The paper discusses how one can try to analyze computations, and maybe computational problems from the point of view of information evolution. The considerations presented here are very preliminary. The long-standing goal is twofold: on the one hand, to find other vision of computations that may help to design and analyze algorithms, and on the other hand, to understand what is realistic computation and what is real practical problem. The concepts of modern computer science, that came from classical mathematics of pre-computer era, are overgeneralized, and for this reason are often misleading and counter-productive from the point of view of applications. The present text discusses mainly what classical notions of entropy might give for analysis of computations. In order to better understand the problem, a philosophical discussion of the the essence and relation of knowledge/information/uncertainty in algorith-mic processes might be useful.