Sep 28, 2020

The mathematics of covid incidence modelling: what's wrong with Taylor v. Newfoundland?

Taylor v. Newfoundland and Labrador, 2020 NLSC 125 (CanLII)

There is a model that public health theorists use to estimate the impact of a contagious disease. There is a lot of math, always a selling point when you’re in the business of telling ordinary people how enormous and dangerous the trivial risk is. Who can confront these public health theorists with their mathematical models, the use of which results in the destruction of civil liberty?

The SEIRS model is all the rage among ‘public health’ types. It was the SEIRS model that estimated millions and millions of deaths than didn’t happen. As Joe Biden would say, ‘probably 200 million people will die before I finish this talk’.

It was the SEIRS model that convinced Burrage J. in Taylor v. Newfoundland and Labrador to revolutionize the constitution and give a province the power to close its border to its own people. What is this all-powerful tool that has superior court judges bending the knee?

SEIRS stands for Susceptible +Exposed + Infected +Recovered. By a hydraulic assessment of these four components ‘public heath’ theorists have successful estimated all 200 million of the 200,000 actual deaths that have taken place in the United States. I hope you get my point.

I say this is hydraulics because, for Rahman, (the ‘expert’ that Burrage relied upon in Taylor), to get to his mathematical result, Rahman had to do a few things to the data. Here’s what Rahman said:

We have assumed that the fraction of exposed and infectious travelers to Newfoundland and Labrador is similar to the exposed and infected fractions at the locations of origin.

Here’s what that little assumption allows Rahman to do:

Now we know that Florida had a total of 153,000 cases (at the time that Rahman did his calculation). That is an infection rate of 0.0007 or 7 in every 10,000 people.

Well we know that Newfies love Florida. There are 46,600 arrivals from Florida to Newfoundland each year. That is roughly 46,600 Newfoundlanders in the Florida covid-soup. Since 7 of every 10,000 Floridans will get covid, then that means there will be 7 for every 10,000 Newfoundlanders who will also get covid from going to Florida. So that is 7 x 4.6 = 32 Newfoundlanders will come back to the province with covid and start spreading it around.

So close the border because 32 Newfoundlanders will bring back covid if permitted to enter Newfoundland. With such math is your civil liberty lost.

There was no expert counterattack in Taylor v. Newfoundland and Labrador. The field was clear for easy acceptance of Rahman’s math. For instance, a skeptic’s attack upon the SEIRS model was not done.

For instance the following chart was not shown to Burrage. I joked about Joe Biden’s incoherent assessment of 200 million dead before he finished his speech. Biden came by such nonsense honestly. Few people are able to effortlessly and accurately analyze statistical inferences. For example, how many readers draw, from the difference between the green line and the orange line below, how utterly unknowable the true infection rate is. Green says that the case peak is a mere 300,000, Orange says the case peak is 600,000. Most people, including judges, are not the types to indict the models because of this arbitrariness. Judges don't discard the models because judges are not fluent in statistical model weakness. But the mere common law of liberty is not good enough for them to operate on. So they grab hold to any carnival barker with a chart and some curves on it.



Below I show the same date for Canada.


What is the judiciary to do with such statistical evidence? One obvious answer is to insist upon the delivery of such opposing statistical analytics that the chance of an absurd result is reduced. There remains the reality, that liberty has been taken away by what amounts to a statistical hypothetical, that has nothing to do with the individual whose liberty has been infringed, is the new thing in Canadian law.