James Surowecki opened up CTC on its second day with a keynote on crowds and what business knowledge can be learned from crowd behavior. As background for his arguments, he gives an example of a crowd of individuals who need to guess the weight of an ox. Taking the average of these guesses - the collective intelligence - they came up with a number that was, in fact, really closed to the weight of the ox. In fact, it was only very marginally off. Was it a coincidence? No. James says that it is merely an example of group intelligence.
Under the right circumstances, groups together can be smarter than even the smartest person amongst them. That they have a power of collective intelligence. As business people, the average group perspective can be a powerful thing. Google has succeeded by taking group opinion into its search - it returns better, more powerful search results because of the PageRank system. PageRank is an Internet vote for relevancy.
Probability will take that groups will be correct in many predictions, including those of business. What is the stock market if not a crowd. But, on a smaller scale, businesses can tap their own employees as a group or crowd. HP set up an online stock market that was used to predict printer sales. Their predictions, as a group, were more effective than the formal predictions the business was running. Internal stock markets are used because they are an easy way to make predictions. Even without incentive or payback for participation. It's like bottom up decision making. And it doesn't need to be chaotic, as many assumptions for teams and crowds go.
What circumstances make groups smarter? People need to be independent, cognitively diverse, have diversity in information, and there needs to be a way to aggregate opinion. Collective intelligence comes out of an aggregate or average of opinion, not from picking the strongest argument. Consensus and imitation are failures of group intelligence.
Where decisions under uncertainty need to me made, we cannot imitate what others around are doing. Imitation is rational, but it does not lead to collective intelligence. And, though we learn from others in our group, it is also important that we share information along, not influence. This is difficult. Some people will be more powerful than others. This can lead to an information cascade that can go against rational self-informed decision making.
Information must come without influence, but it's not an easy step. To avoid influence skewing rational decision making, people must have loose ties to each other. This may sound counter intuitive, since groups that know each other well do work quickly. However, they are more likely to reach consensus rather than forming individual opinions. One step to change this is to have leaders spend more time asking questions than answering them. Diversity and independence are key markers to making groups collectively smarter. Once individual decisions are freely encouraged, those opinions can be averaged or aggregated - this is where collective intelligence is shown. The final answer may be completely different than the individual ones, but it will be the best decision and the best predictor.
