The tyranny of data puts us at the mercy of a new Covid priesthood
Politicians have shown themselves incapable of interrogating statistics or those who wield them
As if we haven’t had enough of graphs, figures and numbers to last a lifetime, yesterday was World Statistics Day. We were invited to prostrate ourselves before the new god, Big Data. I must confess to having been unaware of this occasion until alerted to it by a colleague. It had passed me by, along with 99.99 per cent of the population, though I do not have the evidence to support that assertion.
Perhaps there was bunting up around the globe as people paraded through the streets bearing photos of their favourite number cruncher. Let’s face it, with Hallowe’en, Bonfire Night and Christmas cancelled this year, it might have been the last chance for a good knees-up. Even worse, World Statistics Day only comes around every five years, so we may never get another opportunity to celebrate – especially as the data keep telling us we will all be dead by then.
The theme of this UN “awareness” event was “Connecting the world with data we can trust”. Accurate statistics are, of course, important. Governments play fast and loose with them at the best of times, and beady-eyed regulators are needed to keep watch on how they are being manipulated to suit particular purposes. This matters because our lives are now almost entirely in the hands of data analysts and modellers, whether in the public or private sectors.
It is claimed that the better the data, the more effectively targeted the intervention. Its collection and analysis can ensure the state directs scarce resources towards the right people; companies can better serve their customers; we can be reminded of that film we once thought of watching but forgot about, or encouraged once again to look at that Italian resort we fancied visiting a year or so ago.
Big Data will make us healthier by identifying those most likely to suffer from cancer or diabetes or heart disease. Big Data show trends, associations and patterns that govern human behaviour and interactions. But it is what you do with it that matters. Context is all. After all, a lot of numbers is just that: a lot of numbers. The conundrum is how to interpret them in a way that does not merely reflect preconceived ideas of what should happen.
The great cascade of data for Covid is a classic of its kind. What do all these figures mean? We are assailed with graphs looking like a silhouette of the Alps produced by boffins with worried looks and politicians with furrowed brows. Cases are now at 50,000 a day; they are 200 per 100,000; 350 people are in hospital; ICU beds are filling up. It’s all terrible.
Except, there is no context. The raw data tell you nothing. Set against a population of 65 million, the incidence of infection is tiny. Measured beside the 12,000 deaths that take place every week, 50 or even 100 a day from Covid is minuscule. When you discover that the median age of those who succumb is 82.4 years, which is higher than average life expectancy, what does that piece of data tell you? Moreover, if the increase is really a spike in positive cases predominantly among students, the majority of whom will get the infection only mildly, if at all, what conclusions do you draw?
In the days before Big Data, when the amount of information available to policymakers would have been fairly small and the consequences of locking down the economy unacceptable, the response was entirely different. The flu pandemic of 1968 affected millions and killed thousands, but there was no suggestion that the country should be placed in a deep freeze while it was suppressed. Looking through the parliamentary archives in Hansard, I have struggled to find the outbreak mentioned at all beyond a few questions about NHS preparedness. Even as recently as January 2000, the year of the last serious flu epidemic, no one suggested lockdowns, tiered or otherwise.
The then health secretary, Alan Milburn, reported in the Commons that the worst affected regions were the North and the Midlands, with 600,000 people attending A&E and 200,000 emergency admissions in the previous three weeks. “The evidence is that the patients who are being admitted are more ill than normal and are staying longer than normal … As is usual, and as had been planned, most hospitals have undertaken little routine elective surgery over the past few weeks so as to be able to concentrate their efforts on emergency cases.” There were, he said, only 22 ICU beds available in the entire country. Sick patients were left to die in waiting rooms or parked in converted operating theatres.
Of course, there are differences.
Coronavirus is not the flu, though for most young people its effects are less debilitating. There is a vaccine for flu and not for Covid. The latter is new and, to begin with, its lethality was unknown but has mercifully turned out to be much lower than feared. Yet these factors alone do not explain the diverse approaches of today’s governments (not just here but across western Europe) and those of the past, when life carried on pretty much as normal.
The real change is the tyranny of data and the modern worship of metrics. Our politicians are confronted daily by lots of very large, frightening figures and they also operate on the basis of the “reasonable worst-case scenario”, a relatively new doctrine used to govern pandemic responses. The basis for that scenario is statistical modelling, which does not produce a prediction but a possibility dependent on a range of other factors that may or may not transpire.
Though they pretend to, few politicians really understand any of this stuff. They cannot ask the right questions because they are not sure what assumptions have been made in devising the methodology. They are in thrall to the controllers of the algorithms, the new priesthood, just as in the past if you didn’t speak Latin you could not understand the creed but just accepted it.
At a parliamentary committee hearing on Monday, Sir Patrick Vallance, the chief scientific adviser, said his greatest regret at the outset of this pandemic was a dearth of data. “We were flying blind,” he said. Yet we might have done better without it and just dealt with the reality in front of our eyes. Perhaps the UN could institute a World Common Sense Day. I’d celebrate that.