Criminal behavior is a natural result of metrics management. Yesterday’s episode of This American Life is a horrific example of that. It tells the story of how the much lauded introduction of metrics management into the police department in NYC lead ultimately to horrific criminal behavior by the police.
But that’s not exceptional. Why this happens is full of fascination. Metrics management encourages unethical behavior. In some cases this is intentional but I think that’s rare.
What happens is easy to explain if we set aside the question of ethical behavior.
The manager’s problem is always how to coordinate behavior toward his goals. If the existing behaviors are fine, and often they are, he want’s to add additional behaviors. Metrics management is tool for doing just that. He creates metrics that highlight the desired behaviors and then works to bend the curve. Just making the metrics visible is usually sufficient; but people tend to escalate so it’s common to pile on some carrots and sticks.
Lots of us who have used these methods have found them to be marvelously effective. Early in my work life, for example, I put a chart on my door showing the number of open bugs in the large software system we were building. The line wobbled along, moving up and down, over a period of maybe a year. That was, pretty much all I had to do. In the following six months the curve steady declined until it finally crossed zero. Managers, engineers, customers suddenly became focused on that line. They starting asking about stupid little meaningly wiggles in the trend. It was magical. When we finally crossed zero parties and tee shirts emerged.
Other things suffered. And that’s the thing.
Or no, it’s not. Presumably as things suffer the natural response would be to adjust. And yet people don’t.
There is a weird pattern here. Any organization is skilled at some things, call those A. If we had metrics for those it’s scores would be pretty good. But we typically do not have metrics for these. Not because we are incompetent managers, but because the behaviors that achieve A are so deeply embedded into the organization that they do not require close supervision. These behaviors are sustained as norms, manners, etc. These behaviors are so fundamental that they are part of the organizations social fabric. The social contract it has made with it’s self. This fabric is slow to change.
So when the metrics management card is played into the game, in service of adding or improving a set of behaviors, cal them B, a very odd thing happens. Of course on thing that happens is that B behaviors increase. If they didn’t managers wouldn’t use this trick. For a period of time the overall score A+B rises. The A behavior score does not decline (at least not much) and the B behavior score rises.
But in time the A behaviors start to decline. This alone is an interesting pattern and I don’t think I can recall any instance of metric management that didn’t manifest this pattern. At first, and surprisingly rapidly, everything gets better. This a quite delightful experience. It’s positively addictive. Later A scores decline, but since your not measuring them this tends to go unnoticed.
Except it doesn’t go unnoticed. Individuals in a social fabric are extremely aware of the social contract. If only because it’s enforcing feedback look is implemented thru millions of tiny social cues.
What appears to happen as metric management moves into a organization is that it displaces the existing social contract. This triggers a lot of cynicism. Which in turn undermines ethical behavior.
Exactly. “Tell me how I’ll be measured and I’ll tell you how I’ll behave” said Eliyahu Goldratt in The Critical Chain.
This was a major theme of The Wire (still the best TV show ever made), where it was referred to as “juking the stats“.
Mike – “teaching the next generation how not to do the job” — oh yeah.
Julian – It is amazing, how much the role draws forth behavior.
Pingback: Ascription is an Anathema to any Enthusiasm › Overton Window