March 26th, 2020
Time to reflect a little on the hysteria surrounding the corona virus outbreak. Every day you can hear some frenzied journalist (especially in the US) rabbiting on about how many millions will die because a) President Trump, b) President Trump, c) President Trump. Actually, after a shaky start he seems to have come to grips with the issue quite well; would that we could say that about the political class in general. The attempts by Nancy Pelosi and the Democrats in Congress to insert billions of pork-barrel items into a piece of emergency legislation beggars belief. It’s nothing but venality to the n-th power.
Every day a new “study” appears filled with new prognostications and predictions for the future. Let’s put aside the predictions for the political future, or futures, and take a look at the science picture. The main subject of discussion has been a paper from Imperial College that modeled the likely outcomes in terms of cases and deaths based on certain prior assumptions:
We assumed an incubation period of 5.1 days. Infectiousness is assumed to occur from 12 hours prior to the onset of symptoms for those that are symptomatic and from 4.6 days after infection in those that are asymptomatic with an infectiousness profile over time that results in a 6.5-day mean generation time. Based on fits to the early growth-rate of the epidemic in Wuhan10,11, we make a baseline assumption that R0 = 2.4 but examine values between 2.0 and 2.6. We assume that symptomatic individuals are 50% more infectious than asymptomatic individuals. Individual infectiousness is assumed to be variable, described by a gamma distribution with mean 1 and shape parameter alpha = 0.25. On recovery from infection, individuals are assumed to be immune to re-infection in the short term. Evidence from the Flu Watch cohort study suggests that re-infection with the same strain of seasonal circulating coronavirus is highly unlikely in the same or following season (Prof Andrew Hayward, personal communication).
It was this paper that led to the change of course of the British government.
From this, and the use of the modeling algorithm, they can make predictions of outcomes after making various changes to attempt to modify R-naught and bring down the rate of infection. R-naught is not just a function of the virus, but a function of other things such as the different behaviors of the population like social distancing. Based on certain of these assumptions, this is where the prediction of nearly half a million deaths came from.
Another paper out of Stanford University claims that the prognosis is way over-estimated (can’t find the link right now), but a paper in The Lancet addresses a small study from China. Also, Tomasso Dorigo, an experimental physicist at CERN, thinks that the hype is turning physicists into crackpots.
Although it’s early days, some caveats need to be borne in mind.
First, computer models do not produce evidence of anything. Repeat that to yourself.
Second, computer models produce conjecture—not data.
The models are exactly that, they produce numbers (often displayed with very pretty graphs and diagrams) that are generated by an algorithm operating on a given set of assumptions. The numbers coming out are only related to the initial parameters and the algorithm in the software, which may represent the real world accurately—or not.
Third, data are generated by performing scientific experiments and making measurements and observations of the world around us. This is evidence.
Fourth, when the data match the output of the model then, and only then, can you say that the model may be a reasonably accurate representation of the real world. Note that any change in the parameters in the algorithm or any change in the logic path in the algorithm can lead to radically different computational outcomes. This happens all the time in modeling.
In computer science, the GIGO Principle is undefeated: Garbage In—Garbage Out. Computer models are fine as far as they go, but reality gives data.
Hence, suddenly, claims that the Imperial College model overestimates the numbers of cases. If the coronavirus has infected many more people prior to the panic, who then developed antibodies and they have never shown symptoms of disease, then the case fatality rate will be much lower than heretofore believed. However, we can’t know this until extensive antibody testing is done on the population—all the population not just sick people.
Even if that is true, the tsunami effect on the health care system is still just as real, but the time frame may be much shorter.
The less information we have, the greater the uncertainty. Both these views of the problem may be wrong (they can both be wrong but they cannot both be right!). Millions of tests must be done to make the enemy visible. If there are large numbers of people with antibodies, then they are immune and can get back to work and get the economies moving again, but this can only be ascertained by testing for the antibodies, not just the antigen.
All these discussions between scientists are perfectly normal and good—the science is never settled. That’s only in Al Gore’s fantasy world. Only journalists and politicians think they’re always right. And remember the words of the great physicist Richard Feynman—
…any scientist talking outside his field is just as dumb as the next guy.
When someone you’re talking to keeps dumping on President Trump, remember, he’s the guy making the political decisions—that’s what he was elected to do. After this thing is over, you’ll be glad there was an alpha-male in charge.
[Update: see science20.com article [here]. Could the predictions be out of line?]