It seems plausible to me that the distribution of innate programming ability is a bell curve, but the distribution of programming effectiveness is bimodal. The

It seems plausible to me that the distribution of innate programming ability is a bell curve, but the distribution of programming effectiveness is bimodal. The missing ingredient of course is experience and education. It's similar to an athletic skill, such as the high jump. If you plotted the maximum height that a random assortment of individuals could jump, the resulting graph would be bimodal: the top clump would be those who can do the Fosbury flop and the bottom clump would be everyone else. The innate ability to jump follows a bell curve distribution, but unless you know that trick, you'll end up on the lower clump.