LESSWRONG
LW

475
gjm
31704Ω373073761
Message
Dialogue
Subscribe

Hi. I'm Gareth McCaughan. I've been a consistent reader and occasional commenter since the Overcoming Bias days. My LW username is "gjm" (not "Gjm" despite the wiki software's preference for that capitalization). Elsewehere I generally go by one of "g", "gjm", or "gjm11". The URL listed here is for my website and blog, neither of which has been substantially updated for several years. I live near Cambridge (UK) and work for Hewlett-Packard (who acquired the company that acquired what remained of the small company I used to work for, after they were acquired by someone else). My business cards say "mathematician" but in practice my work is a mixture of simulation, data analysis, algorithm design, software development, problem-solving, and whatever random engineering no one else is doing. I am married and have a daughter born in mid-2006. The best way to contact me is by email: firstname dot lastname at pobox dot com. I am happy to be emailed out of the blue by interesting people. If you are an LW regular you are probably an interesting person in the relevant sense even if you think you aren't.

If you're wondering why some of my very old posts and comments are at surprisingly negative scores, it's because for some time I was the favourite target of old-LW's resident neoreactionary troll, sockpuppeteer and mass-downvoter.

Posts

Sorted by New

Wikitag Contributions

Comments

Sorted by
Newest
You’re probably overestimating how well you understand Dunning-Kruger
gjm5d4728

I've no idea, but I think you should collaborate with someone named Duenning to find out.

Reply312
Why you should eat meat - even if you hate factory farming
gjm9d166

Wait, there's something very strange about those two claims. 75% got more than half their calories from animals. 15% got more than half their calories from not-animals. So what did the other 10% do? (Exactly 50% from each, obviously :-).)

Reply
Accelerando as a "Slow, Reasonably Nice Takeoff" Story
gjm11d207

You write "you can buy it here" but there is no link.

However, you can do better: the whole thing is available for free online. https://www.antipope.org/charlie/blog-static/fiction/accelerando/accelerando-intro.html on the author's website is a good place to start; it has links to various versions of the book and also a little bit of explanatory material.

Reply
Ontological Cluelessness
gjm11d62

Nitpick: This post alternates between using "ontological cluelessness" to mean a certain sort of state of ignorance ("Ontological cluenessness is a state of knowledge that humans could be in, in which they haven't yet discovered ..."), and using it to denote the belief that we are in that state of ignorance. ("Ontological cluelessness is distinct from radical skepticism, pyrrhonism, and mysticism".) I think it would be better not to do that.

Reply11
The Only Option Left
gjm11d1-1

I think it's probably true that most political violence done to rightists is done by other rightists (but I haven't attempted to check; surely there must be some statistics out there) and that it was dead wrong for a bunch of right-leaning commentators to assume that Kirk's killer was an extreme leftist. (I think it would have been dead wrong for them to do so even if, say, 70% of political violence done to rightists were done by leftists.) I was just pointing out that it isn't enough just to point to statistics saying that a large majority of political-violence-as-a-whole is done by rightists.

(Gruesome example: in the ongoing Gaza horror, the IDF has killed a lot more people than Hamas, but if you find a violently-dead Israeli soldier there then "Hamas did it" is a better hypothesis than "the IDF did it".)

Reply
The Only Option Left
gjm12d107

I think there's something a bit off about your discussion of the right's leap-to-conclusions about the guy who murdered Charlie Kirk. I do basically agree with you -- they were absolutely not entitled to talk as if they knew he was some sort of radical leftist before any evidence was in -- but I don't think I buy that the relevant base rate is the fraction of all domestic terrorists or all extremist murderers, because your typical right-wing domestic terrorist isn't trying to murder a right-wing propagandist like Kirk, he's trying to blow up a mosque or an abortion clinic or murder a Democratic congressperson.

The appropriate prior for "guy who shot Charlie Kirk is a radical leftist" certainly wasn't anything like the 99% that might have justified the immediate see-how-the-left-persecutes-us rhetoric from people on the right. But it also wasn't the 5% or whatever you would guess just from the fraction of domestic terrorism that's perpetrated by leftists.

(My understanding of the current state of the evidence is that the killer wasn't in any useful sense a radical leftist, but that his motives for murdering Kirk were more left than right. On the other hand, I suspect that his more-right-than-left background helps to explain why the course of action he chose was shooting Kirk rather than anything else.)

Reply
The Only Option Left
gjm12d74

My political leanings are similar to yours and I agree that the Trump administration is a disaster, and that many people around here seem not to have predicted this, and that its disastrousness gives some evidence for Team Left over Team Right (which suits me fine since Team Left is my team). But I have some concerns about your presentation here, and in particular about how you set forth your claim that Team Left's doing better at making correct predictions than Team Right.

First, looking backward, your "ledger" of implicit predictions. (1) I am not sure you get to ascribe predictions to people who didn't actually make those predictions. (2) These look cherry-picked, in that you've picked specific bad things that turned out to happen and complained that people didn't expect them, without showing predictions of those specific bad things from your side of the table. I think if you'd asked me about several of those things before Trump took office I'd have said "I wouldn't put anything past those guys but I don't know whether they'll do that in particular" rather than "yes, a Trump administration will definitely do that".

And then you say "look at those probabilities". Those probabilities which ... you just made up? If you'd picked smaller numbers, would the case against some LWers' past complacency magically be much stronger?

Second, looking forward, your "forward predictions" are unfortunately not such as to provide much evidence for or against your position whatever happens. The odds ratio for "David is right about everything" versus "David is wrong about everything" is only .65/.35 .6/.4 .6/.4 .75/.25 .55/.45 or about 15:1, and the biggest factor there is from your prediction that the Deputy Director of the FBI will not be replaced, within a year, by someone with a better background; what do the base rates look like on that? (My guess: much the same as your prediction.)

At least one of them doesn't seem like it's "about" the relevant political issues. "At least 2 additional states formalize independent public-health compacts beyond current ones": if this happens, as you predict with probability 0.6, I don't see how that's evidence that the Trump administration is worse than if it didn't (it's a matter of how people who are very much not the Trump administration behave), nor evidence that leftish political views are more accurate than rightish ones (it's maybe evidence that the Trump administration's treatment of healthcare is disastrous, but I doubt many people around here would dispute that anyway).

Reply
davekasten's Shortform
gjm14d250

USCIS says this does not apply to existing H1-B visas, only to new applications: https://www.uscis.gov/sites/default/files/document/memos/H1B_Proc_Memo_FINAL.pdf

(I am not a lawyer nor a spokesperson for the US government and cannot advise on how likely it is that they will somehow backtrack on this.)

Reply
Hyperbolic model fits METR capabilities estimate worse than exponential model
gjm1mo20

I agree. (Or perhaps we should plot in log space when measuring/minimizing errors in log space, and plot in linear space when measuring/minimizing errors in linear space. The former is nearer to the Right Thing here than the latter.)

There are several things I find unsatisfactory in my own analysis here and either I will write a followup or Valentin2026 and I will write one together. (We're still discussing.)

Reply
Models vs beliefs
gjm1mo20

Related: Philosopher Peter van Inwagen on the way that one philosopher can disagree with another while accepting that the other guy is smarter and better informed. (Which he points out both because he finds it interesting in itself, and to argue that there's something inconsistent in complaining about religious believers -- of whom he is one -- believing things when the balance of the evidence is against them.)

Reply
Load More
Crux
3 years ago
(+27/-22)
201Hyperbolic model fits METR capabilities estimate worse than exponential model
2mo
9
133"AI achieves silver-medal standard solving International Mathematical Olympiad problems"
1y
38
21Humans, chimpanzees and other animals
2y
18
68On "aiming for convergence on truth"
2y
55
101Large language models learn to represent the world
Ω
3y
Ω
20
50Suspiciously balanced evidence
6y
24
8"Future of Go" summit with AlphaGo
8y
3
63Buying happiness
9y
34
30AlphaGo versus Lee Sedol
10y
183
8[LINK] "The current state of machine intelligence"
10y
3
Load More