It's the Player Types graph, but an early version. I have the player suits, but the labels on the y-axis are different: active/passive rather than active/interactive. The latter didn't come about until the peer review of the 1996 paper.
I gave the talk to the Cambridge University Computer Society in November 1992. I know the date because the town centre was crowded with gaggles of teenage girls attending a Take That concert at the Corn Exchange, and a quick trip to Google told me the date. By this time, I was using the Player Types model in my own analysis of virtual worlds, but knew I had more to do. I almost didn't try to get the original paper published as I was aware of its inadequacies (which were addressed by the 8-type model).
Some things never change, though:
No, we didn't call the MUGs then either, that was rather the point. Oh, and the "one chap in Japan" was Joi Ito, now head of MIT's Media Laboratory.