Along the lines of this post, I think it would be neat to have a site where people could vote on what they want a budget to look like. The current year's budget would be a starting point, perhaps with automatic adjustments across the board to compensate for projected inflation. Users would have various options in terms of where to start; they could start with the original budget, or with one done by a friend / other person, or perhaps with the current average as determined by previous submissions (although this might be problematic in terms of expressing rationale, since these users will be starting from a different point from other users--see below). The budget would be presented as something like a tree view, with nodes that could be expanded or collapsed. A user could raise or lower the budget of a node, which would have the effect of proportionally adjusting the budgets of all sub-nodes. When doing so, the user would be asked to either provide a rationalization or +1 a rationalization provided by someone else. When a change is proposed, the site would present a list of rationalizations given by other users, sorted by their rating. (This could result in poor visibility for things written in by users, so, to help boost the visibility of items that are similar to other items, the author of one item could request that it be merged with another item, and, if the author of the other item approves, the items and their ratings will be combined.) Sometimes a rationalization for a node will be inherited from its parent, meaning that it was subjected to an across-the-board cut or raise from its parent, and sometimes they will be specific to the node. Users could have the option of importing changes made by other users. In this way, a budget could be planned and submitted collaboratively by several people, with each person taking an area in which they have expertise.
It would be interesting to have a process like that and pass along the outcome to congress for consideration as a baseline, although doing this has problems that I can foresee (and probably some that I can't). The integrity of the system would be a major issue, and, as with electronic voting, I don't know how to ensure it. Whoever administers it could tamper with the results and probably not be detected. If there are security holes, then an outsider might be able to do something similar. Also, I am defining "the outcome" as the average of the budgets passed by users, but taking this as final might cause problems. For one thing, some projects might be possible with a given amount of funding; otherwise they would not be possible, so funding them half-way might not make sense. Also, the results might (or might not) be more variable than the variability in the budgets passed by Congress, and institutions will want to have some amount of stability in their budgets to facilitate planning. Also, such a system would be subject to bias in that it would select for the positions of the people who participate, and some people may be prevented from doing so for reasons other than their willingness, although all forms of democracy are subject to one form of bias or another. Some of these limitations could likely be accommodated, although lack of security seems like a fatal flaw that might make the whole system impractical in terms of influencing public policy.