That sounds like an interesting idea. If implemented, it would also be possible to use patterns of voting to estimate (to some degree) how likely a user is to

That sounds like an interesting idea. If implemented, it would also be possible to use patterns of voting to estimate (to some degree) how likely a user is to up/down-vote based on the author alone. Maybe dye a user's name purple for "groupie"? Of course, stylometric analysis would still give away many of the users here well before the sixty minutes is up. A real solution (to burn through all usability in the name of meritocracy) would be text rewriting ala "A Scanner Darkly".