I don't want to disagree with Shalizi in general, as I think he's broadly right, but the comments in the Crooked Timber article are enlightening. The question of how heritable IQ is seems to be posed almost entirely in terms of race - as if that is the only way that the question of heritability can be framed.*
Many of the commenters have picked up on Shalizi's claim that the question of IQ heritability is "not a well-posed question". I disagree with this, strictly speaking, given that IQ is ultimately a measure, however poor, of neural function then, at some level of environmental equality, it will be substantially heritable (and thus 'genetic') - this doesn't tell us that current differences in IQ between racial groups are substantially due to genetic differences, but it also doesn't rule out the possibility a priori. Nor does it assume that IQ measures any really extant thing, rather than just being a grab-bag of questions that challenge the brain to a greater or lesser extent.
It is perfectly respectable to think that the evidence for racial (or gender) differences in IQ is poor, as are estimates of heritability, and confounded by environmental factors that are not, and perhaps cannot be controlled for, that IQ tests are not measuring any single unitary thing, and that there is substantial genetic overlap between groups we identify by ethnic labels. You could also note that many of the most vocal proonents of such positions appear to do so for less than savoury reasons.
But so many people seem to need to go beyond this position to claim that there can be no differences between racial groups, that IQ cannot have any heritability under any circumstances, that IQ measures nothing at all, that it has no connection at all with anything we might call 'intelligence', and that there can be no genetically identifiable racial groups. All these latter claims are profoundly false, and it never ceases to amaze me that so many liberal people feel compelled to believe them in order to shut out the spectre of racism. [if you want to see the excesses of the other side of the coin try the people at gene expression].
The Crooked Timber comments also include this gem:
"...the evidence that Down syndrome and similar conditions are linked to unremediable IQ deficits is just about as weak as the race-linked evidence, for many of the same reasons; see my article in Disability and Society, here – http://home.vicnet.net.au/~borth/DOWN1.HTM"
* I think this is because people are unable to understand that high heritability estimates for IQ do not mean that racial differences in IQ must be due to genetic differences. My favourite example of this is schizophrenia. We know that schizophrenia is heritable, with much greater chance of developing it if close family members have it, and monozygtic twin concordance of around 50%. There have also been a number of genes identified that confer a risk for developing schizophrenia. Now black people in the UK and the US (African Caribbean/African American ethnicity) have much higher rates of schizophrenia than the rest of the population. And this doesn't seem to be due to racist or culturally insensitive doctors overdiagnosing it (doctors from the Caribbean confirm that these people really do have schizophrenia). So there you go, must be due to black people being genetically predisposed to schizophrenia? Nope, because black people in Africa and the Caribbean don't have this elevated rate of schizophrenia. We don't know quite what causes this elevated rate, it doesn't seem to be due to poverty or migration, and many people suspect it is due to the neuropsychological impact of racism or being in a ethnic minority within a wider society.