To discover something new, two things must be true. The pre-requisite ideas must be known. And the right ideas must be selected and synthesized. The former is a coverage problem, the latter is a search problem. The hard part is you don't know which is your bottleneck.