Call me crazy, but I recently listened to the whole four and a half hours of Lex Fridman’s interview with Sam Harris. I will not comment on the latter’s attempts at gardening in the moral landscape, but seize the opportunity to test Substack’s LaTeX functionality.
At one point, Harris mentions in passing that 300,000 Covid deaths in the United States might have been avoided if only people had not been so hesitant about the vaccines. This is difficult to grasp because it is a claim about one thing (the vaccines) but expressed in a unit (actual deaths) which fits another thing (people). We shall therefore convert his statement into a more appropriate unit, vaccine efficacy.
The total number D of Covid deaths splits into vaccinated and unvaccinated deaths:
Given a vaccination rate V, vaccine efficacy E can be defined as:
Solving this equation for the number of vaccinated deaths gives:
Now we can, by tracking the vaccination rate V and the number of Covid deaths, compute the number of vaccinated deaths, given a vaccine efficacy E. Using OWID data on vaccination rates (two jabs, or “fully”, as the euphemism goes) and Covid deaths (D = 1,112,766 until March 10, 2023), we get:
If the vaccines were 100% efficacious, of course, there would be zero vaccinated deaths. But even if the vaccine was a menace, doubling the chance of dying from Covid given infection (which means efficacy is –100%), only 438,637 deaths would have to be classified as vaccinated. Why is that? Because vaccination only started at the end of 2020, and slowly converged to a vaccination rate of just below 70%. In fact, average vaccination rate over the whole three-year period is around 38%, and 38% of 1,112,766 is in the ballpark of 438,637.
Now, if we want to hypothetically reduce the total number of deaths, we have to increase the vaccination rate (and assume that vaccine efficacy is larger than zero). The vaccination campaign reached maximum momentum around the beginning of May, 2021, and I extrapolated from there to create a maximum-vaccination scenario. This diagram shows both actual and hypothetical vaccination rates:
All else being equal, particularly assuming that vaccination rate has no influence on infection rate (which is not that unreasonable an assumption), we get hypothetical numbers of vaccinated deaths by scaling with the quotient of vaccination rates, and hypothetical numbers of unvaccinated deaths by scaling with the quotient of unvaccinated rates:
We can then compute the hypothetical total numbers of deaths and the differences to the real world:
There it is! If Sam Harris thinks that 300,000 Covid deaths might have been avoided by vaccinating to the max, he believes in vaccine efficacy of around 80% against death. Mind you, no boosters have been involved in this calculation.
For the conversion of Sam Harris we might have to wait a little longer. But LaTeX in Substack is great. Please, Substack developers, consider also inline LaTeX expressions.
I think Sam Harris is implying something like this: There were 1.12M people in the US who have died of covid. Of these, X percent were antivaxxers. If these darned antivaxxers had gotten a vaccine, we would have saved 300,000 lives!
Your calculations take the number of lives saved, 300,000, and uses this to calculate the VE, which was found to be 80%.
What I don't get is how you are looking at deaths through time alongside vaccines through time. Your max possible vaxxed chart makes sense. It seems to me that the possible deaths that could have been avoided are those deaths after the hyphothetical 100% vaxxed date, or 3/9/2021. I see that the max vax rate actual asymptotically approaches 70% from 55% from 3/9/21 to 3/3/23.
Anyway, I don't see how you came up with the distribution of the number of deaths that could have been avoided based on when the vaccines could have been taken.
I have come to realise that the pwoplw I used to think were smart, the smartest even, are actually a bunch of thick bellends; and those that I used to think were crazy, are actually smart as fuck. Sam Harris is an utter douchebag.