Good point.

Also, on typical test I'd expect well educated moderately high IQ person to have 100% success rate on everything that's strongly related to intelligence. So at the top range the differences are driven by the parts that have much less direct relation (e.g. verbal, guess next in sequence, etc). Correlation is a line but real relation we should expect would be more like sigmoid as the relevant parts of test saturate. Furthermore, IQ test doesn't test capacity to develop competence in a complex field.