Automation in the modern world is usually thought of as done by robots or creepily intelligent software, but it is also the way, as this feistily interesting book argues, that bureaucracies composed of human beings increasingly operate. Officials are required to follow recipes (or algorithms) of sorting people into discrete categories and pursuing strictly defined courses of action with no allowance for ambiguity or complexity. They are condemned, if you will, to think inside the box.
This is what David Boyle, an economist and former policy wonk for the Liberal Democrats, refers to simply as “tickbox”: what more usually is called “tickbox culture” or, in US English, “checkbox culture”. For him it covers not only the targets and key performance indicators of official bureaucracy, but phenomena such as pervasive employee surveillance, the culture war over “identity politics”, the rise of management consultancy, the fact that you can never get a simple resolution to your problem from a call centre, and why the trains don’t work. Boyle himself instigated the celebrated “passenger strike” on Southern rail in 2017, convincing most of a trainload of passengers to refuse to show their tickets at Brighton. In an optimistic conclusion, he even claims that Ludwig Wittgenstein was also an enemy of “tickbox”, while exhorting his readers: “Refuse to categorise yourself on feedback or monitoring forms.”
His most potent examples lurk at the heart of government, which he knows all too well, having been an adviser to the Cabinet Office during the coalition government. The appalling stories coming out of the UK’s immigration service since the Windrush scandal, Boyle argues, are in fact inevitable given the “tickbox” structure of the decision systems in place. Such rule sets, he says, “don’t ask why there are discrepancies, or whose mistake it might be, because that would require human intervention and nobody has time for that”. And so, as he shows, people working on asylum applications are incentivised to mark cases that contain such puzzling human detail as “non-straightforward”, which means they go to the bottom of the pile.
But it’s worse than that: in general, management by targets and tickboxes, Boyle says, simply means that stressed employees will find ways to game the system. (In the early days of targets for hospital emergency departments, some designated their trolleys as “mobile beds”.) Indeed, according to “Goodhart’s Law” (named after the economist Charles Goodhart), any measurement that becomes a target ceases, in short order, to become a useful measurement.
Boyle’s book is not a luddite rant against any kind of systematic decision-making. “Rigorous measurement” is of course a good thing, he notes, as long as you are measuring something that it is actually possible to measure, and that measure is meaningful to the task at hand. Surprisingly seldom are both conditions fulfilled: Boyle points, for example, to the late US defence secretary Robert McNamara’s ill-advised use of enemy body count as a metric of success during the Vietnam war, and cites an ingenious suggestion that, during the more recent war in Afghanistan, the best single measure of improvement would have been “the price of exotic vegetables, which showed more accurately than anything else how calm or otherwise the situation was”.
Boyle provides some quick history of the “scientific management” of Frederick Winslow Taylor and the rise in the culture of targets, “performance”, and “compliance” in government, the administration of which is itself eye-wateringly expensive. And he explains how the five-point consumer-satisfaction scale on those customer surveys was invented, by an American sociologist called Rensis Likert in the 1930s. One of the first widespread applications of this technique was to ask German and Japanese civilians how they had felt about being bombed during the second world war. (It turns out they didn’t much like it, but the effect on morale was never such that they were likely to rise up against their own governments.)
People have long dreamed of dehumanising decision-making, purportedly because it will lead to greater objectivity and justice, but at least as much because it prevents those who have, after all, ticked the boxes from being blamed for any bad consequences. But any such system bakes in the biases of the designers and the data it is fed: in 2016, notoriously, Microsoft released an “AI” chatbot called Tay on to Twitter, and then shut it down again within a day, because its cheerfully anarchic human interlocutors had taught it to deny the Holocaust. The clarifying virtue of Boyle’s often very funny book is that he extends this familiar critique of machine-learning systems to other equally algorithmic decision-making processes, even if they are performed by people shuffling paper. Perhaps it would be nice to get rid of messy human judgment completely, but we are eternally condemned to exercise it even if we really rather wouldn’t.
• Tickbox is published by Little, Brown (RRP £18.99). To order a copy go to guardianbookshop.com. Free UK p&p on all online orders over £15.