[2] For an important exceptions, see Young (1986, 1988).

[3] Cohen also argues that two-party systems are ill suited to the aims of epistemic democracy , though his argument is based on the relative ability of political institutions to increase political judgement (Cohen, 35-6). However, my argument against two-party systems above already presumes that voters have good judgement.

[4] I thank Ross Levinsky for the clear inductive derivation which appears from eqns. (6) - (10).

[5] See, for example, Chapters 9 and 10 of (Press, Plannery, Teukolsky and Vetterling, 1986)

[6] So John Rawls writes on the Condorcet theorem, "We must not only be sure that there is a greater chance of a correct than incorrect judgement..., but it is also clear that the votes of different persons are not independent. Since their views will be influenced by the course of discussion, the simpler sorts of probabilisitc reasoning do not apply" (Rawls 1971, p. 358). See Estlund (1994) for a reply.