Reaction to "Apology for an Old Email"
This post is a nearly word for word draft of a comment I wrote and considered posting under this forum post. I ultimately decided against posting it because it was too long, and the tone and substance didn’t seem to be right for the comments section. It is mostly a response to Nick Bostrom’s old racist email from the extropians mailing list, as well as his subsequent apology for it. I also bring up and interact a bit with some of the comments that were on the forum post at the time that I drafted it (around 7:00 AM EST of the day it was posted).
I haven’t changed much in the response since then, as the editing process to get a post up provides a delay as is. As a result much of the discourse has moved on quite a bit, and various things about this post may feel late to the game. I don’t intend to interact much with the subsequent discussion in my intro, there have been takes both more harsh) than mine and more lenient. My own take is probably closest to Habiba Islam’s, or, more simply, Megan Nelson’s. One major point of contention seems to be how good the apology was. I’m in the “apology was bad” camp, but much of this is context more than substance so I can understand disagreement to an extent.
Maybe more noticeable is that a huge amount of the subsequent discussion has seemed to assume that Bostrom’s email was endorsing HBD 1. I make brief note of some confusion here in my comment draft, but didn’t expect this to become such a standard fixture of the narrative (again I wrote this very early on when there hadn’t been much discussion that I could see), and the number of people who disagree with me over this does reduce my confidence in my original interpretation. That said, I do stick by my own view, ultimately, that while it seems plausible from his apology that Bostrom does covertly believe in HBD, this plays no role in the argument of his initial email. To speak briefly in defense of this position, I think two things count against assuming the HBD interpretation:
It goes against the most literal interpretation of his email, and while you could insist that he has a hidden layer of meaning, I see IQ reductionism significantly more often than HBD, so if something looks like IQ reductionism, and doesn’t actually invoke HBD, I think the most reasonable default interpretation is IQ reductionism
Bostrom’s background as someone very interested in AI makes me think he is unusually susceptible to very functionalist views of intelligence, which looks at ability to meet some particular practical benchmarks as what defines “intelligence”. In reality this mixes up what we commonly mean by intelligence with things like motivation, knowledge, focus, practice etc., but it isn’t a bad definition when applied to AI, where you are dealing with a completely alien type of mind, and what seems most significant about it to discussions is functional. It is easy for me to believe that Bostrom sees environmental factors affecting IQ, and rather than concluding that IQ tests have various complications and failure modes, concludes that this just means that environment affects intelligence in basically the same ways that it affects IQ scores.
All this said, if I’m wrong here, much of my criticism of his email is invalid, and the right problems to point out are just the problems with HBD. With that out of the way, here is the take I endorsed shortly after reading the email/apology, and which I mostly maintain now:
Christ this was not fun to read. And I say this as a white EA, to freedomandutility’s point, I can’t imagine what it’s like to read as a black EA, especially when EA is so white to begin with. That said I think the reactions here often seem confused or ambiguous in some unhelpful ways, so here is an overlong distillation of my own takeaways:
The initial email was bad. From some of the reactions here, I went in thinking that what was bad about it, primarily, was that it endorses HBD or something. Nope, the email doesn’t even bring it up. Primarily there were three things about the email that were actually bad:
The use of the n word. This is actually, of the three, the one I think may be least bad about it. It isn’t alright to use now and wasn’t then, but this was the 90s, and if you made a list of all the white people who casually used the word back then (or frankly up until five years ago or so) in the “mention” sense without much thought it would horrify you. For god’s sake watch any of Quentin Tarantino’s widely popular movies from that time, it’s surreal now. This doesn’t rise to the level of “bring historical context to reading Mark Twain”, but it is hard to ignore.
The thesis his racist comments were an example for is dumb. As far as I can tell it comes down to “being edgy with your statements is good because the squares need to be scared out of their complacency with facts and logic”. No. Please just don’t do that. This is like the exact reason you shouldn’t do that. Racism is arguably at its core a result of this type of overly blunt thinking that irons out all the important details, that is what stereotypes are. If your idea of a positive example for your style of communication/thinking is “black people are more dumb than white people” you should probably step back and consider that you have found a candidate for one of the most toxic modes of thought possible, not one of the most helpful.
The conclusion. As far as I can tell the conclusion derives, invalidly, from two premises, one of which is highly controversial, the other of which is, to Tenoke’s point, just an empirical fact (which again, to titotal’s counterpoint, does not redeem the invalid conclusion):
One premise is that the average recorded score of black people on IQ tests is lower than the average recorded score of white people on IQ tests. This was true the last time I heard about it, which was in a unit of a psychology course where the professor brought up HBD in order to present all the evidence that it’s wrong, which probably tells you something about its relevance.
The other premise is that IQ is a good test for intelligence. This is highly controversial. Notably while most people think intelligence of some kind is part of what goes into it, many people think there are other important things you have to control for. Note that the glib defense of IQ tests is fallacious unless used against the very strongest criticisms of these tests: “if IQ doesn’t measure something that matters to us about intelligence, surely we wouldn’t worry about the effects of lead exposure on IQ”, a suitably controlled experiment that showed that lead exposure reliably lowered scores on vocabulary exams would be extremely worrying. Yet vocabulary exams clearly have a tremendous amount to do with cultural context, in fact that is almost all of what they measure. There are more sophisticated defenses of IQ tests, but I have little patience for the glib version. Also notably IQ scores are scores on a test, whereas “intelligence” is usually thought of as an innate disposition of some kind. There is a trivial version of this distinction I can point out, but I think on a less trivial level it drives a common category mistake (calling people “high IQ” as synonymous with “intelligent”), which does let people ignore significant things about IQ tests they pay attention to in any other exam. I’ve taken the IQ test twice, and my scores varied by 10 points between them. I don’t think the most sensible default interpretation of this is that I got nearly a standard deviation more intelligent. If you say that I did, I am suspicious that you are gerrymandering the definition of intelligence around IQ in order to rescue some special status for the IQ exam above and beyond any other test. This gerrymandering is appealing for making science more convenient, but has noxious possible consequences as the email illustrates.
The conclusion seems to derive from this latter, very most extreme interpretation of IQ scores, that they are proxies for intelligence good enough that even when apparently irrelevant things confound them, like cultural context and even day to day personal circumstances, we should simply understand that as intelligence anyway. There is no other way to get straight from average IQ differences to “more dumb”, and Bostrom makes no other stops along the way. He probably understands this in his calmer moments, I would hope, but again the point of the email wasn’t to get at the truth, it was to defend the value of bold edgy statements, and the relevant nuances are just so much less convenient for that.
I honestly think the apology might have been worse. The literal comments and claims in it are less bad, but the email was a 30 year old brief thought from an edgy college kid writing on a niche mailing list. The comments might not be excusable, but the lack of care he put into them was more so. The apology was meant to be a very deliberate public statement, specifically expressing regret for past carelessness, he cannot invoke context as an excuse for any carelessness he put into the apology.
And…look, this apology opens by emphasizing that he wishes people would just let the mailing list be forgotten, and closes by discussing the use of the word “eugenics” in academic bioethics. The apology is relatively light on apology, and heavy on deflection of possible lines of personal attack. Note that this is the only place HBD comes up at all (he expresses agnosticism, which I usually interpret as covert support, maybe that’s unfair, I don’t know), he mentioned eugenics because he knows people will attack him from that angle after this, he mentions his donation habits to signal personal concern for the people he called dumb. He does some apologizing, but it’s pretty clear that that isn’t what this write up is for.
As for what we should actually do with this, I’m not sure? Freedomandutility’s ideas seem basically alright but to be blunt it won’t stop me from reading Bostrom’s better work, or convince me he should be outright fired (maybe from positions of administrative power though). There are already plenty of people in EA I think have some noxious, disappointing view or other (notably Robin Hanson), so the effect of this is mostly about my personal opinion of Bostrom’s character, and on certain topics, his authority. I have a pretty high bar for going beyond that for people, for reasons I talk about at length in some of my earlier blog posts. Maybe I’m wrong and even more should be done, I’m open to ideas but most of them that come to mind are either the sort of public statement 2 stuff freedomandutility mentions, or things that I think just aren’t good ideas. Sometimes I think the best response to bad things happening is disappointingly light, and I think this is especially true of the class of “bad things” that “someone who strongly influenced my thinking said dumb stuff” falls into.
Since posting this I’ve realized that HBD is a more obscure phrase than I thought. So to be clear: “HBD” stands for “human biodiversity”, it is a modern strain of scientific racism that claims large portions of the disparate wealth and outcomes between racial groups today are due to genetics. ↩︎
If , help us write more by donating to our Patreon.