Thread by @danlucraft

I didn’t understand why the UK was not introducing measures now. I know the CMO+CMS said they want to delay and reduce the peak, but so does everyone!

So that doesn’t explain why they are pursuing a different policy to countries that are already controlling the virus well. 1/

Then there's articles like this one suggest that the earlier we introduce controls / change behaviour, the better off we'll be in the long run: medium.com/@tomaspueyo/co…

So I lay in bed last night thinking about exponential curves and I realized what I was missing. 2/

First can I point out very prominently that I'm NOT an expert, so take all of this with a huge pinch of salt. It's just my best guess.

Here's what I think is going on: 3/

The govt thinks they & the public will only be able to maintain social distancing controls for a period of time before fatigue sets in. (More like one month than six months.)

OK, but isn't it better to introduce them earlier, to reduce spread as much and as early as possible? 4/

Won't that both delay and reduce the height of the peak? (The height of the peak being the most critical aspect from a health service capacity and mortality rate point of view.)

Well, no. (This is what I was missing.) 5/

With an exponential (actually logistic) curve, and a time limited set of control measures, earlier controls delay the peak, but hardly reduce it at all. If you want to reduce height of the peak, you have to introduce them later, as the epidemic really begins to take off. 6/

This wasn't intuitive to me. But if there are 10k cases today, doubling each week, then in 4 weeks it's 80k. If we introduce 1 month of controls that halve spread now, then it's 6 weeks to get to 80k. At which point we'll be in the same place and the peak will be just as tall 7/

Whereas if you introduce the controls as the peak begins, you can actually chop off the top of the peak. See imaginary graph of active cases, based on controls that halve the spread

This is based off a v basic logistic curve done in Numbers. Super loose modelling, NOT science 8/

So I *think* this is what the govt is doing. If so this implies some pretty bleak ideas:
- they think countries currently doing well at controlling the virus are actually only delaying the onset of the peak and when controls have to be lifted they'll be back at square one
9/

- we're going to have more cases sooner than countries that have controls earlier, but it'll be worth it
- if they get the timing wrong, things go badly wrong (In the graph above the peak height is wildly sensitive to exactly when controls are introduced and for how long
10/

In theory, if done right, the UK gets to herd immunity much faster than other countries who'll still be struggling later with outbreaks later this year. In theory we control the peak better, because our controls are brought in and are max effective at the right time. 11/

So I think that's what's going on. Again, not an expert, just an interested person trying to understand stuff with my little spreadsheet.

Would love to hear from someone who does know if this is the right way to think about it. 11/