User:RandomP/Importance rating
This is a proposal for a ratings system to judge, without resorting to deletion, how important Wikipedia articles are.
Our input data is preference data collected from users: essentially, every user can decide, for every pair of articles, which one is more important, or that they are (approximately) of the same importance.
Data collection should probably be through adding a little box to assign importance levels to all articles displayed to a user. With XMLHttpRequests and user-provided css, this shouldn't be too much of a problem, though it might be extra load on the mediawiki servers.
Once all the data is collected (register user on a list somewhere, use mvs to get it all), we simply find the minimum epsilon such that there is an article a0 such that at most epsilon votes can agree on any given article that they think is more important - then remove a0 and repeat. Probably O(n2) in the worst case, but perfectly doable in the common case.