Can It Be Wrong To Crystallize Patterns? | Slate Star Codex

In my last post I mentioned my experience of reading over some essays I found really enlightening years ago, finding them on re-examination to be correct but boring, and half doubting that I had ever not-known the things within. In the comments many people said they shared that experience, either with the Less Wrong sequences or something else.

This never happens with other kinds of knowledge. I remember learning the capital of Cuba in my sixth grade Geography class. It is very clear in my memory that before I did this, I didn’t know the capital of Cuba. I can easily conceive of forgetting the capital of Cuba, and indeed if I went long enough without thinking about it I would expect that to happen.

Compare this factual knowledge of the Cuban capital to the idea of mysterious answers to mysterious questions.

The capital of Cuba is a known unknown. Before I learned it was Havana, I knew Cuba must have a capital and that this was a hole in my knowledge. It’s hard to imagine what it would mean for me to understand that I needed a concept like “mysterious answers” without, by that fact itself, having the concept.

Information about Cuba’s capital can be wrong – a vandalized Wikipedia article might tell me it’s Cuba City. I’m not sure what it would mean for the idea of Mysterious Answers to be wrong. The impression I get from that post is that it’s saying “Here is a useful concept, take it or leave it”. If you didn’t like the post, it would be more because you thought very few theories or ideas were really Mysterious Answers, or because Mysterious Answers didn’t have enough extra explanatory power to be worth your time to learn – not because the concept was wrong per se. But I think a more likely reaction is to see the examples and immediately think “Yes, this is a very common failure mode. I can easily think of a lot of times I’ve seen it used. Now I will be able to spot and recognize it more quickly”

Teaching the capital of Cuba imparts new information. Teaching about Mysterious Answers doesn’t, at least for a strict definition of information. Maybe most readers already know all of the particular examples in that post. All they hadn’t done was detect a pattern and give a name to it.

A while ago, a leftist commenter asked me to stop using the term “political correctness”. Without getting into whether or not I should comply, I was struck by the sound strategy behind the request. Without the words “political correctness”, an entire class of things becomes much harder to think about. There are whole debates that would stop happening if the words “political correctness” got removed from the vocabulary. That’s pretty strange. If we removed the word “President” from the vocabulary, people would just start talking about “that guy in the White House” or “the executive branch” or something. But striking out “political correctness” would be a genuine game-changer.

“Political correctness” is also an attempt to detect a pattern (for the purpose of not politicizing this post, I will stay agnostic whether it is real or imagined) and give a name to it. And I think if we stopped using the term, people would stop detecting the pattern and their views about a lot of things would change.

Giving a pattern a name crystallizes it, let people hold it in their minds for a longer period, make it easier to think about and discuss.

This is neat because it lets you formulate a meta-level response to the pattern. If you realize that “God” is sometimes a Mysterious Answer, and “elan vital” is sometimes a Mysterious Answer, then when you’re tempted to use something you like as a Mysterious Answer (“How did the eye form?” “Evolution!”) you’re more likely to recognize it, admit it’s a problem, and apply a general policy of not being satisfied with Mysterious Answers.

And when you encounter it again, it’s much easier to dismiss in a single mental motion. A lot of the problem with debate is time and number-of-mental-steps: the amount of time and energy normal people give to debating things isn’t enough to get past the preliminary outer defenses of bad ideas, and in the rare cases it is, by the time the last defenses have been breached the first ones have already started regenerating. Turning long series of steps into a single convenient package decreases the amount of stuff you need to keep in your mental workspace.

I will now dump a long list of concepts that I think work in more or less this way: “trivial inconveniences”, “Schelling points”, “spoon theory“, “ad hominem fallacy”, “introversion/extraversion”, “social justice”, “ask culture”, “global (versus local) maxima/minima”, “growth mindset”, “logical uncertainty”, “crackpot”, “cult”, “meme”, “self-reference”, “utility”, “warm fuzzies”, “Third World”, “efficient charity”, “entryism”, “witch hunt”, “deontology”, “romantic love”…

…it’s going to be easier to list things that don’t work like this, isn’t it?

“Witch hunt” and “romantic love” seem like especially interesting examples. In olden days, if someone was having a moral panic and decided to unfairly persecute a group of people for something they probably weren’t doing, you would have to come up with some complicated way of explaining this to the public and convincing them that this is a plausible thing that might happen (you can’t borrow my wording; “moral panic” is almost as artificial a crystallized pattern as “witch hunt” is). Good luck getting the public to hold still for that. Now you can just say “They’re holding a witch hunt!” and everyone will understand your objection. I think the existence of the concept “witch hunt” lets everybody coordinate to develop solutions to problems like “How can we best prevent witch hunts from happening?” and really does make witch hunts significantly more difficult.

“Romantic love” shares with the Sequences this pattern of once you’ve got it it’s hard to imagine not having it. We know a lot of our notion of romantic love is Western and modern, but it requires a lot of evidence to prove this to most people because it seems so natural. It’s one possible way of parsing sex and attraction and stuff, and a very natural one for us, but one that was very rarely used in certain circumstances (the crystallization of the pattern “romantic love” probably shouldn’t be confused with semifactual questions like “should we choose partners based on romantic love?”, which already presuppose that we’ve accepted romantic love as a useful concept)

And I’m thinking about these things because of Alexander Stanislaw’s comment on my last post:

Perhaps being immersed in a memeplex such as LW for long enough is sufficient to make the source material seem obvious, whether or not the memes are correct. I’m sure that there are intelligent Christians our age who are re-reading C.S. Lewis and thinking, “wow this is so obvious”. I admit, I find it very hard to argue with the LW stance on what concepts are for instance, and it drives me nuts when people try to define themselves into being correct, or argue about whether X is really Y, or equivocate between definitions of words. But apparently Gilbert of the Last Conformer disagrees with the LW stance on concepts and he seems knowledgeable.

It’s an interesting question: can crystallizing a pattern be wrong?

First, two kinda trivial examples.

Crystallizing the pattern “ley lines” wouldn’t be super useful if you don’t believe ley lines exist. From one perspective, the crystallization is still nice, because you have an extra concept to toss around and you can do things you couldn’t do before like debate the existence of ley lines (and settle on the negative position). From another perspective, the existence of the pattern makes it too tempting to start believing in ley lines or at least thinking they’re important. So in theory crystallizing the concept “ley lines” is an unalloyed positive (minus the five minutes it takes you to read the definition), but in practice it’s a bad idea.

Crystallizing the pattern “mainstream media” in the sense that some conspiracy theorists use it – ie “this exciting archaeological discovery will never be mentioned in the mainstream media” seems more complicated. There definitely exists a “mainstream media” in the sense of “a media that has much different standards and preferences than the local conspiracy theory rag”. And it’s probably useful to talk about – my reaction to the Siberian discovery was “I’ll believe it once I see it in the mainstream media”. But the way the term is used manages to smuggle in questionable connotations – this is some people’s problem with “political correctness” as well, and with other political terms like “patriarchy”, “the Cathedral”, “international Jewry”, et cetera.

I think the problem here might be that pointing out a pattern suggests an agency fiction? There’s clearly international Jewry in the sense of “Jews in many different countries”. One can even make moderately correct statements about international Jewry, like “international Jewry often puts pressure on different countries to support Israel”, which when stripped of its connotations is more or less correct. But the connotations suggest either that they’re all working together in a conspiracy, or that they’re all doing it in about the same way which is becoming a noteworthy pattern, or that it’s something much more interesting than the way other ethnic groups do other things.

I think this was my interlocutor’s problem with political correctness as well – certainly there are things that fit that category, but once the word starts to be used it’s singling it out as A Unique And Interesting Problem.

But I don’t know how to solve this sort of issue in a way that’s easily distinguishable from trying to ban people from ever talking about political correctness or Jewish support for Israel. Making people use a different term than “international Jewry” (“Jews around the world”?) would work for about ten minutes until it got loaded with exactly the same connotations as the original.

But these two examples seem less interesting than the original question about whether you can crystallize patterns, without falling into any traps or connotations – and still have it be a net loss.

I’m trying to think of something that fits Alexander’s example – something written by C.S. Lewis that provides Christians with much more clarity and afterwards they feel like they’re much more advanced and can’t remember how they lived without the concept – but which from my perspective as a non-Christian is wrong and counterproductive.

(my immediate reaction is “ring theory!”, but that’s not dependent on Christianity and probably correct and useful, so it doesn’t count)

The closest I can come (I am not a Lewis scholar) are his trilemma, his use of the different phases of water (ice, liquid, steam) to explain the Trinity, and his famous saying that “going to church does not make you a Christian any more than going to a garage makes you a car”.

But I like all of these!

Although I don’t agree with his formulation of the trilemma, it puts something that people would otherwise spend interminable hours debating without ever reaching the core of the issue – into sharp relief, so that it can be attacked or defended in its strongest form.

And although I don’t accept the premises of the Trinity or the special necessity of figuring out exactly who is or isn’t a “real Christian”, if I admired the problems I would certainly admire Lewis’ solutions to them.

The worst that can be said about Lewis is that he came up with excessively clever solutions to the wrong problems – that is, maybe if someone was wrestling with the idea of the Trinity, that would convince them to second-guess their religion, but Lewis came up with such an elegant metaphor that it prevents them from having to do that.

And I’m not sure that the garage saying or the Trinity analogy are crystallized patterns per se. The trilemma example might be the purest – but that’s also the one I like most.

So provisionally I’m not sure there’s such a thing as crystallizing a pattern and being wrong to do so. You can crystallize patterns in such a way that it ends out misleading people who were already at risk of being misled – like the “ley lines” and “international Jewry” examples – and in practice this is a HUGE HUGE problem. But it seems to me that if you’re good enough at sorting through connotations to handle it that crystallization is usually a good idea