Muller, a historian, argues that organisations across sectors have adopted a belief he calls metric fixation, and that it damages the activities it is meant to improve.
He defines metric fixation as a cluster of beliefs: that it is possible and desirable to replace judgement based on experience with numerical indicators of comparative performance, that making these public ensures accountability, and that motivating people through rewards and penalties attached to measured performance improves it.
The failure mechanisms he identifies are specific. Goal displacement occurs when people pursue the measure rather than the purpose it was meant to represent. Measuring the most easily quantified aspects of a job causes the rest to be neglected, and the easily measured is rarely the most important. Short-termism follows because measurable results arrive faster than the outcomes that matter. Gaming takes several forms: creaming, where organisations select the easiest cases to improve their numbers; lowering standards to raise pass rates; and outright falsification.
The case material covers education, where school and university rankings changed institutional behaviour without improving learning; medicine, where surgeon mortality rankings caused surgeons to refuse high-risk patients; policing, where crime statistics were manipulated through reclassification; the military, where Vietnam-era body counts became the measure of progress; business, where pay linked to short-term financial metrics produced value destruction; and foreign aid and charity.
Muller does not argue against measurement. His position is that measurement should support judgement rather than replace it, that metrics should be developed by the people doing the work rather than imposed, and that they should not be tied to reward and punishment.
The argument is not new, since it restates Campbell's law and Goodhart's law, and the evidence is a collection of cases rather than a systematic assessment of when measurement helps. Muller acknowledges this and the book is short and clearly argued.
Executives designing performance measurement and incentive systems who need the specific failure modes catalogued before they build one.
Chosen by shared topics, the entries that hold them, and the people who recommended both.


