When I said that the US is having one Wuhan a day (50k/day #coronavirus cases in the US vs. 50k for all time in Wuhan), many people responded: "You can't trust the Chinese numbers."

There are so many layers to that answer that it's worth deconstructing. Thread [1/23]

Before I start, and to preempt the hordes: I am no fan of the Chinese authoritarian government. The simple fact that I feel (mostly) safe saying this is a testament to how much I cherish the core of what I think makes our system better: freedom of speech [2/23]

1. Equivalence
It misses that the same thing is true for the US, where the pres. asked to lower testing, positivity is ~15% (of ppl), or the Florida admin decided to fire the head of coronavirus analytics because she didn't want to cook the numbers. [3/ https://www.npr.org/2020/06/29/884551391/florida-scientist-says-she-was-fired-for-not-manipulating-covid-19-data

2. Worldwide accuracy
As if most countries actually had any idea of the true extent of their cases. Country Case Fatality Rates range from 0% to 26%. Which means many countries have no idea what's going on. [4/23]
https://www.worldometers.info/coronavirus/ 

3. CFRs
As of today, the world's average Case Fatality Rate is 4.7%.
The CFR for Hubei is... 4.8%.
The US' is... 4.6%
[5/23]

4. Assumes the Chinese are oracles
Chinese officials would have needed to cook the numbers of both cases AND deaths in real time to get to numbers similar to what the world would experience 4 months later. How crazy! The virus was new. How did they know where to aim? [6/23]

It's impossible. The most likely explanation is... Their CFR reflected what they were seeing, and if they were lying, it was in real time both and proportionally for cases and deaths. That sounds hard. [7/23]

5. Two weeks
Wuhan's cases went down exactly 2 weeks after the imposition of the lockdown. This is exactly what ended up happening in France, Spain, Italy and New York (see charts 7.a and 7.b). Same thing: You can't make that stuff up. It has to come up with real data. [8/23]

6. Chest scans
The Chinese were short on PCR tests. So they used CT scans.

If they were lying so much, why would they add suddenly thousands of cases from diagnostics through CT scans, a method that most countries don't even follow yet?
No reason when you want to cook numbers

7. Orders of magnitude
What these last two points indicate is that, if the Chinese lied, they had to lie in real time, coordinating for Wuhan both cases and deaths so they would remain consistent. The only way they could have lied, therefore, is in orders of magnitude. [10/23]

8. Undercounting cases
Maybe instead of 50,000 cases and 3,900 deaths in Wuhan, there were 500,000 cases and 40,000 deaths?

Except this is exactly what is happening in the West! [11/23]

Eg, NY has a total of 220k confirmed cases. But seroprevalence surveys suggest 20% of the city got infected. With 8.4M ppl —> 1.6M true cases... Or an undercounting of 85%!! [12/23]

9. Undercounting deaths
A paper analyzed crematoriums in Wuhan working 24/7 when they used to work 4h/day, and urn purchases. According to their calculations, that meant total deaths were 10x what was officially communicated. [13/
https://www.medrxiv.org/content/10.1101/2020.05.28.20116012v2.full.pdf

But the true death rate of Mexico's capital DF is probably 5x what is officially communicated. In Spain, it was at least 3x at some point. [14/23] https://twitter.com/tomaspueyo/status/1277903370083155968

Moreover, you can imagine factors such as Wuhan cremating all the dead bodies from all of Hubei, crematoriums left to burn for expediency, or some other factors that might somewhat reduce that 10x to a more realistic number [15/23]

10. Intent
That brings us to intent. I'm a big believer in Hanlon's Razor: "Never attribute to malice what can be explained by stupidity."

I'm sure the Chinese covered up the beginning of the crisis, and probably cooked the numbers of cases and deaths somewhat. [16/23]

But assuming that all their issues were due to malicious data manipulation sounds ridiculous to me.

Most countries in the world have had months to prepare since Wuhan exploded, and yet they are botching their responses, getting their numbers wrong, undercounting cases & deaths..

If everybody is getting this wrong because they can't do any better, why would China, the 1st country in the world to face the virus, have been in a better position? They should have botched the response even worse than any other country. [17/23]

Assuming that they were in a position to be cleverer than any other country gives them much more credit than they deserve.

A more parsimonious explanation is that they did just like every other country. [18/23]

All of the above addresses what I think about China's number cooking. If they did it, they were damn good at it. A more logical explanation is that they just botched it like many others. [19/23]

11. Prevalence
Another response I got was "You can't compare a city of 60M inhabitants with a country of 330M inhabitants."

Ok, then let's compare the US with the EU (500M), or India (1.3B), or China (1.4B), or... No matter how you take it, the US's prevalence is so. much. worse

12. The Forest for the trees
Finally, back to the original comment. If I say "The US has one Wuhan a day", those answering "Do you trust China's data?" are *completely* missing the point. [21/23]

The point of the comparison is:
"We all have a perception of what happened in Wuhan: hospitals collapsed, new ones spun up in days, people dying in droves...

That perception is tied to an interpretation of 50k cases over ~3 months.

We have that every day in the US." [22/23]

The fact that it might have been 300k or 500k cases misses the point. This is not what we thought at the time.

At the time, we thought it was 50k, and it was devastating, and we didn't want that for us.

And yet here we are. [22/23]