It strikes me that it'd be a really cool social meta-experiment to use LJ to organize a serious scientific experiment. Just to use the most recent example I've seen -- apparently a lot of girls find that red meat makes their menstrual side effects worse. Say you got a group of 20 or so women to join a community where during their periods they'd do a daily log entry of flow estimate (heavy, medium, light) and side effects (cramps, grumpiness). You do a year of that to get a baseline for each. Then the next year you have them split into two groups of ten, one of which avoids all red meat in the last few days leading up to, and the days of, their period; the other intentionally has at least a small serving of red meat, at least every other day.
The data would actually be valuable, at least for throwing out there and getting a serious scientist to do a larger study, and I'll bet you could get the meta-experiment -- "Democratizing Science" -- published in a sociology journal of some sort.
Of course, I'm far too lazy to actually run such a thing. Plus, the only topic I've come up with? I am so not qualified to design the details of. (I'd kinda glance over at melebeth, who is qualified, except she is already Frazzled Overcommittment Lady.)