That curve is pure bullshit. No real life event fits a polynomial with that much precision when it is sampled so poorly. They're cooked, both intentionally and because of limitations in equipment, personel, and ability.
Interesting data points from today's presser with Dr. Bruce Aylward of WHO[1]:
"Aylward pointed to an analysis from Guangdong province suggesting that, at least there, most of the infections were coming to the attention of health authorities.
When the virus started to spread in Guangdong — the province where the 2002-2003 SARS outbreak began — worried people flooded fever clinics to be tested. Of 320,000 tests performed, just under 0.5% were positive for the virus at the peak of transmission there, he said — which suggests that only 1 case out of 200 was being missed.
Transmission of the virus has subsided in Guangdong, and the number of positive tests at the fever clinics has declined; now only about 1 in 5,000 people tested at the fever clinics is positive for the virus, he said."
>Of 320,000 tests performed, just under 0.5% were positive for the virus at the peak of transmission there, he said — which suggests that only 1 case out of 200 was being missed
These numbers don't make sense. What, are they assuming 100% accuracy for the tests? China is an authoritarian dictatorship with a very personal need to save face - in order to maintain order it cannot show weakness. China is lying. There is motive, circumstantial, and now with the spread in Korea and Italy, empirical evidence that China is full of it.
I think China was looking at an existential threat to the government, and so made a calculation to trade a recession for survival. Their response was unprecedented, brutal and effective. I think it goes beyond face - lying at this stage would just lead to more Wuhans and eventually collapse.
At least that’s my current thinking - I was in your camp until a week ago or so.
Do you have examples of other epidemics where it was impossible to fit a quadratic? I've seen this argument a lot, but I've never seen anyone attempt to analyze how hard it actually is. I suspect that any data with positive first and second derivatives is pretty easy to fit closely to a quadratic.
That's the first time I've seen n=45,000ish been described as a 'poor' sample.
The only suspicious part I can find is the peak at the first of February, but that just seems like that date has been used as a catch all in case the exact date was unknown.
[1] https://jamanetwork.com/journals/jama/fullarticle/2762130