As of February 21, 2020, the Metaculus community expects that, in the median case, the novel coronavirus will cause fewer deaths than seasonal influenza. However, what is so concerning about this new disease is its tail risk: it has a small but non-negligible probability of being extraordinarily bad.
For various reasons, the standard Metaculus question about the number of deaths is too crude an estimate to get a good sense of the tail risk posed by this virus. There are too few predictions at the extreme ends, so the values tend to fluctuate based on few data points. Moreover, the slider UI has some limitations that make it hard to be precise about low-probability assignments.
Accordingly, it makes sense to ask a specific question about tail risk. That is the goal of this question. The conditional structure requires predictors to focus purely on the scenario where COVID-19 infects 100M people or more.
At the time of writing, the community only assigns ~25% probability to the proposition that the estimated case count will exceed 66 million before 2021.
Resolution critera: If the question about the number of estimated cases resolves with a value greater than or equal to 100 million, this question resolves with the same value as the question about death count. Otherwise, this question resolves ambiguously.