A committee consisting of two factions is considering a project whose distributive consequences are unknown. This uncertainty can be resolved at some unknown future time. By delaying approval, the committee can gradually learn which faction benefits from the project. Because support of both factions is required for approval, it can only happen when there is sufficient amount of uncertainty about the identities of winners and losers. I show that in many situations, a project is more likely to be approved if it gives a lower payoff to everyone. The probability of approval and expected payoffs of both factions are higher if the project is ex ante less likely to benefit the faction that tends to receive good news faster. Equilibrium amount of learning is excessive, and a deadline on adopting the project is often optimal.