Isaiah Berlin’s contrast of the hedgehog & the fox is widely applicable, though I’m not sure it works as well as Philip E. Tetlock does when he applies it to pundits. He is, however, quite correct in his assumption that A) being extreme is likely to be more often wrong, and, B) also more interesting. Partisan pundits (or ones with a fixed idea) are seldom broad, deep, historical thinkers but rather hyperbolic, emotional ones.
He is also right, few are asked to compare their prognostications with what actually came to pass. While our politicians, rightly, are held accountable for assessments and predictions of years ago, the pundits doing the critiques are seldom confronted with their unfulfilled prophesies of even a few months.
Of course, bloggers, with access to a variety of search engines and a keen sense of competition with the “old” media, sometimes entertain themselves with just these discrepancies, but our imaginations prefer broad & graspable generalization rather than more ambiguous modified ones.
This is discussed in more length in Tetlock’s Expert Political Judgment: How good is it? How can we know? (Thanks to A&L.)