Hypotheses

I maintain a list of loose hypotheses, e.g. predictions / expectations, and track (solely empirical) evidence as I find it. It is my belief (hypothesis!) that empirical evidence is more important than theoretical ideas about performance or behavior of networks.