Science relies on formulating, confirming or discarding of sequences new hypotheses. They are usually checked according to their harmony with real-world data. Bayesian methodology suits well to this purpose but needs to assign judiciously prior probability to each hypothesis. A wrong choice may be either harmful to a brilliant hypothesis or may call for too many data to discard a nonsense.
Seminar is devoted to the choice of prior probability and to practical consequences of the proposed solution. They include: a) a view on the problem of “something else”; b) a solution of problem of initial data; c) a way to efficient sequential Monte-Carlo estimators.