A Familiar Asymmetry

Every administrator has felt it. The metric was supposed to describe the work; over time, the work begins to describe itself in terms of the metric. Citation counts replace reading. Response times replace care. Quarterly revenue replaces the slower, less legible question of whether the firm has any business remaining in business a decade from now.

Charles Goodhart’s observation — that any measure, once it becomes a target, ceases to be a good measure — is by now a cliché. What is less appreciated is why the law operates with such reliability across domains. It is not, in the first instance, a story about cheating. Most actors are not cheating. It is a story about attention: about which questions a measurement system makes easier to ask, and which it quietly retires from view.

The Shape of the Distortion

Three features tend to recur whenever the metric begins to eat the mission.

The first is compression. A continuous practice — teaching, research, clinical judgement — is collapsed into a small number of numerical signals. Information is discarded, but the discard is invisible: the dashboard does not show the things it cannot show.

The second is proxy drift. The chosen indicator was once correlated with the thing one cared about. Over time, the correlation weakens, because actors learn to maximise the indicator directly, without bothering to produce the underlying good. The map and the territory move apart.

The third is temporal collapse. Measurement systems privilege what can be observed now, in the current reporting period. The slow-maturing investment — in trust, in tacit skill, in research that may not pay off for a decade — is starved.

None of these failures requires bad faith. They emerge from the structure of the system itself.

What an Editor Can Do

We do not pretend that Performance — a journal whose very name embraces measurement — has solved this problem. We have not. The aim of this issue, and of the issues to follow, is more modest: to host a sustained conversation among economists, computer scientists, and management scholars about the design of measurement systems that resist their own pathologies.

The lead article in this issue, on the mathematical paradigm of knowledge-work performance evaluation, treats the question formally: what is the optimal estimator under a given information set? The essays that follow this one take up the question philosophically, ethically, and historically. The two registers belong together.

A metric eats a mission when nobody is allowed to remember what the mission was. The work of editing — and, we would suggest, the work of management — is in part the work of refusing to let that forgetting happen.