Measuring The Damage

This piece originally appeared in the IPA Review.

Before coming to the IPA, I had a small role in a Victorian Education Department team implementing Gonski school reforms. The department developed school performance targets and a reforms package that would, among other things, help schools identify their strengths and weaknesses. To what extent would the reforms move schools closer to the targets, the minister’s office asked? So we then had to somehow score the reforms against the targets, and estimate how much progress would be provided by funding.

The folly of such exercises is well-captured in Jerry Z. Muller’s The Tyranny of Metrics, released last year. Muller, an historian at the Catholic University of America, argues that metric fixation has overrun bureaucracies, public and private, distorting their behaviour and ultimately frustrating their purposes. Metric fixation replaces experience and discretion with institutional targets and measures, argues that all inputs and outputs should be reported (transparent), and connects rewards and penalties to performance against metrics. The result is that institutions pursue only their most obvious and measurable tasks, leading to the corruption of their internal information flows and ultimately to waste and inefficiency. Muller observes that “measurement may become counterproductive when it tries to measure the unmeasurable and quantify the unquantifiable”.

Not all metrics are useless. Muller notes accurate measurement against readily identifiable ends is desirable when possible. The trick is to distinguish between good and bad uses. To this end, he dedicates the bulk of the book to case studies taken from across society, demonstrating the problem is not limited to the public sphere. Businesses and charities have been just as charmed by made-up numbers and phony rigour as schools, universities, medicine, policing, and the military.

From these cases we learn the incorrect use of metrics reinforces the false belief that more is always better, in place of the law of diminishing returns. For example, if having a certain number of university graduates is good, then twice as many graduates is twice as good. We learn too that all metrics can be gamed—and will be. Standards will be lowered, quantity assumes priority over quality, and data will be skewed or simply invented. As an institution retools for maximising whatever the metrics measure, finite resources will be reallocated towards them and away from other, potentially more important, work. And if in the end, after all the fakery and wastage, results do not budge, then the metrics will be refined and the process begun anew as “the effort demonstrated in gathering and publicising the data satisfies a sense of moral earnestness” and “the progress of measurement becomes a simulacrum of success”.

Nonetheless, Muller does offer guidelines for how and when metrics might be deployed profitably. Notably, Muller makes the unfashionable point that transparency can distort the decision-making process by encouraging image-consciousness and risk aversion. Another point he makes that deserves greater recognition is that as bureaucracies and their actions become more complex, so their costs rise. Standardisation can be costlier than ad hoc procedures as measures must be developed, and data gathered and analysed. This is a boon for the administrators but an impost on everyone else. Muller concludes metrics are best used where goals are easily identified, data is readily available, functions of the institution are limited, and targets are developed from the bottom-up to capture what workers already know about their jobs.

This last point reveals something of Muller’s underlying philosophy. In a previous book, Conservatism, Muller explored a sceptical form of conservatism that he traced from the empiricist philosophy of David Hume (1711-1776) to Michael Oakeshott (1901-1990) and FA Hayek (1899-1992). In his attack on metrics, Muller argues that metrics recreate flaws that Hayek identified in command economies. The substitution of management’s goals for established processes erases the practical knowledge developed by workers and inhibits refinement and organic improvement. Indeed, for any process more complex than a production line, rationalisation towards the abstract goals of management is likely to be self-defeating, as any efficiency gained towards these narrow ends is swamped by inefficiencies elsewhere in the institution.

Given all the above, we might wonder why metrics have become so popular. As Muller recounts, in the late ’80s the UK’s Thatcher government pioneered ‘performance indicators’. This lead the historian Elie Kedourie, a colleague of Oakeshott, to observe that this otherwise inexplicable departure from stated Tory principles must have been an “automatic response to the spirit of the times”. Kedourie was surely onto something, as the decades since have seen rationalism run amok.

In even the smallest and most trivial aspects of our daily lives we are beset by false precision. Your oven cannot be set to 225 degrees, your phone does not have 12 per cent battery left, ball-tracking does not know whether that LBW shout was too high or not. We see too this mendacity at the highest levels of expert analysis. Despite the breathless reporting of laboratory experiments, it is not possible to say that the child who takes the lolly instead of waiting is less likely to be successful in life, or that sugar increases your chance of a heart attack by 31.497 per cent (or whatever suitably non-round number is published this week).

Similarly, the gullible Professor Steven Pinker should not claim the world is getting better based on numbers produced by the Chinese Communist Party, and for the record, it simply cannot be known how teacher training today will affect children’s scores on a maths test in 10 years’ time.

Our fascination with these numbers reveals something wrong deep in the heart of modern life. At one point, Muller observes that metric fixation reinforces the belief that all problems are soluble, and that, as such, it is more like “faith” than “science”.

Behind it all—the media releases, the consultations, the spreadsheets—is the absolute conviction that reality can be made to conform to abstractions, and that if everything were just so then everyone everywhere would be happy. This is a dangerous sentiment. It is not just that our freedom lies in informality, though of course this is true—just contemplate the coercive potential of China’s social credit score, which reduces life to a points tally.

Even more importantly, contained within the experience that abstraction displaces is everything that makes virtue possible. In the accumulated practical wisdom of our culture we can find what we need to know about living good lives; and in the vastness of that experience and the limits of our ability to understand it all, in its nontransparency and its complexity, we find our reasons to tolerate and forgive one another.

Muller writes that “demand for accountability waxes as trust wanes”, and his insight is broadly applicable. Measurement and standardisation have come to dominate our society. For a long time, we have standardised success, glorifying a particular kind of cosmopolitanism and dismissing suburban life as some sort of pitiable remnant. More recently, we have begun to standardise diversity, too, with aggregate statistics forming the basis for redistribution between formalised blocs of people. Social justice is now as much about mathematics as morals.

In this, our lack of trust in one another is supposedly sublimated by mutual subjection to the numbers. But when the institutions that structure our lives are constantly innovating, and when they are the product of conscious construction, we cannot know what we ought to do. Nor, ironically, can we tolerate departures or forgive transgressions, because the rules are now scientific, or democratic, or both, and therefore inflexible and unchallengeable. When experience is abstracted away, institutions that were once wise and tolerant become stupid and harsh.

Rampant rationalism is therefore both a cause and a symptom of a collapse in social trust. Just as a school’s operation is disrupted by targets imposed from outside, so too is a culture disrupted by tinkering from above. The one produces uneducated children, the other unedifying lives.

We must not reduce this phenomenon to a calculation problem. If we make merely an empirical claim—that some scheme is impossible—then we invite the attempt. Instead, the tyranny of metrics must be met with a moral claim: we are better off, as humans and as societies, when we affirm and pass on what we know, and which is encoded in our institutions, and this process can only be corrupted when we pretend to know the unknowable, whether that is the optimum operation of a bureaucracy or the realisation of utopia itself.

The flipside of the limits of our reason is the real value of inherited wisdom and practice, which reveals something true and good and useful about the world. We have placed our trust in numbers, and they have betrayed us. The only solution, then, is recognise again the truth where it lies.