This morning at KM World in San Jose we started off with a memorial to Melissie Rumizen. Steve Barth, Verna Allee and myself read our own and other tributes to the background of photographs collected from her friends. Many people in that audience knew her directly or through her work so it was a sympathetic audience and a moving experience.
James Surowiecki was first up of the keynotes (my turn comes on Thursday) with a well rehearsed powerpointless presentation that provided a good introduction to his Wisdom of Crowds. Now I have heard James several times before and the material is familiar. On this occasion he focused on the three conditions necessary for the crowd to be wiser that any of its individual members. I think it’s worth summarising those and adding some comments on the implications
The three conditions are (my summary)
- There has to be a way of aggregating individual responses to provide collective judgement
- There has to be cognitive diversity in the participating group
- Each of the participants has to make decisions independently of the others.
Now all of these makes sense, it links in with large number mathematics – assuming a chaotic system in which each agent is independent of the other, thus probability and statistical distributions work. Our own work in Singapore with the Seeker module in SenseMaker™ conforms with all three requirements and there is other good work in this area.
One of the illustrations James used is prediction markets. There are many variations on this approach, but generally a group of people treat an outcome or range of outcomes as a market and invest/take options etc on outcomes. The argument is that prediction markets outperformed (to take one example) opinion polls in US elections. Now this was true, but I am not sure if it still is. When prediction markets started they were novel, new, now one knew how they worked so they did (work that is).
However, the results and the implications are now known, so it becomes more likely that prediction markets will be gamed, at least in some cases. We know for example that people want to support a winner, so placing more bets on a candidate will shorten the odds and increase the chance of election. This does not happen in the odds on a horse race as the horses are not influenced by the calls of the bookies. However in an election is more easy to manipulate.
In effect I am arguing that prediction markets can break the third rule. Because the results are visible, behavior is influenced. In effect the system is complex not chaotic. The agents are not independent, they influence each other and the system, and the system in turn constrains the agents. This means that the potential for both entrainment and unpredictability increase. Of course the very concept of prediction is anathema in a complex system I should emphasise here that I think prediction markets are a useful tool, but they are being over hyped, and badly so at the moment.
One might also argue that the second criteria is not present as people prepared to place bets on the outcome of an election may be an unrepresentative subset of the population as a whole. All of this comes before the fact the there is a strong argument to ban all opinion polls (and prediction markets) as they have too much an influence on the outcome of elections. While we are at it, maybe we should ban advertising and sound-bites, instead sending politicians out to speak live from the back of railway carriages or (as in my youth) from the back of a lorry. That way the politician had to engage with people, survive hecklers and generally display abilities to hold and sway a crowd. Such a crowd may well be wiser than the press, and the approach is certainly less expensive. Maybe the money now been spent on the US election could then be deployed to provide health care for the whole population rather than just to an insured elite?
However the relationship between the media and politicians is normally a far cry from wisdom, and the people are quiescent in what they accept. I return again to the Philosopher Kings …..