Thread by @Dierenbach

Fun fact #2:

Colorado's COVID-19 Modeling Group estimates that if we stopped social distancing and stopped wearing masks, 13,370,000 Coloradans would become infected with Covid-19.

Which is weird, as Colorado only has 5,841,000 residents.

Read on.
1/10

1st the model. The COVID-19 Modeling Group (CMG) made an app of their model. @GovofCO Polis said it's the same data used in CMG reports to him. Here is a screenshot of the model with social distancing & mask wearing set to 0.

It results in 5,513,600 Symptomatic Infections
2/10

The model is available at cucovid19.shinyapps.io/colorado/

3/10

BTW 5,513,600 Symptomatic Infections represents 94.4% of Colorado's 5,581,000 citizens. That's a Reproduction number of about 18.

The CDC estimates R at about 2.5.

But wait, there's more...

4/10

The CMG model also calculates Asymptomatic Infections. Here is a portion of a table where the CMG shows the values for the fraction of infected individuals who are symptomatic. Notice the incredible precision with which these values are known!

5/10

Using the data from the chart & the population of each age category, the population-wide fraction of individuals who are symptomatic is about .412

That is, CMG believes 41.2% of infections in Colorado are symptomatic

The CDC estimates that 60% of cases are symptomatic

6/10

If 5,513,600 symptomatic infections represent 41.2% of total infections, total infections = 13,371,000!!!

That is quite an accomplishment for a state of 5,841,000

And the point of all this:

7/10

@GovofCO Polis bases a lot of his decisions on what the CMG tells him. But the CMG model is deeply flawed and produces insane results when SD and masks are set to 0.

And as I wrote in this article, insane models result in insane actions.

8/10

thefederalist.com/2020/08/13/ins…

Final word: Please tell me if you see something wrong with my analysis (I'm looking at you CMG & @CDPHE). I have reviewed the code, parameters & docs, but it is Byzantine and I may have missed/misinterpreted something. I will correct any errors discovered.

Links follow
9/10

Model equations (lots of errors):

github.com/agb85/covid-19…

Model parameters (many seem unchanged since April)

github.com/agb85/covid-19…

Model documentation (mask numbers = wild guesses?):

github.com/agb85/covid-19…

10/10