The similarity of discussion/constraints on the Infected Fataility Rate (IFR) for covid-19 and the Equilibrium Climate Sensitivity (ECS) for carbon dioxide, is interesting.
Both IFR and ECS are model variables that have profound implications for policy but that aren’t directly measurable.
It’s easiest to think of them as constants, but there is a suspicion that they may vary depending on context (background climate for ECS, specific populations for IFR).
Constraints come from imperfect observations, interpreted though imperfect models. Building in the uncertainty in both is key for any robust assessment.
Tying multiple constraints from different approaches together is always tricky. But Bayesian methods are being used in both cases.
But, and here is the rub, we all want the numbers (IFR and ECS) to be low, thus there is an inbuilt bias in how this discussion is conducted.
New studies that come up with low numbers will be seized upon by the media/politicians and the hopeful public. If there are methodological or statistical problems with them or they conflict with other constraints, the domain scientists will find themselves in the uncomfortable...
...position of arguing against optimistic projections that they themselves might prefer. They will then be tagged as alarmists or as inappropriately wanting a bigger policy response.
Large cooperative assessments are somewhat useful in coming up with appropriate conclusions from the multiple lines of evidence including the uncertainty, but this takes a long time and is a significant effort. Can this happen quickly in the covid situation?
It’s important to know that there are some numbers that are probably just too good to be true. The situation in NYC implies IFR > 0.2% even if everyone was infected (which they aren’t) & the Last ice age implies that ECS > 1.5°C. The most-likely values are larger.
So if a study suggests smaller values, they have to consistently explain why these lower bounds are not valid. If they don’t, they will be pretty harshly criticized.
As knowledge grows, the variety of constraints will increase & any new study that suggests a super high or low value will need to do a lot of work to overturn the emerging consensus. At which point, expect a bunch of articles whining about how science isn’t done by consensus! 🙄
