Algorithmic Probability From Scholarpedia

To compute the relative probabilities of different hypotheses can be dropped as it is an independent constant. Furthermore, for a given hypothesis it is often straight forward to compute . A conceptually more difficult problem is to assign the prior probability . Indeed, how can one assign a probability to a hypothesis before observing any data? A theoretically sound solution to this problem was not known until R.J. Solomonoff founded algorithmic probability theory in the 1960s.