We have known for over 30 years that learning to program is surprisingly hard. A series of international studies have shown that measuring learning in programming is challenging, because the demonstrated performance is so small. In my group, we have been developing approaches to improve learning about computing, by improving retention through relevance and by teaching in problem domain context. Our classes and studies have utilized computer-supported collaborative learning. We have learned how anchored collaboration can lead to longer on-topic discussions, but how perceptions of course culture can dramatically inhibit discussion. We have shown that well-designed on-line activities can lead to better learning at reduced cost (including time costs for the student and instructor). We are currently developing an ebook for learning computer science by high school teachers where we are trying to integrate these lessons for a new audience.
