1: \begin{abstract}

2:   The process of consensus voting, or decision making by unanimous

3:   agreement, has many distinct advantages: it fosters discussion and

4:   participation, empowers minorities and independent thinkers, and is

5:   more likely, after a decision has been made, to secure the

6:   participants' support for the chosen course of action. These

7:   considerations, among others, have lead many institutions to adopt

8:   consensus voting as a practical method of decision making.

9:   

10:   The disadvantage of consensus decision making is, of course, the

11:   difficulty of reaching consensus. While this challenge is largely

12:   overcome in many theoretical settings such as Aumann's ``agree to

13:   disagree'' result and its related literature, a hitherto unsolved

14:   difficulty is the lack of a framework offering rational (i.e.,

15:   Bayesian) consensus decision making that can be performed using

16:   simple and efficient calculations.

17:   

18:   We study a stochastic model featuring a finite group of agents that

19:   have to choose between one of two courses of action. Each member of

20:   the group has a private and independent signal at his or her

21:   disposal, giving some indication as to which action is optimal. To

22:   come to a common decision, the participants perform repeated rounds

23:   of voting. In each round, each agent casts a vote in favor of one of

24:   the two courses of action, reflecting his or her current conditional

25:   probabilities, and observes the votes of the rest in order to

26:   calculate an updated conditional probability.

27:   

28:   We prove four results:

29:   \begin{enumerate}

30:   \item Consensus is always reached.

31:   \item Each round of voting improves the aggregation of information.

32:   \item The chance of a correct decision quickly approaches one as the

33:     number of agents increases. This is achieved already at the second

34:     round of voting.

35:   \item Most importantly, we provide an efficient algorithm for the

36:     calculation the agents have to perform.

37:   \end{enumerate}

38: \end{abstract}

39: