Thread by @STWorg

THREAD: The model that influenced UK government policy imperial.ac.uk/media/imperial… is now being critiqued after the UK government’s U-turn and acknowledgement that their initial policy might have caused up to 250,000 excess deaths.
1/n

https://twitter.com/hancocktom/status/1239672456073404416

What follows from this? It does not follow that those scientists, or models generally, should be condemned or ridiculed. The scientists did their best under extremely difficult circumstances. No one approaches corona flippantly.
2/n

On the contrary. The model was valuable. Let’s establish 1 thing: You can't do anything without a model in science. Anyone who says otherwise does not understand what an average of a handful of numbers represents. The average *is* a model. Background: amazon.com/Computational-….
3/n

We thus have to rely on models whenever we make decisions based on data. Whether it's corona or any other project, nothing will happen without a model. The only entity that makes decisions without a model is Donald Trump (+Nigel Farage, if you go down to the miniature shop).
4/n

Oh, you say, but what if the model is wrong? Yes, models can be wrong but if they are wrong, we can figure out why and we can learn from that. For more detail, you can check out this: doi.org/10.1111/j.1467….
5/n

You can learn from models being wrong, but you cannot learn anything from Donald Trump being wrong—because in Pauli’s immortal phrase, he’s “not even wrong“ en.wikipedia.org/wiki/Not_even_….
6/n

None of the carnival barkers around the world who shout jingoistic nonsense can manage to be wrong. Because random noise is just that, an annoying distraction that cannot even be wrong.
7/n

So back to “the” model. One can certainly find things in it that are wrong.
8/n

For example, the model seemingly assumes that the number of ICU beds is invariant between now and 2021 (horizontal red line Figs. 2 and 3). That may not affect the outcome of the decisions based on the modelling, but it nonetheless seems an unnecessary over-simplification.
9/n

But note: the model has created an opportunity for a conversation. We can learn from it. We can change that assumption and imagine a scenario in which the UK is buying more ICU hospital beds (assuming that Grayling doesn’t buy ferries from a pizza company instead).
10/n

We can also question the assumption of the rebound made by the model—which I believe to have been crucial in the decision making. The rebound refers to a recurrence of the virus later in the year, after having been suppressed during spring and summer.
11/n

How likely is this rebound? What is the likelihood of anti-viral drug trials being successful between now and then? Again, the model permits us to explore those different scenarios.
12/n

So if models are so good, what went wrong? Why the sudden U-turn in government policy? I can point to one problem for certain and I can speculate on another one.
13/n

The certain problem is that if this model had been released to the scientific community before it determined policy, the ensuing robust criticism could have prevented a decision that, in retrospect, is now acknowledged to have been premature if not outright false.
14/n

The more speculative issue is that the scientists and the government forgot a simple principle, namely Carl Sagan’s immortal phrase that "extraordinary claims require extraordinary evidence" en.wikipedia.org/wiki/Sagan_sta….
15/n

And in this instance, any radical departure from the established wisdom and recommendations of the WHO and the CDC, all other European countries, and the recent experiences of China, Singapore, South Korea, should have been based on extraordinary evidence.
16/n

The government, having just withdrawn from the European pandemic warning system theneweuropean.co.uk/top-stories/ea… did not seek the extraordinary evidence that their extraordinary policy called for.
17/17

@threadreaderapp - please unroll