Dolphin War!! 🐬 (draft reply to Nate Soares, 16 June 2021)

I noticed my own discomfort as the lines around "fruit" and "berry" started wavering

"Started" wavering? How often does this come up for you? I'm somehwat (not hugely) surprised at your self-report, because I can't actually recall ever having been in a dispute over the extensions of "fruit" or "berry" before!

Furthermore, when I try to imagine how I would have reacted if it did happen, I don't imagine myself feeling any discomfort at wavering lines! I'm imagining mentioning to a friend that I'd been ordering frozen fruit (which keeps longer) in grocery deliveries while avoiding going to supermarkets in person during the COVID-19 pandemic, but that I liked the Cascadian Farm Cherry Berry Blend (of cherries, strawberries, and blueberries) better than the Wyman's of Maine mixed berry bag (of strawberries, blueberries, raspberries, and blackberries).

If the friend were to reply, "Well, actually, strawberries, raspberries, and blackberries aren't berries", I think my reaction would be to look at the Wikipedia page and say, "Oh, it looks like that's true with respect to the botanical definition of the word berry, but the fruit companies were clearly using the colloquial definition. It's possible for the same word to have multiple meanings that can usually be distinguished from context, or explicitly clarified by the speaker if the context alone is insufficient to disambiguate!"

Wikipedia is very clear about the multiple definitions: "In common usage, the term "berry" differs from the scientific or botanical definition [...] The botanical definition includes many fruits that are not commonly known or referred to as berries". They even have a separate page for "Berry (botany)"!

Wiktionary also lists both definitions, with the botanical usage clearly marked as such with a context label, and with a usage note explaining further:

Noun
berry (plural berries)

  1. A small succulent fruit, of any one of many varieties.
  2. (botany) A soft fruit which develops from a single ovary and contains seeds not encased in pits.
    [...]
    Usage Notes
    Many fruits commonly regarded as berries, such as strawberries and raspberries, are not berries in the botanical sense, while many fruits which are berries in the botanical sense are not regarded as berries in common parlance, such as bananas and pumpkins.

The same situation pertains with fruit: both Wikipedia ("In common language usage [...] In botanical usage") and Wiktionary (senses 2 vs. 3) are very clear about the two meanings coexisting!

I don't see any wavering lines here at all: I see two different category boundaries that happen to share the same label (requiring disambiguation in cases of doubt), with all authoritative sources explicitly flagging the label collision!

Based on this evidence, I find your claim that "they're coming for 'fruit'" hard to believe. Who is doing that, in the social spaces you inhabit? Is anyone with equal or greater influence than the executive director of MIRI doing this? If you cannot provide evidence of people using their social power to deprecate the common meaning in favor of the botanical one, then your attempt to use your power to deprecate the botanical meaning reeks of pure DARVO. When you sneer "Invent a new word for your thing" in reference to a usage that already has already been established (e.g., the first definition of fish, or berry/fruit among botanists in particular), then you are the one trying to steal concepts!

It's really weird that your epistemic-status thread laments that not enough people read "A Human's Guide to Words", but you're acting as if you not only haven't read "Where to Draw the Boundary?", but that you also haven't read "Disputing Definitions" and "Variable Question Fallacies". Are strawberries berries? Well, "berry" has two senses—strawberries are a berry(1) but not a berry(2). This is not a difficult concept. Your low quality shitposting makes me think less of MIRI as an institution—and therefore think less of humanity's hopes for a future. If you clowns can't get your own philosophy of language right, why should we trust you to save the world? You betrayed your uniform!

I don't actually believe phylogenetics is generally a good way to carve up the life forms around me

Do you still feel this way after considering the "genetics are at the root of the causal graph underlying all other features of an organism" argument offered in this post? This seems like the kind of deep structural insight that you'd think so-called "rationalists" would be interested in! The variable you use as your category definition being at the root of the causal graph guarantees the conditional independence structure that makes naĂŻve-Bayes-like categories work, as an AI algorithm!!

If you live in human civilization and don't really need to carve up the life forms around you in much detail—if your use-case for thinking about acquatic animals is watching a nature documentary (for entertainment??) rather than living and working with them every day, then you might think the deeper causal structure isn't buying you anything, but I think you would change your mind if you were a veternarian or a zoologist who actually had skin in the game in describing this part of the world.

When people have skin in the game, they care about the underlying mechanisms and want short codewords for them. If you hurt your ankle while running, you would probably be interested to know whether it was a sprain or a stress fracture because that affects your decisions about how to recover. You wouldn't say, "Well, all I know is that my ankle hurts—that's all a child would know—so I'm going to call it a hurtankle; I don't care about anatomy."

Phylogenetics is also a lot more objective: it's easier to stabilize a shared map on the facts of evolutionary history (which we can all agree on in the limit of good science) then the vagaries of what I personally think tastes good on pizza.

And this subjectivity also applies to animal classification. Some might be inclined to argue "bats are birds" (flappy flying animals) on the same grounds as "dolphins are fish" (flappy swimmy animals). But did you know the German word for bat is Fledermaus ("flutter mouse"), which dates back to fledarmĹŤs in Old High German? Apparently, people way back in the tenth century or so (long before evolution was understood) already thought bats were like a mammal-that-happened-to-fly rather than a bird-that-happened-to-be-furry.

the subtle background forces that whisper (at least to blue tribe members in their youth) "phylogenetic classification is the one true way to organize life forms"

Why to Blue Tribe members in their youth, particularly? Yesterday, I mentioned to my mother (born 1951, old enough that the forces shaping her worldview long predate the polarization leading to the local coinage of "Blue Tribe") that the reason I had a terrible week was because I got into a fight about dolphins.

"Dolphins?" she asked, incredulously.

"They're not fish, right?" I asked.

"No, of course not," she said. "They're mammals. Dolphins take care of their young. Dolphins can communicate with each other; they're intelligent."

Why? What changed?

On the object level question?
[...]
Scott's post is I think the source of the first two in me

As for my cynical theorizing of what changed, I'm most interested as to what conversations we would be having in the counterfactual where trans rights didn't become a popular political issue. (Imagine a world where the dice of previous elections and judicial appointments landed such that the gay-marriage fight would have taken much longer to win, such that the LGBT activism machine was still focused on that even through the current year, and didn't have the resources to mount the psychological warfare campaign to replace the public concept of sex with "gender identity" that we're experiencing in this timeline.)

I would be surprised if Scott genuinely, honestly thought dolphins should be fish if you asked him in 2012 (or in my alternate timeline where we still didn't have gay marriage yet). I think it's much more likely that he first felt psychological pressure to come up with a clever casuistry for the bottom line of "trans women are women / trans men are men" (sections IV.–VI.), and then came up with the "whales are fish" thing for the intro section.

I know, I know: caution on bias arguments, we have a terrible track record in guessing Scott's motives. I agree that psychologizing about your interlocutor's political incentives is a fraught endeavor and most people get it wrong! We're probably designed to get it wrong, if there were payoffs in the environment of evolutionary adaptedness to think badly of enemies—either because of asymmetric payoffs (less expensive to fear a nonexistent murder plot than fail to notice a real one), or to search for clever arguments to recruit allies to oppose the enemy.

Nevertheless, think about it: can you imagine Scott making the "whales are fish" case, if he weren't leading to that politically convenient bottom line? Imagine the same post with just the whales-are-fish and categories-are-like-national-borders parts, no parts IV.–VI.? Do you think that post would be popular?

It's notable that the "algorithmic" lie (far too optimized to call a mere "mistake", even if Scott wasn't consciously lying) that makes sections IV.–VI. of "... Not Man for the Categories" mind-poison—the idea that we can legitimately choose categories in order to make people happy (!?!?!), rather than to reflect the territory—isn't something that Scott does anywhere else during that period of his work. For example, "Does the Glasgow Coma Scale exist? Do comas?", published three months earlier, gets the categories issue right: "there is a common-sense concept of being-in-a-coma which is valuable in predicting various things we want to predict, like whether someone is able to talk and able to walk and able to solve math problems and so on"—the "coma" concept should be optimized to reflect the territory, not to make someone feel better.

(It's all so pointless, too—that people think they have to support these Orwellian mind games in order to support trans people. I'm not against trans people, who I'm arguably one of—depending on where you draw the boundary. I'm grateful that I got to try hormone replacement therapy for five months with medical supervision, instead of it being illegal. Body modding to more closely resemble the opposite sex (in the sad absence of the technology to do it for real) is probably a reasonable thing for some people to do in some circumstances, and she is obviously the correct pronoun for people who look like women! But one would have hoped that social support for this could have been won on principled morphological freedom grounds (the same reason people should be free to get a tattoo or a piercing) rather than this culture of mind games bent on destroying the language that I, as a transgender person, need to tell the truth about what I want.)

Anyway, in the world where the gay marriage fight took longer, I don't think "Dolphins are fish, fight me" or "Acknowledge that all of our categories are weird and a little arbitrary" would have become memes. But since Scott won the talent lottery and writes faster and better than everyone else, he has the power to sneeze his mistakes onto everyone else who trusts Scott to do his homework even when he obviously hasn't. (Would section I. of "... Not Man for the Categories" have concluded with "(also, bats)" if Scott knew about die Fledermäuse?)

This is the mechanism that Yudkowsky called dark side epistemology: fake epistemology lessons that people only have an incentive to invent in order to force a conclusion that they can't get on the merits. I don't question your self-report of your motives! I believe that you weren't trying to play political games with me! But your information environment is supersaturated with the toxic waste of everyone else's disinformation, that you don't have the cognitive resources to distangle yourself: if "dolphins are fish, Actually" is suddenly a trendy meme after 2014, and you don't notice that the meme is causally downstream of dark side epistemology, then you won't see anything wrong with running with that thought.

In this specific case—a world where more people are reading "... Not Man for the Categories" than "Mutual Information, and Density in Thingspace", and even the people who have read "Density in Thingspace" (once, ten years ago) are having most of their conversations with people who only read "... Not Man for the Categories"—what happens is that you end up with a so-called "rationalist" culture that completely forgot the hidden-Bayesian-structure-of-cognition/carve-reality-at-the-joints skill! People only remember the specific subset of "A Human's Guide to Words" that's useful for believing whatever you want (by cherry-picking the features you need to include in category Y to make your favorite "X is a Y" sentence look "true", which is easy for intricate high-dimensional things like biological creatures that have a lot of similarities to cherry-pick from), rather than the part about the conditional independence structure in the environment required for "categories" to be a useful AI trick.

As a result of your shitposting, "... Not Man for the Categories" was cited three times by other Twitter users, driving yet more traffic to a post that claims (redacting the specific object-level example)—

I ought to accept an unexpected [X] or two deep inside the conceptual boundaries of what would normally be considered [Y] if it'll save someone's life. There's no rule of rationality saying that I shouldn't, and there are plenty of rules of human decency saying that I should.

Those sentences are a complete betrayal of everything we allegedly stand for. I think you, Nate Soares, need to condemn those sentences, out loud, in public, as a complete betrayal of everything we allegedly stand for. And if you can't do it, then you're a fraud. If the leaders of the so-called "rationalist" community can't condemn it (either because of the politics, or because you're somehow actually that retarded), then the so-called "rationalist" community needs to rebrand, or disband, or be destroyed.