Link to the original post (from 17th March 2020), which is public — reposted here for convenience, with minor formatting edits.

I’ve heard some people refer to the COVID 19 epidemic as ‘Our World War II’ and it’s worth saying that this is not hyperbole.

This may conjure up images of over-the-top war films, but put those aside. In terms of the challenge we face, make no mistake, now that containment has failed, hundreds of thousands of deaths in the UK alone are a very real prospect, one we must avoid at all costs.

(if you think that sounds unlikely, read this — or my summary below)

This is an event that is likely going to affect every one of our lives for the next YEAR or two. Do not think of this on the timescale of weeks as I see many people doing — we’ll be lucky if it’s all over by Christmas. Everyone is going to have to make major behavioural changes. A relatively small number are going to be on front lines — our health service workers (I’m proud to count many of you among my friends and family). But more of us will have some role to play. If you’re young and healthy, consider volunteering. Donate money if you can afford it — or perhaps computing time (see here).

Now, here’s my summary of where we are, based on the paper I linked above. It’s the latest modelling from Imperial College.

What they have determined is that a ‘mitigation’/’cocooning’ strategy is likely not sufficient on its own to prevent the NHS being overwhelmed (because even among the young and healthy, as many as 1 in 20 will require some hospital care). That would cause serious knock-on effects in terms of mortality. This modelling is the reason for the government’s recent moves to more serious, disruptive interventions.

We also know that suppression is not effective in the long term, because when the interventions are lifted, perhaps in a few months, we would not expect the virus to fizzle out. It will come back. The ‘cocooning’ strategy aimed to prevent this by protecting the vulnerable and keeping the NHS within capacity while the natural process of herd immunity in the healthy population started to slow transmission of the virus, but this won’t happen with suppression.

Therefore, the only feasible way identified to prevent disaster is a series of rolling suppression phases, triggered whenever the health service is dealing with a critical caseload. This scenario is shown in the graph I’ve copied from the paper below:

The models suggest lockdown will be in place 2/3rds of the time for at least the rest of the year, although this is affected by what we do in terms of public hygiene. This will, of course, be hugely disruptive — it’s an unprecedented public health intervention — but it’s our best shot collectively.

The endgame is a vaccine — a way to create mass herd-immunity without lots of people being hospitalised. My hope is that this will be available sooner rather than later, and that is why I mentioned donating computing time above — it’s a way to help. But realistically, it could be 18 months before a vaccine can be trialled and production scaled up. We must all hang together until then.

Expect data science + some politics and astronomy. All things data at