Bezos's observation reminds me of Philip Tetlock's conclusions in "Expert Political Judgement" (http://press.princeton.edu/titles/7959.html). Tetlock studied it pretty rigorously and came to a number of conclusions, but the central thesis was that people who got things right the most were people who thought by building up lots of competing models and evaluating all of them (ie, having lots of little ideas) rather than guiding all their decisions by a single ideology.