Get Scott Aaronson to mention the fact that LW 2.0 is a real-life instance of eigendemocracy in one of his “announcements” posts. The credit is his for inspiring the new voting system.
Have you talked about what LW2′s system actually is, in detail, anywhere?
I consider these sorts of things (collaborative filtering) to be incredibly important, it’s become obvious that, say, reddit’s one account, one vote in any context system is inadequate.
It seems to me that eigentrust, or something like it, probably models rank aggregation correctly. That is, I’m getting a sense that you could probably sort content very efficiently by asking users for comparison judgements between candidates, building a graph where each comparison is an edge, then running eigentrust to figure out what’s at the top.
So I’ve been thinking about eigentrust. Gradually working my way through this Eigentrust++ paper (though I have no idea whether this is a good place to start digging into the literature and probably wont make it very far)
Have you talked about what LW2′s system actually is, in detail, anywhere?
I consider these sorts of things (collaborative filtering) to be incredibly important, it’s become obvious that, say, reddit’s one account, one vote in any context system is inadequate.
It seems to me that eigentrust, or something like it, probably models rank aggregation correctly. That is, I’m getting a sense that you could probably sort content very efficiently by asking users for comparison judgements between candidates, building a graph where each comparison is an edge, then running eigentrust to figure out what’s at the top.
So I’ve been thinking about eigentrust. Gradually working my way through this Eigentrust++ paper (though I have no idea whether this is a good place to start digging into the literature and probably wont make it very far)