Measuring performance using single criteria

Defining a single metric to measure the performance of a person, a company, or a project, will incentivize people to behave in ways that may not be aligned with the ultimate goals of the organization[@ridgway1956Dysfunctional Consequences of Performance Measurements].

There is an interesting anecdote in India, in which the British government offered money for dead cobras as a way of eradicating them from cities. This, however, led people to breed them in order to collect the bounty, exactly the opposite from what the government wanted. Today, this is called the cobra effect.

If you judge programmers by the number of lines of code they write, you may arrive to the famous situation of the Negative 2000 lines of code of Bill Atkinson that forced managers to change their approach at judging contributions.

In companies, some of the single-metrics that can be used are cost-per-product, profitability, overall production over set periods, like a month, and therefore neglecting maintenance, for example.

In academia the problem is visible with the problems with citation-based metrics. Mostly because citation-based metrics incentivize competition over collaboration, and incentivize people to work in non risky fields (see: Scientific Exploration and Play).

This is something that is also discussed in Build - Tony Fadell, regarding the incentives of sales teams that may be too short-sighted.