The improbability of probability

"The critical thing first is to get case numbers down, and then I'm hopeful... in a few weeks' time we will be able to move to a regime which will not be normal life, let me emphasise that, but will be somewhat more relaxed in terms of social distancing and the economy, but relying more on testing."  Professor Neil Ferguson (Imperial College London) suggesting that physical distancing could be relaxed in the next week to seven days

We are being subjected to an unprecedented amount of speculation from the science community at the moment. With the government, indeed most politicians, clearly having no clue whatsoever about the coronavirus. With the public largely confused by the various numbers and the methods used to calculate them. The vacuum is being filled by scientists. Including the Chief Medical Officer, their various deputies and particularly the team from Imperial College, London. What is less clear is whether Government policy is being dictated by the science or whether the science is being dictated by Government policy.

From a government that includes somebody who once declared that “the public have had enough of experts” the fact they are listening to scientists at all is quite the turnaround. As a socialist, but also as somebody who taught social statistics for a number of years, I have always argued for evidence-based policy making. Which actually is the exact opposite of what generally happens. 

I have often said in relation to climate change that it would help if we listened to the science. In saying that I’ve not really distinguished science and scientists, tending to see them as the same. But, actually over the past few weeks I’ve come to the conclusion that probably we should trust the science, but not necessarily the scientists. Indeed, we should only trust the science on condition that we can actually understand it. What I’ve felt since the lockdown is that the science has become more and more obtuse as the government have come under increasing pressure.

According to a report by Reuter’s the mishandling of the current crisis can be laid at the feet of the scientists advising the Government. According to John Ashton, a clinician and former regional director of Public Health England, the government agency overseeing healthcare, the government’s advisers took too narrow a view and hewed to limited assumptions. They were too “narrowly drawn as scientists from a few institutions,” he said. That comment about institutions points to institutional rivalry between people studying similar things. It will be familiar to anybody who has ever worked in a university. However, in most instances the failure to listen to others does not cost lives.

The problem according to Reuters was two-fold. First, the scientists who inform governmental policy continued to regard Covid-19 as flu-like in its behaviour long after it was obvious that this was not the case. Second, those same scientists were slow to react to events elsewhere and raise the threat level to high. This early reticence to raise the risk level only changed when Neal Ferguson’s team at Imperial College changed their mind about how many deaths were possible following the ‘herd immunity’ strategy as opposed to the containment strategy now being pursued.

“The catalyst for a policy reversal came on March 16 with the publication of a report by Neil Ferguson’s Imperial College team. It predicted that, unconstrained, the virus could kill 510,000 people. Even the government’s “mitigation” approach could lead to 250,000 deaths and intensive care units being overwhelmed at least eight times over,” Reuter’s reports.

The idea that government policy is being made by scientific experts may be reassuring to a public who have lost faith in politicians, but this places public policy away from democratic, and thus accountable, institutions and into the hands of people whose public persona is of detached pursuers after the truth. Besides we should not forget it is the Government’s job to make decisions, it is the scientists job to give them information on which to make the decision. If the Government don’t understand what the scientists are saying, then it is their responsibility to keep asking for explanations until they do understand. It is not their job to hide behind the scientists because that is easier than taking hard decisions themselves. Part of the problem is that all you have to do to justify a decision is say something like “experts from Imperial College, London” and we are all supposed to roll over and accept that this must be an objective truth. But, scientists not only make mistakes, they are also not always the objectivists we are led to believe.

In 1962 Thomas S Kuhn, a philosopher of science, wrote an influential book called The Structure of Scientific Revolutions. This book argued against the idea that scientists were somehow objective seekers of new knowledge in pursuit of discoveries. Kuhn argued that most of what he called “normal science’ was not looking for new knowledge but rather confirming existing knowledge. What this means is that the majority of scientific activity during these periods of normal science is focussed on affirming that which is already known. According to Kuhn normal science worked within what he describes as a paradigm. Only when it becomes absolutely clear that the data no longer fits do we enter a period of what Kuhn calls “revolutionary science”. This is when a fundamental shift in our way of thinking occurs and this revolutionary period establishes a new dominant paradigm.

None of this should be taken to mean that science cannot come up with new knowledge but if that is going to happen it will be in a particular paradigm. Modelling reality, which teams from Imperial College amongst others do is very much normal science. That is one reason why many people were sceptical when on March 16^th the government changed strategy claiming the science had changed. It is also why Reuters can, with some justification, say that the scientists were slow to act. They were, put simply, carrying on as if Covid-19 was business as usual. Or, to use Kuhn’s terms, they were simply not accepting the challenge to their dominant paradigm. They only did so as it became more and more obvious that the available data did not fit their preconceptions.

It is important not to read too much into 1 or 2 day's data

Based on a one-day fall in the number of deaths recorded in the UK the media began running stories suggesting that the curve was flattening. The following day the number of deaths reached a record daily high, record that is until the following day when it again jumped upwards. Far from flattening the curve was clearly and very visibly rising ever upward. The shape of the curve is important because if the lockdown is having an effect, and scientists estimate that without it around 500,000 would die, the peak will happen at around the mid-point. 

Neil Ferguson is a key player in all this and in terms of death rates he told the Financial Times:

“We will be putting out updated estimates, probably in the next week, both on intervention impact and on growth rate, but we hope that the two cancel out really. We hope we’re in the same sort of regime of mortality — somewhere between about 5,000 and 30,000 deaths, and probably closer to 10,000-20,000.”

I think what is important to note here is the highly speculative numbers being cited. Somewhere between 5,000 (a figure we have already reached) and 30,000 is quite a range particularly for the 25,000 additional dead and their families. The language used is also very speculative: "updated estimates", "we hope" and "somewhere between" do not sound like the phrases of somebody who is totally confident they know what is happening. The problem is, as Neil Ferguson is aware, that in order to accurately model what is likely to happen it is necessary to know at least three things: the size of the population, the infection rate among the population and the death rate of those who contract the virus. We know only one of those and it is the size of the population.

There is a paper which is the basis of what Professor Ferguson said freely available on the internet. It is not easy to read if you are not familiar with academic scientific writing, but it explains if you take the time to read it, not only how they arrive at their figures, but in their own words that there “remains a high level of uncertainty in these estimates.” (p.6).  The paper explains that it uses “a novel Bayesian mechanistic model of the infection cycle” a phrase that is written for other scientists (as most scientific papers are) not for Ministers, journalists or the general public. A simple translation would be “there is a lot we don’t know, so we are using our best educated guesses”. 

Some of the problem with the Imperial College data is its speculative nature. Ministers do not like uncertaintly, neither do the British media. They want figures that they can rely on and which the public will understand. They don’t necessarily have to be right, but reassuring is always good. There is an important statistic which the media have not seized upon largely I suspect because in the UK it does not sound very dramatic. It concerns the number of deaths that they estimate have been avoided as a result of the lockdown. The figures for the UK are not dramatic because they refer to the period when the curve was only just beginning to rise. Nonetheless they say: “Even in the UK, which is much earlier in its epidemic, we predict 370 (73-1,000) deaths have been averted.” (p.10) That is 370 mothers, fathers, sons, daughters, wives, husbands etc who are alive because the Government implemented the lockdown. (The numbers in brackets refer to the margin of error figures so it could be as low as 73 deaths averted or as high as 1000, but as is usual in science the figure of which they are 95% confident is the one quoted).

Why are scientists not pushing the case for testing?

The paper has little to say about testing (perhaps that is covered elsewhere) but accepts that testing figures are likely to be an under-estimate “due to the focus on testing in hospital settings rather than in the community.” (p.12) This means that the populations in most of Europe are nowhere close to herd immunity. Some scientists continue to push the case that the lockdown is actually prolonging the fightback against Covid-19 by reducing herd immunity, which needs to be close to 60% to be effective. For example, Knutt Wittkowski has a video on YouTube in which he states:

“With all respiratory diseases, the only thing that stops the disease is herd immunity. About 80% of the people need to have had contact with the virus, and the majority of them won’t even have recognized that they were infected, or they had very, very mild symptoms, especially if they are children. So, it’s very important to keep the schools open and kids mingling to spread the virus..”

This may or may not be true but it is certainly a dangerous strategy. It relies on the fact that a) children won’t die, except some have already done so; b) that the elderly can be protected by keeping them away from children, which is great in theory but much harder to do in practice without a lockdown; but c) only the elderly and vulnerable should be put in lockdown, which presupposes that the rest of the population is safe from infection when we are daily seeing younger people dying from the virus.

The thing that both the Imperial College team and Professor Wittkowski do not mention is the World Health Organisation’s advice that the only way to defeat Covid-19 is a rigorous testing regime. Although the Imperial College paper notes that Sweden, with one of the lower mortality rates in Europe, is an anomaly because it has implemented no full lockdown their answer is to regard this as “an artefact of our model”. In other words, a case which clearly challenges the model is forced into the model because to do otherwise would clearly undermine the whole basis not just of the model, which may or may not turn out to be correct, but also significantly British government policy. I cannot say with any certainty that the Imperial College team are bending their public pronouncements in order to maintain their own credibility with the Government. But their failure to include a proper testing regime within their model or to allow for the fact that perhaps this explains both Sweden’s and Germany’s far lower mortality rates does seem to be a case of Government policy driving the science, and not science driving Government policy.

Scientists have no more idea than you or I how many people in the population are actually infected, partly because the UK government ignored World Health Organisation advice to “test, test, test”. Since March 5^th when the first person in the UK died from Covid-19 around 6,000 people have been tested each day. To date just under a quarter of a million people in the UK have been tested in a month. In Germany, which started testing much earlier they are capable of testing half a million people a week. In the UK testing has not been carried out in a systematic way allowing for any analysis of which sections of the population are most at risk, or the calculation of a reliable death rate for those testing positive. Testing was originally carried out only on those admitted to hospital and was ramped up, under political pressure, to include NHS workers.

Professor Chris Whitty told the daily press briefing a couple of days ago: “We all know that Germany got ahead in terms of its ability to do testing for the virus, and there’s a lot to learn from that.” Germany did not just do tests because they thought that would be a good idea they learned from China and particularly South Korea, lessons which were there for the UK to learn, especially considering the virus reached Germany before us. Chris Whitty only made that comment to correct his colleague, the government’s chief scientific adviser, Sir Patrick Vallance, who had replied: “The German curve looks as though it’s lower at the moment, and that is important, and I don’t have a clear answer to exactly what is the reason for that.” Here we have, if we need it, evidence that the scientists have one eye on the Government and see their role, in part, as not embarrassing the Government by pointing out that they could have acted much earlier and much more decisively and saved a lot of lives in doing so. But, perhaps we should note as well that the scientists themselves were trapped in their own version of “normal science” and were equally unable to give the advice that the Government needed. 

Science is difficult. It is difficult for scientists to conduct and difficult for non-scientists to understand. Nonetheless, we should not be bamboozled by science. Journalists and opposition politicians need to be prepared to understand the basis of the science in order to do their job of challenging Government. Scientists need to do their jobs free from Government interference to produce the “right results”. What is right from the Governments perspective might well turn out to be wrong from everybody else’s and if scientists don’t feel able to say so, then they really are not scientists at all, but Government propagandists. I don’t know if that is currently the case. Journalists need to ask searching questions and not just accept either scientists or Government versions of events. At the moment, although some journalists are starting to ask questions of government they seem ill equipped to question the science. This relates to a problem I have written about earlier that it is political journalists who are, in the main, covering the story, not journalists with a background in science. And, finally, opposition politicians, particularly in the Labour Party, need to step up and question Government policy rather than, what appears at the moment, to be grovelling for a place in a National Government that is never going to happen. To be fair Keir Starmer has asked about the exit strategy, but we are clearly a very long way from that being relevant. He is not challenging the science neither is he repeating at every opportunity that the UK should be following the lead of Germany or Sweden or South Korea. 

My conclusion from all this is that although we have the right policies in place currently, they took too long to be implemented. That was not an accident and ultimately the responsibility lies with the Government, even if the scientists do not come out of this covered in glory. People are dying because of the Government’s early inaction which was supported enthusiastically by a press which seems to see their main job as being Tory cheerleaders (to the extent that they promoted a clap for Boris event which was actually an insult to the clap for NHS workers who are genuinely putting themselves at risk, rather than somebody who adopted a totally foolhardy approach to the virus and managed to contract it), rather than asking and keep asking questions which make the Government justify both its early inaction and its current inaction over testing. There is a failure too of the Labour Party since Jeremy Corbyn  stepped down as leader. to be fair, in his final days as leader he tried to ask questions but was simply ignored by the press. The main preoccupation of Keir Starmer is to be seen to be responsible and sensible. That means backing the Tories in a time of crisis, which would be fine if it was self-evidently the case that the Tories knew what they were doing, and were acting in the national interest. In fact, the Tories self-evidently do not know what they are doing and although they may feel that they are acting in the national interest (we should not always doubt their sincerity as hard as that seems) they are also clearly continuing to act in the interests of the Tory Party and big business. The Covid-19 virus is a unique and unprecedented attack on our way of life. Many people in the UK will likely die as a result of our ill-preparedness. It is nobody's fault that the virus has occurred. That is simply nature doing what nature does. It is legitimate to argue that people will die who might have survived because of the way this crisis has been managed.  Government failed to act on the science and scientists, the press and the opposition failed to press the Government to do so.