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 in about the last four years. I live near Cambridge (UK) and work for a small technology company in Cambridge. 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 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.

gjm's Comments

ialdabaoth is banned

Maybe I'm confused about what you mean by "the personal stuff". My impression is that what I would consider "the personal stuff" is central to why ialdabaoth is considered to pose an epistemic threat: he has (allegedly) a history of manipulation which makes it more likely that any given thing he writes is intended to deceive or manipulate. Which is why jimrandomh said:

The problem is, I think this post may contain a subtle trap, and that understanding its author, and what he was trying to do with this post, might actually be key to understanding what the trap is.

and, by way of explanation of why "this post may contain a subtle trap", a paragraph including this:

So he created narratives to explain why those conversations were so confusing, why he wouldn't follow the advice, and why the people trying to help him were actually wronging him, and therefore indebted. This post is one such narrative.

Unless I'm confused, (1) this is not "a somewhat standard LW critique of a LW post" because most such critiques don't allege that the thing critiqued is likely to contain subtle malignly-motivated traps, and (2) the reason for taking it seriously is "the personal stuff".

Who's saying, in what sense, that "the personal stuff were separate"?

ialdabaoth is banned

Why would it make sense to "exclude the personal stuff"? Isn't the personal stuff the point here?

Bayesian examination

Sure, 2 knows something 1 doesn't; e.g., 2 knows more about how unlikely B is. But, equally, 1 knows something 2 doesn't; e.g., 1 knows more than 2 about how unlikely C is.

In the absence of any reason to think one of these is more important than the other, it seems reasonable to think that different probability assignments among the various wrong answers are equally meritorious and should result in equal scores.

... Having said that, here's an argument (which I'm not sure I believe) for favouring more-balanced probability assignments to the wrong answers. We never really know that the right answer is 100:0:0:0. We could, conceivably, be wrong. And, by hypothesis, we don't know of any relevant differences between the "wrong" answers. So we should see all the wrong answers as equally improbable but not quite certainly wrong. And if, deep down, we believe in something like the log scoring rule, then we should notice that a candidate who assigns a super-low probability to one of those "wrong" answers is going to do super-badly in the very unlikely case that it's actually right after all.

So, suppose we believe in the log scoring rule, and we think the correct answer is the first one. But we admit a tiny probability h for each of the others being right. Then a candidate who gives probabilities a,b,c,d has an expected score of (1-3h) log a + h (log b + log c + log d). Suppose one candidate says 0.49,0.49,0.01,0.01 and the other says 0.4,0.2,0.2,0.2; then we will prefer the second over the first if h is bigger than about 0.0356. In a typical educational context that's unlikely so we should prefer the first candidate. Now suppose one says 0.49,0.49,0.01,0.01 and the other says 0.49,0.25,0.25,0.01; we should always prefer the second candidate.

None of this means that the Brier score is the right way to prefer the second candidate over the first; it clearly isn't, and if h is small enough then of course the correction to the naive log score is also very small, provided candidates' probability assignments are bounded away from zero.

In practice, hopefully h is extremely small. And some wrong answers will be wronger than others and we don't want to reward candidates for not noticing that, but we probably also don't want the extra pain of figuring out just how badly wrong all the wrong answers are, and that is my main reason for thinking it's better to use a scoring rule that doesn't care what probabilities candidates assigned to the wrong answers.

Why the tails come apart

I can do n=1 (the probability is 1, obviously) and n=2 (the probability is , not so obviously). n=3 and up seem harder, and my pattern-spotting skills are not sufficient to intuit the general case from those two :-).

Changing main content font to Valkyrie?

It renders pretty decently (though I don't love the actual letterforms, at least not at sizes appropriate for body text) for me both on my laptop, which has a high-DPI display (4k at 15"), and on my desktop machine at work, which has a much lower-DPI display (2560x1600 at 30"). It is a bit spindly on the high-DPI display, but not so badly so as to make the text hard to read at any sensible size. (I've checked these things on both Firefox and Chrome.)


Pedantic note: you should do

#define ifnot(x) if (!(x))

(with extra parens around the second instance of "x") because otherwise you will get probably-unexpected results if x is something like a==b.


I remain unconvinced that Lisp's macro facilities are antimemetic. I am not sure exactly how to divide up my disagreement between "it's not antimemetic, it's something more mundane" and "I don't think you should use the exotic-sounding term 'antimeme' for something so mundane" because I am not exactly certain where you draw the boundaries of the term "antimeme".

At any rate, I think the main reasons why Lisp isn't more popular are fairly mundane and don't need an exotic-sounding term to describe them. I think (though I confess I don't have solid evidence) that:

  • most programmers who don't use Lisp never even really look at it;
  • most who look at it but still don't use it are mostly put off by superficial things like the unusual syntax or the Lisp community's reputation for smugness;
  • most who get past those things but still don't use it are mostly put off by genuine drawbacks like the relatively poor selection of libraries available or the difficulty of finding good Lispers.

So if Lisp is tragically underused then some of the tragic underuse will be the result of defmacro (and also, I suggest, set-macro-character and set-dispatch-macro-character) being underappreciated, but I find it very hard to believe that it's a large fraction. And even that doesn't all qualify as antimemetic in the sense of "provoking a self-suppressing response", again unless you're meaning that so broadly that it covers everything whose merits are easy to underestimate. If someone is (say) used to C-preprocessor-style macros, and hears that expert Lisp programming makes a lot of use of macros, and imagines a codebase full of #defines, are you calling that antimemetic? Or if they don't jump to that conclusion but do think something like "hmm, macros aren't really all that powerful, so a language where macros provide a lot of the power must be pretty weak", are you calling that antimemetic?

I think the mental process you have in mind is different from both of those, because you're taking "macro" to include e.g. C's "if" and "for", even though those aren't implemented with any sort of macrology in C, so the reaction you're describing is more like "these people say that in Lisp you can redefine the language itself, but that doesn't make any sense". Again I have no concrete evidence, but I don't think that's a common reaction. After all, C does have macros (of a sort) and you can use them to extend the language (kinda), and users of Ruby, which is at least mainstream-ish, are used to using and making what they call "domain-specific languages" within Ruby even though what you can do there is fairly restrictive compared with Lisp's macros.

I think you're overdramatizing how people react to Lisp. "Lisp is an antimeme!" makes a great story, but I don't think it fits the evidence as well as "Lisp is unfamiliar in various ways, and people are bad at seeing the merits of unfamiliar things".

... Again, unless that or something like it is actually all you mean by "antimeme".

It seems, at least some of the time, as if you're using the term to describe anything whose widespread appreciation is limited by the fact that widespread beliefs or habits of mind get in the way. (Assuming that programming languages can't have adjustable syntax; seeing a clear boundary between one's self and the rest of the universe; being willing to have a boss and do what they say.) With that definition, I'm still not sure Lisp counts (as I said above, I think there are other reasons why many programmers don't use Lisp, and they aren't all antimemetic even in this very broad sense), but there's certainly a case to be made that it does -- but to me, "antimeme" doesn't seem like an appropriate term, for two reasons.

  • It implies that the offputting-ness lies primarily in the thing itself, when in fact it seems much better to see it as a property of the people being put off.
  • It makes "being an antimeme" sound like some exotic SCP-ish property, when in fact a large fraction of underappreciated things are "antimemes" in this sense.
Is Rationalist Self-Improvement Real?

I don't know what the actual causal story is here, but it's at any rate not obviously right that if doctors were good at it then there'd be no reason to increase the age, for a few reasons.

  • Changing the age doesn't say anything about who's how good at what, it says that something has changed.
  • What's changed could be that doctors have got worse at breast cancer diagnosis, or that we've suddenly discovered that they're bad. But it could also be, for instance:
    • That patients have become more anxious and therefore (1) more harmed directly by a false-positive result and (2) more likely to push their doctors for further procedures that would in expectation be bad for them.
    • That we've got better or worse, or discovered we're better or worse than we thought, at treating certain kinds of cancers at certain stages, in a way that changes the cost/benefit analysis around finding things earlier.
      • E.g., I've heard it said (but I don't remember by whom and it might be wrong, so this is not health advice) that the benefits of catching cancers early are smaller than they used to be thought to be, because actually the reason why earlier-caught cancers kill you less is that ones you catch when they're smaller are more likely to be slower-growing ones that were less likely to kill you whenever you caught them; if that's true and a recent discovery then it would suggest reducing the amount of screening you do.
    • That previous protocols were designed without sufficient attention to the downsides of testing.

This article gives three examples of alleged antimemes: Lisp, entrepreneurship, and "stream entry". What it doesn't give is any good reason to think that they are antimemes. (I take it "antimeme" here means something stronger than "thing whose advantages aren't very obvious" or "thing that's unusual".)

I think Lisp is widely ignored not (only?) because it's an antimeme but because

  • languages that happen already to be widely used are easier to discover, have more tutorial materials around, have a better selection of libraries for useful or fun things, are likelier to help you get a job, are easier to hire other people for, etc., etc., etc. So once a language is niche-y it's likely to stay niche-y unless some unusual thing happens to propel it into the limelight.
  • other languages these days offer a lot of the advantages that used to be distinctively Lispy. For instance, exploratory programming is much easier if you have a REPL, a decent selection of built-in types, decent notation for objects of those types, and either dynamic typing or good type inference; this makes Lisp much better for that sort of thing than Fortran or C, but doesn't distinguish Lisp from Python or Javascript.

One other thing that maybe helps Lisp stay relatively obscure, its unusual syntax, does have something of the antimeme about it. At first glance the syntax looks horrible, and that makes it hard to see much else about the language until you've put in a little effort.

I don't think entrepreneurship is widely ignored at all; I think it's very widely admired. Still, most people aren't entrepreneurs; I think that's just plain ordinary risk aversion. (And maybe recognition that successful entrepreneurship requires skills and personality traits that not everyone has.) In any case, entrepreneurship is very far from being invisible in our culture.

(I don't know enough about Buddhism, meditation, altered states of consciousness, etc., to have any useful opinion about whether "stream entry" is an antimeme.)

So what allegedly makes these things antimemes, in lsusr's view? I'll hazard a guess: they are things whose merits seem clear to lsusr but that are widely neglected by others. I suggest, though, that this characteristic doesn't pick out something special about those things. People are just really bad at seeing the merits of things they aren't already in favour of. I'll give a few examples.

  • Presumably some religious or irreligious view is correct. (Like most people around here I think what one might call "atheistic scientific naturalism" is the best candidate, but my point here doesn't change if it turns out to be something quite different.) Call the correct view V. Whatever it turns out to be, the great majority of people are not adherents of V. If V is any of the more likely candidates, the great majority of people have at least had a reasonable chance to be exposed to V. Is V an antimeme? No, people just suck at considering alternative worldviews.
  • Some people think that it's a good idea to live super-frugally in your twenties and thirties, so as to build up financial reserves that let you retire very early (which doesn't necessarily mean never doing paid work again, but means not needing to). Very few people actually do this. Perhaps it's just a bad idea for one reason or another (e.g., maybe you can only actually do it effectively if you're unusually well paid and get lucky with the stock market), but if in fact it's a good idea, is it an antimeme? No, people just suck at considering alternative lifestyles.
  • Lots of people use computers running Windows. Quite a lot use computers running macOS. Some use computers running Linux. Most of these people will tell you that their preferred system is Just Better, at least for their needs. I bet a lot of them are wrong, one way or another. If so, are all these OSes antimemes? No, people just suck at considering alternatives.
Long Bets by Confidence Level

1. The calculation here seems to consider only

  • what would happen to your money if you don't make the bet
  • what would happen to your and your counterparty's money if you do and you win

but (1) if you don't place the bet then presumably your counterparty will do something with that money (which might have value or disvalue from your perspective) and (2) if you place the bet and lose then all the money goes to your counterparty's chosen charity (which almost certainly will have value or disvalue from your perspective).

Unless your expectation is that your counterparty's chosen charity has negligible effectiveness (for good or bad) relative to yours, it seems to me that this calculation is unlikely to be the one you actually want to do.

2. My impression (which could be very wrong) is that making a Long Bet is usually at least as much about raising publicity as it is about directing money where you'd like it to go. It's a thing pairs of people do when they want to get everyone talking about the issue about which they disagree. If I'm right about this, then in many cases the final disbursement of money is likely to matter less than the consciousness-raising.

Load More