We present a model of bargaining in which a committee searches over the policy space, successively amending the default by voting over proposals. Bargaining ends when proposers are unable or unwilling to amend the existing default, which is then implemented. We characterize the policies which can be implemented from any initial default in a pure strategy stationary Markov perfect equilibrium for an interesting class of environments including multi-dimensional and infinite policy spaces. Minimumwinning coalitions may not form, and the set of equilibrium policies may be unaffected by a change in the set of proposers. The set of stable policies (which are implemented, once reached as default) forms a weakly stable set; and conversely, any weakly stable set is supported by some equilibrium. If the policy space is well ordered then the committee implements the ideal policy of the last proposer in a subset of a weakly stable set. However, this result does not generalize to other cases, allowing us to explore the effects of protocol manipulation. Variations in the quota and in the number of proposers may have surprising effects on the set of stable decisions. We also show that equilibria of our model are contemporaneous perfect e-equilibria of a related model of repeated implementation with an evolving default; and that stable decisions in semi-Markovian equilibria form the largest consistent set.