One of the books I read in Belize was The Wisdom of Crowds by James Surowiecki
I was keen to read this book as it had the potential to be relevant at both a macro- and micro-level and books rarely hold this promise. At the macro-level it could address such issues as are we better off allowing important societal decisions to be made by small groups of experts (i.e a technocrat elite) or a larger group of less expert participants (i.e democracy). We live in a world where the media extolls the virtue of experts (CEO as saviour, CNBC market experts), so a book that claims experts are largely irrelevant was intriguing. At the micro-level the subject of group decision making ability is intensely relevant to how VCs operate. Ultimately, in a very Darwinian sense, VCs live or die by the quality of decisions we make. The outcome of any particular investment is probably 75% determined at the time we decide to invest (I shall write a longer post about this in the future). Also one of the fundamental premises of VCs is that a small group of experts can make better decisions than a larger group of less knowledgeable people.
The book does a good job of articulating both the conditions under which groups make smart decisions (diversity, independence, decentralization coupled with the ability to aggregate choices) and the different types of problems (cognition, coordination, cooperation). However ultimately I only graded it *** as after a bright beginning it seemed to lose its way in the final third.
The lesson that resonated with me most strongly was that one of the key to successful group decisions is getting people to pay much LESS attention to what everyone else is saying. This is counter intuitive but a simple experiment conducted by the economists Angela Hung and Charles Plott illustrates the point.
A group is presented with two urns: one has twice as many dark marbles as light, the other has twice as many light as dark, but they don't know which is which. The group is assigned one of these urns and their task is to decide which urn they had been assigned - the majority light or dark urn. The experiment was conducted under two different conditions. Under the first conditions, each member of the group sequentially removed a marble, looked at its colour but did not share this with the group, and then publicly voted (sequentially again) for whether the urn was majority light or dark. Each member was then financially rewarded based on whether their individual vote was correct. Under the second conditions, the procedural rules were the same, but each member was then financial reward based on whether the group vote (as decided by majority) was correct.
With just this small change in the nature of the reward, the group made significantly smarter decisions under the second conditions than the first. Why was this? Well whenever we are making a contribution to a group decision we have two sources of information - "public information" which is what is everyone else is deciding, and "private information" which is what we know but the group doesn't (unless we tell them). In this experiment, imagine that the first two individuals voted for the majority light urn. Now even if the third person draws a dark marble they are likely to still vote for the majority light urn. So an information cascade can easily occur and "public information" dominates "private information". Given the social pressures to conform we need to counteract this to preserve independence of thought among individuals and make sure the "private information" is disclosed and counted. By encouraging participants to use their "private information" individuals, while being individually more wrong, can create smarter collective efforts.
So if you want the groups you participate in to become smarter, be prepared to express your opinion especially when it differs from your colleagues. If you are responsible for setting the norms for how a group makes decisions, then think carefully about how you elicit private information from as many individuals as possible and prevent a few people from dominating the discourse.
Comments