There are some follow up tweets.
To the commenters shocked that Zvi’s probability numbers don’t come from an “objective calculation”:
Not shocked. At this point I am numb to it though.
- Approximating numerical probabilities is required if you want to make coherent decisions.
Average bayesist when they need to pee:
“Hmm, I suspect I need to pee. The decision I now must make is whether or not I should use my toilet or my sink.”
half an hour of examining priors, searching LW for urination sequences and setting up a prediction market later
“Hmm, it appears I have peed my pants.”
- Probability is subjective (a property of our belief-state) for us Bayesians.
Oh my god, he admit it!