> The problem is as always telling the truly good ideas from the just good.

Not only telling the difference, but defining it. The difficulty is avoiding post hoc evaluation, ie. the idea that looks very good in hindsight because it succeeded and the great implementation makes it look compelling.