I've seen two forecasting techniques use for creating predictions from the input of a group:Idea Futures and the Delphi Method. I have not seen any good comparison of these techniques and how each one can be applied most effectively.

The Delphi Method was pioneered at the Rand Corporation in Santa Monica, CA. This technique has been used to help manage large software projects. The essence of the Delphi Method is the use of anonymous, iterative voting voting. Recently, I helped create a lightweight vote processing system that uses a Perl/CGI front-end to support conducting polls on the Web. I've also created software to conduct Delphi Polls.

Idea Futures is an experimental technique that uses market forces to evaluate the probability and interconnectedness of future events pioneered by Robin Hanson. I've been active as a player on Ideosphere's Foresight Exchange, since it started. I have written several technical and financial claims that are now actively trading.

The methodology of the Foresight Exchange is very similar to the methodology traditionally used in insurance by companies like Lloyd's of London when evaluating risk in cases where there is little in the way of real history or data to go by. One of Hanson's big innovations was to propose using this methodology for scientific and technical investigations.

The Chicago Mercantile Exchange has started trading of weather futures. Interestingly, in Robin Hanson's writings one of the most prominent examples showing that establishment science was having problems was a situation in which an obscure British scientist demonstrated he could consistently beat odds determined by the British Weather Bureau.

The combination of Idea Futures and Bayesian belief works will, in my opinion, eventually evolve into much more sophisticated and accurate projection and forecasting than anything we have seen to date.