Agents exert hidden effort to produce randomly-sized innovations in a technology they share. Returns from using the technology grow as it develops, but so does the opportunity cost of effort, due to an 'exploration-exploitation' trade-off. As monitoring is imperfect, there exists a unique (strongly) symmetric equilibrium, and effort in any equilibrium ceases no later than in the single-agent problem. Small innovations may hurt all agents in the symmetric equilibrium, as they severely reduce effort. Allowing agents to discard innovations increases effort and payoffs, preserving uniqueness. Under natural conditions, payoffs rise above those of all equilibria with forced disclosure.