[hackeraccount]

yeah - I think his point was something along the lines of "you can have a useful theory but it might be worth getting rid of it for something less useful that might be more productive in the long run because it has better understanding of the fundamental mechanism."

Or more shortly using predictive ability as the sole criteria can lead to less correct results then alternatives.

If my description is correct I would say that the advantage of using prediction as the sole criteria is that it avoids the burden of making arbitrary selections. I understand the appeal of the choosing a course because it's more elegant or beautiful but if you were getting on a plane and were told that was the guiding light of the designer - or worse the engineers - how happy would you be?