Hintikka's research program for inductive logic started where Carnap's program had left off: laws of nature and their degrees of confirmation on finite amounts of evidence. Many of Carnap's critics took the problem to be the zero probability of laws of nature in infinite universes on finite amounts of evidence. So did Carnap himself. Hintikka was the first to see that the real problem lay elsewhere. His first move was to take as basic not predictive inference but universal inference. This choice had an important consequence. Whereas Carnap's problem was: given a partition of the universe in k cells, and a sample of n individuals such that n _i of them belong to the i-th cell, what is the probability that the next individual to be observed belongs to the i-th cell? Hintikka raised the following problem: given the same data, which is the best (most probable) general hypothesis concerning the whole universe?