Modelers & anti-modelers miss the point. This isn’t physics. We can’t run controlled experiments or make granular time-series forecasts.
There are only two parameters we really care about: how transmissible is the disease and how severe is it? Everything else follows from that.
There are only two parameters we really care about: how transmissible is the disease and how severe is it? Everything else follows from that.
Models for the trajectory of a ball thrown in the air tend to be accurate because we can repeat the experiment and fix what we got wrong.
But forecasting this type of stuff is at best directional.
Molecular biology is what’s crucial here, not models.
But forecasting this type of stuff is at best directional.
Molecular biology is what’s crucial here, not models.
At best, modeling this type of stuff is like looking at a business plan for a startup that forecasts out a few years.
Yes, it’s useful to see that virality is a key parameter. No, you don’t put a lot of confidence in the forecast for Aug 2023 revenue.
Yes, it’s useful to see that virality is a key parameter. No, you don’t put a lot of confidence in the forecast for Aug 2023 revenue.
Even for things like rough projections of how many folks worldwide may eventually get COVID-19 in an uncontrolled scenario, we can initialize and sanity check those models with hard data from molecular biology.
See this thread on serosurveys from Feb.
See this thread on serosurveys from Feb.
