« Score Your Beliefs | Main | Why Refuse To Debate »

April 24, 2009

Comments

I'm in the middle of reading In Search of Excellence, which points out that many of the greatest large companies (at the time it was written) used actual internal markets. This involved ideas like allowing several versions of a new product to be developed in parallel and then subjecting them to a competition to determine which one would be the final product, or promoting those who did initial development on a new product as the sales of that product grew, or even treating other divisons as competitors. The companies studied also apparently had a strong tendency to avoid letting upper management make too many decisions related to things they didn't understand, no matter how much they heard from the employees on the front lines.

If this is a known and effective way to employ the knowledge of those who are closest to the task or customer, any company that tries to use highly centralized control is probably not all that interested in involving lower-level employees in decisions. Any gesture such as using a prediction market to allow everyone to have their input isn't likely to change the way things really work. Overcoming the information barriers created by a bureaucracy while preserving the bureaucracy seems difficult to do.

Prediction markets could be practical for things that aren't big enough to restructure the way decisions are made (such as determining if a deadline will be met), but that can still leave you with deadlines set by managers who are too high up to know how they'll be met. Another example from In Search of Excellence is that the companies that were studied tended to do many small experiments intead of trying to make long-term forecasts. In one case it was common wisdom inside the company (3M) that new uses for a product would often be found after it was selling, making market forecasts meaningless. If it's cheaper to find out how something actually ranks in quality and what demand really is than to have a long planning process, adding more inputs to the planning process won't change that.

In the end one of the central messages of the book is that great companies have little bureaucracy but heavy informal communication, which allows knowledge and experience to be shared in a way that adjusts easily to changing circumstances. An attempt to declare more communication from the top is likely to be based on what works well in some situations, but will still be in force long after the situation changes.

Rhodes' comments on controlling information reminded me of Janis' book on Groupthink. One of the problems leading to groupthink is that information flows within decision making institutions are not linear (passing easily up and down the chain of command). The use of prediction markets, I think, allows for the use of information that some stakeholder may have and not wish to disclose. Without disclosing the information, the stakeholder can use that info to make their prediction. When a large enough group of people make use of their private information as predictions, then it may be possible to reach the best decision, without the stakeholder giving up control over their privately held information.

See also the SNAFU principle.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30
31