The now defunct media watchdog magazine "Brill's Content" had a regular feature in which they'd examine every prediction made by all the big Beltway pundits and then compare those predictions against reality. They'd only award points (plus or minus) for clear results -- vague, hand waving "predictions" didn't count. There was no weighting, so trivial predictions counted just as much as earth changing ones. There were some general findings:

1. "Chimpy" (the magazine's in-house pundit, a chimpanzee) not so surprisingly was about 50% correct. That made him a pretty good pundit -- somewhere in the top quartile.

2. Conservative pundits fared worse -- often much worse -- than their "liberal" counterparts. Margaret Carlson and Eleanor Clift did pretty well, for example. George Will and Fred Barnes, not so good. (In fact, I think "The Beltway Boys" with Barnes and Mort Kondracke was one of the poorest performing programs at the time.)

3. Some pundits made far more predictions than others -- there was quite a bit of variability in their prognosticating tendencies. How much punditing they did didn't seem to be correlated to, say, salaries and appearance fees.

I say, bring back Chimpy!