Moving fast, defining tasks
re 1: Check out how scientist collaborate in universities. Of course it varies a little with disciplines, and if the standing of the universities is not equal - but in general a lot of collaboration is the norm (the currency is citations of a paper, and each author pretty much gets them all). That is a similar dynamic as in the general open source scene methinks.
re 2: Yeah, in most intellectual (or interlectual?) work some diversity is helpful. And once you go past a few layers of specialization, diversity often becomes a rare good and sometimes a limiting factor. I just happened to write down my favorite work approach here. Applied to the book I would suggest to push (on the FAST track) a first tiny set of cases all the way through into the typesetting stage. Delegating as much as possible (make tasks) to relatively green people, who then have time to get going when the real surge comes.