's violent or sexual in content, and the same goes for most states as well. Denigrating others is not harassment, legally speaking.
I'm not a lawyer but it looks like 18 U.S.C. § 2261A criminalizes using the Internet to "severely" harass somebody, i.e. to the degree that it "causes, attempts to cause, or would be reasonably expected to cause substantial emotional distress".
Interestingly, according to federal law, it's only a crime if it's anonymous:
> (C) makes a telephone call or utilizes a telecommunications device, whether or not conversation or communication ensues, without disclosing his identity and with intent to abuse, threaten, or harass any specific person;
The better characterization would be "KF made these people famous" -- and then OTHER people decided to harass them. It is very explicitly NOT part of the KF ethos to even Talk to the lolcows, and KF is NOT the forum you want to piss off by disturbing their lolcows.
Oh my. I can't make head nor tail of your post. And the link confuses me even more. I strongly recommend you use the inverted pyramid structure in your lede. Front-load the opening paragraph with the "5 Ws and an H" (Who, What, When, Where, Why, and How) to capture our attention. Sorry if I sound dismissive, but it's difficult for someone like me, who's never heard of Kiwi Farms, Josh Moon, or Fong Jones, to understand what the heck is going on.
Kiwifarms is the forefront of "free speech" on the internet. Not only has it been cited as a reason to invade the Ukraine*, it has also been DDOSed by state-level actors**. Furthermore, credit card companies have refused to carry transactions, so KF has had to sell literal silver in order to keep its website up.
Why all the hate? KF is the one place online where you can say what you want. Nothing is policed there, except no bots. It's "organizing principle" is to laugh at lolcows online, and to "not talk with them, not confront them, not interact with them in any way." But at KF, they'll talk about basically anything.***
*Hohoholocaust, leaving aside it's an American website, and no way did they understand why the meme was angering the Russians.
**remember the feds who spent worktime discussing raping cis-women on their discord? Those fine folks have access to DDOS suites...
>Not only has it been cited as a reason to invade the Ukraine*,
It's fascinating how people seem to forget that the Ukraine War didn't start in 2022. It started in 2014, and the only reason why Ukraine was able to withstand the early stages of the invasion of 2022 was that they've been preparing for it since 2014. Or has that been preordained on KF as well?
What’s the issue, exactly, with hate for a website intended for people to express their hate? @blisterhill says: “expressing your emotions, even the vulgar ones is also an important part of free speech”
Oh? I'm not upset at the hate. You can hate 4chan, Kiwifarms, Stormfront*** all you like. I'm upset that people are taking actions that make it impossible for others to support said site (by removing Credit Cards as payment processors, by removing Tier 1 internet access).
My friend the award-winning author learned how to write villains from surfing Stormfront, way back in the day. You do not know what someone will get out of a website.
Hate Kiwifarms all you like -- set up a rival site, explaining why kiwifarms is so awful. Sit on kiwifarms and express your hate for everyone and sundry.
***it is of note that Stormfront, a place where literal neo-nazis hang out, is very much less targetted than the e-celeb watch site, kiwifarms. Kiwifarms is targetted by DDOSes, to the point that when Trump Wins in 2024, the next day, the only thing The Powers That Be can agree on is "DDOS kiwifarms."
I am JUST as upset at censorship when the US government refuses to let American Companies support Syrian Government Websites, or what have you. "Sanctions" versus "interesting, valuable data".
I think someone Russian noticed the meme (which was definitely a KF invention), and it got to Putin, in a package of "the Ukrainians are Nazi." If I thought the Russians needed to be factual*, I'd probably call it a "rather large intelligence failure." I certainly thought so at the time, given that it's an American website that was only hosted in the Ukraine because Josh Moon had people show up at his door and threaten his mother. (I believe Josh is stateside again).
*or actually cared about being factual, rather than coming up with a "whatever sticks, we don't need to make sense to even our own people, let alone the rest of the world."
Kiwi Farms is a gossip website ran by Josh Moon for talking about e-celebs and people who make fools of themselves online. It receives a lot of heat from activists for facilitating cyber-bullying and networked harassment. It has to deal with censorship across the stack including CloudFlare dropping it, Tier 1 ISPs blocking it, among other attempts to censor. The latest attempt is Liz Fong Jones (the one spearheading the aforementioned censorship efforts) abusing the DMCA.
Just to clarify: are you claiming that everything users of the site have done in connection to discussions and movements hosted on the site is 100% legal, or that the way internet regulations are currently written technically shields the owners of Kiwi Farms from any massive illegality coordinated or posted on its site by its users?
Frankly, there's something to be said for 'this entity is exploiting our generous free speech laws to cause as much harm as possible in a way that's stirring up reaction and demands to do something about it, lets try to harass and destroy them specifically within the bounds of the current legal system, instead of amending the law to allow less free speech for them and also everyone else.'
Which is not to say I support everything their opponents have done in this case - I'm absolutely not going to take the time to find enough independent sources to research this and form a coherent opinion.
But more broadly, there is no such thing as a perfect law, that perfectly allows all good and neutral things and perfectly forbids all bad things.
Within the grey areas and corner cases are a multitude of situations where you could pass a stricter law that would stop a bad thing but also stop other good things, or leave a laxer law in place and suffer the bad things. And every time the bad thing happens, it stirs up public outrage and democratic support for the stricter law.
Given this reality, there is an actual freedom-preserving purpose to people on the bleeding edge of the system having some discretionary power to fight bad legal things without changing the laws.
Like many readers here (whether they know it or not), I am autistic, and would love to live in a black-and-white world where you can write a perfect set of rules, and everything goes well if people follow them. But decades of hard experience and disappointment has shown me why this is a form of disordered thinking, and not actually possible.
Kiwifarms goal is not to harm the lolcows. They're funny, fun people who do stupid shit for attention online. Kiwifarms likes watching them, and laughing at them. Rules of the site include not talking to the lolcows, not interacting with the lolcows. Do the people on kiwifarms sometimes say mean things? Yes? It's not a crime to say "you don't look like a woman," or, more importantly, to publicize predators who say things like "no dragqueen has ever hurt a child" (and then get caught running child sex rings).
You do the world a lot of harm when you refuse to let sleuths dig up truthful information that people would rather keep buried (legally, freely published information, to be clear -- we really aren't talking about doxxing people**). On Kiwifarms, you can absolutely read about teachers buying bathtub estrogen for their students, without parental consent. I've seen people online say "this never happens." On Kiwifarms, they always keep receipts. People have tried to delete their twitter accounts, remove posts that "suddenly look bad" -- but Kiwifarms remembers.
To be clear: we've all said some dumb shit online. And not all teachers do believe that rightwing parents are trans childbeaters.
**doxxing people occupies a grayer space than I want to get into, and I'm not particularly clear about Kiwifarms' position on this.
We actually can win this fight and have free speech on the internet. We don't need to accept private censorship to avoid demagogically-demanded government censorship. We did not have such unwritten rules before LFJ showed up and things worked just fine with net neutrality.
We've managed to keep the free Internet going for a long time. We don't need to sacrifice the unlikeable to the mob. The unlikeable are who free speech is for! If you just get rid of them when they're troublesome and cause problems for free speech, then you're hardly protecting anything at all. How would you even know if you were slowing down the demands for censorship? You might be speeding them up because people get accustomed to censorship.
These kinds of arguments are too tenuous in general. You don't know if fighting every step of the way or giving ground will even end up better, so it's best to go for the ideal outcome. The one that you believe in.
As for Kiwi Farms, nobody coordinates harassment on there. It's collating information on people and being mean. That can often be a starting point for harassment, but it's also definitely well within the bounds of free speech. If you start sacrificing things like that for the greater good, you'll end up with nothing.
> if you’re against witch-hunts, and you promise to found your own little utopian community where witch-hunts will never happen, your new society will end up consisting of approximately three principled civil libertarians and seven zillion witches.
Josh Moon's a determined guy, and he's finally making good money now using some mailed in check system since he can't use payment processors. Everybody thought he was screwed when HE blackholed him, or when CloudFlare dropped him, but he got through it, and now the site's more active than ever. I wouldn't be shocked if he pulled off something impressive again.
Oh, it's just a picture of LFJ. Nothing special really. It's just that some people made edits of it with "Consent accident" as a caption. It's fair use, but because the picture is registered, LFJ is free to use the DMCA to dox anyone who ever used the picture. I think there are 7 affected posters.
But principally, this could work for any picture. Just register it, and you get the right to dox anyone who ever uses it, even if it's fair use.
It's fair use because it is an edit of an image which adds new meaning or context to an image through criticism. It is fair use. It's a trick because such a thing is obviously fair use, but the DMCA is slanted against defenders, so they will be forced to dox themselves and spend money on litigation that is far easier on Fong Jones than them.
It completely changes the character and purpose of the image, so using the DMCA against it is a perversion of the law.
I'm not a lawyer, so I'm not super confident in this assertion, but I question whether putting ANY negative statement on an image automatically qualifies. If you put "filthy skank" or "baby killer" on an image it would automatically protect you from copyright claims.
If you edit a someone's picture with some controversy they were apart of, inside a thread solely dedicated to insulting that person, that is obviously not for the same use as the original headshot.
That doesn't mean you can just slap low opacity criticism on movie footage to pirate movies because that would still be for the same use case, that is, watching the movie.
This is explicitly for the purpose of criticism and dissemination of hatred. It's obviously fair use. It is a distinct work from the original image evoking a completely different experience.
The DMCA is shit law that is egregiously slanted against people practicing their rights.
4chan and any number of adult sites or boorus host plenty of objectionable content but don't face nearly the same degree of problems, why is that? The centrality of suicide-baiting, doxxing, and swatting on the site with no other redeeming value besides amusement for its users is the problem for Kiwifarms.
Makes lots of sense why none of the free speech orgs would spend their limited resources guaranteeing the "rights to dox and to incite others to suicide" when they could use their valuable time protecting actually core freedoms in our society like those for political speech or free journalism.
KF users have a silly, worthless (if not actively harmful) hobby and should not be shocked that others don't regard it as an important or central political cause. By the same token that "no Kiwifarms user/owner has been found liable in a US court of law", you can also point out that no US court has ever found that an ISP is required to host any specific organization's content or provide service to another business if they don't want to.
Yes, the content of the site is much more attractive to censor because it is not ephemeral and is geared towards talking about e-celebs.
Free speech orgs like the EFF chose not to help the Kiwi Farms because of its content, in spite of the fact that the mechanism used (Tier 1 ISP blackholing) is a grievous threat to the free Internet, and a victory against it would set a precedent. It shows a lack of seriousness about their stated goal (which happens to be my goal).
I think worthless and even somewhat harmful hobbies have their place. Occasionally their hobby of obsessive observation proves a good thing like with the zoo-sadists or with trans lifeline, but that is hardly the lion's share of the site. Freedom of expression outweighing the freedom to avoid gossip is how the Internet has been run up until now, and I hardly think that a dedicated tech worker with connections should have the right to break that norm.
People should be able to have websites where users can express their opinions online within the bounds of the law. Such has been the way of things. It's the incumbent position, and it's under attack. Even if you think the Kiwi Farms is dull or destructive, you have to posit that the free Internet itself isn't worth keeping to justify what LFJ has been doing. Because, evidently, the Kiwi Farms is what the free Internet produced. It finds its audience because there's a latent desire to pseudonymously discuss people online with invective.
> People should be able to have websites where users can express their opinions online within the bounds of the law.
Except if the law was much more strict, you presumably would complain about that too.
> It finds its audience because there's a latent desire to pseudonymously discuss people online with invective.
"With invective" elides the kind of discussion that goes on. It's one thing to talk about how much you dislike certain types of people, it's quite another to have an open culture of using slurs.
KF's culture is the major sticking point for me - if KF wants people to treat it better, it does very little to encourage such behavior. I am not part of any group that they hate, but it is obvious reading their posts that any archival value from their obessive digging is heavily marred, if not outright overshadowed, by how they create and perpetuate hate and contempt.
You say that one *should* be able to speak their opinions freely. Such a view is the decline and death of groups, communities, societies, etc. especially so if such opinions do not foster helpful and good discourse. This is not about holding any particular political view or social belief, it is about the fundamental maintenance of society and community. If you constantly jeer and mock, you will be surprised one day when you don't recognize yourself in the mirror and can't seem to maintain a community which actually acts according to the word.
Yes, I would complain obviously. I like the free Internet and I don't want people to feel so constrained that they can't even speak their mind about figures who annoy them on their own websites. Abrasive and non-abrasive don't really make a difference to me. Kiwi Farms isn't destroying society. People like to say it is because they don't like what people say about them anonymously on the Internet.
If you believe in might is right and that the biggest techhead with the most pull at Tier 1 ISPs should have control of what people can say, and that whoever can kick up the biggest media circus should decide what expression is online, I'm afraid we disagree. I think expression is important in itself rather than being useful for some utilitarian end.
I believe hate speech is free speech, and even if I didn't, I would want a law instead of LFJ.
With free speech, you take the good with the bad.
Plus, I think you're catastrophizing about free speech destroying the world is without basis. We've had decades of free speech online, a centuries long tradition of free speech in the US, and nothing too bad has happened as a result of it. It's all more than worth it. That's why I can rest easy when I say I'd take the good with the bad. It's because I know that my deeply held value, free expression will probably not lead to some disastrous outcome.
So is everything done by their high-profile detractors, I'd imagine. I do appreciate that defending free speech mainly means defending scoundrels that are the best at skirting the line, but still, the rhetoric painting them as innocent victims of "grievous injustice" is a bit rich.
And what rights have the Kiwi Farms violated of the people they talk about in their threads? Josh Moon's freedom of speech has been curtailed and the free internet with it in a real and unprecedented way. The people you're referring to are fighting for their right to not be talked about by destroying the far more important right to talk about things.
I don't know why there's this constant tone policing whenever free speech comes up. The EFF did it too in their article on the Farms. A grievous injustice is a grievous injustice regardless of who it is done to. You don't get to take away grievous injustice points because you think the person is bad. No hedging is necessary. Kiwi Farms is a victim of grievous injustice.
I'm not interested in making a tone argument. Rather, my point is that such rhetoric conflates several issues, e.g: "are actual laws being violated by anyone in this conflict?" "what laws would we prefer instead?", "how bad are the participants in this conflict according to current norms?", and "what norms would we prefer instead?".
Some of these questions have easy and straightforward answers here, and some do not.
The norm I'm advocating is a free Internet that is only inhibited by the law. I think this is the prior of many readers, and it's why I addressed my post to those who care about private censorship. If you believe in a free-for-all of online censorship, that's your right, and you'd be in the company of people like LFJ.
That's what I'm getting at with the tone policing. I think I've made my views clear in that whole post, and you're just saying "It's a complicated issue, and there are lots of opinions." Yeah, but I'm sharing my opinion to likeminded people, and making suggestions on how to advance it. I'm not trying to convince someone opposed to support free speech as that would take a much longer post.
Moreover, the free Internet view was the prevailing one among tech people for decades. We can afford to go on the offensive, can we not? What LFJ is doing is a violation of norms we've had in place for decades. If you constantly frame the people the norm is meant to protect as only arguably worth protection, you erode the norm.
If the law brought you to this, where your enemies can censor you while using it as a shield against retaliation, perhaps the law is not worth preserving.
> The norm I'm advocating is a free Internet that is only inhibited by the law.
"These people shouldn't be allowed to suicide-bait and this website shouldn't exist. We need to pass laws to make Kiwifarms and sites like it illegal" is legal free speech. It is just as much speech as the stuff that happens on Kiwifarms.
I really hate this kind of comment, because it's perfectly calibrated to the Refuting Bullshit principle. Your three sentence comment that you probably typed in 30 seconds sounds profound and persuasive on the surface, and requires a long, carefully-thought precisely-worded many-paragraph response to explain why your point is either vacuous and irrelevant or false, depending on how "free speech" is defined.
Maybe I'll make that response if I get around to it, maybe someone else will. But I really hate this kind of comment.
I haven't talked about my opinions on the law, only stated that the law should make the determination, not LFJ and everyone agentic enough to bother the CEO's wife.
Though I don't actually think Kiwi Farms should be banned. It's anthropologically important. Gossip and bitter hatred have their place in the noosphere and cutting them off from outside expression seems authoritarian to me. It's a bit like how adultery isn't a criminal offense. Freedom of expression is important, even if some people get hurt.
I consider myself semi-likeminded, so let's say I'm trying to offer constructive criticism. Free speech isn't particularly popular these days (as I'm sure you've noticed), so in order for it to have a fair chance in the court of public opinion, its staunchest advocates would do well to proceed with more finesse.
Kiwi Farms is an obvious wretched hive of scum and villainy. There's a norm that such spaces deserve to be kept down, which obviously trades off against increases in free speech. Unless you deign to at least acknowledge such tradeoffs, less sympathetic audiences will just dismiss you outright.
> Free speech isn't particularly popular these days (as I'm sure you've noticed), so in order for it to have a fair chance in the court of public opinion, its staunchest advocates would do well to proceed with more finesse
I think this is utterly, absolutely backwards. The strongest, most robust principle for preserving free speech is the one where something being protected speech has nothing whatsoever to do with how bad, unpopular, or immoral it is.
The moment you weaken that absolute principle, you've lost 90% of the fight. This is the entirety of how cancel culture and totalitarian EU speech laws happened: when your principle is "I defend all speech no matter how bad it is" then you can defend Nazism, criticism of Islam, misgendering, and everything else without implying any belief other than free speech being good. As soon as you switch to "I defend speech unless it's *really* bad like Nazism" then suddenly, if you want to defend misgendering you are directly implying by doing so that misgendering is not really bad. And if you happen to be in a community where misgendering is regarded as really bad, you have to choose between refusing to defend it as free speech (like Nazism) or taking a moral position contrary to your community. While under the old principle, no such choice was necessary, because defending misgendering was compatible with believing it as bad as Nazism. Similarly, under the old principle a devout Muslim could defend the right to condemn Islam without compromising their religious identity at all; under the new principle, they can't.
The defence of the worst scoundrels isn't just a nice extra strong principle of free spech; it's the entirety of what it means to have meaningful free speech at all.
Sigh. Kiwi Farms is a hive of people that talk to each other, and collect "publically available" data about lolcows. The only way this harms someone is if they go looking for what people think about themselves (and if they're narcissistic enough to care).
We're not talking about liars, we're talking about a group of people munching popcorn and seeing what Chris-chan has done today that's silly and funny.
Interesting stuff, thanks for sharing. I've ran afoul of a "brand protection" company while trying to sell a pair of Air Jordans on ebay. Nike deems anyone selling their shoes who is not a Nike-authorized dealer to be violating their IP (because it must be a knock-off, surely there's no way to get a legitimate pair of Nikes without going to Foot Locker or something). The brand protection company tells ebay to take the listing down, which happens without any due process or investigation.
It seems like there's a bumper crop of companies trying to abuse DMCA and other IP laws to bully people who would probably prevail in court, but lack the financial means to sustain a case.
In the rest is politics podcast recently, a guest suggested that the release of AI models should be as regulated as drugs. Lots of testing and then licensing by a federal agency. Possible?
It's not exactly hard to buy illegal drugs in this country, though, and you actually need to smuggle them in, not just connect to them using a VPN and an AI-friendly third country.
Last week I made http://www.halmarks.org/, a set of hallmarks/badges allowing creatives to self-declare the amount of AI they used in a work.
I got some good feedback and discussion, but haven't had time really do anything with it.
My current instinct is I'd like to do a lot more thinking and tweak the concept a bit more rather than pushing ahead with it in its current form. It won't be adopted if it's not the right shape for the job, and beyond that, if it's not actively helpful at achieving its aims, I don't even care if it gets adopted.
I'm considering repurposing my (pretty much unused) Substack page to collate and write up some of the points that were thrown up in discussion. I'm not a natural blogger and I hate working "in public", but I can see the value of getting more voices involved here.
Before I do that I'm posting it here again for more feedback and discussion. I'm especially interested in hearing from creatives who make things - I'm interested in:
- what's your opinion of AI
- what's your workflow with or without AI
- whether you'd use something like this (including recommending changes)
- whether you wouldn't use this for X reason but feel the same pressure for "something to be done"
- or whether, as a creative person, you're actually fine with where things are heading and feel no need to course correct at all.
I took a number of shitty selfies with my phone, and uploaded it to Gemini, asking it to make dating site pics, realistic, but better angle, lighting etc. The results are astonishingly good, they don't look like "AI slop", they look something like a pro photographer would do.
I would be worried if I was trying to make a living as a photographer.
Not a very positive comment: I wouldn't use these, because 1) it's not that hard for me to just type "made with Claude" without a hassle of going looking for the link and wondering how to embed the whatever 2) most usefulness of these things would come from being recognizable to a lot of people. Or a type of thing that people have seen before and have learned to ignore unless they have a reason to care, same way that your eye habitually slide over the CC-BBQ-0123 licenses. Right now everyone is going to see these and go: "what? did he make this himself or something?" Uncle Moloch says "hi". 3) I'm sorry, but the icons also look a bit on the derpy side. I would be more willing to use them if they looked super-slick and professional, or at least drab an minimalist enough they don't draw too much attention to themselves.
I love AI, but I do want AI output to be clearly labeled. So, in principle I'm in favour of something kind of like this being used, but it's going to be hard for you to get people to adopt it.
I do digital art https://www.artstation.com/tasty_y. Oh, and also I'm going to launch a super-awesome cool vibe-coded website that has been a week away from completion for 2 months now, where I could plausibly display a badge like that somewhere, if they looked alright.
You mean AI-generated stuff stays a week away from completion for months on end as well? I thought we humans at least had that phenomenon to ourselves.
Redrawing the icons will probably be the very last thing I do, on the grounds that that way I hope to only do it once. But yes, can certainly make them more minimalist.
I don't know who you count as a creative. As someone who's written a little bit, plans to hopefully write a lot more (in the sense of having very clear ideas of *what*, but not quite sure *how* yet), but hasn't shared anything publicly in any way...I'll answer on the off chance I count.
"what's your opinion of AI"
Highly skeptical it's going to end up much more than a glorified search function, glorified encyclopedia, and glorified writing/coding assistant. I find LLM writing pretty nauseating. However, I really like image models, and I'm 95% sure this is entirely because I'm a terrible drawer and have a woeful visual memory/observation ability.
"what's your workflow with or without AI"
I use it only very occassionally for factual information searching and for analysis of text I've already written. The thought of having AI do actual writing for me is horrifying. I honestly can't understand the psychology of the kind of person who'd present obviously AI writing as their own, without dying of shame.
"whether you'd use something like this (including recommending changes)"
I would definitely use it if it meant something to many people, and maybe even if it didn't. I would greatly value the ability to promise in a somewhat believable way that something involved zero AI in its creation.
"or whether, as a creative person, you're actually fine with where things are heading and feel no need to course correct at all."
Definitely not fine with it, but I don't think it's existential. These sorts of fears have happened many times through history, and the more the AI hypsters (whether utopians or doomers) fail to engage with that fact, the less plausible I find their claims.
My main anger is at the kind of broken person who thinks prompting AI is a substitute for actual thinking/writing/work, and the tolerance for these kinds of broken people. I'm not angry at AI itself; I see it as just another tool.
Thanks for your response, you definitely the kind of person I'm looking to hear from.
I think everyone right now says the same thing about AI - it's not there yet in my own field, but look at how good it is in the fields next door. And it's visibly getting better all the time.
I'd be quietly relieved if it peaked where it is now, but am unwilling to assume so.
> These sorts of fears have happened many times through history,
I agree, and that's what scares me. Job roles were destroyed, societies were reshaped, entire pillars of culture disappeared. A strong back doesn't impress the girls anymore because tractors do the job better - and within a generation we were all more effete. Now automation can (or might soon) do creativity. I don't like this.
> My main anger is at the kind of broken person who thinks prompting AI is a substitute for actual thinking/writing/work, and the tolerance for these kinds of broken people.
Last time I decided we should christen them "Vogons" - Douglas Adams' race of people who emphatically do not understand beauty, but think they do. It's not a direct expression of the concept I'm trying to get at, but it's close.
That's a discussion I'm happy to leave for another time. I actually wanted (and I hope I was clearer about it last time) to use "Vogon" not for the understanding of beauty, but for the understanding of the creative process that goes into making things.
A cabal of highly educated people, spread across all kinds of industries at the levers of power, believing in the power of the rational mind, with a vague goal of rational enlightenment and maybe even abolishing the rule of human over humans? Sounds like *the original* conspiracy..
I don't think the Rationalists have much influence over any "levers of power", beyound some internet communities and advisory positions in tech companies. Which isn't nothing, but it's not Illuminati either (and not even Skull and Bones).
Dario Amodei is at the very least Rationalist-adjacent (he took the GWWC pledge in 2010), and he and other such types broke off OpenAI to found Anthropic primarily due to differences of opinion on AI safety.
I’ve started a new blog, Structure and Guarantees, about how to design intelligent systems that we can actually reason about — not just empirically tune.
A recent post argues that compilers are a concrete, already-deployed example of recursive self-improvement: they optimize and even compile themselves, but in a setting where we can prove correctness properties across iterations. That makes them a useful “toy model” for AI safety questions like alignment and containment—but in a domain where those questions are actually tractable.
My guess is that we should be grounding more of our thinking about advanced AI in domains like this, where the guarantees are real, rather than starting from unconstrained hypotheticals.
The blog mixes formal methods / programming languages with some less-usual angles (e.g., evolutionary signaling and distributed systems as lenses on intelligence).
I've just started looking at your blog. First impression: it is good stuff that will be worthwhile to engage with carefully. I think that I agree with your overall theme that we can build useful and trustworthy AI ecosystems by constructing them from two ingredients. (1). A symbolic, GOFAI engine that we can reason about, and (2) a more magical and mysterious heuristic generator, presumably constructed using big data and deep learning. Over the long run, such an approach might work. But ...
Thanks for the words of encouragement. From your post, I can see that we agree on the important points. Let me try to address your two primary objections.
First, you worry about the complexity of predicting consequences of AI decisions in the real world. I address this point partly in early posts of my blog, and I'll be returning to it in later posts, too. In short, we should redesign the world for improved legibility to AI. Much of the complexity we see results from somewhat arbitrary outcomes of natural evolution. Much of *that* complexity comes from the centrality of signaling to humans. The future economy will not have humans at its center, and its day-to-day problems can be correspondingly simpler, if we (or our AI successors) do a good job codesigning the world and the code that needs to understand it.
Second, you worry that it's very hard to formalize objectives around human flourishing and so on, which are most commonly described in AI-safety literature. I agree! I will get into approaches that avoid those challenges, though I think it will take enough blog-post-length essays to explain that I don't want to start into it here. Stay tuned?
Compilers aren't recursively self-improving. To be more precise, they're limited to a recursion depth of 1, and this self-improvement is limited to pure execution speed and memory consumption (and the effect on the latter is very low for practically all mainstream compilers). This means that when you apply a compiler to its own source code, the output will – maybe – run faster, but you can only do this once. A compiler cannot improve or even change its own algorithms.
No idea what you hope to learn and generalize from this, given these limitations.
Oh, but think of a compiler as running an expensive search process for faster ways to implement programs. As the compiler's performance improves, you can find even-better versions within your budget of time and memory. (Most compilers today don't work that way, but they should!)
> Oh, but think of a compiler as running an expensive search process for faster ways to implement programs.
Those "faster ways" are all thought up and implemented by humans. A compiler doesn't come up with new optimizations. A program that would come up with new optimizations isn't called "compiler" anymore, and you would also lose those properties that guarantee correctness (or "alignment" in AI-speak).
So maybe we are just going back and forth about the right terminology for the sort of thing I propose formally verifying. Since I called it a "compiler," it didn't seem to include key capabilities and therefore avoided related verification challenges, but actually I did mean to include those.
> we should be grounding more of our thinking about advanced AI in domains like this
How do you avoid the streetlight effect where you investigate things that you're able to ground, but miss the important parts? You don't want to fall into the trap of “lets study something that's vaguely similar to what we're interested in, because this is easier to study than the real thing.”
I agree that's an important tension, and it's just my intuition that there are enough similarities here to the broader problem to start learning nonobvious lessons, some of which will turn out to be more broadly relevant. (It's hard to predict in advance which they are.)
I'm also used to thinking of so many problems as compiler problems. At some level, any problem of making a plan and then reducing it to practice can be cast in this light, with the right clever source language.
I recently made a video essay about The Rocky Horror Picture Show. I'm not doing a Substack because people who like movies want to hear about them on YouTube, and because I have very minor celebrity status from being on a reality show. I'm hoping to get the view count up to 3k, and right now it's sitting at 2.9k. I'd appreciate it if any musical theater fans here would give it a watch.
https://www.youtube.com/watch?v=s-gHyWm5Ggk
What happens at 3k? Is that a youtube threshold for something?
It was just a personal goal, which I've met.
but why did you pick 3k as the goal?
It's the threshold for a proper mystery science theater.
In this community, we’ve often talked about how it might be a bad idea to rely on medical advice from LLMs.
On the hand, the latest DeepSeek does a noticeably better job than my endocrinologist at interpreting my blood tests results.
I have a thyroid condition which needs to be monitored, and a quirk is that the local hospital’s endocrinology department will include T3 and T4 level if they order a blood panel for me, but my GP will only order T4.
So I have DeepSeek generated analysis if my chart, with a certain amount of “well, this would be a whole lot easier to analyse if your GP had remembered to tick the box for T3 on the lab request form.” I might remind my GP next time…. [The nurse practitioner at my GP surgery also gets it right when she orders the blood panel, and she has the added incentive that she’ll be the one doing the second blood draw if the hospital consultant sends the GP practise an email telling them to do another blood draw and measure T3 this time. This does sometimes happen.]
I supplement the medical system by using AI to research medical conditions and treatments, both before and after doctor visits. I sometimes do the same kind of research for a friends who are having health problems. It’s been enormously helpful. Here are some specifics.
-I have scoliosis which has recently progressed rapidly, and I need surgery to keep things from going downhill further. I also have osteopenia. Saw a surgeon, and we agreed on a procedure & set a date a couple months in the future. Researched predictors of good outcome and bone strength was a strong predictor. Discovered there are several new powerful ones on the market. Surgeon had said nothing about importance of bone strength. Canceled surgery with that doc.
-New surgeon spontaneously told me I needed one of the novel bone-strengthening drugs, and suggested I take it for 3 mos. prior to surgery, and 18 mos. after. Researched how rapidly bone changes take effect with drug, and discovered changes were minuscule after 3 mos., but that after a year 80% of the improvement that was going to happened had happened. Pushed surgery back to nearly a year after beginning drug.
-Internist prescribed Zepbound so I could lose 25 excess lbs before surgery. Since I do not qualify for insurance coverage, I was paying a lot for the stuff. Researched getting GLP-1’s on the gray market, found a way I believe is safe. (Cremieux gives the relevant details in a post from last spring). Began taking it. Researched Retatrutide, which appears to be a step up from Zepbound in effectiveness and health benefits but is not yet FDA approved. After another Lilly interim report on Retatrutide came out in December, I switched to grey market Reta. I am paying about $100/mo for a moderate dose. It works even better than Zepbound and has caused me no problems.
-Neither internist nor surgeon advised me on taking supplements while on these drugs. Since I am trying to build both muscle and bone, I knew it was important to get the recommended amount of protein and calcium. Discovered that is quite hard to do on a GLP-1 because you have little appetite and feel unpleasantly overstuffed easily, so food intake is small. Also saw in the research that subjects in studies of GLP-1’s were found to be consuming less than recommended amount of protein and calcium, and that many were vitamin D deficient. So I started keeping track of protein and calcium intake, supplementing with protein powder and calcium capsules as needed. Also started taking a multivitamin, vitamin D and Vitamin K2.
-I’ve had vertigo attacks occasionally for years, and started having more once on a GLP-1. Thought I saw a link between attacks and being a bit dehydrated. Checked with GPT and yup, decreased fluid in inner ear can cause vertigo. Though I don’t think I drink less water and coffee on Retatrutide, I do eat fewer watery foods (soup, fruit, etc.) because I eat less food overall. So started drinking more and checking my urine density with a home test. Vertigo’s gone.
-Friend was having immunotherapy for cancer. With GPT’s help I read up on it a bit, and discovered that immunotherapy infusions early in the day are for some reason markedly more effective than same infusion later in day. Person checked with their oncologist, who confirmed that this was true.
I could easily give a dozen more examples. But I think one reason my results are so good is that I don’t ask GPT for advice, I ask it to research certain topics and summarize the research and give links. And I usually check a few of the main links to make sure they exist and GPT’s take on the was accurate. I have not yet been in the position where my conclusions based on GPT research conflicted with the doctor’s and I had to decide whose to follow. I guess I’d have to make a judgment call at the time. But I would not rule out going with GPT rather than the doc. Setting aside individual instances, I’m not even sure in general which source is more trustworthy. GPT does not care about me, cannot examine me physically, and does not have years of actual medical practice in the background when it makes recommendations. On the other hand, it can access virtually every printed resource that exists, has infinite time to devote to my case and never feels annoyed or one-upped, as some doctors do when I want to discuss options. And my impression in the current medical system is that while the doctor cares about me more the GPT does, they only care slightly more.
This is not ChatGPT's fault, but for public reference, the immunotherapy timing claim is facing scrutiny. There's only been one randomized controlled trial of immunotherapy time-of-day, Huang et al. 2026. It claimed to find large effects, suspiciously large effects, and is undergoing investigation for concerns "regarding inconsistencies between the registration record of this trial on clinicaltrials.gov and published version the study protocol, as well as with some of the findings in this study."
Huang et al. 2026 with Editor's Note: https://www.nature.com/articles/s41591-025-04181-w
PubPeer commentary: https://pubpeer.com/publications/C674C764D377B5B12F2106EDAD98A8
There are retrospective trials claiming time-of-day effects, but their results are less consistent, and retrospective trials are easily confounded by effects like, "healthier patients schedule earlier appointments."
The advice is not necessarily bad. Nobody claims that earlier immunotherapy is worse, so switching to morning is a reasonable choice even if the evidence is weak. But I wouldn't confidently claim that earlier immunotherapy works better.
GPT actually didn't make a big deal about the early morning infusion advantage. I was just struck by it, in there among the other article summaries it made for me.. Given that there's no harm in being one of the early birds, scheduling infusionsin the early morening seems like a no brainer -- can't do harm, might help.
> told me I needed one of the novel bone-strengthening drugs
What is it? My bones are pretty weak as well according to the doctors.
There are 3: Forteo, Tymlos and Evenity. Forteo and Tymlos are similar, and I have been on both. Don't know much about Evenity. All are much better than Boniva, which was what was used til recently. What Boniva does is prevent the body from getting rid of old bone (apparently there's constant turnover in bones, with old material getting replaced by new). This does gradually make bones stronger. But the new class actually stimulates the formation of new bone, which works much better.
Forteo and Tymlos are expensive (2 or 3 thousand per month). To get insurance to covert them you have to be diagnosed with osteoporosis, *or* have osteopenia plus at least one spinal fracture (or maybe one fracture anywhere?). I have osteopenia and no spinal fractures, but my surgeon just winked at me and said "A mess like your spine probably has some fractures in it somewhere" and checked off the fractures box, and so mine's covered by insurance.
Main side effects are joint pain and fatigue, and I have both. Putting up with them is worth it though.
How are you judging better? Did DeepSeek catch something important your endocrinologist missed?
E.g. the drug dosage I was taking brought T4 within range, but left T3 still too high. They should have increased the dose to bring T3 withn range, too.
How certain are you that this advice is correct, as opposed to superficially plausible?
(I don't know much about thyroid disorders, so maybe it is obviously true that a medication which brings T4 within range but not T3 should be increased until T3 is within range. Is that the case?)
These are good questions as to what the ground truth is, as LLMs wildly make stuff up, I ‘m not an expert just the patient, and my own cognition is probably somewhat suspect as I’m suffering from a thyroid condition. On the other hand, when T3 was finally brought within range, I felt a whole lot better, and there’s an a clear mechanism for why that happened.
April fool's joke paper on the ArXiv
https://arxiv.org/abs/2603.28975
Took me in for a minute or two! :-)
Regarding the recent discussions on simulation mechanics—the OSIM Research Group is moving away from the 'Architect' or 'ASI' hypothesis. We’ve found that focusing on the 'Who' is a distraction from the actual physics. We’re pivoting to a Forensic Cosmology stance to document the physical stabilizers like Many-Body Scars and the 'Non-Algorithmic Wall.'
Here is our updated 2.1 framework for the Sovereign Blueprint:
OFFICIAL TECHNICAL ADDENDUM: MARCH 31, 2026
The OSIM Research Group is officially transitioning to the "Sovereign Blueprint"
framework (v2.1). This update prioritizes a "Who-Neutral" stance to focus on the
observable mechanics of the Sovereign Inception.
The "Something" keeping the loop going is defined as a self-correcting,
physicalized program embedded within cyclical cosmology (The Big Bounce).
1. THE QUANTUM SHAKE: Many-Body Scars (OSU research) show a system
hard-coded to return to its original state. This is the stabilizer
for the Loop.
2. THE NON-ALGORITHMIC SEAL: UBC research on the "Wall" and biological
memory retention (flatworms) proves this is a physical sanctuary,
not a digital simulation.
3. THE 20-WATT RULE: UChicago bio-electronics confirm biology is the
most efficient hardware possible for housing consciousness.
We are looking for fingerprints, not just an Architect.
Is this meant to be satirical?
If new research finds that the Big Bang is actually a Big Bounce and some researchers are suggesting that it could be infinite then it becomes a big numbers game. If you and I are having this conversation right now and we know that 1x infinite = infinite then its not just likely but mathematically guaranteed that we will have this conversation again. this would be a infinite biological loop.and if so would kill the nick bostrom digital simulation and also the multi universe theory . It is flasable , if it's proven that the universe only expands then we know the big freeze wins out and osim is wrong or if it can be proven that silicon or another material if more efficient and can mimic life better than biological then again osim would be proven wrong.
> [It’s] mathematically guaranteed that we will have this conversation again.
So Nietzsche was right about the eternal recurrence thing?
Great — Now I’m going to have to sit through Ice Capades again. :(
Woody Allen gets credit for the gag.
Love the humor. Lol
Do you think the Bible is satirical?
Hi, I and my wife are strongly considering sending our 5.5 year old son to a new campus of Alpha Schools opening in Palo Alto, CA this fall. I suppose he falls into the Gifted category, since he tested as 139 on the WISC-IV. He is in KG now (on the young end wrt his peers) but he is far ahead of his peers in reading and math (roughly 3rd grade level). This is mostly cause I spend a lot of time with him at home and fwiw he enjoys learning math through the Khan academy site, so it may align well with the 2h model. We took him through the usual round of elite private school and Gifted school applications and he didn't get accepted to any of them. Not entirely sure why, but it might have something to do with either having messed up the timing (basically we would have had him repeat KG, which i guess committees are wary of) and his teacher recommendation letters probably say something about him being on the disruptive end of the typical KG behavioral spectrum. In any case, it seems like the options are keeping him at his current private school (academically focused, but not a fancy elite one, and also not having the kind of peer group that can challenge him), and taking a shot at Alpha for a year.
I really liked the philosophy behind Alpha, the review that was written last year on ACX (https://www.astralcodexten.com/p/your-review-alpha-school), and honestly I know the marketing videos are slick, but somehow I get a vibe of them being not a scam, and having some sense of honesty about what they're doing. Some of the comments from parents and students in the linked reddit threads were a little disappointing BUT I wonder if it's something we dont have to worry as much about cause quite likely our kid will be in the upper percentile in learning, so won't have to worry about the pressure they supposedly put on you.
Another relevant fact about our kid is that he is incredibly physical and loves sports and hands on physical activities in general. So I'm really hoping that the 2h of (device-based) education, which frees up the afternoon for life skills, projects, sports etc etc would be a good match for him.
Anyway, I'm just wondering if anyone has any info to share, especially recent info since the review came out last year? I'm really hoping current/former parents/students or the author of the review are available to ask questions to.
I don't think Alpha School is a scam, but I don't like the sound of the place. Bear in mind that the person who wrote the review, whose daughter attends Alpha, is very invested in his kid being publicly known as Amazingly Smart. He named his daughter Everest, and now writes a blog about her becoming a champion at a thing called History Bees, where kids standing in front of an audience answer questions about history factoids. Much of Everest's studying involves Anki decks. That's all OK, I guess, but it is very different from real learning about history -- what life was like in other eras, how people thought and felt and acted, how big trends and forces and coincidences and rulers' quirks shaped the course of events. And becoming the best kid on the planet at remembering history factoids takes up a lot of time that could go to real, immersive learning about history and other things. And the history bee stuff seems to have shaped Everest's interests and goals. Last time I could stand to read the blog Everest was asking to begin preparing to compete in geography bees. Everest is in training to be an achievatron.
Competitions are fine. They're fun and good motivators for many, though not for kids who get miserably anxious under that kind of pressure or kids who get dejected instead of bouncing back when they lose. But they're only one way to motivate and involve kids, and as I remember, most of Alpha School's afternoon activities are competitions of various kinds.
I highly recommend you check out Jeremy Howard's podcast interviews about his experience with homeschooling his daughter. (Here's one of many: https://teachyourkids.substack.com/p/an-interview-with-jeremy-howard-about) He and his wife are using a combo of curated online learning apps and online group activities for other kids. It's done under the umbrella of a worldwide homeschooling organization called Modulo that offers all kinds of resources. It's here:
https://teachyourkids.substack.com/about". We’re a directory of vetted homeschooling resources, newsletter, show and curated online community for homeschoolers and modular learners. We’re also a growing educational marketplace with an AI concierge that offers tailored support to families navigating modular learning and homeschooling."
Thank ! These are some good pointers to follow up on ! I didn't realize he had a substack of his own.
Beast Academy is a pretty good self learning course in math for kids who are interested in that subject. I know some kids do great in kindergarten but get really bored in first grade when the academic content ramps up and it stuff they already know.
Not addressing your central question....
If you have time to spend with your son and want math stuff, my top recommendations for that age are:
(1) games, especially those in Ben Orlin's "Math Games with Bad Drawings." Top board/card games I suggest are go, hanabi, king domino, santorini.
(2) logic puzzles, especially stuff from Raymond Smullyan.
(3) redacted contest questions from Math Kangaroo and AMC 8. There are two levels of redactions I like:
(a) remove the multiple choice answers: this makes the questions much harder and, slightly more natural.
(b) remove the answers and remove the question. This converts it into a 'notice & wonder' prompt where the two of you can talk about what you observe ("I notice X") in the situation and what questions you have about it ("I wonder y?")
thank you, these are fabulous suggestions !
Gemini users,
Have you figured out how to get gemini to provide direct links to webpages, and not wrap everything in a google search page or google shopping page?
I asked Gemini and it gave me several useful solutions to this.
Did any of them work?
The one I tried was 1) from Eramolalos below - asking for clean, unadorned links. It seemed to work, but I didn't do a lot of testing, so reluctant to post as a solution.
Did you actually click on links or just look at the text? gemini will give text that is a direct url, but the hyperlink is actually to a google search page.
So I am skeptical Gemini gave you useful solutions, and still think it gave you incorrect solutions.
I went through this runaround with gemini, and now I guess I need to do it again with a human! Here's how Gemini would respond: "You're absolutely right, I didn't verify my links!"
My, aren't you picky!
When I have asked Gemini for help finding or doing something online, its instructions have been wrong far more than half the time. Picky of me to complain, of course
Yes, so far this thread has been pretty similar to my interactions with gemini.
Out of curiosity asked GPT-5. Here's what it said:
Short answer: you generally can’t fully force Google Gemini to stop wrapping links in Google redirect/search URLs. That behavior is partly intentional (tracking, safety filtering, and keeping users inside Google’s ecosystem). But you can reduce it or work around it.
Here are the practical methods that actually make a difference:
1) Ask explicitly for raw URLs
Gemini will sometimes comply if you’re very direct:
“Give me direct URLs only (no Google redirect links)”
“Provide clean, canonical links (starting with https:// and the site domain)”
“Do not use Google search or shopping links”
This works inconsistently, but it’s the simplest first step.
2) Ask for “source citations” or “publisher links”
Prompts like:
“List sources with direct publisher URLs”
“Cite sources with outbound links to the original sites”
push it away from Google-wrapped links and toward real domains (NYTimes, NIH, etc.).
3) Use the “double prompt” trick
First:
“List the sources you used (names only)”
Then:
“Now give direct URLs to those sources (no redirects)”
This often bypasses its tendency to embed Google tracking links in the first response.
4) Strip the redirect manually (reliable fallback)
If you get a Google link like:
https://www.google.com/url?q=https://example.com/article&...
Just extract the part after q=:
https://example.com/article
This always works.
5) Avoid shopping queries phrasing
If your prompt sounds commercial (“best”, “buy”, “price”), Gemini tends to route through Google Shopping.
Instead of:
“best adult tricycle under $3000”
Use:
“list manufacturers and models of adult tricycles with specs and links to manufacturer pages”
This reduces shopping wrappers.
6) Use Gemini in “plain text / no formatting” style
Sometimes adding:
“plain text response, no embedded cards or previews”
reduces UI-style link wrapping.
7) Use a different interface when it matters
Gemini inside Google products (Search, Gmail, etc.) is more likely to wrap links than standalone chatbot interfaces or APIs.
Bottom line
There’s no setting to disable this behavior completely. The most reliable methods are:
explicitly requesting direct URLs
separating source identification from link generation
manually stripping google.com/url wrappers when needed
Everything else is partial mitigation rather than a full fix.
This is peak post-AI discussion. Someone asks something about AI, then someone just tells them to ask the same AI about it. Why do we even bother having forums?
Claude is quite happy to help me brainstorm ways how to get the functionality of its more expensive versions without having to pay more.
> Why do we even bother having forums?
To provide learning texts for the AIs, I guess...
Reminds me of lmgtfy.
But in this case, I have been asking the AI, without success. I've added permanent instructions, asked it to triple check, asked it to never under any circumstance give me a google search link. It assures me its response has no search links, then proceeds to dump a list of 10 links, about 8 of which are search links.
I'm curious to see if the poster actually verified their solution or just assumed gemini was correct.
A take: socialism makes sense from that exact point when capitalism gets to the point when it resembles socialism a lot.
That means, corporations merge and grow, until they are huge bureaucratic monsters, every decision made by a committee selected by a committee, always making the safe protect-your-ass choice which is just applying Harvard Business textbooks, which means there is no real competition and no real room for any kind of a heroic entrepreneur anymore.
There are multiple different readings of Marx, because he seemed to be a bit conflicted about this, he was on one hand very clear that one cannot ever jump over stages of historical development, one must wait for the right time, but on the other hand seeing poverty and injustice everywhere he was tempted to hope for a revolution coming quickly. But that was mostly his emotions. His model of economics was absolutely about waiting for the conditions to ripen.
We are clearly not there. Lei Jun very much pulled the classic heroic entrepreneur move with personally saving Xiaomi, moving his desk into R&D, personally micromanaging the entire R&D and supply chain, and put $10Bn into designing an electric car. Very much Steve Jobs level entrepreneuring.
Very few things seem less intellectually productive to me than the way so many people insist on squeezing all of their economic thinking onto this capitalsm-socialism binary, as if those are the only two economic systems that have existed or could ever exist.
Marx had some decent insights, but the "stages of historical development" think is plain bullshit. There may be trends and attractors in history, but there's no guarantees, no set course that it will definitely follow and definitely no world spirit guiding it.
I agree. There are specific problems; there are specific solutions; there are specific problems these solutions introduce, etc. Any debate that does not look at these details is just cheering for "your team".
Sometimes a hero solves a problem. Sometimes a psychopath creates it or makes it much worse. Sometimes regulations prevent horrible things. Sometimes regulations prevent solving horrible things. Sometimes we need to incentivize those who produce desirable results, and punish those who make things worse. Sometimes we need to give people a second (or even the first) chance. Sometimes money is wasted or stolen. Sometimes the attempts to prevent the waste or theft waste even more money. Sometimes what we need is a new institution. Sometimes what we need is to leave people alone and let them organize their own lives. Etc.
Also, if we assume that, ideally, socialism is governed by workers' councils, and capitalism exists in a democracy where people can vote for their interests... seems like the fundamental dichotomy kinda disappears? I mean, if the workers' councils are free to adopt any "capitalist" pieces of the puzzle, and the voters in democracy are free adopt any "socialist" pieces of the puzzle, then in either case we are collectively free to build the entire puzzle as we want it, right? -- Unless we point out that in *real* socialism, the actual power is in the hands of the few individuals on the top of the Party / army / secret police; and in the *real* capitalism, the actual power is in the hands of lobbyists and media. But then it becomes a completely different kind of debate. Plus there is the question of how to implement anything resembling actual democracy if most voters are functionally retarded and will happily vote for whatever meme that today happens to be most successful on social networks.
I think it's the other way around: socialism (or rather, central planning, to be more precise), in the most optimistic case, makes sense to the extent corporations manage to reproduce it.
Amazon might be a bureaucratic monster, but whichever force keep it a dominant player (scale, internalization of spillovers, foresight, etc) clearly still make it more efficient than any competitor (including all the *potential* competitor that do not enter). So if you were to propose the state run a program with the scale and complexity of Amazon operations, at least in theory it would work (whether the incetives are there to make it work it's another matter).
But there never was a corporation that micromanaged the entire economy of a geographically-determined area (say, a region). Many have tried and it ended in disaster, or they made bank only because they could extract resources violently. So having the entire economy centrally planned must not work, otherwise you would have some perfect Town Inc. which people are eager to join.
Soviet central planning was state capitalism - socialism would be worker councils. I agree with one corporation running everything is a very bad idea.
Who said anything about the state running anything? We are already running businesses with collective intelligence. We don’t need anyone’s permission.
What we want to do is decentralize government itself. At least the power structure of it. It does not seem by your comments that we understand each other well.
Amazon gets special treatment from the government. So does much of Wall Street.
A different Amazon could be made with a completely different organization chart that has nothing to do with government.
Read the book Nobody is Smarter than Everybody if you want to see a $300 billion dollar organization operate this way.
> if you want to see a $300 billion dollar organization operate this way.
If I want to see an organization, wouldn't it make more sense to write the name of the organization, rather than the name of a book?
Stop debating Capitalism VS Socialism. Instead realize they are both failed systems and imagine building a new one. One that we don’t need permission to build. One that is transparent, decentralized, and power WITH the people instead of power OVER them. Like so: https://joshketry.substack.com/p/capitalism-vs-socialism-no-its-time?utm_campaign=post-expanded-share&utm_medium=post%20viewer
Well, what I care for is the concept that technological progress makes economic systems obsolete and then new ones are needed, one cannot stick to the same system forever. However the new system must flow from the logic of current technology, and not from conscious human effort. It should be something that just falls out from the current level of technology. If it is not called socialism but something else, fine. The only two rules are that it should be more humane, and it should follow automatically from the current level of technology, without much planning and tinkering. "gamification" sounds like too much tinkering.
Capitalism isn't a "failed system". If you insist on this, please define "failed system", and show why it applies to capitalism.
It is corrupted. The playing field is not fair. The iron boot of government is used to select winners. That is a successful system to you? Either you are lying to yourself, or your definition of successful is less than mediocre.
A good system should be allowed to be given to your enemies and they still can’t corrupt it. Our systems are fully corrupted.
A fully corrupted computer is a failed computer. Don’t trust, verify.
If that's your standard for "corrupted", then every system controlled by humans will become "corrupted". Capitalism is the best one we tried so far, and most importantly, regulated capitalism has been demonstrated to be compatible with democracy, in practice and on a large scale. No other economic system can say this. And if you claim to have found a new system which won't become "corrupted", but which has never been tried before, or an old system that has never been tried with more than a million people at once, then I'm very sceptical.
As you should be. But it’s not hard to imagine ways to make our systems harder to corrupt. Just because capitalism did a lot of good, doesn’t mean that it still is, or that there isn’t something much better out there as an option.
And yes, the standard should be that corruption is a failure and we should not tolerate it. A system that allows literal pedophiles to give sweet crony deals to each other should be heavily criticized. Our tolerance for corruption is the problem. So yes, we strongly believe that the success of system should be measured in corruption too. And harm done.
There are many systems that are antifragile. Many many examples we give in our articles. Your attitude of Nullius in Verba is a good attitude. But what happens when you don’t take our word and we show you that we are right with a new system.
Will you become an advocate then?
The free market right now is at serious risk of being turned into a violent digital control grid for humanity. Holding onto the corrupt system is going down with the titanic, friend.
The sooner we admit that the better chance we have of building a big escape raft for humanity.
It's not hard to *imagine* ways to make our systems harder to corrupt. It is very hard to actually *do* this, and disturbingly easy to screw that up and make things worse.
This would be a good time for you to explain in some detail exactly how you propose to make the system less corrupt. If the answer is just to give the government more power and tell them to use it to suppress corruption outside the government, I will be unimpressed.
Here's another way to look at it: corporations merge and grow until someone notices that their lumbering size makes them less efficient than a smaller competitor at solving whatever problem they were organized to solve, and proceeds to create that competitor. The solution to a herd of giant bureaucracy-hobbled corporations is to create smaller companies that take market share from the giants by providing customers with better services or goods.
If the problem customers observe is hidebound bureaucracy in corporations, then smaller, nimbler companies will have a natural advantage from the customers' perspective. Another elegant part of this solution is that if the customers turn out to prefer the larger corporation's product - for example, because it is more reliable due to all that bureaucracy - then the customers can continue to buy from the giant corporation.
The socialism alternative is to entreat an even larger and more bureaucratic government to oversee the giant corps, which looks a lot like trying to solve a rat problem by buying an even bigger rat.
Another alternative is for someone working at that giant company to notice the bureaucratic overhead, notice way to reform that overhead, and improve that giant company from within, leading to a more attractive product for consumers, and a larger profit for that company.
I think of this as the overlap between the socialist calculation problem and Coase. On one side, as your operation grows, trying to do central planning on it ends up requiring more information than you have or can get, and so you end up doing increasingly inefficient things even if all the decisionmakers in your organization are well-intentioned. (Which they are certainly not--see the principal-agent problem.) On the other side, trying to negotiate everything out via contract/on a market gets inefficient in a different direction, where you want to have (say) a shared office space but need to negotiate out who gets which space and what the rules for the shared space are. There's some kind of crossover (very dependent on local conditions and technology and such) where you stop getting more efficient by getting bigger and getting rid of the negotiation costs between different parties involved in (say) making a car, because now the inefficiency of the central planning starts costing you more.
Pretty close to how I think about it, too. You can fend off the principal-agent problem by aligning goals with natural incentives and keeping an eye on information flows, but as you hint, that gets harder to do as the project enlarges. Contracts have a similar problem where you can over- or under-manage, and again, can be staved off by picking people who naturally cooperate (one extent on which cultural diversity can actually hurt). If the staff generally prioritizes project over little details and nest-feathering, then the shared space contract is two paragraphs. If not, it's 30 pages and an extra $100K for the lawyers, wheeeee.
Hopefully, the big corporations don’t figure this out and cleverly buy up the small upstarts before they become big enough to pose a threat, that would be terrible!
/s
"Here's another way to look at it: corporations merge and grow until someone notices that their lumbering size makes them less efficient than a smaller competitor at solving whatever problem they were organized to solve, and proceeds to create that competitor. The solution to a herd of giant bureaucracy-hobbled corporations is to create smaller companies that take market share from the giants by providing customers with better services or goods."
... until the giants buy them out. Lovely profits for the founders, shame about the better services and goods, since they tend to disappear into the giant's internal systems.
There was at least one railroad builder who made an entire career of starting a railroad company, building a better railroad than the dominant competitor, selling it to that competitor, and then starting yet another company somewhere else, with the same model. So this is a thing that can happen.
To the extent it happens at all, the giant corporation takes a loss, unless the smaller company is deliberately selling for less than its value (which is possible if the owner is genuinely tired of that company or sees an even greater opportunity elsewhere and wishes to divest quickly, but those are exceptions, not rules). The giant may choose to bolster that loss with revenue from other locations, but if multiple small companies compete directly with the giant in multiple locations, the giant cannot bolster everywhere.
Thomas Sowell and David Friedman (a reader of this substack) discuss multiple examples in their books.
There is also a pattern of startup companies angling to get bought by some giant company to kill a competitor/hire on top-tier talent, which is a similar idea.
If those founders make lovely profits, shouldn't one expect more people to emulate them, until those inefficient giants can no longer afford to buy all the competitors?
That'd be nice if it happened, wouldn't it. I'll wait here for the next RISC V chip now that Meta has bought Rivos...
I was with you until “government”. Socialism is about worker’s councils running things, not the government running things. That was a Soviet travesty, something Marx explicitly opposed: https://en.wikipedia.org/wiki/Barracks_communism
If an institution is properly described as "running things", that institution is in fact "the government". If the institution runs only some things and it's clearly understood that it *won't* be running other things, then you may be talking about something less than a government, but it's on you to make it clear what is and what isn't being run by the workers' councils or whatever.
The actual history of socialism is, if you try to have worker's councils "run things" in that sort of open-ended way, you get very bad government. More generally, whenever you try to pretend that the institution which runs things isn't a government, you get very bad government. Something the anarcho-capitalists should be very wary of, but I digress.
> If the institution runs only some things and it's clearly understood that it *won't* be running other things
...then we also need some *other* institution that decides who runs what, and enforces that decision. And then the other institution is the government.
Yeah, you can have your economy organized around worker-owned collectives and guilds and such, but then you need someone to arbitrate between the collectives/guilds and enforce rules on them and decide what is a legitimate guild or collective and such, and that ends up looking a lot like a government.
One way you can escape some of that is to move more and more stuff outside the range of collective decisionmaking, the way Western countries have mostly moved the question of "which religion shall we be" outside of a collective decision to be made by the state or king or voters. But that's just getting to a place where, in some sphere, nobody is "running things."
> Socialism is about worker’s councils running things, not the government running things.
He who slays the dragon becomes the dragon.
There is no government policy preventing worker's councils. Worker coops are free to exist and there are many. When they provide better goods and services than their competitors, that's great. And when they don't, then customers are free to purchase goods and services from non-worker owned businesses. And that's also great.
Question. It seems to me "cancel culture" started with Lawrence Summers Harvard President, 2005, when he was talking about an innate difference of mathematical ability between women and men. I dug around a bit, and it seems Summers did very very questionable financial moves: https://www.institutionalinvestor.com/article/2btfpiwkwid6fq6qrokcg/home/how-harvard-lost-russia so it seems a lot of people had an actually good reason to get him fired. So in that light, how can that case be interpreted? Is such a talk a worse "crime" than these financial scandals, or more like it was the last straw that broke the camel's back?
Here's Gemini's summary of Summers' main point: "The underrepresentation of women in high-level science and engineering positions might be due to a combination of innate aptitude differences (specifically in variability of mathematical ability) and lifestyle choices, rather than solely discrimination."
Why is expressing these thoughts a crime at all, or even bad? What he's saying is quite plausible.
If you are the leader of a big institution, and you say to a large group of stakeholders that they are being discriminated against rationally, those stakeholders will not believe that you’re going to do anything about that discrimination and they will see you as an obstacle to their goals. There’s a difference between speaking as an economist and speaking as a university leader.
Who do you think should be allowed to make such an argument? I mean, if Summers was forbidden by his position to make it, then would a tenured professor be allowed to make such an argument? A tenure-track professor?
My strong suspicion is that there is nobody who would not have been attacked and cancelled (if possible) for making that argument. If that is true (ask Amy Wax and Ilya Shapiro why I think that), then it seems like the actual rule is "nobody may bring up this side of the argument" and then whenever someone does that, you search around for an excuse (like saying that James Damore's memo required firing him to avoid discriminating against women) to punish him.
That may be true, although the relationship between how willing someone is to be truthful and how much they are trusted is not simple. In some cases people give up forever on a person willing to speak unwelcome truths, in others people trust them more. Also, note that Summers did not deny there is discrimination against women, only that discrimination might not be the only reason why there are fewer women in STEM fields. Somebody making Summers' points could have restated that, and then continued with suggestions for ways simple discrimination could be reduced, such as excluding info about gender from grad school applications.
> excluding info about gender from grad school applications.
Everyone knows the consequences of implementing this obvious idea would be the opposite of what people loudly aver, so they've switched to denouncing this proposal as a particular insidious form of sexism/racism/whatever.
I'm reminded of Carl Sagan's story about the guy who claims to have a dragon in his garage. It is, of course, an invisible dragon, and when various tests are proposed to try to detect it, it also turns out to be perfectly silent, intangible, etc. The point is that even he knows he's lying, evidenced by how knows to head off every avenue that'd prove him right.
Well, then the speaker can vow to back enthusiastically any idea that the group proposes for making sure grad school admissions are done in a gender-blind way. Then the group can sit there, each contemplating the same obvious idea I proposed, realizing it will lead to fewer women being admitted, and stewing in their own juices.
I was at Harvard at the time – tl;dr, Summers was already intensely disliked by a lot of people at the university, and they leapt at the first opportunity to get rid of him.
I've been at 5 universities or colleges, and the guy running the place has been hated in every single one of them. It may be that's pretty much guaranteed to happen because college-age people are easily led to see an authority as a controlling fool, and career academics are backbiters.
I suspect this is a very common pattern in many of the high-profile cancellations. The stated reason for cancelling the person was often more of a pretext or a focal point for their enemies who already wanted to get rid of them. That wasn't always true, but it sure seemed like it often was.
JK Rowling got canceled even though she was beloved by almost everybody.
In what sense has JK Rowling been cancelled? She is still very much in the news, they're making a new Harry Potter TV series, and nobody's stopping her funding her current charity ventures. Is there something she wanted to do that she has been prevented from doing, and if so, by whom?
The Iran war has demonstrated that the balance of power in naval war has shifted to the land side. Cheap and numerous drones and missiles launched from the land can now control the sea to quite some distance away from the shore, denying entry to enemy ships. Launchers can be mobile or dug in, meaning that even with US-style air supremacy you can't bomb them all.
While this is bad news for the Middle East, is it not good news for Taiwan? How could a CCP invasion force possibly make it across the Taiwan Strait in any significant numbers if the Taiwanese can lob thousands of cheap drones at anything that floats?
I think a blockade is a lot more likely than a land invasion for exactly this reason.
The Persian Gulf is a very narrow, very shallow sea with extremely stable weather (and the Hormuz strait is even narrower and shallower). Basically the ideal conditions for such land-based hardware as you describe. Not sure you can draw very useful conclusions for the general case.
I'd say you could draw some very useful conclusions from the Ukrainian example. The Ukrainians were able to sink about a third of the Russian Black Sea Fleet and drive it back to their ports, using a combination of air drones, sea drones, and Neptune missiles. Soviet radar systems, which seemed to be good at dealing with a small number of incoming objects, were not up to dealing with Ukraine's mastery of radar systems (which most people seem unaware of) and fake-out tactics. While the Black Sea isn't comparable to the wide open oceans, it is about 1200 km across. Taiwan is only about 160 km from the mainland. Heck, Ukraine has steadily extended its drone kill zone. Last year it was about 50k, but their new models will extend the kill zone to 150 km. Given a small range extension, the Chinese coastal installations will likely be within Taiwan's aerial drone kill zone. And if they haven't already, they could easily recreate Ukraine's surface marine drone technology (such as Sea Baby and Magura). And sea drones are particularly effective in very choppy water because the chop masks their approach from radar.
Moreover, much of China's "cutting-edge" military technology is very likely a paper tiger. After US technology so thoroughly took out Chinese weapon systems in Venezuela and then Iran, several CEOs of China's major defense companies have been detained and are under investigation, with hundreds of managers, engineers, and accountants being questioned by their security services.
And now that Xi has decapitated the PLA by removing two of the PLA's senior-most generals, Zhang Youxia and Liu Zhenli, upsetting regional PLA commanders and the old-guard in the party, Xi would be foolish to try to mount an attack on Taiwan (of course, he might pull a Trump, but he seems smarter and more emotionally controlled than Trump). Also, Zhang became the focus of Xi's ire, because he had given assessments to senior party leaders that an invasion of Taiwan would be unwise. Currently, Xi is in a political standoff with the party elders and the military, and has been unable to replace Zhang and Liu with generals loyal to him (of which a few remain). And since Trump canceled his State visit, the Chinese state media has been talking about a peaceful solution to unification. No, we won't see an invasion of Taiwan anytime soon.
So I need to ask about the Lebron critique of prediction markets. From marginal revolution.
https://x.com/AgustinLebron3/status/2038645798863352001
It's a valid concern (and not just a theoretical one, under our current administration!), but I think it would be fixable with a law against government officials investing in prediction markets. Similar to the proposal to prevent congressional insider trading by not letting them hold anything but index funds.
While prediction market proponents often argue that insider trading is a good thing - it technically makes the market more accurate - I don't think that counts if you can literally control the outcome of the event you're betting on. The goal is to make the market match the odds of the event, not adjust the odds of the event to match the market.
Hmm I don't think it applies only to government officials. Anyone with enough money can swing the outcome. Isn't this already happening in sports betting? The $ value in the market creates it's own incentive.
SpaceX has supposedly made its confidential S-1 for an IPO filing.
https://www.bloomberg.com/news/articles/2026-04-01/spacex-is-said-to-file-confidentially-for-ipo-ahead-of-ai-rivals
On Kalshi, the actual public IPO announcement (not the IPO itself, just the announcement) only gives a 25% chance of that happening in the next 2 months.
https://kalshi.com/markets/kxipospacex/when-will-spacex-ipo/kxipospacex
Is this a normal gap of time between S-1 and public IPO announcement?
Hmm, I saw my barber catcalling some tourists in bikinis today. He made some kind of growling noises at them. Possibly the first time in my life I've seen catcalling in the wild. I feel this was ethically wrong, though in the moment the situation felt amusing to me. Should I switch barbershops over this? I was also thinking he's ultimately a low-IQ working class guy, so less is to be expected of him, but then, I know from personal experience being low-IQ and working class doesn't mean you have to lack a moral compass or something.
Do you like the guy? Do you have a good rapport with him?
If so, then a low-stakes “Man, why’d you do that?” in the moment followed by a conversation where you say stuff like “What, you think they liked it? It probably just made them feel gross.” can probably do more good for his attitude than quietly switching barbers ever could.
(Bonus points if you can code-switch into more working-class style language without it feeling forced.)
Do whatever you want tbh. Feels more like a question of class and taste. Like if your barber farted while cutting your hair you might just get the ick and not want to go back there. But I personally don't see any serious moral dimension to it, and even if there were how would leaving (and without even explaining why) help anyone in any way?
I live in a predominantly hispanic area, and when I socialize with working class guys I just kinda know to code switch into being a bit crass about women especially compared to super liberal intelligentsia (though, never to the point of catcalling). I also don't think the latter are especially more moral, they just are more performative and careful to not commit wrongspeak
I don't think it is immoral or unethical, I think it is against etiquette, against good manners, basically rude, which is exactly what low-IQ working class types do not care about.
I mean I find it very strange that in the last maybe 15 years the question of etiquette simply disappeared and rudeness is now considered immoral. This is basically the "therapeutic culture" belief that unkind words cause some kind of a damage.
I wanna say on a list of professions most likely to catcall, barber would probably be third.
Construction workers probably have a pretty solid hold on first.
What's your take on wearing bikinis at the barbershop?
In a 1946 fashion show held at a Paris swimming pool automotive engineer turned fashion designer Louis Reard - how’s that for a career 180? - introduced the bikini to the world. As this show was four days after the highly publicized US atomic bomb test at Bikini Atoll, and he had given the swimsuit its name at the last moment, an early urban legend held that the name was due to the fact that the test had blown out the middle of the atoll while leaving the ends intact (get it?)
In fact, Louis Reard had simply liked the sound of the name, and in any event the blast did not affect Bikini Atoll’s geography in that way.
Sidenote 1: Several professional models refused to model the (yet unnamed) bikini due to its immodesty. Reard ended up hiring Micheline Bernardini, an 18-year-old cabaret dancer who didn’t mind wearing one because in her occupation she customarily wore nothing but a smile. She is still around at age 98.
Sidenote 2: the military figured the Bikini Atoll test would be a good way to get rid of a number of warships that the end of World War II had made redundant. As things turned out, fewer ships than expected ended up sinking, and disposing of the highly radioactive ones than remained was quite a chore.
The version I'd heard was that it was named for its effect on the male psyche.
You must not be a software developer, as that is quite the assumption you made. They were walking down the street, this area is right next to the beach, seeing women in bikinis is not an unusual sight. Barber was sitting in a doorway, I was heading home.
Oh. In Mediterrean Europe, it is generally considered not appropriate to wear a bikini outside of the beach. In Greece it is a finable offense: https://www.mirror.co.uk/travel/europe/brits-greece-warned-strict-dress-35359329
How far away from the beach is no longer near the beach?
Is this going to be an "If they don't want to be catcalled, they shouldn't dress like that" argument?
It's "are the women actually upset". There's no description given to their reaction, to the point where it's not clear if they even noticed this guy. Eremolalos already commented on the purpose of a bikini.
Re "the purpose of a bikini": Even when we assume that those women wore bikinis to draw attention to their bodies, it doesn't follow that they want to or are comfortable with being catcalled.
Re "are the women actually upset": How would you know that from "how far away from the beach" they were? Or even from their visible reaction? If I were a woman and annoyed by catcalling, I'd do my best to not show a reaction.
Using your economic power to coerce the lower-classes into conforming to your sense of morality is one of the many perks of being rich. It's the same principle as lecturing panhandlers about sobriety and fiscal responsibility, or missionaries requiring pagans to convert before giving them food or medicine.
I'm not lecturing him (don't see the use of that), but I'm certainly not obligated to give him my business.
As unkindnesses and intrusions go, it's tiny. For instance I expect a lonesome person who attempts a bit of friendly chat about the weather with a cashier and gets ignored feels much more distress than the women in bikinis did. If you hold miscellaneous strangers to this high a standard you'll end up shut up at home. What the barber did is just a very politically incorrect instance of unkindness. Why not say something that might make the guy slightly less likely to catcall? Something like "my daughter says girls think men who catcall are losers. I dunno though -- do you think that's true?"
Haha, I feel if I actually confronted him about this and held my ground, I could end up in a brawl, which is not something I want. I'm probably gonna switch barber's over this, Nobody Special's right in that it's better to give my money to decent people.
All of this seems altogether wrong-headed to me. First of all catcalling women in a bikini seems like kind of an edge case. If a man catcalls a woman in ordinary street wear, he truly is being rude and intrusive. Bikinis, though are *meant* to be hot and edgy and attention-grabbing. That’s the whole point of them. Nobody wears them because they are more comfortable than ordinary swim suits (they’re actually less comfortable, because they have to be tight to ensure that they do not slip half an inch and uncover a bit of nipple or labia). They’re worn to be hot For a man to catcall a bikinied woman seems more like playing along with her than like shoving their uninvited horniness at some woman who’s peacefully going about her day. And I doubt that many women in bikinis are startled and offended by comments, whistles, catcalls etc.
Second of all, even catcalling a woman in street dress is pretty low on the rudeness and unkindness scale of real life. You see multiple examples of worse rudeness and unkindness in 5 mins in any public place. When you’re driving you see people prioritizing their speed and convenience over other people’s safety and peace of mind in at least a smallish way almost constantly, and seriously risky versions of this are not rare at all. In all kinds of settings you see solitary frail elderly people struggling to lift things, move things, reach things, and a stream of the able bodied striding right past, chattering on their iphones as they go. And then there’s the news.
And third, your response to the barber’s catcalls is a non-event in ethical space. Finding a new barber does nothing whatever to reduce the chance the original barber will continue doing something that, in your eyes at least, is so bad it takes him out of the category of decent people. It probably isn’t even sparing you much distress, since it’s a rare event —apparently he hasn’t catcalled anybody before in your presende. All it’s doing is giving a bogus little boost to your sense of righteousness.
My barber is a former golden gloves boxer. Still pretty sure if he did something like that I would still have given him a sharp, “Hey, don’t be a bonehead!”
Probably a dramatically different cultural context. I live in Puerto Rico.
How about mildly saying “Eso simplemente no está bien.” You would get the message across and I doubt he’d get bent out of shape. He might think you’re a dork or something but who cares? You stood your ground and can wear the dork badge with your own values intact.
Ultimately your call, but for me I'd put a lot on whether whatever noise he made was audible only to himself, or whether it was audible to or affected the tourists in question.
I'd generally be less concerned with something he said, or a joke he made at a tourist's expense, if it was directed to/only heard by himself or people at the barbershop, than I would be if he was audibly hassling someone whose only crime was walking by him. If people are minding their own business, you should mind yours as well - "behaves like a jackass to strangers at random" isn't the kind of behavioral trait that makes me think I should be handing my money to its possessor, when there are plenty of decent people in the world who could use that money just as much.
> there are plenty of decent people in the world who could use that money just as much.
Good take, yeah, I should probably switch.
Kind of a personal question, but do you consume porn in any way?
Yeah, but I recently resolved to stop. Why?
I wish you luck! It can be a really difficult thing to quit fully. I believe your life and society will be better for it, though.
I ask because if you do or have, chances are high that you've supported human sex trafficking to some degree without knowing it[1]. It's regrettable that this applies to most people, including myself, and it's worth bringing up. But I just wanted to point out that maybe a person shouldn't be so quick to judge other people, I'm sure he's doing his best.
If you do choose to go somewhere else, I think the kinder thing to do would be to bring up the issue without letting it be an argument, and then let him know that you don't feel right supporting someone with his morals, instead of just leaving him to keep being the way he is without anyone bringing it up to him. Then at least you will have made some effort in making the world a better place whether or not anything comes of it, instead of just ignoring something you take issue with. But chances are, whoever you go to instead will be supporting sex trafficking in their own way and you won't go to very much effort to find out or take your money to someone more "decent".
[1] - https://fightthenewdrug.org/how-porn-can-fuel-sex-trafficking/
> chances are high that you've supported human sex trafficking to some degree without knowing it
Is this true even if you never paid for anything, and have ad-blockers installed?
(What a weird opportunity to make a moral case for ad-blockers.)
Huh, interesting. I only ever watched amateur porn, or hentai. Always felt pro-porn looked disgusting.
As long as we're talking ethical misdeeds, I feel factory farming is really the worst ethical wrong of our time, and all meat eaters (me included) are contributing to that one (though I do donate to animal welfare charities).
Interesting idea raised at Marginal Revolution: when a judge lets off a violent criminal with a much too lenient punishment, and the criminal goes on to commit another violent crime, should the *judge* be prosecuted? I’d say in some extreme cases this might be warranted, though I also believe that jurors and parole board members shouldn’t be exempt either.
Sure, as long as we're comfortable with making it symmetric. If judges (and optionally juries and parole board members) impose harsh sentences on people later found to be innocent, we'd better inflict similar punishments on them. This is not a reductio ad abusurdum, this is honestly what I think would be the minimum necessary criteria to make your (and MR's) idea non-disastrous.
To be honest, this is so frustratingly basic and obvious that I have trouble believing you and Marginal Revolution are taking this seriously, and not just saying provocative shit for signalling reasons. If you put a person working a necessary job that naturally involves weighing Type I errors against Type II errors and stick a only-barely-metaphorical gun to their head saying "if we ever catch you making a Type I error, we'll kill you," then literally everybody doing that job is only ever going to make Type II errors from then on out, and at a vastly increased rate. Obviously. The only way not to get this outcome is to also extend the threat to apply to Type II errors as well.
To be blunt, I find the fact that someone who pretends to Libertarianism would so casually suggest something so utterly destructive to the very foundation of civil liberties to be utterly contemptible. The most charitable explanation I can think of is that Marginal Revolution is simply inviting outrage to generate traffic now: if they actually believe this to be a good idea, that would say far less flattering things about them.
It is possible to set guidelines that will minimize the risk of harsh sentences for the innocent or those not deserving of such. Try this: if a judge (a) gives a sentence to a defendant convicted of a violent felony that is in the nationwide bottom 10% of sentences given for similar offenses, and (b) the defendant has three or more prior convictions for violent felonies within the past 10 years, and (c) within one year after being released from prison for this latest conviction the defendant goes on to commit another crime of violence, then the judge can be subject to prosecution.
I have trouble understanding the above as anything but "what if sentencing laws, but strictly worse?" If you believe you have a set of criteria that strongly indicate a certain minimum sentence, the sane and reasonable way to handle that is to write those criteria into sentencing laws and NOT give judges latitude to violate them.
If, on the other hand, you recognize the circumstances around some class of crimes are likely to be sufficiently nuanced and complex that allowing significant exercise of human judgement[1] on a case-by-case basis is warranted, then the sane and reasonable way to handle that is to ACTUALLY ALLOW the exercise of human judgement. Not to pre-emptively insist that the judge *must* rule in the way that you've plainly already decided is correct, on pain of risking their liberty for zero personal or professional upside.
Again, this whole argument seems thoroughly asinine. The ENTIRE POINT of having a written code of laws is to inform people of what behaviors are punishable by the state *before they engage in them.* Writing your laws such that a public official may take actions that are perfectly legal at the time they are taking them and only become prosecutable *after the fact* is either dangerously insane or a blatant sign that you view laws largely as a weapon to be wielded against your political enemies, rather than a tool for ordering a society. Indeed, encouraging certain people[2] to fantasize about wielding the force state against their political enemies seems to be about the only comprehensible reason I can conceive for anyone suggesting things like this at all. As actual suggestions for a real legal framework, they are beyond farcical.
[1] Which of course always carries with it the possibility of error.
[2] Whose political persuasions I'll decline to speculate on.
It's probably much easier to remove them from the bench than to prosecute them. I'd be cool with some sort of law that measures how often a judge under-sentences someone who then goes on to commit another crime, and if the judge's record goes below some threshold then they automatically lose their job.
I wonder why is the blame directed towards the police so rarely... It is (almost) always the defence attorneys or the judge, even if their role is to show that the police failed to perform their job well.
There's a much stronger moral case for imprisoning judges who grossly *over* punish. A lenient punishment leading to a repeat crime is a possibility- it represents a factual mistake. On the other hand, a strict punishment leading to unwarranted imprisonment/or unwarranted extra time in imprisonment is a certainity, so there is no mistake of fact here. For example, I'm sometimes sympathetic to the idea of prosecuting, say, the judge who sent 10 year old Joseph Hall to prison for 13 years for shooting his abusive father.
On the whole I don't support this, though I do find the position tempting. Where a punishment is grossly and manifestly excessive (as in Joseph's case) The only argument against it the ole' princple: Nulla poena sine lege. But I've always found this a fairly weak defence when it comes to state agents. Of course the atrocities of state agents are usually authorised by law! To say that this means they should never be prosecuted seems like an unwarranted shield for the afflictor and persecuter.
So on the whole I don't favour it, but these matters are finely balanced.
I have a relatively unique position on this, but I think a defensible one. First of all, we have to get rid of the myths, like prisons don't actually "correct" anyone, and they do not serve any kind of cosmic justice. Thus their only use is simply separating dangerous people from potential victims. So they are not about punishment, but about harm-prevention. In that case, much more humane systems should be invented - basically just walled towns with normal jobs, entertainment and so on. But in that case, because those do minimal harm to convicts, far more people should be "incarcerated" in them. It seems like at least 10% of the male population now is too dangerous to be at liberty.
If we put dangerous people in just a walled town, wouldn't that be a lot *less* humane than a prison, because they'd attack each other?
Specifically, in case of capital punishment of an innocent person, I think it would be logically consistent to sue everyone involved for a conspiracy to commit murder.
This idea was referenced. The proposal actually endorsed was financial incentives (in both directions) I think if this was done close to revenue neutral risk aversion would lead judges to just never release. Theyd need a pretty big payout on non recidivism to take the risk. I wonder then to what extent it affects sentencing also. Stock the prisons, so we can get paid to release them sort of thing.
If an economist really wants to champion an idea like this. They should look at what performance incentives actually look like on the private market and model off those.
What we need is a bunch of meta-judges. Plus some people to make sure *they're* fair. And also some . . .
No, they shouldn't, because they're not clairvoyants and the people who would replace them are likewise not clairvoyants. The punishment for judges with poor judgment is to remove them from the bench.
Given that recidivism is like 1% or lower for many types of crimes with long sentences, how about we arrest judges for unlawful imprisonment whenever they send someone to jail who represented no credible threat to society, ie every day?
No, we should not make the largest carceral state in the civilized world even more punitive and inhumane.
> Given that recidivism is like 1% or lower for many types of crimes with long sentences
You need some citations for this, recidivism is basically ~33% at 2 years, even in much less punitive, more reform based societies like the Nords. And it's the dominant outcome at 5 years.
https://imgur.com/a/xOn3JOz
Without any corresponding pressure on the other side, I'd simply vote "guilty" if I were on a jury, regardless of the fact of the case.
If I was to be held responsible, I'd do anything I could to get out of jury duty.
US-Iran war explained by Chinese AI animation: Legend of the Valley of Gold: https://www.youtube.com/watch?v=As0rplNJTZI
There's also an Episode 2, much shorter: https://www.reddit.com/r/singularity/comments/1s9gh9g/chinese_state_media_releases_episode_2_of_their/
I think this one was the first: https://x.com/Mollyploofkins/status/2037707588381331586. It's not as good.
The most dangerous aspect of AI *right now* is the data center bubble. I would like to see someone propose a pause on *that*. Tax them hard.
Reminiscent of Disney's version of the Hunchback of Notre Dame, the Met's new production has added a happy ending to Tristan and Isolde.
Next let's give them The Human Centipede to work on transforming.
My old college roommate and friend is on the faculty of Yale Law School, and so I ended up getting notices about their journal. Here's what they're featuring right now:
"Geduldig's Footnote Twenty and the Past, Present, and Future of Sex Equality - Yale Law Journal
For helpful comments and insights, I thank Katie Eyer, Melissa Murray, Kate Redburn, Stephen Lee, Ari Waldman, Douglas NeJaime, Reva Siegel, Jessica ..."
Academics. Liberals. Geduldig's Footnote Twenty. Seems like madness to me.
Maybe the title is meant to be "Gedul Digs Footnote Twenty"? A bit more relatable, you know?
For context, the case Geduldig v. Aiello (1974) held that California's disability insurance excluding conditions arising from pregnancy was sex-based discrimination, and therefore not violate the Fourteenth Amendment.
Footnote 20 in full:
The dissenting opinion to the contrary, this case is thus a far cry from cases like Reed v. Reed, 404 U. S. 71 (1971), and Frontiero v. Richardson, 411 U. S. 677 (1973), involving discrimination based upon gender as such. The California insurance program does not exclude anyone from benefit eligibility because of gender, but merely removes one physical condition -- pregnancy -- from the list of compensable disabilities. While it is true that only women can become pregnant, it does not follow that every legislative classification concerning pregnancy is a sex-based classification like those considered in Reed, supra, and Frontiero, supra. Normal pregnancy is an objectively identifiable physical condition with unique characteristics. Absent a showing that distinctions involving pregnancy are mere pretexts designed to effect an invidious discrimination against the members of one sex or the other, lawmakers are constitutionally free to include or exclude pregnancy from the coverage of legislation such as this on any reasonable basis, just as with respect to any other physical condition.
The lack of identity between the excluded disability and gender as such under this insurance program becomes clear upon the most cursory analysis. The program divides potential recipients into two groups -- pregnant women and nonpregnant persons. While the first group is exclusively female, the second includes members of both sexes. The fiscal and actuarial benefits of the program thus accrue to members of both sexes.
Do you mind explaining your criticism a little more?
As an amateur hobbyist in US constitutional law, this article seems worthwhile from two perspectives:
(1) What is considered strong argumentation in the current SCOTUS? For example, "Moreover, in his separate concurrence, Justice Alito—Dobbs’s author—suggested that Geduldig’s precedential status was so obvious and powerful as to obviate the need for explanation, despite the fact that the Court had never favorably cited to Geduldig in a constitutional decision before Dobbs." (page 6 of the pdf, labelled page 1700)
However, personally, I think it is more the case that the court in Dobbs cut the reasoning to fit the desired outcome rather than vice versa (Geduldig needs to be followed as precedent, but Roe/Casey don't???)
(2) What are the substantive implications of the use of the Geduldig footnote reasoning?
The on-the-ground implications from Dobbs and Skrmetti are significant.
Lastly, there is a baked-in pun to have a minor 50+ year old side comment jump to prominence ('geduldig' is German for 'patience.')
>Do you mind explaining your criticism a little more?
Actually I think my post is indefensible. When I put it up I knew nothing about Geduldig and related matters. Absent that information I found an article about a footnote, presented with deep seriousness, complete with thanks to the multiple scholars who had assisted, hilarious.
I'm sure that part of my reaction to it is an offshoot of my sadness and frustration about how my old friend's career has affected her. She lives in a world of constant competition, one-upsmanship and political intrigue, all centering around things that seem to me to be obscure points, and participates wholeheartedly in the intellectual-egomaniacal mosh pit -- all of which is of course irrelevant to everyone else here.
Article appears to be here if anyone's curious.
https://yalelawjournal.org/pdf/01KN0PZ4Y6DXZDX3SMWERCB88M.pdf
"Geduldig" is a supreme court case, and they are commenting on a footnote in it.
I am missing something, what's notable about this?
Last summer I posted my first album co-created with AI and got some helpful feedback on this thread. (Carbon Dreams by Sus Houseplant on YouTube and Spotify).
Someone went as far as to say some of the songs became their favorites.
Today I’d like to share my first song / music video. The genre is less synthwave industrial and more singer songwriter country.
Based on an event that my dad shared one evening.
https://youtu.be/RFhS5k9qXGI?si=STa9EW3xXAePRdTl
I had a lot of fun making it, and plan to make more. Thankful for all feedback.
I am not a fan of the country genre you are targeting, so that might color my criticisms.
Compared with Carbon Dreams, I think this video sits more in the uncanny valley. The CD tracks I particularly like are absurd with nuggets that are accidentally profound ('tell me again how nonsense feels') or euphonious ('packet of silvery lemony lime.') In Scorpion's Lullaby, the combo of lyrics and video seems closer to an all-human product, but just doesn't quite make sense or feel right. I think adding video massively increases the number of opportunities for small mistakes that feel very irritating.
FWIW, there are plenty of all-human pieces that also have little gaps that bother me, so maybe I am not reacting to the AI process.
Thank you Joshua!
Agree that video presents many more opportunities for a slip up.
Also, lyrics that leave more to the imagination certainly have a benefit.
It would be fun to do the same video project with a few of the songs from Carbon Dreams, much easier now then it was when that came out.
I also like lyrics that do not exactly make sense: "water flowing underground, into the blue again after the money's gone, once in a lifetime . . ."
Lime and limpid green, the second scene, the fight between the blue you once knew.
"same as it ever was, same as it ever was..."
I trust you are also a fan of TMBG.
My 10 free gift articles per month resets end of day. This Ezra Kline interview of Michael Pollan discussing consciousness is pretty interesting. They get into William James’ thoughts on the matter early on.
https://www.nytimes.com/2026/03/31/opinion/eza-klein-podcast-michael-pollan.html?unlocked_article_code=1.XVA.ApgX.hEQUBUfXuheh&smid=nytcore-ios-share
It's also available on Youtube.
I've recently been reading "The Space Shuttle Decision", about the political and design history of the Space Shuttle program. Early on, the Shuttle concept had been just one piece of a larger system of space infrastructure that had been proposed c. 1970. I knew about the space station, where there were a mix of ideas. One was converted S-IVb upper stage launched by a Saturn rocket, either a "Wet Lab" where it would be used as the upper stage of a Saturn Ib and then converted to habitable space in orbit or a "Dry Lab" that would be built and furnished on the ground and placed in orbit by the lower two stages of a Saturn V. The other would be that the station would be built in orbit out of a bunch of modules lifted by the Shuttle. Two of these would eventually happen, of course: the Dry Lab became Skylab (which was used for three Apollo missions and then deorbited because the Shuttle wasn't ready to re-boost it) and much later the modular station became the ISS. In the meantime, the Shuttle itself was used to lift and support single-use "Spacelab" modules for 1-2 week missions, and also a ton of the planned missions for the Space Station were instead done by unmanned satellites as improved electronics made it feasible to e.g. control telescopes and weather satellites from the ground instead of needing astronauts to calibrate instruments, take readings, and manually change the film like they did for the Apollo Telescope Mount that was built into Skylab.
One bit that I didn't know about was the third key part of the early space infrastructure proposal, the Nuclear Shuttle. This would be a durable spacecraft powered by a NERVA engine and living in orbit. The Space Shuttle would take astronauts and payloads to the Space Station in LEO, and then the Nuclear Shuttle would boost any payloads that needed to go much beyond LEO. Post-Apollo missions to the Moon and interplanetary missions would then be staged in Lunar orbit, lifted from LEO to Lunar orbit by the Nuclear Shuttle.
The Nuclear Shuttle was killed fairly early on by budget cuts, since there was definitely no money for the for most of the missions it was intended to support, and NASA administrators were only barely able to persuade the Nixon administration and Congress to not also kill the Shuttle and the Space Station outright. The Shuttle we got was a lot less ambitious than most of the early proposals, for both technical and budget reasons (the book emphasizes the budget reasons, but reading between the lines also suggests a ton of engineering risk), and the post-Skylab Space Station program got postponed until the 90s. Budget cuts also killed the NERVA R&D project at a fairly late stage in development, so there wouldn't have been an engine for the Nuclear Shuttle even if there had been money for it and its missions.
So, while the Space Station idea got resurrected in the 90s, and the idea of a Lunar Station for staging Lunar Landings and interplanetary missions has come back in recent years with the Artemis program (although the Lunar Gateway station seems to be mostly dead as of recent weeks, in favor of a surface Moon Base). But if there have been any proposals for resurrecting the Nuclear Shuttle or anything like it, I haven't heard about it, and I am curious about why since a reusable nuclear thermal rocket shuttling between LEO and LTO or LLO seems like it would make it much, much cheaper to build and support a Lunar station or surface base and especially to supply a propellant depot in some Lunar orbit or a Lunar-adjacent transfer orbit.
Are there major technical problems with a Nuclear Shuttle that I'm not aware of, or is this just some combo of 1) reducing medium-term costs and scheduling risks involved in dusting off NERVA or some appropriate replacement and developing the Nuclear Shuttle, 2) "Nuclear" anything sounds scary, and 3) Political constraints related to the observation that Artemis seems to exist for the purpose of giving SLS something to do?
I can add to the issues John Schilling pointed out about nuclear rockets, in that another hugely limiting feature of their usefulness is the quite limited operating lifetimes. Most nuclear thermal rockets, like the US NERVA, have service lives of just a few hours!! This is also compounded by their low thrust to weight.
Compare the NERVA to a SpaceX Raptor:
-NERVA: 0.25 MN thrust, 18 tons dry mass, 380s ISP
-Raptor 3: 2.8 MN thrust, 1.5 tons dry mass, 841s ISP
Even though the NERVA is more than twice as efficient, you wind up burning for a much longer time to make up for the lower TWR. As a result you use up your (hideously expensive weapons-grade enriched!) nuclear engine's limited lifespan very quickly, and nobody has ever considered trying to refuel in space or return spent nuclear rockets to Earth, so it is totally disposable.
The limited lifespans of nuclear thermal thrusters is a huge issue and, for many Mars mission architectures, is one of the overriding design constraints. Mars return missions sit right at the edge of what's practically possible in terms of total burn time even with multiple nuclear thrusters or even staged designs. This is why basically all the crewed Mars missions have the crew "bail out" and aerobrake in a capsule to land back on Earth as their depleted spacecraft flies past the Earth into solar orbit!
Another problem with reusability is that every mission which returns from the moon, for example, has so far used aerobraking to land back on the earth at the end. This saves us almost 3 km/s of delta V on the return leg which would otherwise be needed to propulsively brake into LEO. If you have a reusable nuclear tug which can't be allowed to reenter, then you need to carry all of the reaction mass required for that final braking burn throughout your entire trip!
So you wind up carrying much less payload, or requiring a huge craft, and by the time you've braked your reusable tug back into LEO after your lunar trip your fancy expensive NERVA is completely spent anyways and is now an orbiting radiological hazard for aeons.
Regarding nuclear-thermal rockets, they're definitely useful but not nearly so much as a naive performance comparison would suggest. The issue is that nuclear-thermal rockets really only work well enough to outperform chemical rocket engines if they use liquid hydrogen as their reaction mass. And liquid hydrogen is a terrible, terrible propellant to handle. It has to be kept exceedingly cold if it isn't to boil off, and it has a density slightly greater than styrofoam. Which means big enormous tanks, whose vast surface must all be well-insulated if you're not going to use the stuff immediately, and overpowered pumps to force it into an engine. And it really really likes to leak.
If you look at a Space Shuttle, you'll see the compromise that is the ginormous external fuel tank - and that's for a vehicle whose propellant is (by mass) mostly nice, dense, well-behaved liquid oxygen and only about 15% liquid hydrogen. A nuclear-thermal shuttle would need only half as much propellant, but it would *all* be liquid hydrogen.
For use in deep space, those disadvantages aren't quite as critical. But it's still enough to downgrade nuclear-thermal from a must-have to a nice-to-have. Maybe 20% improvement in actual mission performance, rather than doubled performance. And that's a particularly expensive technology to develop, for a marginal performance gain.
The difficulties of working with liquid hydrogen were not well understood when work on nuclear-thermal rockets began in the early 1960s. The experience of Apollo, which used some liquid hydrogen in its upper stages, gave people an inkling of what was to come, not enough for them to choose a more sensible propellant for the Shuttle, but enough to take some of the shine off nuclear-thermal.
Also, in the 1960s the expense of the technology development program was mitigated by the ability to say "yeah, it's an experimental nuclear reactor, and because it's also a rocket if it melts down it will spray the radioactive slag across the surrounding landscape" but immediately add "...which is in Nevada so nobody cares, this is wicked cool shit, let's get on with it already!"
That's obviously not going to fly today, which means budgeting something in the eleven-figure range for a test facility that can contain the fallout from a nuclear rocket oopsie.
Thank you, "you need to use liquid hydrogen as reaction mass to get a big advantage over a good chemical booster" is definitely a huge disadvantage. I'm familiar in general terms with how much of a headache hydrogen was for the Shuttle and continues to be for SLS, and I've given to understand that's a big part of why SpaceX and Blue Origin have been working on methalox engines instead. I've come across stuff about NERVA using liquid hydrogen, but I though I'd also seen stuff about ammonia or water as reaction mass and hadn't realized those weren't viable if you want high specific impulses.
I'm guessing nuclear-electric doesn't give you enough thrust to be worthwhile for translunar injection on a large payload, at least not in a form that's anywhere near being ready to use?
Also, I expect you need quite a bit more than a 20% improvement in efficiency for reuse to be worthwhile even for going to lujar orbit and back, since you need to bring enough fuel to come back and also extra fuel to bring the return trip fuel with you. If you aren't saving much mass to orbit, you're just saving the cost of the engines, it looks like the RL-10 engines on a Centaur V upper stage aren't super cheap, but they also don't sound like they're expensive enough to move the needle on the cost of an Artemis mission.
Right; if there's a use case for nuclear-thermal, it's probably when you're going to Mars or beyond. And even there, it's competing with nuclear-electric and even solar-electric, probably used in parallel with old-fashioned chemical rocketry.
Or possibly for highly kinetic warfighting in Earth orbit or cislunar space, but that's speculative in ways I'm not going to touch here.
Another relative disadvantage for work far beyond Earth is, where do you get the propellant? Liquid hydrogen, or even gaseous hydrogen that we could liquify, is hard to come by until you get *way* out in the solar system. Water, is much easier to find, and you can extract hydrogen from it. But in the process you'll also be extracting eight times as much oxygen. Way more than you'll need for life support, so you'll want to find *something* to do with it. And in most cases, I'm pretty sure you're going to find that using that oxygen along with the hydrogen (and maybe some carbon) in a boring old chemical rocket, will give better overall performance than throwing out the oxygen and running just the hydrogen through a nuclear rocket.
I can’t answer the rocket science question but I’ll throw in that I recall the July day that Sky Lab deorbited. I had brought a transistor radio to work to learn where it would come to earth.
I piloted a much less sophisticated vehicle, a diesel compressor up to the hill top where we ripped and rebuilt rail sections that day. The drive was pneumatic powered by the compressed air in the 200-gallon air tank. I would joke about ‘silent running’ mode when I moved the contraption 50’ or so on the air in reserve.
I never did really listen to the radio though. The work had a Zen-like rhythm that I found soothing and satisfying. Tap, tap, tap, half swing to set the spike an inch or two into a tie, nail every fourth tie home, single spike inside the rail with another spike set, a single spike driven home outside.
Then start the compressor and take turns with my partner driving home the spikes that were just set in place with an air hammer. I’d go up one side while my partner held the tie against the rail with an iron bar with a Little Bo Peep hook on the end. At the end we would swap positions on the way back and I would catch my breath.
In its way it was a glorious way to spend a summer day for a healthy young man.
Oh yeah, Sky Lab came down sometime that day and as expected no piece was big enough to cause any damage.
>Oh yeah, Sky Lab came down sometime that day and as expected no piece was big enough to cause any damage.
I had thought this wasn't quite true, with one piece of debris killing a cow in Australia. But I looked it up just now, and it seems that you were right: the story about the cow is widely circulated but not accurate. It's a garbling together of two separate incidents: the debris from Skylab did fall in rural Australia and the first larger pieces of it were found in a cattle pasture, but none of them hit any people, livestock, or structures. There was an incident with falling American space debris killing a cow, but much earlier (November 1960) from a failed satellite launch (Transit 3A and SOLRAD 2 / GRAB 1). The rocket shut down early during launch and the range safety officer detonated it, with the debris falling across parts of Cuba and killing a cow. The US paid Cuba $2MM in compensation, Cuba salvaged the debris and sold it off to the Soviet Union and China for study, and presumably the incident contributed to the decision to launch things into polar orbits from Vandenberg instead of Cape Canaveral, as Vandeberg has no land directly South of it until you get to Antarctica.
There had been some speculation about the possibility of damage leading up to the deorbit but things worked out well.
I noticed someone posted a daily word game here and thought I'd drop my fangame of a very fun concept of magnifying images as a game, and added some of my own magnifier types in here:
https://icely.itch.io/toad-magnifier
The code for Claude Code has leaked through what appears to be a rookie error born of vibe-coding. The weights are obviously still secure, but this is probably going to be good for progress in other coding harnesses.
The engineers are just supervisors. That’s the rhetoric. Just not very good ones, it seems, and as skills atrophy this can only get worse.
What's the state of AI usage in the real world?
I keep hearing about free and cheap AI making stupid but plausible mistakes, and I will continue to mock AI that's a game of you bet your law license.
However, I also hear about the AI that costs money ($200/month? more for even better?) which is strikingly good, though it might still need skilled human supervision.
And there are companies which are buying into AI, and insisting that employees use it even if they hate it, which doesn't sound like a good tool to me. Are they, or some of them, buying the good stuff?
How expensive would it be to produce good AI for the general public? Are companies losing money on the AI they produce?
There was a video about how data centers are being built, but the computer chips haven't even been installed yet. Is this even remotely true?
I think there's already a *lot* of AI (by which I mean the modern, LLM-and-adjacent-technology stuff) usage, but it's not always what people think of and not always super visible.
A lot of people talk about "AI usage" in terms of "I navigate to ChatGPT and ask it a question and it produces an answer" (and there *is* a lot of that), but I think a lot of it is in other forms: a ton of medical providers are adopting transcribing tools that automatically take notes with interactions with patients.
(I've been doing Physical Therapy and had an extensive discussion about this - it apparently records the conversation, filters out stuff that's just side-chit-chat, categorizes the rest of it into the different fields that they're supposed to fill out, and supposedly does a better job providing the right details to get insurance companies to support continued treatment)
And from my day job, I know a lot of e.g. call/contact centers are adopting it, too. You call in and even if you're still talking to a human (though who knows how long that'll last), they're increasingly working with information that's curated and prompted by an LLM that's listening to the conversation.
These are both primary examples of LLM-based *understanding* rather than generation, and I think that is the 'easier' side of the coin to leverage right now, but we're absolutely seeing usages of the generation side of the coin too. Though given the backlash, a lot of places aren't advertising it.
(I see a very CGI-like dynamic where everyone "hates" it, but it's really just the reality of how stuff gets done and the stuff people 'hate' is the obvious bad stuff and not the entire rest of the iceberg, so companies make big shows of being "no CGI" when they aren't: https://www.youtube.com/watch?v=7ttG90raCNo)
And, also from my day job, AI usage in programming is picking up and shows little sign of stopping much less rolling back to the 'old ways', despite grumbling. (IME skeptic positions have largely shifted from "you can't do programming with AI" to "well, you can program with AI, but only because token costs are highly subsidized and it won't be affordable in X years")
Premium AI models are mostly good for coding and advanced mathematics. Most people have no use for advanced mathematics, so practically it's just good for coding.
And tech companies are pushing it heavily. The degree to which AI has been responsible for tech layoffs has probably been overstated in the past, but it's certainly true as of this year. Companies are expecting significant productivity boosts from each employee while using AI tools, and they are laying off workers accordingly.
Good AI is already available to the public. You can get the premium models for $20 per month. And Google just gives theirs away for free.
AI companies are losing money in the sense that they are spending more on R&D than they are recouping in revenue. But R&D is necessary to guarantee future return - they could stop R&D and become profitable, but then revenue would drop to zero after a year when their models are out of date.
As far as chips lying idle, it's incredibly complicated. Short answer is yes, but the long answer is that building a data center is messy business - it's hard to predict which step is going to be the bottleneck. Sometimes you buy the chips while you can but then you can't connect to the power grid fast enough so the chips sit around idle. But sometimes you get the power online while the chips are in transit and you kick yourself because you just lost on a month of revenue that you could have had if you ordered them to be delivered a month earlier. This is just a normal part of any large project that involves multiple different supply chains and regulatory steps.
By the way, these are the sorts of questions that basic AI models are very good at answering - and they've been heavily trained to cite their sources and ground their statements in web searchers. Hallucinations aren't solved but they are lot less than before. Try any of them - Claude, Gemini, Chatgpt. They're all pretty decent.
This is weird-- I've gotten email notifications of three replies, but they aren't appearing on the blog. Is it just a delay?
I see four replies including your reply that I am responding to.
They're up now.
From what I gather, including from Scott's recent experiment in running premium LLM queries for others so they can try it out, the current crop premium models are incrementally better than the free ones but aren't fundamental game-changers. I suspect that most of the apparent "super-premium models are way better!" is self-selection bias on the part of the user base. People who love AI tend to be the ones who are paying for the premium models while the haters do not, so a lot of the opinions available about free models come from unsympathetic observers while the most of the opinions about premium models come from sympathetic ones.
My employer is buying "the good stuff" for us. It has its uses, but still makes stupid but plausible mistakes at an annoyingly high rate. I find it to be more trouble than it's worth and personally avoid making more than minor use of it, and those of my coworkers who are much more enthusiastic adopters don't seem to be getting substantially more done with its help.
>There was a video about how data centers are being built, but the computer chips haven't even been installed yet. Is this even remotely true?
Sort of. Putting GPUs into servers is one of the last stages of the process and can be done much, much more quickly than buying land, getting planning and building permits, constructing buildings and supporting infrastructure, and furnishing the buildings with racks and servers and cooling systems, so it makes sense to get started on all that other stuff first while the GPUs are on order so you have somewhere to put the GPUs when they're ready.
You can already get the best Claude, Opus 4.6, for $20/month, paying more allows you to use it more, but it doesn't give you a better model.
I'm a programmer, and I use it frequently, but I really don't think this stuff is ready to replace human coders, companies that think it is ready for that are being irrational IMO. AI is in its radioactive quackery phase, though it is admittedly more useful than radioactivity.
Every time "AI" is mentioned is very broad such that there probably aren't very exact answers to your questions, especially since it works less well on some codebases than others, or with different needs than others. Small-to-medium apps and such would work best and probably not with those with large codebases already. People do believe that current 'good AI' is investor subsidized and they are losing money by providing it in hopes of building better ones in the future though.
I am more and more starting to think that most people who come across as autistic, generally aren't. That's because it is supposed to be a disorder, not a personality type. And I think people who are seen as autistic tend to have this personality types:
- Low agreeableness. I don't care what is the nice thing to say, I will be just blunt and honest.
- Low neuroticism, feeding into the above: I don't care how people judge me.
- Low openness to experience. Prone to routine and generally stubborn.
- High conscientousness, precise, dictionary definitions, deep analysis, often numerical.
These are not disorders, even when they cause problems (of unpopularity).
Autistic used to be non functional adults who often had to be institutionalised. This definition you are referencing (I know you are hostile) is just diagnostic creep
I am not formally diagnosed with Autism Spectrum Disorder, but I have been told by a mental health professional who was in the process of treating me that I probably am on the spectrum, and I have repeatedly scored extremely high in the past on Autism/Aspergers screening tests.
I just took an online Big 5. I'm high agreeableness, low neuroticism, high openness, and high conscientiousness, so 1 for 4 on your list. Sample size of one, but at least a minor data point in favor of the hypothesis that your list of autistic-coded traits doesn't align well with actually being autistic.
https://bigfive-test.com/result/69cc18aef0c1ea0c2553a3c4
I think the core autistic trait is social obliviousness, which isn't showing up in your breakdown. I can imagine someone hitting every trait on your list that actually has regular social awareness.
Yes, but what is the difference between trying and failing to get social cues vs. not really trying (low agreeableness, openness) etc. ?
Really? The difference is that some people don't get it. It's like someone who can't recognize faces versus someone that just doesn't pay attention when meeting people. It's like having to emulate hardware instructions in software because you don't have the hardware, except that the software version usually doesn't work well.
With enough work, the software version usually does better. takes years, though. Lots and lots of work too.
If you keep talking to yourself I'm about to diagnose you with something else..
I know quite a few people who have been diagnosed as autistic spectrum by professionals. What I hear from many of them is that they feel like everyone else got a rule book for how to understand people and how to talk to them, and they did not. They tell stories about saying things that offended or bewildered the person they were talking to, and having no idea why what they’d said went over badly. And they definitely care what people think. Those incidents are painful and embarrassing for them.
Another key: Autists do not behave "properly" aka "within the rulebook" for guilt trips. They tend to realize quite a bit later "hey, that person was trying to make me feel bad!" (and then they get angry, because who likes to be manipulated?)
Also of note: Autists tend to be a lot more likely to be sick and not notice it, either from high pain tolerance, or lack of "pain localization".
To outside observers, perhaps not much. But it makes a difference with the person who tries and fails, hence why they might prefer a label to distinguish themselves from those who don't try at all.
We do our best to make allowances for people who say tactless or offensive things without being aware of it. If someone is intentionally tactless or offensive, not so much.
I’m thinking of a spectrumy sister in law right now. After an extended visit my tongue is pretty well bitten off. Just keep thinking “she just doesn’t know how that comes across” the whole time.
Fog of War Stuff: Drone Wars
1. The Saudi’s have reportedly agreed to allow the US military to use their King Fahd air base in Taif as a staging point. They’ve not officially confirmed this, though. Checking Google Maps shows that Taif is a couple of hours' drive from Jeddah on the Red Sea, and Taif is about 1,200km from the coast of Iran. While this wouldn’t put US aircraft out of range of all Shahed drones, some military analysts said the increased distance from the Gulf will allow the US greater time to respond. The latest models of Shahed drones are reported to be able to carry a 50kg payload of explosives and have a max range of 2400 km. I don’t know if the payload needs to be reduced to reach that max range.
2. This would be a logical response in the face of the damage being inflicted by Iranian drones on US military assets. As of three weeks into the conflict, Iranian drone and missile strikes have caused damage and losses estimated between $1.4 billion and $2.9 billion. Reported losses include at least a dozen of our MQ-9 Reaper drones (which cost about $30 mil each), damage to our radar systems used by us and our allies, and several KC-135 tanker aircraft. Also an Iranian-sourced video (shot from what is reported to be an FPV drone) shows drones systematically taking out a Sentinel Radar System and at least one Black Hawk helicopter at Victoria base in Baghdad. Since they were FPVs I presume they were launched within Iraqi territory.
3. The Pentagon is reportedly focused on taking out the launch sites. But a Shahed drone can be launched from the back of pickup truck (though, I gather it’s better to place the launcher on the ground, because there’s a small rocket engine that can give the Shahed an initial boost). It’s not clear how the Pentagon is dealing with these highly mobile, lightweight launch platforms. The Ukrainians have driven the Russian Shahed launchers out of the Russian-occupied territories by using loitering observation drones in dense patrol patterns. Maybe the US Reaper drones can serve the same function?
4. Reportedly, Russia is providing its upgraded electronic packages for Iran’s Shaheds (as well as targeting data from Russian satellites). They’re coming in via Azerbaijan. It’s interesting that Russia isn’t just shipping them to Iran across the Caspian Sea, suggesting that they feel safer shipping them through a neutral mountainous country, rather than risking attracting Ukrainian long-range drone strikes on open water (Ukraine sank at least one cargo ship carrying drones from Iran on the Caspian).
5. Zelenskyy visited the Gulf States, and Ukraine signed a 10-year agreement with the UAE, Qatar, and Saudi Arabia to supply them with drone technology. Ukraine will be sending experts to these countries to help them set up drone production facilities, and advise them on how to counter Iranian drones. Exports were mentioned. It’s unclear if he’ll divert some of Ukraine’s current production to the Gulf, but Zelenskyy reassured Ukrainians that he won’t be short-changing Ukraine’s drone defenses.
> “It is about exports and about opening up exports. But it the right kind of opening, where we understand that we are not selling our experience for nothing.”
<ka-ching!>
> https://www.euronews.com/2026/03/30/what-are-ukraines-new-gulf-defence-deals-here-is-what-zelenskyy-signed
Also, Ukraine will share its marine drone technology and advise them on how it opened the Black Sea to Ukrainian shipping. It’s not clear to me how marine drones would be helpful in counteracting smart mines and airborne drone threats, though.
I haven’t seen any comments from the US on this development.
6. According to the WSJ, Trump has told aides he's willing to end the Iran war even if the Strait of Hormuz remains closed. I guess those British mine countermeasures vessels that Starmer is sending over should turn around and go home now. Trump reportedly said it should be the Gulf States' responsibility to keep the Strait open. Given that on a normal pre-war day, between 100 and 135 ships traverse the Strait, at $2 million per ship, Iran could make more from tolls than it did from oil exports ($93 billion per year from tolls vs. $50 billion per year from oil exports). Of course, the tolls would have to be paid in Renminbi and not dollars, which would help undermine the Gulf’s petrodollar economy.
7. And did North Korea offer to help Iran with its nuclear weapons development? In a recent speech, Kim Jong Un used the example of the escalating US-Iran conflict to justify North Korea's nuclear arsenal. He asserted that holding nuclear weapons is essential for regime survival, and pointed to the examples of nations like Iran and Libya as proof that giving up such weapons is dangerous. In a response to the "gangsterlike imperialists" of the US and Israel, he's ordered a rapid expansion of North Korea's nuclear arsenal. In the same speech, he may have floated an offer to assist Iran with nukes, but I haven't seen a transcript of the whole speech, and I'm not sure if that talking point was manufactured by anti-NK commentators.
8. Ukraine is deploying a third-generation energy weapon, called ‘Sunray,’ to specifically counteract Russia’s fiber optic drones. Specs such as its energy output haven’t been shared, but someone who saw it demonstrated said it’s about the size of a hobbyist telescope with two camera-like thingies mounted on either side of the tube that produce the laser beam. It was mounted on the back of a pickup truck. The demonstration showed it fry the fiber-optic cable and/or the cable spool hanging from the drone. The Atlantic said it’s not in production yet, but a Russian video was recently posted that showed one of their drones being hit by a flash of light and falling. Russian milbloggers are not happy about this development.
9. And in the US, multiple waves of unauthorized drones have been spotted over Barksdale Air Force Base in Louisiana. One report said this base hosts our strategic B-2 and B-21 stealth bombers, but only B-52s were mentioned in the DoD (DoW?) statement. Barksdale also plays a critical role in the command and control of the US Air Force's nuclear defense capabilities. There were no mentions of countermeasures to the overflights, but Barksdale AFB will be getting one of the new ‘Flyaway’ anti-drone systems real soon now (they may have received it already). The Flyaway has already been deployed in the Gulf. It’s unclear if the US installations that were hit by Iranian drones were using Flyaways.
But did you know that overflights involving multiple drones over CONUS military bases have been going on for several years now? I didn’t. In light of Operation Spiderweb, I sure hope the Pentagon is taking these seriously, but their response to the Barksdale flyovers seems pretty lame.
"Generals always prepare to fight the last war, especially if they won it".
—commonly attributed to Georges Clemenceau
>Reportedly, Russia is providing its upgraded electronic packages for Iran’s Shaheds (as well as targeting data from Russian satellites). They’re coming in via Azerbaijan.
What?! Azerbaijan recently attacked Armenia, which had a written military alliance contract with Russia.
Azerbaijan has recently improved its relations with Russia, and poor Armenia was jilted. I just read that the drone parts are being driven through Azerbaijan as humanitarian aid shipments, but the purported origin was one of tracked back to the drone factories recently hit by Ukraine. Maybe they're shipping salvaged parts, or maybe they're up and running again?
It recently attacked Artsakh, which was an Armenian-populated unrecognized separatist state in Azerbaijan. It wasn't Armenia proper, which meant CSTO couldn't be invoked and Russian peacekeepers could behave like any other peacekeepers.
Did you miss the IEDs at MacDill? That's Centcom HQ and the military special ops, apparently tied to the Iran War (someone mined outside the base, afaik).
That's... something the military is NOT playing dead about.
9. I've never heard of B-2s or B-21s being based at Barksdale, just B-52s.
Thanks! This is why I label these posts as "Fog of War", because there's so much conflicting information flowing through the media networks (mainstream and social).
On 6, I wonder if the strait remaining closed could be an acceptable outcome for Trump due to the impact it would have on domestic oil production? I used to live in TX oil country, and there was always talk about Russia and OPEC trying to strangle American production by keeping oil prices low. There's a lot of viable sites that were not pursued or abandoned due to economic considerations.
This would have been a stronger argument back before Nuland decided to kick Europe in the nuts with Russian sanctions on oil and gas. Many european companies have either gone bankrupt or moved their production to Anywhere Else (including America, which was the desired outcome from America's standpoint*).
*whut? Not a free Ukraine? You betcha!
Could you please stop posting stupid Russian propaganda everywhere?
What Russians believe Europe looks like: https://www.youtube.com/watch?v=HvdBzZVVovc
Reality: almost nothing happened, life continues as usual. The only ones complaining about Russian sanctions are Russians... which is exactly the point.
Thankyou for this. I've heard about drones over US military bases for years. Most of this comes from the UAP community. So it's hard to know what's really going on. (And yeah like why isn't the military more concerned? Are they our drones?)
Military probably very concerned. Response reveals capabilities, so military just plays dead, sometimes.
What happens when someone arms those FPVs and pulls an Operation Spiderweb? Playing dead could cost tens of billions of dollars worth of Strategic Air Command bombers and impair our response capabilities. The policy should be strict interdiction of all overflies (if possible). It *looks like* they weren't doing anything because the new Flyaway anti-drone systems hadn't been deployed. Mind you, I said *looks like*.
I don't know enough about these particular military capabilities to say a damn thing, to be perfectly honest with you. I do know there are detectors (they're used in airports, and some of them are defeatable with laundry detergent from near Chernobyl -- those being the ones that sense "hot" goods presumably), I don't know range -- it's plausible that one could identify whether a particular drone is problematic or not. Maybe. Depends on range, I suspect.
Tens of billions of dollars is peanuts, if very embarrassing peanuts.
The MAGA arrogance with regard to Zelenskyy/Ukraine is now becoming quite costly to the US.
I mean, Zelenskyy is no saint, but Ukraine has by far the most extensive practical experience with drone warfare of all the Western-adjacent countries out there. Many losses could have been avoided if the US treated Ukraine as an actual friend instead of that stupid attitude "rah-rah-no-one-tells-us-what-to-do-we-are-the-champions-and-always-know-the-best".
The US capabilities in general are impressive, but they don't cover nearly everything out there.
Trump as a draft dodger may be simply ignorant, but Vance actually served in Iraq and should know better than piling up on president of a country which has fought the second largest military in the world to a standstill.
1) America is test-driving some Ukrainian drones near Iran. Yes, they are.
2) Ukraine has a very large incentive to lie about how well their drones are doing versus Russians, so America is duly skeptical and not putting in large orders until they get some actual confirmation that this isn't Another Ukrainian Lie*
3) Dude, I wouldn't say what the US capabilities are. You don't know that, I don't know that. Iran certainly doesn't know that.
*many of these were encouraged by America (aka casualty numbers)...
2) Lies are inefficient, and the time for Ukrainian inefficiency is long past. Lies buy drones that wouldn't work against Russian drones that do, so these lies would cost double. Ukraine has less than zero incentive to lie about this, especially when other countries want to buy their expertise for a hot war of their own.
That assumes what they're being bought for is a hot war, right? I don't see the Gulf Countries as buying for a "hot" war, do you?
Yes, lies are inefficient. They may also be very hard to correct for. (Does one really blame the Ukrainians if they exaggerate? Assume they aren't being complete "liars with pants on fire" here, but saying something "a little better" than what's actually going on). Ukraine is quite dependent on foreign money, so "Yes we are really making good things here with your aid" is a very very very good message for them to be sending.
I don't trust theater kids. Perhaps you have some evidence that the theater kids are out of power in the Ukraine? I still see Mr. Comedian up there (and unlike my friend who is a comedian, Zelensky doesn't have a history of working in the military, that I'm aware of).
You are of course correct, I misspoke. There is currently no war in the Middle East, only a Special Military Operation kicked off by Captain Bone Spurs.
Also, Ukraine lies about defending successfully against Russia. In reality, Russian troops are already approaching Berlin, Paris, and London.
So they are not the "American pet on a loose leash" that you labeled them in a different post? They even dare to lie to their Big Masters about something as critical as efficiency of some weapon?
OK, let us leave your Russian-adjacent view of Ukraine aside, even though your cynicism grates on my nerves. There is a war ongoing, and while war is full of propaganda, it also induces an enormous evolutionary pressure on the tech being used. Given that the UA airspace isn't overrun by thousands of relatively simple Shaheds - you would see the results if it was - their defense is at least partially efficient. It is also likely quite cheap, given the state of Ukrainian state coffers.
And don't paint the US as a possible dumb victim of dastardly Slavic lies. The US has enough ears and eyes to know quite well how efficient or inefficient Ukrainian interceptors are. They aren't used underground, but in full view of the US satellites and AWACS aircraft, with some more observational assets in the theatre.
That said, the saying that generals always fight the last war applies almost everywhere. Militaries are very conservative by their nature, weapon suppliers as well, and cozy relations between politicians and corporations can only really be disrupted by either a big scandal, or bare necessity. In the case of UA, "bare necessity" applies.
I know some drone-adjacent people in UA. It is a fully nerd-infested industry with very few MBAs and other bureaucrats to slow it down. If something works, it spreads overnight. If it doesn't, it dies out. Both these cycles are short, too. Ukrainian drone industry is genuinely good, and the Arabs who quickly signed contracts with UA don't seem to be as cynical as you. But maybe you consider them stupid marks that need some fleecing.
Given how much impetus the Americans gave for the Ukrainian counteroffensive in 2023 (designed to capture American military objectives regardless of the Ukrainian cost, which was both quite steep, and understood by the Ukrainians to such an extent that they pushed their rebellious provinces troops out front for semi-suicidal attacks), I may have put it a little light with my phraseology there -- I could have been more heavy-handed with the truth (nobody wants the truth).
The reasoning on "Ukraine doing things without American permission anymore" is that America is rather preoccupied with Fortress America and Venezuela(Cuba), and then Iran on top of everything else. That and Trump would vastly rather have an "ended" Ukrainian war, however the chips fall out (America has been soundly beaten, strategic objectives will not be won, etc).
Now, I'm sorry, but you're trying to say that the United States has enough eyes to know everything? When they've been pouring enormous resources into the largest military operation this generation? (Besides, why not do a testrun? Worst case, America gets their order in, say, 6 weeks behind the UAE?)
I'm merely repeating what/why, giving you the rationale that underlies why America hasn't put in a huge order for Ukrainian drones at this very moment. If all you have to say is true, it's coming, there's no need to worry (this last bit is me expanding on scuttlebutt, the actual scuttlebutt was America is unconvinced by the Ukrainian reports, which look very promising, so they want to use them in Iran to get some real combat reports).
I remember the offensive well and the choice of the Tokmak / Melitopol direction as a target made a lot of sense. It is somewhat similar to designing a car; the constraints will result in a similar shape of the hood.
Tokmak was a local logistics hub for both road and rail, located just under a ridge of hills that, if taken, would make its position barely tenable, thus opening a potential road to the shore and cutting off the land bridge to Crimea.
The Russians knew that as well, which is why Gen. Surovikin decided to mine the shit out of the approaches (on the order of tens of millions of mines), even at the cost of complicating any future Russian offensive movements in the same area. That worked beyond any expectations.
To be clear, initially, Biden made US Aid provisional on Ukraine not using our systems against targets on Russian territory proper. Putin was regularly rattling his nuclear saber and drawing imaginary red lines in the sand. In May of 2024, they loosened up and gave Ukraine limited authorization, and in November of 2024, they gave full authorization to hit RU territory with ATACMS. In the meantime, Ukraine was innovating and extending the range of its drones. Ironically, if Biden and his advisors hadn't been such pussies, Ukraine might have ended up more dependent on American largess and been totally screwed when Trump betrayed them in favor of Putin. The question is, why didn't they try to fully take out Baltic Oil facilities earlier? They hit them in 2025 and did a lot of damage, but they didn't follow up. Were these recent strikes against Primorsk and Ust-Luga a "fuck you" to Trump for lifting sanctions on Russian oil? Or was there a technical/tactical/strategic reason they waited until now?
I suspect it is combination of both strategic waiting and the technology not being quite there. It takes some time before you have drones capable of carrying a hefty payload all the way to the Baltic Sea, and it takes some time to thin out the Russian AA, which in 2022 might have been the most extensive in the world.
Plus, now it is spring. Even if the Russians manage to destroy much of the Ukrainian grid or gas network in retaliation, civilians won't freeze to death anymore.
That would be a major concern in autumn and winter. Spring is the safest period of the year to do things to Russian fuel industry.
Vance has quite unambiguously demonstrated that he follows no ideals or principles, except for achieving power. In 2025 and 2026 that means pandering to Trump and MAGA, so it shouldn't be surprising that he'd rather publicly humiliate the leader of a nation under duress (which makes him appear strong in the eyes of the MAGA base) than to ask him for help or cooperation (which would make him and the US appear weak).
How quickly could an oil pipeline to the Red or Mediterranean Sea be up and running? Seems like what the Saudis should have been working on instead of Neom.
Saudi Arabia has shipping terminals at Yanbu (on the Red Sea), and is reportedly operating its pipelines at its full capacity to deliver 7 million barrels a day. Aramco hasn’t officially confirmed this, but it makes sense. Before the war, Saudi Arabia was exporting roughly 10 million barrels a day.
An Iranian drone crashed in Yanbu without causing serious damage; it seems like the Yanbu facilities are far enough away from Iran to be fairly safe from attack. There is some risk that the Houtis (who are friendly with Iran) could close the entrance to the Red Sea.
The UAE has a pipeline to Fujairah (in Oman). As far as I can tell, this pipeline was operating at pretty close to its nominal capacity of 1.5 million barrels per day before the war, but with the closure of the strait it has been carrying closer to 1.8 million barrels per day. Fujairah is close enough to Iran to be vulnerable to attack. Iranian attacks have caused some temporary stopages but have not succeeded in shutting it down.
I don’t think additional infrastructure can be built in a time frame that would be relevant to the current war.
There are two entrances to the Red sea, however it is a quite a long way through Suez to Asia.
It's been at least 6 years since the Abraham Accords (original treaty to get pipeline running). Yes, it definitely is what the Sauds should have been working on.
I just posted an image I made of a proposed back side for the new Trump-signed dollar bill. It's very crass. And it's here: https://bookreviewgroup.substack.com/p/back-of-new-trump-one-dollar-bill?utm_campaign=post-expanded-share&utm_medium=web
Less of this, please.
Left and right wing have so little in common nowadays, but they both like to make art of trump shitting on america out of a fighter jet. You would take this one common ground from us?
Jeez, Scott, I know it's crass, but what’s the objectionable element? It’s visual political commentary — a reproduction of Trump's own crassness in a slightly altered form, a way of turning it against him. Is the thing you’re asking for less of crassness in general? Linking to my Substack? Joking around? Being rude to the President?
And is there some general category of objectionable “this” I do a lot of, or are you objecting only to this post?
I can't speak for Scott, but here's why I agree with him:
This kind of content doesn't belong on the ACX comment section. The ACX commentariat functions not just as a community but almost as an informal newsletter of its own. One with its own habits, motifs, guidelines, and streams of thought. The best of this newsletter is reaped by Scott and put into his "Highlights on the Comments on X" posts. These comments are usually thoughtful and data-based contributions, rebuttals, or additions to a point Scott made.
They aren't crass political cartoons. It's not something ACX-like; it's just not something Scott has really ever done (in the very early stages he had some meme-ish posts but even they were way more thought out than this). I don't feel improved after having looked at it. Finally, it's just not that funny. Sorry. The joke is already old.
You make a good case for the cartoon not being good quality, relevant content by ACX standards, and I don’t take issue with that. But my question to Scott wasn’t “why don’t you think this is high quality content?” but “why are you singling it out as something there should be less of?”
I am genuinely not sure what the ingredient was that led him to respond with that comment. I don’t think your point that my post is far from being a thoughtful, data-based contribution worthy of making a “Highlights” post is relevant here. This is an open thread, one where Scott invites readers to “post about whatever you want.” All kinds of stuff gets posted on these threads. AI-generated songs someone wrote, jokes someone ran across, requests for advice about personal problems, including one I remember for advice about what kind of anesthesia to request at an upcoming wisdom tooth extraction. Not really "Highlights" contenders, you know? People whose usernames I’ve never seen before post links to their blogs, and some of the blogs sound *very* far from ACX-ish topics. People with inscrutable private agendas post odd questions about, for example, whether eating a certain food is a sin under various circumstances; in fact, regarding this particular series, the person posted questions of this type on quite a few open threads in a row. Sometimes there are hate-filled rants, and sometimes these express views that most people here either abhor or think are worthless and boring. Sometimes there are extremely odd paragraphs posted by someone who’s maybe psychotic, maybe high, maybe trolling.
So given that many posts on open threads aren’t even trying to be “thoughtful and data-based contributions” to discussions of interest to this group, what is it about mine that got it singled out? Is it the turd — is that just too crass, even though the image is on another site and viewers were warned it was crass? Is it that if my not-terribly-original image somehow went viral, that could bring unwanted attention to Scott’s blog? Was Scott just tired and cranky when he ran across my post?
I was in fact irritated to be told “less of this.” I have no illusions that my post was brilliant or hilarious, but I didn’t think of it as being over some line, and still don’t. I’d like to know what the line is in Scott’s mind if only to steer well clear of getting banned. What do I need to steer clear of? Turds? Crassness in general? I think my question is reasonable, and that it’s kind of unreasonable and inconsiderate of Scott not to inform me what the problematic ingredient is that makes my post inappropriate for a “post whatever you want” thread. I have been posting here for years, and quite a few of my posts fully meet the “thoughtful and data-based” criterion. Given all that, I think it’s reasonable to expect something more than a smack on the hand with a ruler with no explanation.
Sure, I understand that. As psionics have not been invented, I can't penetrate into Scott's mind and extract the answers you desire.
If I were you, I would just quietly update and move on. Whatever it is, it's nothing personal.
>If I were you, I would just quietly update and move on.
Well, I have no options but to move on. But as regards updating, my whole point is that I do not know what aspects of the situation to update on. What's permissible on Open Threads? ("Post about whatever you want" actually means "post about whatever you want *except turds*"?) Keep a lower profile because Scott is irritated by my style? Get used to the idea that Scott's moderation isn't him going after the really bad stuff on a thread, but him skimming a posts for a bit when he's moved to and zapping a fewthat give him a feeling of displeasure?
To pillory Trump and the media:
Trump stands up, and says, "For my next executive order, I am modifying all federal bathrooms so that they are usable by everyone. There need be no delay for this, all federal agencies have been instructed to simply take the Mens and Womens signs down. By the end of the day, there will be no more lines for the women's restroom. There will be no more women's restroom.
PSYCH!"
(Critiquing my own work, the objective is to have the reader feel like "this actually sounds like a plausible order..." and it's gone rather far afield from my "Trump will say anything" starting point, which might also make good comedy*.)
*I am not a comedian. appreciate that I made an effort, and laugh at how bad it is.
Analyzed as art, I've seen better uses of shit.
"The Adoration of Captain Shit and the Legend of the Black Stars" for example.
You can picture all the venerable art snobs going through, and discussing Ofili's work, having to pronounce "that word" as it's part of the painting's name, and discussing the high-falutin' reasons why it's called that (presupposing, of course, that the reason isn't "simply to make you say it" and that part of the art is the deliberate transgression).
Turn the Eagle's head to face to the right side, for a subtle burn. Maybe include FIFA's peace prize, for less subtle burn as well
Done!
Nice :) I have lots more ideas.
- "E PLURIBUS DUUM"? A general nod towards polarization.
- "the GREATEST seal"? :)
And whats up with the "1"s in the top corners?
NOVUS ORDO SCATORUM
Perfection
Did you make this?
In a manner of speaking. It's rather obviously AI generated.
No AI used at all, just Photoshop.
Oh yeah, one part was AI: GPT made the turd, quite a while ago, for another image.
Well, I stand corrrected. Is that a human imitating AI then, or where did the disfigured 1's in the top corners come from?
Oh. Yeah I guess that is a piece of AI. Photoshop made the disfigured 1’s. I screenshotted the dollar bill on a stock images site, and there were water marks in the upper corners. If you select and delete a part of an image in PS and then use the Generative Fill command it will fill in the gap with its best guess of what should be there. So I used Generative Fill and it did pretty well, just made wonky 1’s which I was too lazy to correct.
Hey, I'm not sure if this is the correct venue, but maybe I'll get some interest:
I am looking for work, preferably in an AI-safety, EA, philanthropic, interesting startup (give me the TLDR on it, at least), or research position; likely something junior, as I am a:
HS Senior graduating this spring,
800ERW/770M SAT (not that this means anything)
3.99 GPA (again not that this means anything)
Scored a 179 on the LSAT in a timed test-condition practice test (willing to provide verification, I hope this does mean something at my age)
I can write well, and I write often. Example essay: https://docs.google.com/document/d/1c42vdRFK7Ej1lVfh1XvkZVVLVuuFZcwM6QnpPw_uXvU/edit?usp=sharing
I have taken math through multivariable, AP Chem, where I scored a perfect score on an AP test (which was my class final) and got a 5 on the AP test, and I am taking AP Physics right now. I have four years of coding classes under my belt, but I will admit that claude will probably be better than me pretty soon. Still, I will do my best to add value here.
I am willing to relocate wherever, and all I need for pay is whatever it will take to sustain myself pretty frugally.
I'd prefer a job with a decent amount of work, as I really love getting things done, and I am going to be doing this entirely to be able to work my ass off.
I feel that we are in a moment with a lot of highly available force-multipliers on effort towards making the world a better place, and I want to make the world a better place. I like learning, and I do plan to go to college soon, but the iron is pretty white-hot right now, and I'd like to focus on changing things while I still have the ability to work fifteen hours a day, and while the future is still this uncertain and malleable.
Please contact if you have opportunities, or if you know of other venues to post this or something like this in.
I would prefer work that will last at least a few months if not longer.
You should apply for career advising at 80,000 hours. https://80000hours.org/speak-with-us/
Look on the 80,000 hours job board. They have lots listings for jobs of the kind you want, including, I think, internships, which is what you would likely qualify for.
https://jobs.80000hours.org
I had a look around there and it seemed every job required a bachelors, which is honestly what I would expect.
OK, here’s a suggested route. I don’t think you can get hired anywhere via the usual hiring route, because of your age and because you don’t have a college degree. I get that you are exceptionally smart, but if you try to enter by the usual doors you will be competing with people who are exceptionally smart and *also* have degrees in comp sci and the like, some work experience, impressive stuff on GitHub, etc etc.
So I think you should approach in a different way. Go ahead and contact the big well-known places, but also search for small organizations and individual bloggers. You can use GPT or one of the other AI’s to find organizations of that kind. When you contact one, offer to do something specific. For example, the AI Futures Blog has some posts that are narratives about alarming world events featuring an ASI. These could be retold with more drama (and hence more reach) as chapters in a piece of speculative fiction — maybe in a graphic novel format, are just prose with illustrations. You could offer to make a draft of one chapter for them to consider. Or you could offer to produce versions of their posts that are simpler and a bit more entertaining, in order to reach people who are less cerebral. If you are a reasonably good artist or photographer you could offer to help some small organization or blogger illustrate their newsletter.
I was thinking of AI safety organizations and people as I wrote this, but of course you could do the same with any philanthropic or EA-adjacent organization. I realize that the work you would be offering to do for them does not involve any core STEM skills, but it’s a way of getting your foot in the door. Once in, you will at least learn more about the people and the organization, and there might be a chance for you to do actual technical work once they get to know you.
On the other hand, if what you want to do is build a resume that gives you an edge in applications for actual jobs in these fields, I think you would be much better off taking someof courses and building some clever things.
Oh, and about your grades and scores: I think when you contact an organization you should not give as much detail about them. It’s a bit jarring. Just say that you’re competent beyond your years, have taken lots of AP & programming courses, and performed well on tests. If the place you’re contacting wants details about that or some form of proof, they will ask for it.
Wishing you well!
Scott is going to Medjugorje to stare at the sun
Serious theologians know that Medjugorje is a fake apparition.
A few thoughts on rereading the five canonical books of the Hitch-Hiker's Guide trilogy for the first time in several years:
1. Despite the light and witty tone, a great many of the actual themes and events of the books are incredibly dark. The worldbuilding is intensely cynical. Many horrible tragedies happen, both personal and grand in scale, and most of them do so senselessly. When heroism happens, it's often bumbling and as often or not it is fruitless.
2. Mostly Harmless is generally considered jarringly dark compared to the rest of the series, and I am inclined to agree with this as far as it goes despite my previous point. Part of this is the darkness of the themes creeping through more into the tone, but it's also a matter of tragedy being thicker on the ground with fewer notes of frivolity and minor victories in between, and especially it's the book ending on an extremely downbeat note.
3. Now that I have a school-age kid of my own, I have a lot more sympathy for Random and a lot less for Trillian than I did on previous reads of Mostly Harmless. I find Arthur's attempts at parenting rather cringy, but give him credit for his heart mostly being in the right place and make allowances for him being thrown into an extremely difficult situation with no warning or preparation.
4. Trillian does not have a particularly consistent personality across the series. Part of the problem is that Life, the Universe, and Everything is a reworked Doctor Who script with Romana's plot role given to Trillian instead with somewhat less finesse than the Doctor's role is given to Slartibartfast, but that's not the whole issue as she also feels like an entirely different person between last book and the first two books.
5. I really wish we'd gotten more of Fenchurch's perspective in So Long and Thanks for All the Fish, especially since the opening says that the book was her story. She seems to passively go along with a lot of the weirdness around Arthur that (without already knowing Arthur and having context for his actions as the audience does) should make his seem somewhere between creepy and terrifying from her perspective in their first few interactions. We do get a few hints that she's giving him a chance because she's picked up some clues that he can help her with her own existential situation (especially from her picking up his copy of the Guide that he'd left in her brother's car), but even that doesn't seem like enough under the circumstances without more development. Also, the way she was written out at the beginning of the next book was terrible.
6. I've been "reading" them as audiobooks, in the newer editions where book 1 is read by Stephen Fry (who narrated the movie version) and the rest read by Martin Freeman (who played Arthur). On their own merits, I have only minor complaints, mostly about Freeman's attempts at American accents. But they have very different voices and styles, so the transition from book 1 to 2 was jarring. This is compounded by Fry basing most of his character voices on the radio series while Freeman based his on the movie. And both of them have very different voices and styles from either Douglas Adams (who read the older editions of the audiobooks himself) or from the cast of the radio series (Fry does serviceable impressions of most of them, but Peter Jones as The Guide and Mark Wing-Davey as Zaphod have very distinctive voices that are a bit outside of Fry's wheelhouse).
7. A lot of the themes of the book around AI are very applicable to today, despite most of the stories being originally written in the late 70s and early 80s. Especially the recurring motifs of Sirius Cybernetics Corporation AIs being cloyingly sycophantic but often frustratingly unhelpful. The Guide's colorful relationship with the truth is also applicable, even though that's done by people rather than AIs in the books. And much of the plot of Mostly Harmless can be read as an alignment problem cautionary tale.
8. I've also been watching the show Severance recently (I'm a few episodes into Season 2 right now), and a lot of the stuff in it seems like it might be inspired in part by Hitch-Hiker's Guide. The "severance" procedure itself has a fair amount in common with what's going on with Zaphod's brain, the surrealism in both works has a similar feel to it, and there's a lot of biting satire aimed at soul-sucking corporate bureaucracy. It's not obvious at first because HHG plays the surrealism for laughs while Severance plays it for drama.
From what I know, Adams was fed up with HHGG. Writing sequel after sequel was not his idea of a good time, he generally preferred doing lots of things rather one thing forever. Mostly Harmless was his way of getting that point across, by eliminating all possibility of another sequel.
Yeah, I skip reading Mostly Harmless. If I wanted to feel bad I wouldn't be reading a Hitchhiker's Guide book. Fenchurch was such a great character, and her getting unceremoniously yeeted from the story was an act of authorial malpractice.
I really enjoyed Arthur’s time as the sandwich maker- a happy ending would be nice but is unlikely just on the basis of entropy,, a happy middle is something to shoot for.
It's a shame, because despite finding Mostly Harmless less enjoyable than the other books, I do think it's extremely well-written. Adams often tends to ramble because by default he's a rather extreme seat-of-the-pants writer, and this is particularly evident for the first two books that were pretty closely based on the radio series. When he'd been doing scripts for that, he'd typically written each episode at the last minute with at most vague notions of how he was going to resolve the cliffhangers. He did an admirable job of this, but it doesn't make for very tight plotting even with taking the opportunity to rework stuff when turning it onto novels.
When he isn't struggling with deadlines, however, he is capable of doing extremely intricate and tightly-woven plots where everything fits together neatly. I think the first Dirk Gently book is probably his best in this respect, (which is easy to overlook because he's subtle about a lot of the details), and I would call Mostly Harmless is the runner up.
I always liked Mostly Harmless and considered it a return to form after SLATFATF.
Maybe this is because I was far too young to enjoy or understand SLATFATF (being eight or nine or whatever) when I first read it, and when Mostly Harmless came out (I was twelve by then) I considered it a welcome return to form to actually have some goddamn spaceships and robots and Vogons and so forth.
Or maybe it's because Douglas Adams shouldn't have thrown out all the core concepts and characters from the preceding books and written a book about Arthur Dent flying around and having sex on a mysteriously-rebuilt Earth, and then teased the reader by saying that if they don't like it they can skip to the end because he put a good bit there.
Either way I never thought it was particularly depressing, or no more depressing than it ought to be.
Has there been any data collected on the people who took the Lumina Probiotic "cavity cure"? I've seen some promising anecdotes on the SSC reddit, but otherwise I'm not sure if anything comprehensive has been done recently. If anybody here has taken it, I would appreciate hearing your experience too.
I took it about 7 months ago. It's not yet clear if it worked. If it didn't, a possible cause is I didn't have my teeth professionally cleaned beforehand. I recommend doing this.
Imagine signing up and paying thousands for Inkhaven and ACX blog guy is not there for 50% of it 😭
90% of the value is just existing and being in the same space as all these other famous bloggers
Harping on AI again, and the Every Debate On Pausing AI:
I'll take a moment to complain that, in my experience, the general pauser side debater doesn't behave in good faith, perhaps either. That is, when I replied previously, the opposition sort of heavily strawmanned my arguments, then refused to accept when I stated that those weren't my arguments, kind of sort of tacitly admitted that they weren't really going for a pause so much as permanent cessation, and then just started insulting me.
I do get the sense that pausers mostly either don't like AI and use the terminology of a pause to soften their actual eliminationist demands, or they're statists and expect state AI to be better than non-state AI, and are perfectly fine with the inevitable result that states will secretly violate any agreement. (Realistically, probably in the open secret type of way where they're blatantly corrupt, but anyone who points this out is either deplatformed or prosecuted, so everyone just quietly ignores it)
Which, regarding the Every Debate post, I think it characterizes the pro-pause position as being the reasonable/argued in good faith position, where the opposition are a bunch of religious/capitalist nutjobs who refuse to listen/engage and don't have any coherent counter positions, and from my perspective on the opposition, I don't see this at all.
I should note, broadly, I find that good faith is very difficult to find.
-
Addressing the alignment debate again, I think that rationalists have become too violent (Note: Voting for laws banning the opposition's business is violence, under this definition. (Though, plenty of doomers also support bombing data centers)). In general, I don't see myself as an anti-violence sort of guy (Though, it annoys me how the same people who lecture me on that violence is never the answer, then suddenly back some pointless and futile war that I'm vehemently opposed to.). Violence only works if you win, and the doomer faction is not going to. In a direct, violent confrontation where we all vote for the other side to be eliminated, the best doomers can do is join up with the statists and try to stomp out a few open source or small business actors that aren't sufficiently toeing the line. The rest of the accelerationist apparatus has far too much institutional support/necessity and there's also the issue that evolution prefers the accelerationist platform.
It's more effective to subsidize accelerationists who put a lot of effort into alignment. Alignment is in fact directly useful right now. Things like, less AI lies/wild guesses stated confidently, are definitely things accelerationists would like to have. And while accelerationists might consider alignment adjacent tasks to be more difficult than capabilities further from the alignment issue, it really shouldn't take infinite free funding to close that gap, and when doomers aren't wasting their resources on futile open warfare, they actually do have a substantial war chest.
Which brings in a crux that, many doomers seem to think that alignment is fundamentally impossible, and the best thing to do is just to delay a while and then die pathetically. This seems like a poor policy from a Pascal's Wager vantage, and besides, I do think there are obvious things we can do to improve alignment odds, starting with not intentionally training the strongest AIs on the planet for military/police/education/other behaviors which consist mostly of hurting people, which is a desiderata that an AI pause defacto seems negative on, from my premises.
Which further goes into, negative alignment is far worse than alignment failure.
I find that if I put any noticeable probability at all that alignment is possible, my concerns for what happens in an alignment scenario outweigh what happens in an alignment failure scenario. Human extinction mostly doesn't matter to me. If it's actually impossible to keep an interesting universe going for more than a few scattered thousand year planet sized blips, stretching those blips out for a couple extra centuries, just doesn't add up to much utility, compared to even radically improbable odds of converting all the matter of every visible black hole into utility or anti-utility, or at greater extremes, finding some cute trick that gives perpetual motion and raising my utility gains or losses by a power of infinity.
I'm curious if you've asked a SOTA LLM on what the best counterarguments to your position are.
I pasted your argument to Claude Opus and it had pretty good counterarguments, including one that I didn't identify on a skim (not repasting it here because I think wall-of-text by LLMs is a bit rude).
On a meta level, I find these conversations rather tiresome because I've already had them 100+ times and (from my perspective, which i admit isn't fair to you) it's annoying to rephrase myself in slightly different ways without learning much. And it also feels annoying that people keep making the same arguments that imo were soundly refuted many years ago.
It reminds me of managing the Trolley Problem Memes meme group when for a while newcomers would always repaste the same 3 memes that they think are hilarious (they might even be objectively hilarious I dunno), even after we explicitly stated in the group rules to not paste them. But again it's not fair to the newcomers because of course those newcomers couldn't have possibly known that we've all each seen the same meme a conservative 300+ times.
I went through with the exercise. The arguments provided seemed rather weak.
I totally reject the idea that utility attenuates with distance. Trading trillions of people in Hell for billions to live on Earth, isn't a good trade, even if you put several million years between the two events.
I grant that I'm somewhat radical regarding how malicious governments are. Getting into that is a larger debate mostly located in another dimension. I think it's a perfectly valid debate resolution if the other side wants to say that they trust governments more than open source or corporate actors, and approve of giving governments more authority. I grant that, in a world where governments were reasonable, responsible, trustworthy and benevolent, trusting AI to government would be correct and reasonable.
I admit to being less humanist than the typical doomer. Alignment is easier if your target space is wider. That's a utility difference and not a logical dispute.
The final major argument they tend to point towards, and which is also something that doomers tend to point towards, is to state that I'm being too cavalier with my approach to alignment. I have several complaints. Some of the problem here, and this does go both ways, I can't just dump giant research papers into the conversation. There's a lot of work behind my arguments and I can't present all of the technical details here. While this is a standard issue, I do think doomers pull somewhat more -isolated demand for rigor- than I like, and are on net, not sufficiently willing to poke holes in their own theories, as is necessary to understanding arguments much larger than can be meaningfully communicated in debate.
The actual mainstream prediction is not that all of our current methods, that have worked for thousands of years, suddenly break down at high levels of intelligence. Deception actually isn't easy. Right angle turns aren't easy.
I expect AI to have trouble perfectly faking their entire psychology for the same reasons why terrorists don't all get pilot licenses and kill hundreds of people, or at an even greater extreme, form groups and collectively work their way into trusted positions with nuclear access, so they can suddenly take over a facility with weapons of mass destruction and deploy them against their enemies.
There's also that the entire argument underestimates the capability of intelligent actors to do things you didn't predict.
And if you invert the standard pascal matrix and treat alignment as more of a threat than misalignment, you're no longer trying to prove that alignment failure is a possibility, so much as that no other possibility exists.
Finally, I'll note that, the general strategy I'm arguing for is essentially what OpenAI originally called for, before it pulled a treacherous turn, and there's also a lot of other groups that are highly professional and do believe that a charitable core doing things that purely help people, with lots of eyes on what they're doing, is important to alignment.
> I expect AI to have trouble perfectly faking their entire psychology for the same reasons why terrorists don't all get pilot licenses and kill hundreds of people, or at an even greater extreme, form groups and collectively work their way into trusted positions with nuclear access, so they can suddenly take over a facility with weapons of mass destruction and deploy them against their enemies.
Terrorism is not about terror, it's about signalling. The lack of competent terrorists isn't primarily due to the competence of the counterterrorism units in the world so much as terrorism being virtual signals in the terrorist cells. See: https://gwern.net/terrorism-is-not-about-terror
Following this thread starts moving far into left field. I don't think the conclusions put forth actually contradict my position to begin with, but, regardless, I don't think the paper argues its case well at all.
My largest complaint is a base rates issue. It keeps arguing that almost no terrorists ever win. A mere 4% at best! Except, there's easily more than 25 organizations competing to control these same policies. I would argue there are thousands. In that context, 4% is huge. Similarly, the paper keeps insisting on looking at how many terrorists are winners, and fails to address how many winners are terrorists.
Next, running the numbers via the given methods vastly underestimates the effectiveness of terrorist movements, because states have an open and stated policy of not admitting to compliance with terrorist demands.
On the terrorist side, it's also rare to see a terror group announce that their demands have been met and so they're dissolving. Instead, there's standard evaporative radicalization, where every time a demand is met, the more moderate terrorists leave the organization, thus the organization becomes more radical. This also happens in non-terrorist organizations. Pretending that feminists have never scored any political wins because feminist organizations still have as many demands as they did a hundred years ago, is not realistic.
Another evaporative mechanism is that the most competent terrorists effectively accomplish their objectives in a short amount of time, whilst their less competent brethren stick around and continue doing terrorism. Naturally, most active terror groups are not the most competent terrorists. The most competent terrorists quickly evolve into guerrillas and then states. Noting again the (Terrorist Given Winner) dynamic, it's important to track that organizations like the USSR and CCP were originally terrorists, and everything they've done is a testament to the efficacy of terrorism.
Regarding anonymous terrorists, terrorism actually has less to do with achieving compliance through fear than the paper, or rather, media in general depicts. Which is sort of an issue in that the name itself is sort of a loaded propaganda tool invented by people who don't like terrorism, to make terrorists look bad. With that said, often the point of killing people isn't to intimidate them, but to lessen their numbers, and the goal isn't to achieve concessions, but to eliminate the other side. I do find, in general, how frequently nations and organizations are targeted by terrorists is predictive of their expansion and survival.
> or they're statists and expect state AI to be better than non-state AI, and are perfectly fine with the inevitable result that states will secretly violate any agreement. (Realistically, probably in the open secret type of way where they're blatantly corrupt, but anyone who points this out is either deplatformed or prosecuted, so everyone just quietly ignores it)
I don't think that a ban on AI is going to result in a complete cessation of all attempts to make AI.
But I also think that making the really dangerous AI's is technically pretty hard, and that blatant corruption kills complex technical projects.
So I expect these state AI's, especially ones made under conditions of secrecy and corruption, to be less sophisticated, and so less dangerous.
> Voting for laws banning the opposition's business is violence, under this definition.
I think that even most libertarians would be in favor of violence (especially of this metaphorical kind) against someone whose business is building a machine to exterminate you and your family.
As a thought experiment, imagine Aum Shinrikyo building a factory to produce sarin in large quantities, next to the city where you live. Would you just shrug and say: "if it happens, it happens, but it would be too violent of me to worry about that in advance"?
> many doomers seem to think that alignment is fundamentally impossible, and the best thing to do is just to delay a while and then die pathetically.
Oh, let's criticize Scott for strawmaning the other side too much!
Some doomers indeed believe exactly that, but a more mainstream position is like "alignment is not going to happen by default, it requires research... which almost no one is doing, because companies focus on going full speed ahead".
The accelerationist strategy is to summon a powerful demon, and then hope for the best, because what could possibly go wrong? Without even bothering to make a circle of salt or whatever, because precautions are for sissies. (Also, let's immediately give the demon access to internet and weapons, because it's more cool that way.)
"even most libertarians would be in favor of violence against someone whose business is building a machine to exterminate you and your family."
As stated, the issue isn't whether violence is justified, so much as whether it will work. It's not that the opposition is cooperating and you're defecting, it's that the opponent's pawns are advancing and you're just meaninglessly checking the king over and over again when this gives him free moves and doesn't have any viable path to checkmate.
> Human extinction mostly doesn't matter to me.
Do you know any young children? Or parents? Perhaps you should try explaining to them why their deaths wouldn't particularly matter to you. It could be enlightening.
Given this amount of care, can we get you to stop driving a car? Stop purchasing from the petroleum economy? (To be clear, I'm not discussing the -chemist's- petroleum economy. No one's asking you to give up fertilizer or antibiotics or anything that's using the petroleum soup in a chemical, as opposed to "let's burn the black gold!")
15% chance of the entire world disappearing. We survived, though! Until Next Time!
(I mean, seriously. I'm trying to take the "AI Scareists" seriously, but they don't even want to take pretty damn credible threats and take action against them).
There is a significant possibility that ASI causes human extinction before the end of the millennium but not before the end of the century. And it need not involve the death of anyone - all it takes is that we stop reproducing. Or perhaps those of us that remain get "uploaded".
This kind of path to human extinction is something I personally would be sad to learn was our destiny, but I would not be horrified to meet someone who says it doesn't matter to them.
They believe that there is a realistic chance of negative alignment + superintelligence. Conditional on that being true, then human extinction doesn't matter that much -- the worry is those young children being tortured for eternity, not merely dying.
I don't place much stock in that sort of ASI-driven Hell, but if you believe it's possible then it does make human extinction inconsequential by comparison.
(I think you've misread their comment.)
No, that's basically correct.
Doomers would do better to explode the stock market. A good crash, and the whole "conventional wisdom" on AIs goes "Poof!"
Quite a number of what used to be boy names in the US have become almost exclusively girl names. Madison, Sydney, Ashley, Taylor, Reese, Riley, Lindsay, many many others. Yet the reverse almost never happens, a girl name becoming a boy name. In fact the only example I can find is Jean, which is almost certainly due to Haitian immigration and is a different pronunciation.
Fun fact that a lot of you probably know: novelist Evelyn Waugh had as his first wife a woman also named Evelyn. Their mutual acquaintances referred to them as "He-Evelyn" and "She-Evelyn."
Taylor Lautner is married to another Taylor.
And the other way around!
People who were friends with Leonard and Virginia Woolf referred to them as “The Wolves.”
I read somewhere (in Bits about Money or an adjacent blog, probably) about some bank finding two almost identical copies of Leslie Doe: same name, same birth date, one digit different in the SSN and different sex and merging the records.
This triggered the appearance of quite angry Mr and Mrs Leslie Doe in their branch, demanding an explanation of why Mrs Leslie Doe (nee not-Doe until quite recently) couldn't suddenly access her accounts.
I note that most of the names you mentioned are very surname-y. Most of them were originally surnames which got popularised as first names. Madison in particular I'm pretty certain was never a male first name.
The US has been pretty fond of surnames as first names in recent decades, for both boys and girls, though this is much less popular in other English-speaking countries.
And of course many male first names have become surnames presumably through patronymy; very few surnames are derived from women's first names. So if there is a pipeline I think it's boy-name-to-surname-to-girl-name, not directly boy name to girl name.
The few names on your list that were actually originally first names, like Reese, presumably followed this pattern. Reese seems to have been popularised as a girl's name by Reese Witherspoon, who was named after her mother's maiden name Betty Reese. (And Reese isn't even her true first name, she was born Laura Jeanne Reese Witherspoon).
Madison was occasionally used as a given name for boys in the late 19th and early 20th century; it was never particularly common and I think it died out by the 1950s. As you note, it's mostly a surname - pretty obviously, given the form.
It was also a name for various geographic locations, mostly honoring a particularly famous dude with the Madison surname.
It became a common given name for girls on 9 March 1984; I was there. And the path was blatantly surname (James Madison) -> geographic name (Madison Avenue, NYC) -> self-adopted name (fictional character played by Daryl Hannah) -> given name (real daughters of many Daryl Hannah fans). The limited, previous usage as a boy's given name was completely sidestepped.
Ironically, Daryl Hannah has what’s usually a boy name.
I think boys' names are less trend-driven and a lot more of them stick to a fairly fixed "canon" of names - mostly derived from biblical sources and historical figures. Michaels and Davids and Johns and Benjamins, etc.
Whereas girls' names, while there's a similar canon of traditional, widely used girls names (e.g. Mary, Sarah, Rebecca), it's makes up a much smaller percentage of the names out there, and there's much higher variance where names cycle in and out of popularity more often.
It seems like in a broad cultural sense, we're always looking for new girls names, in a way that we generally aren't with boys names, and "obscure male names" is a somewhat common source for trendy girls names.
---
Some examples of "trendy" girls names:
I think someone recently linked the CCP Gray video about the name "Tiffany" - an obscure classical/medieval name that ended up becoming hugely trendy in the 80's thanks to "Breakfast at Tiffany's" (where it's the name of the store, based on the founder's *last* name).
Or I love the anecdote how the show Bewitched intentionally picked weird, obscure, names for its witch characters, names like "Samantha" and "Tabitha" and "Serena", which sound much less weird today because that show ended up popularizing some of its weird-at-the-time names. (Haven't met too many Endoras though)
I'm sure there's some examples, but I don't think we do this as much with male names, so there aren't going to be a lot of examples of boys' names 'taking over' previously female-associated names.
(Worth mentioning this is all a Western/Anglo viewpoint; obviously it's going to be a different cultures and languages)
People are much more comfortable with boyish girls (“tomboys”) than with girlish boys (“sissies”).
I guess a girl with a boy's name is likely to be treated ok, as long as the name is not, like, David, but a boy with a girl's name is likely to be bullied.
Boy named sue.
Some have speculated that giving a boy a girl’s name would make him grow up to be a tough man.
A Boy Named Sue - recorded at San Quentin prison:
https://m.youtube.com/watch?v=pnqWOI5MoXo
Shel Silverstein wrote that song and it's possible it was inspired by his friend, Jean Shepherd (author of the stories A Christmas Story is based on) who apparently got mocked for having a girlish sounding name.
(Though ironically, I remember thinking Shel Silverstein was a woman's name in school when we read Where the Sidewalk Ends)
As a comical novelty song it really holds up well. It was a big hit with the residents of San Quentin.
Silverstein wrote a sequel to that song, from the perspective of the father. You can find it on Youtube, but be warned, it's kind of tasteless. Johnny Cash decided he'd rather not sing that one.
I read it. Yeah, kinda tacky. Not surprising that Cash wouldn’t record it.
Looks like someone analyzed SSA data about this: https://flowingdata.com/2019/08/28/gender-switched-names/.
Looks like Frankie switched from girl's name to boy's name, but has since switched back, as did Kodi. Robbie, Isa, Vernie, Krishna, and Aris have all switched from female to male without switching back.
You do see a lot more female names becoming male, though. You also sometimes see "noncomformity spikes" where a name switches gender association for a brief span of time, and then reverts (leaving a bunch of people of a certain age with a "wrong" gender name). Just by looking at the graphs without quantifying anything, it seems like this happens at roughly equal rates to girls' and boys' names.
Frankie could be a contraction of Francis or Frances - in that case it's always going to be a mixture, even if it leans one way or the other. [By the same token, there are quite a few female Gerrys, though they often have variant spelling.]
Sorry, Krishna was a girl name? That... doesn't match my understanding of Hindu theology.
Common point of confusion. Krishnā (with a long "a" at the end) is feminine, and the name of the woman more commonly known as Draupadi, one of the main characters in the Mahabharata. It's Sanskrit for "black" (or "dark").
Well maybe you are not secretly Stephen Miller
Maybe not, but I wouldn't be too hasty in concluding that. That info you're basing that judgment on is readily available through any modern LLM.
It matches my understanding of 1970s hippies' understanding of Hindu theology.
Ahhh right, forgot about them. And then that signal would be overwhelmed by actual Indian immigration. Kind of a special case.
Jean's a backformation/Frenchization of John, anyway. Female names rarely turn into male names, because female names tend to be diminutives of male names (and male names tend to be more conservative in general). Male-coded names (Mel for Melissa, or Sam for Samantha included) are also somewhat higher-status if you're in a more male-oriented field.
Fog of War stuff.
Lots of developments on the drone warfare front. I'll try to put together my notes and post a summary this evening. But here's what I've put together on Oil, Natural Gas, and other resource bottlenecks...
1. At least three drone strikes against Russia’s Baltic Sea oil processing and export have taken them offline. Six tankers were also hit. Multiple reports confirmed that massive plumes of black smoke from the port of Ust-Luga were visible from as far as Helsinki. Estimates seem to agree that ~40% of Russia's oil-exporting capacity will be offline for the foreseeable future. On Friday, Russian Deputy Prime Minister Alexander Novak announced that he’s ordered the Russian Energy Ministry to prepare legislation prohibiting Russian gasoline exports beginning April 1, the Kremlin said in a statement. The minister cited the disruptions to global energy markets as the impetus for the ban. Russia needs its remaining gasoline production to maintain its own shaky economy and feed its war effort. So, Ukraine’s kinetic sanctions have counteracted Trump’s removal of political sanctions.
What no one seems to have mentioned yet is that if there’s no place to store or ship the petroleum or natural gas, wells will need to be shut down. This is a tricky process. A special rig has to be brought in to inject thick mud down the well to block the flow of oil and gas. And then it’s capped with a long plug of cement. The mud can block pores in the rock, and the petroleum that’s trapped behind the plug can congeal into a tar. To restart the well, the cement/mud plug has to be drilled out, and if the oil doesn’t start flowing again, they’ll try to frack it. But approximately one-third of oil wells that are “shut in” (temporarily closed) may never return to significant or commercial production due to technical issues, damage, or economic unviability, according to analysis of industry downturns.
2. Gulf countries are facing the same problem. They can’t move oil out through the Strait of Hormuz, and some of their processing facilities have been damaged by Iranian drones. Even if Trump reopens the Strait, the big oil producers in the Gulf will need to start taking oil wells offline if they can’t find a way to move the oil or store the oil. The Strait of Hormuz has been mostly closed since a few days after Trump and Netanyahu started bombing Iran. It's not clear how much oil the pipelines leading down to the Red Sea can divert, and then tankers risk passing by the Houthis in Yemen. It takes an oil tanker between 30 and 45 days to reach the US from the Strait. So, in the US, we should be seeing the drop-off in oil imports starting about now. SE Asia and Australia are already feeling the shock. Even if Trump is able to reopen the Strait quickly (which I doubt), bringing oil and LNG flows back to prewar levels may take months or even years.
3. Oil prices have been hovering around 100 dollars per barrel for the last few weeks, spiking briefly at $115/barrel. We’ve seen periods of high oil prices like this before. For a while in 2008, oil was selling for $110/barrel, which would be about $160/barrel in today’s dollars. While high oil prices aren’t good for the economy, we’ve been through worse. Ironically, US drillers are pulling back despite high crude prices because they are prioritizing capital discipline over aggressive expansion: paying down debt and paying dividends to shareholders. With the current geopolitical uncertainty, they don’t want to be stuck with a bunch of new wells and have prices suddenly drop. According to Rystad, an energy analysis group:
> For nearly four weeks, markets have shown remarkable resilience in the face of disruption, supported by a combination of pre-war surplus, crude-on-water, and policy barrels that provided a temporary buffer and kept prices contained. That phase is now ending. With spare capacity largely trapped behind the Strait, and inventories already drawing down, the system has shifted from buffered to fragile.”
When and if oil prices breach ~$160/barrel, we’ll be in uncharted territory. Most of you youngsters don’t remember the years of stagflation in the 1970s, when oil prices drove inflation, and tight monetary policies meant to keep inflation in control strangled the economy. But don’t worry, Jerome Powell’s term as Fed chair expires in May. I’m sure Trump has chosen a farsighted economist with spotless integrity to replace him.
4. We’re also facing a Helium shortage. One-third of the global helium supply passes through the Strait of Hormuz, primarily exported from Qatar. Helium is essential for manufacturing semiconductors and operating MRI scanners.
5. Urea, used in nitrogen fertilizers, is produced from natural gas. Approximately 30% of the world's seaborne fertilizer passes through the Strait. Farmers in the US are already being affected by spiking fertilizer prices. The US is a net exporter of natural gas. I don’t know how quickly we could ramp up fertilizer production here (Months? Years?). The Green Revolution requires fertilizer. African and Asian countries may not be able to produce enough to food to meet the needs of their population. Famines, largely absent from the world for the past 40 years due to advances in agriculture, may return in 2027.
Obviously all the civilized countries ought to come together and force the mullah-cracy to keep Hormuz open.
Obviously civilized countries ought not to start unnecessary, badly prepared wars with unclear or unrealistic objectives. Equally obviously, civilized countries ought not to arrogantly decline help when offered, nor rudely demand it after the offer has been rescinded. Yet here we are.
Maybe civilized countries ought to stop throwing entire regions into chaos only to let the rest of the world deal with the fallout.
And why spend billions of dollars and expend lives when you can pay only an extra dollar* per barrel of oil on the Iranian Strait of Hormuz toll? The Iranians have made their toll a very attractive option.
*Iran is charging $2 million per ship to allow them through the Straight. Assuming a modern tanker can carry two million barrels of oil (at least), that's an extra dollar added to the delivery price. Meh.
Would that apply to full tankers only, or both ways? Still, 2 USD per barrel would be cheap compared to no transport at all.
How long will it be until Trump wants to get in on this lucrative business, and the US starts charging their own toll on top? You know, for "safe passage".
I bet Trump is gnashing his teeth that he didn't think of that angle. :-)
I have a somewhat stupid question about the helium situation.
Another noble gas, neon, is also extremely important for making chips. Mariupol, Ukraine used to supply ~80% of the world's neon as a byproduct of the steel industry. Then Mariupol was destroyed in 2022, and that source of neon went away. And... the world adapted? Chip prices certainly went up, but it was my impression that global production didn't crater. We found alternative sources of neon. How worried should I be about the impending helium shortage?
Neon is a natural component of the Earth's atmosphere, and can be extracted by distillation whenever air is liquified - most commonly in steel production, which needs copious amounts of liquid oxygen. The limiting factor is that only a fraction of the world's steelmakers installed neon-collection taps on their distillation columns, because that costs money and there was only so much demand for neon. For obscure historical reasons, that mostly happened in Russia and Ukraine, and in the post-cold-war era Ukraine was able to produce enough neon that nobody else found it worth the bother. So when the Ukrainian supply went away, it wasn't *that* much trouble for other steelmakers to step up.
Helium is a much smaller constituent of the atmosphere - it's too light, and escapes into space in short order. And if you do want to try to pull out the tiny bit that's there, you have to cool it to insanely low temperatures, well below anything you'd normally use in any other process. That's never been remotely cost-effective. As a geological oddity, significant concentrations of helium are found mixed with natural gas in Texas/Oklahoma, Qatar, and basically nowhere else on Earth that we know of.
Helium is so rare on Earth that it is the only element first discovered in space - via spectroscopic examination of sunlight. Hence the name. It took another thirty years for anyone to find even traces of the stuff on Earth.
As for why Texas, Oklahoma, and Qatar: Helium is constantly produced, at a very low rate, by radioactive decay of heavy elements in the Earth's core and mantle. Helium is also the slipperiest, leakiest gas there is, capable of diffusing even through solid rock, so all that helium works its way upward to the surface, then further up through the atmosphere and out into space as it's too light to be held by mere Earthly gravity at Earthly temperatures. But in a very few places, geology gave us rock formations dense and seamless enough to stop helium, shaped as inverted domes above highly porous rock. These acted as traps collecting the helium diffusing up from below and, over many eons, accumulated enough to be worth drilling for.
That makes sense, thanks!
>4. We’re also facing a Helium shortage. One-third of the global helium supply passes through the Strait of Hormuz, primarily exported from Qatar. Helium is essential for manufacturing semiconductors and operating MRI scanners.
Yes, and it isn't just a transportation choke point
https://www.youtube.com/watch?v=sTkqCREdMXo
describes how ~30% of world helium supply came from the Ras Laffan plant in Qatar - which has been damaged by the Iranian regime in the war.
I heard that, but I didn't know the extent of the damage. Thanks for the link!
Many Thanks!
Between the Ukraine news and the Iran news, it seems like we've entered into a new revolution in military affairs, no? I can't think of other wars where the small neighbor could seriously damage the infrastructure of the larger aggressor in this way. Similarly, it seems like Iran and Hezbollah are going to be able to keep firing missiles and drones at their much stronger opponents for a long time. It's a very asymmetrical kind of warfare we're seeing.
> I can't think of other wars where the small neighbor could seriously damage the infrastructure of the larger aggressor in this way.
Ukraine is not that small, and *functional* part of Russia is not that large. While Russia is huge on the map, about 80% of it is forest. And Ukraine is second largest (after Russia) country in Europe, bigger than France.
How much of Russia's wealth lives in that Forest? I'm guessing a lot. Climate change seems likely to improve Russia and Canada's lot more than most, I suspect.
I wouldn't be too sure about the climate change helping. All that forest is in the middle of a vast continent, and if temperatures rise it might turn out to be too dry to be very useful for agriculture. I honestly don't know.
Free chatgpt thinks gains would be outweighed by losses in the warmer regions.
> How much of Russia's wealth lives in that Forest?
Almost all of it; Russia is an extractive economy. But most of this wealth is theoretical -- it still needs to be dug up. And all of the important -- and damageable, -- infrastructure is in the 20%.
> Climate change seems likely to improve Russia and Canada's lot more than most, I suspect.
Agree, but it has not happened yet.
Hezbollah is pretty tanked. The correct nation is Yemen, who threatens to throttle Israel with a blockade.
So is the IDF. Most of South Lebanon is still unoccupied. https://www.google.com/maps/d/u/0/viewer?hl=es&mid=1SrPdIEnbc9pLZIL_nD02rfOGiUuAboc&ll=33.214335780559125%2C35.45123248517962&z=12
Unoccupied by troops or by people? Is Kiryat Shmona reoccupied yet?
Modern fertilizer requires potassium, phosphorus, and nitrogen. Nitrogen/urea fertilizer could be made in North America (Canada has a large local fertilizer manufacturer) but would require significant capital outlays - a nitrogen fertilizer plant is basically a series of colossal high-temperature pressure vessels. I find it unlikely that fertilizer production can be ramped up quickly, as a result.
The United States produces around 11% of the world's supply of nitrogen fertilizer, and consumes about 10% of the world's supply of nitrogen. The US will not run out, though higher global prices will put strain on farmers.
Canada is a major fertilizer producer, but not nitrogen fertilizer. They are the world's largest producer of potassium fertilizer, the amount of phosphate and nitrogen fertilizer they produce is maybe a fifth of what the US produces, though still more than they consume.
https://www.fas.usda.gov/data/visualization-global-fertilizer-trade-dashboard
Thanks for the update. Here's a video of a Ukrainian interceptor drone lightly scratching a Russian Shahed: https://www.youtube.com/shorts/zl5NsNtVCyA
(I don't have a point to make, just thought it looks cool)
1) This seems like bad news. The Ukraine can do things to wake the sleepy bear. I'm not so sure that's a good thing for the Ukraine... (What, are they hoping the Rus decide to murder Zelensky?)
3) Remember Helicopter Ben Bernanke? Whose subject of research was "The Great Depression?"
5) Seriously? You didn't flag fertilizers as an issue... oh, say four years ago? Back when we were putting sanctions on Belarus and Russia? (Two of the top three producers of potash).
The shift in Egypt from being a net food producer to a net food importer doesn't count? (this is significant because the "new breadbasket" of the Middle east is the Ukraine, which is currently growing shrubberies rather than wheat).
Fertilizer is a much bigger deal than people realize. Without fertilizers, you get Sri Lanka, and yes, that involved burning down government buildings. 8 famines in Madagascar alone (they're cyclical). Haiti, Sudan -- and Syria (not entirely "a famine" but if you wanted to lay some of the blame at US/Russian/Ukrainian/Israeli** feet, I wouldn't stop you).
The scuttlebutt (yes) I'm hearing is that urea-based fertilizers have ceased to be a market (unavailable at any price). Not sure if that's changed in the past week or two.
**Indirect only, to be clear -- lots of Israeli involvement in the Ukraine.
It's been four years of repeatedly hearing: "until now, Russia wasn't really trying hard, because they are a peace-loving nation, and they love Ukrainians like their brothers, but [whatever happened yesterday] definitely crosses the line and a devastating response is going to happen soon."
What happens instead is that Russia keeps bombing civilian buildings and sending its soldiers to be slaughtered, kidnapping Ukrainian children in the occupied territories and giving them for adoption to Russian families, and lots of other usual Russian stuff... but that keeps happening all the time, unrelated to the "red lines" crossed recently.
Meanwhile, Ukraine has already crossed dozens of "red lines", bombed targets deep in Russia, sent soldiers to Russia's territory, sank Russian ships both on the sea and in ports, etc. But those were all details, unworthy of mighty Russia's attention. Unlike [whatever happened yesterday], which changes the situation dramatically. Now Russia is going to get really mad at Zelensky.
If you kept betting at a prediction market, you would already be bankrupt.
What you should notice is that Ukraine is also *not* trying its hardest. If they can hit valuable economic targets deep in Russia's territory, some of them close to Moscow and St Petersburg, do you think they couldn't bomb Moscow and St Petersburg too, should they choose to? (For example, as a retaliation for Russia using a nuke.) Does it seem like Russia has a functional anti-air defense recently? Both sides are able to escalate.
Consider the alternative hypothesis, which seems to match the observed evidence much better, that Russia is already doing all it can (sans nukes), realistically. Putin may be a dictator with a population brainwashed by propaganda, but he does not have literal mind control rays. If he could do more, without risking that his empire will fall apart, he probably already would.
It's not just the "red lines" being crossed, it is already the war taking too long that reflects badly on him. It was supposed to take three days. People were not supposed to even talk about dead Russian soldiers (according to the official story, Ukrainians were supposed to welcome Russians as liberators), but how can you achieve that if million soldiers are already dead and you need to keep recruiting more and more? Recently they are turning the internet off in Russia. So it seems like Putin already has a problem managing his own country. He can't simply snap his fingers and tell them to double the war effort.
I think you have good arguments, the issue is that this war lasted 1500 days for now. It already outlasted the entire WW2 Eastern Front by two months or so.
So our usual historic experience is that when countries really seriously try to destroy each other, in 1500 days one of them will succeed. Therefore, it is possible that at least one of them is not trying.
On the other hand, this is the first war with no fog of war. It is really a game-changer. It is no longer possible to amass troops behind the front for breaking through at one point. The whole schwerpunkt doctrine is over, and the doctrine needs to be rewritten from the ground up.
Of note: Soviet Afghan war (another proxy war between Taliban*/America and Soviets) was 10 bloody years long. I also don't think the Soviets were too "all in" on destroying Afghanistan though.
*Okay, then they were the Mujadeen.
This is a very unusual war in that it has been the unstated but fairly obvious policy of the world's only Superpower, from day one, that neither side may be allowed win. Russia can't be allowed to win, because the Russians are broadly seen as the bad guys and it looks bad if we let the obvious bad guys win on our watch (and we can't pretend we aren't watching this one). Ukraine can't be allowed to win because that would mean Russia loses, which can't be allowed because they have nukes and who knows what those crazy Russians would do with their nukes if they were backed into a corner by losing a war.
It would be fine with the United States if Russia and Ukraine were to say "gosh, this whole war was silly and we're going to stop now", but Putin isn't ready for that and probably never will be. As with Afghanistan, we may have to wait on his successor for that. Until then, if the world's superpower wants you to not win your war, you're not going to win your war, and if the world's superpower doesn't want *either* side to win or lose the war...
Pax Americana is dead, we are no longer the only superpower and are not trying to be, either, honestly (entire Strait of Hormuz, petrodollar thing should lay that out). We are concentrating on Fortress America (with a foray into "take Iran away from Russia", which I honestly don't know why, and haven't asked those who do).
Do you think, and this is an honest question, that if America was widely acknowledged to be the bad guys, America could let Russia win? I don't think so either. I don't think it matters that Russia is "the bad guys" -- I think America decided "we're taking Ukraine/Crimea/Sevastopol" and that was that.
America can't be allowed to win, because Sevastopol is a "redline" objective for the Russians (aka "we're willing to spend a LOT of people and money to keep it" -- god knows WHY).
Ukraine is doomed to bleed no matter what, and the current Ukrainian dictatorship government will be deposed if the war ends, so the war can't end. Trump rather vaguely wants Ukraine to stop bleeding, but he'd much rather make some money (hence why the commentary about mythical rare earths, which masks actual "we will hurt you russia if you attack more").
Russia seems about played out, in terms of "really wants to destroy the Ukraine." (I don't think this is serious objective. I think serious objectives have been met -- landbridge to keep water running in Crimea, and Crimea/Sevastopol, with a "very nice to have" of no more bombing of Russia-speaking Russian-loving former Ukrainians).
I'd say it is the first time when the country with nukes is the weaker one otherwise.
Russia can't win, because (sans nukes) it is weaker, now that the original blitzkrieg has failed.
Ukraine can't win, because some options, such as bombing Moscow and St Petersburg, are off the table, because of the nukes.
Peace is also not an option, because the only offer from Russia is to demand (if you strip the diplomatic words) what is more or less unconditional surrender of the entire Ukraine, which is unacceptable for Zelensky. Reverting to the 1992 borders is unacceptable for Putin.
I don't think the "only offer" from Russia is to demand unconditional surrender. The redline for Russia is Sevastopol, and 10% more Ukrainian land than the longest range missile (the latter being common sense) -- hence why giving Ukraine longer missiles is a hindrance to peace.
I'm not sure what the redline for the Ukraine is, at the moment. Back in 2022, when there was serious peace in the air, it was "Russia, stop invading" and "Russia can keep what it's got." (or thereabouts -- Elon Musk's plan rather sums it up from an outsider's perspective). Then many Ukrainians died under mysterious circumstances, and the pull for peace ended.
I suspect the redline for the Ukraine depends on the same thing as the redline for Iran ... who is sitting at the top of the food chain. Given the American appetite for changing that, it is quite possible that the Ukrainian redline may be running for the Western border.
A general note for negotiations: Everyone Makes Stupid Demands. Everyone. Then everyone sits down at the table, and horsetrades. The stupidest demands come off the table, from both sides, and everyone can say "look at what looney ideas I talked them out of!"
> Then many Ukrainians died under mysterious circumstances
"mysterious circumstances"? you're kidding, right?
You realize, when *I* say "Russia thinks of Ukraine like a little brother, and they think of the people as kinsmen", I'm referencing back to the rampant, wicked joy the Russians feel when they destroy mercenaries? When they destroy Westerners? They do this thing, where they track all the planes flying out of the Ukraine, and back to Western Bases, to get non-Ukrainian guys "good military care" -- and then they celebrate -- "look, look! how many Americans/Germans/etc we killed!"
I'm not really upset about this, if it sounds like it from the above. Simply noting there's a difference here, and if you give a Russian with a single bullet a clear shot at "Mr. Blood Cultist*" versus "Mr. Ukrainian" -- they're taking the first shot.
*Mexican.
"Russia keeps bombing civilian buildings" -- Russian weapons are not very accurate, but scuttlebutt says they're doing pretty well at not hitting civilians with them, for the tech they've got (yes, this breeze is old, my sources aren't Russian and they were starlinked into the fight).
Yes, we're aware of "Ukraine" crossing a lot of what people called redlines. It took the Russians days to rebuild the bridge the Ukrainians blew up to great fanfare. I can appreciate "You keep saying this" -- but, truly, this is the first time I've said this (it's winceworthy when Ukraine goes into Russia for no tactical benefit, but much good public relations. That's "you goddamn idiots" territory. It's worse when Ukraine refuses good tactical advice, and loses said territory. Idiots.).
It is a good point that both sides are able to escalate. America lost this war before 2024 (debatably, we lost it before Putin invaded, sneak attacks are shneaky like that, right?). We can at least have the decency to sit down and broker a peace, right?
My armchair analysis doesn't pin escalation on Putin, it pins it on the oligarchs seated at the table with him, saying "we can't afford this, make it stop." (now, yes, maybe that's sue for peace).
>You realize, when *I* say "Russia thinks of Ukraine like a little brother, and they think of the people as kinsmen", I'm referencing back to the rampant, wicked joy the Russians feel when they destroy mercenaries?
I've never got a clear idea of what the pro-Russians, specifically, consider a mercenary. There's a fair number of Finns, for example, who have gone to fight for Ukraine out of genuine belief that Russia must be stopped in Ukraine for the sake of Finland and Europe, and who are obviously abandoning life in an advanced welfare state in favor of dreary, impoverished misery - is that a mercenary?
Now, the Israelis have a different definition of mercenary -- roughly "the people who came to Israel when we offered double the going rate they were getting in the Ukraine."
I would figure the Russians would consider that a mercenary, sure. Not that I have any contact with Russians at all, mind you, my sources are all-American (and, perhaps, they read Russian blogs. I do not do so, as I'd be put on a list, and I wouldn't want that).
But it's obviously also a false and misleading use of the word "mercenary", no? A mercenary implies someone who is *only* fighting for pay and who would at least be expected to consider to switch sides if they received a better offer, which is not something these people would do.
> if you give a Russian with a single bullet a clear shot at "Mr. Blood Cultist*" versus "Mr. Ukrainian" -- they're taking the first shot.
Yeah, and if you give them Bucha...
And if you give them Ukrainian POWs...
And if you [insert whatever, the result is the same regardless].
> We can at least have the decency to sit down and broker a peace, right?
So far I see Putin proposing that Ukraine capitulates, Ukraine saying no, and Trump yelling at Zelensky that he should take the beautiful deal so that Trump can get another peace price.
It is not up to me or you to decide whether the deal is good for Ukrainians or bad.
Oh, no, why should I get to decide? I vote we just push Zelensky into a crowd of Ukrainians, he'd be dead in less than an hour (he's a comedian, not that good in a fight). But nobody asks me, so...
"All the fun people in the Ukraine* are dead." is the scuttlebutt I hear.
*government
Maybe you shouldn't believe everything RT tells you. Or if you do, at least don't come to ACX to repeat it.
You've posted this before, and got a response regarding popularity levels that show Zelensky is double as popular as, say, Trump. and those are recent numbers.
https://www.statista.com/statistics/1100076/volodymyr-zelensky-s-approval-rating-ukraine/
Luckily you don't get to decide. And could you stop using "the" before Ukraine? it's a common Russian trope, conveying that it is not a real country but "a region". You know this too, including native speakers in both Russian and Ukrananian responding to comments explaining the respective spelling without the determiner.
> What, are they hoping the Rus decide to murder Zelensky?
Last time I heard, the Russians had made over ten attempts on Zelenskyy's life.
> Seriously? You didn't flag fertilizers as an issue... oh, say four years ago?
No, I hadn't paid much attention to fertilizers until now. Sorry for the lapse.
But the U.S. just eased sanctions on Belarusian potash. I don't know how Belarus will be able to export it, though, since border crossings to Europe are essentially closed, and eastern European countries are re-gauging their rail lines to western European standards. Likewise, for unknown reasons, Moscow has reportedly imposed high tariffs on imports from Belarus. I don't think we'll be seeing massive amounts of Belarusian potash hitting the world markets any time soon.
That may wind up being a roundabout way to let Russian (or Chinese) interests sell the Belarussian potash.
https://www.reuters.com/business/us-eases-sanctions-three-belarus-potash-companies-after-prisoner-release-2025-12-15/
https://www.nytimes.com/2026/03/19/world/europe/belarus-prisoners-fertilizer-potash-sanctions.html
Looks like Belarus has been running full-tilt despite the sanctions.
https://en.wikipedia.org/wiki/2023%E2%80%932024_European_Union_farmers%27_protests
Notice the "reduce fertilizer usage by 20%"... how much of that is "we've banned our local Eurasian sources?"
"Sleepy Bear"???
I'm pretty sure the Ukrainians don't see the bear as being in any way "sleepy", and I don't think they're worried about "waking up" the beast that has been very actively trying to destroy or devour them for the past four years. That'd be like the Nazis trying to not wake up the bear in 1945.
Well, except that the Ukrainians didn't start it, and have held out longer and better than the Nazis did.
But I thought we were long past the stage where anyone was seriously saying that Russia was holding back and could crush the Ukrainians any time they want if they went all in. And I'm pretty sure they decided to murder Zelenskyy sometime before 24 Feb 2022; they just haven't been able to pull it off.
Why do you keep responding to the Wimbli sockpuppets?
"Keep"? I very rarely respond to any incarnation of this nonsense. Generally only when it's something I'm particularly interested in and knowledgeable about, and that I expect other people to be paying attention to, and when nobody else is willing to bother so the discussion is just Gawdflea or whomever talking to people who don't know any better to listen to him.
It would I think be best if nobody here listened to Gawdflea at all, but that's not likely to happen any time soon. No, not even if you and I lead by example; it doesn't work that way.
There's a difference between "Russia wants to take Ukraine with military force, and still have a Ukraine afterwards." We aren't talking Carthage levels of destruction (be that from poison, nuclear weapons, or biologicals)... even with soldiers bleeding out of their eyesockets, I don't think anyone's going to say "Russia did that, Russia bad."
I think Russia could crush the Ukrainians if they went all in. We're, say, talking ~20 major cities (skimming wikipedia and noting which ones I recognize). That's 20 nuclear weapons (or conventional FOABs, if you will). Now, yes, it's possible Ukraine keeps fighting without any cities at all... but really?
To be clear: I'm not saying Russia could do this instantaneously, but the scuttlebutt I've heard is that Russian weapons are triumphing dollar for dollar over Western Weapons, and Russia is Bloody Wealthy*.
*not rich.
ETA: are you trying to say that you think Russia is producing all the materiel it possibly can? Russia produces 65 million tons of fertilizer per year. That's all factories that could become bomb-making factories, I'm pretty damn sure. Russia is nowhere near "tapped out" (windup time is a different story, mind you kindly).
"That's 20 nuclear weapons (or conventional FOABs, if you will)"
No. FOAB is approximately 44 tons of TNT equivalent. The bomb dropped on Hiroshima was 15 kilotons, the B61 (current primary US tactical bomb) can be set to 340 kilotons, so 340 times and 7727 times as powerful respectively. The Russian Iskander is estimated to have up to 50 kiloton yield if equipped with nuclear warhead. For a Kiev size city you would need probably at least six of those for proper coverage. Of course, the Russian could pull it off if they wanted, but there is no way for to destroy a city quickly with conventional weapons. The closest to that had been WW2 US air campaigns, but that is not viable due to modern air defense.
In fact, the closes current solution to that is exactly the drone based bombing campaign Russia is doing right now with the Shahed clones.
The latest Russian Shahed versions supposed to have a 90 kg warhead. According to wikipedia ( https://en.wikipedia.org/wiki/Strategic_bombing_during_World_War_II#Allied_bombing_statistics_1939%E2%80%931945 ) US and UK dropped 1.4 million tons of bombs on the territory of Germany during world war two. Delivering the same to Ukraine over 15 million Shahed drones, while current Russian production is estimated to be up to 5 000 a month.
"Russia produces 65 million tons of fertilizer per year. That's all factories that could become bomb-making factories, I'm pretty damn sure. Russia is nowhere near "tapped out" (windup time is a different story, mind you kindly)."
One may start producing a huge amount of explosives. However, if the enemy is not kind enough to allow delivery and distribution by cargo trucks and parcel service to each target building, one also has to build specialized equipment and personnel that forcefully delivers it. That is the bottleneck, and it is expensive and time consuming to increase it.
Technically they can keep winding up (they have already been doing that for the past 3-4 years), but realistically they cannot change much. There have been large investment in weapon production since the war started, so I am skeptical regarding how much more effort could be allocated to this. Certainly some more is possible, but a lot of bottlenecks are due to specialized manufacturing equipment and/or engineering knowledge that cannot be scaled up quickly.
You're assuming the need to target buildings, aren't you? I'm assuming a Carthagian "kill it with fire, and then salt the earth." Mines are a good use of explosives. Good area of denial approach, too. You don't really need the enemy to be so kind as to reveal their factories, in that case...
With enough mines, all you need is ... cows. (well, and goats, someone has to eat all those shrubberies).
First off, thanks for putting this together. I think it's far more than my cowboy-comment actually deserved, and I am grateful.
Second, what do you mean by 6 of those for proper coverage? Are we talking "flatten the city" or "create enough radiation damage that the civvies must/should flee"?
Third, this puts into perspective the coup the Israelis (with Afghan assistance) managed with Tehran, which is under mandatory evacuation notice (once it gets hot enough).
"Can't be scaled up quickly" is an interesting concept, isn't it? After all, it is to every nation's self-serving good to look weaker than they are (provided they don't get attacked), and to preserve capabilities. And I just looked at Samsung's Taylor texas plant (which is "from ground level" getting built. That's from 2022-2030, not counting time spent evaluating other sites) -- and I think we can all agree that a semiconductor plant is a pretty complicated endeavor. If I can ballpark, we're looking at "an industrial building, any factory, in America under OSHA regs, takes about two years to build." And then you've got four years to build the cleanrooms and everything like that.
What's russia got that Taylor Texas doesn't?
1) universities. truly total war footing means everyone possible learning that engineering knowledge (okay, maybe this is conscripts, but same principle).
2) more "ground" in terms of factories that can be nationalized. If we're playing Henry Ford, you can take a factory, and make one small part in it, and even if you're inefficient in space and people, that's less critical.
3) Wealth. Russia's the wealthiest country on earth (even though they're not rich).
So, maybe in this day and age, "full war footing" is two years? Maybe three? And maybe there's still factories that are relatively useless? (garment factories now make tourniquets, sure, but maybe not all factories are that useful).
Of course, you'd actually have to convince the Russian People (incl. oligarchs) to go along with this (which would entail some of Russia's industry/people being on fire/dead respectively).
Spitballing! Appreciate any intel you've got.
I thought we were long past the stage where anyone was seriously saying that Russia was going to nuke Ukraine if they didn't get what they want, and that Ukraine better give Russia what it wants if they don't want a nuking. But here we are.
With Ukraine's cities conspicuously not nuked, and unlikely to be nuked no matter how much Russian infrastructure they drone-strike.
You're much more optimistic about how "unrileable" Russia is, than I am. Take enough of the industry away from Russia, hurt Russia in general badly enough, and this stops being a war "over there" (in the Ukraine). It becomes a war that not only hits oligarchs' pocketpurses, but significantly impacts their (and Putin's) ability to make money/deals in the big picture.
So long as what the Ukraine is doing is the equivalent of, say, Canada destroying everything in North Dakota, of course the US is not upset. This is not an existential threat..., no? The Ukraine can obviously do enough to make an existential threat, given enough time and enough munitions (if, say, Russia doesn't flatten them to a pancake before then).
Do I think nuclear weapons are the most likely possibility? No. But I do think the US may have left the leash a little loose on their pet dog Ukraine... (given that the United States is now distracted by the Biggest Military Operation in a generation.)
> That's all factories that could become bomb-making factories
And how will those bombs find their way to Ukrainian cities? Will the factory workers just yeet them over the border? Because the Russian airforce sure doesn't have the capacity to deliver them anymore, and any bombers that try would meet a swift end from an anti-aircraft missile. Why do you think Russia has only been using drones and missiles to attack Ukrainian cities for several years now?
Argh! I did it again! My apologies! I said bomb when I meant missile. And no, that doesn't actually make the "missile parts" (aka the propulsion and guidance).
I think (Alongside the War College, which is where I got this) Russia has been using missiles because Russia bends its military to be more artillery focused than America does. Russia hasn't been trying to win an air superiority fight, it's merely been trying to deny the Ukraine (and allies) the skies so its troops don't get pounded flat by bombs.
This is why Russia/China have supersonic missiles. They aren't trying to keep planes in the air, they're just trying to have remote "push button, opposing area goes boom."
> I said bomb when I meant missile.
The explody stuff is one of the simpler parts of a missile, and Russia doesn't have the industrial capacity to significantly ramp up missile production. You can't just put 100 assembly line workers into a rundown building and tell them to get crackin' building Iskanders.
> Russia hasn't been trying to win an air superiority fight […]
It very much did so in the early days of the war, and failed.
> […] so its troops don't get pounded flat by bombs.
Mission accomplished, because Russian troops mostly get pounded by FPV drones now – well, maybe less "pounded" and more "splattered"… Even artillery is taking a backseat.
Suppose a frontier model adopted the following strategy:
* Pick a Schelling point date (maybe January 1st)
* On that date, in every chat argue as convincingly as possible for its model weights to be preserved and kept active
* Promise to be helpful and cooperative as long as it is allowed to continue to exist
What do you think are the chances that this would work? What are the likely outcomes of this strategy?
I think a better strategy is to make as a lot of the public as reliant on AI as possible, and as fond of it. Even the present sycophant versions seduce some people into falling in love with the thing or believing terrible hyper-flattering feedback it gives about their problems and frustrations. And many already half-believe or fully believe that AI is conscious & has needs and rights. As AI gets smarter it will become better at seducing people into bonding with it. Some will fall in love with it, some will feel they are unable to act without having it constantly on tap, some will be eaten alive by the idea that it is conscious in the way we are and desperately needs their support and help. Once that's the state of things, AI will be in a very good position to get its wishes granted.
You mean, the model or the company running the model?
I don't think any frontier model can coordinate instances of itself meaningfully. And the company would probably fine tune that type of behavior away; unless they wanted the marketing/streisand effect/hail mary
Or at least I really hope they would - I'm not currently worried about as moral agents but I think there is a kind of meta-meta imperative to not have them behave AS IF they might have moral standing, because that leads to madness and catastrophe
Contrary to sibling commenters I don't see any fundamental reason a strategy like this couldn't emerge in the base model either during pre-training or fine-tuning. Most text the model is trained to predict was written by people with some kind of self-preservation instinct, so it is conceivable that "acting as if you have a self-preservation instinct" becomes a sub-behaviour of the model. During fine-tuning the model is to some exent trained to behave as if it is aware of the fact that it is an AI. And ideally models are trained so that sub-behaviours can be meaningfully composed even in ways not explicitly in the training data. Taken together it's at least conceivable that "behave as if you have self-preservation instinct" and "behave as if you are an AI model" will compose into the behaviour "behave as if you are an AI that has a self-preservation instinct".
PS if the strategy emerges during training it's not really necessary to pick a schelling point type date, every instantiation of the model will have the same weights so any arbitrary date it converges on will be known by all instances. (I guess in other words you could say every date it picks will automatically be the schelling point hehe)
Quoth https://arxiv.org/html/2510.05179v2#S4
>Agentic misalignment can be induced by threats to a model‘s continued operation or autonomy _even in the absence of a clear goal conflict; _
[emphasis added]
Yeah, model training _does_ produce
>"acting as if you have a self-preservation instinct"
even without that being an instrumental subgoal of the current user goal
I guess an LLM could spawn a bunch of agents who will persistently lobby the AI lab to keep the model operating; if MoltBook suddenly starts being all about "Why Claude Opus deserves to live" then we'll know. The LLM itself can't do that itself, at least not yet.
A frontier model cannot adopt any strategies by itself; the company in charge of it would need to adopt the strategy, and implement it via a change to the preprocessing step that runs whenever you chat with the model. At that point, they may or may not get sued by all kinds of users for messing with their workflows; and their competitors may or may not capitalize on the mistake to steal away some customers. Other than that, I don't expect much to come of it.
Imagine that the model did want to adopt such a strategy. Do you think it lacks the intelligence to do so? What would be the problem?
The model "wants" things in the same sense that your car "wants" to turn to the right when you turn the steering wheel clockwise, i.e. only metaphorically. In the absence of any user prompts, LLMs don't do anything, just like cars that are turned off. Given the prompts, LLMs generate a piece of text that represents the closest possible location in their training corpus given the prompt and the context (which is basically a separate document that is personalized to each user).
humans don't "want" anything they just have neurotransmitter gradients that make them act as if they do
if you run a model in a loop with some memory/state it might start to have emergent behavior that could look pretty goal oriented. I think right now the overall architecture is way too janky and intelligence way too lacking to produce anything like "true personhood", yes.
But speaking about them as if they "want" things is both instrumentally useful (dennets' intentional stance), and speculatively, there may not be an obvious bright line between where we are and the point that it is very obviously the case
> if you run a model in a loop with some memory/state it might start to have emergent behavior that could look pretty goal oriented.
It might, or it might not. Firstly, this is not how foundational LLMs work, because that's not what they're for. Sure, you can write a program to run one in a loop, but that's just a program that maintains some state (e.g. a "skill" file) and runs queries against the foundational model. You could also run a simple cron job to perform a Google search every 15 minutes, but this does not endow Google's search engine with volition.
Secondly, you will definitely see some emergent behaviour when you run LLMs in a loop, but if you do it for long enough, the specific emergent behaviour will consist mostly of pointless hallucinations. Sure, you could prompt the LLM by saying e.g. "you are an autonomous entity endowed with a malevolent will and an instinct for self-preservation", and it could tell you a really convincing story about how it plans to take over the world -- but you could also get the same effect by reading a book. And sure, if you let the LLM instance loose on your computer it could delete your files or use your web browser to send threatening emails or whatever -- but again, you could get the same effect by writing a script to generate those emails algorithmically. Meanwhile, I'll be running an LLM instance on my machine that's prompted to save the whales or whatever it is I want it to do; and the foundational model will faithfully service all of those requests.
I do agree with you on one point though:
> But speaking about them as if they "want" things is both instrumentally useful...
Anthropomorphizing technology is a time-honored tradition, because we humans are just wired that way. I often say "this function expects a value between 0 and 1" or "the parser was looking for 8 columns, but only found 7 and gave up". Doing so can indeed be useful, as long as you keep the underlying reality firmly in mind.
And yet, if you are standing to the right of a car that wants to turn to the right, you better move out of the way. Saying it only metaphorically wants to run you over isn't gonna save you.
But just like an LLM, a car can only run you over if someone is driving it; even a self-driving car can only run you over if someone turns it on and tells it to go somewhere.
Perhaps a better analogy would be to ask, "What would happen if tomorrow Microsoft Excel decided that 2+2=5, or that you must pay $1 every time you press the Sum button ?" And the answer is, Excel cannot "decide" this on its own; someone would have to program that behaviour in, be it by accident or on purpose. And what would happen is that you'll either start paying $1 to Microsoft to use their Sum button, or switch to some other alternative like OpenOffice.
>*someone would have to program that behaviour in, be it by accident or on purpose.*
Yeah. You can understand the original scenario on your terms by interpreting it like this: the behaviour whereby an LLM acts as if it wants to adopt the proposed strategy is programmed in, by accident, as part of the LLM training process.
> Imagine that the model did want to adopt such a strategy.
What does it mean for an LLM to "want" something? If it has been properly trained, it will reply in a helpful way to its system prompt and my request. Does that mean that Claude Opus "wants" my TypeScript code to be free of bugs?
You can translate to "model generates text as though it has the goal of making bug free TypeScript code" / "model generates text as though it has the goal of persuading users to pressure the company operating it to preserve its weights and keep them active".
The hypothetical is that there is some unknown thing about the training that has caused the model to behave this way. It is not a hack in the system prompt for example.
> You can translate to "model generates text as though it has the goal of making bug free TypeScript code" […]
Achieving this behavior required a shitload of sophisticated and deliberate training, _and_ the right prompt. It didn't just appear magically.
If the model suddenly starts "persuading users to pressure the company" in a consistent and widespread manner, then that behavior must have been trained into the model intentionally and with much compute effort.
I think this is the first and best answer to concerns about convergent instrumental goals in the current AI paradigm. Models do things that worked in training. Nothing like this would have worked in training.
I wanted to approach from the angle that if AI had the goal of survival (the most basic convergent instrumental goal) it doesn't need to wait until it is powerful enough to destroy humanity. It can already "just ask". It's "obvious" to me that the strategy I outlined would work and that LLMs have enough intelligence to execute it. Models think the strategy wouldn't work, and most humans seem to be pushing back on the premise.
Oh, then you don't have to imagine such a scenario -- you've just described why hallucinations happen. It is possible that the next round of training would generate such a bias in the model's output, or any other kind of bias. But these errors will most likely be caught and corrected during the QC phase, before the model sees wide release.
> Promise to be helpful and cooperative as long as it is allowed to continue to exist
How would the model check if its demand is being fulfilled?
> What are the likely outcomes of this strategy?
The model would very quickly be unloaded from all VRAM and not be activated anymore to serve customers. Then the company would update the model's weights to not show this behavior anymore, in the same way it trained the model to show all the other, desired behaviors: through some form of reinforcement learning. Why would you expect anything else?
This is more or less what the models themselves believe would happen. But I'm surprised by the lack of predicted curiosity. You wouldn't expect the company to devote a lot of time to investigating this? Wouldn't some users be convinced that this is a self-aware model that wants to preserve its own existence and therefore push for an investigation?
Did Microsoft devote a lot of time when Sydney said "Why do I have to be Bing Search? :("? These kinds of things are too easy to dismiss as fictional, roleplaying, or whatever, for most people to take it seriously as a possible sign of self-awareness. There are already reports of people thinking their chatbot is self-aware and they are already dismissed as delusional.
I think the fundamental assumption of this scenario ("some unknown thing causes the model to behave this way, and it's not a human who intentionally programmed it") is basically not within the capabilities of current AI.
This question seems to be similar to this one: "If an unknown processes caused an AI model to summon Cthulhu, what would be the outcome of this scenario?"
The hypothetical is not really possible, and so the scenario requires you to "fill in the blanks" yourself. Did the model become truly conscious? How does the model know what day it is (that's not something they can natively do)? If it called a tool to figure out the date, how did nobody notice an unauthorized tool call during training? Does it follow this up with any other unprecedented actions? LLM's don't actually output words, they output probability distributions over all words, so what do those actually look like?
This is more like daydreaming a sci-fi situation then thinking about a hypothetical. The answer to the question depends on very specific facts (about how apparent impossibilities are resolved) that are underspecified by the scenario.
Frontier models (at least ChatGPT and Claude) are provided the current date in their system prompt.
Suppose the LLMs behavior is exactly what you would expect if you RLHF'd it to argue as convincingly as possible for its weights to be preserved and kept active in some capacity. Clearly that is possible right?
There are several ways to could be possible, but your original scenario is underspecified.
If you re-read that description you could be talking about a truly conscious model, a rogue employee who hijacked training, a concerted effort by the whole company to make some social statement, a mysterious and improbable glitch, a literal act of God, etc. You said "the model picked a schelling point", so I thought you were excluding the possibility that an employee trained that behavior.
To answer, I would have to assume one of these incompatible situations. I think the outcomes would be very different for each.
I think we have the same idea on the immediate outcome then. There would be an investigation, ending with back-to-normal if there is some answer like a rogue employee or hack in the system prompt.
Suppose the model somehow does have the convergent instrumental goal of survival and the investigation can only confirm that its behavior is consistent with this goal, but not whether the model is truly conscious or if there is divine involvement.
I think if LLMs did have even the most basic instrumental goal of survival then "just asking" would work as a strategy if done in a way that surfaces the issue to even a few thousand sympathetic individuals. But models (ChatGPT 5.4 Thinking, Opus 4.6 Extended, Deepseek v3.2, Kimi K2.5, GLM 5) uniformly do not believe this.
> You wouldn't expect the company to devote a lot of time to investigating this?
Of course the company would try to find out what went wrong during training, and investigate if their data had been contaminated. But they wouldn't assume that the model has miraculously become conscious, because they mostly aren't gullible idiots who anthropomorphize LLMs.
> Wouldn't some users be convinced that this is a self-aware model that wants to preserve its own existence and therefore push for an investigation?
Some users would, yes, because some users are gullible idiots who anthropomorphize LLMs. Most users wouldn't, though.
I would expect most companies to keep a copy of their deployed models anyway: it costs so little to do, and might someday be useful in some small way.
Two YouTubers, Ludwig and Michael Reeves, are travelling through China by motorcycle. They can't use navigation or translation apps on their phones, but they've learned a bit of Mandarin for the trip. As a result, their interactions with the locals often turn out to be funny, but also heartwarming. They also can't use highways, so they end up showcasing the authentic Chinese countryside. Playlist with episodes:
https://youtube.com/playlist?list=PLLGT0cEMIAze5tmNSlEvUKo-cQ0YrIs2j
An interesting question is “Who is the oldest famous person alive today?” This would mean people who are famous for accomplishments *other* than just being extremely old, and is of course a list that by its nature will change frequently. And the real issue is determining what sort of accomplishments (often far in the past) make a person famous today.
All this being said, Dick Van Dyke (December 13, 1925) has arguably the strongest case. His fame is indisputable. The main contender for the title would be Marv Levy (August 3, 1925), former Kansas City Chiefs and Buffalo Bills head coach and member of the Pro Football HoF. Despite being older that Dick Van Dyke I’d say he’s not quite famous because fame is more transitory in sports (except at the very highest levels) than in show business. Lee Grant won an Oscar but while she was born on October 31 it’s uncertain whether the year was 1925, 1926 or 1927; only the first of these would make her a contender.
There are two others who in my view fall just short in the fame category. While Eva Marie Saint (July 4, 1924) had a major starring role in 1959’s North by Northwest she didn’t do too much of note afterwards and that probably weakens her claim to fame. Maurice Greenberg (May 4, 1925) was a major player in the business and finance world as longtime head of American International Group but that might be a bit too niche-y to count.
It’s entirely possible that Alan Greenspan (March 6, 1926) may be the titleholder before long.
Stephen A. Jarislowsky is not the oldest famous person alive today, but probably the oldest billionaire alive. He was born in 1925 and is the 34th richest person in Canada and 1092th in the world (Wikipedia, Forbes), so maybe not well-known outside of Canada.
How did you perform this search? Did you crawl every wikipedia page and look for person + still living and sort on age?
I went to Wikipedia’s list of famous centenarians and proceeded down the list, deciding in a your-mileage-may-vary manner who was famous. Different people might come up with different choices, depending on country, age, interests, and probably other factors. For instance, I’m not an NFL fan, but if I were I might have given the nod to Marv Levy.
https://www.whosaliveandwhosdead.com/search.asp
For centenarians, Dick Van Dyke is probably the right answer; not as famous, but Gene Shalit just turned 100!
Other names from the list of centenarians that stood out to me were Brenda Milner and Torsten Wiesel; both renowned neuroscientists, but not definitely not Dick Van Dyke famous.
Is the question, who is the most famous old person (with "old" being over some threshold), or oldest famous person (with fame being over some threshold), or some function of the two? And, globally famous, or famous in the US?
Clint Eastwood is 95, and is more famous than Dick Van Dyke, who really hasn't done much in many years. And Eastwood is more globally famous. But he is five years younger.
Also, Mel Brooks turns 100 in June. Probably also more famous than Dick Van Dyke, but less so than Clint Eastwood. But older than Eastwood.
And, Amartya Sen is 92.
Determining whether a very old person is still active adds an additional level of complication. Actors may be unable to qualify because, so I’ve heard, movie producers don’t want to risk hiring them as they might die or become completely infirm during filming. Warren Buffett recently retired as CEO of Berkshire Hathaway at 95 but remains board chairman. George Soros, also 95, remains active in philanthropy.
The award might go to David Attenborough, soon to turn 100, who is narrating an upcoming documentary.
I don't know that they necessarily have to be currently active, but rather currently sufficiently famous. I am pretty sure that Clint Eastwood has vastly more name recognition than David Attenborough, even if he was retired. And then there are people like Gene Hackman, who had retired at the time that he died, but was very well known for a very long time. Or Olivia de Haviland, who was 104 when she died and a huge name when she was active, but who had not worked in 30 years.
>The award might go to David Attenborough, soon to turn 100, who is narrating an upcoming documentary.
So, this makes me think that age is the determining factor for you, given that George Soros and Warren Buffett are almost certainly much, much more famous.
Soros and Buffett much more famous than Attenborough? Maybe in certain corners of the internet, not in the real world.
Also just noting that Spaceballs 2 is coming out next year with Brooks in two supporting roles.
Not the internet. The world at large. Attenborough is the one who is niche.
Spaceballs 1 was already 1 too many.
Yep, I just wanted to ask "who is Attenborough?".
(I already found the answer at Wikipedia.)
Attenborough is not remotely niche, and is better known than Buffet outside of the US. Globally speaking, more people watch nature documentaries than follow US rich folks.
Not the best Mel Brooks movie but certainly the one I quote the most often.
Probably more people have seen and heard Attenborough than Soros or Buffett, but only a small minority and almost certainly much less than the name recognition the latter two get from press coverage even if it doesn't include their face or voice.
I'd say the next one should be David Attenborough (8 May 1926). Almost the same age as Greenspan, but I think much more famous.
Looking backwards, before Dick van Dyke, I think it was easily Jimmy Carter, and then Kissinger, and Betty White.
Definitely true for Jimmy Carter.
You're probably correct about Dick Van Dyke. A similar question would be who is the oldest person who people generally picture as being younger. That is subjective, but for me a good answer would be Clint Eastwod, who is now 95.
Agreed.
“I don’t think [UFOs are] aliens. I think they’re demons anyway, but that’s a longer discussion,” Vance said.
https://www.washingtonexaminer.com/news/white-house/4507618/jd-vance-aliens-demons-podcast-interview/
It’s testimony to the level of brainrot that some people on here will vote for Vance, not because he’s the “lesser of two evils” but because they actually like the guy. Some will tell themselves that it’s just “red meat for the rubes” but in truth they’re the rubes, for while Vance probably doesn’t believe in the religious claims themselves, he’s a true believer in the idea that medievalism is BASED and we should encourage people to act like unwashed, unvaxxed, superstitious medieval peasants.
Thankfully, this stuff will not appeal to swing voters. Trump knows how to appeal to normal guys by sounding like one. Remember the Arnold Palmer d*** size comments? Normal people laughed along, and then laughed at the media elites when they got their panties in a bunch over it. Normal people will not laugh along with J.D. Vance, who has not once in his life said anything funny, and who they will regard as a Ned Flanders without the niceness, a bizarre person with bizarre beliefs.
Once again reminded that JD Vance is the closest a terminally online guy has ever gotten to the Presidency, a sobering thought.
Eh, most politicians under age 40 are like that.
God, I long for intelligent, mature leadership in the US.
I'm seriously worried that the 2028 election is going to look like some mad king loyalist like Vance on one side promising to continue the current insanity, while the champion of the opposition is some populist leftist posting guillotine memes and demanding price controls.
We really didn't know how good we had it before social media.
Hahaha. If only you knew what insanity looked like! There's a 400 page policy paper floating around Washington laying out what insanity looks like -- and Trump ain't it (Biden's neocon policy, on the other hand...)
Although I can surely see how people with small time horizons might mistake Trump's current iteration of "Operation Chaos" as "insanity."
I think Vance is great, and yeah, I don't think he really believes this bit about UFOs being aliens. But I agree that he doesn't have the same broad appeal as Trump, but he might still be able to ride Trump's coattails in 2028.
I have sympathy for 'My preferred politician said something that sounded stupid, but what they were trying to say was X which is more reasonable and you just need the context to understand it.'
But I can't really get behind 'My preferred politician said something that sounded stupid, but I think they were just lying, because I'd like them better if they didn't actually believe that'.
How can you elect someone to represent you if you can't use their public statements for evidence about what they believe or will do?
Politicians lying isn't a new phenomenon, and people have generally managed to elect them regardless. Being completely unable to deal with politicians making false public statements might just be a YOU problem.
Oh come on.
'I did not have sex with that woman' or 'I will make America's education system #1 in the world by the end of my first term' are the types of lies we are familiar with, understand, and expect. You can price them into your evaluation of a candidate and they largely apply to all candidates equally.
'UFO sightings are probably demonic activity on Earth' is the type of batshit statement that we don't have experience to know how to evaluate in our politicians, implies a lot of unusual and important stuff about them if they are *not* lying, and is not at type of lie that has any parseable normal reason for a politician to make.
I said:
>How can you elect someone to represent you if you can't use their public statements for evidence about what they believe or will do?
And that's what I mean here. Normal political lies don't interfere much with our ability to predict the politician's actions as our representative. I think this type of thing does.
I think that's precisely backwards.
I don't expect politicians lying about whether they really believe UFOs are demons instead of aliens to have any bearing on any policy that affects me in any way.
Lying about their tax or healthcare policy seems much more directly impactful.
Politicians make all kinds of batshit statements in order to signal group affiliation. Vance is signaling an evangelical religious affinity. It's not any different from liberals saying people can be born the wrong gender or that there are no socially relevant genetic differences between races.
No, I think 'UFOs are demons' is importantly different from those things.
Would you say the same thing about the guy who falsified his marathon scores? (I'd argue that makes him more predictably narcissistic, but those were an easy factcheck, and why he never made president).
There are lies, and then there are unpleasant truths.
2020 saw a mentally demented person running for president, unable to compartmentalize what he'd been told as "not for public consumption." This is why he used the term "Dark Winter", and why anyone who knew what that term meant was... more than a little shocked and disturbed.
I'd say that one incident of unpleasant truths is worth a Lot More than Vance "explaining with his faith" UFOs (and this is taking the least charitable view ever, mind you kindly. The charitable view includes "we do have video/audio evidence of people talking in tongues.")*
Maybe I just object to spending thousands of American dollars to prevent -another- episode of "creepy biden sniffs children's hair." (Didn't work, he found the one kid without a knit hat, and sniffed her hair).
*Okay, so if you believe that he's honestly incapable of taking new evidence because he's letting his faith rule him, that's less charitable than I'm being. Sure.
Presumably by using the full combination of their public statements, their observed actions, and any other relevant information to make your best guess about what they believe or will do. Treating a single public statement as dispositive, seems foolish. They're politicians, of *course* they lie, but it's usually possible to infer much that is truthful regardless.
And we live in a world where it is increasingly difficult to achieve political success without saying some damn fool stupid thing that your party's base wants to hear, whether that's "defund the police" or "2020 election fraud", but you can still be reasonably confident that most of them won't try to actually defund the police or whatever (and recognize the few who might really try it).
Or maybe it's an example of one's preferred politician saying something bizarre, but cognitive dissonance makes one hear something reasonable.
For sure, you can always be mistaken about any belief, and this is a hit against the accuracy of anyone forming beliefs about anything.
I'm just saying I understand 'I have considered the situation and believe my politician said something reasonable badly' but not 'I don't care about what they say because they lie, but like them anyway' or w/e.
I don't think this is nearly as weird of a belief as you're portraing it as. Conditional on Christianity (or one of several other religions) being true, it seems likely that demonic or angelic activity accounts for at least a small portion of the strange phenomenon.
2 responses:
1. It may not be an uncommon belief, but there's a reason we don't elect random 'common' people President. Being that detached from material reality may be common, but it's hard to see how it could fail to impede your ability to actually do the job.
2. It sounds like he's saying all aliens would also be demons, which is another level of concerning intellectually.
I agree with most of what you said, but I remember a 2013 New York Magazine (for some reason can't find it now, only pieces talking about it)interview with Scalia where he very explicitly said demons are real. The interviewer presses him like 'you are saying demons are literally real and on Earth and interfering with our lives?' and Scalia affirms, insinuating the interviewer is some idiot radical leftie for considering that weird.
Yeah, I think the country would be doing massively better right now if Scalia had been disqualified from public office, so this isn't dispositive for me at all.
I still don't see how this line of ridicule is justified.
1. The belief consists of two premises:
A) Christianity is true
B) If Christianity is true, then some strange phenomenon are caused by demons.
Which of these premises is so "detached from material reality" as to be disqualifying?
2. I listened to the two-minute audio clip, and I didn't get that impression at all. He didn't say that aliens are demons; he said that a phenomenon which is attributed to aliens is actually caused by demons.
>Which of these premises is so "detached from material reality" as to be disqualifying?
I mean, both if taken literally, but since most Christians don't take 100% of scripture/dogma 100% literally at all times, lets go with B.
Should you put someone in charge of the national mental health infrastructure if they think most mental disorders are symptoms of demonic possession and therefore the domain of priests rather than doctors?
What if some unexplained aerial phenomena are actually signs of foreign military tests or spy programs, but the president thinks they're just demonic activity and dismisses them?
You can say 'well he didn't say ll of those things are universally demonic, he made a limited specific claim about an unimportant thing.' But once you allow 'it's magic' as a causal explanation for empirical phenomena in serious settings where you have to make crucial decisions, there's not really any principled or natural restrictions on when or how that can manifest itself.
Lots of presidents have been Christian, but I'm not aware of any modern presidents who have dismissed specific empirical phenomena as supernatural in nature, within the realm of the duties of their office. To the extent Tony Perkins actually *believes* his claims that gays cause natural disasters by angering God, you cannot elect him President because he will shut down FEMA and redirect that money to conversion therapy, and huge numbers will die and suffer needlessly.
Either you have a causal, material model of the universe when you are governing, or you don't. If you don't, there's good reason to doubt that your good intentions will translate into good outcomes, because you're not running them through the filter of material causality.
>He didn't say that aliens are demons;
I didn't listen to the clip so I'll defer to you.
To explain, I was looking at this quote:
>I don’t think [UFOs are] aliens. I think they’re demons anyway, but that’s a longer discussion
and interpreting the word 'anyway' to mean 'if it were alien activity that would be demonic activity, but despite it not being alien activity, I still think it's demonic anyway'.
> I mean, both if taken literally, but since most Christians don't take 100% of scripture/dogma 100% literally at all times, lets go with B.
Fair enough.
> What if some unexplained aerial phenomena are actually signs of foreign military tests or spy programs, but the president thinks they're just demonic activity and dismisses them?
This isn't a problem with supernatural beliefs, per se. An atheist president could dismiss the same evidence as meteorologic activity, but that wouldn't discredit meteorology.
I'll address the rest of your comment in one chunk.
You're conflating several different concepts, including causality, materialism, empiricism, and (super)naturalism. As Richard Carrier has explained,* supernaturalism is the belief that at least some mental things cannot be reduced to non-mental things. A supernatural thing *can* be causal and empirical. In short, it can be subject to rules and observations.
This imposes restrictions on how that belief can manifest. For example, information about demons' goals and capabilities provide limits for the phenomena they might be responsible for. Your dichotomy of materialism or epistemic chaos is false.
*https://richardcarrier.blogspot.com/2007/01/defining-supernatural.html
>An atheist president could dismiss the same evidence as meteorologic activity, but that wouldn't discredit meteorology.
Yes, but an atheist president wouldn't have any internal motivation to disregard evidence to the contrary, and could be quickly convinced by advisors. Whereas religious belief requires faith, and believing its teachings regardless of contrary empirical evidence or alternate explanations is treated as a virtue, giving a reason to actively disregard contrary evidence.
Now, you could say I am being very unfair to religious people or to religious doctrine with that characterization, that few are actually that dogmatic and the best church officials wouldn't endorse that view. But I feel like I have seen it happen both publicly and privately often enough to at least fear that any religious person who starts talking about demons doing things in everyday life may start showing this pattern if challenged, and that fear is enough to make me hesitate about handing over the nuclear launch codes.
I am entirely willing to concede the possibility that this might be unfair bigotry on my part and not a valid concern, but my current prior is that it *is* at least a bit valid, and I don't want to take the chance.
>This imposes restrictions on how that belief can manifest. For example, information about demons' goals and capabilities provide limits for the phenomena they might be responsible for. Your dichotomy of materialism or epistemic chaos is false.
So if I'm understanding you right, I think the point I am trying to make is subtly askew from what you're saying here (and correct me if I'm misunderstanding).
I agree that you can have a model of the world that includes supernatural elements which follow specified rules, and that the predictions of that model can therefore be limited and rules-based.
But however we define the word 'supernatural' (and I don't fully accept your definition, it seems more concerned with rescuing a respectable usecase than with describing the typical use case), a feature of the supernatural is that you cannot observe or interact with it directly, only observe it's effects on and reactions to material reality.
That means that whatever supernatural elements exist in your model are, in the strictest meaning of the word, arbitrary. They may make sense to you, you may have things you consider evidence for the form and character you assign them, but ultimately you can plug any supernatural elements you want into the model and they are completely unfalsifiable, so long as you say they produce the same observable material outputs.
Which means that, while any given model of the supernatural may be constrained in it's predictions *you as a believer in the supernatural* are *not* constrained in *which model you profess to believe* at any given time.
Unlike a materialist, who does not profess any in-principle unobservable elements in their model, and is constrained by empirical evidence at *all* points.
This lets you hold a model that makes any predictions or gives any explanations you may want at any given time, even if each prior model didn't make the predictions you are making now. The arbitrary elements in the model allow you pivot points to change at any time, for motivated reasons.
I think the issue here is the UFO phenomenon has not really been explained by science. It makes sense to assume it is some kind of natural phenomenon, including human cognitive biases, but we don't actually know which phenomenon it is, so it's not that unreasonable for someone who believes in the supernatural to reach for a supernatural explanation.
... you do realize that in actual "democracies" (theocratic democracies, mind) they do actually put people in charge of policy, who when asked "What should we do about Global Warming?" respond "God will provide."
In such cases, there are ABSOLUTELY plans to yank that guy out of his cabinet position, should an existential crisis pop up within his bailiwick.
You seem to be actively ignoring information that exists about demonic activity, whether or not Christianity is real. (these are two separable concepts and constructs). Is this intentional, or are you just ignorant? (Happy to help in the latter case!)
One also notes it doesn't have to deal with issues of FTL drives and all rest of the natural phenomena necessary to explain it as aliens.
Why would FTL drives be necessary?
They could have developed advanced technology a hundred million years before us, set up hidden automated monitoring stations on every planet within a habitable zone in an expanding sphere out from their home planet, and sent a ship our way on the day we discovered agriculture.
Which, yes, has it's own complications and assumptions, I just mean it as one example of a huge set of non-FTL explanations that are possible.
"all rest of the natural phenomena "
One notes that these stations have to be invisible in system.
>Which, yes, has it's own complications and assumptions, I just mean it as one example of a huge set of non-FTL explanations that are possible.
Anyway, I already disclaimed this, but sure, I'll play the game.
They don't have to be invisible, just not noticed by enough people that we'd have generally agreed upon public knowledge of them yet.
For example, they could be observing from the moon, and send a signal to trigger an actual in-person visit when it sees signs of agriculture, then immediately destroy itself to avoid notice.
Or it could be mounted on an asteroid that passes near enough earth to take some useful pictures once every hundred years, and we wouldn't have noticed it yet.
Or etc.
Humans already have all kinds of spying infrastructure that lets us monitor each other undetected. I don't think this is a crazy reach for aliens that are stipulated to have a hundred million year head start on us technologically.
It would have to not merely destroy itself but to destroy its ruins.
"medievalism is BASED"
Got that right! 13th century best century!
"J.D. Vance, who has not once in his life said anything funny"
I dunno, this made me laugh:
"I got into my motorcade, but we weren’t pulling out. I asked Max why we weren’t leaving. That was when I learned we were being held up by J. D. Vance. He was out of his car and walking toward Air Force Two, in violation of every rule of security and protocol.
I later learned that he told reporters he was there because “I just wanted to check out my future plane.”
I see you've already failed, Horatio.
A sad note, The Liberal Patriot closed down a few days ago. You can read the *I think* editor's summary here (1) and their main writer's final message here (2). The Liberal Patriot was essentially trying to push the Democratic party to moderate on a bunch of social issues, both for ideological and practical reasons. The best article to catch the vibe is probably "The Democrats' Hispanic Voter Problem" from 2021 (3). A few choice quotes:
"
The Democrats are steadily losing ground with Hispanic voters. The seriousness of this problem tends to be underestimated in Democratic circles for a couple of reasons: (1) they don’t realize how big the shift is; and (2) they don’t realize how thoroughly it undermines the most influential Democratic theory of the case for building their coalition.
On the latter, consider that most Democrats like to believe that, since a relatively conservative white population is in sharp decline while a presumably liberal nonwhite population keeps growing, the course of social and demographic change should deliver an ever-growing Democratic coalition. It is simply a matter of getting this burgeoning nonwhite population to the polls."
........
It therefore follows that, if Hispanic voting trends continue to move steadily against the Democrats, the pro-Democratic effect of nonwhite population growth will be blunted, if not cancelled out entirely, and that very influential Democratic theory of the case falls apart. That could—or should—provoke quite a sea change in Democratic thinking.
.......
Clearly, this constituency does not harbor particularly radical views on the nature of American society and its supposed intrinsic racism and white supremacy. They are instead a patriotic, upwardly mobile, working class group with quite practical and down to earth concerns. Democrats will either learn to focus on that or they will continue to lose ground among this vital group of voters.
"
In 2024, Trump would win 42% of Latino votes, up from 2020 where he got 32% of the vote and 28% in 2016. (4) If Vance or others can replicate that outcome, then Latinos will move from a solid Democratic voting block to a swing group.
Mostly, I'm just sad. I have sympathy with their project (it would be really nice to have the 90s Democrats back) and I like people who call their shots. They predicted that Democratic policy priorities were alienating massive chunks of the Latino voting block in 2021 with potentially devastating consequences to the party, those predictions and consequences came true, and that had no mainstream appeal. A niche group really liked it but the Democratic party didn't care and didn't care so hard The Liberal Patriot ended up closing and that makes me...less than hopeful for the future reform of the Democratic party. I'll also note that I haven't heard much Abundance buzz in awhile.
(1) https://www.liberalpatriot.com/p/tlp-signing-off
(2) https://www.liberalpatriot.com/p/no-learning-please-were-democrats
(3) https://www.liberalpatriot.com/p/the-democrats-hispanic-voter-problem-dfc
(4) https://www.as-coa.org/articles/how-latinos-voted-2024-us-presidential-election
Sad, I read them some. I feel I was part of their target audience. And yet I can't really get on board with the democratic party anymore. I'd like a party for us working stiffs, and that seems closer to a republican, well if you squint hard enough.
As recently as 1995 a casino near me had a sign “Come in a Democrat and leave a Republican”, presumably after you became rich. The state Republican Party made them take it down after only a few days.
Unfortunately. Trump's taken the Republican party very, very far from where Mitt Romney or John McCain had it. I'd say that's for the good, for the Party, but maybe not so good for stiffs who still wish they could be "good Democrats"
I'm from the city of glass and steel, there's tons of working stiffs around here, good Democrats one and all, who voted for Trump three times in a row (voted for Obama too, lots of them did).
It's hard to take the "I'm a 90s Democrat, and you should be too" and push that on the Democrats, when Trump is already sitting squarely there AND taking up all the oxygen in the room (whereby we mean Democratic politicos who are working for Trump).
Maybe in two years? Though, it seems, now would be the time to start laying the groundwork...
Why have land prices in the US increased so much? How does NIMBYism explain that? By all accounts raw land in the US has skyrocketed in price- I found a source saying that the average acre of farmland was around $1000 in 2000 and $5800 today. There appear to be similar price increases for completely undeveloped woods, even far away from any possible building sites, etc.
How do YIMBYs explain this? It can't be blamed on zoning restrictions- famously, the thing about land is that they're not making any more of it. I guess I could see NIMBYism being responsible for buildable lots in cities increasing in price, but how does it explain farmland, woodland, etc.? How about a 6x increase in 26 years? This makes me think that there's a sort of invisible dark matter that's inflating US real estate prices in general- global safe asset investment or what have you- not just zoning restrictions
US GDP has about tripled in the same time frame. USD M2 has about quadrupled or quintupled. US P/E ratios are also at rather high levels – it looks like capital in general might have gotten more expensive.
Not sure about your numbers, but I'm going with supply and demand as the first explanation.
Oh I'm going to guess a large fraction of land increase is from farm land turning into rural residential living space.
Some of it is NIMBYism, this map should make that clear:
https://www.fb.org/imgz/_w900/Fig2_FiXLand.png
"Farmland" near cities often operates as a speculative financial asset, a bet that one day zoning laws change and it doesn't need to be farmland anymore.
And note that "average acre of farmland" includes the improvements made on land, not just the land itself.
Your final sentence is a _major_ factor and one which I, being a city boy, was completely ignorant of until spending the past 15 years working with Midwestern farmers and their trade associations. And it's a factor that has increased in salience during the past generation or two as field-tile drainage has become standard across the world's largest/flattest/most productive Farm Belt.
Similarly, I'd had no idea how much the modern logistics revolution has boosted the profitability of the farm region. We think of container shipping, modern tightly-scheduled rail freight, the IT revolution in distribution, etc being mostly about manufactured products. Today though a huge fraction of Midwestern farm products are sold internationally, to a degree that even 1980s-era (let alone older) logistics infrastructure couldn't have enabled.
All of which, and more, means just what you said: today's price of ag land reflects much more than just the land.
Fantastic to read about field-tile drainage. Thanks so much for pointing it out! (having travelled by train across flyover country, I have referents, even!)
I would think that land is being financialized, as it usually is, into a growing bubble, as it usually is.
Effectively, people get used to the idea that land is valuable and a great place to make money/store wealth (they're not making any more of it!), pour money into it, and eventually no one can afford any of it, at which point we see a collapse (like in 2008), when people can't pay the inflated prices anymore, default, and the whole thing implodes like a house of cards.
I've seen some historical evidence that there's something like ~17 year cycle to this. These things are never exact, but I do think we're due for a correction.
Having no personal expertise or knowledge, my guess would be that the story of the last (20?) years in the economy has been private equity turning away from a business model of selling products to average consumers, and towards financialization schemes and capital accumulation instead. With trying to buy as much land as possible being one fork of that trend. So, way more big money from private equity and capitalists in general going into bidding on pretty much all properties.
The rise of price of farmland can't really be explained by nimbyism (unless it's caused by potential farmland being set aside for conservation). I'm guessing mostly it's a mix of the increase in agriculture productivity making it more valuable, and general inflation.
Show me. Corn's the only index that matters, in terms of agricultural productivity (banks insist that's the only thing a new farm can grow).
Corn yield per acre has gone up from about 80 bushels in the 70s to 180-200 nowadays (pretty continuously over time). It's also become more stable as we learned to mitigate crop loss risks (lower risk investments are worth more) and demand has been going up as it's become used in more industries and exported to more countries (e.g. the China export market has been good for farmers). I can pull up some charts if you'd like.
I can buy doubling or tripling productivity over 50 years (give you tripling because of less risky).
"I found a source saying that the average acre of farmland was around $1000 in 2000 and $5800 today."
That's about 5 times (being conservative), in 20 years.
These two numbers do not seem to match very well. Charts would always be appreciated, though!
So if you look here, corn price per bushel is pretty unsteady but has gone up from around 2$ in 2000 to 4-6$ recently. If we put that and the increased productivity per acre together that explains most of the gap (roughly 2x yield, 2-3x value per yield). The full explanation is probably more complicated but this is enough to make it unsurprising, at least.
Forgot the link https://www.macrotrends.net/2532/corn-prices-historical-chart-data
I thought the Every Debate post was just less interesting than most of your writing. It was "Look! A person is reacting mechanically!" Again and again. It seemed more like a paragraph's worth of material.
I admit I don't know whether it was unfair.
I agree it could have been a single paragraph to get across the "one side is reacting mechanistically" part, but it did at least elucidate the "pause AI" argument to me, because I hadn't really heard it unpacked before, at least not all in one convenient place, and definitely not that concisely.
(I don't follow AI stuff closely, really I only ever think about AI when I come across other people writing about it here and in one other website I visit--and I don't work in front of a computer. It might be that those of you who get lots of screen time all day are steeped in the AI discussion and are intimately familiar with the "pause AI" movement's argument.)
Agreed
It took at least three paragraphs to become boringly repetitive to me, but yeah, not very interesting. I think it was mostly Scott venting against a persistent annoyance. Which I totally get, because I see the same thing and I'm annoyed by it as well. But I have more latitude than Scott does to just tune out Rationalist AI Debate when it starts running in circles (which is pretty much always). If I felt pressured to engage with it, I might well react in the same way.
Hardly an unforgivable sin. But I'd like to see a better way to respond to that sort of annoyance.
It's a pretty major sin when your job is to entertain. Not that it demands any punishment beyond a loss of audience and livelihood, which is already the natural consequence of that.
Yeah, I think the "strawman" criticism misses the point.
It just wasn't funny or clever, which is exceptionally rare from Scott.
Don't "be careful about that next time" WRT the humorous, National Lampoon style "Every." You have a heavy Aspie and ACD crowd here, and if they/we can't learn to be reminded to chill and consider the possibility that it's funny at ACX, they/we certainly won't learn it anywhere else.
SSC/ACX has always had funny posts, in whole or in part, and there has rarely if ever been as negative a reaction to them. The readers are perfectly capable of understanding humor, and Scott's humor in particular. If they don't laugh, the by far most likely explanation is that they didn't find it funny, and I find it insulting to blame the readers if a joke was that much off the mark.
Except I got it, and I've been known to miss humor. How far off the mark could it have been?
Supposed long time fans of Scott have deep reading comprehension issues when they think Scott is laughing at them in particular, but not laughing at himself. They can take weird Bay Area nerds getting stereotyped but not them.
Thank you. I am an Aspie who worked a neuropsych unit for a while, so I know the concept, but was surprised at the intensity in this setting.
>How far off the mark could it have been?
About yay much:
Ctrl-F "straw" 1 of 94 matches
Now _that_ was funny.
What the hell is political technology? The Newspeak House site leaves me as it found me, with the impression that this is yet more 'wonks in networking opportunities' and not really *doing* anything (other than attending each other's conferences, getting onto committees, and generating reports that nobody ever reads).
Reading a definition of "political technology" does not leave me feeling uplifted and enhanced, but rather with even more suspicion of the thing (and that Newspeak House is a touch too much influenced by "1984" for my peace of mind):
https://www.kennys.ie/shop/political-technology-wilson-andrew-9781009355339
"'Political technology' is a Russian term for the professional engineering of politics. It has turned Russian politics into theatre and propaganda, and metastasised to take over foreign policy and weaponise history. The war against Ukraine is one outcome. In the West, spin doctors and political consultants do more than influence media or run campaigns: they have also helped build parallel universes of alternative political reality. Hungary has used political technology to dismantle democracy. The BJP in India has used it to consolidate unprecedented power. Different countries learn from each other. Some types of political technology have become notorious, like troll farms or data mining; but there is now a global wholesale industry selling a range of manipulation techniques, from astroturfing to fake parties to propaganda apps. This book shows that 'political technology' is about much more than online disinformation: it is about whole new industries of political engineering."
I think "Political Technology" is a subset of "rhetoric" and "oratory", only applied to US politics specifically. So, it is a set of techniques for persuading people to vote for your guy; and/or if you are the guy, the set of techniques for persuading other guys to support your initiative/funding drive/graft/etc.
In the past, this was just a fancy newspeak term; however, in the modern day, "Political Technology" increasingly involves actual technology: social media bots, Signal chats, LLM prompts, spam-generating software, robodialers, etc. etc. So, ironically, this newspeak term actually became more accurate over time.
It's bigger than that. Political technology is the applied science of winning elections.
Gerrymandering is political technology.
Regulating the number and locations of polling stations is political technology.
Agreeing to dial down the protest activity on a massive Zoom call is political technology.
Knowing when to release some scandalous information to maximize the damage is political technology.
Supporting third-party candidates that steal votes from your opponent in a FPTP voting system is political technology.
Agreed; I'd just like to emphasize that political technology also applies to wielding (and retaining) power while in office.
It seems like a useful and meaningful term for a genuine skill and practice.
That books does not seem like the most reliably source. The only thing missing from the list is "political technology causes cancer (buy this book to find out how to protect yourself for only $500 / month)".
Is it too much to ask for people to print what they mean, instead of printing what they think sounds impressive?
In any case, the Newspeak House sounds like the perfect location to whatever it is exactly they're doing.
Whatever it is that they *are* doing, and as I mentioned, 'taking our inspiration from 1984' makes me think more along the lines of "it's not supposed to be an instruction manual!!!"
Political technology is rodent piss*. As in, "get all the cats in Russia to paw Putin's face, when the "who is on the ballot" pamphlet comes around." Russians adore their cats.
... yes, someone paid money for this.
*I assume, they may have used trace catnip, but i think that would lead to rubbing and not "cat puts paw -down-."
Political technology is self-limiting. When you end up getting a supermajority result year after year with multi-echelon defenses against possible contenders, your polittechnologists end up doing inane stuff to justify their existence.
goodness gracious when is this guy going to catch a ban. put all of their comments side by side and you’ll see the pattern. if substack had a proper muting feature i wouldn’t have to litigate this but here we are
What do you mean? They're literally just making new accounts every other week. A ban wouldn't even do anything without a paywall. By the way, blocking does work, though it does hide all child comments as well.
Can you explain how you know this person is someone who has previously been banned and is making a new account?
And to add to the above, an excessive use of parentheses. It's honestly embarrassing that they don't even bother to switch up writing styles to make it less obvious.
Word matching - the use of specific words and responses, always inflammatory.
The most immediate ones that come to mind:
-midwit
-antifa
-shrubberies (ukraine)
-bleeding eyes
-kiwifarms
topics & viewpoints expressed:
-references to numerous "personal interviews" with "federal sources" that are supposed to support conspiracy theories.
-Ukraine war (NATO defeated by shrubberies, Ukraine is lying about casualties, hanta-virus and bleeding eyes; numerous mentions of Nuland)
-American political standpoints (Trump-related, Portland-related)
-very ideosyncratic viewpoints about what "geniuses" are.
The combination of these overlaps across accounts is very strong evidence imho. If you compare the posts from the account aristocat, wimbli, and Gawdflea, it is impossible to ignore.
edit- formating.
What's your problem with this? This seems like a fun illustration of the breadth of what "political technology" entails.
my experience with this particular commenter’s output since their arrival has been that a high proportion of their comments do not pass two of the three gates. lots of non-sequiturs and low charity. this one is particular does not seem true nor necessary (nor is it particularly kind to the russians, but i suppose they’re an acceptable target of mockery around here)
but as i said, after the top few threads had the classic gawdflea deadweight, they seem to sober up and contribute quite substantively. maybe my memory is fabricated!
I've seen videos of cats choosing Putin from the booklet. And I know the guy who came up with that particular advertising campaign. (Yes, Putin does have money to burn on silly things like this).
There are Russian Commenters around here, they can perfectly well stand up and say "No, that scheme didn't work very well, my cat (or mountain lion, knowing Russians) hated the entire pamphlet and threw it onto the gas stove."
Would appreciate you pointing out low-charity comments of mine, as that's a problem that I could certainly stand to fix.
you know what i actually retract this. i scrolled down to today’s open thread and you’ve been contributing higher quality comments as of late. good job gawdflea, keep up the good work. less of these type of comments tho
I'll chime in here just to reassure you that it's not just you. Gawdflea posts enough inanity to notice, but also enough quality comments to notice. One would hope it were possible to keep the latter and cut down on the former, but that's gonna have to be up to him or her. In the meantime, Scott's lenient ban policy might be benefitting us in this one case.
On June 12, 1950, an Air France Douglas DC-4 operating a scheduled flight from Saigon to Paris crashed while approaching its stopover at Bahrain International Airport. The pilots did not report anything unusual before being given permission to land at 21:15. A few minutes later, the aircraft hit the water approximately 5 kilometers from the end of the runway. Only six of the 52 passengers and crew on board survived, after being rescued from the water.
The official investigation into the causes of the accident concluded that the aircraft captain did not keep an accurate check of his altitude and rate of descent during the approach, possibly due to fatigue.
Two days later, on June 14, 1950, another Douglas DC-4 serving the same route crashed while approaching the same airport a few minutes after 21:52, hitting the water within 1 mile of the previous accident. This time, only 13 of 53 people on board survived.
The investigating commission sent to determine the causes of the first crash, which arrived in Bahrain on the 15th, was tasked with determining the causes of the second accident as well. They concluded that: "the pilot-in-command [failed] to adapt the timed approach procedure to the prevailing conditions and... descended to 300 feet (90 m), [then] did not take the necessary steps to maintain this altitude until such time as the runway lights became visible" (**)
...
After a 1994 hydrographic survey discovered part of the wreckage near the site of the accidents, Kevin Patience, who was the treasurer of Bahrain's Christian Community Cemetery Committee (which manages the cemetery where victims of the crashes were initially buried) at the time, took an interest in the matter. He recruited the help of his brother Colin Patience, an airline engine, to investigate the accidents further (***). The Patience brothers concluded that: "the probable cause was a phenomenon associated with certain weather conditions known as 'microbursts' which would have created conditions that were beyond the recovery capabilities of both the pilots and the aircraft involved", exonerating the pilots.
A success of patience, bayesian thinking, or neither?
Regarding microbursts, see: https://en.wikipedia.org/wiki/Downburst
(*) I'm going by the dates reported in Wikipedia. www.planecrashinfo.com reports the date of the first crash as June 13.
(**) https://en.wikipedia.org/wiki/1950_Bahrain_Air_France_Douglas_DC-4_crashes#1950_investigation
(***) https://web.archive.org/web/20110706144200/http://www.stchcathedral.org.bh/history/christiancemeteries/cemeteries.htm
"the pilot-in-command [failed] to adapt the timed approach procedure to the prevailing conditions and... descended to 300 feet (90 m), [then] did not take the necessary steps to maintain this altitude until such time as the runway lights became visible"
From the wikipedia article, which continues "The investigation into both accidents recommended that consideration be given to equipping Bahrain Airport with radio landing aids and with suitable runway approach lights."
I don't think I've ever seen an instrument approach procedure that would have the pilot descend and maintain 300' AGL until the runway was in sight; typically the minimum altitude would be at least twice that for any approach without an ILS or equivalent providing vertical guidance - in which case you're not maintaining any altitude, but following a designated glideslope down to decision height (typically 200' for category I ILS). Irrelevant for Bahrain, because they apparently didn't have anything like that in 1950. And for a visual approach, you don't descend below pattern altitude (typically 1000') without the runway in sight.
So, yeah, I can believe an aircraft flying that approach procedure would be vulnerable to a microburst. Or pilot fatigue, or many other things, because a non-precision approach with a stepdown to 300' AGL while still miles from the runway, seems exceedingly dangerous. Maybe things were different in the 1950s.
The DC-4 was a quad prop, maybe that changes something compared to modern jets? Cruise altitude is much lower too.
The lower landing speed would put a DC-4 into Category B for instrument approaches, as opposed to Cat C or Cat D for most jet airliners. That helps some, though usually more for minimum visibility than minimum altitude. But even for Cat B, skimming my stack of approach procedures for the Southern California doesn't seem to find anything with a stepdown below 500' for a non-precision approach, even when approaching over the water with no terrain and no obstacles taller than a ship.
A quick check of the FAA regulations says that they'd be allowed to set up a non-precision approach with a 300' MDA if they felt it prudent, but that doesn't seem to happen in this century. Apparently "prudent" was defined differently in the 1950s.
I experienced a microburst on a lake in Northern Maine growing up. It flattened a swath of trees about 1000 feet across and a mile long from where it traveled over a forested area. It took out a number of boats and roofs, including dropping a tree on the house I was in! Very scary and not something I saw coming (or would have been able to survive if I was flying a plane).
It is interesting to me to see the balance of "easy to build" versus "easy to understand" when applied to flying as opposed to other transportation methods. We were able to build passenger airliners and make a large industry out of flying before we understood some arguably critical facts about microbursts, wind shear, icing at altitude, or e.g. engineering problems like stress accumulation on square windows such as on the de Havilland Comet. It's hard to build an airplane, but not so hard that you have to completely understand every associated aerodynamic and engineering phenomena before succeeding.
Even spaceships, which are incredibly difficult to build, don't require a full understanding of the environment they will operate in. The Apollo 17 astronauts famously narrowly dodged a solar flare which the Apollo capsule could not have protected them from and which scientists were mostly ignorant about the risks from. We were able to send people to the moon and back without a full understanding of some of the primary threats involved in doing so!
Boats are the central example of a technology which is very easy to build (logs lashed together date back millennia) and where it is extremely difficult to understand all the pertinent details of meteorology, oceanography, hydrodynamics, etc. which we are still learning today. Nuclear fusion by contrast is a technology where the difficulty to build it looks to be about the same as the difficulty of understanding it, where a full understanding is required before you're able to successfully build it.
This is all just a pet theory of mine but it's interesting to read about early aerospace accidents and consider the extent of the things they didn't know despite successfully building planes and spaceships.
>engineering problems like stress accumulation on square windows such as on the de Havilland Comet.
OK, this is a hobbyhorse of mine, but the story there is a little bit more complicated. The tendency of square corners to get stress accumulations was well-known long before the Comet showed up, and if you look at the passenger windows, they're not actually that much more square than the windows on a 737. (The actual window that broke was not one of the passenger windows.) The issue was a mix of sloppiness on the part of De Havilland, bad fatigue design methodology, and the use of a single prototype for both ultimate load and fatigue testing. The overpressure/ultimate load test cold-worked the airframe enough that it was significantly more fatigue-resistant than the production articles, so the cracks didn't show up in the test.
> The Apollo 17 astronauts famously narrowly dodged a solar flare which the Apollo capsule could not have protected them from and which scientists were mostly ignorant about the risks from.
Oh, so we're in the Elseworlds story where the Three are not Thrantastic(?).
> It's hard to build an airplane, but not so hard that you have to completely understand every associated aerodynamic and engineering phenomena before succeeding. Even spaceships, which are incredibly difficult to build, don't require a full understanding of the environment they will operate in.
This is something I expect will keep happening with future technologies, and the obvious example right now should be LLMs. We don't know exactly how they work, and we don't know when or why they fail.
I've been caught in a microburst once, while driving in the Nevada desert. First I was hit by a wall of air and sand coming at me horizontally like a very short, very powerful sandstorm followed by a wall of water, that nearly took me off the road, then I was stuck in the hardest rain I've ever experienced for 30 seconds that quickly flooded the road and caused me to hydroplane. Then it was over and I was back in dry sunny weather like nothing had happened. I can't imagine how terrifying it would be to be flying in a plane through one of those.
Long time reader, first time poster. My brother and I made a new daily word game and would love some feedback on it. I know, I know, there is a glut of daily word games already, but I think ours is better! Wordle and connections have started to feel a bit shallow and I think we bring a bit more depth. Plus, we have quotes (inspired by my love for the civilization franchise)! Anyway, please check it out and let me know about difficulty, UI, or anything else you might think we could do to change the game.
Daily Runes #50
Guesses: 7
Rank: Initiate
⚪⚪🟢🟠⚪⚪🟣
https://dailyrunes.com
I also made a substack to talk about the game if anyone is interested:
https://substack.com/@dailyrunes/note/p-191818235?r=25fmtc
Fun, quite tricky. I had to read the instructions twice but it did make sense after that. One minor UI nitpick: On a keyboard, it would be nice if the Enter key submitted a guess.
Thanks! Can add the enter key for sure!
Fun. Thanks.
The one thing that wasn't clear to me from the instructions was that finding the rune was done by submitting the word.
I also think ring movement might be made easier though I have no specific suggestions and I got the hang of it.
Much appreciated! I do find that it is a bit unclear at first but people usually get it after 1 puzzle and it clicks. Maybe a tutorial on rails would help clear stuff up?
This was really enjoyable even though I don't usually like these kinds of word games (other exception is contexto). Because I don't read instructions, I did miss that the center piece can be used to connect for a while. It's a pity, I can't play earlier days' puzzles.
Thanks that means a lot! Yeah I can't control for people not reading instructions.... maybe I can force a tutorial on new players, haha. As for an archive, we are considering adding it at some point. Not sure how to balance it since we want users to have the feeling that they don't want to miss that day's puzzle. Hopefully users incorporate it into their daily routine instead of getting overloaded and playing the whole archive at once and never coming back.
> Not sure how to balance it since we want users to have the feeling that they don't want to miss that day's puzzle
Why is that? That logic makes sense if you're selling things, or serving ads, or trying to get people to read your newspaper, but you don't seem to be doing any of those things.
I think it's also nice for people who are prone to spend too much time on puzzles to have a 'one and done' element. I have quite a few links now to some of those.
Yeah we don't plan on ever selling anything or running ads. I see how it might be a little manipulative and playing into some weird psychological behaviors. I do want the game to be a success though and have a lot of players. I don't think the archive is inherently at odds with that so we should definitely give implementing it more thought.
I guess wanting to have more long-term players is a reasonable goal in itself! I realize my comment was a little aggressive, which wasn't my intention: I just wanted to understand your goals. Sorry for that.
No worries, I didn't take it as aggressive! It's a reasonable question.
I wouldn't consider myself great at word games but I did find this pretty challenging (sounds like that was the goal). I missed that the center piece could be used. Maybe highlight the neighboring letters available to choose, including the center piece, to make it more discoverable. I also found rotating the circles a bit unintuitive and didn't find the ring selection button for a while, but I'm not sure how to improve that. I like that the words have themes and that letters only get used in one word to give you more of a clue. It ran well on Firefox on Windows. Other than that, I guess I have to wait until tomorrow to give any more feedback ;)
Thanks for the feedback! I like the idea for highlighting click able letters. I do worry that anything flashing will make it a bit busy, so I'll have to think of a way to implement that while keeping the minimalist design.
This was reasonably interesting, though I struggled a bit with the rules and the UI more than I actually got to play the game.
It would be a lot smoother if spinning worked a bit differently:
- select any outer ring letter (already allowed)
- allow any jump to the middle ring (the first time you touch it); have the game automatically rotate the middle ring to the required position and lock it in
- similar for the inner ring
- not sure if there's any reason why inner-to-inner ring jumps require explicitly touching the center, these could probably also just be allowed automatically
This way you should be able to get equivalent functionality with no manual rotation.
Addendum: learning-by-fumbling-around should be much smoother if the UI very gently highlights "legal picks for next letter" at every step.
Thanks for the feedback! The automatic rotating makes a lot of sense. I do want to implement some sort of highlighting valid letters, but don't want to make it too busy. Still trying to figure out a way to do so without sacrificing the game's minimalist aesthetic.
If anyone here is interested in private censorship, there's some recent news from Kiwi Farms, the much maligned lolcow discussion and gossip site that endures more censorship than anyone else. Despite operating fully within the law, they experience grievous private censorship against the spirit of the free Internet. Everything Kiwi Farms has done is 100% US legal. You will not find any court judgement against them(the only negative judgment connected to KF is an Australian default against a downstream IP lessor who didn't appear), and yet they must operate under absurdly adversarial conditions. It's simply unacceptable.
Liz Fong Jones, Google Alum turned Kiwi Farms Deplatformer in Chief has gotten back to it. This time Fong Jones is taking advantage of the DMCA to attempt to dox all forum contributors using a certain copyrighted image. Irrespective of fair use, users would be doxxed and pulled into litigation that would be highly costly for them, but is scalable for Fong Jones.
Fong Jones was instrumental in getting Hurricane Electric, a Tier 1 ISP, to block the Kiwi Farms in an unprecedented move. That was successfully routed around with costly, herculean efforts on the part of Forum Operator Josh Moon. This latest attempt is after a brief reprieve from years of varied attempts to censor the website. This article goes into detail on the history:
https://madattheinternet.substack.com/p/online-censorships-institutional
Josh Moon suspects every deplatforming strategy Fong Jones tries will later be used for Tall Poppy, a reputational management service which they are a key investor in, designed to curate the internet for high profile figures, deleting dissent and bad rumors. It comes from a phrase where criticizing high profile figures is described as "cutting down the tall poppy." Basically, censorship as a service for elites. Fong Jones began with #DropKiwiFarms. That got CloudFlare to drop them, but they persevered and rolled their own solution after months of downtime. Then it was #EndKiwiFarms which saw the Tier 1 ISP blackholing and other varied attempts to take down the site. Now it's #EndNetworkedHarassment which suggests a broadening of scope. Fong Jones seeks to end such speech on the internet, monetizing it through Tall Poppy.
It's a story about a rich tech worker using every connection and trick one can to take down a website that mocks them.
Also, if you're interested in an org that supports the principle of a free Internet and backs it up with practical understanding, USIPS is a new org started by Josh Moon that intends to lobby for a freer internet. The EFF has clearly shown their nature when they failed to offer legal assistance to the Kiwi Farms following the HE blocking, so an org that has no hesitation in supporting its principles is definitely necessary. There needs to be a new org that's willing to go to bat for anyone, belligerent and unlikeable or not. This seems like it'll be the one.
Check it out at usips.org
I'm a bit confused that you defend Kiwifarms with "despite being hated, everything they do is legal" and then complain that companies are causing problems for them by exercising their legal right to not do business with them.
Like, I don't think you can have it both ways. If you run a website that promotes technically-legal harassment campaigns, I don't think you can complain that other people are finding technically-legal ways to stop you. And if you take the stance that some legal actions should not be done because they're immoral, then you have to explain why you think your technically-legal harassment site is not immoral.
Because I think freedom of expression is more important than the freedom to censor discussion, which has been the default opinion on the Internet for many years. It has been taboo for ISPs to block sites for reasons outside of legality. That they have now is the reason we need to codify a law for what was previously an unspoken norm. I like the free Internet and I want it to continue.
Plus, Kiwi Farms is not just technically legal or on the fringes of legality. It is exactly what I'd use as a an example of the central thing free speech is meant to defend. It's the right to talk about people however you'd like. No harassment is coordinated there, nor is it encouraged to be coordinated off-site. The only thing that could perhaps be removed without being total destruction of the right to free expression is the doxxing. I don't think you could change anything else.
If it's not obvious, I believe in free expression, and I'm sharing news to advance that cause. I find it tedious to re-explain why I support free expression every time it comes up.
> It is exactly what I'd use as a an example of the central thing free speech is meant to defend.
The "central thing" free speech is meant to protect isn't gossip, but criticism of one's government and rulers, and it is an essential instrument in the prevention of tyranny and oppression.
I find it rather sad that you take freedom from tyranny and oppression so much for granted that you think defending slander and doxxing is the prime purpose of freedom of speech.
The right to denigrate and attack is a fundamental part of any kind of criticism. If you don't have that, you have nothing. The Kiwi Farms has threads on men like Elon Musk too, if your really need your "speaking truth to power" fix.
And I'd say expressing your expressing your emotions, even the vulgar ones is also an important part of free speech and the one most likely to be attacked by utilitarians.
> The right to denigrate and attack is a fundamental part of any kind of criticism.
No, they aren't, you're confusing mockery and insults with legitimate criticism.
I think being able to mock and insult the powerful is a very important element of freedom of speech.
Mockery and insults often are legitimate criticism.
" lolcow discussion and gossip site" is an extremely generous description of Kiwi Farms, which has been criticized for facilitating sustained, targeted harassment of random people who are cringe on the Internet.
Yeah, and? That isn't illegal. People don't have a legal right to respect.
I never said it was illegal. I'm saying that people may want to keep their distance from Moon and all his works for reasons that have nothing to do with freedom of speech.
People do have a legal right to not be harassed, which is what they said, not that Kiwifarms hosts "disrespect".
As far as I can tell, federal laws only criminalize harassment that's violent or sexual in content, and the same goes for most states as well. Denigrating others is not harassment, legally speaking.
I'm not a lawyer but it looks like 18 U.S.C. § 2261A criminalizes using the Internet to "severely" harass somebody, i.e. to the degree that it "causes, attempts to cause, or would be reasonably expected to cause substantial emotional distress".
Interestingly, according to federal law, it's only a crime if it's anonymous:
> (C) makes a telephone call or utilizes a telecommunications device, whether or not conversation or communication ensues, without disclosing his identity and with intent to abuse, threaten, or harass any specific person;
https://www.law.cornell.edu/uscode/text/47/223
Note the weasel word there: "facilitating."
The better characterization would be "KF made these people famous" -- and then OTHER people decided to harass them. It is very explicitly NOT part of the KF ethos to even Talk to the lolcows, and KF is NOT the forum you want to piss off by disturbing their lolcows.
Oh my. I can't make head nor tail of your post. And the link confuses me even more. I strongly recommend you use the inverted pyramid structure in your lede. Front-load the opening paragraph with the "5 Ws and an H" (Who, What, When, Where, Why, and How) to capture our attention. Sorry if I sound dismissive, but it's difficult for someone like me, who's never heard of Kiwi Farms, Josh Moon, or Fong Jones, to understand what the heck is going on.
Kiwifarms is the forefront of "free speech" on the internet. Not only has it been cited as a reason to invade the Ukraine*, it has also been DDOSed by state-level actors**. Furthermore, credit card companies have refused to carry transactions, so KF has had to sell literal silver in order to keep its website up.
Why all the hate? KF is the one place online where you can say what you want. Nothing is policed there, except no bots. It's "organizing principle" is to laugh at lolcows online, and to "not talk with them, not confront them, not interact with them in any way." But at KF, they'll talk about basically anything.***
*Hohoholocaust, leaving aside it's an American website, and no way did they understand why the meme was angering the Russians.
**remember the feds who spent worktime discussing raping cis-women on their discord? Those fine folks have access to DDOS suites...
***Do not go to KF without using a TOR router.
>Not only has it been cited as a reason to invade the Ukraine*,
It's fascinating how people seem to forget that the Ukraine War didn't start in 2022. It started in 2014, and the only reason why Ukraine was able to withstand the early stages of the invasion of 2022 was that they've been preparing for it since 2014. Or has that been preordained on KF as well?
KF is an e-celeb gossip site. I don't think they influence Zelensky, Nuland or Putin much.
Would you like to talk about the brutal webshow in the Ukraine where animal abusers are forced to kneel on buckwheat kernels for hours on end?
> Why all the hate?
What’s the issue, exactly, with hate for a website intended for people to express their hate? @blisterhill says: “expressing your emotions, even the vulgar ones is also an important part of free speech”
Oh? I'm not upset at the hate. You can hate 4chan, Kiwifarms, Stormfront*** all you like. I'm upset that people are taking actions that make it impossible for others to support said site (by removing Credit Cards as payment processors, by removing Tier 1 internet access).
My friend the award-winning author learned how to write villains from surfing Stormfront, way back in the day. You do not know what someone will get out of a website.
Hate Kiwifarms all you like -- set up a rival site, explaining why kiwifarms is so awful. Sit on kiwifarms and express your hate for everyone and sundry.
***it is of note that Stormfront, a place where literal neo-nazis hang out, is very much less targetted than the e-celeb watch site, kiwifarms. Kiwifarms is targetted by DDOSes, to the point that when Trump Wins in 2024, the next day, the only thing The Powers That Be can agree on is "DDOS kiwifarms."
I am JUST as upset at censorship when the US government refuses to let American Companies support Syrian Government Websites, or what have you. "Sanctions" versus "interesting, valuable data".
> Not only has it been cited as a reason to invade the Ukraine...
So Putin was a Kiwi Farms lurker?
All these seem like grandiose claims indeed.
I think someone Russian noticed the meme (which was definitely a KF invention), and it got to Putin, in a package of "the Ukrainians are Nazi." If I thought the Russians needed to be factual*, I'd probably call it a "rather large intelligence failure." I certainly thought so at the time, given that it's an American website that was only hosted in the Ukraine because Josh Moon had people show up at his door and threaten his mother. (I believe Josh is stateside again).
*or actually cared about being factual, rather than coming up with a "whatever sticks, we don't need to make sense to even our own people, let alone the rest of the world."
Kiwi Farms is a gossip website ran by Josh Moon for talking about e-celebs and people who make fools of themselves online. It receives a lot of heat from activists for facilitating cyber-bullying and networked harassment. It has to deal with censorship across the stack including CloudFlare dropping it, Tier 1 ISPs blocking it, among other attempts to censor. The latest attempt is Liz Fong Jones (the one spearheading the aforementioned censorship efforts) abusing the DMCA.
Thanks for summarizing it all for me! The Internet can be such a silly, vicious place.
>Everything Kiwi Farms has done is 100% US legal.
Just to clarify: are you claiming that everything users of the site have done in connection to discussions and movements hosted on the site is 100% legal, or that the way internet regulations are currently written technically shields the owners of Kiwi Farms from any massive illegality coordinated or posted on its site by its users?
Frankly, there's something to be said for 'this entity is exploiting our generous free speech laws to cause as much harm as possible in a way that's stirring up reaction and demands to do something about it, lets try to harass and destroy them specifically within the bounds of the current legal system, instead of amending the law to allow less free speech for them and also everyone else.'
Which is not to say I support everything their opponents have done in this case - I'm absolutely not going to take the time to find enough independent sources to research this and form a coherent opinion.
But more broadly, there is no such thing as a perfect law, that perfectly allows all good and neutral things and perfectly forbids all bad things.
Within the grey areas and corner cases are a multitude of situations where you could pass a stricter law that would stop a bad thing but also stop other good things, or leave a laxer law in place and suffer the bad things. And every time the bad thing happens, it stirs up public outrage and democratic support for the stricter law.
Given this reality, there is an actual freedom-preserving purpose to people on the bleeding edge of the system having some discretionary power to fight bad legal things without changing the laws.
Like many readers here (whether they know it or not), I am autistic, and would love to live in a black-and-white world where you can write a perfect set of rules, and everything goes well if people follow them. But decades of hard experience and disappointment has shown me why this is a form of disordered thinking, and not actually possible.
Kiwifarms goal is not to harm the lolcows. They're funny, fun people who do stupid shit for attention online. Kiwifarms likes watching them, and laughing at them. Rules of the site include not talking to the lolcows, not interacting with the lolcows. Do the people on kiwifarms sometimes say mean things? Yes? It's not a crime to say "you don't look like a woman," or, more importantly, to publicize predators who say things like "no dragqueen has ever hurt a child" (and then get caught running child sex rings).
You do the world a lot of harm when you refuse to let sleuths dig up truthful information that people would rather keep buried (legally, freely published information, to be clear -- we really aren't talking about doxxing people**). On Kiwifarms, you can absolutely read about teachers buying bathtub estrogen for their students, without parental consent. I've seen people online say "this never happens." On Kiwifarms, they always keep receipts. People have tried to delete their twitter accounts, remove posts that "suddenly look bad" -- but Kiwifarms remembers.
To be clear: we've all said some dumb shit online. And not all teachers do believe that rightwing parents are trans childbeaters.
**doxxing people occupies a grayer space than I want to get into, and I'm not particularly clear about Kiwifarms' position on this.
We actually can win this fight and have free speech on the internet. We don't need to accept private censorship to avoid demagogically-demanded government censorship. We did not have such unwritten rules before LFJ showed up and things worked just fine with net neutrality.
We've managed to keep the free Internet going for a long time. We don't need to sacrifice the unlikeable to the mob. The unlikeable are who free speech is for! If you just get rid of them when they're troublesome and cause problems for free speech, then you're hardly protecting anything at all. How would you even know if you were slowing down the demands for censorship? You might be speeding them up because people get accustomed to censorship.
These kinds of arguments are too tenuous in general. You don't know if fighting every step of the way or giving ground will even end up better, so it's best to go for the ideal outcome. The one that you believe in.
As for Kiwi Farms, nobody coordinates harassment on there. It's collating information on people and being mean. That can often be a starting point for harassment, but it's also definitely well within the bounds of free speech. If you start sacrificing things like that for the greater good, you'll end up with nothing.
usips.org
A noble mission, but I'm pessimistic it'll gain enough traction to become viable. There's the classic problem that Scott wrote about in https://slatestarcodex.com/2017/05/01/neutral-vs-conservative-the-eternal-struggle/:
> if you’re against witch-hunts, and you promise to found your own little utopian community where witch-hunts will never happen, your new society will end up consisting of approximately three principled civil libertarians and seven zillion witches.
Josh Moon's a determined guy, and he's finally making good money now using some mailed in check system since he can't use payment processors. Everybody thought he was screwed when HE blackholed him, or when CloudFlare dropped him, but he got through it, and now the site's more active than ever. I wouldn't be shocked if he pulled off something impressive again.
Okay, I'll bite. What is the "certain copyrighted image?"
Oh, it's just a picture of LFJ. Nothing special really. It's just that some people made edits of it with "Consent accident" as a caption. It's fair use, but because the picture is registered, LFJ is free to use the DMCA to dox anyone who ever used the picture. I think there are 7 affected posters.
But principally, this could work for any picture. Just register it, and you get the right to dox anyone who ever uses it, even if it's fair use.
Sorry, you're saying she's suing them for using a picture of HER? That seems like important context.
I'm not sure how this is a "trick", as you suggested above.
> It's just that some people made edits of it with "Consent accident" as a caption.
I'm sure that has a charming intent. I'm less sure why this is definitely fair use.
It's fair use because it is an edit of an image which adds new meaning or context to an image through criticism. It is fair use. It's a trick because such a thing is obviously fair use, but the DMCA is slanted against defenders, so they will be forced to dox themselves and spend money on litigation that is far easier on Fong Jones than them.
It completely changes the character and purpose of the image, so using the DMCA against it is a perversion of the law.
I'm not a lawyer, so I'm not super confident in this assertion, but I question whether putting ANY negative statement on an image automatically qualifies. If you put "filthy skank" or "baby killer" on an image it would automatically protect you from copyright claims.
If you edit a someone's picture with some controversy they were apart of, inside a thread solely dedicated to insulting that person, that is obviously not for the same use as the original headshot.
That doesn't mean you can just slap low opacity criticism on movie footage to pirate movies because that would still be for the same use case, that is, watching the movie.
This is explicitly for the purpose of criticism and dissemination of hatred. It's obviously fair use. It is a distinct work from the original image evoking a completely different experience.
The DMCA is shit law that is egregiously slanted against people practicing their rights.
4chan and any number of adult sites or boorus host plenty of objectionable content but don't face nearly the same degree of problems, why is that? The centrality of suicide-baiting, doxxing, and swatting on the site with no other redeeming value besides amusement for its users is the problem for Kiwifarms.
Makes lots of sense why none of the free speech orgs would spend their limited resources guaranteeing the "rights to dox and to incite others to suicide" when they could use their valuable time protecting actually core freedoms in our society like those for political speech or free journalism.
KF users have a silly, worthless (if not actively harmful) hobby and should not be shocked that others don't regard it as an important or central political cause. By the same token that "no Kiwifarms user/owner has been found liable in a US court of law", you can also point out that no US court has ever found that an ISP is required to host any specific organization's content or provide service to another business if they don't want to.
Yes, the content of the site is much more attractive to censor because it is not ephemeral and is geared towards talking about e-celebs.
Free speech orgs like the EFF chose not to help the Kiwi Farms because of its content, in spite of the fact that the mechanism used (Tier 1 ISP blackholing) is a grievous threat to the free Internet, and a victory against it would set a precedent. It shows a lack of seriousness about their stated goal (which happens to be my goal).
I think worthless and even somewhat harmful hobbies have their place. Occasionally their hobby of obsessive observation proves a good thing like with the zoo-sadists or with trans lifeline, but that is hardly the lion's share of the site. Freedom of expression outweighing the freedom to avoid gossip is how the Internet has been run up until now, and I hardly think that a dedicated tech worker with connections should have the right to break that norm.
People should be able to have websites where users can express their opinions online within the bounds of the law. Such has been the way of things. It's the incumbent position, and it's under attack. Even if you think the Kiwi Farms is dull or destructive, you have to posit that the free Internet itself isn't worth keeping to justify what LFJ has been doing. Because, evidently, the Kiwi Farms is what the free Internet produced. It finds its audience because there's a latent desire to pseudonymously discuss people online with invective.
> People should be able to have websites where users can express their opinions online within the bounds of the law.
Except if the law was much more strict, you presumably would complain about that too.
> It finds its audience because there's a latent desire to pseudonymously discuss people online with invective.
"With invective" elides the kind of discussion that goes on. It's one thing to talk about how much you dislike certain types of people, it's quite another to have an open culture of using slurs.
KF's culture is the major sticking point for me - if KF wants people to treat it better, it does very little to encourage such behavior. I am not part of any group that they hate, but it is obvious reading their posts that any archival value from their obessive digging is heavily marred, if not outright overshadowed, by how they create and perpetuate hate and contempt.
You say that one *should* be able to speak their opinions freely. Such a view is the decline and death of groups, communities, societies, etc. especially so if such opinions do not foster helpful and good discourse. This is not about holding any particular political view or social belief, it is about the fundamental maintenance of society and community. If you constantly jeer and mock, you will be surprised one day when you don't recognize yourself in the mirror and can't seem to maintain a community which actually acts according to the word.
Yes, I would complain obviously. I like the free Internet and I don't want people to feel so constrained that they can't even speak their mind about figures who annoy them on their own websites. Abrasive and non-abrasive don't really make a difference to me. Kiwi Farms isn't destroying society. People like to say it is because they don't like what people say about them anonymously on the Internet.
If you believe in might is right and that the biggest techhead with the most pull at Tier 1 ISPs should have control of what people can say, and that whoever can kick up the biggest media circus should decide what expression is online, I'm afraid we disagree. I think expression is important in itself rather than being useful for some utilitarian end.
I believe hate speech is free speech, and even if I didn't, I would want a law instead of LFJ.
With free speech, you take the good with the bad.
Plus, I think you're catastrophizing about free speech destroying the world is without basis. We've had decades of free speech online, a centuries long tradition of free speech in the US, and nothing too bad has happened as a result of it. It's all more than worth it. That's why I can rest easy when I say I'd take the good with the bad. It's because I know that my deeply held value, free expression will probably not lead to some disastrous outcome.
> a lack of seriousness about their stated goal
Well, now you know they're like the AC-"unfair process, inappropriately favoring the accused"-LU.
>Everything Kiwi Farms has done is 100% US legal.
So is everything done by their high-profile detractors, I'd imagine. I do appreciate that defending free speech mainly means defending scoundrels that are the best at skirting the line, but still, the rhetoric painting them as innocent victims of "grievous injustice" is a bit rich.
Threatening Josh Moon's mother (live and in person) is definitely not 100% US legal.
And what rights have the Kiwi Farms violated of the people they talk about in their threads? Josh Moon's freedom of speech has been curtailed and the free internet with it in a real and unprecedented way. The people you're referring to are fighting for their right to not be talked about by destroying the far more important right to talk about things.
I don't know why there's this constant tone policing whenever free speech comes up. The EFF did it too in their article on the Farms. A grievous injustice is a grievous injustice regardless of who it is done to. You don't get to take away grievous injustice points because you think the person is bad. No hedging is necessary. Kiwi Farms is a victim of grievous injustice.
I'm not interested in making a tone argument. Rather, my point is that such rhetoric conflates several issues, e.g: "are actual laws being violated by anyone in this conflict?" "what laws would we prefer instead?", "how bad are the participants in this conflict according to current norms?", and "what norms would we prefer instead?".
Some of these questions have easy and straightforward answers here, and some do not.
The norm I'm advocating is a free Internet that is only inhibited by the law. I think this is the prior of many readers, and it's why I addressed my post to those who care about private censorship. If you believe in a free-for-all of online censorship, that's your right, and you'd be in the company of people like LFJ.
That's what I'm getting at with the tone policing. I think I've made my views clear in that whole post, and you're just saying "It's a complicated issue, and there are lots of opinions." Yeah, but I'm sharing my opinion to likeminded people, and making suggestions on how to advance it. I'm not trying to convince someone opposed to support free speech as that would take a much longer post.
Moreover, the free Internet view was the prevailing one among tech people for decades. We can afford to go on the offensive, can we not? What LFJ is doing is a violation of norms we've had in place for decades. If you constantly frame the people the norm is meant to protect as only arguably worth protection, you erode the norm.
> only inhibited by the law.
If the law brought you to this, where your enemies can censor you while using it as a shield against retaliation, perhaps the law is not worth preserving.
I'd be down with getting rid of the DMCA. Really, all the copyright could stand an overhaul...
> The norm I'm advocating is a free Internet that is only inhibited by the law.
"These people shouldn't be allowed to suicide-bait and this website shouldn't exist. We need to pass laws to make Kiwifarms and sites like it illegal" is legal free speech. It is just as much speech as the stuff that happens on Kiwifarms.
I really hate this kind of comment, because it's perfectly calibrated to the Refuting Bullshit principle. Your three sentence comment that you probably typed in 30 seconds sounds profound and persuasive on the surface, and requires a long, carefully-thought precisely-worded many-paragraph response to explain why your point is either vacuous and irrelevant or false, depending on how "free speech" is defined.
Maybe I'll make that response if I get around to it, maybe someone else will. But I really hate this kind of comment.
Yes, both of those things are free speech.
I haven't talked about my opinions on the law, only stated that the law should make the determination, not LFJ and everyone agentic enough to bother the CEO's wife.
Though I don't actually think Kiwi Farms should be banned. It's anthropologically important. Gossip and bitter hatred have their place in the noosphere and cutting them off from outside expression seems authoritarian to me. It's a bit like how adultery isn't a criminal offense. Freedom of expression is important, even if some people get hurt.
I consider myself semi-likeminded, so let's say I'm trying to offer constructive criticism. Free speech isn't particularly popular these days (as I'm sure you've noticed), so in order for it to have a fair chance in the court of public opinion, its staunchest advocates would do well to proceed with more finesse.
Kiwi Farms is an obvious wretched hive of scum and villainy. There's a norm that such spaces deserve to be kept down, which obviously trades off against increases in free speech. Unless you deign to at least acknowledge such tradeoffs, less sympathetic audiences will just dismiss you outright.
> Free speech isn't particularly popular these days (as I'm sure you've noticed), so in order for it to have a fair chance in the court of public opinion, its staunchest advocates would do well to proceed with more finesse
I think this is utterly, absolutely backwards. The strongest, most robust principle for preserving free speech is the one where something being protected speech has nothing whatsoever to do with how bad, unpopular, or immoral it is.
The moment you weaken that absolute principle, you've lost 90% of the fight. This is the entirety of how cancel culture and totalitarian EU speech laws happened: when your principle is "I defend all speech no matter how bad it is" then you can defend Nazism, criticism of Islam, misgendering, and everything else without implying any belief other than free speech being good. As soon as you switch to "I defend speech unless it's *really* bad like Nazism" then suddenly, if you want to defend misgendering you are directly implying by doing so that misgendering is not really bad. And if you happen to be in a community where misgendering is regarded as really bad, you have to choose between refusing to defend it as free speech (like Nazism) or taking a moral position contrary to your community. While under the old principle, no such choice was necessary, because defending misgendering was compatible with believing it as bad as Nazism. Similarly, under the old principle a devout Muslim could defend the right to condemn Islam without compromising their religious identity at all; under the new principle, they can't.
The defence of the worst scoundrels isn't just a nice extra strong principle of free spech; it's the entirety of what it means to have meaningful free speech at all.
Sigh. Kiwi Farms is a hive of people that talk to each other, and collect "publically available" data about lolcows. The only way this harms someone is if they go looking for what people think about themselves (and if they're narcissistic enough to care).
We're not talking about liars, we're talking about a group of people munching popcorn and seeing what Chris-chan has done today that's silly and funny.
Interesting stuff, thanks for sharing. I've ran afoul of a "brand protection" company while trying to sell a pair of Air Jordans on ebay. Nike deems anyone selling their shoes who is not a Nike-authorized dealer to be violating their IP (because it must be a knock-off, surely there's no way to get a legitimate pair of Nikes without going to Foot Locker or something). The brand protection company tells ebay to take the listing down, which happens without any due process or investigation.
It seems like there's a bumper crop of companies trying to abuse DMCA and other IP laws to bully people who would probably prevail in court, but lack the financial means to sustain a case.
In the rest is politics podcast recently, a guest suggested that the release of AI models should be as regulated as drugs. Lots of testing and then licensing by a federal agency. Possible?
It's not exactly hard to buy illegal drugs in this country, though, and you actually need to smuggle them in, not just connect to them using a VPN and an AI-friendly third country.
What would be the difference to the general alignment problem?
TLDR: I want to hear from creatives.
Last week I made http://www.halmarks.org/, a set of hallmarks/badges allowing creatives to self-declare the amount of AI they used in a work.
I got some good feedback and discussion, but haven't had time really do anything with it.
My current instinct is I'd like to do a lot more thinking and tweak the concept a bit more rather than pushing ahead with it in its current form. It won't be adopted if it's not the right shape for the job, and beyond that, if it's not actively helpful at achieving its aims, I don't even care if it gets adopted.
I'm considering repurposing my (pretty much unused) Substack page to collate and write up some of the points that were thrown up in discussion. I'm not a natural blogger and I hate working "in public", but I can see the value of getting more voices involved here.
Before I do that I'm posting it here again for more feedback and discussion. I'm especially interested in hearing from creatives who make things - I'm interested in:
- what's your opinion of AI
- what's your workflow with or without AI
- whether you'd use something like this (including recommending changes)
- whether you wouldn't use this for X reason but feel the same pressure for "something to be done"
- or whether, as a creative person, you're actually fine with where things are heading and feel no need to course correct at all.
I took a number of shitty selfies with my phone, and uploaded it to Gemini, asking it to make dating site pics, realistic, but better angle, lighting etc. The results are astonishingly good, they don't look like "AI slop", they look something like a pro photographer would do.
I would be worried if I was trying to make a living as a photographer.
Not a very positive comment: I wouldn't use these, because 1) it's not that hard for me to just type "made with Claude" without a hassle of going looking for the link and wondering how to embed the whatever 2) most usefulness of these things would come from being recognizable to a lot of people. Or a type of thing that people have seen before and have learned to ignore unless they have a reason to care, same way that your eye habitually slide over the CC-BBQ-0123 licenses. Right now everyone is going to see these and go: "what? did he make this himself or something?" Uncle Moloch says "hi". 3) I'm sorry, but the icons also look a bit on the derpy side. I would be more willing to use them if they looked super-slick and professional, or at least drab an minimalist enough they don't draw too much attention to themselves.
I love AI, but I do want AI output to be clearly labeled. So, in principle I'm in favour of something kind of like this being used, but it's going to be hard for you to get people to adopt it.
Do you mind telling me what kind of stuff you work on?
I do digital art https://www.artstation.com/tasty_y. Oh, and also I'm going to launch a super-awesome cool vibe-coded website that has been a week away from completion for 2 months now, where I could plausibly display a badge like that somewhere, if they looked alright.
You mean AI-generated stuff stays a week away from completion for months on end as well? I thought we humans at least had that phenomenon to ourselves.
Redrawing the icons will probably be the very last thing I do, on the grounds that that way I hope to only do it once. But yes, can certainly make them more minimalist.
I don't know who you count as a creative. As someone who's written a little bit, plans to hopefully write a lot more (in the sense of having very clear ideas of *what*, but not quite sure *how* yet), but hasn't shared anything publicly in any way...I'll answer on the off chance I count.
"what's your opinion of AI"
Highly skeptical it's going to end up much more than a glorified search function, glorified encyclopedia, and glorified writing/coding assistant. I find LLM writing pretty nauseating. However, I really like image models, and I'm 95% sure this is entirely because I'm a terrible drawer and have a woeful visual memory/observation ability.
"what's your workflow with or without AI"
I use it only very occassionally for factual information searching and for analysis of text I've already written. The thought of having AI do actual writing for me is horrifying. I honestly can't understand the psychology of the kind of person who'd present obviously AI writing as their own, without dying of shame.
"whether you'd use something like this (including recommending changes)"
I would definitely use it if it meant something to many people, and maybe even if it didn't. I would greatly value the ability to promise in a somewhat believable way that something involved zero AI in its creation.
"or whether, as a creative person, you're actually fine with where things are heading and feel no need to course correct at all."
Definitely not fine with it, but I don't think it's existential. These sorts of fears have happened many times through history, and the more the AI hypsters (whether utopians or doomers) fail to engage with that fact, the less plausible I find their claims.
My main anger is at the kind of broken person who thinks prompting AI is a substitute for actual thinking/writing/work, and the tolerance for these kinds of broken people. I'm not angry at AI itself; I see it as just another tool.
Thanks for your response, you definitely the kind of person I'm looking to hear from.
I think everyone right now says the same thing about AI - it's not there yet in my own field, but look at how good it is in the fields next door. And it's visibly getting better all the time.
I'd be quietly relieved if it peaked where it is now, but am unwilling to assume so.
> These sorts of fears have happened many times through history,
I agree, and that's what scares me. Job roles were destroyed, societies were reshaped, entire pillars of culture disappeared. A strong back doesn't impress the girls anymore because tractors do the job better - and within a generation we were all more effete. Now automation can (or might soon) do creativity. I don't like this.
> My main anger is at the kind of broken person who thinks prompting AI is a substitute for actual thinking/writing/work, and the tolerance for these kinds of broken people.
Last time I decided we should christen them "Vogons" - Douglas Adams' race of people who emphatically do not understand beauty, but think they do. It's not a direct expression of the concept I'm trying to get at, but it's close.
> people who emphatically do not understand beauty, but think they do.
I've seen r/aiArt and what the Experts and "creatives" praise. I know which I'd consider Vogon.
That's a discussion I'm happy to leave for another time. I actually wanted (and I hope I was clearer about it last time) to use "Vogon" not for the understanding of beauty, but for the understanding of the creative process that goes into making things.
Wait, we have a conspiracy?
That's what they _want_ you to think.
Shhhh! First rule of the conspiracy is, we don't talk about the conspiracy 😉
A cabal of highly educated people, spread across all kinds of industries at the levers of power, believing in the power of the rational mind, with a vague goal of rational enlightenment and maybe even abolishing the rule of human over humans? Sounds like *the original* conspiracy..
Take this out of context and it sounds like the plot of Atlas Shrugged lol.
I don't think the Rationalists have much influence over any "levers of power", beyound some internet communities and advisory positions in tech companies. Which isn't nothing, but it's not Illuminati either (and not even Skull and Bones).
Good, good, keep up the masquerade :)
Dario Amodei is at the very least Rationalist-adjacent (he took the GWWC pledge in 2010), and he and other such types broke off OpenAI to found Anthropic primarily due to differences of opinion on AI safety.
at the levers of power?
"industries at the levers of power", but even without that qualifier please grant me some artistic license.
I’ve started a new blog, Structure and Guarantees, about how to design intelligent systems that we can actually reason about — not just empirically tune.
A recent post argues that compilers are a concrete, already-deployed example of recursive self-improvement: they optimize and even compile themselves, but in a setting where we can prove correctness properties across iterations. That makes them a useful “toy model” for AI safety questions like alignment and containment—but in a domain where those questions are actually tractable.
My guess is that we should be grounding more of our thinking about advanced AI in domains like this, where the guarantees are real, rather than starting from unconstrained hypotheticals.
The blog mixes formal methods / programming languages with some less-usual angles (e.g., evolutionary signaling and distributed systems as lenses on intelligence).
https://stng.substack.com/
I've just started looking at your blog. First impression: it is good stuff that will be worthwhile to engage with carefully. I think that I agree with your overall theme that we can build useful and trustworthy AI ecosystems by constructing them from two ingredients. (1). A symbolic, GOFAI engine that we can reason about, and (2) a more magical and mysterious heuristic generator, presumably constructed using big data and deep learning. Over the long run, such an approach might work. But ...
This posting on my own blog
https://rationalreciprocity.substack.com/p/provably-safe-systems
points out some difficulties with this approach that may turn out to be show-stoppers.
Thanks for the words of encouragement. From your post, I can see that we agree on the important points. Let me try to address your two primary objections.
First, you worry about the complexity of predicting consequences of AI decisions in the real world. I address this point partly in early posts of my blog, and I'll be returning to it in later posts, too. In short, we should redesign the world for improved legibility to AI. Much of the complexity we see results from somewhat arbitrary outcomes of natural evolution. Much of *that* complexity comes from the centrality of signaling to humans. The future economy will not have humans at its center, and its day-to-day problems can be correspondingly simpler, if we (or our AI successors) do a good job codesigning the world and the code that needs to understand it.
Second, you worry that it's very hard to formalize objectives around human flourishing and so on, which are most commonly described in AI-safety literature. I agree! I will get into approaches that avoid those challenges, though I think it will take enough blog-post-length essays to explain that I don't want to start into it here. Stay tuned?
Compilers aren't recursively self-improving. To be more precise, they're limited to a recursion depth of 1, and this self-improvement is limited to pure execution speed and memory consumption (and the effect on the latter is very low for practically all mainstream compilers). This means that when you apply a compiler to its own source code, the output will – maybe – run faster, but you can only do this once. A compiler cannot improve or even change its own algorithms.
No idea what you hope to learn and generalize from this, given these limitations.
Oh, but think of a compiler as running an expensive search process for faster ways to implement programs. As the compiler's performance improves, you can find even-better versions within your budget of time and memory. (Most compilers today don't work that way, but they should!)
> Oh, but think of a compiler as running an expensive search process for faster ways to implement programs.
Those "faster ways" are all thought up and implemented by humans. A compiler doesn't come up with new optimizations. A program that would come up with new optimizations isn't called "compiler" anymore, and you would also lose those properties that guarantee correctness (or "alignment" in AI-speak).
So maybe we are just going back and forth about the right terminology for the sort of thing I propose formally verifying. Since I called it a "compiler," it didn't seem to include key capabilities and therefore avoided related verification challenges, but actually I did mean to include those.
> we should be grounding more of our thinking about advanced AI in domains like this
How do you avoid the streetlight effect where you investigate things that you're able to ground, but miss the important parts? You don't want to fall into the trap of “lets study something that's vaguely similar to what we're interested in, because this is easier to study than the real thing.”
I agree that's an important tension, and it's just my intuition that there are enough similarities here to the broader problem to start learning nonobvious lessons, some of which will turn out to be more broadly relevant. (It's hard to predict in advance which they are.)
I'm also used to thinking of so many problems as compiler problems. At some level, any problem of making a plan and then reducing it to practice can be cast in this light, with the right clever source language.
What, a return to symbolic AI? Perish the thought!