First -- great post.

In fact, one my key motivations for writing it -- and a point where I strongly disagree with people like Kuhn and Feyerabend -- is that I think heliocentrism was more plausible during that time.

I think this could be made clearer in the post itself, because whether or not there were good reasons around at the time is prior to whether or not we should try be like the heliocentrists.

Rather, I think they did something very right (and very Bayesian). And I want to know what that was.

This reasoning is itself quite non-Bayesian: exploring possibility-space, rather than updating a probability distribution over known unknowns. And maybe it's part of what the heliocentrists were doing right.

Do you think selection bias might play a role? Maybe the biggest breakthroughs tend to come from headstrong and philosophically-inclined scientists, but most such scientists we never hear about, and being headstrong and philosophical isn't epistemically hygenic in general.

I can't help thinking of the great Red Dwarf novel "Better Than Life", whose concept is almost identical (see http://en.wikipedia.org/wiki/Better_Than_Life ). There are few key differences though: in the book, so-called "game heads" waste away in the real world like heroin addicts. Also, the game malfunctions due to one character's self-loathing. Recommended read.