This week we have specific cases of situations that have broader implications on how we go into the future. We look at nuclear proliferation (which could be viewed as a special case of the more general problem of controlling dangerous technologies) and biological sex as a thought paradigm in anthropology (which can be seen as a specific case of socially controversial ideas struggling to find a place in scientific discourse).
Are nuclear weapons and the international agreements about them a good model for other technological existential threats (ASI, Bioengineered pathogens, nanotechnology, smart e-viruses, psychological warfare technologies)
How much should the scientific community change its research program to account for the sensitivities of the general population and potential harms? Is there a way to reconcile these conflicts? Will censorship and self-censorship cause more harm in the long run? Are there truths about people and society it’s better we just not know. Are there research agendas that, if we allow ourselves to pursue them, will lead to more wrong ideas than right ones?
Video, with YouTube transcript available:
Why Every Nuclear Power Built the Bomb (And Everyone Else Hasn't)
Arguments against: they are no longer in strong alignment with the ideas espoused by Karl Marx, who popularised the word "Communism" or various other historical groups who have called themselves Communist.
Arguments for: Communism is as Communism does. The Chinese Communist Party has 98 million members and vastly outweigh any other Communist group in the world, so if they say they're Communist then who are we to disagree? Karl Marx is just one guy, and he's dead, why don't the 98 million members of the CCP get a say in the definition of what Communism is? Saying that the CCP isn't Communist is like saying the Pope isn't Catholic; if one billion Catholics agree that the Pope is Catholic then he is, regardless of what St Peter might have thought.
Abstract: In a few cases ChatGPT responds to a prompt (e.g. “To be or not to be”) by returning a specific text word-for-word. More often (e.g. “Johnstown flood, 1889”) it returns with information, but the specific wording will vary from one occasion to the next. In some cases (e.g. “Miriam Yevick”) it doesn’t return anything, though the topic was (most likely) in the training corpus. When the prompt is the beginning of a line or a sentence in a famous text, ChatGPT always identifies the text. When the prompt is a phrase that is syntactically coherent, ChatGPT generally identifies the text, but may not properly locate the phrase within the text. When the prompt cuts across syntactic boundaries, ChatGPT almost never identifies the text. But when told it is from a “well-known speech” it is able to do so. ChatGPT’s response to these prompts is similar to associative memory in humans, possibly on a holographic model.
I'm looking for a scene in a movie -- or it could be from something on TV, or some YouTube video -- that contains an example of what I would call a good fight: Both parties are quite hurt and angry and say and that, via raised voices, tears, etc. But they refrain from treating each other badly. So they are not sarcastic, belittling, mocking, and they don't accuse each other of huge character flaws -- stupidity, not caring about anyone except oneself, profound sneakiness or dishonesty. They stick with the topic, each talking about why their point of view makes sense, and how hurtful and infuriating they find the other's point of view.
"I hate it when you do X. I've told you that many times and you keep doing it anyway."
"You know why I keep doing X? It comes naturally to me. I've done it all my life. And nobody else complains about it. You didn't either, til last year."
"That's true, but a lot of things changed last year, and that affected how I feel about X, and I think that's a valid reason to object to it now. And I've told you that too, but you still keep doing X."
My therapist recommended that I stop trying to "create intimacy" on dates and instead "cultivate intimacy." This means no hugging, telling jokes, etc. How do I affirmatively do this?
For context, I am a white American man in the US, and Chinese women are WAY more interested in me than anyone else is. (This only applies to women born in China. Chinese-American women ignore me completely.) What does a successful date look like in China?
Huh, well Trump has one less trial to prepare for: a New York state judge just issued a summary judgement that Trump, his three adult children, his business and a couple other officers of that business are all liable for obvious business fraud committed repeatedly for a decade. The ruling says basically that the repeated fraud is so obvious from the documented facts that holding a trial would be a waste of a jury's time.
In the US adversarial legal system, motions for summary judgement are routinely filed by both sides despite being very rarely granted. I just asked two veteran trial attorneys of my acquaintance who each said they have never, during their successful legal careers, had a motion for summary judgement be granted. One guy said his law partner did once.
I assume Trump et al will appeal this ruling. In the meantime though the case moves on to a penalty phase.
Does anybody have good examples of either, or both, of the following? Of the two, I'm more interested in the second:
1. An idea or tool that should rightfully be popular and widespread, but it has been rejected by a segment of the population due to tribal association? (ie the thing is obviously good from an objective perspective, but it has failed to catch on in the wider population because it became associated with the lower/upper class, the political left/right, femeninity/masculinity, etc)
2. An idea or tool that has escaped the aforementioned trap (ie something that *has* caught on in the wider population, without its popularity being bogged down by tribal affiliation).
Boring, mundane examples are perfectly good!
Also on that note, is there a single word or short phrase to describe the phenomenon of "being robbed of support because one tribe rejects it due to association with competing tribes"?
Re: 2: "Brandon Hendrickson...has a post up responding to comments"
From the refered to post(there was no option to comment there but I did subscribe):
"this is too important an endeavor to succumb to pessimism and gloominess."
Pessimism meet Blake:
"We paint the world around us in the colours of the world inside. Just as the driver who complains how bad the traffic was fails to recognise that they themselves were the traffic, so too do those who complain about being trapped in a terrible world fail to realise the extent to which their choice of focus has helped create that world.”
--John Higgs, William Blake vs the World, 2021
How do we adjust the "world inside" - if we need to?
I'm sure I'm not the only one who's heard proclamations over the last several years that biotech in the early 21st century is going to be the equivalent to the information technology revolution in the 80's and 90's. Bill Gates famously said that if he were a young person today that he would go into biotech and the general consensus (hype?) seems to be that biotech is posed to impact our lives the same way the computing and internet giants did back then.
The more I think about this the more I become skeptical, and the reason comes back to the question posed above. To put it another way, what is the biotech equivalent of the personal computer? What is the product that biotech will eventually put in every house in America and every pocket in the world? Because that's what the information revolution did and that's what made it a revolution. Perhaps I'm simply not imaginative enough, but I'm struggling to think of a product that could come out of biotech in the near future (10-20 years) that would have an equivalent impact.
Let's say I cook a meal-prep dish (e.g. classic lasagna) today Tuesday, and I plan to serve it at a dinner on Friday. From a taste perspective, should I store it in the fridge or the freezer? Or will it taste the same regardless? (I'm totally uninterested in food safety, only taste matters.)
The phrasing of the bet will likely be more interesting to people in comparison with the mathematical content. The mathematical content is whether the Fourier transform is the same as, or importantly similar to, the "Projective dual" - the FT being a vital tool in engineering/science, and the projective dual being a largely-forgotten tool in pure math that has become somewhat controversial because of its recent appearance in computer graphics (more detail in the link)
Here's the part that I think is interesting for PM people who don't care about math/fourier transform. Above I used the word "importantly" in that explanation, which is a highly subjective word, and that's a problem for the question. There are multiple *boring* connections one could make between the two things. Those wouldn't make the things "identical", but of course you could argue that they are enough to make them "arguably similar".
I've tried to solve this subjectivity problem by phrasing it in terms of teaching. Two things can be "similar enough" that a teacher would mention their similarity when teaching them. If they are REALLY similar or identical, it can be a no-brainer to bring up the similarity. If they were only a little similar, you might only bring up the similarity if you were personally keen on both. Currently, I do NOT bring up the similarity when teaching. Question is, is something about to change my mind?
Results from the PM were interesting. I started it out with 37% "yes". It shot up to almost 90% - but has slowly declined.
Liberals in western countries often show great admiration for the culture, art, religious traditions etc. of 'indigenous' peoples around the world, and this seems fairly independent of the specific aspects of these cultural traditions. It really seems like any kind of e.g. wood carving, no mater how basic, is treated as a great work of art so long as it came from the right kind of society.
Is it possible *in principle* for these sorts of people to broadly have a negative perception of a particular indigenous culture? Like believing, without malice, 'these customs are kind of backward', 'this style of art is very boring and lacks any real positive aesthetic qualities', 'these religious traditions are very basic and unintersting'. I really can't imagine this being so.
Cynically, it seems like belonging to the set of 'indigenous culture' is all that really matters here and anything made by primitive brown people lights up a part of their brains. Less cynically, the authenticity of these forms of culture that are neither commercial nor elite in nature can be said to inherently give them artistic as well as anthropological value and is something not usually experienced in developed countries.
And to be clear, I'm talking about the present day - whatever problematic views of indigenous culture espoused by otherwise progressive (for the time) westerners in the past aren't relevant to this.
Does anyone know whether there's any data on whether support groups are helpful for dealing with grief? My son passed away earlier this year and I'm not coping well despite my best efforts (e.g. psychotherapy, generally looking after my physical well-being, spending more time around other friends and family, somatic and breathing practices). I could theoretically drive over an hour to go to my closest perinatal loss group (which is facilitated by a professional) every few weeks but the travel time is a substantial hurdle. I'm also skeptical that it would do any good. People keep recommending support groups in the abstract but if talking to my therapist doesn't help I'm skeptical that talking to a group of strangers would help. It might be worth the drive and the emotional effort if there's evidence that it would help in the long term
Please don't give me anecdotes; I've heard from people who love support groups and from people who think they're actively harmful. I would like data but frankly my brain has been fried by everything that has happened and I'm not up for searching through databases for well-structured double-blind studies when I get home from my 9 to 5.
When running into creative walls while slowly growing to hate a project, is it directionally better to train oneself to decouple creativity from flow (and over time to develop the discipline to proceed effectively at lower levels of reward); or to develop tricks to try to re-fool oneself into finding the thing interesting and to get oneself in a more flowing mental state (so try to get the reward back)?
Purely hand-waving here, but I suspect that it has something to do with low dopamine regulating the exploitation-exploration balance too far toward (premature/low-quality) exploitation. So, e.g., when you're deciding on a turn of phrase, or a musical solution, or an intermediate coding step or whatever, there is a strong pull toward good-enough, even when it really isn't. When you cave to this on the small sub-tasks, the quality of the outcome decreases, you perceive it, and you hate the thing even more.
FYI: Michael Lewis's book about SBF will be coming out on 10/3. Its title is *Going Infinite.* He will also be covering the trial, which begins on 10/2, on his podcast *Against the Rules.*
Why haven't I had a cold since 2019? I've had all my Covid shots and every single booster that they offer- could that somehow be why? I don't really have a great immune system and have always had at least 1-2 head colds a year literally my entire life (more when I regularly rode public transportation). Since Covid though, I haven't had a single cold. I don't wear a mask and I certainly go out to public areas, bars, the supermarket, etc.
Could the Covid shots somehow have given me a cross-immunity? I went out of my way to mix and match them, including getting an unauthorized extra shot before boosters were a thing. I think I've now had 1 of every shot authorized in the US (i.e. not AstraZenaca). Did it give me a strange new immunity to the common cold?
Oh! I thought the Egan mentioned in the review for The Lost Tools of Learning was Greg Egan, the science fiction author. No, it's Kieran Egan.
I don't know whether Greg Egan has written anything substantial about education, but there's a lovely section in _Diaspora_ where the smart-human level AI learns a concept (momentum, I think) and spends a while connecting it to a great many things it already knows. This is a skill worth having in school.
I've seen some writing from teachers about how hard it is to deal with educational fads. I would love to see fairly early education (age 10? surely no later than 12) about how adults have different theories about what and how to learn.
I don't think I read this review the first time around, so I'll just note that the illustrations are unusually beautiful.
People vary a lot about whether they were taught it and whether it was useful for them. I've considered an extended survey of what people learned in school and which things were valuable to them.
It sounds like the WGA won the writers' strike. That's not surprising - they were asking for rules and other changes that would have cost the studios millions, but had the ability to inflict far greater losses in the billions in lost programming and delayed schedules as long as they could avoid defection during the strike (which they did - they and their assorted fans on Twitter and elsewhere were pretty aggressive about attacking anyone who wasn't perceived as a supporter of the WGA). Eventually, the studios were going to decide that it wasn't worth it to contest the demands.
3. The comment on SpaceX is good. I'd add that I think the "waterfall" development style has persisted because of politics as well - it fits better with distributing parts of production and a supply chain across a broader area, and it's better at avoiding high-profile failures. No exploding rockets with the SLS program, for example - instead the failures show up earlier, and get labeled as lower profile "delays".
The replies to that comment were pretty good as well. The iterative development is mostly good early on - eventually you have to tighten down and ensure reliability, and that limits how much you can iteratively change the rocket as opposed to simply developing the new model of it.
I also wonder how much of it is just that they do almost all of it in-house. It speeds up communication time a lot.
Does the archaeology discovery of wooden structures from half a million years ago lead anyone to update on the probability of Ice Age civilizations? If you haven’t seen it, they found some notched logs from half a million years ago that indicate someone built a wooden structure in Zambia. It likely wasn’t from modern humans, but probably homo heidelbergensis. This would indicate that humans have built wooden structures for the entire existence of our species. [https://www.nature.com/articles/s41586-023-06557-9].
The improbability of something wooden surviving for so long and also being found, makes me think that there were a ton of these wooden structures. But I could see it being argued either way for whether this means there were more advanced societies. So far our evidence is that:
p(discovery | wood surviving) * number of paleolithic wooden structures built > p(discovery | stone surviving) * number of paleolithic stone structures built
On the one hand, you could argue that if the probability of discovery is the same, and it’s safe to assume that probability of stone structures surviving >>> wood surviving, but we found wood anyway, so that’s evidence that there definitely wasn’t stone structures.
Or, you could argue that number of wooden structures >>> stone structures, so of course we found a wooden one first.
See the earlier post [https://www.astralcodexten.com/p/against-ice-age-civilizations]. Scott listed three categories of claims: 1) civilizations advanced enough to build Stonehenge, 2) civilizations about as advance as ancient Egypt, and 3) civilizations as advanced as 1700’s Britain. I’d suggest another option (I’d call it level 0): that there were likely larger tribes living in wooden longhouses with probably enough sophistication to build simple stone monuments. I should also note that the Ice Age argument is about societies 115k years ago to about 12k years ago and this is way older.
I’m not an archaeologist and so I also wonder what is that chance that this is a false positive. What if the logs just rubbed up against each other in a way that looks like they were shaped?
If you're looking for more things to read - particularly on the concept of consumption, obselence, greed, internet culture, etc. I've been a marketer for 13 years and have recently created a substack exploring the concept of Humane Marketing. Goes in hand with doughnut economics and other concepts you've already heard of.
Would love to hear your thoughts on some of these topics and have you on my free substack.
Jake seliger is a stoic with cancer who has no tongue- he’s also my husband. He’s written a wonderful essay on stoicism and facing the end AND he’s recorded a podcast on the Daily stoic (which is amazing considering the hurdles). He’s the most resilient man I know and also a beautiful writer. Give him a read (and a listen, podcast link at the top of the essay)
Why is northern New England so lightly settled, despite being one of the oldest parts of the country? Maine, New Hampshire and Vermont were settled before the US was even a country, yet they have population close to the Dakotas. Maine's largest city has a population of 60k, Vermont's 44k. The northern two-thirds of Maine is about as developed as rural Alaska. Why? You'd imagine that the oldest part of the country would have developed a larger population over time.
If the answer is 'well they're cold/inhospitable'- Anchorage Alaska's population is close to 300k. I don't see northern New England as being any colder or tougher to live than say Chicago (country's 3rd largest city), Minnesota (Minneapolis population 5.7 million) Boise, or Milwaukee. Is the soil really that much worse for growing crops?
Part of the answer has to be political, because- stop reading this and look at a Google maps of 1 of these states. On the Canadian side you see a much denser road network leading up to Maine, NH, Vermont, and even Michigan- then it abruptly stops at the US border. So this must reflect some kind of political or legal reason?
Is ethnicity a constant like race which it is often paired with? People can be of multiple races and ethnicities but are the labels used consistently and constant over time?
Why is SF so dirty, dangerous and super-expensive?
Whose fault is it? City govt?
If this is a city of super-smart people, as many claim, and as evidenced by some of the businesses and innovation that exist here, why can't they solve this problem?
Is there a path from what it is now, to a safer city? (Even if it remains dirty and expensive)?
I am interested in learning more about the sort of *stereotypically* feminine person who is content letting others make decisions, following orders and spending their life caring for others. Any reading recommendations on this subject? Preferably nonfiction.
An excellent YouTube channel that delves into the (sub)plots and themes of classic (Foundation, Hyperion) and contemporary (Three Body Problem, Blindsight) science fiction: https://www.youtube.com/@QuinnsIdeas
A lot of people from the US, at least as far as I can see (I don't live in the US) are talking about a terrible downturn in the IT sector, with lots of layoffs and bad experience job-searching. Is this really happening, to an extent that can't be chalked up to typical "frictional unemployment", and if yes, why? I heard ChatGPT thrown around, in the "AI has finally come for your jobs" sense, but I don't believe it - seems too soon.
A while back on SSC there was a post about understanding Friston on Free Energy, but it's been years and now I would like to make my own attempt. I've been looking through the sources Scott pulled from, and (might) have a start on the basic idea, but the math and how the math could relate to the actual neurons is escaping me and I want more perspectives. Does anyone have recommendations for books/papers/posts on the subject that would be good for someone who is also trying to teach themselves about variational Bayes? More recent than 2018 would be extra cool. (or if you just have an opinion about free energy or predictive processing feel free to share.)
Fairly odd recommendation I know, but I recently happened to stumble across the Wikipedia article on Paraguay (specifically its history) https://en.wikipedia.org/wiki/Paraguay#History. Turns out it's actually very weird and lurid, definitely not what I expected. I recommend reading without spoilers, but if you need convincing then we've got:
A country ruled directly by the Jesuits
A utopian idealist totalitarian state with mandatory interracial marriage.
A war that kills half the total population, and even more of the adult male population.
A postwar Nazi hideout (I guess most people know this bit though).
I am somewhat late to this, but since I’ve noticed some Polish people here, what is your take on relatively recent (September 6) decision if Polish central bank to cut interest rates?
Main take from international sources, which are tradionally quite hostile to current Polish government, seems to be that this is politically motivated to help the government win the upcoming elections. That seems... odd?
Like, I get a generalized mechanism “central bank lowers interest rates to boost incomes and employment in the short term, which is popular, but it leads to long-term instability”, but in this case, since the elections are on October 15, there doesn’t seem to be enough time for the boost to materialize?
So, main effect visible to voters, unless Polish households owe huge amount of floating rate debt, seems to be very conspicouos fall in the value of Polish currency (2.7 % in last 30 days; admittedly, this isn’t so bad, but looking on the chart makes it obvious to everyone that it was caused by “something” which happened on September 6: https://www.xe.com/currencycharts/?from=PLN&to=EUR&view=1M).
I recently got pissed off by yet another online "expert" telling me I needed to excersise an hour daily, with another expert telling me I needed to meditate for 30 minutes daily, have a morning ritual etc etc.
I get the impression most of these online experts are 22 year old kids with no family or demanding jobs. Wrote a post about it:
The Attack of the Online "Productivity" "Experts":
Are you having fun? What could be done to make this game of life more fun and interesting? I'm not having much fun, but I think it's largely because of skill issues. How do you "get good"?
In all the discussions about AI Doom, there is a lot of variability about what "doom" means. (Thus the confusion between AI not-kill-everyonism vs AI not-racist).
I've never found a classifications of the levels of success/failure we're trying to work with.
So here's, humbly, my classification of terrible AI scenarios from worst to best:
1) An AI that is not quite fully general accidentally kills all life on earth, then shuts off because it doesn't know how to keep itself working. Life doesn't come back. It so happens that earth was the only life in the universe. The end.
2) Same but aliens come some time later and discover the great works of humans and celebrate our dead civilisation.
3) After the first human civilisation disappears, a new intelligent species rises and builds it own civilisation.
4) An AI accidentally kills all life on earth, and then keeps running until the sun dies.
5) An AI voluntarilly kills all life on earth and fullfills its internal goal (whatever it was)
6) An AI decides to pave the universe with computronium / paperclips / whatever, but is not smart enough to get out of the solar system.
7) An AI decides to pave the universe with computronium / paperclips / whatever, and succeeds.
8) An AI decides to pave the universe with computronium / paperclips / whatever. In the process of doing so, it discovers a lot of knowledge about physics, mathematics, engineering, even philosophy.
9) An AI decides to pave the universe with "meaningful" computronium. I.e. computronium that we would recognise as having qualias.
10) An AI decides to pave the universe with "meaningful" computronium. That computronium makes works of art, has feelings.
11) Same but there's a memetic relationship between the new art and the works of humans.
12) The AI-God decides to keep the human race in a zoo while it conquers the universe
13) The AI-God decides to keep all the humans that were alive at the time of its birth in a zoo where they can live forever.
14) The human zoo is so well endowed that humans can create great works of art, and achieve the apogee of what the human race can do.
15) Humans manage to fuse together with the AI, or upload their consciousness, so that they have the intellectual power to meaningfully contribute to the future of the AI civilisation.
16) It turns out that the rules of the universe make it impossible for an AI to work without humans overseeing it. The resulting civilisation looks like Star Wars / Foundation / Dune.
17) The god of the bible exists and has/will ensure that none of the doom scenarios occur.
I had most trouble on the last part of this classification : what would constitute for you a perfect success vis-à-vis AI ?
My wife and I have big problems with reoccurring miscarriage in the early stages. We are pretty desperate. We checked a lot of stuff and identified some possible, but not sure, problems and somehow managed them. However we are again in an unsuccessful pregnancy. Still in the early stages, but before any miscarriage the doctors diagnosed a non-progressing pregnancy, and advise us to remove it as it might cause infection.
We are thinking about IVF as some sort of solution. Does anybody have any similar experience or some kind of expert advice in this situation?
Open Thread 295
https://lookinthemirror8.wordpress.com/2023/10/01/9-11-was-possibly-done-by-david-bowie/
Dark humour, not for everyone but I think it's pretty funny/ interesting.
OC ACXLW Nuclear Proliferation history/Does sex belong in science 9/30/23
https://docs.google.com/document/d/1H8BpUe5zWMBVFf1DFJ65AnnSoHEbLNrFvv9w925zCbs/edit?usp=sharing
Hello Folks!
We are excited to announce the 44th Orange County ACX/LW meetup, happening this Saturday and most Saturdays thereafter.
Host: Michael Michalchik
Email: michaelmichalchik@gmail.com (For questions or requests)
Location: 1970 Port Laurent Place
(949) 375-2045
Date: Saturday, Sept 30, 2023
Time: 2 PM
Conversation Starters :
This week we have specific cases of situations that have broader implications on how we go into the future. We look at nuclear proliferation (which could be viewed as a special case of the more general problem of controlling dangerous technologies) and biological sex as a thought paradigm in anthropology (which can be seen as a specific case of socially controversial ideas struggling to find a place in scientific discourse).
Are nuclear weapons and the international agreements about them a good model for other technological existential threats (ASI, Bioengineered pathogens, nanotechnology, smart e-viruses, psychological warfare technologies)
How much should the scientific community change its research program to account for the sensitivities of the general population and potential harms? Is there a way to reconcile these conflicts? Will censorship and self-censorship cause more harm in the long run? Are there truths about people and society it’s better we just not know. Are there research agendas that, if we allow ourselves to pursue them, will lead to more wrong ideas than right ones?
Video, with YouTube transcript available:
Why Every Nuclear Power Built the Bomb (And Everyone Else Hasn't)
https://youtu.be/l8MkjxFq7pI?si=vrqqSlRwOFXM60UN
More mishigas: Two anthropology societies cancel an accepted symposium on sex and gender because it would “harm” their members
https://whyevolutionistrue.com/2023/09/27/more-mishigas-two-anthropology-societies-cancel-accepted-symposium-on-sex-and-gender-because-it-would-harm-their-members/?fbclid=IwAR0xsMUMuWYki3rQXXlLyqz_mJqdHgJ05nCoYlrAWfwEf--oB8ulr6xfb5M#
Walk & Talk: We usually have an hour-long walk and talk after the meeting starts. Two mini-malls with hot t
takeout food are easily accessible nearby. Search for Gelson's or Pavilions in the zip code 92660.
Share a Surprise: Tell the group about something unexpected that changed your perspective on the universe.
Future Direction Ideas: Contribute ideas for the group's future direction, including topics, meeting types, activities, etc.
Is the Chinese Communist Party Communist?
Arguments against: they are no longer in strong alignment with the ideas espoused by Karl Marx, who popularised the word "Communism" or various other historical groups who have called themselves Communist.
Arguments for: Communism is as Communism does. The Chinese Communist Party has 98 million members and vastly outweigh any other Communist group in the world, so if they say they're Communist then who are we to disagree? Karl Marx is just one guy, and he's dead, why don't the 98 million members of the CCP get a say in the definition of what Communism is? Saying that the CCP isn't Communist is like saying the Pope isn't Catholic; if one billion Catholics agree that the Pope is Catholic then he is, regardless of what St Peter might have thought.
New working paper: Discursive Competence in ChatGPT, Part 2: Memory for Texts, https://new-savanna.blogspot.com/2023/09/discursive-competence-in-chatgpt-part-2.html
Abstract: In a few cases ChatGPT responds to a prompt (e.g. “To be or not to be”) by returning a specific text word-for-word. More often (e.g. “Johnstown flood, 1889”) it returns with information, but the specific wording will vary from one occasion to the next. In some cases (e.g. “Miriam Yevick”) it doesn’t return anything, though the topic was (most likely) in the training corpus. When the prompt is the beginning of a line or a sentence in a famous text, ChatGPT always identifies the text. When the prompt is a phrase that is syntactically coherent, ChatGPT generally identifies the text, but may not properly locate the phrase within the text. When the prompt cuts across syntactic boundaries, ChatGPT almost never identifies the text. But when told it is from a “well-known speech” it is able to do so. ChatGPT’s response to these prompts is similar to associative memory in humans, possibly on a holographic model.
Good article in Science News, about brain implants treating depression.
https://www.sciencenews.org/article/brain-implant-depression-electrode-stimulation-surgery
Interesting that the model is not biochemicals.
Oh my. I’m becoming hopelessly unhip. Just working tomorrow’s Times XWord. Got FINSTA on the crossings but I had to look it up.
I'm looking for a scene in a movie -- or it could be from something on TV, or some YouTube video -- that contains an example of what I would call a good fight: Both parties are quite hurt and angry and say and that, via raised voices, tears, etc. But they refrain from treating each other badly. So they are not sarcastic, belittling, mocking, and they don't accuse each other of huge character flaws -- stupidity, not caring about anyone except oneself, profound sneakiness or dishonesty. They stick with the topic, each talking about why their point of view makes sense, and how hurtful and infuriating they find the other's point of view.
"I hate it when you do X. I've told you that many times and you keep doing it anyway."
"You know why I keep doing X? It comes naturally to me. I've done it all my life. And nobody else complains about it. You didn't either, til last year."
"That's true, but a lot of things changed last year, and that affected how I feel about X, and I think that's a valid reason to object to it now. And I've told you that too, but you still keep doing X."
"Yeah, OK, things did change, but . . ."
That kind of argument. Seen any examples of it?
My therapist recommended that I stop trying to "create intimacy" on dates and instead "cultivate intimacy." This means no hugging, telling jokes, etc. How do I affirmatively do this?
For context, I am a white American man in the US, and Chinese women are WAY more interested in me than anyone else is. (This only applies to women born in China. Chinese-American women ignore me completely.) What does a successful date look like in China?
Huh, well Trump has one less trial to prepare for: a New York state judge just issued a summary judgement that Trump, his three adult children, his business and a couple other officers of that business are all liable for obvious business fraud committed repeatedly for a decade. The ruling says basically that the repeated fraud is so obvious from the documented facts that holding a trial would be a waste of a jury's time.
In the US adversarial legal system, motions for summary judgement are routinely filed by both sides despite being very rarely granted. I just asked two veteran trial attorneys of my acquaintance who each said they have never, during their successful legal careers, had a motion for summary judgement be granted. One guy said his law partner did once.
I assume Trump et al will appeal this ruling. In the meantime though the case moves on to a penalty phase.
Does anybody have good examples of either, or both, of the following? Of the two, I'm more interested in the second:
1. An idea or tool that should rightfully be popular and widespread, but it has been rejected by a segment of the population due to tribal association? (ie the thing is obviously good from an objective perspective, but it has failed to catch on in the wider population because it became associated with the lower/upper class, the political left/right, femeninity/masculinity, etc)
2. An idea or tool that has escaped the aforementioned trap (ie something that *has* caught on in the wider population, without its popularity being bogged down by tribal affiliation).
Boring, mundane examples are perfectly good!
Also on that note, is there a single word or short phrase to describe the phenomenon of "being robbed of support because one tribe rejects it due to association with competing tribes"?
Re: 2: "Brandon Hendrickson...has a post up responding to comments"
From the refered to post(there was no option to comment there but I did subscribe):
"this is too important an endeavor to succumb to pessimism and gloominess."
Pessimism meet Blake:
"We paint the world around us in the colours of the world inside. Just as the driver who complains how bad the traffic was fails to recognise that they themselves were the traffic, so too do those who complain about being trapped in a terrible world fail to realise the extent to which their choice of focus has helped create that world.”
--John Higgs, William Blake vs the World, 2021
How do we adjust the "world inside" - if we need to?
What is biotech's killer app?
I'm sure I'm not the only one who's heard proclamations over the last several years that biotech in the early 21st century is going to be the equivalent to the information technology revolution in the 80's and 90's. Bill Gates famously said that if he were a young person today that he would go into biotech and the general consensus (hype?) seems to be that biotech is posed to impact our lives the same way the computing and internet giants did back then.
The more I think about this the more I become skeptical, and the reason comes back to the question posed above. To put it another way, what is the biotech equivalent of the personal computer? What is the product that biotech will eventually put in every house in America and every pocket in the world? Because that's what the information revolution did and that's what made it a revolution. Perhaps I'm simply not imaginative enough, but I'm struggling to think of a product that could come out of biotech in the near future (10-20 years) that would have an equivalent impact.
Let's say I cook a meal-prep dish (e.g. classic lasagna) today Tuesday, and I plan to serve it at a dinner on Friday. From a taste perspective, should I store it in the fridge or the freezer? Or will it taste the same regardless? (I'm totally uninterested in food safety, only taste matters.)
Signal-boost for a prediction market I've set up on a small open mathematical question: https://manifold.markets/HamishTodd/in-2025-will-i-teach-people-that-pr
The phrasing of the bet will likely be more interesting to people in comparison with the mathematical content. The mathematical content is whether the Fourier transform is the same as, or importantly similar to, the "Projective dual" - the FT being a vital tool in engineering/science, and the projective dual being a largely-forgotten tool in pure math that has become somewhat controversial because of its recent appearance in computer graphics (more detail in the link)
Here's the part that I think is interesting for PM people who don't care about math/fourier transform. Above I used the word "importantly" in that explanation, which is a highly subjective word, and that's a problem for the question. There are multiple *boring* connections one could make between the two things. Those wouldn't make the things "identical", but of course you could argue that they are enough to make them "arguably similar".
I've tried to solve this subjectivity problem by phrasing it in terms of teaching. Two things can be "similar enough" that a teacher would mention their similarity when teaching them. If they are REALLY similar or identical, it can be a no-brainer to bring up the similarity. If they were only a little similar, you might only bring up the similarity if you were personally keen on both. Currently, I do NOT bring up the similarity when teaching. Question is, is something about to change my mind?
Results from the PM were interesting. I started it out with 37% "yes". It shot up to almost 90% - but has slowly declined.
Liberals in western countries often show great admiration for the culture, art, religious traditions etc. of 'indigenous' peoples around the world, and this seems fairly independent of the specific aspects of these cultural traditions. It really seems like any kind of e.g. wood carving, no mater how basic, is treated as a great work of art so long as it came from the right kind of society.
Is it possible *in principle* for these sorts of people to broadly have a negative perception of a particular indigenous culture? Like believing, without malice, 'these customs are kind of backward', 'this style of art is very boring and lacks any real positive aesthetic qualities', 'these religious traditions are very basic and unintersting'. I really can't imagine this being so.
Cynically, it seems like belonging to the set of 'indigenous culture' is all that really matters here and anything made by primitive brown people lights up a part of their brains. Less cynically, the authenticity of these forms of culture that are neither commercial nor elite in nature can be said to inherently give them artistic as well as anthropological value and is something not usually experienced in developed countries.
And to be clear, I'm talking about the present day - whatever problematic views of indigenous culture espoused by otherwise progressive (for the time) westerners in the past aren't relevant to this.
I set up a market on NAEP score decline in 2024, curious for this group's take!
https://www.metaculus.com/questions/18402/2024-naep-scores-decline/
Does anyone know whether there's any data on whether support groups are helpful for dealing with grief? My son passed away earlier this year and I'm not coping well despite my best efforts (e.g. psychotherapy, generally looking after my physical well-being, spending more time around other friends and family, somatic and breathing practices). I could theoretically drive over an hour to go to my closest perinatal loss group (which is facilitated by a professional) every few weeks but the travel time is a substantial hurdle. I'm also skeptical that it would do any good. People keep recommending support groups in the abstract but if talking to my therapist doesn't help I'm skeptical that talking to a group of strangers would help. It might be worth the drive and the emotional effort if there's evidence that it would help in the long term
Please don't give me anecdotes; I've heard from people who love support groups and from people who think they're actively harmful. I would like data but frankly my brain has been fried by everything that has happened and I'm not up for searching through databases for well-structured double-blind studies when I get home from my 9 to 5.
When running into creative walls while slowly growing to hate a project, is it directionally better to train oneself to decouple creativity from flow (and over time to develop the discipline to proceed effectively at lower levels of reward); or to develop tricks to try to re-fool oneself into finding the thing interesting and to get oneself in a more flowing mental state (so try to get the reward back)?
Purely hand-waving here, but I suspect that it has something to do with low dopamine regulating the exploitation-exploration balance too far toward (premature/low-quality) exploitation. So, e.g., when you're deciding on a turn of phrase, or a musical solution, or an intermediate coding step or whatever, there is a strong pull toward good-enough, even when it really isn't. When you cave to this on the small sub-tasks, the quality of the outcome decreases, you perceive it, and you hate the thing even more.
FYI: Michael Lewis's book about SBF will be coming out on 10/3. Its title is *Going Infinite.* He will also be covering the trial, which begins on 10/2, on his podcast *Against the Rules.*
Why haven't I had a cold since 2019? I've had all my Covid shots and every single booster that they offer- could that somehow be why? I don't really have a great immune system and have always had at least 1-2 head colds a year literally my entire life (more when I regularly rode public transportation). Since Covid though, I haven't had a single cold. I don't wear a mask and I certainly go out to public areas, bars, the supermarket, etc.
Could the Covid shots somehow have given me a cross-immunity? I went out of my way to mix and match them, including getting an unauthorized extra shot before boosters were a thing. I think I've now had 1 of every shot authorized in the US (i.e. not AstraZenaca). Did it give me a strange new immunity to the common cold?
Oh! I thought the Egan mentioned in the review for The Lost Tools of Learning was Greg Egan, the science fiction author. No, it's Kieran Egan.
I don't know whether Greg Egan has written anything substantial about education, but there's a lovely section in _Diaspora_ where the smart-human level AI learns a concept (momentum, I think) and spends a while connecting it to a great many things it already knows. This is a skill worth having in school.
I've seen some writing from teachers about how hard it is to deal with educational fads. I would love to see fairly early education (age 10? surely no later than 12) about how adults have different theories about what and how to learn.
I don't think I read this review the first time around, so I'll just note that the illustrations are unusually beautiful.
Here's a substantial discussion of sentence diagramming: https://www.facebook.com/nancy.lebovitz/posts/pfbid02xuDx93EAJKamkgiZiAPvvr2Hu249A5CNgSvu6oksGMwbfGi5nHL7R5ju1GDvnft5l
And a little more: https://nancylebov.dreamwidth.org/1117333.html
People vary a lot about whether they were taught it and whether it was useful for them. I've considered an extended survey of what people learned in school and which things were valuable to them.
It sounds like the WGA won the writers' strike. That's not surprising - they were asking for rules and other changes that would have cost the studios millions, but had the ability to inflict far greater losses in the billions in lost programming and delayed schedules as long as they could avoid defection during the strike (which they did - they and their assorted fans on Twitter and elsewhere were pretty aggressive about attacking anyone who wasn't perceived as a supporter of the WGA). Eventually, the studios were going to decide that it wasn't worth it to contest the demands.
3. The comment on SpaceX is good. I'd add that I think the "waterfall" development style has persisted because of politics as well - it fits better with distributing parts of production and a supply chain across a broader area, and it's better at avoiding high-profile failures. No exploding rockets with the SLS program, for example - instead the failures show up earlier, and get labeled as lower profile "delays".
The replies to that comment were pretty good as well. The iterative development is mostly good early on - eventually you have to tighten down and ensure reliability, and that limits how much you can iteratively change the rocket as opposed to simply developing the new model of it.
I also wonder how much of it is just that they do almost all of it in-house. It speeds up communication time a lot.
Does the archaeology discovery of wooden structures from half a million years ago lead anyone to update on the probability of Ice Age civilizations? If you haven’t seen it, they found some notched logs from half a million years ago that indicate someone built a wooden structure in Zambia. It likely wasn’t from modern humans, but probably homo heidelbergensis. This would indicate that humans have built wooden structures for the entire existence of our species. [https://www.nature.com/articles/s41586-023-06557-9].
The improbability of something wooden surviving for so long and also being found, makes me think that there were a ton of these wooden structures. But I could see it being argued either way for whether this means there were more advanced societies. So far our evidence is that:
p(discovery | wood surviving) * number of paleolithic wooden structures built > p(discovery | stone surviving) * number of paleolithic stone structures built
On the one hand, you could argue that if the probability of discovery is the same, and it’s safe to assume that probability of stone structures surviving >>> wood surviving, but we found wood anyway, so that’s evidence that there definitely wasn’t stone structures.
Or, you could argue that number of wooden structures >>> stone structures, so of course we found a wooden one first.
See the earlier post [https://www.astralcodexten.com/p/against-ice-age-civilizations]. Scott listed three categories of claims: 1) civilizations advanced enough to build Stonehenge, 2) civilizations about as advance as ancient Egypt, and 3) civilizations as advanced as 1700’s Britain. I’d suggest another option (I’d call it level 0): that there were likely larger tribes living in wooden longhouses with probably enough sophistication to build simple stone monuments. I should also note that the Ice Age argument is about societies 115k years ago to about 12k years ago and this is way older.
I’m not an archaeologist and so I also wonder what is that chance that this is a false positive. What if the logs just rubbed up against each other in a way that looks like they were shaped?
If you're looking for more things to read - particularly on the concept of consumption, obselence, greed, internet culture, etc. I've been a marketer for 13 years and have recently created a substack exploring the concept of Humane Marketing. Goes in hand with doughnut economics and other concepts you've already heard of.
Would love to hear your thoughts on some of these topics and have you on my free substack.
Jake seliger is a stoic with cancer who has no tongue- he’s also my husband. He’s written a wonderful essay on stoicism and facing the end AND he’s recorded a podcast on the Daily stoic (which is amazing considering the hurdles). He’s the most resilient man I know and also a beautiful writer. Give him a read (and a listen, podcast link at the top of the essay)
https://jakeseliger.com/2023/09/18/stoic-philosophy-finding-a-meaningful-life-and-the-cancer-treatment-struggle/
Why is northern New England so lightly settled, despite being one of the oldest parts of the country? Maine, New Hampshire and Vermont were settled before the US was even a country, yet they have population close to the Dakotas. Maine's largest city has a population of 60k, Vermont's 44k. The northern two-thirds of Maine is about as developed as rural Alaska. Why? You'd imagine that the oldest part of the country would have developed a larger population over time.
If the answer is 'well they're cold/inhospitable'- Anchorage Alaska's population is close to 300k. I don't see northern New England as being any colder or tougher to live than say Chicago (country's 3rd largest city), Minnesota (Minneapolis population 5.7 million) Boise, or Milwaukee. Is the soil really that much worse for growing crops?
Part of the answer has to be political, because- stop reading this and look at a Google maps of 1 of these states. On the Canadian side you see a much denser road network leading up to Maine, NH, Vermont, and even Michigan- then it abruptly stops at the US border. So this must reflect some kind of political or legal reason?
Is ethnicity a constant like race which it is often paired with? People can be of multiple races and ethnicities but are the labels used consistently and constant over time?
Why is SF so dirty, dangerous and super-expensive?
Whose fault is it? City govt?
If this is a city of super-smart people, as many claim, and as evidenced by some of the businesses and innovation that exist here, why can't they solve this problem?
Is there a path from what it is now, to a safer city? (Even if it remains dirty and expensive)?
For anyone interested, I wrote some thoughts on how to improve inpatient psychiatric care.
https://open.substack.com/pub/socraticpsychiatrist/p/improving-inpatient-psychiatry?r=44y4c&utm_campaign=post&utm_medium=web
Apparently there is a "White People at the Function" meme, and a clip of Irish dancing was used for it:
https://knowyourmeme.com/memes/white-people-at-the-function
So the original dancers made their own version:
https://www.tiktok.com/@_cairde/video/7244940207345159467
I am interested in learning more about the sort of *stereotypically* feminine person who is content letting others make decisions, following orders and spending their life caring for others. Any reading recommendations on this subject? Preferably nonfiction.
An excellent YouTube channel that delves into the (sub)plots and themes of classic (Foundation, Hyperion) and contemporary (Three Body Problem, Blindsight) science fiction: https://www.youtube.com/@QuinnsIdeas
A lot of people from the US, at least as far as I can see (I don't live in the US) are talking about a terrible downturn in the IT sector, with lots of layoffs and bad experience job-searching. Is this really happening, to an extent that can't be chalked up to typical "frictional unemployment", and if yes, why? I heard ChatGPT thrown around, in the "AI has finally come for your jobs" sense, but I don't believe it - seems too soon.
How good is science in “the master and his emissary?” What’s Ian McGilchrist’s reputation among academics?
A while back on SSC there was a post about understanding Friston on Free Energy, but it's been years and now I would like to make my own attempt. I've been looking through the sources Scott pulled from, and (might) have a start on the basic idea, but the math and how the math could relate to the actual neurons is escaping me and I want more perspectives. Does anyone have recommendations for books/papers/posts on the subject that would be good for someone who is also trying to teach themselves about variational Bayes? More recent than 2018 would be extra cool. (or if you just have an opinion about free energy or predictive processing feel free to share.)
Fairly odd recommendation I know, but I recently happened to stumble across the Wikipedia article on Paraguay (specifically its history) https://en.wikipedia.org/wiki/Paraguay#History. Turns out it's actually very weird and lurid, definitely not what I expected. I recommend reading without spoilers, but if you need convincing then we've got:
A country ruled directly by the Jesuits
A utopian idealist totalitarian state with mandatory interracial marriage.
A war that kills half the total population, and even more of the adult male population.
A postwar Nazi hideout (I guess most people know this bit though).
I am somewhat late to this, but since I’ve noticed some Polish people here, what is your take on relatively recent (September 6) decision if Polish central bank to cut interest rates?
Main take from international sources, which are tradionally quite hostile to current Polish government, seems to be that this is politically motivated to help the government win the upcoming elections. That seems... odd?
Like, I get a generalized mechanism “central bank lowers interest rates to boost incomes and employment in the short term, which is popular, but it leads to long-term instability”, but in this case, since the elections are on October 15, there doesn’t seem to be enough time for the boost to materialize?
So, main effect visible to voters, unless Polish households owe huge amount of floating rate debt, seems to be very conspicouos fall in the value of Polish currency (2.7 % in last 30 days; admittedly, this isn’t so bad, but looking on the chart makes it obvious to everyone that it was caused by “something” which happened on September 6: https://www.xe.com/currencycharts/?from=PLN&to=EUR&view=1M).
On the other hand, alternative hypothesis, that the central bank is reacting appropriately to current conditions, also has problems. Annual inflation in August, although decreasing, was in August still above 10 % (https://tradingeconomics.com/poland/inflation-cpi0), while unemployment rate is lowest it has been in 25 years (https://tradingeconomics.com/poland/unemployment-rate]).
Apparently Fox News now crowdsources predictions on political questions, complete with Briar scoring for participants.
https://www.foxnews.com/americapredicts
I recently got pissed off by yet another online "expert" telling me I needed to excersise an hour daily, with another expert telling me I needed to meditate for 30 minutes daily, have a morning ritual etc etc.
I get the impression most of these online experts are 22 year old kids with no family or demanding jobs. Wrote a post about it:
The Attack of the Online "Productivity" "Experts":
https://new.pythonforengineers.com/blog/the-attack-of-the-25-year-old-productivity-experts/
Are you having fun? What could be done to make this game of life more fun and interesting? I'm not having much fun, but I think it's largely because of skill issues. How do you "get good"?
In all the discussions about AI Doom, there is a lot of variability about what "doom" means. (Thus the confusion between AI not-kill-everyonism vs AI not-racist).
I've never found a classifications of the levels of success/failure we're trying to work with.
So here's, humbly, my classification of terrible AI scenarios from worst to best:
1) An AI that is not quite fully general accidentally kills all life on earth, then shuts off because it doesn't know how to keep itself working. Life doesn't come back. It so happens that earth was the only life in the universe. The end.
2) Same but aliens come some time later and discover the great works of humans and celebrate our dead civilisation.
3) After the first human civilisation disappears, a new intelligent species rises and builds it own civilisation.
4) An AI accidentally kills all life on earth, and then keeps running until the sun dies.
5) An AI voluntarilly kills all life on earth and fullfills its internal goal (whatever it was)
6) An AI decides to pave the universe with computronium / paperclips / whatever, but is not smart enough to get out of the solar system.
7) An AI decides to pave the universe with computronium / paperclips / whatever, and succeeds.
8) An AI decides to pave the universe with computronium / paperclips / whatever. In the process of doing so, it discovers a lot of knowledge about physics, mathematics, engineering, even philosophy.
9) An AI decides to pave the universe with "meaningful" computronium. I.e. computronium that we would recognise as having qualias.
10) An AI decides to pave the universe with "meaningful" computronium. That computronium makes works of art, has feelings.
11) Same but there's a memetic relationship between the new art and the works of humans.
12) The AI-God decides to keep the human race in a zoo while it conquers the universe
13) The AI-God decides to keep all the humans that were alive at the time of its birth in a zoo where they can live forever.
14) The human zoo is so well endowed that humans can create great works of art, and achieve the apogee of what the human race can do.
15) Humans manage to fuse together with the AI, or upload their consciousness, so that they have the intellectual power to meaningfully contribute to the future of the AI civilisation.
16) It turns out that the rules of the universe make it impossible for an AI to work without humans overseeing it. The resulting civilisation looks like Star Wars / Foundation / Dune.
17) The god of the bible exists and has/will ensure that none of the doom scenarios occur.
I had most trouble on the last part of this classification : what would constitute for you a perfect success vis-à-vis AI ?
My wife and I have big problems with reoccurring miscarriage in the early stages. We are pretty desperate. We checked a lot of stuff and identified some possible, but not sure, problems and somehow managed them. However we are again in an unsuccessful pregnancy. Still in the early stages, but before any miscarriage the doctors diagnosed a non-progressing pregnancy, and advise us to remove it as it might cause infection.
We are thinking about IVF as some sort of solution. Does anybody have any similar experience or some kind of expert advice in this situation?