Hacker News Reader

Comment by CapmCrackaWaka | Hacker News Reader

Ehh I attribute it to the rise of wannabe data scientists who want to fit a sigmoid to everything, as well as influencer culture. People would be trying to get their LinkedIn likes no matter what leadership came out with.

View on HN

Well, if the government was matching a curve to it, on a public website that had updates daily, supported by professionals, then I think there would be less demand for random people to do it.

The problem is that the curves anyone today draws are guaranteed to be wrong. We have almost no idea about how transmissible the virus is, how many people have it, how many people die from it. Even worse, we have a lot of pieces of data that look like information but aren't. For the most obvious example, a lot of people are tempted to divide the count of known deaths by the count of known cases, and they calculate a number which looks like a mortality rate but is multiple times higher than any actual estimate of the mortality rate.

Any curve you see today which claims to outline the progress of COVID-19 (rather than the progress of COVID-19 testing and response) is extremely misleading.

Part of the problem is people who think mortality rate is the same as case fatality risk (CFR). You are one of them too, so look it up.

Every article I've read reported the WHO's statement using the words "death rate" or "mortality rate". I understand that in the technical sense, the mortality rate is always relative to the total population whether or not they're infected, but I don't think anyone is getting confused by this, and since I'm not myself an expert I'm going to use colloquial terminology over epidemological jargon. Nobody thought the WHO's 3.4% number was saying that 3.4% of the world population is dead.

Then have a rolling history of curve predictions, day-over-day, maybe with some commentary about what seems to be driving changes in the curve (increased tests, changes in transmissibility estimates, a sudden outbreak, etc)

Can't fit a sigmoid if you haven't found the inflection point yet.

Yes and I've heard the sigmoid described as "the first half being exponential, the second half being logorithmic" all over the place and it bothers me to no end. The natural log of x goes to infinity... I sure hope that doesn't describe infection rates.

Reply on news.ycombinator.com