Measuring the of a public service was intended to replace trust with evidence.
It produced a large literature on what happens when a measure becomes a .
The pattern is consistent across health, policing, education and welfare administration.
Performance on the measured dimension improves, and performance on unmeasured dimensions falls.
An ambulance service judged on response times will reach more patients within the and may reach the sickest later.
Hospitals judged on waiting lists will treat the easy cases that clear the list quickly.
None of this requires cynicism from the staff involved, which is the part most commonly misunderstood.
A reallocates attention, and attention is finite in any organisation.
follows when the is attached to a consequence and the definition can be interpreted.
a patient, a crime or an examination entry changes the number without changing the world.
the definition rather than the number is the response that works and the one that is rarely funded.
A second family of remedies reduces the weight placed on any single figure.
Rotating indicators, unannounced sampling and qualitative inspection all raise the cost of gaming relative to the cost of improving.
They also make the system harder to summarise, which is why they are politically unattractive.
A minister can defend a single number in an interview and cannot defend a portfolio of partial evidence.
The deepest objection is not technical but constitutional.
Choosing what to measure is choosing what the service is for, and that decision is made by whoever designs the framework.
It is therefore a political act performed in a technical , which is the most durable way to avoid being argued with.