The Economist blog Free Exchange argues that journalistic pundits are biased toward outlandish predictions: Pundits are almost never punished for being wildly wrong about something. Nor are they rewarded for being right about something—along with 7,000 other pundits. For journalists, a prediction pays off only if it is both right, and unusual. This gives them an incentive to take unnecessary risks, making somewhat outlandish predictions on the off chance that they will be right. Those pundits who "got Iraq right" or "predicted the tech bubble collapse" are feted with speaking engagements and special television appearances, while those who made sensible-but-dull arguments labour in obscurity. The public is then surprised when their famous pundit makes some other spectacular prediction that doesn’t turn out so well.
Jim Cramer is right 50% of the time, and wrong 50% of the time, but makes more money than any of us because of his flair.
In the case of the Iraq war wrong people got promoted and right people are still in obscurity.http://www.radaronline.com/...
My point is that getting something surprising right is far from enough to be "feted with speaking engagements and special television appearances." The fact that someone somewhere remembers you is not the same.
I'm not really sure I'd count that as "ignored". I know about Xanadu, and I can't personally name a pundit who's famous for "getting Iraq right". It's probably safe to say that Xanadu is well-known among people interested in such matters.
Eliezer, how about the whole Xanadu group predicting in 1985-1990 that the web would soon arrive?
Some examples of unusual predictions gotten right by ignored folk? (If it's not too much bother.)
One area in which this could be tested is sports punditry. The outcomes are objective, but the difficulty would be defining "outlandish" predictions. Although one way would be to look at pundits predictions of scores and then compare how the distribution of predictions compares with the distribution of outcomes. Given what's said above we would expect the predictions to be more dispersed than the outcomes.