Reading the Ron Unz essay from July (via Steve Sailer) was interesting enough in itself, pointing out the rural/urban split in IQ scores worldwide that seems to disappear quickly when the country mouse becomes a city mouse.* But it included, almost in passing, a weakness in twin studies I had never noticed. And I should have.
Much is made of the fact that identical twins reared apart have IQ's much more similar to each other than to the others in the family they grew up in. This has been solid evidence for the "hard-genetic" folks for decades, and I have relied on it myself at times. Yet it neglects the evidence that the separated environments are still quite similar in most cases. We can note that one family has many books, the other few; one family is composed of garrulous researchers while the other of silent, morose, neglectful unemployed - but the children were the same age and grew up in the same country. Their media exposure and general stimulation might have been similar, the same popular songs, popular TV shows, even same school curriculum. Ideas would have been introduced at the same age. From within our own culture, the homes might appear divergent. But imagine running into an an age-mate from rural Laos and trying to find a point of connection in childhood reminiscence.
I have switched to soft-to-medium-genetic pretty quickly. Certain minima of stimulation and ideas introduced at critical periods must be present before we are comparing apples to apples. I still believe that some of the apples will tend to be bigger because of their DNA, year after year, and differences will remain no matter how equally all are tended and fed. But the comparisons can only be made within certain parameters.
BTW, Unz kicks Stephen Jay Gould pretty hard, which is always fun.
*It's more complicated than that, as it doesn't seem to always work and especially not always work fully; and expanding media culture available to rural areas seems to have dampened the effect - yes, yes. But the effect is still dramatic in comparing say Greece 1961 to Greece 1979, a span that should certainly not show any genetic effect.