Welcome to the Bayesian track. First, a question about what a probability is.
When someone says “the probability of X is 70%”, there are two things they might mean:
- Long-run frequency: if you repeated the situation many times, X would happen in about 70% of the repeats. This is the reading behind casinos, quality control, and most intro stats courses.
- Degree of belief (the Bayesian reading): given everything I currently know, I’d treat X as 70% plausible — I’d take one side of a bet at those odds but not the other.
The frequency reading needs a repeatable situation: flip the coin again, pull another widget off the line. The belief reading doesn’t — it applies to any proposition, including one-off events that will happen exactly once or already happened.
Here’s the useful part: both readings obey the exact same math — probabilities are between 0 and 1, mutually exclusive options sum, and so on. The rules don’t care which reading you use. What changes is scope: which claims you’re allowed to put a number on at all.
Look at the four claims in the question. Three of them describe repeatable situations. One of them is about a single, unrepeatable fact that is either true or false right now — and yet putting a number on it is exactly what you do every time you debug.