We study communication in committees selecting one of two alternatives when consensus is required and agents have private information about their preferences. Delaying the decision is costly, so a form of multiplayer war of attrition emerges. Waiting allows voters to express the intensity of their preferences and may help to select the alternative correctly more often than simple majority. In a series of laboratory experiments, we investigate how various rules affect the outcome reached. We vary the amount of feedback and the communication protocol available to voters: complete secrecy about the pattern of support; feedback about this support; public communication; and within-group communication. The feedback no-communication mechanism is worse than no feedback benchmark in all measures of welfare - the efficient alternative is chosen less often, waiting cost is higher, and thus net welfare is lower. Our headline result is that adding communication restores net efficiency, but in different ways. Public communication does poorly in terms of selecting the correct alternative, but limits the cost of delay, while group communication improves allocative efficiency, but has at best a moderate effect on delay.