If you think of playful activities, software testing may not occur to you, but it is useful - especially in the case of distributed systems - to see (and organize!) software testing as a game that the tester plays with the implementation under test (IUT). We will argue the importance and indeed the necessity of the game-theoretic approach to testing. Does it relate to the automation of testing? It does. In fact, it is a testing tool that plays with IUT. Using good math (linear programming, Markov decision processes), we will analyze a particularly ubiquitous game and will find out when exactly optimal strategies exist and how to construct one. We will also discuss related software engineering issues and in particular this one: Does it pay to optimize a strategy instead of using the time directly to find bugs?