|
[width=120mm]figs/netrek-robots-bw |
Netrek is a game of conquest with a Star Trek motif. The game is normally played by up to 16 players organized into two teams. A team wins by transporting friendly armies to each of the opposing team's planets. Figure 3.1 illustrates some of the basic elements of the game: planets, ships, and armies. The first benefit of Netrek as a test-bed for user interfaces is that it is a game: so it provides built in metrics for the success of a new interface design. If the interface allows a player to play the game more effectively (to win more), then the interface can be said to be successful with little room for argument.
Netrek is very much a team-oriented game. Winning requires a team that works together as a unit. This fact , in particular, provides a rich set of interface opportunities ranging from low-level tactics to high-level strategy. There has been some work on new tactical interfaces, but these interfaces were blocked by the Netrek community with an authentication system to keep games fair. We will concentrate on the opportunities for building interfaces for high-level communication regarding strategy since these provide the most room for novel interface design.
Netrek is usually played by groups of players on wide area networks spread over large geographic areas. The standard interface requires the user to directly control their ship. Communication with teammates is accomplished by type-written messages: this means that in order to send messages the user must temporarily give up control of their ship. So players must communicate about strategy and complex maneuvers in an efficient manner. This necessity led to the creation of a set of staccato jargon and an associated set of what we'll call programs. The programs are essentially little plays with a few roles that can be filled by appropriate players.
The existence of these programs is good for research in several ways. First there is pre-existing code, called a robot, that is capable of running a small set of these programs as well as game basics like tactics and navigation. These robots provide a solid base on which to build a research system. The jargon also represents a strict codification (enforced by the difficulty of communication) of human play that might indicate that recognition of plays and machine learning of plays through observation of human players would be tractable. This codification also means that there may be opportunities for novel, expressive interfaces to encourage the formation of new strategies.
One last aspect of netrek is the virtual embodiment of the robots in the game. The ships obey a simple dynamic model and they have limited control. This is particularly interesting given the proposal to represent behaviors as control signals to dynamic systems. This creates a satisfying duality between the mode of expression and the target of that expression.