Some people think now that we& #39;ve & #39;flattened the curve& #39;, it& #39;s time to open.  If COVID-19 is bad for health, and lockdown is bad for the economy, shouldn& #39;t we open as much as we can keeping below hospital capacity?

Our working paper discusses why this is flawed thinking. 

1/n https://twitter.com/Annayork91/status/1265313015689347076">https://twitter.com/Annayork9...
You might have heard:

1. We need time to boost contact tracing capacity.
2. People won& #39;t go out if they& #39;re scared of getting sick.
I believe these reasons. 

But let& #39;s say you& #39;re skeptical of contact tracing.  You want to get out.  We give another way to look at it. 

2/n
To keep cases below hospital capacity, we could:

1. Take Thor& #39;s hammer on the curve, let up, then #Stomp on the curve again.

2. Walk a #TightRope to keep cases *right* at the critical level.

Obviously #Stomp keeps disease lower.  But would it make us happier?

3/n
We show that for a broad range of utility functions, #Stomp > #Tightrope.

COVID-19 grows (nearly) exponentially.  This means that little changes in physical distancing have BIG impacts on COVID-19.

--> e.g. 2 months at 90% lockdown = ~7 months at 50% lockdown

4/n
If we pay a little more upfront for disease control, we get a *lot* more breathing room.  This means that we can be much *more* open (or open for longer).
Our paper doesn& #39;t give a specific optimal policy. Rather we provide simple intuition **against** moderation. Strict intermittent policy feels weird. But with COVID-19 control, it may be better, not just for disease control but other outcomes too.

https://www.nytimes.com/2020/05/11/opinion/coronavirus-reopen.html

6/n">https://www.nytimes.com/2020/05/1...
Some nerdy notes: The more we thought about it, the harder it was to figure out the right utility function associated w/COVID-19 policies.  We considered a range of options but would love thoughts and feedback (p.4 + Appendix G.1).

7/n
We were curious why optimal control papers often find smooth solutions -- we found the form of the optimal solution is sensitive both to the utility fxn and to if/how a latent period is included. (Appendix G). A reason to be mindful of the epi ( @EpiEllie, @johngraves9)!

8/n
One last note -- the appendix is technical.  We tried to make it accessible. But we believe math communication is like public health communication: if reader doesn& #39;t get it, it& #39;s on you, not them.  So if you read the appendices (bless you!) and have q& #39;s, reach out!

10/10
You can follow @ambilinski.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: