This post is inspired by recent discussion about Nick Bostrom but it is not about Nick Bostrom. It is about EA in general.
It's clear that many EAs believe that there are population-level differences in average intelligence between ethnic groups. Some representative comments from recent discussions:
"IQ distributions differ across races. IQ proxies [are] a general factor of intelligence. Afaik intelligence researchers believe both sentences are correct. I care more about being correct than about being culturally compliant [...]" (1)
"The finding that generally many groups of black people have on average lower IQ than than the median is pretty robust, and he does make it clear even in the initial emailthat [sic] he doesn't think they are morally inferior or that he doesn't hold the actual negative views about them that one might associate with this language." (2)
"[...] the focus on labeling people who lean towards there being a genetic difference in population means as bad is mistaken given that the threat is actually people who try to leverage this claimed difference politically or attempt to inject their belief in this difference into as many conversations as possible." (3)
Some EAs draw a distinction between thinking black people are stupider on average than white people, and thinking they are worth less:
"[...] Bostrom makes clear that differences in intelligence do not alter the moral value or human dignity of each person.
"For him, as for many, the issues of intelligence and moral worth are distinct; he never claims that black people are worth less, you are ascribing your own notion that IQ=moral worth, and then blaming him for not responding to it." (4)
I am not going to relitigate the question of whether population-level differences in IQ exist between races. My own view is the evidence for the existence of such differences is weak and entirely unconvincing. I do not believe such differences exist. I am uninterested in debating this, however: it is in fact irrelevant to my post. My interest is in the fact that, regardless of whether such differences exist in actuality, a subset of EAs clearly believe in them and think they are important.
This has profound implications for the ability of the movement to work in the interests of all humanity. For example, EAs who believe it is important that the human population expand but who think that African people are likely to be stupider than the human average may divert resources away from them. EAs making hiring decisions may prioritise applications from people whom they have not subconsciously assigned to the schema "stupider than me."
These risks are especially salient for a movement like EA, which prides itself on its rationality and adherence to rigorous logical thinking. Belief that members of certain groups are likely to be less capable of this type of thinking is likely have a pernicious effect upon the good work EA is trying to do.
I suggest that:
- EA leaders (executives and board members of EVF, GWWC, FHI, etc.) should publicly share whether they believe that there are population-level differences in intelligence between ethnic groups.
This will allow potential donors to make informed decisions about whether EA work is something they wish to support.
[note to moderators: please do not tag this Nick Bostrom; as noted above the post is not about him or his email, but rather about the wider issues his email highlights]
It is not clear that many EAs believe this, unless you can point to a representative survey that shows otherwise. I would suggest changing 'many EAs' to 'some commenters on the EA Forum'.
Most of the linked commenters do not even have an extensive comment history on this forum. Not to say these people aren't EA'ers (I don't think the email drama has reached the mainstream yet), but if the issue keeps getting attention, there is a real possibility of increased brigading from outsiders. Neo-nazis and scientific racists do this all the time.
I agree a representative survey of EAs would be useful data. In its absence, this survey is (I believe) a reasonable proxy, showing the popularity of "race realism" beliefs among rationalists:
https://mobile.twitter.com/IneffectiveAlt4/status/1613821366318338049
I'm not convinced that is a reasonable proxy. There seems to be a big difference between 'believe that races differ genetically in socially relevant ways' and 'believe that there are population-level differences in average intelligence between ethnic groups'. The latter statement is much more specific.
For example, I would agree with the first statement on the grounds that certain genetic diseases are more prevalent in some ethnic groups than others (e.g. cystic fibrosis is higher in white people, sickle-cell disease is more prevalent in people of African and African-Caribbean origin), and this is socially relevant for how healthcare resources are used. This seems to be scientific consensus - and indeed I think I first learnt about these differences as a biology student, either in high school or university.
The survey question wasn't alluding to cystic fibrosis and it's disingenuous to pretend otherwise! You and I both know this!
I'm not going to respond further, I don't think this conversation is productive.
It is commonly theorized that having friends who hold a viewpoint should make one more charitable to that viewpoint. This has not been the case with HBD. I have a close friend of around decade who has gotten increasingly obsessed with HBD. In general they are a smart and friendly person. But the things they have started espousing have become really shocking.
Example of their beliefs: If Black people are not heavily under-represented in a 'cognitively demanding' organization that is very strong evidence the organization is racist against White and Asian individuals!
Obviously this point of view is completely at odds with any sort of fair and inclusive community or organization. They have also moved further and further rightwing. This resulted in a lot of personal problems when I came out as trans. They don't 'just' have some abstract objections, they were quite toxic to an old and supportive friend when she was having a hard time. They explicitly admit that a huge driving force for them moving rightward in general is belief in HBD. The logic for why is not hard to see. If you believe in HBD you can start to feel 'persecuted' by people on the left or center-left. It's easy to start sympathizing with the right and far right.
I've been in the rationalist community for over a decade and the EA community for a somewhat shorter period. I have seen tons of seemingly kind and reasonable friends become increasingly far-right after they got into HBD. Im honestly not surprised FLI was considering funding an explicit far-right nazi-adjacent group. The sympathies run deep. Neo-reaction has been close to the EA and rationalist communities for a very large fraction of our history.
It would be extremely hypocritical for me to hold people to views they no longer support. I endorsed HBD in 2015 and 2016. Like many rationalists I was introduced to HBD by reading Slatestarcodex. Promoting HBD in anyway, including privately exposing people to the ideas, is one of the biggest regrets of my life. It is a seriously harmful philosophy. Im very, very sorry for any negative impact my actions may have caused. For obvious reasons I have sympathy for people who have gotten into the racist pipeline. I honestly only got out because the right is so shitty to trans people and is pretty anti-vegan. Like many eggs I had a lot of trans friends. Independently I was quite convinced veganism was a positive lifestyle. them being so shitty on slam dunk issues like trans rights made me start rethinking other parts of the ideology I had started adopting. HBD is a very harmful pseudo-science and it is totally unacceptable that people with power in Effective Altruism believe in it.
I truly hope the EA movement can move toward a better future free from this toxicity.
>If Black people are not heavily under-represented in a 'cognitively demanding' organization that is very strong evidence the organization is racist against White and Asian individuals!
Any way of measuring cognitive ability shows that blacks have lower cognitive ability on average than whites and asians. Not only that, they have a lower variance in IQ, meaning fewer people 3 standard deviations above the black mean than whites/asians 3 standard eviations above the white/asian means respectively.
Any sufficiently large and non-niche organization (e.g. not some black focused charity or anything else that may legitimately have some metirocratic basis for favoring black applicants, or is located somewhere where there's mostly black people in the local labor market) that hires people at least significantly on the basis of cognitive ability and does not consider race should hire far fewer black people per capita than whites or asians. This has to be the case.
If this isn't the case, then this organization simply then factors other than cognitive ability must be dominating the hiring criteria, especially if we average this out over a whole sector, type of organization or the whole economy to smooth out any idiosyncrasies.
And we know for an indisputable fact that A) Universities engage in affirmative action on the basis of race which results in black applicants being admitted to selective colleges with academic scores below that of other races who get admitted and B) Black college graduates at all levels of qualification have significantly lower IQ, literacy and numeracy than graduates of other races at the same qualification level. So yes, its trivially true that institutions discriminate against whites and asians on the basis of race. You may think this is a good thing, but if you deny it exists then you are objectively incorrect.
>Obviously this point of view is completely at odds with any sort of fair and inclusive community or organization.
Hiring people on the basis of their cognitive ability is necessary NOT inclusive. Hiring the person who is most intellectually capable is necessarily discriminatory, even if we're talking about picking between people of different races.
As for fair, what is "fair"? Who decides what that is? Is a high IQ asian applicant missing out on a position at harvard to a lower IQ black applicant on the basis of their race "fair"?
I think this is highly unfair. But perhaps even more importantly, this will lead to worse social outcomes because there is a strong correlation between job performance and IQ.
Anything other than picking people at random is going to be not "inclusive", but doing so would make the maintenance of a modern economy impossible.
Thanks for this perspective, sapphire. I'm glad you managed to get out. I honestly am not hopeful about the prospects for a wider detangling of EA and white supremacy but I will keep my fingers crossed.
I have written a post that probably answer some of those concerns:
https://forum.effectivealtruism.org/posts/r3f45au7ewEkypygf/wokism-rethinking-priorities-and-the-bostrom-case
Across large racial groups, however, the observed differences are statistically significant but modest (as they are for other anthropometric characters), and given the social and cultural disparities across racial groups and the social plasticity of IQ the share of biological causality on the observed racial IQ gap is a difficult and controversial research problem. Bostrom has not scientific legitimacy to give an opinion in such a technical and disputed issue.
Regarding feelings and human sympathy, the big question is: whose feelings? How can a modest difference in the mean of the IQ distribution offend a black man with an IQ of 130? And if you are a white person with an IQ of 85, how useful is for you the white black IQ gap for which you make a negative contribution? IQ is a personal characteristic, and means are completely irrelevant for every single person.
If we believe that IQ is important for some application, we can measure it directly at the individual level instead of looking at skin color. Anti-racist arguments do not depend on the distribution of psychometric variables. They depend on treating each person as an individual, ignoring skin color, and exclusively considering the abilities and limitations relevant to the case considered of the person under scrutiny. Any “anti-racism” in excess of this, is, in fact, racism
>Across large racial groups, however, the observed differences are statistically significant but modest (as they are for other anthropometric characters)
The balck white IQ gap is approximately one standard deviation, which is by no means modest. If we could raise a population's mean IQ by one standard deviation, this would be utterly transformative in a way that almost no other intervention could be.
I'm not saying it's not modest because its not a small number - it's not modest because such a gap has profound social implications in a way that black people being slightly taller than asian people doesn't. Controlling for IQ literally eliminates most of the inequality between races in things like income.
And it's not just a difference in the mean. Black people have a lower variance in IQ, meaning there are less black people 3 standard deviations above the black mean than there are asian people 3 standards above the asian mean. So not only are black cognitive elites less intelligent than e.g. Asian cognitive elites on average (because +3σ from a lower mean is a lower number than +3σ from a higher mean), they're also less common. The further away we move above the mean, the greater the intelligence discrepencies between races become.
>and given the social and cultural disparities across racial groups
Cultural explanations have consistently failed to account for IQ gaps, either in the form of test bias or genuine intellectual ability and its development. And understandably so: Are Ashkenazi jews much more intelligent than asian on average because asian culture isn't intellectual enough?
And again for social disparities. A full two thirds (!) of the black white IQ gap exists in children before they even begin school (with the rest increasing in line with the general increase in the heritabilty of traits with age), and adoption studies show a much greater correlation between biological parental IQ than adopted parental IQ, largely eliminating the possibility that home environment explains a majority of IQ gaps.
>and the social plasticity of IQ the share of biological causality
No such plasticity exists.
Twin studies conclusively show that IQ is highly heritable. When test reliability is accounted for, this heritability may be as high in countries like the US as 80%.
Maybe it's different between races, but the burden of proof is on you to show that this general rule for heritability doesn't apply in this specific case. And all main environmental explanations have studied and convincingly argued against.
>is a difficult and controversial research problem.
It's controversial for ideological, not scientific reasons. As for difficulty, perhaps. But breakthroughs in molecular genetic studies of intelligence have not only identified many of the genes associated with cognitive ability, but it turns out the alleles are unequally distributed between races (and in a way that is congruent with observed IQ differences): https://www.mdpi.com/2624-8611/1/1/34/htm
>Bostrom has not scientific legitimacy to give an opinion in such a technical and disputed issue.
This is a textbook, slamdunk case of 'isolated demands for rigor'. Such objections are never, and I truly mean never, raised by people like yourself against people denying the existence of heritable differences in intelligence between races. Most people who object to the existence of such differences are not only not scientists, but don't even have a rudimentary understanding of the literature on the topic. And sorry to say, but this includes yourself. Cultural explanations for IQ differences lack any support in the literature and this has been the case for decades now.
If virtually anyone is allowed to deny these differences exist, I see no reason why Bostrom cannot similarly express (an almost certainly more well informed than average) opinion on the topic too.
>Regarding feelings and human sympathy, the big question is: whose feelings? How can a modest difference in the mean of the IQ distribution offend a black man with an IQ of 130? And if you are a white person with an IQ of 85, how useful is for you the white black IQ gap for which you make a negative contribution? IQ is a personal characteristic, and means are completely irrelevant for every single person.
Groups differences in intelligence explain group differences in social outcomes - that's why this matters.
Racially conscious black people aren't angry just because they personally have less e.g. money than white people. They think black people generally having less money than white people generally is a grave social injustice. And considering these differences are explained by things like group intelligence differences, that's why said differences are relevant and important. And it also explains why we should never expect African countries as a whole to achieve anywhere near the same levels of economic development pre-singularity as western or North-East Asian countries, even if plenty of high IQ Africans can be successful.
If everybody was as politically individualistic as you're implying, there largely wouldn't be any problem. But that's not the case.
>If we believe that IQ is important for some application, we can measure it directly at the individual level instead of looking at skin color. Anti-racist arguments do not depend on the distribution of psychometric variables.
Yes, they do!
All of anti-racist thought today as practiced by anyone with any kind of institutional legitimacy and/or power is dependant on biological racial egalitarianism. They believe that without "racism", whatever they happen to mean by that, the races would have equal outcomes because they're biologically equivalent in intellectual ability.
>They depend on treating each person as an individual, ignoring skin color, and exclusively considering the abilities and limitations relevant to the case considered of the person under scrutiny. Any “anti-racism” in excess of this, is, in fact, racism
This is a wonderfully admirable belief to have, but unfortunately it is an extreme minority position institutionally. Hence why affirmative action exists.
>This has profound implications for the ability of the movement to work in the interests of all humanity. For example, EAs who believe it is important that the human population expand but who think that African people are likely to be stupider than the human average may divert resources away from them. EAs making hiring decisions may prioritise applications from people whom they have not subconsciously assigned to the schema "stupider than me."
Let's say that it is true that not only are africans today lower in cognitive ability than whites on average, but that this gap is also mostly heritable (the latter is debatable but there is strong evidence to support it - the former is indisputable unless you think literally every single measure of cognitive ability that has strong predictive validity is radically and fatally flawed) .
Should this be acknoledged? Should this affect how we view the world?
Assuming it's true, do you believe, as a rationalist, that ignoring fundmental characteristics of human ity that explain more about society than almost any other factors will lead you towards making better decisions?
Speaking only for myself here, I think the answer is an unqualified hell no.
If you assume African Americans are congitive equivlanet to Asian Americans on average, then you will assume that sending more of them to college (in excess of those who can get in on the basis of academic merit alone currently) will result in improved socioeconomic outcomes (even if no affirmative action in hiring existed). But the real result is that many black people will have wasted 2-4 years of their life accumulating debt and forgoing earnings to gain an "education" that did nothing to close the cognitive gaps between themselves and asian americans. The debt, loss of income and wasting of time that could have been used developing valuable skills will hurt these graduates.
If instead, given that we know that intelligence varies between people, and we're interested in maximising the earning potential of black kids from low income families, we had identified their IQ early on and put our effort into getting them onto the optimal career track for their specific level of cognitive ability.
(The same is true for kids of any race - the difference is people are generally much more accepting of the fact that some white kids are just plain less intelligent than other white kids and can't have the same careers, and much more of acepting income inequality within than between races. And so there is a much bigger push to get black people specifically into college rather than low income/IQ kids generally, which is why it makes sense to think along racial lines here).
The same logic applies to entire countries. If you assume that Nigerians are no less intelligent than Chinese people, then you're likely to expect that Nigeria is going to follow the same development path as China once some threshold is passed for e.g. education health and so on. This is almost certainly not going to happen, and expecting it to will mean you will make suboptimal decisions when thinking about how to help countries in Africa.
More controversially, it also would mean that people in Africa are likely going to be of much lower instrumental value to humanity than people in Japan. But choosing to ignore this is actually the decision that is being made on the basis of race. Choosing to factor this into your worldview and decisions means you're acting on the basis of utility alone and not making decisions aimed at helping people because of their race more than you would of.
Liberal americans obviously aren't categorically opposed to helping low income black americans more than low income white americans in rural areas/the south, and it's likely this is because they think that black americans are better people and have more to offer the country, whereas white americans are just dumb and ignorant and cannot be reformed (not everyone believes this entirely, but I think some implicit form of this belief colors much of liberal thought in this area). All I'm saying is this type of thinking also applies more generally, except in this case the difference is based on data rather than intution and in-group favoratism.
If you, as you say, care about "working in the interests of humanity", you need to think about the interests of humanity collectively. Maximixing the interests of humanity will necessarily involve not helping all groups equally.