[agumonkey]

Ambiguity driven development would make for a good book.

And not to quote pg's old crowd. The middleground approach of crafting little theories that will help you solve a larger problem through easy variation seems a good starting point.