Is it time for the board to be replaced?

Here’s a thought…how good are boards of directors at making predictions about the future?  This is essentially their job, anticipating the future and what investments should be made as a result. But is the boardroom an effective forum to make these types of decisions? Could there be a better way? Could prediction market science transform boardroom decision making?

For the past year or so we have been studying the science of prediction, trying to identify the ingredients for making effective predictions about the future. This has involved studying the effectiveness of small groups of people making different types of predictions. They are given the opportunity to make these decisions using a variety of approaches, allowing us to observe their collective ability accurately to look ahead.  

What I conclude (and I am sorry to disappoint those hoping I would say the opposite) is that a well-constructed board of directors with a mix of independent thinkers appears to be pretty close to an ideal model for effective decision making. Why? Well, they are about the right size: prediction market science tells us that collective forecasts made by groups of individuals with access to the same information peaks out at around 16 people, pretty close to the size of an average board. The mix of executive and non-executive opinions provides independent and unbiased input. It is the wisdom of crowds approach in action.

However, boardroom decision making processes are not perfect. They suffer from one underlying flaw, which is the impact of herd effects. Perfect prediction protocol relies on individuals making completely independent and unbiased judgements and we all know that doesn’t always happen when emotionally delivered arguments and partisan loyalties are involved.

In our simple experiments upwards of 15% of group predictions went off course as a result of herd decision making, one person’s opinion having a cascading influence on the others, so I think it’s likely to be quite a significant factor in board room decision making too.

During the course of the research we have found some solutions to the problems of herd effects that potentially could be applied to make board room decision making more predictive.

My first suggestion would be a move from voting to betting. Give each board member a budget, just as in predictive markets trading. If the issue they are voting on allows for measurement, and proves to be the right decision in hindsight, those who were correct are rewarded in some fashion. I am not sure what that might be, although a harsh chair might consider removing and replacing those who consistently make the wrong decisions.

There is published evidence that demonstrates that prediction market trading techniques deliver more reliable predictions that simple polling or voting as a board tends to do. 

Next think ‘dark trading’ where the voting/betting process is conducted in two rounds, the first secret. Once everyone’s bets are revealed, people may change their minds, and also raise their bets if they wish. This has proven to deliver more reliable forecasts than open betting from the start because it prevents herd behaviour.

I would recommend establishing a series of company-wide prediction markets made up of an even broader cross section of individuals in the company to make decisions before the board votes on key issues. Using the wisdom of crowds approach, it allows for the sharing of opinions and experience, condensing the knowledge from across the whole company. All our research shows that good information sharing among the decision making group is the key factor in making effective predictions.

Finally, in an ideal world, I would recommend having three boards. I would ask each to make exactly the same set of decisions and then compare the results. If they all reach the same conclusions, there is an 87% chance that their prediction will be the best it could be. If there is conflict aggregate the votes, or consider opening it up to future debate. In our experiments this improved prediction accuracy by 13%.

Jon Puleston is vice-president of innovation, Lightspeed GMI