Indeed, one has to scale that probability by our confidence in the model itself -- so the probability itself has a wide confidence interval. Ie., a more accura

Indeed, one has to scale that probability by our confidence in the model itself -- so the probability itself has a wide confidence interval. Ie., a more accurate claim is something like, `15x to 1500x`, with `150x` the best mean according to the model. (Or even `1.5x to 15,000x` -- it would be useful to know).