Thread by @soncharm

when stuff like this hits INTERNET i feel kinda bad for whichever rando GS-11 with a title like 'Public Health Analyst' was basically just told to cook up these scenarios with these assumptions & throw them on a slide

what i really wanted to ask tho is what's the y-axis?

https://twitter.com/kerpen/status/1420920749435592708

no seriously what's the y-axis

it reads, 'Probability Of Increase'. slide seems to be inviting reader to see p>.5 as bad and p<.5 as good. but increase of what?

i think, increase in 'incidence'? like, from this hypothetical baseline (50/100k/wk)?

then here's my story of this slide:

Under ballpark assumptions we like about Delta's R0, vax rate, & mask effectiveness, it's likely (p~1) that incidence will increase, unless all mask. (And it's not enough - p>.5 still - if only unvax mask)

This is the 'bottom right' graph

so the story/flow of the slide seems pretty clearly inviting onlookers to have the takeaway, We should recommend masks, and for vax'd too

and then that's what they did (so I guess my interpretation is right?)

but then I have a question: so what, bottom-right? who cares?

slide headline is: NPIs are 'needed' 'to reduce' Delta

this punchline is driven home by bottom-right which shows, unless all mask, 'incidence' will probably increase

maybe that's all true (given assumptions)!

but so what? why is it absolutely imperative to reduce incidence?

don't get me wrong. all else equal, do i wanna reduce incidence? sure

is reducing incidence positive-utility? of course

but it's not *infinite* utility. you don't just get to say: this goal is good, and we 'need to' do XYZ to get to the goal, so therefore XYZ

It just seems CDC now thinks that any policy, any action, any command is justified if it 'reduces incidence' of one particular virus. And on those grounds.

Or if that's not what they explicitly think, they're acting on such a principle. (which is a value judgment)

and that's before we even get into whether the mask effectiveness parameters they plug in here make any sense. obviously, if they're off, these graphs change a lot. my guess: p>0.5 no matter what, for a while. delta wave will rise, then break.

so that's reality. so what?