Archive for December, 2007

I found this a little bit like “The Tipping Point” in that it’s full of interesting anecdotes, many of which I’ve heard before, but by the end of the book I’m not clear that it all adds up to something. So doing a book review was helpful. At some level, this is a book about economics – how we synchronize our activities, often via money. I started off on audiobook and that was somewhat enjoyable. The author reads the opening, and then a professional voice reads the book. He reads fairly slowly though, so I ended up switching to the book itself.

At the book’s core is the contention that under the right circumstances, groups can be smarter than the smartest individuals among them, and the book is about the situations where that is true. The author supports the anecdotes with academic studies.

In the opening the author identifies 3 types of problems:

Cognition problems – ones that have right answers (at some point in the future, the answer can be judged correct or incorrect), like who will win the Superbowl or who will win the next election. Voting is an established way of aggregating judgments.

Coordination problems – getting people to work together like traffic, or connecting buyers and sellers. Coordination problems can be solved by everyone working in their self interest once the rules are set up – like traffic driving on the right hand side of the road in the US. Social Norms and conventions minimize coordination problems.

Cooperation problems – people need to work together like a in a tax system, or a salary negotiation. Cooperation problems require a broader definition of self interest and the use of trust. Fortunately, studies show that people care about fairness and strong reciprocity is a deeply embedded norm that helps with cooperation. Capitalism needs transparency & trust to minimize transaction costs.

There are three conditions that help a group be wise:

Diversity – particularly in thinking and experience. Someone less “smart” can make the group smarter because they’re less redundant. Diversity is more important when trying to generate alternatives than when trying to choose among a fixed set, usually in cognition problems. Diversity can actually cause problems in solving coordination problems (fewer shared norms & conventions) or cooperation problems (less trust).

Independence – members of the crowd need to rely on their own thinking and experience to make their decision, rather than changing their opinion based on what other members of the crowd think. While this is critical for cognition problems, it’s less helpful for coordination or cooperation problems.

Decentralization – related to the above two, decisions are made where the knowledge is. For good decision making in large groups or communities, aggregation of the decentralized decision making is the next key. It’s important to aggregate judgments, not just raw data, to maximize taking advantage of decentralized knowledge.

Dangers that make groups unwise:
Lack of diversity and independence can lead to groupthink and herding. Poor leadership can create confirmation bias – a tendency to evaluate new data against an already made decision. Dissent, conflict and independence are key to good decision making. Too much dependence in our decision making can lead to bubbles and crashes.

My net: Diversity is key to good decision making, but making decisions at scale requires coordination and cooperation which is challenged by diversity. Figuring out how to aggregate the diversity into coordinated cooperation without losing the value that was generated is a key challenge we face

Read Full Post »