Gotta be fair to Trump here. He didn’t come up with the 60,000 figure himself; that was the projection made by the IHME model in early April, when the curve began to bend thanks to Americans staying home from coast to coast. Even Anthony Fauci, who’s grumbled behind closed doors about the imperfection and variability of models, blessed the 60,000 figure on April 9:

“The real data are telling us that it is highly likely that we’re having a definite positive effect by this mitigation things that we’re doing – this physical separation – so I believe we are gonna see a downturn in that,” Fauci, the director of the National Institute of Allergy and Infectious Diseases, told NBC’s “Today” show.

“And it looks more like the 60,000 than the 100,000 to 200,000,” he said.

The IHME model has been rapped all along for being too optimistic. (About the death toll, I mean. It was far too pessimistic in projecting the number of hospitalizations.) In reality, we passed 60,000 deaths in the U.S. a few days ago. The model currently foresees 72,000 U.S. deaths by August 4, a ludicrous underestimate. If we average the same number of daily deaths this week as we did last week we stand a chance of passing 72,000 just seven days from now.

All of which is to say: (1) We should stop paying attention to IHME, at least when it comes to forecasting deaths, and (2) Trump lowballed the death toll initially only because the experts around him were doing so as well.

That said, I don’t know why he’d lowball it *again* by predicting 80-90,000 deaths in his townhall last night with Fox News, as you’ll see below. There’s a chance we’ll pass that number by the end of *this month,* especially with many states beginning to relax restrictions. All he’s doing by continuing to offer forecasts is handing Democrats’ attack-ad material. The script writes itself: First shot is of him projecting 60,000 deaths, then the second shot is of him in this Fox segment projecting 80-90,000, then the third shot is of him later this month saying, “It looks more like 100-120,000 deaths,” then the fourth shot…

You get the point. Dems will use it as a microcosm of incompetence. “The administration bungled its response so badly it had to keep raising its estimates of how many would die.”

Trump: That’s one of the reasons we’re successful, if you call losing 80k-90k people successful
Bret: That number has changed Mr. President. You said 60
Trump: I used to say 65k and now I’m saying 80 or 90 and it goes up and it goes up rapidly pic.twitter.com/v4FbbygF5Q

— Acyn (@Acyn) May 3, 2020

Birx was asked about the projections yesterday on Fox News as well and wisely played it conservative, reminding viewers that the White House model projected 100-240,000 deaths, and that was with “full mitigation,” i.e. sustained social distancing. Hopefully we’ll come in under 240,000. Although in a worst-case scenario where weekly deaths remain at around 10,000 per week throughout the summer, we could conceivably pass that number in September.

TRUMP, April 20: "We’re going toward 50 or 60,000 [deaths]."

BIRX, today with death toll at 68,000: "Our projections have always been between 100,000 and 240,000 [deaths]." pic.twitter.com/js68n55bSW

— Aaron Rupar (@atrupar) May 3, 2020

Why was the IHME model off so badly? Specifically, why are there still 30,000 new cases and 1,500-2,000 deaths daily despite the draconian measures Americans have taken to crush the curve and bend it downward? David Wallace-Wells speculates:

It could be there is some feature of the disease we do not yet understand, or something about how long it incubates that we have failed to recognize. It is possible the disease remains infectious even after a patient has recovered, perhaps for a significant amount of time. It is possible that sheltering at home is a flawed quarantining method, since family members can infect one another, and that some meaningful share of the new cases come from within quarantines. It is possible that some meaningful share comes from within hospitals, with doctors and nurses and administrators all serving as vectors of the disease. It is possible that some meaningful share comes from the 50 million exposed frontline workers — employees at grocery stores and meat plants and pharmacies and the post office and so on — who are laboring (without hazard pay) to make the quarantines possible for the rest of us. Quite likely, a few of these factors, and possibly all, are contributing, to some degree (and 900 workers at a single Tyson Foods plant did just test positive). Together, they are preventing anything resembling a decline of deaths or cases as rapid as the ascent a month or so ago.

Birx went on Jeanine Pirro’s show on Saturday night and was pressed about whether maybe COVID deaths are being overcounted. (You can watch that at 1:53 here.) Isn’t it true, said Pirro, that patients infected with the virus who have a severe underlying illness are being counted as coronavirus deaths even though we don’t know for sure in some cases that the virus contributed to that death? That argument, that people are dying with COVID but not of COVID, has been pushed by Trump-friendly skeptics since the outbreak began as a way of suggesting that maybe he’s done a much better job reducing fatalities than anyone thinks.

Birx reminded her that we know people with underlying health problems are more susceptible of succumbing to the virus, but that just leaves Pirro-ites in the position of arguing again that it’s the underlying problem, not the complications from the virus, that have caused the death. So here’s a data-driven way to think about it: According to two separate studies, people with underlying health problem are dying much sooner than we’d expect even by the standards of lower life expectancy associated with their underlying problem.

“Some people think that these people dying would have died this year anyway,” said one of the lead researchers to the WSJ. “That’s simply not the case.” As you can see, even septuagenarians with “severe” health problems are losing eight years or so off their projected life span. If they’re dying “with” COVID rather than “of” COVID, why is that?

Relatedly, there’s the now famous phenomenon of “excess deaths.” If people are dying “with” COVID rather than “of” COVID, we would expect the number of deaths from all causes in April 2020 to look similar to the number in April 2019. After all, even in normal times we expect a certain number of deaths each month from cancer, heart disease, auto accidents, and so on. If coronavirus isn’t causing more people (or many more people) to die this year then the latest overall mortality figures shouldn’t look wildly different from how they looked last year. In reality, they look wildly different:

If people are dying “with” COVID, we’d expect the number of U.S. deaths to track within that gray band. What we’re actually seeing is a massive spike from confirmed coronavirus deaths (that’s the shaded orange area) plus many thousands more deaths — 37,000 so far, per WaPo — above that in the unshaded orange area. Those are the “excess deaths,” ones that haven’t been confirmed via testing as caused by COVID but which nonetheless happened, waaaaaay above the number of deaths we’d expect for this time of year. This is why epidemiologists continue to insist that the death toll from the disease is being undercounted, not overcounted. We can’t say for certain that every “excess death” was due directly due to the pandemic, but given the baffling spectrum of symptoms doctors have seen in patients that could cause sudden death at home (strokes and “happy hypoxia” are the two that are most well known), it’s inescapable that the true death toll is higher than the official toll.