I went climbing with Dominic Cummings last week. Not literally of course. I don't imagine he has much time for leftie remoaners like me. And I'm picky about climbing partners, favouring those who I don't think will make things worse in an emergency. But it was cold and wet, so, between tedious training exertions on an indoor climbing wall, I thought his written enquiry evidence might be interesting. It was. Excellent on many points, but I was more intrigued by some others.
Firstly, he claims that the infamous report 9 from Professor Neil Ferguson's Imperial College Covid response team had almost no influence on the decision to lock down. That was all down to Cummings and the 'incredibly able people' he recruited to advise him, independent of SAGE. The advice from these 'top people' was apparently so stark that between 12 and 19th March Cummings repeatedly warned the prime minister of the risk of a 'zombie apocalypse'. On the 18th he was telling the PM to lockdown London immediately or the 'NHS in London collapses in 15 days'.
What was actually happening at that time was less dramatic. The best evidence suggests that new infections had almost certainly peaked sometime between 15th and 18th March. Earlier in London. These figures come from my published work on inferring infection rates directly from deaths, and separately on replicating Imperial College analyses, to remove some biases built into their models. Also from the REACT-2 study's reconstructions, based on asking randomly sampled people, who had had confirmed Covid, when their symptoms started. They also align with the pattern hinted at by the drop in calls to NHS 111 for respiratory problems, and by google mobility data showing a huge drop in mixing well before 24th March.
Despite being critical of the governmental pattern of 'debacles brushed under the carpet and no serious learning', Cummings does not comment on this mismatch between the catastrophic rhetoric leading to lockdown and reality. Yet shortly after the 9th or 10th of April, when deaths, hospital occupancy and ventilator occupancy had all peaked, it must have been clear that the doctrine of lockdown necessity had dubious foundations. On page 37 of his evidence Cummings reports that, having implemented lockdown, they predicted the peak between 15-20 April. That it was actually a week earlier strongly implies that it was not lockdown that turned around the wave of infections. That had already happened. Surely 'incredibly able people' must have realized this?
This isn't the only place where Downing Street's abdicated king of data ignores the data when it is inconvenient. We are repeatedly told that locking down earlier would have saved lives. But despite injunctions to learn from other countries, the fact that early lockdown Eastern European countries then suffered heavier second wave losses is apparently not one of the lessons to attend to.
The only hint of doubt comes, perhaps, in the tortuous straw man argument that he uses to dismiss the criticism that long before lockdown people had already reduced their contact rates sufficiently to turn around the wave of infections. He argues that people reduced contact rates, not in response to the perceived risk from the virus, but because 'the failure of government to act was so scary that people out of fear retreated to their homes'. Therefore government had to act as it did. Perhaps this perspective of a passive population continually waiting for guidance from the great helmsmen of government is an occupational hazard of working in Westminster. I don't know.
But the nadir of self awareness comes on page 99 of the evidence where Cummings complains of the government tendency to have crises managed by 'A small group of people excluding practically all the smartest and most knowledgeable at the top of highly centralised and closed/secretive institutions'. So who were the small group of 'excellent people' who Cummings portrays as being the real brains behind the shift to lockdown? Four are singled out in the evidence. Tim Gowers is mentioned repeatedly. A Fields medallist and Fellow of the Royal Society, he is as able and distinguished as pure mathematicians come. He also appears to have no experience in epidemiology, statistical data analysis, medical science or risk management. Winner of a Nobel prize for Chemistry, Venki Ramakrishnan, was president of the Royal Society and is an extra-ordinary scientist. He would surely have been a fantastic member of any team addressing a crisis in which protein synthesis or ribosomal structure played a major part. Perhaps less so when the crisis was about public health. Demis Hassabis and Mark Warner are scarcely less impressive figures from AI. Experts in data at least, just not the medical or epidemiological sort, for which a statistician might have been a better bet.
Elsewhere Cummings asserts that 'the public had the right to expect [questions] were being pursued by the best qualified people in the country'. Was that expectation being met by Cummings' informal team? Does being superbly qualified in one domain make somebody 'best qualified' to decide policy in a crisis unrelated to that domain?
An indication that it might not comes from Cummings' obsession with exponential growth, which he claims to have first learned about in the spring of 2020, from Tim Gowers. As anyone whose knowledge of epidemiology and exponential growth preceded the spring of 2020 will tell you, exponential growth is a reasonable approximation to the very early growth in the number of cases of a new disease. How early is 'very early'? Before people start reacting by reducing their chance of exposure, for one thing. And secondly, before noticeable effects from the highly gregarious catching it first, thereby removing themselves from the infection process. And before the infection risk from a seasonal virus starts dropping as the weather improves for a third. All these factors were at play in the ending of the UK exponential growth phase, and the turn around of the first wave of infections before lockdown, as they were in the turn around of Sweden's first wave a couple of days later, without lockdown. In what proportion, I have no idea.
For a pure mathematician grappling with epidemiological models for the first time, exponential growth may well seem like all that needs to be considered, leading him to recommend that Cummings should 'push extreme suppression immediately' as apparently happened on 14th of March. Someone who claims to have himself only just learned about exponential growth is unlikely to be in a position to question this advice. In the end this, and much of what Cummings presents, was back of an envelope reasoning, and it remains so no matter how distinguished and able the envelope's owner was.
An old climbing partner, a retired GP, complains that Covid unleashed an epidemic of the Dunning-Kruger effect. Cummings written evidence, I think, does little to refute this view.