Revisiting the Covid Models on August 1
When the Covid-19 pandemic began, we all became amateur epidemiologists, learning about R-naughts, exponential growth, and death models. When March dawned, there were fewer than a hundred cases in the US, but things were bad in other places. Sometime in that period the model at the University of Washington became the go-to for doomscrollers who obsessed over predicted peak dates and numbers. Models were magnetizing because they seemed to give us some control over an overwhelming situation.
For whatever reason, researchers chose August 1 as the time horizon, predicting what the numbers would be in each state and the nation. In those days of late winter, that seemed like an impossibly distant date; the numbers of deaths those early models offered seemed outlandish. As cases exploded, our perception changed radically and by the end of March, when Anthony Fauci predicted total deaths from the pandemic would be between 100,000 and 200,000, it had start to seem plausible. Still, the models showed peaks in April and May, with deaths comfortingly tailing off by August.
We have arrived at that arbitrary date.The results?
- 145,477 (Covid Tracking)
- 151,499 (CDC)
- 153,391 (Johns Hopkins)
- 156,747 (Worldometers)
Yesterday the US recorded 68,000 new cases and 1,300 deaths. We did see peaks in April and May, but things didn’t tail off — now they’re peaking again.
Models, of course, depend on accurately predicting a variety of variables. In March, even infectious disease doctors weren’t sure how Covid spread. Did masks help? Early advice out of the CDC said no. Could people become infected through surface contact? It wasn’t clear for weeks. Washington came in for a fair amount of abuse when, in early May, it readjusted its August 1 deaths prediction from 72,000 to over 100,000. It was around that time we quite looking at the predicted deaths and instead looked at the actual numbers.
The biggest variable, of course, turned out to be the one we got the most wrong: what would federal and state governments do? Models predicted the US would address the pandemic the way other countries have. A nearly complete national shutdown to bring the infection rate down followed by contact tracing, the use of masks, strict limits on gatherings, social distancing, and other protocols. When those measures are in place, the infection rates follow a predictable pattern of decline. Here’s Italy, which had one of the earliest and most deadly spikes:
They are currently recording about 35 deaths a day. Or how about South Korea, which identified their first case on January 20th, a day before the US received its first? They have a current 7-day rolling death rate of less than one case — and have only had 300 cases in total (!).
That brings us to the US, which has been a disaster. The federal government, riven by an internal war between epidemiologists and a president trying to conceal bad news, abandoned leadership almost immediately. States, counties, and cities were forced to make their own decisions, and we ended up with a patchwork response no statistical modeler could ever predicted. The result?
It’s curious that we no longer think much about the models. We have shifted our doomscrolling to the actual numbers instead. If you’re curious, the University of Washington has adjusted its timeframe and now predicts 230,000 deaths by November 1. That is a convenient date, because it arrives two days before the election. We will see both what the actual number is and whether Americans feel it is disqualifying for a president. Trump’s numbers have softened this hot, deadly summer, but not the way one might expect for the country with by far the most mortality.
In the six months the pandemic has been galloping around the globe, we have continually readjusted our expectations. I remember in March wondering how we could possibly make it to August 1st if the virus was still active. It didn’t seem possible for a country to just shut down for five months. Now, of course we would love to think it would be over in five months. Instead, we worry that it will be a year: that August 1, 2021 might be a reasonable time to hope for normalcy to return. The models don’t extend that far into the future, and they wouldn’t be worth much if they did. Without a coherent national response, we can’t know what awaits.
We know there is not controlling this virus, not in the US and not under GOP rule. Now we just endure.