This article reports on an unconventional collaborative event called Real-Time Research, a project that brought 25 participants together from radically divergent fields for a playful and somewhat improvisational investigation of what it means to do games and learning research. RealTime Research took the form of a two-part workshop session at the 'Games, Learning & Society' (GLS) conference during the summer of 2008, in which attendees collaboratively designed and then conducted five research experiments that took place over the course of the two-day conference. The article reports on the results of those studies and the processes that generated them.Few models exist for the successful collaboration among research academics, let alone among research academics and game designers and practicing teachers and students. Traditional conferences are often unable to provide much time and space for genuinely substantive conversation within disciplines, let alone across disciplines. There is the occasional follow-up conversation after a particularly good PowerPoint presentation with the obligatory exchange of business cards (that rarely survive the trip home), perfunctory post-session discussions, and the chance run-in at the hotel bar, but beyond these mostly accidental exchanges, we wanted to create a context designed explicitly for productive collaboration. The Games Learning and Society Conference (GLS), held annually in Madison, Wisconsin, was the perfect occasion for such an experiment. GLS already straddles several domains: academics, industry designers, policy makers, practicing teachers, and gamers themselves. At GLS, conference designers place a premium on creating opportunities for different constituencies to congregate and share concerns through a number of design elements. This mix of participants can lead to entertaining and informative sessions; in one example from the first GLS, a gaming consultant suggested that games could one day replace teachers, and a bevy of objections shot up: Designers said it was too difficult to replace teacher practices with artificial intelligence; educational theorists raised the historical failure of such efforts, and teachers laughed and said, 'good luck with that'.These discussions are a good start, but we often learn best when we are engaged in meaningful activity working side by side with others (Vygotsky, 1978;Lave, 1988;Lave & Wenger, 1991). We need opportunities not just for cross-disciplinary conversation but cross-disciplinary collaboration. This kind of collaborative activity enables us to see how others view problems, to test our understandings against others, and to talk through key issues in a context grounded in specifics (rather than mired in the general). If it is true that we get to know each other and our world-views by actually engaging in activities with each other (Steinkuehler, 2004), then why not create a context within a conference in which this can happen? We hoped that such exercises, no matter how small, might seed future work not just among those who...