a thoughtful web.
Good ideas and conversation. No ads, no tracking.   Login or Take a Tour!
comment by Devac
Devac  ·  1747 days ago  ·  link  ·    ·  parent  ·  post: Hubski COVID-19 Round-Up #1

My institute, university and school where I'm TAing closed until mid-April. Same goes for nurseries, kindergartens and primary/secondary schools in Warsaw, Crakov, Poznań, Łódź and a bunch of other cities. Poland only has 22 confirmed cases as of the time of writing, and I hope it'll dampen the spread.





am_Unition  ·  1747 days ago  ·  link  ·    ·  

Because America is the only country that matters in the world (HUGE fan of American Exceptionalism, myself), you've probably heard we just hit 1,000 infections. We're undoubtedly way behind in assessing and tracking the true number, despite a growth rate matching the rest of the non-China world closely (b ~ 1.2). Alotta this comment is for posterity, not Devac.

If we have identified 1k+ cases, and we have reportedly used perhaps, optimistically 8k testing kits to this day. Let's say there's a two-day test delay for specimen handling, testing, and reporting (again, pretty optimistic, I think). We might infer that somewhere around 1 in 5 people that get tested are testing positive, but this invests an unwise amount of faith into our testing and reporting system. Because, OHhh, that's right, we don't even fucking know for sure that the other 4 out of 5 aren't positive yet (per theAtlantic.com article above), or if the tests even always work. We can't infer a damn thing because of the botched response, which obviously needed to include accurate statistical reporting.

You want good statistics? The feds and states should've talked about how they would coordinate testing and coordination in like January. We knew this was more contagious than SARS or Ebola because of China's (relatively) swift and massive response, and then their transparency. We talk shit about China, and maybe for good reason sometimes (sorry), but they deferred to scientists, and now, the source country of the outbreak is on track to have less cases than the U.S. within a month or two. Thank goodness covid-19 didn't originate here in the U.S. under Trump, the rest of the entire world would be 10x more clueless about it. Not-America saved our asses by generating the statistics we need. This time, it's not just another instance of U.S. disregard of science on the world's stage, our naive response negatively impacts health at a global scale. An international organization may investigate Trump and his genius squad for criminal conduct in their response to covid-19. Or, at least, one may try.

K, back to strict math. Let's play nice for almost a few contiguous paragraphs, and assume that U.S. measurements are going OK. Here they are:

I'mma fit an exponential function, f(x, A) = A*(b)^x, to data between Feb. 23 to now. I'm just using A=50, and x={0, 17}, and I get a b value somewhere between 1.19 and 1.2 that satisfies f(x=17) = 1,000. That actually agrees with what is currently being observed in the larger dataset of all the non-China's right now (source: the YouTube video in my post far above, by 3Blue1Brown).

Fitting an exponential to this larger dataset (which has muuuuuuch better statistics) over roughly the same time interval as before also gives me b ~ 1.2. Pretty cool, but not necessarily indicative of accurate measurements in the 'States. Using the covid-19 tracking map and basic knowledge of population sizes, a comparison of population percentage diagnosed with covid shows that the United States is shockingly low compared to e.g. Europe (don't use the circles to do this, use the raw numbers in the sidebar). Perhaps it came to the U.S. last, or some initial quarantine efforts were not in vain, but we cannot make any statements with a high confidence when only a very small subset of folks presenting with symptoms are being tested. For whatever reason. The cloud of uncertainty regarding wtf is going on inside the Trump administration doesn't help, and casts further doubt on the veracity of the numbers they offer (or don't).

Ok, angry me, again, sorry not sorry. This absolute shit job of almost zero covid-19 testing preparation means we will, FOR A LONG TIME, continue to have much worse statistics compared to almost every other developed country. We have missed an opportunity to respond intelligently to this. It frankly doesn't much matter how many little hotspots you quarantine if you're not tracking the spread of the virus through the population. That's like simply ignoring a flaming bag o' poo on your doorstep, but your porch is wooden, and your house burns down. Shoulda sprung for a fire extinguisher.

I have yet to see any documentation or explanation regarding why the Trump admin. refused testing kits offered to the U.S. from the WHO (probably not the band, in this instance):

    But neither the CDC nor the coronavirus task force chaired by Vice President Mike Pence would say who made the decision to forgo the W[orld] H[ealth] O[rganization] test and instead begin a protracted process of producing an American test, one that got delayed by manufacturing problems, possible lab contamination and logistical delays.

Again, the theme: You need good science, president dipshit. Good science needs data. You've ensured that we won't get it in time to help choke this thing early, and compared to the rest of the developed world, we're fucked. You've endangered the entire world. Great job.

Prediction: The U.S. will lead the developed world in covid-19 cases within a month or two, surpassing China. Italy's condition is unfortunate, and a reminder that luck is always on the playing field during an epidemic.