Participant here from the beginning and from OB before that, posting under a throwaway account. And this will probably be my only comment on the race-IQ issue here.
I don’t think the view that there are genetic racial differences in IQ is popular here, if that’s what you’re referring to. It’s come up a few times and the consensus seems to be that the evidence points to cultural and environmental explanations for the racial IQ gap [emphasis mine].
The vast majority of writers here have not given their opinion on the topic. Many people here write under their real name or under a name that can be matched to their real name by spending a half hour with Google. In the U.S. (the only society I really know) this is not the kind of opinion you can put under your real name without significant risk of losing one’s job or losing out to the competition in a job application, dating situation or such.
Second, one of the main reasons Less Wrong was set up is as a recruiting tool for SIAI. (The other is to increase the rationality of the general population.) Most of the people here with a good reputation are either affiliated with SIAI or would like to keep open the option of starting an affiliation some day. (I certainly do.) Since SIAI’s selection process includes looking at the applicant’s posting history here, even writers whose user names cannot be correlated with the name they would put on a job application will tend to avoid taking the unpopular-with-SIAI side in the race-IQ debate.
So, want to start a debate that will leave your side with complete control of the battlefield? Post about the race-IQ issue on Less Wrong rather than one of the web sites set up to discuss the topic!
Downvoted for not even giving your opinion on the issue even with your throwaway account.
Some have pointed out that cultural and environmental explanations can account for significant IQ differences. This is true.
It doesn’t follow that there aren’t racial difference based on genetics as well. In fact, the idea that there might NOT be is quite absurd. Of course there are. The only question is how large they are.
“It doesn’t follow that there aren’t racial difference based on genetics as well. In fact, the idea that there might NOT be is quite absurd. Of course there are. The only question is how large they are.”
And what direction they’re in. If social factors are sufficient to explain (e.g.) the black-white IQ gap, and the argument for their being some innate differences is “well, it’s exceedingly unlikely that they’re precisely the same,” we don’t have reason to rate “whites are natively more intelligent than blacks” as more likely than “blacks are natively more intelligent than whites.” (If we know that Smith is wealthier than Jones, and that Smith found a load of Spanish dubloons by chance last year, we can’t make useful conclusions about whose job was more renumerative before Smith found her pirate booty.) Of course, native racial differences might also be such that there are environmental conditions under which blacks are smarter than whites and others in which the reverse applies, or whatever.
In any event I don’t think we need to hypothesize the existence of such entities (substantial racial differences) to explain reality, so the razor applies.
Even if cultural factors are sufficient, in themselves, to explain the black-white IQ difference, it remains more probable that whites tend to have a higher IQ by reason of genetic factors, and East Asians even more so.
This should be obvious: a person’s total IQ is going to be the sum of the effects of cultural factors plus genetic factors. But “the sum is higher for whites” is more likely given the hypothesis “whites have more of an IQ contribution from genetic factors” than given the hypothesis “blacks have more of an IQ contribution from genetic factors”. Therefore, if our priors for the two were equal, which presumably they are, then after updating on the evidence, it is more likely that whites have more of a contribution to IQ from genetic factors.
I’m not sure that this is the case, given that the confound has a known direction and unknown magnitude.
Back to Smith, Jones, and Spanish treasure: let’s assume that we have an uncontroversial measure of their wealth differences just after Smith sold. (Let’s say $50,000.) We have a detailed description of the treasure Smith found, but very little market data on which to base an estimation of what she sold them for. It seems that ceteris paribus, if our uninformed estimation of the treasure is >$50,000, Jones is likelier to have a higher non-pirate gold income, and if our uninformed estimation of the treasure is <$50,000, Smith is likelier to.
Whites and blacks both have a cultural contribution to IQ. So to make your example work, we have to say that Smith and Jones both found treasure, but in unequal amounts. Let’s say that our estimate is that Smith found treasure approximately worth $50,000, and Jones found treasure approximately worth $10,000. If the difference in their wealth is exactly $50,000, then most likely Smith was richer in the first place, by approximately $10,000.
In order to say that Jones was most likely richer, the difference in their wealth would have to be under $40,000, or the difference between our estimates of the treasures found by Smith and Jones.
I agree with this reasoning, although it does not contradict my general reasoning: it is much like the fact that if you find evidence that someone was murdered (as opposed to dying an accidental death), this will increase the chances that Smith is a murderer, but then if you find very specific evidence, the chance that Smith is a murderer may go down below what it was originally.
However, notice that in order to end up saying that blacks and whites are equally likely to have a greater genetic component to their intelligence, you must say that your estimate of the average demographic difference is EXACTLY equal to the difference between your estimates of the cultural components of their average IQs. And if you say this, I will say that you wrote it on the bottom line, before you estimated the cultural components.
And if you don’t say this, you have to assert one or the other: it is more likely that whites have a greater genetic component, or it is more likely that blacks do. It is not equally likely.
And if you don’t say this, you have to assert one or the other: it is more likely that whites have a greater genetic component, or it is more likely that blacks do. It is not equally likely.
Often when people say “equally likely” they mean “I don’t know enough to credibly estimate which one is greater, the probability distributions just overlap too much.” (Yes, the ‘bottom line’ idea is more relevant here. It’s a political minefield.)
But that’s the point of my general argument: if you know that whites average a higher IQ score, but not necessarily by how much (say because you haven’t investigated), and you also know that there is a cultural component for both whites and blacks, but you don’t know how much it is for each, then you should simply say that it is more likely (but not certain) that whites have a higher genetic component.
I mean “equally likely” in wedrifid’s sense: not that, having done a proper Bayesian analysis on all evidence, I may set the probability of p(W>B)=p(B>W}=.5 (assuming intelligence works in such a way that this implied division into genetic and environmental components makes sense), but that 1) I don’t know enough about Spanish gold to make an informed judgement and 2) my rough estimate is that “I could see it going either way”—something inherent in saying that environmental differences are “sufficient to explain” extant differences. So actually forming beliefs about these relative levels is both insufficiently grounded and unnecessary.
I suppose if I had to write some median expectation it’s that they’re equal in the sense that we would regard any other two things in the phenomenal world of everyday experience equal—when you see two jars of peanut butter of the same brand and size next to each other on a shelf in the supermarket, it’s vanishingly unlikely that they have exaaaactly the same amount of peanut butter, but it’s close enough to use the word.
I don’t think this is really a case of writing things down on the bottom line. What reason would there be to suppose ex ante that these arbitrarily constructed groups differ to some more-than-jars-of-peanut-butter degree? Is there some selective pressure for intelligence that exists above the Sahara but not below it (more obvious than counter-just-so-stories we could construct?) Cet par I expect a population of chimpanzees or orangutans in one region to be peanut butter equal in intelligence to those in another region, and we have lower intraspecific SNP variation than other apes.
“I could see it going either way” is consistent with having a best estimate that goes one way rather than another.
Just as you have the Flynn effect with intelligence, so average height has also been increasing. Would you say the same thing about height, that the average height of white people and black people has no significant genetic difference, but it is basically all cultural? If not, what is the difference?
In any case, both height and intelligence are subject to sexual selection, not merely ordinary natural selection. And where you have sexual selection, one would indeed expect to find substantial differences between diverse populations: for example, it would not be at all surprising to find significantly different peacock tails among peacock populations that were separated for thousands of years. You will find these significant differences because there are so many other factors affecting sexual preference; to the degree that you have a sexual preference for smarter people, you are neglecting taller people (unless these are 100% correlated, which they are not), and to the degree that you have a sexual preference for taller people, you are neglecting smarter people. So one just-so-story would be that black people preferred taller people more (note the basketball players) and so preferred more intelligent people less. This just-so-story would be supported even more by the fact that the Japanese are even shorter, and still more intelligent.
Granted, that remains a just-so-story. But yes, I would expect “ex ante” to find significant genetic differences between races in intelligence, along with other factors like height.
The reason I did not even give my opinion on the race-IQ issue is that IMHO the expected damage to the quality of the conversation here exceeds the expected benefit.
It is possible for a writer to share the evidence that brought them to their current position on the issue without stating their position, but I do not want to do that because it is a lot of work and because there are probably already perfectly satisfactory books on the subject.
By the way, the kind of person who will discriminate against me because of my opinion on this issue will almost certainly correctly infer which side I am on from my first comment without really having to think about it.
It doesn’t follow that there aren’t racial difference based on genetics as well. In fact, the idea that there might NOT be is quite absurd. Of course there are. The only question is how large they are.
That is not the only question. The question that gets people into trouble, is “which groups are favored or disfavored”. You can’t answer that without offending some people, no matter how small you think the genetic component of the difference is, because many of the people who read it will discard or forget the magnitude entirely and look at only the sign. Saying that group X is genetically smarter than group Y by 10^-10 IQ points will, for many listeners, have the same effect as saying that X is 10^1 IQ points smarter. And while the former belief may be true, the latter belief is false, harmful to those who hold it, and harmful to uninvolved third parties. True statements about race, IQ, and genetics are very easy to simplify or round off to false, harmful and disreputable ones.
That’s why comments about race, IQ, and genetics always have to be one level separated from reality, talking about groups X and Y and people with orange eyes rather than real traits and ethnicities. And if they aren’t well-separated from reality, they have to be anonymous, to protect the author from the reputational effects of things others incorrectly believe they’ve said.
(Edited to add: See also this comment I previously wrote on the same topic, which describes a mechanism by which true beliefs about demographic differences in intelligence (not necessarily genetic ones) produce false beliefs about individual intelligence.)
It seems clear to me that much of the time when people mistakenly get offended, they’re mistaken about what sort of claim they should get offended about, not just mistaken about what claim was made.
Since SIAI’s selection process includes looking at the applicant’s posting history here, even writers whose user names cannot be correlated with the name they would put on a job application will tend to avoid taking the unpopular-with-SIAI side in the race-IQ debate.
What makes you think “the unpopular-with-SIAI side” exists? Or that it is what you think it is?
Participant here from the beginning and from OB before that, posting under a throwaway account. And this will probably be my only comment on the race-IQ issue here.
The vast majority of writers here have not given their opinion on the topic. Many people here write under their real name or under a name that can be matched to their real name by spending a half hour with Google. In the U.S. (the only society I really know) this is not the kind of opinion you can put under your real name without significant risk of losing one’s job or losing out to the competition in a job application, dating situation or such.
Second, one of the main reasons Less Wrong was set up is as a recruiting tool for SIAI. (The other is to increase the rationality of the general population.) Most of the people here with a good reputation are either affiliated with SIAI or would like to keep open the option of starting an affiliation some day. (I certainly do.) Since SIAI’s selection process includes looking at the applicant’s posting history here, even writers whose user names cannot be correlated with the name they would put on a job application will tend to avoid taking the unpopular-with-SIAI side in the race-IQ debate.
So, want to start a debate that will leave your side with complete control of the battlefield? Post about the race-IQ issue on Less Wrong rather than one of the web sites set up to discuss the topic!
Downvoted for not even giving your opinion on the issue even with your throwaway account.
Some have pointed out that cultural and environmental explanations can account for significant IQ differences. This is true.
It doesn’t follow that there aren’t racial difference based on genetics as well. In fact, the idea that there might NOT be is quite absurd. Of course there are. The only question is how large they are.
“It doesn’t follow that there aren’t racial difference based on genetics as well. In fact, the idea that there might NOT be is quite absurd. Of course there are. The only question is how large they are.”
And what direction they’re in. If social factors are sufficient to explain (e.g.) the black-white IQ gap, and the argument for their being some innate differences is “well, it’s exceedingly unlikely that they’re precisely the same,” we don’t have reason to rate “whites are natively more intelligent than blacks” as more likely than “blacks are natively more intelligent than whites.” (If we know that Smith is wealthier than Jones, and that Smith found a load of Spanish dubloons by chance last year, we can’t make useful conclusions about whose job was more renumerative before Smith found her pirate booty.) Of course, native racial differences might also be such that there are environmental conditions under which blacks are smarter than whites and others in which the reverse applies, or whatever.
In any event I don’t think we need to hypothesize the existence of such entities (substantial racial differences) to explain reality, so the razor applies.
Even if cultural factors are sufficient, in themselves, to explain the black-white IQ difference, it remains more probable that whites tend to have a higher IQ by reason of genetic factors, and East Asians even more so.
This should be obvious: a person’s total IQ is going to be the sum of the effects of cultural factors plus genetic factors. But “the sum is higher for whites” is more likely given the hypothesis “whites have more of an IQ contribution from genetic factors” than given the hypothesis “blacks have more of an IQ contribution from genetic factors”. Therefore, if our priors for the two were equal, which presumably they are, then after updating on the evidence, it is more likely that whites have more of a contribution to IQ from genetic factors.
I’m not sure that this is the case, given that the confound has a known direction and unknown magnitude.
Back to Smith, Jones, and Spanish treasure: let’s assume that we have an uncontroversial measure of their wealth differences just after Smith sold. (Let’s say $50,000.) We have a detailed description of the treasure Smith found, but very little market data on which to base an estimation of what she sold them for. It seems that ceteris paribus, if our uninformed estimation of the treasure is >$50,000, Jones is likelier to have a higher non-pirate gold income, and if our uninformed estimation of the treasure is <$50,000, Smith is likelier to.
Whites and blacks both have a cultural contribution to IQ. So to make your example work, we have to say that Smith and Jones both found treasure, but in unequal amounts. Let’s say that our estimate is that Smith found treasure approximately worth $50,000, and Jones found treasure approximately worth $10,000. If the difference in their wealth is exactly $50,000, then most likely Smith was richer in the first place, by approximately $10,000.
In order to say that Jones was most likely richer, the difference in their wealth would have to be under $40,000, or the difference between our estimates of the treasures found by Smith and Jones.
I agree with this reasoning, although it does not contradict my general reasoning: it is much like the fact that if you find evidence that someone was murdered (as opposed to dying an accidental death), this will increase the chances that Smith is a murderer, but then if you find very specific evidence, the chance that Smith is a murderer may go down below what it was originally.
However, notice that in order to end up saying that blacks and whites are equally likely to have a greater genetic component to their intelligence, you must say that your estimate of the average demographic difference is EXACTLY equal to the difference between your estimates of the cultural components of their average IQs. And if you say this, I will say that you wrote it on the bottom line, before you estimated the cultural components.
And if you don’t say this, you have to assert one or the other: it is more likely that whites have a greater genetic component, or it is more likely that blacks do. It is not equally likely.
Often when people say “equally likely” they mean “I don’t know enough to credibly estimate which one is greater, the probability distributions just overlap too much.” (Yes, the ‘bottom line’ idea is more relevant here. It’s a political minefield.)
But that’s the point of my general argument: if you know that whites average a higher IQ score, but not necessarily by how much (say because you haven’t investigated), and you also know that there is a cultural component for both whites and blacks, but you don’t know how much it is for each, then you should simply say that it is more likely (but not certain) that whites have a higher genetic component.
I agree.
I mean “equally likely” in wedrifid’s sense: not that, having done a proper Bayesian analysis on all evidence, I may set the probability of p(W>B)=p(B>W}=.5 (assuming intelligence works in such a way that this implied division into genetic and environmental components makes sense), but that 1) I don’t know enough about Spanish gold to make an informed judgement and 2) my rough estimate is that “I could see it going either way”—something inherent in saying that environmental differences are “sufficient to explain” extant differences. So actually forming beliefs about these relative levels is both insufficiently grounded and unnecessary.
I suppose if I had to write some median expectation it’s that they’re equal in the sense that we would regard any other two things in the phenomenal world of everyday experience equal—when you see two jars of peanut butter of the same brand and size next to each other on a shelf in the supermarket, it’s vanishingly unlikely that they have exaaaactly the same amount of peanut butter, but it’s close enough to use the word.
I don’t think this is really a case of writing things down on the bottom line. What reason would there be to suppose ex ante that these arbitrarily constructed groups differ to some more-than-jars-of-peanut-butter degree? Is there some selective pressure for intelligence that exists above the Sahara but not below it (more obvious than counter-just-so-stories we could construct?) Cet par I expect a population of chimpanzees or orangutans in one region to be peanut butter equal in intelligence to those in another region, and we have lower intraspecific SNP variation than other apes.
“I could see it going either way” is consistent with having a best estimate that goes one way rather than another.
Just as you have the Flynn effect with intelligence, so average height has also been increasing. Would you say the same thing about height, that the average height of white people and black people has no significant genetic difference, but it is basically all cultural? If not, what is the difference?
In any case, both height and intelligence are subject to sexual selection, not merely ordinary natural selection. And where you have sexual selection, one would indeed expect to find substantial differences between diverse populations: for example, it would not be at all surprising to find significantly different peacock tails among peacock populations that were separated for thousands of years. You will find these significant differences because there are so many other factors affecting sexual preference; to the degree that you have a sexual preference for smarter people, you are neglecting taller people (unless these are 100% correlated, which they are not), and to the degree that you have a sexual preference for taller people, you are neglecting smarter people. So one just-so-story would be that black people preferred taller people more (note the basketball players) and so preferred more intelligent people less. This just-so-story would be supported even more by the fact that the Japanese are even shorter, and still more intelligent.
Granted, that remains a just-so-story. But yes, I would expect “ex ante” to find significant genetic differences between races in intelligence, along with other factors like height.
The reason I did not even give my opinion on the race-IQ issue is that IMHO the expected damage to the quality of the conversation here exceeds the expected benefit.
It is possible for a writer to share the evidence that brought them to their current position on the issue without stating their position, but I do not want to do that because it is a lot of work and because there are probably already perfectly satisfactory books on the subject.
By the way, the kind of person who will discriminate against me because of my opinion on this issue will almost certainly correctly infer which side I am on from my first comment without really having to think about it.
That is not the only question. The question that gets people into trouble, is “which groups are favored or disfavored”. You can’t answer that without offending some people, no matter how small you think the genetic component of the difference is, because many of the people who read it will discard or forget the magnitude entirely and look at only the sign. Saying that group X is genetically smarter than group Y by 10^-10 IQ points will, for many listeners, have the same effect as saying that X is 10^1 IQ points smarter. And while the former belief may be true, the latter belief is false, harmful to those who hold it, and harmful to uninvolved third parties. True statements about race, IQ, and genetics are very easy to simplify or round off to false, harmful and disreputable ones.
That’s why comments about race, IQ, and genetics always have to be one level separated from reality, talking about groups X and Y and people with orange eyes rather than real traits and ethnicities. And if they aren’t well-separated from reality, they have to be anonymous, to protect the author from the reputational effects of things others incorrectly believe they’ve said.
(Edited to add: See also this comment I previously wrote on the same topic, which describes a mechanism by which true beliefs about demographic differences in intelligence (not necessarily genetic ones) produce false beliefs about individual intelligence.)
It seems clear to me that much of the time when people mistakenly get offended, they’re mistaken about what sort of claim they should get offended about, not just mistaken about what claim was made.
The important thing for me is that the standard deviations swamp the average difference, so the argument against individual prejudice is valid.
What makes you think “the unpopular-with-SIAI side” exists? Or that it is what you think it is?