I present a method for optical computing based on white-light interferometry. The problem to be solved is coded by optical path lengths and the superposition of all possible paths that a photon can travel is used for computing the solution. The solution itself is chosen by interference with the reference light. Several gedankenexperiments demonstrate how this method can be used for solving computational hard problems. Especially, I will introduce
the basic principle with two maze-type puzzles and then concentrate on how the method can be employed for performing ultra-fast (sub-ps) digital-optical arithmetic with arbitrary precision.