A few more papers, still trying to catch up.
There have been a million papers with a million models on mitigation of spread tactics and what happens if we lift lockdowns. They are mostly abstractions with little basis in real data. Here is an example, just so you know I read everything, even stuff I don’t thing is worth paying much attention to. (Medrxiv Paper) The authors looked at Washington State’s experience and used an agent-based model to examine the effects of “non-pharmaceutical interventions”. That is a BS euphemism; it’s an extreme lockdown. They say that these tactics as used in Washington “flattened the curve”. They use the usual bad assumptions about infectiousness, severity of illness, etc. They lump the frail elderly into a group with everyone else. Their conclusion based on the model is that any relaxation of the lockdown results in a bump up of infections. Yeah, if you don’t allow some measure of spread, you are going to live with these lockdowns forever.
Next up, another paper assessing strategies to suppress spread. (Medrxiv Paper) These authors are one of several groups who assess the usual tactics. They used that standard SEIR model approach to assess three broad categories: social distancing; active protection and active removal of carriers. Active protection means identifying and removing exposing individuals, and active removal is quarantining. Although this is basically just simulations, and again the assumptions don’t appear to jive with the data, interestingly there is a suggestion that just isolating vulnerable populations can work as well as other approaches.
And yet another paper on models, looks at improving how various segments of the population mixes affects the results the model delivers. (Medrxiv Paper) It compares a model which basically treats the population as homogenous units versus one that looks more at infection as occurring within localized clusters. Then the authors engage in a curve-fitting exercise with data from actual epidemic course in certain countries, which of course is totally inadequate data, so the output is just worthless.