Wisdom of the Crowd

As a follow-up to my Patent Office use of the ‘wisdom-of-the-crowd’ post:
I went to a Web 2.0 Conference in Virginia a while back. I hopped the ‘business’ freight.
James Surowiecki, author of ‘The Wisdom of Crowds’, described the anecdotal evidence that forms the backbone of his book.

The internet has enabled a method to capture large scale predictions on multifarious issues from political races, Oscar winners, Stock Market, and sports. Without exception, surveys that follow the rules below out perform ‘experts’ in the field and surveys that are statistically derived.
Evidence was also provided to support the notion that small group dynamics should follow a similar ruleset for maximum effectiveness.

Conditions for success:

  1. Tool of aggregation – Be specific during the collecting process. Voting, averaging, betting, or prediction markets should be used over an open-ended suggestion box.
  2. Diverse populations – Smart groups tap a wide range of problem solving (not just high IQ, or perceived experts). This excerpt from Scott E Page, University of Michigan – Ann Arbor, Professor of Complex Systems, paper ‘The Interplay of Difference’ summarizes his findings:
    ‘Diverse collections of agents perform incredibly well. In fact, under a set of fairly unrestrictive assumptions, we find (and prove) that a relatively small group of randomly chosen diverse but intelligent agents will outperform a group of the best agents on a hard problem.’

  3. Independence – The group must not influence each other – avoid peer pressure

Leave a Comment


Leave a Reply


John, thanks for the summary.


I too love the book, love the concept, and can imagine its relevance in government.

The question is how are we going to apply it in government? Any thoughts?

John Gladstone

A few come to mind. Since efficacy seems linked to scale its best to capture these pearls of wisdom as a sub-set of standard user interaction with your web site. This site from HHS uses a dynamically generated list of the top five searches of the database. Weblogs could be parsed at intervals to point to the most popular portions of your site. A Digg-like thumbs-up/thumbs-down could be used as an informal survey vehicle bypassing the dreaded and arcane Paperwork Reduction Act.