I started work on a "somewhat" similar project* around the beginning of the year. This was for the state level, not the federal level. There were quite a number

I started work on a "somewhat" similar project* around the beginning of the year. This was for the state level, not the federal level. There were quite a number of issues. The most annoying being that the state archives only went back so far. Checking it now, it goes back to the mid-70s (very encouraging, 9 months ago, it only went back to the mid-90s). If you need a reference on why this is important go back and read the Patriot Act (hint, without access to the legal code going back very far, it is futile). As far as your bullet points, we already do much of that (ok, not in "real time", not with commit messages (you could have it, but everyone would quickly fill it in with "cause it's good for America"), but much of this is publicly available). Anyone who adds or votes for or against is very well documented. For my state it is a real PITA to parse out and apply but that information is provided and in a usable manner. The only thing missing is the speed and pretty ui. Also, Sunlight has already done a great deal of work in this area (when I last checked they were still working on the API). *Since, I've let the idea lay dormant for months now, I'll share it. The original idea was to setup a Facebook type "game" where bored housewives or whoever get a list of 5 random bills. They get the summary and it links to the bill's text and any current laws that will be affected by that bill. What they don't see is the bill's name and who the sponsor and co-sponsors are. After reading the summary they fill out a short survey of who and what party they think the bill came from, do they think the bill is a good or bad idea and any tags they think are appropriate for the law ("pork","insurance","nice", etc) and even add their own notes. After they do that they find out who the sponsor/co-sponsors are. The idea was to give people an incentive to read pending legislation without an instinctive bias. Though my code is ugly, I may GitHub what I have so far.