Does a puppet show count as theater? My theory is that it does: the puppet constitutes a form of displaced embodiment. This extends to online RP – the avatar is a type of displaced embodiment. Because an agent is controlling these alternative bodies, the requirements for my definition of theater are met.
But what if the puppets are controlled by a machine? Is this still theater? Here things get squishy. If the audience is unaware that the puppets are machine-controlled, then as far as it’s concerned, the performance is theater. But if the spectator is told that the performance was mechanized, they almost inevitably reevaluate what they experienced. They might feel a sense of dismay, they might be thrilled by the ingenuity of the mechanics, or something else, but it’s highly unlikely they wouldn’t reevaluate the performance in some manner.
In terms of my definition of theater, the mechanized puppetshow wouldn’t be theater because there’s no possibility of intersubjectivity or a reflexive social relationship. Experientially, however, it’s not so clear. Some people might still want to call it theater.
One possible solution is to say that any definition of a cultural practice is going to have some fuzziness around the edges. Although we often would like definitions to be clear-cut, in reality they often aren’t. Take biology for example. From an evolutionary point of view, the answer to the “chicken and the egg” problem is straightforward: what came first was the egg because what laid it was, genetically speaking, not quite a chicken. A few genes were slightly different. That means there are some fuzzy borders around the genetic definition of chicken. There’s no reason why something is extraordinarily complex as a cultural practice should have a definition with absolutely borders.
Still, we know what distinguishes the egg from the not-quite-chicken, and it would be nice if we could identify the areas of fuzziness in definition of theater. It would seem from my comments above that whereas the ontology of theater is fairly strict, its phenomenology is not.
But Second Life raises further problems. Avatars’ movements are all controlled by programmed animations. One popular form of entertainment in SL is dance performance. In this genre, avatars perform choreographed movements, but all the person behind the avatar has done is set the animations running (typically, by “sitting” on a disk or other object containing the animations). I would wager that audiences generally respond to these performances much like they do to a live performance in RL. But the dancing could be performed equally well by a bot; in fact, arguably, the person behind the avatar has simply turned their avatar into a bot. There have been various performances that are quite explicitly by bots, in which the artist has left her computer on, running multiple instances of a program to keep the bots operating (I have in mind Jo Ellsmere’s work particularly; see for instance Mask, Biomechanical, and Obedience). In any case, not only has embodiment been displaced, but we have a question of how much performer agency actually remains when most or all of the performance has been automated.
Ultimately, then, the concept of intersubjectivity is probably going to be reconfigured if artificial intelligence produces computer behavior that passes the Turing test, or – as in the movie Her – meets the criteria of reflexiveness and intentionality that we associate with human agency.