It's a slightly tangential idea, but my idea of human contributors extracting the best bits of tmo (and the rest of the Maemo contentsphere) could help.

However, it probably doesn't scale to the "tens of pages with about a dozen useful posts (and which dozen is constantly changing)" threads that this seems to be dealing with.