Astral Codex Ten Podcast

Plus FTX charges, scandal markets - and oh yeah, wasn't there some kind of midterm recently? https://astralcodexten.substack.com/p/mantic-monday-twitter-chaos-edition Twitter! This is all going to be so, so obsolete by the time I finish writing it and hit the “send post” button. But here goes: 395 traders on this, so one of Manifold’s biggest markets, probably representative. The small print defines a major outage as one that lasts more than an hour. See here for a good explanation of why some people expect Twitter outages. https://www.metaculus.com/questions/embed/13499/Polymarket is within 2% of Manifold. Metaculus here has slightly stricter criteria but broadly agrees. 71 traders, still pretty good, but I find it meaningless without a way to distinguish between “everything collapses, Elon sells it for peanuts to scavengers” vs. “Elon saves Twitter, then hands it over to a minion while he moves on to a company building giant death zeppelins”. Oh, here we go. 20 traders, they think Musk will stay in charge. 23 traders. Twitter was profitable in 2018 and 2019, then went back to being net negative in 2020 and 2021 (I don’t know why) . I don’t think it’s been very profitable lately, so it would be a feather in Musk’s cap if he accomplished this. 24 traders. Twitter’s mDAU have consistently gone up in the past. DAU is slightly different and I think more likely to include bots. 26 traders. One thing I like about Manifold is that it lets you choose any point along the gradient from “completely objective” (eg Twitter’s reported DAU count) to “completely subject” (eg whether the person who made the market thinks something is better or worse). This at least uses a poll as its resolution method. But the poll will be in the comments of this market, which means it will mostly be by people who invested in this market, who’ll have strong incentives to manipulate it. Maybe Manifold should add a polling platform to their service?

https://astralcodexten.substack.com/p/the-psychopharmacology-of-the-ftx Must not blog about FTX . . . must not blog about . . . ah, $#@% it         Tyler Cowen linked Milky Eggs’ excellent overview of the FTX crash. I’m unqualified to comment on any of the financial or regulatory aspects. But it turns out there’s a psychopharmacology angle, which I am qualified to talk about, so let’s go. I wrote this pretty rushed because it’s an evolving news story. Sorry if it’s less polished than usual.1 1: Was SBF Using A Medication That Can Cause Overspending And Compulsive Gambling As A Side Effect? Probably yes, and maybe it could have had some small effect, but probably not as much as the people discussing it on Twitter think. Milky Eggs reports a claim by an employee that Sam was on “a patch for designer stimulants that mainlined them into his blood to give him a constant buzz at all times”. This could be a hyperbolic description of Emsam, a patch form of the antidepressant/antiparkinsonian agent selegiline. The detectives at the @AutismCapital Twitter account found a photo of SBF, zoomed in on a scrap of paper on his desk, and recognized it as an Emsam wrapper.

https://astralcodexten.substack.com/p/contra-resident-contrarian-on-unfalsifiable I. Contra Resident Contrarian . . . Resident Contrarian writes On Unfalsifiable Internal States, where he defends his skepticism of jhana and other widely-claimed hard-to-falsify internal states. It’s long, but I’ll quote a part that seemed especially important to me: I don’t really want to do the part of this article that’s about how it’s reasonable to doubt people in some contexts. But to get to the part I want to talk about, I sort of have to. There is a thriving community of people pretending to have a bunch of multiple personalities on TikTok. They are (they say) composed of many quirky little somebodies, complete with different fun backstories. They get millions of views talking about how great life is when lived as multiples, and yet almost everyone who encounters these videos in the wild goes “What the hell is this? Who pretends about this kind of stuff?” There’s an internet community of people, mostly young women, who pretend to be sick. They call themselves Spoonies; it’s a name derived from the idea that physically and mentally well people have unlimited “spoons”, or mental/physical resources they use to deal with their day. Spoonies are claiming to have fewer spoons, but also en masse have undiagnosable illnesses. They trade tips on how to force their doctors to give them diagnoses: > In a TikTok video, a woman with over 30,000 followers offers advice on how to lie to your doctor. “If you have learned to eat salt and follow internet instructions and buy compression socks and squeeze your thighs before you stand up to not faint…and you would faint without those things, go into that appointment and tell them you faint.” Translation: You know your body best. And if twisting the facts (like saying you faint when you don’t) will get you what you want (a diagnosis, meds), then go for it. One commenter added, “I tell docs I'm adopted. They'll order every test under the sun”—because adoption means there may be no family history to help with diagnoses. And doctors note being able to sort of track when particular versions of illnesses get flavor-of-the-week status: > Over the pandemic, neurologists across the globe noticed a sharp uptick in teen girls with tics, according to a report in the Wall Street Journal. Many at one clinic in Chicago were exhibiting the same tic: uncontrollably blurting out the word “beans.” It turned out the teens were taking after a popular British TikToker with over 15 million followers. The neurologist who discovered the “beans” thread, Dr. Caroline Olvera at Rush University Medical Center, declined to speak with me—because of “the negativity that can come from the TikTok community,” according to a university spokesperson.   Almost no one who encounters them assumes they are actually sick. Are there individuals in each of these communities that are “for real”? Probably, especially in the case of the Spoonies; undiagnosed or undiagnosable illnesses are a real thing. Are most of them legitimate? The answer seems to be a pretty clear “no”. I’m not bringing them up to bully them; I suspect that there are profiteers and villains in both communities, but there’s also going to be a lot of people driven to it as a form of coping with something else, like how we used to regard cutting and similar forms of self-harm. And, you know, a spectrum of people in between those two poles, like you’d expect with nearly anything. But it’s relevant to bring up because there seem to be far more Spoonies and DID TikTok-fad folks than people who say they orgasm looking at blankets because they did some hard thinking (or non-thinking) earlier. So when Scott says something that boils down to “this is credible, because a lot of people say they experience this”, I have to mention that there’s groups that say they experience a lot of stuff in just the same way that basically nobody believes is experiencing anything close to what they say they are. Granting that this is not the part of the article RC wants to write, he starts by bringing up “spoonies” and people with multiple personalities as people who it’s reasonable to doubt. I want to go over both cases before responding to the broader point. II. . . . On Spoonies “Spoonies” are people with unexplained medical symptoms. RC says he thinks a few may be for real, but most aren’t. I have the opposite impression. Certainly RC’s examples don’t prove what he thinks they prove. He brings up one TikToker’s advice: In a TikTok video, a woman with over 30,000 followers offers advice on how to lie to your doctor. “If you have learned to eat salt and follow internet instructions and buy compression socks and squeeze your thighs before you stand up to not faint…and you would faint without those things, go into that appointment and tell them you faint.” Translation: You know your body best. And if twisting the facts (like saying you faint when you don’t) will get you what you want (a diagnosis, meds), then go for it. One commenter added, “I tell docs I'm adopted. They'll order every test under the sun”—because adoption means there may be no family history to help with diagnoses. This person is using a deliberately eye-catching title (Lies To Tell Your Doctor) to get clicks. But if you read what they’re saying, it’s reasonable and honest! They’re saying “If you used to faint all the time, and then after making a bunch of difficult lifestyle changes you can now mostly avoid fainting, and your doctor asks ‘do you have a fainting problem yes/no’, answer yes!” THIS IS GOOD ADVICE. Imagine that one day you wake up and suddenly you have terrible leg pain whenever you walk. So you mostly don’t walk anywhere. Or if you do have to walk, you use crutches and go very slowly, because then it doesn’t hurt. And given all of this, you don’t experience leg pain. If you tell your doctor “I have leg pain”, are you lying ? You might think this weird situation would never come up - surely the patient would just explain the whole situation clearly? One reason it might come up is that all this is being done on a form - “check the appropriate box, do you faint yes/no?”. Another reason it might come up is that a nurse or someone takes your history and they check off boxes on a form. Another reason it might come up is that everything about medical communication is inexplicably terrible; this is why you spend umptillion hours in med school learning “history taking” instead of just saying “please tell me all relevant information, one rational human being to another”.

https://astralcodexten.substack.com/p/can-people-be-honestly-wrong-about A tangent of the jhana discussion: I asserted that people can’t be wrong about their own experience. That is, if someone says they don’t feel hungry, maybe they’re telling the truth, and they don’t feel hungry. Or maybe they’re lying: saying they don’t feel hungry even though they know they really do (eg they’re fasting, and they want to impress their friends with how easy it is for them). But there isn’t some third option, where they honestly think they’re not experiencing hunger, but really they are. Commenters brought up some objections: aren’t there people who honestly say they don’t feel hungry, but then if you give them food, they’ll wolf it down and say “Man, that really hit the spot, I guess I didn’t realize how hungry I was”? Yes, this sometimes happens. But I don’t think of it as lying about internal experience. I think of it as: their stomach is empty, they have low blood sugar, they have various other physiological correlates of needing food - but for some reason they’re not consciously experiencing the qualia of hunger. Their body is hungry, but their conscious mind isn’t. They say they don’t feel hungry, and their description of their own feeling is accurate. This is also how I interpret people who say “I’m not still angry about my father”, but then every time you mention their father they storm off and won’t talk to you for the rest of the day. Clearly they still have some trauma about their father that they have to deal with. But it doesn’t manifest itself as a conscious feeling of anger. This person could accurately be described as “they don’t feel conscious anger about their father, but mentioning their father can trigger stress-related behaviors”. Linch gives an especially difficult example: I think it's possible for people to fool themselves about internal states. My favorite example is time perception. You can meditate or take drugs in ways that make you think that your clock speed has gone up and your subjective experience of your subjective experience of time is slowed down. But your actual subjective experience of time isn't much faster clock speeds (as could be evidenced by trying to do difficult computational tasks in those stats). But I think this can be defeated by the same maneuver. Just as you can be right about feeling like you’re not hungry, when in fact your body needs food, so you can be right about it feeling like time moves slowly for you, when in fact it’s moving at a normal rate.

https://astralcodexten.substack.com/p/highlights-from-the-comments-on-brain [original post here] On What Kind Of Thing Brain Waves Are: Loweren writes: In my undergrad biology program we visited a brain research lab near Moscow. The brain scientist gave us a brief intro to Fourier transforms, which made me understand how beautiful they are - something that 2 years of undergrad math classes didn't manage to do. Then he explained the brain waves to us like this: "Imagine you are standing outside the football stadium. You don't see what's happening inside, but you hear the chatter of the crowd. All the individual words blend together into indistinct mess and although there's definitely a local information transfer going on, from the outside you can't make out anything specific. Then imagine one of the teams scored a goal. The crowd behavior is now very different! The fans of the winning team start to cheer and sing. You can easily pick this up from outside and infer what's happening. This is because the individuals behave in a globally coordinated manner, so their signals amplify each other in tune. From this perspective, brain waves are a byproduct of globally coordinated neuronal activity, and it's the first one we historically learned to pick up. They appear when neurons stop chatting with each other and start chanting in unison." Then he plopped some probes on my head and announced I have beautiful epileptic spikes (I'm not an epileptic)

https://astralcodexten.substack.com/p/highlights-from-the-comments-on-my [Original post here] I know I don’t usually publish on Saturdays, but I wanted to get this out before people filled in their mail-in ballots. So: Is Prop 31 Another Attack On Vaping?   Maximum Limelihood Estimator is concerned that Prop 31 (against flavored tobacco products) is meant to target vaping: The flavored tobacco ban is mostly a ban on vaping; the vast majority of vape products are flavored, while most cigarettes aren't. About 40% of cigarettes are flavored, compared to about 85% of vape juice. A study suggests that a ban on flavored tobacco would increase cigarette consumption (by making cigarettes relatively more desirable than vaping). Limelihood writes that “The statistics [in the study] are great, which is honestly shocking to me, since it's the first time I've said this about an experiment in . . . ever.” There is also a study purporting to show that flavored cigarette bans do decrease smoking, but Limelihood says that: …it's got some big problems. The study there only compares tobacco sales in a single city (San Francisco) before and after a ban on menthol cigarettes. However, because there's no comparison to other cities, it's essentially worthless; tobacco sales throughout the US dropped at this time, and I don't know how this compares.

https://astralcodexten.substack.com/p/acx-grants-project-updates         Thanks to everyone who got ACX Grants (see original grants here) and sent me a one-year update. Below are short summaries of the updates everyone sent. If for some reason you want one of the full updates, which are longer and more technical, let me know and I‘ll see if I have permission to send them to you. I’ve also included each grantee’s assessment on a scale of 1-10 for how well they’re doing, where 5/10 is “about as well as expected”. A few grantees are asking for extra help - I’ve included those requests in italics at the end of the relevant updates, and I’ve collected all of them together below. Updates   1: Discover Molecular Targets Of Antibiotics (8/10)Pedro Silva planned to use in silico screening to identify the biochemical targets of seven promising natural antibiotics, which could potentially help develop better versions of them. He says he's finished most of the simulations and determined the 5-20 most stable complexes for each antibiotic. Once he finishes this, he can start additional simulations on the best complexes to obtain better estimates of their stability and construct hypotheses on which of these is most involved in the antibiotic's efficacy. 2: Ballot Proposition For Approval Voting In Seattle (?/10)They have asked me not to discuss their progress until after the November election. 3: Software To Validate New FDA Drug Trial Designs (10/10)Michael Sklar and Confirm Solutions have gotten further funding from FTX and now have 2-3 people working full-time on the project. They are building new statistical techniques and software to help regulators quickly assess designs for clinical trials. Here is a recent conference poster on the methods. They have written proof-of-concept code and are writing a white paper to show regulators and pharma companies.  They also claim to have developed software that has "sped up their simulations for some standard Bayesian trial designs by a factor of about 1 million." They are looking for more employees and collaborators; if you’re interested, contact research@confirmsol.org 4: Alice Evans’ Research On “The Great Gender Divergence” (?/10)Dr. Evans has done over four months of research in Morocco, Italy, India, and Turkey. You can find some of her most recent thoughts at her blog here. Her book is still on track to be published from Princeton Press, more details tbd. 5: Develop Safer Immunosuppressants (7/10)Trevor Klee planned to continue his work to develop a safer slow-release form of cyclosporine. He realized this would be too expensive to do in humans in the current funding environment, and has pivoted to getting his medication approved for a feline autoimmune disease as both a proof-of-concept and as a cheaper, faster way to start making revenue. He recently raised $100,000 in crowdfunding (in addition to getting $200,000 from angel investors to run a feline trial, which will finish in January. He still anticipates eventually moving back to humans. Trevor wants to talk to bloggers or writers who might be interested in covering his work. 6: Promote Economically Literate Climate Policy In US States (4/10)Yoram Bauman and Climate 24x7 have written a policy paper about their ideas. They were able to get a bill in front of the Nebraska Legislature, but it died in committee. They have a promising measure in Utah, and an off chance of getting something rolling in Pennsylvania. Overall they report frustration, as many of the legislators they worked with have been voted out or term-limited. If you are a legislator or activist interested in helping with this project - especially in Utah, Pennsylvania, or South Dakota - please contact Yoram at yoram@standupeconomist.com. 7: Repository / Search Engine For Forecasting Questions (8/10)Nuno Sempere at metaforecast.org was able to hire a developer to “make the backend significantly better and add a bunch of functionality” - you can see a longer list of updates here. The developer has since left for other forecasting-related work and the project is moving more slowly. 8: Help [Anonymous] Interview For A Professorship (8/10)[Anonymous] was a grad student who wanted to interview for professorships at top schools where he might work on AI safety in an academic environment. The grant was to help make it financially easier for him to go on a long round of interviews [Anonymous] successfully got a job offer from a top school, and will be going there and researching AI safety.

https://astralcodexten.substack.com/p/my-california-ballot-2022 Previously: 2018, 2020 General Philosophy Of Voting   This is California, so the Democrats always win. When I vote, I mean to send a signal somewhere in between “you are the candidate I really prefer for this office” and “I will vote for the Democrat if I approve of her and want her to have a mandate; otherwise I will vote for the Republican as a protest”. I try to have a weak bias towards voting “NO” on state constitutional amendments, because unless there’s a compelling reason otherwise I would rather legislators be able to react to events than have things hard-coded for all time. I lean liberal-to-libertarian; the further you are from that, the less useful you’ll find my opinions. State Propositions   Proposition 1: Constitutional Amendment Enshrining Right To Abortion California will never decide to ban abortion. If the federal government decides to ban abortion, California’s state constitution won’t matter. So you would think that having a right to abortion in the Constitution is a purely symbolic matter. The people arguing for the proposition don’t address this concern. The people arguing against the proposition claim that this is a Trojan Horse intended to sneak in support for using taxpayer funding for late-term and partial-birth abortions, which California doesn’t currently do. Is this true? It’s true that California currently doesn’t allow abortions past 24 weeks. It’s true that the exact text of the proposed amendment is: The state shall not deny or interfere with an individual’s reproductive freedom in their most intimate decisions, which includes their fundamental right to choose to have an abortion and their fundamental right to choose or refuse contraceptives. This section is intended to further the constitutional right to privacy guaranteed by Section 1, and the constitutional right to not be denied equal protection guaranteed by Section 7. Nothing herein narrows or limits the right to privacy or equal protection …which sure doesn’t sound like it’s saying the state can continue to ban abortion after 24 weeks. But this article quotes law professors who reassure us that courts would totally understand that this amendment has to be interpreted in the context in which it was written - ie a state which supports a 24-week abortion ban - so no court would ever interpret it as making 24-week abortion bans unconstitutional. So apparently our defense against this is . . . that all California judges will be die-hard originalists completely immune to the temptation of judicial activism even when the text is begging them to do it. A friend brings up that late-term partial-birth abortions happen more often in Republicans’ imaginations than in real life. When they do happen in real life, it’s usually for sympathetic medical reasons. I interpret this as a purely symbolic measure that has no real benefits, probably also has no real risks, but writes a poorly-worded thing whose explicit text nobody wants into the state constitution. I vote NO. Proposition 26: Legalize In Person Sports Gambling At Racetracks And Indian Casinos Allows four racetracks in the state to offer in person sports betting, and tribal casinos to allow “sports betting, roulette, and games played with dice”. California is truly the dumbest state. I believe this for many reasons, but my reason for believing it today is that apparently the law allows tribal casinos to offer slot machines, but not roulette or dice games. Nobody comes out and says exactly why, but I think it’s because of this paragraph in the California constitution, from 1872 Every person who deals, plays, or carries on, opens, or causes to be opened, or who conducts, either as owner or employee, whether for hire or not, any game of faro, monte, roulette, lansquenet, rouge et noire, rondo, tan, fan-tan, seven-and-a-half, twenty-one, hokey-pokey, or any banking or percentage game played with cards, dice, or any device, for money, checks, credit, or other representative of value, and every person who plays or bets at or against any of those prohibited games, is guilty of a misdemeanor, and shall be punishable by a fine not less than one hundred dollars ($100) nor more than one thousand dollars ($1,000), or by imprisonment in the county jail not exceeding six months, or by both the fine and imprisonment. Since roulette existed in 1872 but slot machines didn’t, the Constitution banned roulette but not slot machines, and that rule has continued to the present day. Now if slot-machine-filled casinos want to also have roulette, they need a Constitutional amendment. DID I MENTION THAT I WISH PEOPLE WOULD STOP ADDING EVERY LAW THAT THEY LIKE TO THE CONSTITUTION? But this law also allows random people to sue “card clubs”, ie small-scale private gambling establishments. We originally thought this was a Texas-style “bounty” law that gave the random people part of the winnings, but it seems this isn’t true. I’m not sure if the idea is that legal gambling establishments would fund these lawsuits, or if they just expect private citizens to do this out of the love of suing people. Although I think the first prong of the law - allowing roulette and sports betting at casinos - makes sense, the second prong seems to be casinos making it easier to shut down their competitors. These competitors are probably ordinary people who want to gamble in a backroom somewhere without hurting anyone else. And the argument against on the ballot is by the Black Chamber of Commerce, saying that these card clubs are a useful source of revenue for poor minority communities. I don’t want to help giant casinos put a bounty on their heads. I vote NO.

https://astralcodexten.substack.com/p/moderation-is-different-from-censorship This is a point I keep seeing people miss in the debate about social media. Moderation is the normal business activity of ensuring that your customers like using your product. If a customer doesn’t want to receive harassing messages, or to be exposed to disinformation, then a business can provide them the service of a harassment-and-disinformation-free platform. Censorship is the abnormal activity ensuring that people in power approve of the information on your platform, regardless of what your customers want. If the sender wants to send a message and the receiver wants to receive it, but some third party bans the exchange of information, that’s censorship. The racket works by pretending these are the same imperative. “Well, lots of people will be unhappy if they see offensive content, so in order to keep the platform safe for those people, we’ve got to remove it for everybody.” This is not true at all. A minimum viable product for moderation without censorship is for a platform to do exactly the same thing they’re doing now - remove all the same posts, ban all the same accounts - but have an opt-in setting, “see banned posts”. If you personally choose to see harassing and offensive content, you can toggle that setting, and everything bad will reappear. To “ban” an account would mean to prevent the half (or 75%, or 99%) of people who haven’t toggled that setting from seeing it. The people who elected to see banned posts could see them the same as always. Two “banned” accounts could still talk to each other, retweet each other, etc - as could accounts that hadn’t been banned, but had opted into the “see banned posts” setting. Does this difference seem kind of pointless and trivial? Then imagine applying it to China. If the Chinese government couldn’t censor - only moderate - the world would look completely different. Any Chinese person could get accurate information on Xinjiang, Tiananmen Square, the Shanghai lockdowns, or the top fifty criticisms of Xi Jinping - just by clicking a button on their Weibo profile. Given how much trouble ordinary Chinese people go through to get around censors, probably many of them would click the button, and then they’d have a free information environment. This switch might seem trivial in a well-functioning information ecology, but it prevents the worst abuses, and places a floor on how bad things can get.

https://astralcodexten.substack.com/p/highlights-from-the-comments-on-jhanas "I think it’s the first time half the commenters accused the other half of lying" I. Is Jhana Real?   This was a fun one. I think it’s the first time half the commenters accused the other half of lying. Okay, “half” is an exaggeration. But by my count we had 21 people who claimed to have experienced jhanas (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21), and 7 who said they were pretty sure it wasn’t real as described (1, 2, 3, 4, 5, 6, 7). The former group include people like Tetris McKenna, who wrote: I've experienced samatha jhanas. I don't do it so much now. The first few times you get on the edge of 1st jhana, it's difficult to achieve, because you see the wave of pleasure approaching, and grasp for it, and that grasping takes you away from it. So it's a careful balancing act of pleasure/desire in the first place to get there, which you have to master to some degree. To even get to 1st jhana, you have to internally figure out some stuff about the craving/pleasure dynamic on a subconscious, mechanical level. 1st jhana is, as the author describes, intensely pleasurable. Sublime. His descriptions are spot on imo. But in some ways, it's also too much pleasure. It can feel agitating once you get used to it and aren't so awestruck by it anymore. Indeed, the latter jhanas are associated with letting go of certain aspects of the initial jhana, to more and more refined states that are more calm and equanimous than intensely pleasurable. Again, this is internalising and mastering the skill of balancing pleasure/craving. Those calm and equanimous states of 2nd-4th jhana become much more satisfying than the initial pleasure wave of the 1st jhana. Cultivating them to that degree is a process of gaining valuable insight into the pleasure/craving dynamic in your mind. Even if you don't get to those stages, just practising 1st jhana alone will help the mind normalise the intensity of the pleasure, such that it's no big deal any more. You don't need or even want pleasure all the time, because you've seen it with such clarity, over and over again, just by setting the conditions up correctly in your mind.