Astral Codex Ten Podcast

The Tourist Board of Xanadu Did recently impose a fee On those who travel far from home To visit Kubla's pleasure dome Of $20, 9 – 3 So twice five miles of fertile ground With fence and wire are girdled round And signs proclaiming "ENTRY AT THE GATE" Where gather many a camera-bearing crowd And here are docents, who in solemn state Explain the Mongol histories aloud But oh! That deep romantic chasm protracting Into a hill, athwart a cedarn cover A savage region, visitors attracting By actresses, forever reenacting A woman wailing to her demon-lover And from this chasm, with ceaseless turmoil spilling Crowds of old men in fat thick pants are milling And there, a fountain momently is forced: Amid whose swift half-intermitted burst Groups of eight to ten people, screaming ever White-water-raft upon the sacred river

[Previously in sequence: Fundamental Value Differences Are Not That Fundamental, The Whole City Is Center. This post might not make a lot of sense if you haven't read those first.] I. Thanks to everyone who commented on last week's posts. Some of the best comments seemed to converge on an idea like this: Confusing in that people who rely on lower-level features are placed higher, but the other way would have been confusing too. We need to navigate complicated philosophical questions in order to decide how to act, what to do, what behaviors to incentivize, what behaviors to punish, what signals to send, and even how to have a society at all. Sometimes we can use theories from science and mathematics to explicitly model how a system works and what we want from it. But even the scholars who understand these insights rarely know exactly how to objectively apply them in the real world. Yet anyone who lives with others needs to be able to do these things; not just scholars but ordinary people, children, and even chimpanzees. So sometimes we use heuristics and approximations. Evolution has given us some of them as instincts. Children learn others as practically-innate hyperpriors before they're old enough to think about what they're doing. And cultural evolution creates others alongside the institutitions that encourage and enforce them. In the simplest case, we just feel some kind of emotional attraction or aversion to something. In other cases, the emotions are so compelling that we crystallize them into a sort of metaphysical essence that explains them. And in the most complicated cases, we endorse the values implied by those metaphysical essences above and beyond whatever values we were trying to model in the first place. Some examples: People and animals need a diet with the right number of calories, the right macronutrient ratios, and the right vitamins and minerals. A few nutritional scientists know enough to figure out what's going on explicitly. Everyone else has evolved instincts that guide them through this process. Hunger and satiety are such instincts; when they're working well, they make sure someone eats as much as they need and no more. So are occasional cravings for some food with exactly the right nutrient – most common in high-nutrient-use states like pregnancy. But along with these innate heuristics, we have culturally determined ones. Everyone has a vague sense that potato chips are "unhealthy" and spinach is "healthy", though most people can't explain why. Instead of asking ordinary people and children to calculate their macronutrient and micronutrient profile, we ask them to eat "healthy" foods and avoid "unhealthy" foods. There's something sort of metaphysical about this – as if "health" were a magic essence that adheres to apples. And in fact, sometimes this goes wrong and people will do things like blend a thousand apples into some hyper-pure apple-elixir to get extra health-essence – but overall it mostly works. EXPLICIT MODEL: Trying to count how many calories and milligrams of each nutrient you get EMOTIONAL EXPERIENCE: Feeling hungry or full REIFIED ESSENCE: Some foods are inherently healthy or unhealthy ENDORSED VALUE: Insisting on only eating organic foods even when those foods have no quantifiable benefit over nonorganic Every society has some kind of punishment for people who don't follow their norms, whether it's ostracism or community service or beheading. There's a good consequentialist grounding for why this is necessary, with some of the most academic work being done in the field of prisoners' dilemmas and tit-for-tat strategies. But again, we don't expect ordinary people, children, and chimpanzees to absorb this work. The solution is the (innate? culturally learned? some combination of both?) idea of punishment. Punishment relies on a weird metaphysical essence of moral desert; people who do bad things deserve to suffer. The balance of the Universe is somehow off when a crime goes unavenged. Take this too far and you get the Erinyes and the idea that justice is the most important thing. There are references from ancient China to Hamlet that if you have something important you need to avenge, you need to do that now or you're a bad person. None of this follows from the game theory, but it's a really good way to enforce the game-theoretically correct action. EXPLICIT MODEL: Trying to figure out how to best deter antisocial behavior and optimize society EMOTIONAL EXPERIENCE: Feeling angry when someone wrongs you REIFIED ESSENCE: Justice: the world is out of balance when crimes go unavenged ENDORSED VALUE: Wrongdoers must suffer whether or not that prevents future crimes

Related to yesterday's post on people being too quick to assume value differences: some of the simplest fake value differences are where people make a big deal about routing around a certain word. And some of the most complicated real value differences are where some people follow a strategy explicitly and other people follow heuristics that approximate that strategy. There's a popular mental health mantra that "there's no such thing as laziness" (here are ten different articles with approximately that title: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10). They all make the same basically good point. We shame people who don't work very hard as "lazy", and think they should have lower status than the rest of us. But actually, these people don't just randomly choose not to work. Some of them have psychological issues, like anxiety and trauma that are constantly distracting them from their work, or a fear of success, or self-defeating beliefs about how nothing they do matters anyway. Others have biological issues – maybe hypothyroidism, or vitamin deficiencies, or ADHD, or other things we don't understand that lower their energy and motivation. Still others just don't want to do the specific thing we are asking them to do right now and can't force themselves to work uphill against that gradient. When we call people "lazy", we're ignorantly dismissing all these possibilities in favor of a moralistic judgment. A dialogue: Sophisticus: I don't believe in laziness. Simplicio: What about my cousin Larry? He keeps promising to do important errands for his friends and family, and then he never does them. Instead he just plays video games all the time. This has happened consistently over the past few years, every time he's promised to do something. One time my aunt asked him to go to the DMV to get some paperwork filled out, he promised he would do it, and then he kept putting it off for a month until it was past the deadline and she almost lost her car. He didn't forget about it or anything, he just couldn't bring himself to go out and do it. And he's been fired from his last three jobs for not showing up, and… Sophisticus: Yes, yes, I'm sure there are people like this. But he probably has some self-defeating beliefs, or vitamin deficiencies, or mental health issues. Simplicio: Okay. Well, my mother is going to be away for the next week, and she needs someone to dog-sit for her. Her dog is old and sick and requires a lot of care each day. She's terrified that if he doesn't get his food and medication and daily walk on time, something terrible will happen to him. She's willing to pay a lot of money. Do you think I should recommend she ask my cousin Larry? Sophisticus: No, of course not. Simplicio: Why not? Sophisticus: He probably won't do it. He'll just play video games instead. Simplicio: Why do you think so? Sophisticus: Because he has a long history of playing video games instead of doing important tasks. Simplicio: If only there were a word for the sort of person who does that! Sophisticus: Oh, I see. Now you're making fun of me. But I'm not saying everyone is equally reliable. I'm saying that instead of denouncing someone as "lazy", we should look for the cause and try to help them. Simplicio: Hey, we did try to help him. Larry's family has taken him to the doctor loads of times. They didn't anything on the lab tests, but the psychiatrist thought he might be ADHD and gave him some Adderall. I would say now he pulls through on like 20% of the things we ask him to do instead of zero percent. We also tried to get him to go to therapy, but the therapist deferred because ADHD has a very low therapy response rate. His parents tried to change the way they asked him to do things to make it easier for him, or to let him choose a different set of tasks that were more to his liking, but that only worked a little, if at all. Probably there's some cause we don't understand, but it's beyond the reach of medical science, incentive design, or the understanding that exists between loving family members to identify. Sophisticus: See! The Adderall helped! And letting him choose his own tasks helped a little too! Simplicio: I agree it helped a little. So should I recommend him to my mother as a dog-sitter? Sophisticus: No, of course not. Simplicio: Then I still don't see what the difference between us is. I agree it was worth having him go to the doctor and the therapist to rule out any obvious biological or psychological issues, and to test different ways of interacting with him in case our interaction style was making things worse. You agree that since this still hasn't made him reliably fulfill his responsibilities and we don't have any better ideas, he's a bad choice for a dog-sitter. Why can't I communicate the state of affairs we both agree on to my mother using the word "lazy"?

Ozy (and others) talk about fundamental value differences as a barrier to cooperation. On their model (as I understand it) there are at least two kinds of disagreement. In the first, people share values but disagree about facts. For example, you and I may both want to help the Third World. But you believe foreign aid helps the Third World, and I believe it props up corrupt governments and discourages economic self-sufficiency. We should remain allies while investigating the true effect of foreign aid, after which our disagreement will disappear. In the second, you and I have fundamentally different values. Perhaps you want to help the Third World, but I believe that a country should only look after its own citizens. In this case there's nothing to be done. You consider me a heartless monster who wants foreigners to starve, and I consider you a heartless monster who wants to steal from my neighbors to support random people halfway across the world. While we can agree not to have a civil war for pragmatic reasons, we shouldn't mince words and pretend not to be enemies. Ozy writes (liberally edited, read the original): From a conservative perspective, I am an incomprehensible moral mutant…however, from my perspective, conservatives are perfectly willing to sacrifice things that actually matter in the world– justice, equality, happiness, an end to suffering– in order to suck up to unjust authority or help the wealthy and undeserving or keep people from having sex lives they think are gross. There is, I feel, opportunity for compromise. An outright war would be unpleasant for everyone…And yet, fundamentally… it's not true that conservatives as a group are working for the same goals as I am but simply have different ideas of how to pursue it…my read of the psychological evidence is that, from my value system, about half the country is evil and it is in my self-interest to shame the expression of their values, indoctrinate their children, and work for a future where their values are no longer represented on this Earth. So it goes. And from the subreddit comment by GCUPokeItWithAStick: I do think that at a minimum, if you believe that one person's interests are intrinsically more important than another's (or as the more sophisticated versions play out, that ethics is agent-relative), then something has gone fundamentally wrong, and this, I think, is the core of the distinction between left and right. Being a rightist in this sense is totally indefensible, and a sign that yes, you should give up on attempting to ascertain any sort of moral truth, because you can't do it. I will give this position its due: I agree with the fact/value distinction. I agree it's conceptually very clear what we're doing when we try to convince someone with our same values of a factual truth, and confusing and maybe impossible to change someone's values.

Yesterday I wrote about melatonin, mentioning that most drugstore melatonin supplements were 10x or more the recommended dose. A commenter on Facebook pointed me to an interesting explanation of why. Dr. Richard Wurtman, an MIT scientist who helped discover melatonin's role in the body and pioneer its use as a sleep aid, writes: MIT was so excited about our research team's melatonin-sleep connection discovery that they decided to patent the use of reasonable doses of melatonin—up to 1 mg—for promoting sleep. But they made a big mistake. They assumed that the FDA would want to regulate the hormone and its use as a sleep therapy. They also thought the FDA wouldn't allow companies to sell melatonin in doses 3-times, 10-times, even 15-times more than what's necessary to promote sound sleep. Much to MIT's surprise, however, the FDA took a pass on melatonin. At that time, the FDA was focusing on other issues, like nicotine addiction, and they may have felt they had bigger fish to fry. Also, the FDA knew that the research on melatonin showed it to be non-toxic, even at extremely high doses, so they probably weren't too worried about how consumers might use it. In the end, and as a way of getting melatonin on to the market, the FDA chose to label it a dietary supplement, which does not require FDA regulation. Clearly, this was wrong because melatonin is a hormone, not a dietary supplement. Quickly, supplement manufacturers saw the huge potential in selling melatonin to promote good sleep. After all, millions of Americans struggled to get to sleep and stay asleep, and were desperate for safe alternatives to anti-anxiety medicines and sleeping pills that rarely worked well and came with plenty of side effects. Also, manufacturers must have realized that they could avoid paying royalties to MIT for melatonin doses over the 1 mg measure. So, they produced doses of 3 mg, 5 mg, 10 mg and more! Their thinking–like so much else in our American society–was likely, "bigger is better!" But, they couldn't be more wrong. So he's saying that…in order to get around a patent on using the correct dose of melatonin…supplement manufacturers…used the wrong dose of melatonin? I enjoy collecting stories of all the crazy perversities created by our current pharmaceutical system, but this one really takes the cake.

[I am not a sleep specialist. Please consult with one before making any drastic changes or trying to treat anything serious.] Van Geiklswijk et al describe supplemental melatonin as "a chronobiotic drug with hypnotic properties". Using it as a pure hypnotic – a sleeping pill – is like using an AK-47 as a club to bash your enemies' heads in. It might work, but you're failing to appreciate the full power and subtlety available to you. Melatonin is a neurohormone produced by the pineal gland. In a normal circadian cycle, it's lowest (undetectable, less than 1 pg/ml of blood) around the time you wake up, and stays low throughout the day. Around fifteen hours after waking, your melatonin suddenly shoots up to 10 pg/ml – a process called "dim light melatonin onset". For the next few hours, melatonin continues to increase, maybe as high as 60 or 70 pg/ml, making you sleepier and sleepier, and presumably at some point you go to bed. Melatonin peaks around 3 AM, then declines until it's undetectably low again around early morning. Is this what makes you sleepy? Yes and no. Sleepiness is a combination of the circadian cycle and the so-called "Process S". This is an unnecessarily sinister-sounding name for the fact that the longer you've been awake, the sleepier you'll be. It seems to be partly regulated by a molecule called adenosine. While you're awake, the body produces adenosine, which makes you tired; as you sleep, the body clears adenosine away, making you feel well-rested again. In healthy people these processes work together. Circadian rhythm tells you to feel sleepy at night and awake during the day. Process S tells you to feel awake when you've just risen from sleep (naturally the morning), and tired when you haven't slept in a long time (naturally the night). Both processes agree that you should feel awake during the day and tired at night, so you do. When these processes disagree for some reason – night shifts, jet lag, drugs, genetics, playing Civilization until 5 AM – the system fails. One process tells you to go to sleep, the other to wake up. You're never quite awake enough to feel energized, or quite tired enough to get restful sleep. You find yourself lying in bed tossing and turning, or waking up while it's still dark and not being able to get back to sleep. Melatonin works on both systems. It has a weak "hypnotic" effect on Process S, making you immediately sleepier when you take it. It also has a stronger "chronobiotic" effect on the circadian rhythm, shifting what time of day your body considers sleep to be a good idea. Effective use of melatonin comes from understanding both these effects and using each where appropriate.

The rationalist community started with the idea of rationality as a martial art – a set of skills you could train in and get better at. Later the metaphor switched to a craft. Art or craft, parts of it did get developed: I remain very impressed with Eliezer's work on how to change your mind and everything presaging Tetlock on prediction. But there's a widespread feeling in the rationalist community these days that this is the area where we've made the least progress. AI alignment has grown into a developing scientific field. Effective altruism is big, professionalized, and cash-rich. It's just the art of rationality itself that remains (outside the usual cognitive scientists who have nothing to do with us and are working on a slightly different project) a couple of people writing blog posts. Part of this is that the low-hanging fruit has been picked. But I think another part was a shift in emphasis. Martial arts does involve theory – for example, beginning fencers have to learn the classical parries – but it's a little bit of theory and a lot of practice. Most of becoming a good fencer involves either practicing the same lunge a thousand times in ideal conditions until you could do it in your sleep, or fighting people on the strip. I've been thinking about what role this blog plays in the rationalist project. One possible answer is "none" – I'm not enough of a mathematician to talk much about the decision theory and machine learning work that's really important, and I rarely touch upon the nuts and bolts of the epistemic rationality craft. I freely admit that (like many people) I tend to get distracted by the latest Outrageous Controversy, and so spend way too much time discussing things like Piketty's theory of inequality which get more attention from the chattering classes but are maybe less important to the very-long-run future of the world.

I'm late to posting this, but it's important enough to be worth sharing anyway: Sandberg, Drexler, and Ord on Dissolving the Fermi Paradox. (You may recognize these names: Toby Ord founded the effective altruism movement; Eric Drexler kindled interest in nanotechnology; Anders Sandberg helped pioneer the academic study of x-risk, and wrote what might be my favorite Unsong fanfic) The Fermi Paradox asks: given the immense number of stars in our galaxy, for even a very tiny chance of aliens per star shouldn't there should be thousands of nearby alien civilizations? But any alien civilization that arose millions of years ago would have had ample time to colonize the galaxy or do something equally dramatic that would leave no doubt as to its existence. So where are they? This is sometimes formalized as the Drake Equation: think up all the parameters you would need for an alien civilization to contact us, multiply our best estimates for all of them together, and see how many alien civilizations we predict. So for example if we think there's a 10% chance of each star having planets, a 10% chance of each planet being habitable to life, and a 10% chance of a life-habitable planet spawning an alien civilization by now, one in a thousand stars should have civilization. The actual Drake Equation is much more complicated, but most people agree that our best-guess values for most parameters suggest a vanishingly small chance of the empty galaxy we observe. SDO's contribution is to point out this is the wrong way to think about it. Sniffnoy's comment on the subreddithelped me understand exactly what was going on, which I think is something like this:

Chris Stucchio recommended Matt Rognlie's criticisms of Piketty (paper, summary, Voxsplainer). Rognlie starts by saying that Piketty didn't correctly account for capital depreciation (ie capital losing value over time) in his calculations. This surprises me, because Piketty says he does in his book (p. 55) but apparently there are technical details I don't understand. When you do that, the share of capital decreases, and it becomes clear that 100% of recent capital-share growth comes from one source: housing. I can't find anyone arguing that Rognlie is wrong. I do see many people arguing about the implications, all the way from "this disproves Piketty" to "this is just saying the same thing Piketty was". I think it's saying the same thing Piketty was in that housing is a real thing, and if there's inequality in housing, then that's real inequality. And landlords are a classic example of the rentiers Piketty is warning against. But it's saying a different thing in that most homeowners use their homes by living in them, not by renting them out. That means they're not part of Piketty's rentier class, and so using the amount of capital to represent the power of rentiers is misleading. Rentiers are not clearly increasing and there is no clear upward trend in rentier-vs-laborer inequality. I think this does disprove Piketty's most shocking thesis. Rognlie also makes an argument for why increasing the amount of capital will decrease the returns on capital, leading to stable or decreasing income from capital. Piketty argues against this on page 277 of his book, but re-reading it Piketty's argument now looks kind of weak, especially with the evidence from housing affecting some of his key points. Grendel Khan highlights the role of housing with an interesting metaphor: Did someone say housing? As an illustration, the median homeowner in about half of the largest metros made more off the appreciation of their home than a full-time minimum-wage job. It's worst in California, of course; in San Jose, the median homeowner made just shy of $100 per working hour. See also Richard Florida's commentary. See also everything about how the housing crisis plays out in micro; it is precisely rentier capitalism. In the original post, I questioned Piketty's claim that rich people and very-well-endowed colleges got higher rates of return on their investment than ordinary people or less-well-endowed colleges. After all, why can't poorer people pool their money together, mutual-fund-style, to become an effective rich person who can get higher rate of return? Many people tried to answer this, not always successfully. brberg points out that Bill Gates – one example of a rich person who's gotten 10%+ returns per year – has a very specific advantage: Not sure about Harvard's endowment, but it's worth noting that the reason Gates, Bezos, Zuckerberg, and other self-made billionaires have seen their fortunes grow so quickly is that each of them has the vast majority of their wealth invested in a single high-growth company. This is an extremely high-risk investment strategy that has the potential to pay off fantastically well in a tiny percentage of cases, but it's not really dependent on the size of the starting stake. Anyone who invested in Microsoft's IPO would have seen the same rate of return as Gates. This is a good point, but most of Piketty's data focuses on college endowments. How do they do it? Briefling writes: I'm not sure you can take the wealth management thing at face value. The stock market since 1980 has 10% annualized returns. Instead of trying to replicate whatever Harvard and Yale are doing, why don't you just put your money in the stock market? Also a good point, but colleges seem to do this with less volatility than the stock market, which still requires some explanation.

[Original review is here. Don't worry, people who had interesting comments on the review – I'll try to get a comments highlights thread up eventually.] For Ricardo, who published his Principles of Political Economy and Taxation in 1817, the chief concern was the long-term evolution of land prices and land rents. Like Malthus, he had virtually no genuine statistics at his disposal. He nevertheless had intimate knowledge of the capitalism of his time. Born into a family of Jewish financiers with Portuguese roots, he also seems to have had fewer political prejudices than Malthus, Young, or Smith. He was influenced by the Malthusian model but pushed the argument farther. He was above all interested in the following logical paradox. Once both population and output begin to grow steadily, land tends to become increasingly scarce relative to other goods. The law of supply and demand then implies that the price of land will rise continuously, as will the rents paid to landlords. The landlords will therefore claim a growing share of national income, as the share available to the rest of the population decreases, thus upsetting the social equilibrium. For Ricardo, the only logically and politically acceptable answer was to impose a steadily increasing tax on land rents. This somber prediction proved wrong: land rents did remain high for an extended period, but in the end the value of farm land inexorably declined relative to other forms of wealth as the share of agriculture in national income decreased. Writing in the 1810s, Ricardo had no way of anticipating the importance of technological progress or industrial growth in the years ahead. Like Malthus and Young, he could not imagine that humankind would ever be totally freed from the alimentary imperative. One underappreciated feature of Piketty is his engaging presentation of economic history. A constant feature of the theorists he discusses is that they are all brilliant thinkers, they all follow the trends of their time to their obvious conclusions in ways deeper and more insightful than their contemporaries – and they all miss complicated paradigm shifts that make the trends obsolete and totally ruin their theories. Rationalists take note. Like Ricardo, Marx based his work on an analysis of the internal logical contradictions of the capitalist system. He therefore sought to distinguish himself from both bourgeois economists (who saw the market as a self-regulated system, that is, a system capable of achieving equilibrium on its own without major deviations, in accordance with Adam Smith's image of "the invisible hand" and Jean-Baptiste Say's "law" that production creates its own demand), and utopian socialists and Proudhonians, who in Marx's view were content to denounce the misery of the working class without proposing a truly scientific analysis of the economic processes responsible for it.7 In short, Marx took the Ricardian model of the price of capital and the principle of scarcity as the basis of a more thorough analysis of the dynamics of capitalism in a world where capital was primarily industrial (machinery, plants, etc.) rather than landed property, so that in principle there was no limit to the amount of capital that could be accumulated. In fact, his principal conclusion was what one might call the "principle of infinite accumulation," that is, the inexorable tendency for capital to accumulate and become concentrated in ever fewer hands, with no natural limit to the process. This is the basis of Marx's prediction of an apocalyptic end to capitalism: either the rate of return on capital would steadily diminish (thereby killing the engine of accumulation and leading to violent conflict among capitalists), or capital's share of national income would increase indefinitely (which sooner or later would unite the workers in revolt). In either case, no stable socioeconomic or political equilibrium was possible.