https://insights.som.yale.edu/insights/in-defense-of-mathema...
"For those readers who are unfamiliar with it, the Imperial College model is the one that projected that 2.2 million Americans (and 510,000 residents of the UK) could die from COVID-19 over the next two years without public health interventions."
https://www.nature.com/articles/d41586-020-01003-6
"When updated data in the Imperial team’s model indicated that the United Kingdom’s health service would soon be overwhelmed with severe cases of COVID-19, and might face more than 500,000 deaths if the government took no action... The same model suggested that, with no action, the United States might face 2.2 million deaths..."
Here's the March 26, 2020 report, covering the rest of the world, with far more specific predictions:
https://www.imperial.ac.uk/media/imperial-college/medicine/m...
In this report, they predict 2.2 million deaths in the US in the unmitigated scenario, and over 1 million in the "social distancing" scenario. Their projection if the US "suppressed" the virus (by uniformly cutting contacts by 75% as soon as the death rate exceeded 1.6 per 100k; by my math, that's about 5,000 deaths) was 474,000 dead.
So, uh...that was wrong.
They projected 40 million deaths, worldwide (current actual total 3.14M) in the unmitigated scenario, with between 1.9M and 10.5M deaths in the "total suppression" scenario (depending on how soon lockdowns occurred).
If you go to the spreadsheet [1], you'll see lots of other fun projections:
* 470,000-670,000 deaths in Japan in the "social distancing" scenarios (current total: 10,031, despite living mostly normally throughout 2020).
* 330k-740k deaths in the "social distancing" scenarios for France (620k-760k "unmitigated"; current total: just over 3M, despite a year of strict interventions).
* Let's not forget their projection of 60,000-90,000 deaths in Sweden, in the "unmitigated" scenario (current total: 13,968).
Overall, their predictions bear very little resemblance to reality.
[1] https://www.imperial.ac.uk/media/imperial-college/medicine/s...
Okay. And the revised reports?
This is why, in most fields of modeling with any intellectual rigor, we have blinded prospective tests.
Those projections were still wildly off the mark. There is no way to look at their projections and take them seriously as modelers. They were simply wrong. They made bad parameter assumptions, didn't question their results or cross-validate, and ran with the scariest projections of their models.