In the discussion on what players learn from digital games, there are two major camps in clear opposition to each other. As one side picks up on negative elements found in games the other side focuses on positive aspects. While the agendas differ, the basic arguments still depart from a shared logic: that engagement in game-related activities fosters the development of behaviors that are transferred to situations beyond the game itself. With an approach informed by ethnomethodology, in this paper we probe the underlying logic connected to studies that argue for such general effects of games. By focusing on proficient gamers involved in the core game activity of boss encounters in a massively multiplayer online game, we examine the fundamentals that must be learnt and mastered for succeeding in an ordinary collaborative gaming practice where aggression is portrayed. On the basis of our empirical analysis we then address the contentious links between concrete instances of play and generic effects. As expected, the results point to "aggression" as well as "collaboration" as major components in the gaming experience, but our analysis also suggests that the practices associated with these notions are locally tied to the game. Based on these results, we propose that to reverse this relationship and claim that game environments foster collaboration or aggression "in general" first assumes strong theoretical claims about the nature of cognition and learning, and second, risks confusing the debate with hyperbole.