Google+ Followers

Monday, March 2, 2009

Fuzzy Dots are ... well ... Fuzzy. Let's keep them that way.

In College I studied Philosophy and Accounting. An odd combination, but there is an intersection with regard to Fuzzy Dots.

In Philosophy class, we learned that if a dot is truly fuzzy, not just blurry because of your eye piece, then no matter how fine a microscope you use, it will always be fuzzy.

Over in the business school -- and now in the business community -- I see a lot of business analysts trying to get better microscopes to increase their resolution and decrease the fuzziness around their dots. Now, if you are doing accounting of historical numbers, this may be a good idea. There is one true number that you can solve for. However, in many cases, the number the analyst is pondering is itself a fuzzy dot. Estimating next quarter's earnings, the NPV of a capital project, the ROI on an investment, the cost of a major project or the completion date a of major project are examples of numbers that are fuzzy dots.

To resolve them to a clear dot is to ignore their underlying reality which only creates an illusion of certainty. Unfortunately, many enterprise planning applications require the analyst to put in a 'naked number' for a fuzzy dot rather than a more realistic probability range. Worse, many managers require their analysts to just give them a number. See "Why Can't You Just Give Me the Number?" (Patrick Leach, 2006) for an excellent summary of this issue.

Many would prefer to have the illusion of accuracy (the NPV will be $20.687M) fostered by false precision. The only thing we know about such numbers is that they will be wrong: we just don't know by how much and in which direction. For further reading, see Phil Rosenzweig "The Halo Effect" (New York: Free Press, 2007).

No comments:

Post a Comment