I think it's an interesting idea, but I don't like theauto-extrapolating part. The challenge suffers in that it only"changes" your predictions when you predict a winnerincorrectly...however, matches that you call correctly but areextremely far off what you expected are also very significant, yetthose couldn't alter your predictions for any of the later rounds.

Also,another thing, kinda tough to explain, but here goes. Right right nowthe auto-calculations are based on keeping your original winnerconstant in strength, meaning any incorrect guesses would be a resultof purely "underestimating" that character that pulled off the upset,yet clearly many upsets are a result of people overestimating thecharacter that lost. It's also often a combination of the two. I thinkit might be better if there wasn't any auto-extrapolation at all incases of upsets, but to just keep your percentage predictions exactlythe same. So, if you have Link over Raziel with 75%, but Strider beatRaziel previously (with, say, 55%) you'd have a "new" prediction ofLink with 75% over Strider. So, yeah, nice and simple...not that Ithink it'd be much better. At least it doesn't have the sloppy methodof only changing character's extrapolated values if you predict a matchincorrectly.

Part of the fun of the Oracle Challenge is totake previous results in the current contest and see how you interpretthem. It's almost blind luck guessing on how characters have changedfrom past years. Sure, some characters will obviously change (Cloudwith Advent Children), but others have changed in popularity for noapparent reason (Solid Snake). Besides that, I wouldn't want to seeextrapolated standings take control over a contest at all...

Iapologize for the slamming...well, I had to speak my mind. In allhonesty, I'd probably participate in that contest regardless. =P

Oh,and I'd love to hear moreideas on how to improve a contest like that,but I can't think of any at the moment. I was thinking about Aprosenf'sidea of just making a list of predictions in case of any possibleupset, but that's also flawed because you still can't take into accountmargins of victory at all, but strictly only victories. I suppose itCOULD work, but it would still be very sloppy...

I would love tosee an adjustment to the Oracle Challenge by having all predictionssubmitted in secret (you know, through e-mail) so that you can't useothers' predictions as a guide for your own.

---
Yoblazer beat me in the Guru Challenge! Good job!