This paper doesn't necessarily answer why people can't program, it just points out some pretty interesting discoveries. http://www.cs.washington.edu/homes/reges/mystery/mystery.pdf

It's a paper that focuses on some strange patterns that came out of a statistical analysis of the 1988 Computer Science AP test. It turns out that there are a few questions that end up being a great indicator towards someone's natural programming ability.

Definitely worth the read.