Pruning away excess information is about fitting into a niche, while keeping the extraneous is about flexibility. The selection should be based on how uncertain you are about the future and nothing else.

Me personally, I'm aware of four big picture questions that predict 16 possible (and incompatible) futures of great divergences, so I figure flexibility is the way to go.

This was quite a satisfying read.