Step 2: Figuring out the probabilities of hypotheses (Priors)URL copied
The wisdom in thinking like a bayesian lies in setting the probabilities to this hypothesis space. Firstly the sum of all the probabilities of these hypotheses sum upto one. Generally speaking, we set larger probability to events that have happened more in the past.
To give an idea of what I'm talking about - 'experience' tells us that the probability that I own a 373 faced dice is really small. The number of times one has seen such a dice is non-existent (what about a 10 Million faced dice?) but it's not impossible. So we set a probability of seeing that hypothesis as almost zero. What we do often see are dice with a fewer faces and the probabilities better reflect that. (Strictly speaking, the [priors don't matter]({% post_url 2017-04-23-priors-do-not-matter %}) when we have enough evidence)