Understanding adaptivity: Random systems revisited

Abstract

We develop a conceptual approach for probabilistic analysis of adaptive adversaries via Maurer's methodology of random systems (Eurocrypt'02). We first consider a well-known comparison theorem of Maurer according to which, under certain hypotheses, adaptivity does not help for achieving a certain event. This theorem has subsequently been misinterpreted, leading to a misrepresentation with one of Maurer's hypotheses being omitted in various applications. In particular, the only proof of (a misrepresentation of) the theorem available in the literature contained a flaw. We clarify the theorem by pointing out a simple example illustrating why the hypothesis of Maurer is necessary for the comparison statement to hold and provide a correct proof. Furthermore, we prove several technical statements applicable in more general settings where adaptivity might be helpful, which can be seen as the random system analogue of the game-playing arguments recently proved by Jetchev, Özen and Stam (TCC'12). © International Association for Cryptologic Research 2012.