I think the fundamental mistake in brainstorming is thinking that out-there ideas are a good approach. And that comes from underestimating the size of idea-spac

I think the fundamental mistake in brainstorming is thinking that out-there ideas are a good approach. And that comes from underestimating the size of idea-space. Exploring idea-space without a map doesn't break you out of your own limitations, because combinations are infinite but your own mind is not. So you flail generating unworkable ideas within the box rather than hill-climbing towards a workable one that might be out of the box. What is missing is a heuristic. Like, nobody would expect an AI to generate good ideas by throwing random numbers in the air. They would demand to see some heuristic for iterative improvement.