That may be the case, but I still don't find the explanation satisfactory from the point of view of the classic general intelligence theory (not that I have a better alternative, though).

To clarify, the traditional theory of general intelligence, which is taken as a background assumption in most IQ-related research, assumes that general intelligence is normally distributed in the general population, and any reasonable measure of it will be highly correlated with IQ test scores (which are themselves artificially crafted to produce a normal distribution of scores). Moreover, it assumes that people whose intellects stand out as strikingly brilliant are drawn -- as a necessary condition, and not too far from sufficient -- from the pool of those whose general intelligence is exceptionally high. Now, if the scores on IQ tests are rising, but there is no visible increase in outstanding genius, it could mean one or more of these things (or something else I'm not aware of?):

• We're applying higher criteria for genius. But are we really? Has the number of people at the level of von Neumann, Ramanujan, or Goedel really increased by two orders of magnitude since their time, as it should have if the distribution of general intelligence has simply moved up by 2SD since their time? (Note that for any increase in average, ceteris paribus, the increase in the rate of genius should be greater the higher the threshold we're looking at!)

• The average has moved up, but the variance has shrunk. But this would have to be implausibly extreme shrinkage, since the average of IQ scores today is roughly at the z-score of +2 from two generations ago.

• The modern culture is making common folks smarter, but it drags geniuses down. I believe there might be some truth to this. The pop culture everyone's supposed to follow, however trashy, has gotten more demanding mentally, but true intellectual pursuits have lost a lot of status compared to the past. Still, such effects can't explain the severity of the effect -- remember, the Flynn increase is greater than the difference between borderline retardation and being above average in the way the scores are used for diagnostics!

• The IQ scores say a lot about people who are average or below average, but not much about smart people. This seems like the most plausible option to me, and the only one compatible with evidence. But this means that the standard model based on the normal distribution is seriously broken when it comes to the right side of the distribution, and it also makes the results of many heritability studies much more murky.

All in all, the situation is confusing, and unlikely to get clearer in the near future.