Brenda Laurel responds (excerpt)
Mateas's new causal chains (as represented in his Figure 3.2) are based on a very interesting analysis and a good inversion. Speaking in Aristotelian terms, however, the player does not provide formal causality, but contributes efficient causality to the extent that she shares authorship of the plot, and also contributes materially to the plot by presenting thought and character (patterns of choice) that can influence the shape of the particular plot. Later, Mateas correctly implies that knowledge (or intuition) about the form can act as a constraint on the player's actions. Mateas's key contribution, in my view, is his novel and extremely useful observation that "in order to invoke a sense of agency, an interactive experience must strike a balance between the material and formal constraints."
I think it's brilliant to place "the mechanism of interaction" (as an affordance) at the level of spectacle, referring specifically to what the interface presents to the player as the possible modes of interaction. It is consistent with Aristotelian spectacle in that it must be available to the senses. We should add that a player may be enabled to impinge on the evolving plot at various levels, and that this may suggest some new criteria for judging the robustness of interactivity. An interface that enables language or symbolic communication enables a player to contribute material at the level of diction. How far-ranging the player's choices may be (from predetermined one-button responses to inventive solutions) can be seen as the range available at the level of thought. But thought counts for naught (rhyme intended) unless it can be expressed in action, and action doesn't count unless it is consequential at the level of plot. This means that enabling blathering, hand-wringing, and random smiting does not constitute robust interaction design, unless these behaviors change the course of the plot itself.