I fully agree. Also, I think (as embarrassing as it is) we need raw data and code. Would make peer review much easier as you could perform the same analysis on the same data. Makes it easy to correlate results and validate premises of data sampling methods.

(Warning: In Scientist-Engineer limbo)