This famous maxim by Yogi Berra, the great baseball player and creator of aphorisms, applies perfectly to computer research. The talk will discuss four types of case through a variety of examples. First, cases where the theory is clear but everyday practice confusing. This is the case with highly counter-intuitive probabilities, where we analyze in particular the strange reactions to Loto type games and the " Monty Hall Problem ", which made headlines in the 1990s. Then, we'll look at cases where theory is not very effective but practice is surprisingly efficient, in particular the Boolean calculus, a typical NP-complete problem that was never thought to be solvable in practice, but which is nevertheless solved surprisingly well in large-scale real-life cases - although we don't really know why. We'll also talk about some pseudo-theoretical false beliefs that are widely held but have been overturned in practice, such as the end of Moore's Law for circuits, or the infinite power of quantum computers. We'll then look at a case where theory and practice alternately cooperate or conflict over the course of time: the evolution of programming and its languages. We conclude with cases where theory and practice are virtually fused, for example in modern randomized algorithms, in the formal verification of programs and theorems, or in the simulation of physical and biological phenomena.