Good in Theory

Leon Martell, of Duck's Breath Mystery Theater, once told me a story about a meeting he and some other scriptwriters had with a Producer. The producer had read their script and pronounced it "not good enough."

"Wait a minute," somebody said. "This script is better than the last three movies you made."

"Yeah," said the Producer. "But it's not better than the scripts for those movies. Movie making is a tough business. Scripts get cut for length, actors mangle the lines, somebody decides that a different ending will test better, all kinds of crap happens. I don't need scripts that are merely good. Those I have plenty of. I need scripts that are so good that we'll still have a good movie no matter how much I screw with it."

Honesty is pretty refreshing, huh?

Metrics Management sounds like a good idea. After all, management is about controlling an organization, and having data about how the organization is functioning should help in managing what is going on. That's basic engineering control theory, and I suspect that you can trace a lot of metrics-based management theories back to General Electric and other companies where a lot of the managers began as engineers.

But people aren't servo motors and business metrics aren't really control signals, either. Focus on a single metric and you're going to optimize for that quantity—at least in the short run. Of course, what a company would really like to focus on is making money, but there is seldom a direct link between any given action taken by an employee and the profitability of the enterprise. If I were a management consultant I'd call that "too many transitional states between intermediate state variables," which is the same as saying "many a slip between cup and lip," and almost as informative.

Attempting to focus on money alone can lead to Joel Spolsky calls "The Econ 101 Management Method."

Spolsky notes two significant problems with this approach. The first is that it substitutes extrinsic motivation for intrinsic motivation, i.e., the natural desire that most people have for doing a good job is trumped by external financial motivation of incentives. This has always been a problem for salesmen on commission, who try to optimize their commissions even if the company loses money on the sales. Econ 101 Management spreads this problem to a larger group of people.

The external incentive problem doesn't apply only to financial motivations, of course. Yelling at people is a form of motivation, as is threatening their jobs. The point is that taking people who take pride in their work and undercutting that motivation with other, controlling, incentives can have pernicious effects.

The other problem Spolsky notes is even more generic: working to a metric encourages employees to "game the metric." They will alter their working methods to affect the metrics without necessarily improving the work. Software bug metrics therefore encourage either lumping several bugs into a single bug report (making them harder to fix) or simply failing to report the bug. Customer support personnel wind up either passing disgruntled customers off to someone else, or lying to them, or managing to avoid picking up the calls in the first place.

Many managers start off with an adversarial attitude between them and their employees, and there are few things that will kill pride of labor than having someone always trying to beat down your wages and motivate you through fear and intimidation. Add a little bit of ideological muzziness to the deal and you get the bizarre belief held by some managers that metrics + incentives will substitute for training and organizational support.

So the bottom line I seem to have gotten to is this: out in the working world I have often been amazed at the degree to which professional pride has overcome enormously poor managerial decisions, to let projects squeak through on the sheer competence of the people doing the work, as opposed to the incompetence of the people setting strategy and making decisions. The employees are generally better than the companies that employ them.

Just as Americans are currently so much better than their leaders. Someone should do something about that.