An article in Cal. Mgt. Rev., Vol. 50, Fall 2007 at 146, does a good job of explaining prediction markets, which in part are a way to aggregate information (See my posts of Feb. 16, 2006 on collective prediction markets; Sept. 13, 2005 on an early version of this; Dec. 20, 2005 about an online futures market for litigation; and March 17, 2006 on internal “markets.”). The article describes three shortcomings of another way to aggregate information, meetings, in terms of the problem of accurate information aggregation.
“First, members in the meeting may not have incentives to provide unbiased information. Worse yet, they often have incentives to provide biased input. Second, members often yield to theirs superiors because of a hierarchical power structure. Third, there is no systematic way to assign relative importance to each input. As a result, whoever argues most eloquently usually has his or her input weighted significantly more. However, a person’s ability to communicate may not have any direct bearing on whether they have relevant information.”
I should add that some people are stubborn so they eventually grind down the others at a meeting. These imperfections of meetings have come to light on this blog (See my posts of June 5, 2007 on politics; Feb.1, 2006 on the chilling effect of seniority; and March 12, 2006 on electronic voting software that permits anonymous opinions.).