During the course of the discussion several difficult problems were mentioned that we would not be able to solve, but were worth highlighting. These include scaling high touch experiences, the role of authorial voice, and ethical issues,
Mediation of player experiences ultimately leans towards creating psychological situations where players choose of their own free will to act in a desired manner. In ARGs and narrative spaces, the most powerful tool available is the puppet master, a human actor who guides and adjusts the experience to the player's current needs. Many years of experience have taught game designers that players will always react mostly strongly to the prods, suggestion and presence of other human beings.
Unfortunately, finding skilled people to fill these mediation roles is time consuming and often futile. It is a new role that most practitioners stumble upon by accident, not one supported by schools or trade organizations. Once you do find a competent game master, you run into problems of scale. A single game master can handle a few hundred players in an emotionally meaningful manner. With games hoping to reach millions of players, all those game masters start costing you big bucks. High touch experiences, mediated by live human beings, are exceedingly costly.
Luckily many puppet master
activities are relatively mundane, such as giving new quests or acknowledging
the completion of real world activities. Also, they are time limited
to a few minutes or less of actual contact with the players. This has
lead several participants in the group to attempt to create high touch
experiences by building emotionally engaging, highly social NPC characters.
General purpose AI is an unsolved problem, but perhaps if we limit the
scope of what we are attempting, we can create an effective helper.
We all need a little AI Jesus to help us make the right decisions
The goal is to have each game master control a herd of AIs that handle most moment to moment interaction with the player. Based on the user responses, computer simulations can be run of the user's mood, and the AIs can adapt their responses accordingly. Extreme situations can be brought to the Game Master's personal attention, at which point the GM can seamlessly take over. The entire time, the person on the other end operates under the assumption that they've been talking to a real human being.
Attempts thus far build on the psychological models of the player we mentioned earlier. Designers build proactive NPCs that poke and prod the player in ways the AI models suggest will result in desired behavior. Current research in this area includes:
With such tools, the hope is that a game masters can use its socially-aware NPC army to support thousands of players in a high touch, emotionally meaningful manner.