Obviously hearing imaginary sounds is generally no fun, but it seems to me like its heavily implied (although I have found no clear source) that causality goes the other way around: being anxious and or depressed can make you hear imaginary sounds.
WTF?
Like no seriously under which model for the supposedly simple neurological process behind hearing does this make sense?
"Transparent. We care less about telling our readers what we think than showing them why we think it. Every part of the process – from our reasoning to our raw data – will always be out in the open."
Generally, I think this is the wrong priority and clarifying what you think is more valuable. The recruited writers are better than most, but I think this is the wrong goal even for them.
Anybody know of any good literature discussing the effects of low-skilled immigration in advanced economies?
Emil Kierkegaard's claim that low-skilled immigration is only good for the immigrant made me curious, and Caplan's rebuttal didn't seem that convincing.
1) Why does methylphenidate work immediately contrary to SSRIs despite that both serotonergic and dopaminergic neurons have autoreceptors that inhibit the recapture of their respective neurotransmitters?
2) Why hasn't anybody seriously tried to combine MAOIs with drugs that would prevent these side-effects from happening?
3) Why don't we prescribe exocytosis-promoting molecules such as MDMA when initiating SSRI treatments?
4) Why don't we prescribe autoreceptor antagonists such as pindolol when initiating treatment to make patients respond faster to the treatment and augment SSRIs?
Unfortunately I couldn't follow the more mathy parts. but my key takeaways are:
Memorization is easy but it can't predict unseen data. Generalization is hard but can often predict unseen data (or sometimes just predict seen data more efficiently).
The memorization circuit tends to grow in proportion to the amount of data it can predict. Generalization circuits scale better with large amounts of data.
Models may initially get "stuck" with memorization circuits but this can usually be overcome by increasing the amount of data. (Perhaps over-parameterizing the model helps with this?) There are pathways or interpolations between the two regimes. This allows partial generalization circuits that can't yet outdo the memorization circuit on their own, to get a toe hold.
They reverse engineered the algorithm the model uses to perform modular arithmetic. It's not what I would have predicted. It involves trigonometry and Fourier transforms.
But also they reverse engineered it! From the inscrutable list of parameter weights! What are the implications for interpretability?
Anyway there's a bunch more. They go into a lot of detail.
It's now required by law that you watch this video before you claim "Jan 6 USA was a miltiary coup". I passed this law, its domain of jurisdiction is all internet arguments that I see or hear or read about.
Substack need to fix their comments. I can click on a comment and have no hope of finding it if its merely a sub thread of 3 deep. The app doesnt even try. Reddit has solved this, its not technically that difficult.
So I'm attempting to calculate the plausibility of nuclear war causing prompt human extinction due to radiation alone.
est total global nuclear stockpiles: 5000Mt. Assume the war uses all of those but destroys the capability the manufacture more.
est total energy of radioactive decay of fallout: 10% of the energy of the blast
10% of 5000mt divided by the surface area of earth to a thickness of 1m is 4J/L or a dose of 4 grays versus the 5 gray LD50 for an instantaneous dose. But that's with the very unrealistic pessimistic assumption that the fallout spreads perfectly evenly over the earth before any of it decays, and then it all decays simultaneously. Realistically it'd be unevenly distributed and it'd take weeks for the winds to carry a small percentage of the dose to remote areas that didn't get nuked. Much of the fallout would decay before that. Also the dose would be spread out over time, making it less lethal. A dose of 1 gray is only a 5% chance of eventual death, so those remote areas are probably not dying off. A weakness in my argument is the arbitrary choice of 1m thickness for converting radiation per m^2 into radiation per litre or kg. Another weakness is not considering specific isotopes that get concentrated in specific organs, like iodine. The nuclear winter will suck but the San or amazon rainforest hunter gatherers will probably cope just fine with the collapse of the civilization they didn't depend on anyway and several degrees decline in temperature won't stop those areas from being habitable. So I'm ~95% confident that nuclear war wouldn't directly cause human extinction. Maybe it kills us in some other way like preventing coordination on AI or bioweapons.
I don't know if someone has already brought this up, but ACX posts are extremely slow to open from my phone, much more than other substacks, and comment threads give even more trouble (slow to load, slow scrolling, "expand full comment" button doesn't work). Sometimes it's slow on laptop as well.
This may be caused by the ACX theme or special settings (no likes etc) or possibly by the size of the comment thread.
I imagine it's not only me with this problem, if so I thought Scott should know about it.
The TLDR is that we used to have a mind split in two. One half generated social commands (eg. 'share', 'build a pyramid to store the chief's body') and the other half executed these commands. There was no introspection or rumination. Scott rejects both the timing and completeness of our psychological change, but gives Jaynes credit for unearthing quite a different psychology in the past.
I wrote a post speculating on what could have caused the development of a bicameral mind. When we had outer speech but not inner speech, I think the most fit subsequent adaptation would be to access reciprocal altruism. Something to remind us to play nice and fit in. Auditory hallucinations of societal demands--a bicameral mind--would be such a mechanism. https://vectors.substack.com/p/consequences-of-conscience
I had been moving in that direction for quite a few years previously - I had committed to being as fair-minded as possible; to avoid ad hominem criticisms; to try to speak of others in such a way that, had they inadvertently overheard what I'd said, I would not be ashamed; to accept justified blame for my errors graciously; and to try to restate an opponent's argument in non-pejorative language such that they could hear me state their case and agree that I had presented their beliefs fairly.
This is all somewhat aspirational, but I think Scott's post helped confirm I was on the right track.
After watching the YouTube channel of Sabine Hossenfelder, where every video seems like an advertisement for her paid courses... it seems to me that every controversy about this "brave female provocative contrarian physicist" is just a part of a marketing campaign.
I mean, she seems to be an expert, and her lessons are probably good. It's just all this supposed controversy that feels purely manufactured. She is bravely fighting... where no one is fighting back. She challenges the scientific establishment... by saying exactly the same things the scientific establishment was saying. Which is good, because it means she is promoting the established science... except for the part where she pretends that she is the only one doing so.
A question for other hypersensitives (I mean susceptible to sensory overload more than extra-sensitive senses, though I'd suspect they're correlated). I am curious if anyone else employs the same One Weird Trick: I'm nearsighted, and need glasses for things like driving or classroom learning. (What is the deal with overhead projectors being so much harder to read than black/whiteboards?) But sight unaided is enough to mostly navigate the everyday environment, minus some occasional squinting for roadsigns and other small-font-at-a-distance. So I just...don't wear glasses unless strictly necessary. Everything gets kinda fuzzy more than a few feet out, and that's comforting, cause there's significantly less visual data to (over)process. One small fragment of sensory deprivation, that most comforting of states...
Are all claims of AI bias (to do with e.g. race) baloney?
Nearly all of the ones I've seen are basically complaints that AI is characterizing reality accurately in a way that is incongruent with left-wing narratives.
A system that disproportionately associates criminality with black people (compared to white people) is called "biased", even though this is a correct description of reality. And we know that this is 100% what is happening, because literally none of these people have ever complained about e.g. men being considered men more associated with criminality than women (and before anyone interjects here, the black/white crime ratio is about the same as the male/female crime ratio - and in the most recent data it was higher for homicide).
We saw this earlier with policing algorithms that kept saying police should be in black neighbor hoods (and then when this was forbidden, it kept saying police should be in neighborhoods with things which are a proxy for being black neighborhoods). This is because that's where a majority of the crime is, and yet the left were furiously proclaiming that these systems are "racist"
It's really funny when you think about it. When conservatives would complain that CNN, MSNBC etc had a liberal bias, the epic comeback was that "reality has a liberal bias". Well, terrabytes of data have shown us that reality actually has a...non-woke bias? But no, I'm sure its the machines that are wrong.
If this is the kind of nonsense that is stealing so much attention in the AI space, any chance on genuine problems like alignment being solved seem like a demented pipe dream at this stage. The prospect of humans being wiped out is literally less alarming to people than an accurate characterization of race crime rate differences.
I realize this may be too unspecific a question, but does anyone have strong recommendations for print magazines that are not highly political (insofar as that’s possible), ideally just good, long form writing on subjects that would be interesting to read on a Sunday afternoon?
I’m asking because: (1) I’ve been through several of the mainstream mags but find myself disappointed or uninterested (Paris Review, etc.) in the content and (2) I like reading things in the physical world rather than on my phone.
I see the “women are less happy now” argument brought up in the comments section regularly as if women’s rights advances were a mistake. In my circles, it is considered self-evident that women are less happy because support for family care has not increased at pace with women’s participation in the workforce, so the burden of life for women is heavier than it used to be when a majority of women were homemakers. Women are dissatisfied because they want high quality and affordable child care so they can thrive in their professional lives the way men did when they had homemaker wives. (I have often joked that I need a wife.) Is there a seriously held belief in the community here that the answer to “women are less happy now” is a return to the midcentury housewife model? Or am I misinterpreting?
Looking at online tide data, I noticed that the peak high tides in my area are always higher when the high tide occurs at night than during the day, and the later the peak, the higher it is. Anyone know why this might be? Is it a general phenomenon or something unique to my local geography?
I thought it was very convincing (tho' to be fair I didn't need much in the way of convincing) and would be interested to hear a response from any EAs who want to mount a defense against the criticism.
The link in the mail and in the article (curently) seems to send me to the http (instead of the https) version of Asterisk mag. http is generally frowned upon, especially when you ask for information such as email. The usual practice is to redirect all http traffic to https. I have no idea what kind of control you have over the Asterisk website itself, but you should be able to edit the Substack link by putting "https://asteriskmag.com", which should redirect to the https version of the website. Cloudflare has an article on why it matters https://www.cloudflare.com/en-gb/learning/ssl/why-is-http-not-secure/, and I'm sure lots of people here can explain way better than me why it's necessary. In shrort: it's about protecting your users.
I'm excited about this magazine. I have a personal bias for orange and your writing, which is already a good start, but more than that, I can feel like the authors behind were smart, and dedicated to the readers. The desaturated colors are easy on the eyes, the headline is short and impactful, the "We are" part delves deeper on what I can find without overstaying its welcome, and saying what people will write about instead of who they are (achievements, stuff like that) while leaving a link for the curious is proof that you put your money where your mouth is: it's a space where people are judged on their actions rather than on who they are, that you're already applying the principles that you stated before. I also love the favicon.
That's why I think taking care of the https stuff is important. I'm being dramatic here, but without it it kind of ruins what you've built up until this point. I personally don't fear much outside of a few more spams than usual if my email addresss ends up somewhere I didn't want to, but not everyone is in this situation.
After reading Slime Mold, Time Mold's "Chemical Hunger" series and following their ongoing potato study I was reminded of the saying "Any diet works as long as you stick to it" (or maybe that was just the name of the meta-study that demonstrated this result: https://jamanetwork.com/journals/jama/fullarticle/1900510)
In psychotherapy there is a similar hypothesis—the "Dodo bird verdict"—that says that all legitimate forms of therapy produce about equivalent results. If we look around further I'm sure there are similar phenomena in e.g. weight training, software engineering (the "how to best develop software" flameware probably predates the Internet), etc.
If you squint a bit it looks like it's the same phenomenon repeating itself—following a protocol produces the desirable result, no matter what the protocol actually is. Given this I have two questions:
1. Is there a more general name for this phenomenon? I personally like the "Dodo bird verdict", so if you broaden the definition it could be applied to subjects outside of psychotherapy
2. Is there something else going on here? Given that the chosen protocol doesn't seem to matter, could it be that the positive effects are a result of being more mindful of your actions? E.g. if you've committed to a Paleo diet you start checking whether your meals follow the Paleo protocol; but maybe any weight loss you experience is not a result of eating like a troglodyte but simply because you're more mindful of what you're eating?
Do you know of a place I can read about the history of various countries but in a format that is shorter than a book ? Like 30-50 page summary of the history of various countries instead of 400 or 600. Basically long blog posts. If you know a history blogger that does this let me know. Also if you know similar history youtubers I'd be down too !
Has anyone ever successfully purged an earworm infection?
I mean of the auditory variety. Irritating unpleasant songs you heard once (or are made to hear often, thanks to low variance in ambient radio setlists) and can never truly purge from the brain's cache. That sometimes arise unbidden like not thinking of a pink elephant, and then must be consciously chased away by flooding the senses with other music, to much consternation.
This is the one thing I'd use the Pensive artifact for (from __Harry Potter__ universe). If the technology/therapy actually does exist to selectively suppress or erase unwanted memories, I'd like to...hear about it.
I was recently reading something of Scott's and Google prompted me with a related search for "scott alexander author". It was accompanied by a thumbnail of a guy who was clearly not "our" Scott - tanned, mustache, big grin - so I was curious enough to click through.
This guy apparently writes self-help books with a rhinoceros theme, like "Rhinoceros Success: The Secret to Charging Full Speed Toward Every Opportunity", "Advanced Rhinocerology" and "Rhinocerotic Relativity". I hope "rhinocerotic" is just an adjective from "rhinoceros", and not a portmanteau of "rhinoceros" and "erotic", but I'm not certain.
The whole thing is bizarre enough that I wondered if "our" Scott had created it (the Amazon and Goodreads pages, etc) as a joke, or as a distraction during the NYT doxxing debacle.
I'd never heard of this guy before - but presumably, in some specific niche and/or in terms of absolute numbers, he is the better-known Scott Alexander...?
FAO Scott and anyone else who likes traditional art, dislikes modern art, and is rich:
I came across this thread https://mediachomp.com/spite-patronage-and-cringe-commissions/ about how tech nerds could be commissioning their own operas, sculptures, etc - and how the Renaissance happened because of newly-rich merchants commissioning the type of art they wanted to see.
Dave Karpf writing on his difficulties with Longtermism:
"If we take the Longtermist/Long Now perspective seriously, then it is absolutely true that when we cure cancer just isn’t that important. 500 years from now, it simply won’t matter when the breakthroughs occurred. Every life saved will have long since ended.
But I have friends and loved ones who have died of cancer. I’d like there to be a cure sooner rather than later thankyouverymuch. Our world—the world you and I inhabit today—will be much more improved by potential breakthroughs in cancer research than by a fucking art project in a mountain owned by Jeff Bezos."
Any WW2 buffs here?: Is there a map of the population density in Eastern Europe just before WW2? My understanding is that most of the Soviet population was in the west, but then I don't understand how the Soviet Union could send millions and millions of people to the front in WW2 if most of their population were occupied? And wouldn't their industry suffer catastrophically (I know they sent it east of the Urals but how did they have time for that if it was all located in the west, as I presume)? Also, how did the Nazis occupy and keep control over such a large area and population: just through sheer brutality (I guess they didn"t have much of a plan)?
One year is too early yet to call it a trend, but maybe?
The U.S.A. had a spike in births (after a long general decline for over a decade) in 2021 that appears to be lead by college educated women working remotely.
It’s long been noticed that higher IQ’s (and educational attainment) increases the likely number of children men have, but decreases the likely number of children women have.
I've heard a number of social conservatives say regarding the overturning of Roe something to the effect of "men who pretend to care about this passionately are phonies; it doesn't affect them directly".
I've also heard, over the years, many social conservatives argue that the sexual revolution was better for men than women because it allowed more men to have sex outside marriage, which, according to the arguments, tended to hurt women in the long run.
I, a man, tend to believe the latter. The sexual revolution was much better for men than women, and legal abortion, along with the pill, was a big part of it. The notion of having sex for pleasure and without consequences was great for men who like having casual sex. But men are the gender more worried about the immediate consequences of sex: babies and such being a negative if it's a casual hookup; women are the gender with all their eggs in their ovaries.
As such, it makes sense to me, a libertine, that men should care more about legal abortion than women.
Why do Americans get so much spam / scam calls? It seems that this is such a widespread problem that people have generally abandoned the practice of picking up phone calls from unknown numbers.
Is it a technological problem in that we don’t have the technology to fight number spoofing?
Is it a legal/procedural problem that we don’t have the resources to adequately punish scammers from other countries?
Is it a political problem in that telecom companies don’t want to invest in technology/practices to reduce spam calls?
It seems like something that annoys everyone and devastates a small number of people who get scammed. Given that the median voter and politics are older, it seems like they should care about phone calls being trustworthy and not annoying. Why isn’t there more pressure/movement to defeat spam calls?
Anybody here making money as a freelance writer or translator? If so, I would appreciate some tips on how to start. I have some technical papers to my name (low double-digit Google citation indices), but no other published text that I can put on a resume.
Is anyone else having the issue where Astral Codex Ten does not show up in the Substack Inbox on desktop? And is the only one which does not? I've gone through all my settings to try and find out if I accidentally changed something, but I didn't see anything.
It seems to me that usage of reflexive pronouns has become incontinent over the past 20 years. People used to say or write, for instance: "James and I robbed the bank." whereas now they will say "James and myself did it."
Now maybe I just don't know what proper grammar is. I never really paid attention in school, so maybe I misunderstood what I thought I learned. I thought one should only use "myself" in cases where the subject is also the first person, such as: "I accidently shot myself while James and I were robbing the bank."
Today I read that "James is greater than myself." I really, really thought "I" was the appropriate pronoun in that case.
I also notice that public speakers use "myself" like it's going out of style. It would apparently kill them to say either "I" or "me", so it's always "James, Irene and myself" never "James and I" or "James and me".
Or am myself wrong about this over-usage of reflexive pronouns these days?
I visited it for 5 weeks, and I love the vision & the team behind it!
I decided to start a VC fund focused on startup cities, because I think we can build great startups there enabled by better regulations (e.g. peer country regulation, 3D on-chain property rights).
If you're an entrepreneur or innovator, I'd like to show it to you:
Interested to hear your thoughts… I’m going through a breakup. I initially thought this person was my soulmate. However, after a year of dating, we realized we were incompatible and we broke up. I went through the five stages of grief in less than a week, and I now feel completely fine, and if anything, relieved… People who’ve heard about the breakup have said that they’ve never seen someone handle a breakup so rationally. (I.e. since I know him and I aren’t compatible, I feel no need to be sad about the relationship ending.) That being said, why do you think I’m so indifferent to the break up? When most people make a logical decision, they still battle the emotional aspects of their decision.
Can anyone explain in technical terms why catecholamine testing isn't part of ADHD diagnosis? I read a study where they tested dopamine levels of kids before and after exercise and Richard Saul claims he diagnosed kids by having catecholamine tests done by mayo clinic so testing dopamine seems possible? I've asked a number of drs and psychiatrists and they all say haha that would be nice but it's impossible and I just want to understand why.
I've got a data-gathering question that will be simple for many of you. But I'm not too proud to ask it: I am gathering reports people make on sites that provide a covid PreP med called Evusheld. Will be compiling reports and making them available to later Evusheld-hunters, to help them find the best sites quickly.
I've made a form for people to give a report on. It comprises about 8 questions, followed by a spaces for people to write in answers. Some answers might be a paragraph long. Some items will not apply and people will leave them blank.
How can I collect these reports? There will be links on the site where people hunt for providers saying, "To leave a report on this provider click here." I could have the link go to a google drive document that's in suggestion mode, so that people's answers appear on the page as edits. The problem is that if 2 people try to report simultaneously, I don't know what will happen, but nothing good. Also, each new reporter needs a fresh document. They can't make a report on the one that somebody else just used
What's a simple way to do this? I do not need a system that compiles results for me some fancy way. This will work fine if I just end up with a bunch of separately filled out report forms. But how do I do that?
Please so not make any suggestions that involve hacking, writing code or anything on that level. I won't be able to implement them. I think I either need a clever idea about how to do this using google drive, which I'm comfortable with, or else an app or a site that will host the process and present a new report form to each new reporter.
I've written a blog post on when the COVID pandemic should be officially over: https://nsokolsky.substack.com/p/the-many-definitions-of-a-pandemic. Most interesting learnings were that the WHO changed their official "pandemic" definition in 2009 for the sake of H1N1 and that monkeypox is somehow not officially a "pandemic" yet. Feedback and more datapoints are welcome.
Longtime rationalist/weirdo here! I write short fiction and post it on ashkie.com (and eventually I’ll stop being neurotic and get around to cross-posting it to my substack)
ashkie.com/nowhere-man is my fave story I’ve done recently — & the themes of agentiness, difficulty connecting with others, power of knowledge, etc make me think ACX readers would enjoy it! check it out, let me know what you think :)
People used to be squeamish intensely judgmental about sex. We had the virgin/whore dichotomy. Active sexual desire was seen, in popular culture, as a kind of deviance to be judged.
Now we are squeamish about power. We have the victim/privilege dichotomy. Active desire for power is now seen in popular culture as a kind of deviance to be judged.
In what ways is this true? In what ways is this false? Am i wrong that there're an American tribal valence to this pair of dichotomies?
I just got a job at a small (~20 ppl) tech-startup! Check out airforestry.com if you're interested. Located in Uppsala, Sweden.
Does anyone have any tips or thoughts? Either about working at a startup in general, about their idea or that company in particular, anything is welcome :)
I will be working mostly in the workshop with physical prototypes, though I understand that in a small team everyone needs to do a bit of everything. I have a life-long burning interest in all things tech and have had several relevant jobs over the last decade, though never at a startup specifically. I don't have any formal education, but impressed them during the interview by showing some prototypes and projects I've done in the past. I start in a few weeks!
What makes GMU economics department so special? How did it happen that so many greats (Tyler Cowan, Robin Hanson, Bryan Caplan, maybe Garrett Jones) landed there?
Imagine you have these two alternatives: (A) eat half of a medium-size pizza, and (B) eat a whole medium-size pizza. Pizza is not a healthy eating choice, so both alternatives are bad, but obviously, B is worse than A. Also, for the sake of this question, let's say this is a one-off thing. Otherwise, you have pretty healthy eating habits.
I have the idea that option B is not twice as bad as option A, even though it involves twice as much pizza, but I don't have much theory or evidence that this is the case. Intuitively, if you eat half a pizza your body will process it normally, but if you eat a whole pizza your body will pass a lot of it relatively quickly. As a result, a whole pizza is worse than half a pizza, but not twice as bad. Is this roughly right?
(Note: I chose the half pizza/full pizza dilemma to keep things within reason. I imagine eating 10 pizzas is more than 10 times as bad as eating one pizza.)
Something that took me off guard in the last couple of years is the difference of my ability to remember things I've observed vs participated in. I used to listen to podcasts and be befuddled that I could remember details that the hosts themselves would forget or be hazy on from episodes that happened a year or so earlier. Similarly, as an aficionado of Diablo 2 speedruns, I'd be confused that streamers would forget that they had certain layouts in the map they were playing, or get details mixed up between different runs in the same day (D2 has randomised maps and items, so every run is different).
In the last year and a half I have begun doing both of these things myself (hosting and editing a podcast, performing D2 speedruns) and found that it's not that I had a better memory than the hosts or speedrunners, it's that my memory of events as an observer is far superior to my memory as a participant! This seems really counter intuitive to me, surely you should have better knowledge of what you yourself have done rather than what you've seen other people do.
So I just saw Garett Jones, a GMU economist, tweet this out- a proposed new system for selecting political candidates for office. The idea I guess came from Reddit, link below.
He proposes convening juries of randomly selected citizens to interview political candidates. It'd probably have to start at the very local level, so maybe 25-50 folks who are impaneled for a couple weeks for their local mayor, DA or sheriff's office. They review CVs, interview candidates, and have them do policy presentations, just like for a job. "Mr. Stewart and Ms. Jones, affordable housing is a major issue in our town. Please prepare a 20 minute presentation on your solutions, and then be prepared to answer questions afterwards." For transparency's sake all of this should probably be recorded/posted online.
You could have the jury then select the candidate, or you could have their selection go to a referendum, or at a minimum this process could be the new 'primary' with multiple winners and then the public could vote on 1 candidate from there. A new future jury could then examine their job performance at the end of their term and consider new candidates, etc. My guess is that having the jury select the winner would be fine for very local offices, and the higher up you go (state, federal office), it'd be more likely that this process serves as the primary instead. I would argue this process is about 100x more effective than current American political primaries.
Yes there are certainly some logistics to work out (it'd be tough for regular people to take such a large chunk of time off, though paying them could certainly help). I'd be curious to hear people's thoughts. Ideally the system could help avoid the demagoguery inherent in the current process. Being a legislator is a job, and it makes more sense to treat hiring them like a regular job interview than based on one's ability to give demagogic speeches or what have you.
Realistically, this would probably start at the hyperlocal level, maybe with District Attorney or sheriff elections (which really shouldn't be highly politicized/partisan). If it's proven to be effective, maybe it could work its way up the electoral ladder from there
This may be a dumb question, but in Singer's pond thought experiment, are there any arguments in favor of helping domestically versus internationally that aren't related to neglect or tractability?
On this page from Singer's non-profit, their argument for foreign aid is that 95% of American charity is already domestic. But assuming that a dollar spent at home could save the same DALYs as another abroad, is there still a case to be made for giving locally?
Scott mentions that the Pennsylvanian Quakers are the first historical culture to have recognisably modern moral norms, in regards to opposition to slavery/ treatment of women/child raising and other issues, in this post: https://slatestarcodex.com/2016/04/27/book-review-albions-seed/
I'd like to know if this is a fair characterisation of their culture and if they really were so different to neighbouring cultures/every other pre-modern society.
Also he seems to speculate that we inherited modern norms from the 17-18th century Quakers, does that suggest our current attitudes aren't the product of some kind of historical progress and are just as arbitrary as any of the other moral system that have existed?
It does seem like the Quakers had more historically novel views than modernity in some regards, like animal rights and pacifism, so maybe they were on a Whigish trajectory but were actually ahead od modern society in some ways.
I poked around in the dataset that Stable Diffusion is using as a corpus for training their model; LAION. It's basically constructed by scraping the internet, associating images with the text surrounding it, then providing a giant list or URL:s plus text snippets that anyone can download for themselves.
The first thing you notice by sampling random entries is how bad the text is. Picking something by random is likely to just be labelled "Funny images for laugh humorous HD wallpaper" than an actual description of the image content itself. I'm impressed that the model was able to suss out any content from this at all. The complicated queries people use to make the model create "good" images seems to be very much out-of-distribution!
There is also other AI-generated images in the corpus itself! Literally the first entry that contains the token decoded from "dragon" is this: https://live.staticflickr.com/65535/51363644206_c3e2880fa8_o.png - maybe this will lead to a future where the training corpus is "poisoned" by older models, leading to newer ones targeting the inferior output of older ones?
A solution for this that I can imagine is that people band together to curate content that they want generated; e.g. D&D people collecting all images of fantasy elves/dwarves/etc on the internet and writing detailed descriptions of them; which reasonably should improve images on that subject a great deal. You'd "just" have to create a website for it that people can use to collaborate and submit their work, and have the latest models regularly train themselves on on the output.
“religions are attempts to define what 'good' means, and it's impossible for a person to live and function without some definition of 'good', either implicit or explicit.”
If you like the piece I'd love if you would consider subscribing -- It's free, and I'm at least 75% confident the median ACX reader will enjoy my writing.
(P.S., last I checked self promotion was cool in these threads, Scott, sorry if it's not)
i haven't seen anyone else using it for this purpose, which makes me wonder how much of a bubble we're still in with these tools⏤i.e. just based on midjourney's discord, it's still >99% nerds creating bland fantasy renders. so what happens when people with actual visual-cultural literacy get into the game?
Asterisk sounds interesting but not I'm willing to go through the bother of a mailing list- hopefully Scott will link to it again when there's actually something to read.
I tried to find out whether Asterisk has a Twitter account, or whether it's already being discussed on Twitter. I discovered that another Asterisk Magazine launched just last month. This is an unfortunate name clash :(
Disclaimer: please don’t turn this into a culture war thread. I’m asking this to try and find information, not get into a big debate over the nature/existence of gender. Thank you for your consideration.
I have been confused about the idea of “passing” vs many transgender people’s seeming openness about being trans. It strikes me that being open about being trans would be opposed to my understanding of what passing is.
Obviously, trans people are not a monolithic hive-mind, so maybe some people just care about one significantly more than the other. But is there some other interaction that I’m missing, or am I misunderstanding the meaning of passing, or is it some other thing? Thanks
If the bet is aesthetic in nature, Stable Diffusion should suffice, which is publicly released soon (weights available to researchers already). If not then you probably want Imagen
It seems to be a commonly accepted medical fact that Tinnitus is associated with anxiety and depression.
( https://en.wikipedia.org/wiki/Tinnitus )
Obviously hearing imaginary sounds is generally no fun, but it seems to me like its heavily implied (although I have found no clear source) that causality goes the other way around: being anxious and or depressed can make you hear imaginary sounds.
WTF?
Like no seriously under which model for the supposedly simple neurological process behind hearing does this make sense?
https://francisquinn.substack.com/p/utilities-and-utilitarianism?sd=pf
Asterisk:
"Transparent. We care less about telling our readers what we think than showing them why we think it. Every part of the process – from our reasoning to our raw data – will always be out in the open."
Generally, I think this is the wrong priority and clarifying what you think is more valuable. The recruited writers are better than most, but I think this is the wrong goal even for them.
Anybody know of any good literature discussing the effects of low-skilled immigration in advanced economies?
Emil Kierkegaard's claim that low-skilled immigration is only good for the immigrant made me curious, and Caplan's rebuttal didn't seem that convincing.
I'm looking for feedback on my ADHD/depression pharmacology questions and hypotheses:
https://docs.google.com/document/d/1rJW5TInpW2rvd6CFkKCdlPxv-IMefBJ4PdQuff1uo6U/edit?usp=sharing
Thanks!
CC @Scott Alexander
—————————
Outline:
1) Why does methylphenidate work immediately contrary to SSRIs despite that both serotonergic and dopaminergic neurons have autoreceptors that inhibit the recapture of their respective neurotransmitters?
2) Why hasn't anybody seriously tried to combine MAOIs with drugs that would prevent these side-effects from happening?
3) Why don't we prescribe exocytosis-promoting molecules such as MDMA when initiating SSRI treatments?
4) Why don't we prescribe autoreceptor antagonists such as pindolol when initiating treatment to make patients respond faster to the treatment and augment SSRIs?
I found this article, "A Mechanistic Interpretability Analysis of Grokking" fascinating.
https://www.lesswrong.com/posts/N6WM6hs7RQMKDhYjB/a-mechanistic-interpretability-analysis-of-grokking
Unfortunately I couldn't follow the more mathy parts. but my key takeaways are:
Memorization is easy but it can't predict unseen data. Generalization is hard but can often predict unseen data (or sometimes just predict seen data more efficiently).
The memorization circuit tends to grow in proportion to the amount of data it can predict. Generalization circuits scale better with large amounts of data.
Models may initially get "stuck" with memorization circuits but this can usually be overcome by increasing the amount of data. (Perhaps over-parameterizing the model helps with this?) There are pathways or interpolations between the two regimes. This allows partial generalization circuits that can't yet outdo the memorization circuit on their own, to get a toe hold.
They reverse engineered the algorithm the model uses to perform modular arithmetic. It's not what I would have predicted. It involves trigonometry and Fourier transforms.
But also they reverse engineered it! From the inscrutable list of parameter weights! What are the implications for interpretability?
Anyway there's a bunch more. They go into a lot of detail.
Watch PolyMatter's "How Coup d’états REALLY Work" :
https://www.youtube.com/watch?v=_7nIqdwhdqA
It's now required by law that you watch this video before you claim "Jan 6 USA was a miltiary coup". I passed this law, its domain of jurisdiction is all internet arguments that I see or hear or read about.
Substack need to fix their comments. I can click on a comment and have no hope of finding it if its merely a sub thread of 3 deep. The app doesnt even try. Reddit has solved this, its not technically that difficult.
So I'm attempting to calculate the plausibility of nuclear war causing prompt human extinction due to radiation alone.
est total global nuclear stockpiles: 5000Mt. Assume the war uses all of those but destroys the capability the manufacture more.
est total energy of radioactive decay of fallout: 10% of the energy of the blast
10% of 5000mt divided by the surface area of earth to a thickness of 1m is 4J/L or a dose of 4 grays versus the 5 gray LD50 for an instantaneous dose. But that's with the very unrealistic pessimistic assumption that the fallout spreads perfectly evenly over the earth before any of it decays, and then it all decays simultaneously. Realistically it'd be unevenly distributed and it'd take weeks for the winds to carry a small percentage of the dose to remote areas that didn't get nuked. Much of the fallout would decay before that. Also the dose would be spread out over time, making it less lethal. A dose of 1 gray is only a 5% chance of eventual death, so those remote areas are probably not dying off. A weakness in my argument is the arbitrary choice of 1m thickness for converting radiation per m^2 into radiation per litre or kg. Another weakness is not considering specific isotopes that get concentrated in specific organs, like iodine. The nuclear winter will suck but the San or amazon rainforest hunter gatherers will probably cope just fine with the collapse of the civilization they didn't depend on anyway and several degrees decline in temperature won't stop those areas from being habitable. So I'm ~95% confident that nuclear war wouldn't directly cause human extinction. Maybe it kills us in some other way like preventing coordination on AI or bioweapons.
I don't know if someone has already brought this up, but ACX posts are extremely slow to open from my phone, much more than other substacks, and comment threads give even more trouble (slow to load, slow scrolling, "expand full comment" button doesn't work). Sometimes it's slow on laptop as well.
This may be caused by the ACX theme or special settings (no likes etc) or possibly by the size of the comment thread.
I imagine it's not only me with this problem, if so I thought Scott should know about it.
A couple years back Scott reviewed Julian Jaynes' Origin of Consciousness in the Breakdown of the Bicameral Mind. https://slatestarcodex.com/2020/06/01/book-review-origin-of-consciousness-in-the-breakdown-of-the-bicameral-mind/
The TLDR is that we used to have a mind split in two. One half generated social commands (eg. 'share', 'build a pyramid to store the chief's body') and the other half executed these commands. There was no introspection or rumination. Scott rejects both the timing and completeness of our psychological change, but gives Jaynes credit for unearthing quite a different psychology in the past.
I wrote a post speculating on what could have caused the development of a bicameral mind. When we had outer speech but not inner speech, I think the most fit subsequent adaptation would be to access reciprocal altruism. Something to remind us to play nice and fit in. Auditory hallucinations of societal demands--a bicameral mind--would be such a mechanism. https://vectors.substack.com/p/consequences-of-conscience
How did you all discover ACX (or its predecessor, SSC)?
In my case, my son* sent me a link to this post, which I really liked, reread recently, and still find very profound:
https://slatestarcodex.com/2014/02/23/in-favor-of-niceness-community-and-civilization/
I had been moving in that direction for quite a few years previously - I had committed to being as fair-minded as possible; to avoid ad hominem criticisms; to try to speak of others in such a way that, had they inadvertently overheard what I'd said, I would not be ashamed; to accept justified blame for my errors graciously; and to try to restate an opponent's argument in non-pejorative language such that they could hear me state their case and agree that I had presented their beliefs fairly.
This is all somewhat aspirational, but I think Scott's post helped confirm I was on the right track.
* And I should ask him how he discovered SCC.
After watching the YouTube channel of Sabine Hossenfelder, where every video seems like an advertisement for her paid courses... it seems to me that every controversy about this "brave female provocative contrarian physicist" is just a part of a marketing campaign.
I mean, she seems to be an expert, and her lessons are probably good. It's just all this supposed controversy that feels purely manufactured. She is bravely fighting... where no one is fighting back. She challenges the scientific establishment... by saying exactly the same things the scientific establishment was saying. Which is good, because it means she is promoting the established science... except for the part where she pretends that she is the only one doing so.
LineLoop: Hyperloop for The Line
Rodes.pub/LineLoop
A question for other hypersensitives (I mean susceptible to sensory overload more than extra-sensitive senses, though I'd suspect they're correlated). I am curious if anyone else employs the same One Weird Trick: I'm nearsighted, and need glasses for things like driving or classroom learning. (What is the deal with overhead projectors being so much harder to read than black/whiteboards?) But sight unaided is enough to mostly navigate the everyday environment, minus some occasional squinting for roadsigns and other small-font-at-a-distance. So I just...don't wear glasses unless strictly necessary. Everything gets kinda fuzzy more than a few feet out, and that's comforting, cause there's significantly less visual data to (over)process. One small fragment of sensory deprivation, that most comforting of states...
Totally Crazy Unique Thing, or a well-known hack?
Are all claims of AI bias (to do with e.g. race) baloney?
Nearly all of the ones I've seen are basically complaints that AI is characterizing reality accurately in a way that is incongruent with left-wing narratives.
A system that disproportionately associates criminality with black people (compared to white people) is called "biased", even though this is a correct description of reality. And we know that this is 100% what is happening, because literally none of these people have ever complained about e.g. men being considered men more associated with criminality than women (and before anyone interjects here, the black/white crime ratio is about the same as the male/female crime ratio - and in the most recent data it was higher for homicide).
We saw this earlier with policing algorithms that kept saying police should be in black neighbor hoods (and then when this was forbidden, it kept saying police should be in neighborhoods with things which are a proxy for being black neighborhoods). This is because that's where a majority of the crime is, and yet the left were furiously proclaiming that these systems are "racist"
It's really funny when you think about it. When conservatives would complain that CNN, MSNBC etc had a liberal bias, the epic comeback was that "reality has a liberal bias". Well, terrabytes of data have shown us that reality actually has a...non-woke bias? But no, I'm sure its the machines that are wrong.
If this is the kind of nonsense that is stealing so much attention in the AI space, any chance on genuine problems like alignment being solved seem like a demented pipe dream at this stage. The prospect of humans being wiped out is literally less alarming to people than an accurate characterization of race crime rate differences.
If we don't want spoilers for Dune, does the inclusion of God Emperor of Dune as a review contest finalist mean we're not allowed to vote?
I noticed a typo in your melatonin article from a couple years ago (https://lorienpsych.com/2020/12/20/melatonin/): "When a this kind of late sleep schedule"
I realize this may be too unspecific a question, but does anyone have strong recommendations for print magazines that are not highly political (insofar as that’s possible), ideally just good, long form writing on subjects that would be interesting to read on a Sunday afternoon?
I’m asking because: (1) I’ve been through several of the mainstream mags but find myself disappointed or uninterested (Paris Review, etc.) in the content and (2) I like reading things in the physical world rather than on my phone.
Thanks in advance for any recommendations :)
I see the “women are less happy now” argument brought up in the comments section regularly as if women’s rights advances were a mistake. In my circles, it is considered self-evident that women are less happy because support for family care has not increased at pace with women’s participation in the workforce, so the burden of life for women is heavier than it used to be when a majority of women were homemakers. Women are dissatisfied because they want high quality and affordable child care so they can thrive in their professional lives the way men did when they had homemaker wives. (I have often joked that I need a wife.) Is there a seriously held belief in the community here that the answer to “women are less happy now” is a return to the midcentury housewife model? Or am I misinterpreting?
Is a Trump/Desantis ticket constitutionally prohibited?
Looking at online tide data, I noticed that the peak high tides in my area are always higher when the high tide occurs at night than during the day, and the later the peak, the higher it is. Anyone know why this might be? Is it a general phenomenon or something unique to my local geography?
I am interested in a discussion based on the following premise:
Consciousness can be defined as the ability to imagine you are someone or some thing else.
Anybody read Erik Hoels piece on effective altruism?
https://erikhoel.substack.com/p/why-i-am-not-an-effective-altruist
I thought it was very convincing (tho' to be fair I didn't need much in the way of convincing) and would be interested to hear a response from any EAs who want to mount a defense against the criticism.
The link in the mail and in the article (curently) seems to send me to the http (instead of the https) version of Asterisk mag. http is generally frowned upon, especially when you ask for information such as email. The usual practice is to redirect all http traffic to https. I have no idea what kind of control you have over the Asterisk website itself, but you should be able to edit the Substack link by putting "https://asteriskmag.com", which should redirect to the https version of the website. Cloudflare has an article on why it matters https://www.cloudflare.com/en-gb/learning/ssl/why-is-http-not-secure/, and I'm sure lots of people here can explain way better than me why it's necessary. In shrort: it's about protecting your users.
I'm excited about this magazine. I have a personal bias for orange and your writing, which is already a good start, but more than that, I can feel like the authors behind were smart, and dedicated to the readers. The desaturated colors are easy on the eyes, the headline is short and impactful, the "We are" part delves deeper on what I can find without overstaying its welcome, and saying what people will write about instead of who they are (achievements, stuff like that) while leaving a link for the curious is proof that you put your money where your mouth is: it's a space where people are judged on their actions rather than on who they are, that you're already applying the principles that you stated before. I also love the favicon.
That's why I think taking care of the https stuff is important. I'm being dramatic here, but without it it kind of ruins what you've built up until this point. I personally don't fear much outside of a few more spams than usual if my email addresss ends up somewhere I didn't want to, but not everyone is in this situation.
Will there be a "Meetups Everywhere" thing soon? Last year it was in August. Wondering if I should wait until that to do another meetup in my city.
After reading Slime Mold, Time Mold's "Chemical Hunger" series and following their ongoing potato study I was reminded of the saying "Any diet works as long as you stick to it" (or maybe that was just the name of the meta-study that demonstrated this result: https://jamanetwork.com/journals/jama/fullarticle/1900510)
In psychotherapy there is a similar hypothesis—the "Dodo bird verdict"—that says that all legitimate forms of therapy produce about equivalent results. If we look around further I'm sure there are similar phenomena in e.g. weight training, software engineering (the "how to best develop software" flameware probably predates the Internet), etc.
If you squint a bit it looks like it's the same phenomenon repeating itself—following a protocol produces the desirable result, no matter what the protocol actually is. Given this I have two questions:
1. Is there a more general name for this phenomenon? I personally like the "Dodo bird verdict", so if you broaden the definition it could be applied to subjects outside of psychotherapy
2. Is there something else going on here? Given that the chosen protocol doesn't seem to matter, could it be that the positive effects are a result of being more mindful of your actions? E.g. if you've committed to a Paleo diet you start checking whether your meals follow the Paleo protocol; but maybe any weight loss you experience is not a result of eating like a troglodyte but simply because you're more mindful of what you're eating?
What is the quickest way to learn history ?
Do you know of a place I can read about the history of various countries but in a format that is shorter than a book ? Like 30-50 page summary of the history of various countries instead of 400 or 600. Basically long blog posts. If you know a history blogger that does this let me know. Also if you know similar history youtubers I'd be down too !
Has anyone ever successfully purged an earworm infection?
I mean of the auditory variety. Irritating unpleasant songs you heard once (or are made to hear often, thanks to low variance in ambient radio setlists) and can never truly purge from the brain's cache. That sometimes arise unbidden like not thinking of a pink elephant, and then must be consciously chased away by flooding the senses with other music, to much consternation.
This is the one thing I'd use the Pensive artifact for (from __Harry Potter__ universe). If the technology/therapy actually does exist to selectively suppress or erase unwanted memories, I'd like to...hear about it.
The Other Scott Alexander: Rhinoceros Guy
I was recently reading something of Scott's and Google prompted me with a related search for "scott alexander author". It was accompanied by a thumbnail of a guy who was clearly not "our" Scott - tanned, mustache, big grin - so I was curious enough to click through.
This guy apparently writes self-help books with a rhinoceros theme, like "Rhinoceros Success: The Secret to Charging Full Speed Toward Every Opportunity", "Advanced Rhinocerology" and "Rhinocerotic Relativity". I hope "rhinocerotic" is just an adjective from "rhinoceros", and not a portmanteau of "rhinoceros" and "erotic", but I'm not certain.
The whole thing is bizarre enough that I wondered if "our" Scott had created it (the Amazon and Goodreads pages, etc) as a joke, or as a distraction during the NYT doxxing debacle.
I'd never heard of this guy before - but presumably, in some specific niche and/or in terms of absolute numbers, he is the better-known Scott Alexander...?
FAO Scott and anyone else who likes traditional art, dislikes modern art, and is rich:
I came across this thread https://mediachomp.com/spite-patronage-and-cringe-commissions/ about how tech nerds could be commissioning their own operas, sculptures, etc - and how the Renaissance happened because of newly-rich merchants commissioning the type of art they wanted to see.
I'm really curious, what bet do you have going that depends on Imagen/Parti access?
Anybody worked with AlphaSights before?
Dave Karpf writing on his difficulties with Longtermism:
"If we take the Longtermist/Long Now perspective seriously, then it is absolutely true that when we cure cancer just isn’t that important. 500 years from now, it simply won’t matter when the breakthroughs occurred. Every life saved will have long since ended.
But I have friends and loved ones who have died of cancer. I’d like there to be a cure sooner rather than later thankyouverymuch. Our world—the world you and I inhabit today—will be much more improved by potential breakthroughs in cancer research than by a fucking art project in a mountain owned by Jeff Bezos."
See here: https://davekarpf.substack.com/p/against-jackpot-longtermism
I’m interested in learning more about Chinese history, and am wondering if anyone here has any (text)book recommendations?
Why are they "against easy answers" at asteriks? What is wrong with an easy answer as long as it's correct?
Any WW2 buffs here?: Is there a map of the population density in Eastern Europe just before WW2? My understanding is that most of the Soviet population was in the west, but then I don't understand how the Soviet Union could send millions and millions of people to the front in WW2 if most of their population were occupied? And wouldn't their industry suffer catastrophically (I know they sent it east of the Urals but how did they have time for that if it was all located in the west, as I presume)? Also, how did the Nazis occupy and keep control over such a large area and population: just through sheer brutality (I guess they didn"t have much of a plan)?
Maybe I should just play HOI4.
https://rationalwiki.org/wiki/Self-refuting_idea#Kettle_logic
Does a blog search website exist?
- Feedly lets you search your own list of RSS feeds well, but their global search returns a ton of junk from various websites, not just blog articles
- Google Blog Search has been dead for a decade
- Normal search for "topic" + "blog" usually returns a bunch of junk and not just blogs
- Even site:substack.com returns a ton of junk on Google
One year is too early yet to call it a trend, but maybe?
The U.S.A. had a spike in births (after a long general decline for over a decade) in 2021 that appears to be lead by college educated women working remotely.
https://ifstudies.org/blog/is-remote-work-behind-the-spike-in-us-birth-rates
It’s long been noticed that higher IQ’s (and educational attainment) increases the likely number of children men have, but decreases the likely number of children women have.
Until now (maybe?)
I've heard a number of social conservatives say regarding the overturning of Roe something to the effect of "men who pretend to care about this passionately are phonies; it doesn't affect them directly".
I've also heard, over the years, many social conservatives argue that the sexual revolution was better for men than women because it allowed more men to have sex outside marriage, which, according to the arguments, tended to hurt women in the long run.
I, a man, tend to believe the latter. The sexual revolution was much better for men than women, and legal abortion, along with the pill, was a big part of it. The notion of having sex for pleasure and without consequences was great for men who like having casual sex. But men are the gender more worried about the immediate consequences of sex: babies and such being a negative if it's a casual hookup; women are the gender with all their eggs in their ovaries.
As such, it makes sense to me, a libertine, that men should care more about legal abortion than women.
Why do Americans get so much spam / scam calls? It seems that this is such a widespread problem that people have generally abandoned the practice of picking up phone calls from unknown numbers.
Is it a technological problem in that we don’t have the technology to fight number spoofing?
Is it a legal/procedural problem that we don’t have the resources to adequately punish scammers from other countries?
Is it a political problem in that telecom companies don’t want to invest in technology/practices to reduce spam calls?
It seems like something that annoys everyone and devastates a small number of people who get scammed. Given that the median voter and politics are older, it seems like they should care about phone calls being trustworthy and not annoying. Why isn’t there more pressure/movement to defeat spam calls?
Anybody here making money as a freelance writer or translator? If so, I would appreciate some tips on how to start. I have some technical papers to my name (low double-digit Google citation indices), but no other published text that I can put on a resume.
Is anyone else having the issue where Astral Codex Ten does not show up in the Substack Inbox on desktop? And is the only one which does not? I've gone through all my settings to try and find out if I accidentally changed something, but I didn't see anything.
It seems to me that usage of reflexive pronouns has become incontinent over the past 20 years. People used to say or write, for instance: "James and I robbed the bank." whereas now they will say "James and myself did it."
Now maybe I just don't know what proper grammar is. I never really paid attention in school, so maybe I misunderstood what I thought I learned. I thought one should only use "myself" in cases where the subject is also the first person, such as: "I accidently shot myself while James and I were robbing the bank."
Today I read that "James is greater than myself." I really, really thought "I" was the appropriate pronoun in that case.
I also notice that public speakers use "myself" like it's going out of style. It would apparently kill them to say either "I" or "me", so it's always "James, Irene and myself" never "James and I" or "James and me".
Or am myself wrong about this over-usage of reflexive pronouns these days?
Scott wrote repeatedly about Prospera.
I visited it for 5 weeks, and I love the vision & the team behind it!
I decided to start a VC fund focused on startup cities, because I think we can build great startups there enabled by better regulations (e.g. peer country regulation, 3D on-chain property rights).
If you're an entrepreneur or innovator, I'd like to show it to you:
Prospera Healthtech Summit, September 23-25, 2022: https://infinitafund.com/healthtech2022
Prospera Edtech Summit, October 28-30, 2022: https://infinitafund.com/edtech2022
Prospera Fintech Summit, November 18-20, 2022: https://infinitafund.com/fintech2022
Interested to hear your thoughts… I’m going through a breakup. I initially thought this person was my soulmate. However, after a year of dating, we realized we were incompatible and we broke up. I went through the five stages of grief in less than a week, and I now feel completely fine, and if anything, relieved… People who’ve heard about the breakup have said that they’ve never seen someone handle a breakup so rationally. (I.e. since I know him and I aren’t compatible, I feel no need to be sad about the relationship ending.) That being said, why do you think I’m so indifferent to the break up? When most people make a logical decision, they still battle the emotional aspects of their decision.
Can anyone explain in technical terms why catecholamine testing isn't part of ADHD diagnosis? I read a study where they tested dopamine levels of kids before and after exercise and Richard Saul claims he diagnosed kids by having catecholamine tests done by mayo clinic so testing dopamine seems possible? I've asked a number of drs and psychiatrists and they all say haha that would be nice but it's impossible and I just want to understand why.
I've got a data-gathering question that will be simple for many of you. But I'm not too proud to ask it: I am gathering reports people make on sites that provide a covid PreP med called Evusheld. Will be compiling reports and making them available to later Evusheld-hunters, to help them find the best sites quickly.
I've made a form for people to give a report on. It comprises about 8 questions, followed by a spaces for people to write in answers. Some answers might be a paragraph long. Some items will not apply and people will leave them blank.
How can I collect these reports? There will be links on the site where people hunt for providers saying, "To leave a report on this provider click here." I could have the link go to a google drive document that's in suggestion mode, so that people's answers appear on the page as edits. The problem is that if 2 people try to report simultaneously, I don't know what will happen, but nothing good. Also, each new reporter needs a fresh document. They can't make a report on the one that somebody else just used
What's a simple way to do this? I do not need a system that compiles results for me some fancy way. This will work fine if I just end up with a bunch of separately filled out report forms. But how do I do that?
Please so not make any suggestions that involve hacking, writing code or anything on that level. I won't be able to implement them. I think I either need a clever idea about how to do this using google drive, which I'm comfortable with, or else an app or a site that will host the process and present a new report form to each new reporter.
I've written a blog post on when the COVID pandemic should be officially over: https://nsokolsky.substack.com/p/the-many-definitions-of-a-pandemic. Most interesting learnings were that the WHO changed their official "pandemic" definition in 2009 for the sake of H1N1 and that monkeypox is somehow not officially a "pandemic" yet. Feedback and more datapoints are welcome.
On the topic of human artists who are quickly becoming obsolete -- who are some living artists worth paying attention to?
Longtime rationalist/weirdo here! I write short fiction and post it on ashkie.com (and eventually I’ll stop being neurotic and get around to cross-posting it to my substack)
ashkie.com/nowhere-man is my fave story I’ve done recently — & the themes of agentiness, difficulty connecting with others, power of knowledge, etc make me think ACX readers would enjoy it! check it out, let me know what you think :)
Agree or disagree:
People used to be squeamish intensely judgmental about sex. We had the virgin/whore dichotomy. Active sexual desire was seen, in popular culture, as a kind of deviance to be judged.
Now we are squeamish about power. We have the victim/privilege dichotomy. Active desire for power is now seen in popular culture as a kind of deviance to be judged.
In what ways is this true? In what ways is this false? Am i wrong that there're an American tribal valence to this pair of dichotomies?
I just got a job at a small (~20 ppl) tech-startup! Check out airforestry.com if you're interested. Located in Uppsala, Sweden.
Does anyone have any tips or thoughts? Either about working at a startup in general, about their idea or that company in particular, anything is welcome :)
I will be working mostly in the workshop with physical prototypes, though I understand that in a small team everyone needs to do a bit of everything. I have a life-long burning interest in all things tech and have had several relevant jobs over the last decade, though never at a startup specifically. I don't have any formal education, but impressed them during the interview by showing some prototypes and projects I've done in the past. I start in a few weeks!
What makes GMU economics department so special? How did it happen that so many greats (Tyler Cowan, Robin Hanson, Bryan Caplan, maybe Garrett Jones) landed there?
Question for people with knowledge of nutrition.
Imagine you have these two alternatives: (A) eat half of a medium-size pizza, and (B) eat a whole medium-size pizza. Pizza is not a healthy eating choice, so both alternatives are bad, but obviously, B is worse than A. Also, for the sake of this question, let's say this is a one-off thing. Otherwise, you have pretty healthy eating habits.
I have the idea that option B is not twice as bad as option A, even though it involves twice as much pizza, but I don't have much theory or evidence that this is the case. Intuitively, if you eat half a pizza your body will process it normally, but if you eat a whole pizza your body will pass a lot of it relatively quickly. As a result, a whole pizza is worse than half a pizza, but not twice as bad. Is this roughly right?
(Note: I chose the half pizza/full pizza dilemma to keep things within reason. I imagine eating 10 pizzas is more than 10 times as bad as eating one pizza.)
Something that took me off guard in the last couple of years is the difference of my ability to remember things I've observed vs participated in. I used to listen to podcasts and be befuddled that I could remember details that the hosts themselves would forget or be hazy on from episodes that happened a year or so earlier. Similarly, as an aficionado of Diablo 2 speedruns, I'd be confused that streamers would forget that they had certain layouts in the map they were playing, or get details mixed up between different runs in the same day (D2 has randomised maps and items, so every run is different).
In the last year and a half I have begun doing both of these things myself (hosting and editing a podcast, performing D2 speedruns) and found that it's not that I had a better memory than the hosts or speedrunners, it's that my memory of events as an observer is far superior to my memory as a participant! This seems really counter intuitive to me, surely you should have better knowledge of what you yourself have done rather than what you've seen other people do.
So I just saw Garett Jones, a GMU economist, tweet this out- a proposed new system for selecting political candidates for office. The idea I guess came from Reddit, link below.
He proposes convening juries of randomly selected citizens to interview political candidates. It'd probably have to start at the very local level, so maybe 25-50 folks who are impaneled for a couple weeks for their local mayor, DA or sheriff's office. They review CVs, interview candidates, and have them do policy presentations, just like for a job. "Mr. Stewart and Ms. Jones, affordable housing is a major issue in our town. Please prepare a 20 minute presentation on your solutions, and then be prepared to answer questions afterwards." For transparency's sake all of this should probably be recorded/posted online.
You could have the jury then select the candidate, or you could have their selection go to a referendum, or at a minimum this process could be the new 'primary' with multiple winners and then the public could vote on 1 candidate from there. A new future jury could then examine their job performance at the end of their term and consider new candidates, etc. My guess is that having the jury select the winner would be fine for very local offices, and the higher up you go (state, federal office), it'd be more likely that this process serves as the primary instead. I would argue this process is about 100x more effective than current American political primaries.
Yes there are certainly some logistics to work out (it'd be tough for regular people to take such a large chunk of time off, though paying them could certainly help). I'd be curious to hear people's thoughts. Ideally the system could help avoid the demagoguery inherent in the current process. Being a legislator is a job, and it makes more sense to treat hiring them like a regular job interview than based on one's ability to give demagogic speeches or what have you.
Realistically, this would probably start at the hyperlocal level, maybe with District Attorney or sheriff elections (which really shouldn't be highly politicized/partisan). If it's proven to be effective, maybe it could work its way up the electoral ladder from there
https://old.reddit.com/r/EndFPTP/comments/wbesg1/sortition_election_hybrid_let_juries_elect/
Is an exercise like freewriting really tapping into a subconscious or unconscious mind or is it something else (the conscious mind with no editor perhaps)? Reference https://howaboutthis.substack.com/p/a-curious-realizer-guide-to-freewriting
This may be a dumb question, but in Singer's pond thought experiment, are there any arguments in favor of helping domestically versus internationally that aren't related to neglect or tractability?
https://www.thelifeyoucansave.org/common-objections-to-giving/
On this page from Singer's non-profit, their argument for foreign aid is that 95% of American charity is already domestic. But assuming that a dollar spent at home could save the same DALYs as another abroad, is there still a case to be made for giving locally?
Wild animal remediation-- giving legs to a snake.
https://www.youtube.com/watch?v=1SgGfMlbCoM
Very silly.
Scott mentions that the Pennsylvanian Quakers are the first historical culture to have recognisably modern moral norms, in regards to opposition to slavery/ treatment of women/child raising and other issues, in this post: https://slatestarcodex.com/2016/04/27/book-review-albions-seed/
I'd like to know if this is a fair characterisation of their culture and if they really were so different to neighbouring cultures/every other pre-modern society.
Also he seems to speculate that we inherited modern norms from the 17-18th century Quakers, does that suggest our current attitudes aren't the product of some kind of historical progress and are just as arbitrary as any of the other moral system that have existed?
It does seem like the Quakers had more historically novel views than modernity in some regards, like animal rights and pacifism, so maybe they were on a Whigish trajectory but were actually ahead od modern society in some ways.
I poked around in the dataset that Stable Diffusion is using as a corpus for training their model; LAION. It's basically constructed by scraping the internet, associating images with the text surrounding it, then providing a giant list or URL:s plus text snippets that anyone can download for themselves.
The first thing you notice by sampling random entries is how bad the text is. Picking something by random is likely to just be labelled "Funny images for laugh humorous HD wallpaper" than an actual description of the image content itself. I'm impressed that the model was able to suss out any content from this at all. The complicated queries people use to make the model create "good" images seems to be very much out-of-distribution!
There is also other AI-generated images in the corpus itself! Literally the first entry that contains the token decoded from "dragon" is this: https://live.staticflickr.com/65535/51363644206_c3e2880fa8_o.png - maybe this will lead to a future where the training corpus is "poisoned" by older models, leading to newer ones targeting the inferior output of older ones?
A solution for this that I can imagine is that people band together to curate content that they want generated; e.g. D&D people collecting all images of fantasy elves/dwarves/etc on the internet and writing detailed descriptions of them; which reasonably should improve images on that subject a great deal. You'd "just" have to create a website for it that people can use to collaborate and submit their work, and have the latest models regularly train themselves on on the output.
What do people think of this definition + claim:
“religions are attempts to define what 'good' means, and it's impossible for a person to live and function without some definition of 'good', either implicit or explicit.”
***Shameless self promotion***
I wrote about why, contrary to popular opinion, I think creative fields will lose jobs to AIs sooner and faster than other endeavors. You can read about it here: https://commentariat.substack.com/p/ai-is-coming-for-artists-first?sd=pf
If you like the piece I'd love if you would consider subscribing -- It's free, and I'm at least 75% confident the median ACX reader will enjoy my writing.
(P.S., last I checked self promotion was cool in these threads, Scott, sorry if it's not)
i've been using midjourney to design textile patterns, and as someone who works in fashion and knows a lot of great print designers, i'm *astounded* by the results (see https://twitter.com/itsjaneflowers/status/1553207619552157697 and https://twitter.com/itsjaneflowers/status/1554227557385617408 for a few samples)
i haven't seen anyone else using it for this purpose, which makes me wonder how much of a bubble we're still in with these tools⏤i.e. just based on midjourney's discord, it's still >99% nerds creating bland fantasy renders. so what happens when people with actual visual-cultural literacy get into the game?
Asterisk sounds interesting but not I'm willing to go through the bother of a mailing list- hopefully Scott will link to it again when there's actually something to read.
I tried to find out whether Asterisk has a Twitter account, or whether it's already being discussed on Twitter. I discovered that another Asterisk Magazine launched just last month. This is an unfortunate name clash :(
The other one is at https://www.asteriskmagazine.net/home
Does the name "Asterisk" mean anything special? Astral risk?
Disclaimer: please don’t turn this into a culture war thread. I’m asking this to try and find information, not get into a big debate over the nature/existence of gender. Thank you for your consideration.
I have been confused about the idea of “passing” vs many transgender people’s seeming openness about being trans. It strikes me that being open about being trans would be opposed to my understanding of what passing is.
Obviously, trans people are not a monolithic hive-mind, so maybe some people just care about one significantly more than the other. But is there some other interaction that I’m missing, or am I misunderstanding the meaning of passing, or is it some other thing? Thanks
If the bet is aesthetic in nature, Stable Diffusion should suffice, which is publicly released soon (weights available to researchers already). If not then you probably want Imagen
What is the best critique of Bayesian Reasoning? When should we not use it?