This looks like an interesting idea: Phygg, a cross between the arXiv and Digg. The idea is that people submit papers (to arXiv, as usual), and then Phygg lets its members vote them up or down. Presumably, the papers that are most worthy of your attention will wind up with the highest vote totals.

Here’s my thought, though: what really qualifies as worthy of your attention? It depends on who you ask: someone working on the mathematical foundations of quantum gravity, for example, will be interested in a whole different type of publications than an experimentalist searching for extrasolar planets. The way Phygg is set up right now, it’s hard to separate those different fields out, so when you look at the top voted papers, you get a whole bunch of results that aren’t relevant to you, simply because they’re off-topic for whatever you’re researching. This isn’t such a big issue for Digg, because when it comes to random website links, most people are interested in the same stuff. But I think what science needs more than a voting system is a reliable categorization system. Easier said than done, of course.

What I think …