The same model predicted 96K deaths for Sweden, a country that eschewed lockdown measures or mask mandates [1]. The deaths topped at around 6K in Sweden. Sweden did take some precautions like restricting large gatherings but not enough to have an effect of this magnitude.

[1]: https://www.aier.org/article/imperial-college-model-applied-...

Ensure you're tracking whether the populace is taking additional measures that are not mandated, and suddenly it starts to explain the error almost completely.

Remember, policies do not execute themselves, and lack of policy does not imply lack of action.

Other problems are also caused by wrong, simplistic models. E.g. SEIR might have estimated the total death count correctly, but not the timing or location at all. (Because the model is isotropic.) Plugging in annualized data for various airborne viruses into a Bayesian anomalous diffusion model gives much closer timings. Though there are a few models with varying base assumptions and the estimator used a mixture. (One using Spanish flu, next using common cold, another using seasonal flu. Separately the worst case is... common cold variant.)

Or

> The coronavirus could kill millions of Americans: ‘Do the math,’ immunization specialist says

How is that failed? Do we have a crystal ball that says it's all over/ this virus can't possibly mutate into something deadlier? I do agree it feels "unlikely" but marking it as "failed" is just a prediction at this point.

what's even funnier is that the next prediction is already marked as "correct" even though we haven't touched 1Million global deaths yet.

> 'We Are At War,' WHO Head Says, Warning Millions Could Die From COVID-19

--- on a different note, this is also strange:

> 21 Indian cities will run out of ground water by 2020 Link Failed Resources 2030-40 2019

So it's made in 2019, but is it about 2020 or about 2030-40? Seems unlikely that it really is about 2020 if it was made in 2019.... unless it was all but guaranteed to become true (and then I'd like to know how they avoided it).

Similarly

> Coronavirus pandemic could cause 40 million deaths if left unchecked

What data can we base this on? Humanity has taken measures to ensure that the pandemic doesn't get to those levels, so any real-world data is obviously going to be much less than 40 million.

Thank you for pointing that. The evidence does suggest that deadly virus mutations aren't that likely, but yes, it is more accurate to say 'Unlikely'.