tag:blogger.com,1999:blog-3722233.post116309154363842011..comments2022-12-01T07:04:29.377-06:00Comments on Computational Complexity: Prediction Map Post MortemLance Fortnowhttp://www.blogger.com/profile/06752030912874378610noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-3722233.post-1163257559313934392006-11-11T09:05:00.000-06:002006-11-11T09:05:00.000-06:00I understand that combinatorial markets have been ...I understand that combinatorial markets have been proposed as a way out of this - however the complexity of that (taken over 50 states) would be mind-boggling at best. <BR/><BR/>Would it not be best to identify (Bayesian?) parameters that help predict the races? Something like a basis set of outcomes - and the predictions on the individual races could be made out of combinations of those basis functions?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3722233.post-1163172480969878002006-11-10T09:28:00.000-06:002006-11-10T09:28:00.000-06:00So how did the markets do for the governors?So how did the markets do for the governors?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3722233.post-1163113367173174912006-11-09T17:02:00.000-06:002006-11-09T17:02:00.000-06:00But what about 24 hours before your snapshot? Or ...But what about 24 hours before your snapshot? Or 48 hours? In other words, at what point did the map settle on the actual outcome in terms of blue v. red?<BR/><BR/>A prediction is more powerful the larger the gap in time, it would seem. And if the map is essentially showing the same thing that the polls at the same time showed, then it would seem much less useful.<BR/><BR/>So I'd really like to see this kind of analysis.Anonymousnoreply@blogger.com