In Part 1 yesterday I examined Dominic Cummings’s project for government reform and his operational plan for improving government decision-making in uncertain situations. Today I will look at how his operational plan worked in early 2020, resulting in his Red Team advisers persuading him to support lockdown.
THE first death attributed to coronavirus in the UK was recorded on March 5, 2020, two days after the Health Department published the UK Coronavirus Action Plan. The victim was a 70-year-old woman with underlying health conditions who had been in and out of hospital for non-coronavirus reasons, and who had no known contact with an infectious person. A positive test result for coronavirus came back after she died.
Dominic Cummings says that two days later Marc Warner, the founder and CEO of Faculty AI a company working on the NHSX digital transformation project, approached him with concerns. ‘He said to me, “I’m really, really worried about this. It seems to me that this plan could easily be mad. It could be incredibly destructive. Has this really been tested? Have you really thought it all through? Should I and some others start thinking about a plan B?”’
Marc and his brother Ben Warner worked on Vote Leave and were two of the very smart people whose judgements Cummings trusted. Cummings relied on his old team and used them as his own Red Team.
Ben emailed Sir Patrick Vallance, then the government’s Chief Scientific Adviser, on February 27 saying, ‘In the SAGE meeting today, I was a little concerned the NHS didn’t seem to know what they needed for their models and didn’t seem to have started modelling. Did you have the same feeling?’ (this is from Cummings’s Covid Inquiry written testimony) By March 6, his brother’s company was tasked with building the NHS a Data Dashboard for managing covid.
Cummings had long been an advocate for the more sophisticated use of data analytics, modelling and data visualisation in government decision making, arguing that ‘model-driven material can serve as a kind of enhanced imagination’ (Cummings’s emphasis).
During his May 2021 appearance before the joint Parliamentary Committees, Cummings explained his reliance on Ben Warner as the coronavirus crisis unfolded. ‘I am not a technical person. I am not a smart person. I could not understand a lot of the models and things like that that were being discussed [by Sage],’ he said. ‘I sent Ben. I thought that it was far better to have a PhD physicist there who could actually properly look at the documents and really understand what was being said. He had a lot more to contribute to that than I did. I listened to lots of the conversations in February and March [2020], but a lot of it was over my head.’
Ben Warner was in turn relying on Sir Jeremy Farrar, Professor John Edmunds, Angela McLean (chief scientific adviser at the Ministry of Defence) and Sir Patrick Vallance to further his understanding of the science. As Farrar revealed, Warner was being influenced in other, less ethical ways, particularly by Edmunds who he says ‘devised his own deliberate behavioural strategy in Sage meetings, which was to look political advisers directly in the eye while repeating the phrase, “we are talking about hundreds of thousands of deaths”.’
Although the parameters for the NHS Covid Dashboard were still being scoped, Marc too was in full super-forecaster mode. ‘I recall messaging Dominic Cummings, and likely Dr Ben Warner and others, with stark warnings around this time [week of March 9 2020], but I cannot find them (presumably they were via WhatsApp or NHSX email),’ he said. ‘I don’t remember the specific details of many of them, but I think at least one of them was something like: “Code Red – the system has f*cked it. There is no plan”.’
Cummings said, ‘In the week of the 9th, when people like the Warners were saying, “Panic, panic. Advance”, everyone was thinking, “We’ve got ages to think about this. The peak is not until June. We’ve got weeks to potentially make decisions about these things”.’
Sage, the Department of Health and the Cabinet Office weren’t panicking but there was a countervailing force: the brothers and Marc’s girlfriend Dr Laura Pimpin, an epidemiologist who worked for Babylon Health, which operated a telehealth service. Believing that the Prime Minister was being ill-advised by Government officials, Cummings became contemptuous. ‘In that week it became clear neither Hancock/CABOFF understood herd immunity effects: 100s of 1000s choking to death + no NHS for *anybody* for months + dead unburied + econ implosion; so we moved to Plan B: suppression + Manhattan Project for drugs/vaccines + test&trace etc.’
Emotion at times had greater effect than reason. ‘Laura was one of the people shouting at me that week that I needed to stop this somehow’ (Farrar’s book). He subsequently brought Pimpin in to No 10 for three months as an unpaid adviser because he said the Cabinet Office lacked specialist staff who understood disease and genetics or data and analytics but it had been tasked with setting up the Joint BioSecurity Centre and it ‘could not figure out that the DHSC plans were full of holes’. (Covid Inquiry statement)
On March 12, Cummings had dinner with the Warners. He recalled, ‘They kind of hit the total panic button with me. They said, “We are looking at all this data; we are looking at all these graphs. We are heading for total and utter catastrophe. We need to have plan B.” Ben then went off and spoke to Patrick Vallance about that on Friday the 13th.’
‘I do not remember the conversation precisely,’ said Ben, ‘but I know that at this time Marc was strongly of the view that the plans [mitigation not lockdown] were incorrect.’ His notes, he said, expressed Marc’s concerns at the COBR [the Cabinet Office system for dealing with emergencies] process.
In fact, the process was never followed properly. The concept of operations behind it is meant to guide UK government responses to three pre-defined levels of emergency:
· Level 1 significant emergencies do not require coordinated central government response but COBR might be activated to brief Ministers.
· Level 2 serious emergency requires some involvement under the leadership of the lead government department. In the event of ‘health emergencies’ the Health Department is the designated lead. Level 2 was activated for 2009 swine flu, the first purported public health emergency to receive a disproportionate central government response.
· Level 3 emergencies are described as catastrophic with widespread impact such as a natural disaster or a Chernobyl-scale industrial incident which requires immediate central government support as local services are overwhelmed and can’t cope or emergency measures are required.
Although Cummings purports to be in favour of decentralisation and distributed controls and against centralised bureaucracies, when faced with what he was told was a crisis he did the opposite. The Plan B response measures imposed on March 23 2020 pre-emptively escalated Covid to a Level 3 catastrophic disaster and a data management dashboard was developed to micromanage it.
Cummings would have done well to remember what he wrote on his blog in June 2019: ‘In some ways the use of technology now makes management worse as it encourages Presidents and their staff to try to micromanage things they should not be managing, often in response to or fear of the media.’ It is the best example of his own super-forecasting abilities, although he ignored his own advice.
The final element of Cummings’s rational plan for improving decision-making and ‘not fooling yourself’ is quantitative problem solving. Just as he’d suggested in his 2014 blog, he shored up his resolve by consulting his chief rationalist of choice, Tim Gowers, a Cambridge University mathematician.
Explaining the reasoning he gave Cummings about lockdown, Gowers is quoted in Jeremy Farrar’s book as saying: ‘I felt very strongly at the time that a herd immunity strategy was wrong – a simple back-of-the-envelope calculation made it clear that in order to implement it, either hospitals would have to be massively overwhelmed or the strategy would take years, and depend on reinfection not being possible . . . Given that a herd immunity strategy couldn’t possibly work, lockdown was inevitable, and that given the lockdown was inevitable, it was madness not to start immediately.’
The retrospective certainty expressed is stronger than it was when Gowers set out the quantitative problem solving behind his reasoning in a blog post on March 28 2020 saying, ‘Again I want to stress that these conclusions are all quite tentative, and should certainly not be taken as a guide to policy without more thought and more sophisticated modelling. However, they do at least suggest that certain policies ought not to be ruled out without a good reason.’
In the public memory, Neil Ferguson’s modelling is often blamed for lockdown but it wasn’t Sage who asked him for it. Ben said, ‘Marc and I decided that I should phone Neil Ferguson. Neil went on to produce some work modelling stringent measures and the effect on bed demand.’
Ferguson’s modelling was used to justify lockdown publicly but he had in fact cautioned Ben on the 10th saying, ‘Lockdown could be worse than the disease.’ His modelling wasn’t presented until the 15th after Boris Johnson had already been pushed into Plan B at the infamous March 14 meeting when Ben Warner’s hand-drawn graphs were used to enhance the imagination of the Prime Minister in order to persuade him. Cummings says ‘one of the things being screamed at the PM was “there is *no plan for lockdown* & our current official plan will kill at least 250k & destroy the NHS”. Cf the graph: “optimal single peak strategy” with 3 interventions. That was the official plan, which was abandoned.’
Data analytics are tools to enhance decision-making, but on the evidence, data experts are no better at making decisions during crises. Emotions, not reasoning, prevailed and Cummings and his Red Team were no match for the Machiavellian characters manipulating them.