Dave: I think one lesson you can take from Wikipedia’s success is the value of extreme modularity. The bite sized components of individual term entries lowers barriers to participation and barriers to editing and curation. Open content inevitably “decays” and the structure of most traditional textbook means that as soon as significant aspects of the content become outmoded, outdated, or broken (links, images, etc). the entire project is rendered suspect and/or worthless.
I would suggest that an open textbook project should be divided into clear micro genres and organized like blog posts or wikipedia entries (with tags and categories). You could have Topic Primers (introductory/overview/background material) Deep Dives (exploring essential concepts) and Activated Learning (activities, exercises, applications). The great thing is that each new cycle of a MOOC-based writing process could balance revising old entries with simply adding new ones. Old, decayed material would drift to the sea floor while new material would be churned up to the surface by new engaged participation.
I think the clear model here is the DS106 daily create bank with its open-ended user and editorial community. This example might also suggest that a successful open textbook may also need to connect directly with a network of practice that uses it and contributes back to it (even beyond a particular set of scholars). Maybe, creating a system for “playlists” that would allow individual instructors to curate paths through the best/most relevant material? Maybe, a link-back method for sharing student uses of the material back with the community?
In any case, my two cents argue for extreme modularity, a both/and revision, new content strategy to iterations, and a focus on user interactivity in design.