I'm mainly kidding, although we were certainly not oblivious to the fact that Firefox users are an early adopter oriented crowd who would likely be more interested in what we are doing than the average web user. Another major consideration was the fact that we wanted to be multiplatform, but didn't want the bloat of Java, the licensing complexity of Qt, etc.
At the same time, someone was speculating here (I think) that we only based this on FF because XULRunner is not yet available. This isn't true. First of all, we probably could have used XULRunner in its existing form. Secondly, we feel that there are huge synergies between AllPeers functionality and FF. In essence, we've felt for a while that we are building a new generation web browser, and it doesn't make sense to reinvent the wheel. FF is an amazingly extensible platform that enabled us to save a huge amount of time by leveraging their existing functionality. I hope that the synergies between FF and AllPeers will be obvious once you use it.
Personally I leave FF running all the time, but the problem of memory bloat is a real one. We've discussed having a lightweight daemon running when FF isn't, and we'll probably do this sooner or later. I'm curious to find out how big a problem it will be for people to leave FF running. I'm suspecting it won't be as large as you suggest, but we'll see.
If you're not part of the solution, you're part of the precipitate.