Exponentials, extrapolation, and prudence

It's been a remarkable week.  There seems to be a consensus among US universities, based in part on CDC guidelines, and in part on the logistically and legally terrifying possibility of having to deal with dormitories full of quarantined undergraduates, that the rest of the 2019-2020 academic year will be conducted via online methods.  This will be rough, but could well be a watershed moment for distance education techniques.  The companies that make the major software platforms (e.g. zoom, canvas) and their web storage are facing a remarkable trial by fire when the nation's large universities all come back from break and hundreds of thousands of students all try to use these tools at once.

At the same time that all this is going on, many doctoral programs around the country (including ours) that had not already done their graduate recruiting visitations were canceling open houses and trying to put together virtual experiences to do the job.  

There is a lot to unpack here, but it's worth asking:  Are people over-reacting?  I don't think so, and over-reacting would be better than the alternative, anyway.  Different estimates give a range of values, but it would appear that the age-averaged mortality rate of covid-19 is somewhere between 0.7% and 3%.  (The current number in the US is something like 2.9%, but that's probably an overestimate due to appallingly too little testing; in the non-Wuhan parts of China it's like 0.6%, but in Italy it's over 3%.)  The disease seems comparable in transmission to the annual influenza, which in the US is estimated to infect 35-40M people every year, and with a mortality rate of around 0.1% leads to something like 35-40K deaths per year.  Given this, it's not unreasonable to think that, unchecked, there could be between 250K and 1.2M deaths from this in the US alone.  A key issue in Italy stems from the hospitalization rate of around 10-15%.  If the cases come too rapidly in time, there just aren't enough hospital beds.  This is why flattening the curve is so important.

It annoys me to see some people whom I generally respect scientifically seem to throw their numerical literacy out the window on this.  We shouldn't freak out and panic, but we should understand the underlying math and assumptions and take this appropriately seriously.

Update:  Notes from a meeting at UCSF (web archive version of link) hosted by, among others, Joe DeRisi.  I first met Joe when we became Packard Fellows back in 2003.  He's a brilliant and very nice guy, who with colleagues created the viral phylogeny chip that identified SARS as a previously unknown coronavirus and pinpointed its closest relatives.