You're always able to evaluate an experiment to a full-fledged startup. And, however odd it sounds, very frequently I'm much more effective when I know that no

You're always able to evaluate an experiment to a full-fledged startup. And, however odd it sounds, very frequently I'm much more effective when I know that no one counts on me, and I'm free to experiment. It gives a little competentive edge as well.