This resonated a lot with me as well -- elementary school computer science education in the '90s wasn't very developed. I went to a high school that had one of the better computer science programs for that level (in that it existed), but it still wasn't anything compared to the university level. The people who really learned the stuff were a self-motivated group who interacted outside of class as a club. These kinds of extracurricular studies were at best ignored by school faculty, and for students like myself, who was falling behind the curve in the school's demanding math/science/language program, were actively discouraged.

There's a crucial distinction between my experience and hers, though. I had the support of a core group of computer nerds, egging each other on and making simple games and the like. This group was almost entirely male, and were mostly interested in other stereotypical geeky hobbies like video games, anime, and tabletop role-playing games. We weren't consciously exclusionary, but anyone who didn't match the profile was probably going to feel very out of place.

I remember in my CS course, there were two girls in the class, and they were most definitely not part of the clique, even though they were brilliant by most objective definitions. One of them was definitely in the "gifted" category, and while the rest of the class was concerned with making video games, she was writing an equation plotter. We didn't talk to her all that much.

My point with all this rambling is that there's a lot of bundling of interests that goes on, all of them male-dominated, and if you aren't into those things, you don't get the same peer education as somebody who is. I think this explains a lot of the gender disparity we see in CS education today, and I don't know what to do about it. Education should not be tied to social cliques, but in reality, it often is.