Jeff Zeitlin writes: > Currently, the ballot requires a straight ranking: I like Zendo better > than Ice Towers better than Martian Backgammon better than Spicklehead. > It doesn't take into account that a particular voter may feel that Zendo > is only marginally better in his opinion than Ice Towers, but both are > WAY COOL, and that Spicklehead is WAY below Martian Backgammon. His > vote is completely countered by someone who feels "Eh" about all four > games, and rates them Spicklehead, Martian Backgammon, Ice Towers, > Zendo. Coincidentally, just last night I watched something my Tivo recorded for me on its own: Tavis Smiley interviewing William Poundstone, the author of a new book called _Gaming the Vote_. Poundstone (whose earlier book _Prisoner's Dilemma_ I own and enjoy-- but I don't think my Tivo knows that!) was recommending "range voting", where each voter rates every candidate on a scale (e.g., 0 to 10). The winner is then the candidate who gets the highest average rating. I'm not convinced this is really the best method (for one thing, it doesn't satisfy the Condorcet criterion, according to Wikipedia), but I like the fact that it allows voters to provide more information than just an unweighted list of preferences. > The system I encountered says "You have x points. Allocate them as you > see fit among the choices." X is a function of how many choices there > are on the ballot - the particular example I encountered said "two > points per choice on the ballot". I believe this is equivalent to range voting (if you allow non-integer ratings), but range voting seems easier-- and more familiar, since it's essentially used in everything from Olympic voting to Boardgamegeek to HotOrNot. http://en.wikipedia.org/wiki/Range_voting --dougorleans@xxxxxxxxx