Christian AlthausKonto zweryfikowane

There were certainly problems with that model, but we should be careful when criticizing a model that assumes no significant interventions for failing to predict what happened when there were interventions.

I'm criticizing the CDC decision's to publish that model, later claims that their over-prediction was due to intervention, and our community's defense of unrealistic social mixing models that we know lead to over-prediction.

From talking to modelers, I got the impression the CDC's model was just way to simple and basically they didn't know what they were doing. Is that not true?

Whether the CDC intentionally published over-estimated projections is still a matter of debate. But I would certainly argue that the simplifying assumptions used in their model accounted for much of the overestimation and strongly biased their results on interventions.

is there any evidence they "intentionally" overestimated? Or is that just a hypothesis?

None.

Great thread. I'd also add that ~who (or in this case, what entity) publishes results from a lousy model matters, too. As Sam's histogram notes, many forecasts performed well (enough) – but because it was the CDC that overshot, we're still talking about it today. 1/2

Don't get me wrong; I think it's great that we're still talking about it, precisely for the reasons Sam raised... But I don't think we would be if it hadn't been the CDC (or an equally powerful stakeholder group) that overshot and then (unethically) tried to explain away. 2/2