I do not know BeOS that well, but I disagree with that "accumulated wisdom and academic research up to that point" claim. The prime reason is this: http://2f.ru/holy-wars/fbc.html. Although that probably helped performance-wise, I see that as a gross hack, not as a thing to follow. If BeOS had lived on, I think that decision would eventually have haunted them.

I also remember that, in its time, BeOS was innovating quite a bit in its pervasive multi-threading ('a thread for each window').

Or were both based on research papers or older systems? If so, I would like to know which these were.