From 1c4be9a714deacdfe311d4838e9975e339c237b8 Mon Sep 17 00:00:00 2001 From: "Zack M. Davis" Date: Fri, 27 Oct 2023 22:04:04 -0700 Subject: [PATCH] memoir: pt. 3 edit sweep ... --- .../if-clarity-seems-like-death-to-them.md | 31 ++++++++++--------- notes/memoir-sections.md | 2 +- 2 files changed, 18 insertions(+), 15 deletions(-) diff --git a/content/drafts/if-clarity-seems-like-death-to-them.md b/content/drafts/if-clarity-seems-like-death-to-them.md index 79fe376..81dbbfc 100644 --- a/content/drafts/if-clarity-seems-like-death-to-them.md +++ b/content/drafts/if-clarity-seems-like-death-to-them.md @@ -670,6 +670,7 @@ But what was in question wasn't the observations of Hans's performance, only the * Felix Fix-It, Jr. anecdote * lemon anecdote * fascination with forklifts at the same time the pronoun change is going on, but with wisdom of previous generations, and d=2.44 in the literature, has no weight + * http://zackmdavis.net/blog/2020/02/relationship-outcomes-are-not-particularly-sensitive-to-small-variations-in-verbal-ability/ ] Anyway, that's just a hypothesis that occured to me in early 2020, about something that _could_ happen in the culture of the current year, hypothetically, as far as I know. I'm not a parent and I'm not an expert on child development. And even if the "Clever Hans" etiological pathway I conjectured is real, the extent to which it might apply to any particular case is complex; you could imagine a kid who _was_ "actually trans", whose social transition merely happened earlier than it otherwise would have due to these dynamics. @@ -688,45 +689,47 @@ Given that I spent so many hours on this little research and writing project in ----- -On 1 June 2020, I received a Twitter DM from _New York Times_ reporter Cade Metz, who said he was "exploring a story about the intersection of the rationality community and Silicon Valley". I sent him an email saying that I would be happy to talk, but that I'd actually been pretty disappointed with the community lately: I was worried that the social pressures of trying to _be_ a "community" and protect the group's status (_e.g._, from _New York Times_ reporters who might portray us in an unflattering light??) incentivize people to compromise on the ideals of _systematically correct reasoning_ that made the community valuable in the first place. +On 1 June 2020, I received a Twitter DM from _New York Times_ reporter Cade Metz, who said he was "exploring a story about the intersection of the rationality community and Silicon Valley". I sent him an email saying that I would be happy to talk, but that I'd actually been pretty disappointed with the community lately: I was worried that the social pressures of trying to _be_ a "community" and protect the group's status (_e.g._, from _New York Times_ reporters who might portray us in an unflattering light?) incentivize people to compromise on the ideals of systematically correct reasoning that made the community valuable in the first place. He never got back to me. Three weeks later, all existing _Slate Star Codex_ posts were taken down. A [lone post on the main page](https://slatestarcodex.com/2020/06/22/nyt-is-threatening-my-safety-by-revealing-my-real-name-so-i-am-deleting-the-blog/) explained that the _New York Times_ piece was going to reveal Alexander's real last name, and that he was taking his posts down as a defensive measure. (No blog, no story?) I [wrote a script](/source?p=Ultimately_Untrue_Thought.git;a=commitdiff;h=21731ba6f1191) (`slate_starchive.py`) to replace the _Slate Star Codex_ links on this blog with links to the most recent Internet Archive copy. ------ -I continued to work on my "advanced" philosophy of categorization thesis. The disclaimer note that Scott Alexander had appended to "... Not Man for the Categories" after our Christmas 2019 discussion had said: +I continued my philosophy of language work, looking into the academic literature on formal models of communication and deception and writing a [couple](https://www.lesswrong.com/posts/4hLcbXaqudM9wSeor/philosophy-in-the-darkest-timeline-basics-of-the-evolution) [posts](https://www.lesswrong.com/posts/YptSN8riyXJjJ8Qp8/maybe-lying-can-t-exist) encapsulating what I learned from that—and I continued work on my "advanced" philosophy of categorization thesis, the sequel to ["Where to Draw the Boundaries?"](https://www.lesswrong.com/posts/esRZaPXSHgWzyB2NL/where-to-draw-the-boundaries) + +The disclaimer note that Scott Alexander had appended to "... Not Man for the Categories" after our Christmas 2019 discussion had said: > I had hoped that the Israel/Palestine example above made it clear that you have to deal with the consequences of your definitions, which can include confusion, muddling communication, and leaving openings for deceptive rhetorical strategies. -This is certainly an _improvement_ over the original text without the note, but I took the use of the national borders metaphor here to mean that Scott still hadn't really gotten my point about there being underlying laws of thought underlying categorization: mathematical principles governing _how_ definition choices can muddle communication or be deceptive. (But that wasn't surprising; [by Scott's own admission](https://slatestarcodex.com/2013/06/30/the-lottery-of-fascinations/), [he's not a math guy](https://slatestarcodex.com/2015/01/31/the-parable-of-the-talents/).) +This is certainly an improvement over the original text without the note, but I took the use of the national borders metaphor to mean that Scott still hadn't gotten my point about there being underlying laws of thought underlying categorization: mathematical principles governing _how_ definition choices can muddle communication or be deceptive. (But that wasn't surprising; [by Scott's own admission](https://slatestarcodex.com/2013/06/30/the-lottery-of-fascinations/), [he's not a math guy](https://slatestarcodex.com/2015/01/31/the-parable-of-the-talents/).) Category "boundaries" are a useful _visual metaphor_ for explaining the cognitive function of categorization: you imagine a "boundary" in configuration space containing all the things that belong to the category. -If you have the visual metaphor, but you don't have the math, you might think that there's nothing intrinsically wrong with squiggly or discontinuous category "boundaries", just as there's nothing intrinsically wrong with Alaska not being part of the contiguous U.S. states. It may be inconvenient that you can't drive from Alaska to Washington without going through Canada, and we have to deal with the consequences of that, but there's no sense in which it's _wrong_ that the borders are drawn that way: Alaska really is governed by the United States. +If you have the visual metaphor, but you don't have the math, you might think that there's nothing intrinsically wrong with squiggly or discontinuous category "boundaries", just as there's nothing intrinsically wrong with Alaska not being part of the contiguous U.S. states. It may be inconvenient that you can't drive from Alaska to Washington without going through Canada, and we have to deal with the consequences of that, but it's not "wrong" that the borders are drawn that way: Alaska really is governed by the United States. -But if you _do_ have the math, a moment of introspection will convince you that the analogy between category "boundaries" and national borders is shallow. +But if you do have the math, a moment of introspection will convince you that the analogy between category "boundaries" and national borders is shallow. -A two-dimensional political map tells you which areas of the Earth's surface are under the jurisdiction of what government. In contrast, category "boundaries" tell you which regions of very high-dimensional configuration space correspond to a word/concept, which is useful _because_ that structure is useful for making probabilistic inferences: you can use your observastions of some aspects of an entity (some of the coordinates of a point in configuration space) to infer category-membership, and then use category membership to make predictions about aspects that you haven't yet observed. +A two-dimensional political map tells you which areas of the Earth's surface are under the jurisdiction of which government. In contrast, category "boundaries" tell you which regions of very high-dimensional configuration space correspond to a word/concept, which is useful _because_ that structure is useful for making probabilistic inferences: you can use your observastions of some aspects of an entity (some of the coordinates of a point in configuration space) to infer category-membership, and then use category membership to make predictions about aspects that you haven't yet observed. -But the trick only works to the extent that the category is a regular, non-squiggly region of configuration space: if you know that egg-shaped objects tend to be blue, and you see a black-and-white photo of an egg-shaped object, you can get _close_ to picking out its color on a color wheel. But if egg-shaped objects tend to blue _or_ green _or_ red _or_ gray, you wouldn't know where to point to on the color wheel. +But the trick only works to the extent that the category is a regular, non-squiggly region of configuration space: if you know that egg-shaped objects tend to be blue, and you see a black-and-white photo of an egg-shaped object, you can get close to picking out its color on a color wheel. But if egg-shaped objects tend to blue _or_ green _or_ red _or_ gray, you wouldn't know where to point to on the color wheel. -The analogous algorithm applied to national borders on a political map would be to observe the longitude of a place, use that to guess what country the place is in, and then use the country to guess the latitude—which isn't typically what people do with maps. Category "boundaries" and national borders might both be _illustrated_ similarly in a two-dimensional diagram, but philosophically, they're different entities. The fact that Scott Alexander was appealing to national borders to explain why gerrymandered categories were allegedly okay, showed that he Didn't Get It. +The analogous algorithm applied to national borders on a political map would be to observe the longitude of a place, use that to guess what country the place is in, and then use the country to guess the latitude—which isn't typically what people do with maps. Category "boundaries" and national borders might both be illustrated similarly in a two-dimensional diagram, but philosophically, they're different entities. The fact that Scott Alexander was appealing to national borders to explain why gerrymandered categories were allegedly okay, demonstrated that he didn't understand this. -I still had some deeper philosophical problems to resolve, though. If squiggly categories were less useful for inference, why would someone _want_ a squiggly category boundary? Someone who said, "Ah, but I assign higher utility to doing it this way", had to be messing with you. Where would such a utility function come from? Intuitively, it had to be precisely _because_ squiggly boundaries were less useful for inference; the only reason you would realistically want to do that would be to commit fraud, to pass off pyrite as gold by redefining the word "gold." +I still had some deeper philosophical problems to resolve, though. If squiggly categories were less useful for inference, why would someone want a squiggly category boundary? Someone who said, "Ah, but I assign higher utility to doing it this way" had to be messing with you. Where would such a utility function come from? Intuitively, it had to be precisely _because_ squiggly boundaries were less useful for inference; the only reason you would realistically want to do that would be to commit fraud, to pass off pyrite as gold by redefining the word "gold". -That was my intuition. To formalize it, I wanted some sensible numerical quantity that would be maximized by using "nice" categories and get trashed by gerrymandering. [Mutual information](https://en.wikipedia.org/wiki/Mutual_information) was the obvious first guess, but that wasn't it, because mutual information lacks a "topology", a notion of _closeness_ that made some false predictions better than others by virtue of being "close". +That was my intuition. To formalize it, I wanted some sensible numerical quantity that would be maximized by using "nice" categories and get trashed by gerrymandering. [Mutual information](https://en.wikipedia.org/wiki/Mutual_information) was the obvious first guess, but that wasn't it, because mutual information lacks a "topology", a notion of "closeness" that made some false predictions better than others by virtue of being "close". -Suppose the outcome space of _X_ is `{H, T}` and the outcome space of _Y_ is `{1, 2, 3, 4, 5, 6, 7, 8}`. I wanted to say that if observing _X_=`H` concentrates _Y_'s probability mass on `{1, 2, 3}`, that's _more useful_ than if it concentrates _Y_ on `{1, 5, 8}`—but that would require the numerals in _Y_ to be _numbers_ rather than opaque labels; as far as elementary information theory was concerned, mapping eight states to three states reduced the entropy from lg2 8 = 3 to lg2 3 ≈ 1.58 no matter "which" three states they were. +Suppose the outcome space of _X_ is `{H, T}` and the outcome space of _Y_ is `{1, 2, 3, 4, 5, 6, 7, 8}`. I wanted to say that if observing _X_=`H` concentrates _Y_'s probability mass on `{1, 2, 3}`, that's more useful than if it concentrates _Y_ on `{1, 5, 8}`. But that would require the numerals in _Y_ to be numbers rather than opaque labels; as far as elementary information theory was concerned, mapping eight states to three states reduced the entropy from lg2 8 = 3 to lg2 3 ≈ 1.58 no matter "which" three states they were. -How could I make this rigorous? Did I want to be talking about the variance of my features conditional on category-membership? Was "connectedness" intrinsically the what I wanted, or was connectedness only important because it cut down the number of possibilities? (There are 8!/(6!2!) = 28 ways to choose two elements from `{1..8}`, but only 7 ways to choose two contiguous elements.) I thought connectedness was intrinsically important, because we didn't just want _few_ things, we wanted things that are _similar enough to make similar decisions about_. +How could I make this rigorous? Did I want to be talking about the variance of my features conditional on category-membership? Was "connectedness" intrinsically the what I wanted, or was connectedness only important because it cut down the number of possibilities? (There are 8!/(6!2!) = 28 ways to choose two elements from `{1..8}`, but only 7 ways to choose two contiguous elements.) I thought connectedness was intrinsically important, because we didn't just want _few_ things, we wanted things that are similar enough to make similar decisions about. I put the question to a few friends in July 2020 (Subject: "rubber duck philosophy"), and Jessica said that my identification of the variance as the key quantity sounded right: it amounted to the expected squared error of someone trying to guess the values of the features given the category. It was okay that this wasn't a purely information-theoretic criterion, because for problems involving guessing a numeric quantity, bits that get you closer to the right answer were more valuable than bits that didn't. ------ -I decided on "Unnatural Categories Are Optimized for Deception" as the title for my advanced categorization thesis. Writing it up was a large undertaking. There were a lot of nuances to address and potential objections to preëmpt, and I felt that I had to cover everything. (A reasonable person who wanted to understand the main ideas wouldn't need so much detail, but I wasn't up against reasonable people who wanted to understand.) +I decided on "Unnatural Categories Are Optimized for Deception" as the title for my advanced categorization thesis. Writing it up was a major undertaking. There were a lot of nuances to address and potential objections to preëmpt, and I felt that I had to cover everything. (A reasonable person who wanted to understand the main ideas wouldn't need so much detail, but I wasn't up against reasonable people who wanted to understand.) -In September 2020, Yudkowsky Tweeted [something about social media incentives prompting people to make nonsense arguments](https://twitter.com/ESYudkowsky/status/1304824253015945216). Something in me boiled over. The Tweet was fine in isolation, but I rankled at it in the context of his own incentive-driven nonsense remaining unaddressed. I left [a pleading, snarky reply](https://twitter.com/zackmdavis/status/1304838486810193921) and [vented on my own timeline](https://twitter.com/zackmdavis/status/1304838346695348224) (with preview images from the draft of "Unnatural Categories Are Optimized for Deception"): +In September 2020, Yudkowsky Tweeted [something about social media incentives prompting people to make nonsense arguments](https://twitter.com/ESYudkowsky/status/1304824253015945216), and something in me boiled over. The Tweet was fine in isolation, but I rankled at it given the absurdly disproprotionate efforts I was undertaking to unwind his incentive-driven nonsense. I left [a pleading, snarky reply](https://twitter.com/zackmdavis/status/1304838486810193921) and [vented on my own timeline](https://twitter.com/zackmdavis/status/1304838346695348224) (with preview images from the draft of "Unnatural Categories Are Optimized for Deception"): > Who would have thought getting @ESYudkowsky's robot cult to stop trying to trick me into cutting my dick off (independently of the empirical facts determining whether or not I should cut my dick off) would involve so much math?? OK, I guess the math part isn't surprising, but— diff --git a/notes/memoir-sections.md b/notes/memoir-sections.md index 8fd8a23..77d78eb 100644 --- a/notes/memoir-sections.md +++ b/notes/memoir-sections.md @@ -19,6 +19,7 @@ pt. 3 edit tier— ✓ "I" statements ✓ we can go stronger than "I definitely don't think Yudkowsky thinks of himself ✓ cut words from December 2019 blogging spree +✓ mention "Darkest Timeline" and Skyrms somewhere ----- _ Ben on "locally coherent coordination": use direct quotes for Ben's language—maybe rewrite in my own language (footnote?) as an understanding test _ "Not the Incentives"—rewrite given that I'm not shielding Ray @@ -38,7 +39,6 @@ _ in a footnote, defend the "cutting my dick off" rhetorical flourish _ choice quotes in "end of the Category War" thank you note _ do I have a better identifier than "Vassarite"? _ maybe I do want to fill in a few more details about the Sasha disaster, conditional on what I end up writing regarding Scott's prosecution?—and conditional on my separate retro email—also the Zolpidem thing -_ mention "Darkest Timeline" and Skyrms somewhere _ footnote explaining quibbles? (the first time I tried to write this, I hesitated, not sure if necessary) _ "it was the same thing here"—most readers are not going to understand what I see as the obvious analogy _ first mention of Jack G. should introduce him properly -- 2.17.1