One knew that Andrew Sullivan would not be able to resist the James Watson/African IQ controversy --
No one disputes that the raw data overwhelmingly show clear bell-curve (geddit?) differences between racial groups on IQ. That applies just as much to bell-curve variations between Asians and Caucasians as between Africans and Caucasians. No one disputes either that the IQ variations within ethnic and racial groupings exceed any differences between them. What's disputed is the relative influence of genes and environment - and their interaction - on these results.
No No No No No No No. What's disputed is that there is any meaningful sense in which one can speak of a single, numerical, rankable characteristic called "IQ" which adds anything to our understanding of issues like educational outcomes, achievement, wages, anything. Once you're down the road, as Sully is, of trying to put everyone on this single scale and use it to explain other stuff, you've already lost.
UPDATE: Henry at Crooked Timber elaborates. See especially this excellent comment.