3 ms·

If I predict an earthquake in San Francisco "tomorrow" every single day, at some point, I will be correct. The problem with "naturopathic" treatments is that p

by bsder 3y ago

If I predict an earthquake in San Francisco "tomorrow" every single day, at some point, I will be correct. The problem with "naturopathic" treatments is that people forget that there are 4 outcomes--not 2. There is treated-got better, didn't treat-didn't get better. Then there are also treated-didn't get better and didn't treat-got better anyway. A treatment needs to show appropriate statistics in all 4 outcomes. "Treated-got better" loses a lot of its causative power when there are also a lot of "Treated-didn't get better", for example.

Terr_ 3y ago
> If I predict an earthquake in San Francisco "tomorrow" every single day, at some point, I will be correct. Not quite the same issue, but a similar logical space: I sometimes give the example of someone living by a volcano and maintaining a sign that predicts it will not erupt today, claiming 100% accuracy. Thirty years later, it erupts, and they amend their sign: "99.99% accuracy". Point being that context and Bayesian prior probability are pretty important.