Hm, I'm not so sure I like the idea of getting no points if your winnerin a round 2 match didn't win the round 1 match. The problem with thisis that it rewards people who get the most matches right far too much.We already have a contest like that. The person who did get the match'swinner right could have an absolutely TERRIBLE prediction, like, say,Link over Ganondorf with 58%, and still come out an entire 20 pointsahead. That's no small defecit, and would be rather hard to make up(ignoring the questionable integrity of that person's ability if theypicked Link with 58%). This is not a problem in the Oracle challenge,because you're not locked into your winners; even though I pickedRidley over Diablo in round 1, I was still able to save myself fromtotal loss by making good Diablo picks in the next two rounds.

Don'tget me wrong, I love the idea, I'm just voicing my concerns about howwell this would work in practice, since I agree with the concerns thathave risen over other scoring methods. But if you can think of a goodway to do this, I'd love to see it. Even with flawed scoring, it couldbe very interesting to try at least once.