[13ren]

PageRank (based on publishers' idea of relevance, though linking) can yield false positives, too. But, I guess someone who goes to the trouble of publishing a document will have a more considered opinion than readers of that document.

BrowserRank is a bit like Web 2.0 PageRank, in that it operates on user voting. It would be nice if you could nominate whose reading habits you would contribute to the ranking you see. Maybe a new job title could be "browser leader", where you sell your browsing data, if you are really good at wasting time. Average that over a million browser leaders, and you have an interesting dataset for Ranking.

The other problem is scale (as someone said). Modeling a web of documents is hard enough; modeling all the readers of those documents is a lot more data - but certainly possible to go some way towards it. Hardware gets cheaper and cheaper, but web adoption will saturate soonish, so we'll get there.