Deals in the Dark: multi-agent moral hazard and learning