A test I would like to see is learning abstract algebra by reading Wikipedia articles versus learning it by reading the better textbooks, such as the Artin textbook. My bet is that the textbook users would come out ahead by any reasonable means of assessing their understanding and performance.
I did much better in some of my classes learning math, CS, and EE concepts through Wikipedia than I did from the textbook or lectures.
Edit: possible cause: textbooks designed for selling to universities rather than designed to educate, and courses designed to maintain the professor's royalties.
I would guess that it was more likely due to the pick-and-choose nature of most courses. When I take a class, the material that is actively tested is usually only the general concepts and only the ones the professor deems important. Wikipedia is designed for you to be able to get a general idea about specific topics, where a text book is designed to include in-depth and comprehensive knowledge.
People would come to a deeper and more complete understanding of a subject by reading a document whose actual intent is to bring them to a more complete and deep understanding of that subject.
