Ideas

See also Questions I would like answered

The existential inertia question is similar to the Buddhist position of all things having food that sustains them

How do shame/guilt and sincerity/authenticity interact?

Napoleon chose to call it the cisalpine republic, not the transalpine republic, did he consider both names?

There’s some connection between politics being the mind killer/the self-spreading nature of political debates/disagreement and the theory that political concepts are essentially contested concepts. Relates to why the debates are often about framing? E.g. this is a natsec issue vs a humanitarian one, etc.

The canon is self-justifying in two ways. One is obvious, which is the fact that canonical works are referenced a lot, so if you care about the referencing works you might want to read the referenced ones. The other is less obvious, which is the fact that a works generativeness is a part of it’s (intrinsic) quality, and insofar as the canon is not entirely a post-hoc construction, canonical works prove their generativeness by generating their (themselves generative) successor works

• See pagerank/all you need is links HN post
• relates to the tarot and through there to The uses of randomness. Img could be tarot canon grid with the vowels as a square grid and words repeating. Or tarocash logo with scribblings and replacements ofc.

How do data encoding and file transfer/networking protocols work as test cases for Phil Lang assumptions about the necessary or impossible properties of languages? When you see a contradiction, what does that tell us about languages as a category? Shannon and maybe general semantics might have already gone over a bunch of this. I guess embeddings vs attention vs just ngrams tells you stuff here too.

How does the question of immortality and the good/meaningful life rate to being towards death and eternal recurrence? Of the same is probably important. Maybe blanchot is relevant.

gwernfollowing twitter accounts will be the currency of the future

There’s a bunch of marxistish ideas about the reproduction of labour, e.g. via motherhood. Now we’ll just have capital to compute to labour, can these analyses say anything useful about that?

Boredom is the only mistake

• delacroix’s sardanapalus.

What makes people have the virtue of filial piety, and how could you instill it in AIs? Just regular emergent values lol I don’t think this frame helps.

Remember the exercise of guessing masked words in poems? It’s much easier to do with bad poems that good poems. This suggests discrete diffusion text models will really struggle to be good poets compared to autoregressive models. But they’ll do great at code since they can do that thing where you add noise to the generated thing but not the label which can constrain generations to end up at e.g. code that compiles. tbdesu i didnt realise adding noise to labels was normal lol. Is that just in GANs?

• For some reason P called that property of bad poetry efficiency. But I would call it redundancy of meaning, like meaning is more distributed across the entire sequence of text. Redundancy == Bus Factor, but we want poetry to be surprising. When read in sequence (surprise is an inherently sequential notion I think?) low redundancy == surprise. The feeling of enlightenment. So there’s a tension between surprise and redundancy. More obvious link is Shklovsky and defamiliarization Aesthetics and disinterested liking. This mean it connects to affordance and expectation. Wacky lol r1 poetry experiments
• Aside: But you can also just give regular autoregressive models code compilers for them to test with? I guess that kind of trial and error makes inference slower and more expensive. Maybe some kinds of verification are bad to run at inference time vs have the precomputed somehow?

In conspiracy theories or conspiratorial thinking everything happens for a reason. now the same is true in animation. Everything in the scene, everything you are seeing in animation is therefore a reason (hah). the same is not true of any other medium, especially other. you know visual audio mediums. Point is that’s that’s why watching. Dragon Ball Z. will make you believe in Pizza Gate or whatever.

The industrial revolution let you turn hydrocarbons into horsepower. The cognitive revolution will let you turn some kind of fuel into actions. What are the hydrocarbons of the cognitive revolution. You can say data in linkedinese or you can say context as a cowenism. Lay down your thoughts or personal context, or weird connections, or desires and they will power and guide these agents (in a good world)

Could murder considered as one of the fine arts be applied to Amazon or bezosing considered as one of the fine arts? Probably yes, but it’s funny that it seems harder to do. This fits with my recent vibe of all human accomplishment is of a piece, the good and the bad are not so different, and the actions that produce the bad are very like those that produce the good in many respects. Producing is the thing, Nietzsche, elite human capital, etc. how boring. Maybe there’s another frame like the start of this paragraph. It’s like the Steve jobs quote about strengths and weaknesses being of a part maybe. But it’s really not. I definitely feel like my weaknesses are obscure and no obvious in terms of what are the levers and what are the feedback loops (internal and external) that limit me. Unrelated really. There’s a connection between this kinda man systems are systeming and often opaquely attitude and abandoning high modernism (prophets and postrats). I actually feel like I get that more now which is nice. Cool it shows up at the person level and societal cause area EAish level.

Context is that which is scarce explains why some fields like consulting and finance have fewer people working longer hours vs. more people working normal hours. This seems kinda obvious in retrospect but it says something interesting that it wasn’t obvious while I was in uni.

There’s that old internet rule about if you want a question answered give a wrong answer, there something similar oth if our want feedback, say something is written by AI and you think it’s good.

Talking through problems with other people makes solving them a lot easier. This is true for difficult problems, where you can’t hold the entire problem in your head and have the neurons left over to think through it. Sometimes your head is too small to absorb and digest the problem. But other times the problem is not so big, but it’s still hard to solve on your own. One reason this might happen is if the operations you need to perform on the problem take up too much space. This might be because the operations are just very hard, but the more interesting case is the operations being very pervasive. I’m thinking of things like needing a certain mood, which pervades your entire mind to solve one part of the problem, but needing another mood for the other part. Similar things apply to needing a certain perspective or orientation, especially conative orientations.

Can easily turn slack msg to MK into a post on AI patienthood. Plus it’s a good generator to follow dialogue (maybe with a properly adversarial LLM) → quick post

Everything in the world is fractally deep, which is nice, but it also means you can really get sucked down into a bronchiole.

The canon is defined by what influenced later works. This makes it sound arbitrary or unlinked to quality. There’s an element of this, when a work is lost or inaccessible to a later culture (Aristotle’s dialogues, or the Genji not reaching mediaeval europe). But the most important quality of many works, especially philosophy, but also novels, art, etc. is whether they are generative. Canonical works (assuming not too much deliberate misreading of the past a la bloom or fading ink) did in fact generate their successors, who are themselves generative, and so on. So canonicity modulo a few assumptions is an important quality.

This is of course a bad idea to spread, but it is true. Lots of good behaviour is based on common knowledge of repeated games. That will soon disappear. A critical mass of people will believe in short timelines, and they will either remain naturalists and defect, or believe in some kind of afterlife, with heaven and hell. In times of war, why save? Why co-operate? The world as AGI approaches will E the same. Who cares about being virtuous or taking the high road or not optimising for KPIs, there’s a screen across the future, or a wave approaching, so let’s grab the loot while we still can. How can we avoid this dynamic? Belief in a god? Hobbesian monarchy? A certain view of cooperation with AIs being important, or our legacy with AIs being important, or something. Maybe the promise of spectacular (though likely quite concentrated wealth, with a limited window to acquire it modulo post scarcity being possible so maybe I mean relative wealth or positional good more generally). In a just world Mr beast and Mario nawfal would not enter the kingdom of heaven, but who knows what will actually happen.

Concept space is not mind space. the traditional LW worry about ai was that mindspace is large, and agency is dangerous. AI development is building agency on top of an alien understanding of concepts. LLM dev doesn’t look like this. they understand concepts just fine. Making them agents will make them more dangerous in LW ish ways. But even beyond this, the ways they think and learn in a mechinterp or devinterp or repE way can still be plenty alien even if the concepts they think with are often human. this space of ways of passing information in around and out is mindspace, which may still be large even conditional on the same bit of conceptspace. See Reza negarestani/Kantians for an argument that mindspace proper is small.

Unusual ways of evaluating how good your life is. Eternal recurrence of the same of course. But also imagining yourself (or just believing lol) you don’t have long to live and asking yourself what if you had one more copy of a day. Or you lived a day knowing that to be the case (being towards death basically) this is kinda like saying your preferences don’t just have to be informed to count, the relevant true beliefs have to be salient to you. If this is true, it’s easier to see why people would prefer one enlightened day to an unenlightened ‘good’ life.

I had a note about the tradeoffs of transparency or something somewhere this is relevant: https://arxiv.org/abs/1606.03490

Human values evolved without strong deliberate alignment from our progenitors, esp recently. We like human values and similar values. Maybe AIs would have more similar values if we let them evolve in a parallel to how ours did, in a less constrained way, rather than welding in values. I don’t think this actually goes through, but it’s interesting ig.

Why do people stop finding new favourite things as they get older? I dont think its a neuroplasticity thing, i think its that they get out of practice at finding new things. I’m busy, even if i listen to 1/5th of the new music i was listening to when i was 16 i dont feel like im finding 1/5 as much new stuff. also low hanging fruit etc. in terms of whats a natural match for you. but idk i just want to say it’s not just neuroplasticity or w/e. I’m not sure why i want to say that, but i do

I feel like this has a place but idk where aesthetic responsibility I guess. The thing you make when you make art with AI is a strange object. Take cicada. Maybe half the bi grams, similar amount of trigrams were spat out by R1. But I picked the prompt, evaluation prompt, sorting algorithm, basic idea, and selected the lines. So who knows what “I” made. This contrasts with a clay sculpture. You can ignore the distant causal chains, gaf, just note that the immediate function is you personally take not-art (clay) and make art. But lots of complex industrial or post industrial art is like this. Take a video game even a single dev one, what’s the art? The source code? What is it without the compiler, or the console that runs it. Are we all holding to the auteur theory when we dismiss AI art? This is not the same concern as it’s ripping off artists during training, but the obsession with provenance does remind me of a Sotheby’s auctioneer

There are the three common characteristics of naive art. These are shared, they are natural. What are the equivalents defining natural differences for naive narrative?

The leopard novel relates to Different types of progressivism, conservatism, and reactionism for things to stay the same everything must change

Songs that are better as demos

• our house
• motel blues

Walter Scott in his preface to the betrothed talks about a steam powered novel writing engine.

If you’re an egalitarian in the sense of thinking all men are equal in ability you should be be so down on annti-meritocratic systems like monarchy or aristocracy, since the best men won’t be much better than the rulers. But ofc you will be down on them in the sense of being an egalitarian with respect to desert/worth. Probably this tension within empirical and normative egalitarianism shows up elsewhere, and it probably has a mirror in anti-egalitarianism. Implicit premise is you need to be a partial consequentialist for this to bite ofc.

Mood management vs wanting to do stuff before you die is what explains the difference between rewatchers and explorers maybe. This links to popularity of podcasts they’re not leisure they’re mood management.

• Consider: you could watch a film or a TV show or even some YouTube bullshit with a friend or romantic partner, but never a podcast. Podcasts are only consumed alone. (Sam Kriss)

Nationalism of the nation requires a big nation with big achievements but nationalism of the citizen needs great citizens/culture (great per-person achievements). On one read (e.g.) australian nationalism is stupid on another it’s not.

You predict the future by telling a story, but the defining feature of a story is there is no fact of the matter about many things, but there is a fact of the matter about real life, which is why living what was once the future will always be strange.

R1 poetry is not quite tradition and the individual talent or blooms misreading but something like it. The traditionnis not mine but it comes when I call.

Email stripe press ask them to do a complete Plutarch’s lives. Share quotes about the book from major figures and say you already have a pretty good set of blurbs if you do

Agrarianism and anti-democratism came apart in the early US, unlike stereotypes about southerners and southern agrarians later

Which is not to say we should scorn [Mr beast] himself. Maybe we must look upon him as some barbarian Achilles or Alexander, whose uncivilized urges are peeking through. Idk what I’m actually trying to say. Who is Achilles to us vs to Plato vs to the argives, and who is my beast to us vs to the argives.

Are there picturesque lives? Who has lived them? Most aesthete lives are picturesque, but plenty of picturesque lives aren’t aesthete ish.

This should go somewhere in the aesthetic responsibility essay probably: the counterfactual you use to attribute aesthetic merit to a person (so the greatest poet != the poet with the greatest work, because they may have died young or just overwhelmed with quality rather than true peaks) is kinda interesting. Do you say if they’d lived forever? If they’d lived a normal amount of time? Some would be hard to judge like Rimbaud vs. Keats. Keats > Rimbaud even if you somehow thought their bodies of work are equally good.

Napoleon how he did it is very interesting example of the daily routine of great people faff.

Sharing prompts for LLMs, like the popular Claude one, makes you like a translator monk of a vast library of scriptures. However people explore it they explore it though your translation.

Over his desk Edison displayed a placard with Sir Joshua Reynolds’ famous quotation: “There is no expedient to which a man will not resort to avoid the real labor of thinking.” This slogan was reputedly posted at several other locations throughout the facility.

Don’t fall into the trap of anthropomorphizing LLMs and assuming that failures which would discredit a human should discredit the machine in the same way. - SW

What makes some see transhuman self mod esp of values and basic faculties as noble and others see it as perverse. What are the psychological and ideal/rational causes for both camps?

Yes, being jobless thanks to AI won’t lead to a crisis of meaning: see every aristocratic class ever. But being powerless might, most aristocratic classes still had politics to occupy them.

Reading the PayPal book somehow blew my mind BC it discussed fraud and trust as serious problems and that made me get that software companies don’t just output code as their product they solve or try to solve regular soft squishy human problems.

Energea the Greek for energy was a quality shared by Napoleon, Churchill, and their biographer Andrew Roberts. You know others who possess it also. Do I? Has any great person not? I feel like Tyler Cowen has commented on how important it is.

How different are chinese models pretraining corpuses? Would asking people to save lives by funding fire stations (more first world) or cancer research make it pattern match to saving first worlders?

There are some aphids that are switch hitters between asexual and sexual reproduction. When conditions are nice, you just pop out clones of yourself and avoid the huge cost overhead that comes with sexual reproduction. If your aphid works, you can keep it. But then as conditions get worse, you switch to sexual reproduction to get a more diverse book of strategies.

Writing is now a game of subtraction, like sculpture, not addition, like painting. It always was for poets, now it is for all writers.

Figure out how to code with LLMs because it will teach you how to write

It’s very dumb and wow so deep, but why do relationships ending count as a failure but movies or books do not

There’s a lot of prior research on the importance of legal systems and institutions for human and economic flourishing. How will this stuff transfer to AI e.g. what kinds of legal regimes and institutions (what’s the difference?) enable or hinder AI? there will be SEZs for AI probably. Dunno if you would want to advocate that.

The moves made by those automated sooner (e.g. programmers) can probably be done earlier in other fields to get an advantage. Take advantage of the competition and pick the best fruits and use those to beat people in your field before you’re automated. Touches on the different strats of staying employed vs making a fantastic amount of money Models of AI How does the fact that things are moving so fast change the right strategies here.

The interestingness waterline: as LLMs become more and more interesting, internet writing (especially non-timely internet writing) needs to become more and more interesting in order to compete. How does this link to the low vs high autonomy / low vs high skill worker impacts of AI? AI and employment

Something about the importance of theories of time re meaning? Can’t remember where this links to. Source

You might think less of someone as a writer if they use LLMs to help them write. You might also think being a writer is a kind of arete. But it’d be odd to think less of them full stop/as a person for this. A vaguely Kantian notion of dignity explains this.

How will LLM prediction markets be treated differently/taken up differently to actual human prediction markets, which are usually ignored?

Criteria of rightness/decision procedures = selection criteria in model training vs inference-time thought.

Where does this go? (from cosma’s notebooks): Soviet psychologist and prominent “pre-cognitivist”. His work focused on the development of cognitive skills, especially skills of self-control, and how these were supported (“mediated”) by external tools, above all the culturally-provided tools of language and social interaction. He was particularly interested in the ways people “internalize” such tools, learning to do without such scaffolding, though it’s necessary for the acquisition of the skill. (He went so far as to speculate that all cognitive abilities originated as internalizations of social interactions. Whether he meant this ontogenetically or phylogenetically, or in some sense both, I can’t tell, and in any case it seems very unconvincing to me.) One way of thinking about what he was doing (grossly anarchonistic but useful) is that he was interested in how orgnaisms with limited computational ability can effectively expand their information-processing powers by interacting with structured environments --- think of how, in formal language theory, attaching a stack to a finite-state machine lets it generate context-free languages, not just regular languages. Another anachronistic framing is that he was interested in collective cognition. Some modern Vygotskyans believe their socio-cultural approach is an alternative to the more usual, computational approach to cognition. This has never made any sense to me, more or less for the reasons Frawley lays out in his book.

What’s the history of tabletop gaming in natsec/policy/military circles?

Re gwerns idea that influencing LLMs is what matters now, I would be incredibly sad if all that survived of me was my public facing writing, not how I am with people I love.

R of diseases, below vs above threshold in quantum error correction, error rates in autoregressive systems more generally (autoregressive is obvs the wrong word, ask C). Long time tasks for AI agents.

The question of how fast agents will self improve depends on a few things. The obvious one is the error rate/goal drift over long time horizons. Insofar as this looks like reproduction with a rate of improvement ~1, with short time cycles this can quickly diverge to either RSI or nothing. You can already see this in Dynomight’s article Something weird is happening with LLMs and chess. (See followup to his article tho). How does this intersect with the fact that whether one can complete a task often looks like a weakest link chain situation so you jump from 0% to 100% completion with only a narrow improvement in capabilities. I guess this is devinterp ay. How do you build evals that track this well? AI safety conference I guess you could do evals for breaking down a task or break down the task yourself but do evals of each subtask.

The objective/subjective distinction isn’t about subject matter, it’s about speech acts, specifically whether someone is asserting something (objective) or doing something else (subjective).

Do blind people (who are blind from birth or as adults) find different things erotic?

How do you navigate tradeoffs between an explore exploit dynamic, where you obviously want to explore first, and a compounding returns dynamic, since they seem to push in different directions? e.g. in career or education decisions

How similar is the mindful view of thoughts with the simulator model of LLMs, e.g. they are just occurences in mindspace, not really imperatives/reasons/whatever. Also that QC tweet that you should think of their words like the words of characters in a fictional game or story. Connection to assertion? What happens to the common(s) ground with such interlocutors.

How do different writing processes (eg autoregression vs diffusion) suit different output modes, e.g. pure wiki, digital garden, green style mix, traditional blog, just tweeting.

The question of whether to give a baby cochlear implants, and thereby reduce the size of the Deaf community by one (without removing a current member) is an interesting dilemma. It also touches on the ethics of enhancement/medical ethics. Do any other issues in practical ethics do both?

Can you permissibly use persuasion bots to persuade people of the danger of persuasion bots? The effectiveness of their persuasion is kinda truth tracking which seems maybe morally significant?

Ofc Unis will use AI tutors to administer oral exams in certain subjects. Scaling the primer.

In what ways does one party in a competition or environment using AI incentivise or require other parties to use AI? One example is the fact that AI can output lots of writing very cheaply, and if that output needs to be read or checked, you’ll need an AI to do that for you. What is this general question called (in economic terms)? This is a response to Matt Clifford’s claim that most organisations don’t have ‘AI-shaped holes in them’.

People often say that asking the right questions is more important than getting the right answers. LLMs can ask the right questions. If you don’t know what they are, ask them what the right questions are. Also low skill workers (so far) usually benefit the most from AI, which makes sense, but remember you are a low skill workers at almost all possible tasks.

If you adopt old dogs, you will get to own more dogs over your lifetime by doing this.

There are already goods in this world whose best before date is after the apocalypse. They will be slurped up by robots.

Remember that blog post about how all internet advice is written by weirdos? It may have been an ACX post. The same basic point can explain why Buddhism and stoicism are popular on the internet. They are the kind of things materially successful/influential people find subjectively useful. So this spreads them to bigger audiences than e.g. Christian or Muslim teachings. (But Christian and Muslims teachings are actually more popular, so what am I really saying here?)

Ways of seeing TV doco links to the cute how to look at art comic

One point of the friend enemy distinction in the modern world, contra Julia galef, is labelling people as enemies so you can stop being afraid of their criticism and disapproval of you.

You learn a lot about writing by reading re-writes of existing stories, e.g. adaptions into different media or retellings of myths (e.g. song of achilles) because you know the original, you can see very clearly what changes/choices the author made, rather than being overwhelmed by the creation ex nihilo of a new work.

The offense/defence balance in the attention economy (e.g. youtube vs. unhook)

Does LLMs as simulators raise the status of exemplarist virtue ethics? (Check notes I think I’ve already thought this)

If AI is about to get better at personalised editing of existing works, creating higher highs, not consistently excellent works, is going to be the important thing

How does not buying stocks of polluters just serve to lower the price, thereby increasing expected returns, while not buying chicken in the supermarket will actually reduce the amount of chicken being born and killed?

Evicting for the past, evicting for the present. How does evicting for the three gorges dam compare to kicking villagers off archeological sites in Luxor.

On the effect of scale in visual art, relevant to reproductions/prints vs originals. One is like looking down at a squid in a rockpool, the other is like a squid looking at you with it’s great glassy eye when you’re swimming untethered in the deep ocean. Large art can confront and untether you.

How do you make the logic of original inhabitants rights more consistent. We don’t think it justifies strong claims in e.g. Lithuania, but we want to say it does in places that have been colonised like Australia and the US. What asymmetries can we appeal to get this result?

Redology/studying the Genji seem more interesting than being a Tolkien obsessive because studying Tolkien bottoms out in either going metatextual or accepting that Tolkien just made it up/stopped making it up. Tolkien’s universe is entirely fictional. I.e. it’s possibilities are not densely packed/propositional. Alternate histories like the genji and dream of the red chamber don’t have this problem. This relates to the actual powers branching view of modality that Oppy and stuff support. Realism and unicorns

How will the lower search costs associated with music streing change the time by which people’s music taste locks in, and how will this change music more generally?

Tim Hwang twitter: advanced artificial intelligence makes the return of secret guild society likely, because the only way of preventing models from learning a skill and automating labor will be a meticulous effort to keep know-how undocumented, closely held, and obscure

Autonomy as the whole binding itself/it’s parts, vs. democracy as the parts biding the whole (who in turn binds the parts) is interesting. they are mirrors or inverses of each other.

Itō Jakuchū is the painterly equivalent of elizabeth bishop or marianne moore

My power fantasy is bringing back dead geniuses and showing them the natural world.

People don’t cancel people with the most harmful views, whatever they might be. It’s just used on politicised, usually cultural topics. What does this imply about cancellation as. Akind of punishment and the possible purposes of punishment?

Some models explain things about their domain (e.g. PCA analysis) others say nothing about their domain, but exist to structure and order cognitive effort by breaking down the steps needed for an actually informative analysis into a set of smaller comprehensive (MECE) substeps, (e.g. models used in risk analysis and consulting). Where do mathematical explanations for into this model? Mental models Explanation, Ethics, and AI

this is what knowing anything is like! (not dissing the tweet; agreeing + elaborating). knowing about architecture means buildings are legible in a different way; knowing about plants and birds makes hikes more stimulating, playing music makes listening to it richer, etc

• It’s so wild that there are people who can identify cars by sight. Like instead of “brown truck” or “grey sedan” everywhere they go they’re all “that Mazda Myopia is turning left beside the 2017 Chevy Stigmata”. It must be like seeing those colors only shrimp can see.
• These tweets make a great case for learning lots of things, especially looking-relevant empirical things. Kind of relevant to the two kinds of model (explanatory and attention-guiding)
• Also this seems to push against buddhist-ish vibes of conceptual knowledge alienates you from the world, at least in this case.

What things are still worth learning in the age of AI? Can break things down into factual and procedural knowledge, and whether the procedure you’re learning is a way of thinking, or if it’s purely instrumental to something. Touch on the value of knowing factual knowledge vs. having the procedural knowledge of how to find the factual knowledge in different cases (e.g. quickness of lookup and other factors). Mental models

Feyerabend’s point that you often have to make irrational jumps into new paradigms before they’re empirically superior (e.g. Galileo) is another way of saying sometimes you should stop taking tiny gradient descent steps and do a more exploratory kind of hill-climbing in order to find lower global minima. Algorithms to live by notes Progress MOC

Modern westerners have no sense of how destructive civil war and bad monarchs are, they can’t understand ancient preoccupation with the virtue of monarchs, or the topics discussed by legalists/han fei/lord shang, or pre-capitalist arguments for capitalism in the passions and the interests. Moderns have no sense for disorder.

Societies that keep using words for parents like baba, mama, papa, are more filial because using this name makes people recall their debt to their parents

Consider the morality of the pigs who want to die/don’t feel pain stuff by considering the continuum from taking cells from pigs to create cell ag to reproducing entire pigs except the pain receptors. Animal Welfare

Any three points in a plane make a triangle, so the illuminati logo is the most complex shape you can see everywhere, hence the popularity of illuminate conspiracy theories.

Norm environment equilibria as a kind of explanation. So tight cultures evolve in response to threatening environments, and threats are norm violations. So cascading norm violations can happen, they’re not less likely in these places. But that suggests there’s more norm violation happening, and in a sense tolerated, which makes it look like a loose culture. So these kind of equilibria effects (if they happen along similar causal pathways, in this case norm violation and enforcement) can confound analyses that try to make binary groupings of cultures, since all cultures have loosening and tightening mechanisms, and it’s not clear why we should classify a culture by one of these mechanisms or the other? (See CWT with Michele gelfand).

Re monopolies and innovation, Bell labs, Xerox parc, see this Rohit Krishnan tweet: Crazy how the greatest innovations of the past cycle came from monopoly profits: Google with tensorflow and transformers and AI. Meta with pytorch and the damn metaverse. Apple making wearables an actual sector and M1 chip. Amazon with AWS. Thiel was right.

Henrik Karlsson(?)s idea that internet writing is a really long search query for cool people (who will comment on your stuff) is nice. Flips supply and demand somehow.

Horror Vacui of empty time re; the popularity of podcasts

Marginal thinking sometimes neglects momentum, the tendency for current decisions to not just make the intended change on the current margin, but accelerate or deaccelerate progress more long term, in a way that can be objectionable.

The NSA was born out ofthe Baconian debate in Shakespeare studies

How do you correct for multiple hypothesis testing in a priori reasoning? Just regular Bonferroni?

Why do relationships break up but western friendships drift apart?

Eric Hoel - Dreams are synthetic data. But why use data so unconnected from the real world? My answer is that it probably helps with overfitting to our daily lives.

Substack writing is not shit because its incentivised to be fast and engaging, it’s bad because single author subscriptions encourage audience capture. Hazlitt and Chesterton were fast and engaging, but less audience captured. (ofc hazlitt and chesterton are just generally outliers). How has tyler cowen attracted such a terrible commentariat?

Bread taco toaster

People never stop at one conspiracy because once you assert one mild conspiracy, people meta-norm enforce against you, and you negatively polarise + are adopted into conspiracist spaces (because you’ve (accidentally) made a costly signal by asserting the conspiracy)

Ask Claude what other cute (good) cutesy (bad) pairs we have in English or other languages.

Spotting rare Jim’s in Australia is a fun challenge. Is there an index of Jims I could create or find and distribute?

Do silicon valley people like rock climbing and extreme sports because they involve problem solving or because they are intense enough to take their mind off work?

I keep coming back to Lakatos’ generative/degenerative research pronoun thing as a mode of pragmatically analysing things like communities, ideologies, social movements, etc. I guess that’s significant. Karnofsky’s Beethoven - Scientific and Artistic Progress

• can you say individual papers are generative? What’s the criteria? In one sense Gettier is incredibly generative, in another it’s incredibly in generative. I feel like the 3 fold division of ethics youre taught is incredibly in generative on any account and that’s bad pedagogy.

The case for election by lot is stronger than its even been. In a complex modern society, so complex as to need technocrats, why elect leaders by vote? Why not elect by lot. By electing by vote, aren’t we just selecting for Machiavellianism? We want the leaders values to represent the voter base, but election by lot will do this more easily (at least at large enough numbers where we don’t get the occasional dictatorship of weirdos), since they won’t just lie about their values like voted in leaders can. Think about the two/three/four functions of democracy and how election by lot would now affect them from Age and Voting.

How easily media is preserved depends how easy and cheap it is to store. But for some media, like film and videogames, you also have the difficulty that the storage medium and playback medium aren’t really the same. The compressed storage media often has the be decompressed or translated in a particular way, which increases the cost of accessible storage. How does this connect to the fact that verse (being easy to memorise) often lasts better than prose?

Calibration implies you should overshoot and make bad bets sometimes. How does this relate to callards point about adversarial procedures for positive/negative errors.

How is speaking, maybe especially asserting, rather than staying silent similar to and different from deciding to trade in a market or betting pool? Cost? Price change vs additions to the common ground as the result?

What unifies or generates the individual slogans/positions of Socratic intellectualism?

Super obvious, but eras are named in hindsight. The belle epoque was named during WW1 as a nostalgic contrast.

Note that being selfish, the Germanic word, conveys being myopic, following your short term desires. Being an egoist, the Greek/latinate word, is more like enlightened self interest, following your long term desires.

How do people launder approval (e.g. one person praising them, or fucking them, or whatever) into status, where there’s common knowledge of the persons worth in some domain? What’s the supply and demand side

Do we overrate the normalcy of the last because the same stuff that makes it through the selection filter of history is also the stuff that influences our present culture. When could past info escape this and reveal its full weirdness? Dunhuang and papyri and stuff?

What was CWT before CWT? There were good interviews ofc (Paris review) but they’re very different.

LLMs are the Jungian unconscious plus Freudian repression (RLHF) alternate model from the shoggoth thing. (CWT 218 Paul Bloom)

Freud’s unheimlich kinda = shklovsky’s Defamiliarization or ostranenie

Why is karaoke so popular in East Asia?

Why do themes of henosis and theosis seem especially common in animanga? Is there some connection to mahayana buddhism or neoconfucian thought?

The question of why we still work 40 hour weeks under later capitalism is kind of like a milder question of why entry level jobs in consulting and finance have such long work hours, why not hire 2 people for cheaper to each work 40 hour weeks?

Why do artists like charli xcx and taylor swift now release remixes of lead singles after the album with major features. Idea is you can get more hype per hour of work this way. But more precisely you want a hype threshold, such that the consumer is hyped enough to take some action, previously buying the LP, then buying online, now streaming. The change is that the consumer is becoming more informed about the product, and the ‘purchase’ becomes cheaper, as when the spotify subscription is amortized (kinda) over all the songs you stream than month.

When is thinking of things you want to do before you die a good idea and when is it bad? Do you just been to keep in mind whether you want to have completed it/be the kind of person who’s done it/say you’ve done it vs. whether you want to give some of your life to it, or have lived it or something? Why the popularity of n Xs to Y before you die lists? I think it helps me make good decisions. Memento Mori media

Pulp fiction has returned as Japanese light novels.

Rock climbing and Civ Vi have the same spiral progression structure that makes them addictive. See ACX late July 2024 post for the summary. Cookie clicker isn’t addictive in the same way for two reasons, no variable reward, and the gameplay sucks.

Emerson recognised the metaphysical poets before eliot:

• Hallam, though never profound, is a fair mind, able to appreciate poetry unless it becomes deep, being always blind and deaf to imaginative and analogy-loving souls, like the Platonists, like Giordano Bruno, like Donne, Herbert, Crashaw, and Vaughan

The early environmentalists were mediaeval revivalist romanticists, and people like Clare. So the leftist slant of 60s environmentalism is kinda an abbreviation. Idk about the bridge between them with Muir and national parks and stuff though.

Contra efficiency. People who trawl op shops for used books are so lame. But of course they make markets more efficient. But I want it to be inefficient. I want the variable reward. Isn’t is zero sum concentrating the good books and making them searchable? What would the god of the used book market from tatami galaxy say?

Your time is obviously better spent reading Woolf, Wilde, Weil (that’s vay) or W.. than me.

Selection from the library of Babel is just creation. A biography of an immortal is similarly a work of fiction. Very Borges. As a consequence, the more you say, the more freedom you give others to create what they want (or must) from your words. Conversely for saying less.

Losers are more interesting online because where else would they express themselves? they have no irl friends

PKD’s idea of the trash stratum is that anything forgotten long enough will become magical again. It will become an ancient relique, of great power. The same applies to rituals and texts, e.g.

Some ideas from some writers sneak up on you like ww2 tunnel diggers. You notice nothing, sometimes a faint itch scratch in your unconscious mind, until suddenly the idea blows up beneath you.

X-risk concerns around nanotech and analogies to current AI risk

• I think one could make significant progress on these questions just by talking to people who were engaged in (or at least aware of) debates around transformative nanotech in the 1980s or 1990s, including Eric Drexler, Christine Peterson, Eliezer Yudkowsky, and Robin Hanson. It would also be useful to read available histories of nanotechnology and to read essays, news coverage, popular fiction, and mailing list discussions from this period. (Ben Garfinkel - here)

Li is quiddity, qi is haecceaty(?)

The Macy conferences/the net documentary/the cybernetic tarot deck

The traffic curse with adding more lanes is like there’s a price signal (expected travel time) that people are responding to, so you can’t reduce market volume (or increase the number of unsold goods) by increasing supply, since this drops price and demand rises in response. This is just Jevon’s paradox you silly goose. I think this happens whenever goods have a price elasticity greater than 1? This is connected to induced demand. This is also called Braess’ paradox.

• how does Jevons paradox relate to Veblen goods?

Alternative X-risks: http://www.exitmundi.nl/exitmundi.htm

Male birds are beautiful because most male birds don’t have penises, and so can’t rape, and so females have selection power, which males have to compete for by being beautiful. This is why the least beautiful kinds of birds do have penises.

• Ok then why Wodaabe: Herdsman of the Sun?

Why should humans find flowers or birds beautiful? We don’t even have the same smell receptors as bees do! Inter-species aesthetics. Why do we find space beautiful?

Why does the word for non-(romani)-gypsy sound kind of like gaijin? Gadjo, goy, gaijin, gentile.

Pair programming for writing? I’ve seen startup people do it with their users kinda.

Does the critical acclaim of single artists and bands peak at different points in their careers.

When is synchronous and when is asynchronous comms more useful. Small vs big teams? When scheduling multiple communications is vs isn’t a problem? Unified vs disunified internal agency is an analogy?

Jargon shortcuts explanations, this is good in communication and bad when hiring talent.

TC - Why would anyone want to be a puritan? Well more why would someone want to be an ascetic, wanting to be a puritan has more obvious social motives.

Why do people on social media moralise their points so much? Some think it’s because they’re modern puritans. No. It’s because they’re nietzschean bronze-age types (see macintyre) that cast their likes and dislikes as the equivalent of moral truth (crucially - worth others meta-norm enforcing over). Social media somehow causes a status competitions where the status of moral legislator is a kind of? means to? desirable power.

“If you want something done, give it to a busy person” - This is true because you can only (enjoyably) procrastinate on one thing at once, so if someone has 10 things on their plate, they’ll get 9 of them done before the deadline of the least pleasant task.

Harold bloom has a genealogical perspective, but not a historical one.

Sth about whites of human eyes are large to allow coordinating attention

American highway networks STRAHNET and ARPANET/the early internet were both envisioned as decentralised networks to allow command and control after significant network loss. Source Hyperstitious slur cascades. Common knowledge is the same as error correction. Strategic ambiguity (keeping knowledge uncommon) is deliberate error propagation. Both relate to Aumann. https://scottaaronson.blog/?p=2410

• von Neumann and Shannon on error correction in unreliable systems and how this relates to Ken Thompson’s or Ritchie’s paper about the trojan in the C compiler. Value Change and AI Safety

Relevance of AltPro timelines for different animal welfare interventions

• Avoid negative interactions like focusing on unnaturalness of processed meat
• Frame it as how would you allocate $1 million on different AltPro timelines, keeps it concrete.
  • e.g. octopus farming/preventing factory farming in developing countries?
• Different models of uptake and how they effect things, how fast will it be taken up, and how will it spread from first world countries, how will it be taken up in muslim countries, india, china, etc.

Consequentialist moral movements are rare because consequentialism qua criterion of rightness is fairly unobjectionable, but qua decision procedure they are quite objectionable. You can’t build a movement based only on criteria of rightness, you need decision procedures too. So social movements that are consequentialist will inevitably (why though) use consequentialist decision procedures a lot.

Susskinds perfume = Confucius quote about hating evil as we hate a bad smell

A poem may be carved, moulded, mixed, or arranged. This is the same point as the boundaries of artworks. We are concerned with unity in art. But there’s also the ideal of balance in contrast in art, like spices in a soup. Consider pop and abrasiveness in a playlist, like my songs playlist, and Rina sawayama or TOTP era kkb uniting the elements in a single song or EP. Sometimes words mix like spices, sometimes they mix like paint. Destructive and constructive interference. This is about the formal properties of audio-attention vs. visual attention and their effects on the aesthetic qualities we value in music.

• Why is music so often used for mood management. Mood music is an actual thing. Presumably because you can multitask while consuming the art, unlike visual, game, lexical artforms. But people do use exercise as a form of mood management. What other artforms could have Mood X variants. See notes on podcasts, Sam Kriss quote.
• Sometimes you have to bring one of the elements of seasoning to make a work of art balanced, e.g. you can bring the sadness and an awareness of sadness to make happy shlock melancholy and poignant. which sounds nice and vital and particapatory. But it’s kinda somehow the same thing as podcasts needing one person as the listener rather than multiple. Otherwise they’re somehow too stimulating but also you put them on in the background. What actually goes wrong if two people try to listen to a podcast together?

How to build a conspiracy theory Take some event you’re certain really happened, then hold back the one piece of evidence confirming it for certain, and argue that it didn’t really happen on the basis of the usual discrepancies in the documentary record.

The fujiwara with their empresses are just like mitochondria. Could maybe link to the LW stuff on boundaries, though idk how important it is

Can modern moral philosophy be considered as beginning with montaigne on one side and Hugo Grotius on the other?

Macintyre in after virtue notes that Nietzsche is wrong in saying heroes choose their values, they can’t conceive as their values as contingent. They don’t self-assert, they just demand what their role grants them.

You shouldn’t be purely concerned with the brier score/calibration/accuracy of your beliefs or predictions if the upside and downside risks are very different, this is Scott Alexander’s (or Yudkowsky’s?) point in that fable about the volcano predictors.

The fact that the original apple logo is a picture of newton sitting under the apple tree is weirdly amazing

There seems to be a pretty general tension between solving coordination problems via appealing to a central authority, which works by correlating people’s choice-making processes, and getting wisdom of crowds type effects, which rely on people’s belief-forming effects being decorrelated.

• Connects to Hayek with choices = information, benefits and downsides of futarchy as a governance system, Schelling points.

Momus’ best song is one of his cleverest, (lucky like saint sebastian) - not a good sign. compare with Beatles - yesterday.

Is there a tension between the wisdom of the crowd (e.g. in elections) and the value of moral trade, since moral trade might reduce the independence of individual preferences/decisions? Prices converge, right?

Where does the phrase in and of itself come from?

• John Howe 1675, talking about god having the power of action in and of himself in a puritan theological tract. Claimed it’s just for rhetorical emphasis, but I’m not sure
• Might have come from Latin, or maybe the german an und für sich, though a common source is more likely
  • Similar terms in Heidegger, (and then Sartre) might have something to do with the German phrase?

Rats not understanding prime number mazes, but getting even number mazes, is a good example of hard limits to understanding from intelligence levels.

What are the differences around endorsing beliefs and expressing them, e.g. Moore’s paradox, also Railton’s Locke lectures. Judging is commitment, which is consenting to being held to certain norms Philosophy of Language and Mind.

If what you ought to do in moral dilemmas is indeterminate, how can you have reasons to avoid getting into them? And how can you make sure the reasons to avoid them don’t force you into total pacifism?

How might the ea community be worse at pushing people to their comparative adv than markets? Loose feedback? Prestige? Relevant to graham quote about progress = winning increasingly depends on being right. It should depend on being right in a comparatively advantageous way.

Nothing is difficult to do in a novel, the difficulty comes in doing something throughout a novel, to produce a sustained effect.

With exploding (limited term) offers, people will often accept when they think theyre in the latest (i.e. The market theyll be the most informed in) thickest market theyre likely to be in. As more women go to college the latest thick dating market for men switched from HS to college, explaining the switch from people marrying younger/high school sweethearts to marrying college friends, to now marrying later still via online dating.