Ken, a very interesting topic and a good article. In my current job, I get to deal with numbers and metrics on a daily basis and those who like to make "knee jerk" reactions based on the numbers. Numbers are comforting, they provide a certainty that a subjective "better" or "worse" doesn't always give.

The problem is that you're dealing with humans who collect, interpret, the numbers and decide which numbers are meaningful. Humans who (in my world) have pay increases directly related to the numbers they collect. So, there's an active incentive to report things in the best possible light when what's really needed is to see where and how things are going wrong or what else is going wrong because the number is better. Humans who actively manage the system to try to get the best numbers possible- even at the expense of the overall goal (often not realizing the effect on the overall goal).

I'll provide an example from my manufacturing background. "They" decided we needed to get into the 20th Century and measure Inventory Turns (essentially a metric to measure the dollar value of what we have sitting in the warehouse compared to what we ship out in a given month). "They" read in a management book that 9-10 inventory turns is a world class organization and the metric said we were at 6. So, the beatings will continue until inventory turns get up to 9. Well, our product mix was not changed. Customer lead times were not changed (to allow us to actually plan a schedule and run in campaigns of similar products and reduce setup time). The process was not changed or sped up. So, we ended up with a lot less inventory, a customer base used to getting things right away that no longer could, and constant break-in orders in production to keep a customer from running out of material. That's what blind management to numbers gets you from folks who don't get the numbers, process, or customer. I'm sure that staff folks have more military related anecdotes but I think the principles are similar.

Numbers can be incredibly useful. But some thought has to be given to how the numbers are obtained, what incentives you're providing (and are they all good?), what the numbers actually say (and what they don't say). Another fun exercise (probably easier to do in manufacturing than military operations) is to say, "OK, I think the numbers we measured are telling me "X". What else should I expect to see if "X" is really happening? Is that happening too? Or is it something else entirely?".

Just my two cents.

Ken