From Penicillin to MySpace, balancing structured research and build with openness to failure has proven a successful approach.