From: Zack M. Davis Date: Tue, 19 Nov 2024 06:47:51 +0000 (-0800) Subject: finish drafting "Prediction Markets Are Not a Drop-In" X-Git-Url: http://unremediatedgender.space/source?a=commitdiff_plain;ds=sidebyside;p=Ultimately_Untrue_Thought.git finish drafting "Prediction Markets Are Not a Drop-In" --- diff --git a/content/drafts/prediction-markets-are-not-a-drop-in-replacement-for-concepts.md b/content/drafts/prediction-markets-are-not-a-drop-in-replacement-for-concepts.md index 94fa39c..cb654e0 100644 --- a/content/drafts/prediction-markets-are-not-a-drop-in-replacement-for-concepts.md +++ b/content/drafts/prediction-markets-are-not-a-drop-in-replacement-for-concepts.md @@ -1,10 +1,10 @@ Title: Prediction Markets Are Not a Drop-In Replacement for Concepts -Date: 2024-10-14 20:00 +Date: 2024-11-20 20:00 Category: commentary -Tags: Eliezer Yudkowsky, literary critcism, worldbuilding, prediction markets +Tags: Eliezer Yudkowsky, literary criticism, worldbuilding, prediction markets Status: draft -In ["Comment on a Scene from _Planecrash_: 'Crisis of Faith'"](http://unremediatedgender.space/2022/Jun/comment-on-a-scene-from-planecrash-crisis-of-faith/), I critiqued a scene in which Keltham (a [magical universe-teleportation victim](https://en.wikipedia.org/wiki/Isekai) from alternate Earth called dath ilan) plays dumb about why a pre-industrial Society would only choose males for military conscription. +In ["Comment on a Scene from _Planecrash_: 'Crisis of Faith'"](http://unremediatedgender.space/2022/Jun/comment-on-a-scene-from-planecrash-crisis-of-faith/), I criticized a scene in which Keltham (a [magical universe-teleportation victim](https://en.wikipedia.org/wiki/Isekai) from alternate Earth called dath ilan) plays dumb about why a pre-industrial Society would only choose males for military conscription. _Planecrash_ coauthor [Eliezer Yudkowsky comments](http://unremediatedgender.space/2022/Jun/comment-on-a-scene-from-planecrash-crisis-of-faith/#isso-583): @@ -55,7 +55,7 @@ def prediction_market_sort(my_list): next_comparison_markets = { (i, j): prediction_markets.create( f"Will this list be sorted with no more than {op_budget} comparisons, if the" - f"next comparision is between indicies {i} and {j}?", + f"next comparison is between indices {i} and {j}?", static_data=my_list, ) for i in range(n) @@ -124,43 +124,44 @@ Barbara objects. "Stop! What are you doing?" I wrote the restaurant-choice scenario as an illustrative example for this blog post, but I want you to imagine how you would react if someone _actually behaved that way in real life_: stopped you when you searched for Italian restaurants and insisted you start a prediction market instead. That would be pretty weird, right? -It's not that there's anything particularly wrong with the idea of using prediction markets to get restaurant suggestions. I can easily believe that you might get some good suggestions that way, even in a sparsely-traded play-money market on Earth. (In a similar vein, Yudkowsky has [a "What books will I enjoy reading?" market](https://manifold.markets/EliezerYudkowsky/what-book-will-i-enjoy-reading).) +It's not that there's anything particularly wrong with the idea of using prediction markets to get restaurant suggestions. I can easily believe that you might get some good suggestions that way, even in a sparsely-traded play-money market on Earth, _i.e._, the real world. (In a similar vein, Yudkowsky has [a "What books will I enjoy reading?" market](https://manifold.markets/EliezerYudkowsky/what-book-will-i-enjoy-reading).) -The weird part is the implication that the form of reasoning you would use to make a decision in the absence of a prediction market can be dismissed as "a way based not on predicted outcomes" and regarded as obviated by the existence of the market. I don't think anyone really believes this, as contrasted to [believing they believe](https://www.lesswrong.com/posts/CqyJzDZWvGhhFJ7dY/belief-in-belief) it in order to ease the cognitive dissonance of trying to simultaneously adhere to the religious committments of both _Less Wrong_ "rationalism" and American progressivism. +The weird part is the implication that the form of reasoning you would use to make a decision in the absence of a prediction market can be dismissed as "a way based not on predicted outcomes" and regarded as obviated by the existence of the market. I don't think anyone really believes this, as contrasted to [believing they believe](https://www.lesswrong.com/posts/CqyJzDZWvGhhFJ7dY/belief-in-belief) it in order to ease the cognitive dissonance of trying to simultaneously adhere to the religious commitments of both _Less Wrong_ "rationalism" and American progressivism. The `prediction_market_sort` code doesn't obviate standard sorting algorithms like quicksort, because if you run `prediction_market_sort`, the first thing the traders in the market are going to do is run a standard sorting algorithm like quicksort to decide which comparisons to bet on. -The restaurant-enjoyment market doesn't obviate the concept of Italian food, because if you post a market for "Where should we go for dinner given that Vinnie's is closed?", the first thing traders are going to do is search for "Italian restaurants near [market author's location]"—not because they're fools who think that "Italian food" is somehow ontologically fundamental and eternal, but because there contingently do happen to be [approximate conditional indpendence relationships](https://www.readthesequences.com/Conditional-Independence-And-Naive-Bayes) between the features of meals served by different restaurants. A decision made on the basis of a [statistical compression](https://www.lesswrong.com/posts/mB95aqTSJLNR9YyjH/message-length) of meal features is based on predicted outcomes insofar as and to the extent that meal features predict outcomes. +The restaurant-enjoyment market doesn't obviate the concept of Italian food, because if you post a market for "Where should we go for dinner given that Vinnie's is closed?", the first thing traders are going to do is search for "Italian restaurants near [market author's location]"—not because they're fools who think that "Italian food" is somehow ontologically fundamental and eternal, but because there contingently do happen to be [approximate conditional independence relationships](https://www.readthesequences.com/Conditional-Independence-And-Naive-Bayes) between the features of meals served by different restaurants. A decision made on the basis of a [statistical compression](https://www.lesswrong.com/posts/mB95aqTSJLNR9YyjH/message-length) of meal features is based on predicted outcomes insofar as and to the extent that meal features predict outcomes. -To be sure, there are all sorts of nuances and caveats that one could go into here about exactly when and why categorization works or fails as a cognitive algorithm—how categories are sometimes [used for coordination](https://www.lesswrong.com/posts/edEXi4SpkXfvaX42j/schelling-categories-and-simple-membership-tests) [and not](http://unremediatedgender.space/2019/Oct/self-identity-is-a-schelling-point/) [just predictions](http://unremediatedgender.space/2019/Dec/more-schelling/), how categories should [change when the distribution of data in the world changes](https://www.lesswrong.com/posts/WikzbCsFjpLTRQmXn/declustering-reclustering-and-filling-in-thingspace) (_e.g._, if [fusion cuisines](https://en.wikipedia.org/wiki/Fusion_cuisine) become popular), whether categories might perversely distort the territory to fit the map via self-fulfilling prophecies[^self-fulfilling] (_e.g._, if entrepreneurs only open restaurants in established ethnic categories because that's what customers are used to, thereby stifiling culinary innovation) ... +To be sure, there are all sorts of nuances and caveats that one could go into here about exactly when and why categorization works or fails as a cognitive algorithm—how categories are sometimes [used for coordination](https://www.lesswrong.com/posts/edEXi4SpkXfvaX42j/schelling-categories-and-simple-membership-tests) [and not](http://unremediatedgender.space/2019/Oct/self-identity-is-a-schelling-point/) [just predictions](http://unremediatedgender.space/2019/Dec/more-schelling/), how categories should [change when the distribution of data in the world changes](https://www.lesswrong.com/posts/WikzbCsFjpLTRQmXn/declustering-reclustering-and-filling-in-thingspace) (_e.g._, if [fusion cuisines](https://en.wikipedia.org/wiki/Fusion_cuisine) become popular), whether categories might perversely distort the territory to fit the map via self-fulfilling prophecies[^self-fulfilling] (_e.g._, if entrepreneurs only open restaurants in established ethnic categories because that's what customers are used to, thereby stifling culinary innovation) ... [^self-fulfilling]: Although this is _also_ a potential problem for prediction markets [and other cognitive systems](https://www.lesswrong.com/posts/SwcyMEgLyd4C3Dern/the-parable-of-predict-o-matic). But, bluntly? The kind of person who asks what use there is in "creating a law that asymmetrically treats different sapient beings in a way based not on predicted outcomes" is not interested in the nuances and caveats. (I know because [I used to be this kind of person](/2021/May/sexual-dimorphism-in-the-sequences-in-relation-to-my-gender-problems/#antisexism).) It's not an honest question that expects an answer; it's a rhetorical question asked in the hope that the respondent doesn't have one.[^isolated-demands-for-rigor] -[^isolated-demands-for-rigor]: Which, to be fair, is usually a good bet in the service of the goal of suppressing anti-egalitarian memes. I've specialized in patiently answering [isolated demands for rigor](https://slatestarcodex.com/2014/08/14/beware-isolated-demands-for-rigor/), but most people aren't philosophically sophisticated enough to do that, and the ones that are have competing demands on their time. +[^isolated-demands-for-rigor]: Which is usually a good bet in the service of the goal of suppressing anti-egalitarian memes. I've specialized in patiently answering [isolated demands for rigor](https://slatestarcodex.com/2014/08/14/beware-isolated-demands-for-rigor/), but most people aren't philosophically sophisticated enough to do that, and the ones that are have competing demands on their time. -"If there's more information to be gained from measuring biological sex than by just measuring height, it'll show up in the military prediction markets," Yudkowsky writes. I agree, of course, that that sentence [is literally true](https://www.lesswrong.com/posts/MN4NRkMw7ggt9587K/firming-up-not-lying-around-its-edge-cases-is-less-broadly), but the conditional mood implies such a bizarre prior. "If"? "Just measuring _height_"? Are we pretending to be uncertain about whether a troop of 5'6" males (15th percentile) or 5'6" females (80th percentile) would prevail in high medieval warfare? (Yes, _on average_—but outlier groups are exponentially[^exponentially] rarer than outlier individuals!) Does Yudkowsky want to bet on this? +"If there's more information to be gained from measuring biological sex than by just measuring height, it'll show up in the military prediction markets," Yudkowsky writes. I agree, of course, that that sentence [is literally true](https://www.lesswrong.com/posts/MN4NRkMw7ggt9587K/firming-up-not-lying-around-its-edge-cases-is-less-broadly), but the conditional mood implies such a bizarre prior. "If"? "Just measuring _height_"? Are we pretending to be uncertain about whether a troop of 5'6" males (15th percentile) or 5'6" females (80th percentile) would prevail in high medieval warfare? (Yes, _on average_—but outlier groups are exponentially[^exponentially] rarer than outlier individuals!) Does anyone want to bet on this? [^exponentially]: In the size of the group. ----- -Perhaps at this point the advocate of prediction markets will complain that I'm the one performatively missing the point: the claim isn't that sex is useless for predicting military performance or that restaurant category is useless for predicting meal enjoyment; the claim is that prediction markets can do better by incorporating all other sources of information not encapsulated in a crude, lossy categorization. If the only other Italian place in town is lousy, then we probably want to go somewhere else. +Perhaps at this point the advocate of prediction markets will complain that I'm the one performatively missing the point: the claim isn't that sex is useless for predicting military performance or that restaurant category is useless for predicting meal enjoyment; the claim is that prediction markets can do better by incorporating all other sources of information not encapsulated in a crude, lossy categorization only tangentially relevant to the decision being made. Some men would make terrible soldiers. If the only other Italian place in town is lousy, then you and Barbara obviously want to go somewhere else. Therefore, use prediction markets to pick the actually best soldiers/restaurant, don't just pick the ones with a penis or from Italy. Right? -[TODO: The claim isn't that ethnic or sex categories are useless, but that prediction markets are better, because you can scoop up the exceptions. I've already agreed that if you ignore efficiency, prediction markets can do anything you can do by any other procedure—by construction! So my reply here is that efficiency actually does matter. (If efficiency didn't matter, there would be no gounds to object to the prediction market sort—in principle, it should work.) anti-discrimination law prohibits not just stupid forms of discrimination, but also subtler systems that use categories but also have exception handling. Actually existing militaries don't literally go "anyone with a penis, that's our only criterion"! But they also don't bother drafting women, which was the point under dispute. The choice is "category only" versus prediciton market, you can also search for Italian with good reviews, and that's _doing most of the work of the market_] +But what's at issue isn't whether making decisions _solely_ on the basis of category membership is a good idea, but whether institutions should be able to take category membership into account as a matter of explicit policy (rather than only implicitly via the black box of a prediction market, whose traders are allowed to notice things that the policy isn't). -[TODO: manipulation of small markets?] +Real-world militaries that practice conscription _don't_ just take males indiscriminately with no other requirements than having a Y chromosome, because that would be crazy. The draft board does administer fitness tests and psych evals, consider relevant skills, _&c_. (Similarly, real-world diners craving Italian food also take Yelp ratings into account.) But one of the features real-world militaries _do_ consider is sex, thereby running afoul of the principle of not asymmetrically treating different sapient beings in a way (allegedly) "based not on predicted outcomes." For example, [Israel drafts women](https://en.wikipedia.org/wiki/Women_in_the_Israel_Defense_Forces), but for a shorter term, and [not in all combat roles](https://www.ynetnews.com/articles/0,7340,L-5089566,00.html). (There are a few mixed-sex battalions, like [the 33rd "Caracal"](https://en.wikipedia.org/wiki/Caracal_Battalion), named after [a species of cat with low sexual dimorphism](https://en.wikipedia.org/wiki/Caracal), but it's an exception rather than the norm.) -[TODO: "just go by the presence of penises, don't use anything else" would be stupid, but it's also _not what real-world militaries do, replacing the draft board with a prediction market doesn't seem like it would satisfy the principled antisexist thing Keltham was trying to do ] +If Keltham's objection to Osiriani patriarchy (which restricts women's education and right to hold property) also condemns real-world Israel (which doesn't), it would appear that something has gone wrong with his reasoning. If the problem is that excluding women from education and property ownership is oppressive and not justified on the empirical merits, Keltham should expect to make that case on the empirical merits: that women's economic liberty works great in dath ilan, and the Osiriani don't seem to be a different species for which the empirical merits would be different. -[TODO: you might say that the market is superior because it can account for exceptions—which can indeed exist— -some restraurants or books or solidier defy classification +That's not what we see in the text. Keltham challenges a native to explain things that would go wrong if foreigners imposed a strict Equal Rights Amendment, and when presented with a sensible example (military conscription), rather than saying, "okay, I can see how that one makes pragmatic sense, but that doesn't explain or justify the property thing", he persistently refuses to acknowledge the point. "[H]ow is [unusually strong women being drafted] more terrible than strong men being forced to join an army for less than their self-set wage for that?" [he asks](https://www.projectlawful.com/replies/1817425#reply-1817425), he asks, and when he receives a sensible answer to _that_ (that the women might get taken advantage of sexually, which would have lasting consequences for them), he [objects that truthspell-enabled governance would prevent rapes, and patronizingly wonders whether the Osiriani are aware of the human gestation period](https://www.projectlawful.com/replies/1817432#reply-1817432). -but the original context was Ketlham proposing that the law _cannot_ take categories into account, which is different from having an exception-handling procedure. Real-world conscription systems don't just take everyone with a penis, they also have a draft board which issues nuanced classifications on an individual basis, -https://en.wikipedia.org/wiki/Selective_Service_System#Classifications -] +Of course nothing about Keltham's behavior in this scene entails him affirmatively believing any unambiguously false atomic statement. [Word of God](https://tvtropes.org/pmwiki/pmwiki.php/Main/WordOfGod) claims that "Keltham isn't proposing to actually enforce that prohibition [of laws that mention sex] on Osirion, he's trying to figure out _what the laws are trying to do and why_".[^word-of-god] As I mentioned in ["Comment on a Scene"](http://unremediatedgender.space/2022/Jun/comment-on-a-scene-from-planecrash-crisis-of-faith/), anti-discrimination policy makes sense as game theory: [if you don't trust decisionmakers not to misconstrue group differences in a way that benefits them, forcing them to behave as if all groups were equal is the obvious Schelling point for preventing exploitation](http://unremediatedgender.space/2020/Apr/book-review-human-diversity/#schelling-point-for-preventing-group-conflicts). (The government can't oppress people on the basis of sex if the government isn't allowed to see sex.) ------- +[^word-of-god]: [Eliezerfic Discord server, #dath-ilan channel, 12 June 2022](/images/yudkowsky-keltham_isnt_proposing.png) + +One could construe Keltham's line of questioning as deliberately trying to play that strategy against Osirion: it's not that Keltham is denying that predictively useful categories are also useful for making decisions; he just doesn't trust Osirion's irrationally sexist culture to do that sanely and is eager to explain how good decisions can be recovered in terms of lower-level features at some extra expense. + +I just don't _buy it_. Almost everywhere else in the dath ilan mythos that dath ilan is compared to Earth (_i.e._, the real world) or Golarion, the comparison is unflattering; we're supposed to believe that dath ilan is a superior civilization, a utopia of reason where average intelligence is 2.6 standard deviations higher, where everyone is trained in Bayesian reasoning from childhood. One of the rare places in canon that dath ilan is depicted as not having already thought of something good and useful in the real world is in [the April Fool's Day confession](https://yudkowsky.tumblr.com/post/81447230971/my-april-fools-day-confession), when [NGDP targeting](https://en.wikipedia.org/wiki/Nominal_income_target) is identified as a clever and characteristically un–dath ilani hack. Dath ilan is accustomed to solving coordination problems by the effort of "serious people [...] get[ting] together and try[ing] not to have them be so bad": the mode of thinking that would lead one to propose automatically canceling out the sticky wage effect by printing more money to keep spending constant is alien to them. -[TODO: the original context was about liteary flaws in dath ilan, fictional aliens who "just happen" to share the author's predjudices break suspension of disbelief. (And there should be a careful way to word this principle that's agnostic about whether the author's pet beliefs are true.) Anti-discrimination makes sense as game-theoretic Earth-craziness, but that's not supposed to be dath ilan's M.O.! (It sounds like Keltham actually believes in anti-discrimination as a principle, it's not just a pragmatic bias canceler like NGDP targeting, which dath ilan isn't suppose to have; "cheap hacks to route around other people's irrationality" is not supposed to be what they're going for); Keltham should be able to say "women own property in dath ilan and it works fine _on the empirical merits_" and concede re: conscription; the fact that he can't marks the text as being from Berkeley, not another world] +Anti-discrimination norms are like NGDP targeting: prohibiting certain probabilistic inferences in order to cancel out widespread irrational bigotry is similar to printing money to cancel out a widespread irrational tendency to fire workers instead of lowering nominal wages in that it's not something you would think of in a world where people are just doing decision Bayesian decision theory—and it's not something you would _portray as superior_ if you came from a world that prides itself on just doing Bayesian decision theory and were trying to enlighten the natives of a strange and primitive culture. Yudkowsky's reply to "Comment on a Scene" tries to patch the problem by suggesting that Civilization doesn't need to make those probabilistic inferences anyway because it has prediction markets, but this is an obvious rationalization. (If you disagree, I have an amazing new sorting algorithm that may interest you ...) diff --git a/content/images/yudkowsky-keltham_isnt_proposing.png b/content/images/yudkowsky-keltham_isnt_proposing.png new file mode 100644 index 0000000..da4f4e0 Binary files /dev/null and b/content/images/yudkowsky-keltham_isnt_proposing.png differ