The Failure of Imperial College Modeling Is Far Worse than We Knew

Figure II: Performance of Imperial College Modeling in 4 Non-Lockdown Countries & the United States

Country (assumed R0 = 2.4) Imperial Model projected deaths – social distancing (lockdowns) Imperial Model projected deaths -unmitigated spread 1 year actual deaths (3/26/21) Overestimate, Lockdown scenario Overestimate, Unmitigated scenario Overestimate Percentage – Lockdowns Overestimate Percentage – Unmitigated
Sweden 30,434 66,393 13,496 16,938 52,897 126% 392%
Taiwan 93,712 179,828 10 93,702 179,818 937020% 1798180%
South Korea 141,198 301,352 1,716 139,482 299,636 8128% 17461%
Japan 469,064 1,055,426 8,967 460,097 1,046,459 5131% 11670%
United States 1,099,095 2,186,315 563,285 535,810 1,623,030 95% 288%

As can be seen, Imperial College wildly overstated the projected deaths in each country under both its “unmitigated” scenario and its NPI-reliant “social distancing” scenario – including by orders of magnitude in several cases.