Cognitive Design

Prediction markets have emerged as crisp architecture (structure and dynamic) for designing artifacts with collective intelligence. On-line trading sites let us buy and sell shares on an uncertain outcome (e.g. who will win an election, which product will be successful, if a stock will go up or down tomorrow, etc.).  The outcome with the greatest share price at the end of trading reflects the market’s prediction.

This is how the capital markets work as the stocks with the highest share prices reflect the market’s prediction of which companies will generate the most free cash flows in the future. 

The prediction market will on average be more accurate than the best expert judgment and therefore the claim that intelligence emerges from the collective trading activity by those in the market. Note this is not a form of collaboration or brainstorming where people get together and try and figure through conversation or group problem solving which outcome is most likely. Instead it involves little to no direct communication and relies on isolated individual judgment being aggregated and reflected as a single piece of information – the price of a share. 

Successfully designing prediction markets requires a careful understanding of the underlying cognition at work.

John McQuaid makes this point well in his recent Wired essay, Prediction Markets are Hot, But Here’s Why They Can Be So Wrong.

Two key points McQuaid makes is that prediction markets require a diversity of individual cognition to work (insider, noise traders, etc.) and that they need to have real skin in the game (e.g. significant dollars or stakes at risk).  Without those factors traders are not motivated to seriously focus nor can the market aggregate the diversity of knowledge needed to outperform individual experts.  Excellent advice on the cognitive design factors for prediction markets!