Sunday, July 17, 2011

Diversification slips away

For the most part, we value diversification.

The usual justification for this belief (assuming beliefs need justification--some would say beliefs are beyond justification) is that diversity reduces risk to the mean behavior. That is: if I have two quantities (also: behaviors), the mean is half their sum, but how stable is this mean position?

After all, we also value stability, from which we get predictability.

To get a handle on stability, we turn to an examination of how each quantity (also: behavior) influences the mean. So, like so many phenomenon that are sensitive to distance, the distance from either quantity to the mean is important. Outliers, a long distance from the mean, can pull the average off in their direction.

Do we want an outlier in control? Usually not.

We value the 'yellow stripe': something down the middle
Enter: diversification.

We seek more quantities (also: behaviors) for the composition of the mean. Perhaps there are a total of N. Now, each individual quantity is somewhat diluted by the presence of others. Indeed, we find that the prediction of the mean improves by square-root(N)

Ah, but wait! What if it's really behavior (not quantities) and the objects of behavior are a team of project practitioners. What then?

If the team is co-located, we can expect near-continuous collaboration, a ready exchange of ideas, casual debate on the pro's and con's, and over time two things will happen:
  1. Leaders will emerge; by corollary, followers will emerge in the '1 - L' space.
  2. Leader/follower relationships establish correlation and cause-effect: movement by one causes movement by others, usually in the same direction.

If we were doing regression analysis, trying to predict the next move, the r-squared coefficient would be near 1 vis a vis leader impact on follower.

Now, to the extent that leaders influence followers, we have a sort of coupling.  Effects are transferred from one to the other; errors are not trapped at the boundaries.

The overall effect is that diversification, which depends on independence, slips away.  It's as if the followers are subsumed into the leader orbit and we're back to having only a few elements to set the average.  The opinion of one becomes the opinion of all.

Also, stability is now vulnerable to the appearance of a new leader 'star'

So, be aware that in the drive for diversification, it's self-defeating to establish the circumstances and environment for tight coupling.  There's something to be said for the virtual team after all!

(Did I mention I was describing the loss of diversity in a sequestered jury?  Perhaps I failed to mention how we citizens of Orlando reacted to a recent murder trial outcome.  And, it wasn't because the jury was from Tampa)

 Bookmark this on Delicious  
Are you on LinkedIn?    Share this article with your network by clicking on the link.