Platform Mechanics Entry #0283 Classified Declassified

The reason a metric becomes a target and then a lie

Any measure good enough to manage by is good enough to game, and the gaming arrives through exactly the same channel as the improvement.

No visual record attached The written record below is complete.
Plate 16 — a dashboard rising while the quantity it stands for does not.

Intuition test — answer before you read on

A metric rises by thirty per cent after it becomes a bonus target. What can be concluded about the underlying goal?

A support team is judged on tickets closed per day. Within a quarter the number rises and the customers are angrier. Nobody cheated in a way an audit would find: long problems were split into several tickets, and every ticket was closed on schedule.

What everyone sees

The usual diagnosis is bad people or a badly chosen measure, and the usual remedy is a better one. Managers replace tickets closed with satisfaction score, then with resolution time, treating each failure as evidence that the previous measure was simply the wrong measure.

What is actually happening

The failure is structural. Goodhart described it in monetary management and Campbell in social programmes: once a statistical regularity is used for control, the relationship it rested on decays, because effort flows to the measured proxy rather than to the unmeasured goal. The proxy stays honest as a count and becomes false as a description.

Why it stays hidden

Two things conceal it. The early phase looks like success, since real improvement and gaming both raise the number and are indistinguishable on a dashboard. And whoever chose the measure now owns it, so the incentive to reinterpret decay as progress sits with the person best placed to detect it.

A measure used as a target stops describing the thing it was chosen to describe.

A measure used as a target stops describing the thing it was chosen to describe.

The hidden part — entry #0283

Collect this card

A measure used as a target stops describing the thing it was chosen to describe.

0 / 10,000 collected

Sources & further reading 3
  1. Goodhart, "Problems of Monetary Management: The UK Experience", 1975
  2. Campbell, "Assessing the Impact of Planned Social Change", Evaluation and Program Planning, 1979
  3. Muller, "The Tyranny of Metrics", 2018

Circulate this file

Annotations are reserved for archive members.

Sign in to annotate