Metrics as Monkey’s Paw

It’s been noted that quality programming is hard to measure, and it’s no surprise that companies measuring programmers based on the volume of code turned out have larger products but not necessarily better ones. I’m far from the first person to note that you get what you measure, but despite that people continue to use lousy metrics.

Perhaps the notion that form without content is void should be considered a special case of you get what you measure: if you demand well-structured, well-argued essays without measuring their content, you’ll end up with essays lionizing Nazis as intellectual heroes; if you demand a certain number of abstracts per week according to formula, you’ll get that number of abstracts but they will have no value for their intended audience.

In both cases, one must ask whether the goal can be measured appropriately, and if you’ll be satisfied when you get exactly what you ask for.