(AND THE CATCH-22 OF GROUP THINK) ...
Here's a tip of Awesome Value, adapted from James Surowiecki's magnificent The Wisdom of Crowds:
You've got a huge marketing decision to make post haste. (Or a decision about War & Peace if you're, say, President.) You gather 10 experts in the field. Lock them in a room for 72 hours. Ask them to come up with a best estimate of, say, success of a New Product you're close to launching. The process is better than nothing—maybe.
Alternate: Select 10 experts from disparate fields, some closely associated with the decision at hand, some not. Tell each one to stay isolated in his or her individual office, lock and bar the door, turn off all phones and computers—and come up with a best estimate in 72 hours, which will then be emailed to you. You, in turn, average their estimates and take the result as the collective output. This process/result is likely to be ... Solid Gold!
Surowiecki's argument (supported by a ton of evidence and research, from every field you can name and some you can't) is that crowds, even crowds of non-experts, are wise beyond measure. IF JUDGEMENTS ARE TRULY INDEPENDENT ... AND 100% PROTECTED FROM PRESSURE AND GROUP-THINK. It's a lot more complicated than that, of course. (Read the book!) But the second—successful—approach I described is an adaptation of a process Surowiecki reports the U.S. Navy first used in 1968 to find the lost submarine Scorpion. (And, alas, the questionable first approach is not far from the 9/11 Commission's group-think conclusion that centralizing intelligence activities & power in a Mega-bureaucracy with a solo Czar who reports directly to POTUS is the answer to getting piercing, imaginative, independent results that thwart wily, inherently unpredictable terrorists.)
The way I laid this out makes it sound as if you've got to be a Big Cheese to take advantage of it. Not so. At all. You are running a 6-person project team, and you'd like to get an estimate of something or other of monumental importance to your work. Use your network (ever so relatively easy to do in WebWorld), and dig up 5 disparate experts or interesting folks in general, reward them with a dinner for their trouble, and ask them to work solo and send you their Best Guess in 24 hours; you, in turn, process their answers-estimates. (In the lost sub case, experts were betting bottles of Chivas Regal over who would come closest to being right when the job was done.) Just don't gather them in a conference room, real or virtual, and ask for a "consensus view"!