"think that one can draw from the results obtained the conclusion that markets will display the same erratic judgment as crowds do...with markets, unlike with large numbers of individuals voting on a question, there's a built-in incentive for individuals to get the answers right".
He is of course right, and indeed, incentives are important, they probably even form a necessary condition, to make markets effective. And it's more than that. Markets let you see the "answers" of others, prices are shown together with volumes and sometimes also order depth. In this context, one often thinks about purely speculative markets, like betting exchanges (Bush or Kerry, Tottingham vs Leeds, etc...), but for "real" markets, non-speculative participants (a.k.a "liquidity traders") contribute with even more information. By buying or selling the goods they afford or can produce given the current price, they reveal important information about fundamental aspects concerning the good such as demand, substitution possibilities, production costs etc.
Finally, there is the selection process. Successful speculators are rewarded, thereby gaining financial strength; less successful ones are soon forced out of the market. Survival of the fittest eventually helps to enforce efficiency. At least insofar skill might with some precision be told apart from luck, and then also might be rewarded fairly enough. These are rather strong assumptions I guess, and one should probably not expect market to regulate themselves or even be the best solution for all "social information processing". But they are surprisingly often the best solution.