---------------------------------------------------------------------- Cooperation through Narrow Interfaces - Laser Theater Written by Ray Jones ---------------------------------------------------------------------- Consider the power of a group of ants. Though they have limited communication abilities, they are able to cooperate to achieve complex goals, such as the discovery and collection of food, caring for young, and defending the ant hill. Narrow communication channels mean that some effort is wasted, but still, the work gets done. Humans, with much wider communication channels, are able to cooperate more efficiently. It is difficult for ten ants to cooperate to lift with ten times their individual strength, but ten humans can synchronize their efforts and come close to applying ten times their original ability. Such shared effort is likely to involve preplanning ("You lift there.") and verbal cues ("One, two, three, lift!"). In some situations, such high-level organization is not possible, and we must rely instead on more indirect communication. For example, SIGGRAPH's Electronic Theater has explored cooperation with single-bit controllers, the well-known paddles. Given information about the state of some shared world, such as a game of Pong, cooperation between large groups of individuals emerges. However, each individual is only aware of their own effort in the grand scheme and does not have information about what other members of their team are doing. Thus, although the effort is cooperative, the cooperation is of the simplest form, that of a shared goal. We propose to widen the interface, allowing a broader class of tasks to be undertaken by large groups, and a wider range of cooperation to emerge. Each participant is given two continuous degrees of freedom, and everyone is aware of the current efforts of other participants. There is a simple, inexpensive method for delivering such freedom to multiple users: the laser pointer. By shining their pointers on a shared screen, participants communicate with the computer system controlling a simulated world. Simultaneously, they provide information to each other about what their current contribution to the overall effort is. Changing the interface from on-off to left-right-up-down makes several new classes of tasks available, selection from a large number of options on the screen, for example, or arranging points to form a particular shape. The immediate feedback of what other members on the team are doing helps achieve these goals; it is hard to imagine cooperating to fill a shape without knowing where the gaps are. There are several possibilities for cooperative tasks and games that are possible. A short list: - Shape drawing - trying to achieve some objective metric - competing between two or more teams - Concentration, the memory game - Moving ball games, where the pointers exert force on a ball - Soccer (competitive) - Basketball (cooperative) - Experiments in creating individuality out of a sea of anonymity - "One person in box A, everyone else in box B" - Choice games, such as the Monty Hall Problem - Asteroids (and other "laser" games) There are many more possibilities; nearly everyone that we explain the basic concept to comes up with a new activity or game. The most interesting activities might be the ones that are the most undirected, and therefore the most reliant on emergent phenomena. In the ETs where paddles have been used, several spontaneous games usually crop up before the "actual" show starts. We expect the same sort of emergent play to occur, and would like to find shared worlds that facilitate play without imposing structure. For example, we could simulate the flow of water over a surface whose shape is controlled by the pointers, or put multicolored blocks on the screen and allow pointers to manipulate them (blocks are a favorite toy for unstructured play, after all). Implementation -------------- The interface is made up of four elements: - A large display screen - A digital camera - A computer taking input from the digital camera and controlling the images displayed on the screen - Some large number of participants, each with a laser pointer The digital camera is aimed at the screen, and registered to it to minimize distortion between a pixel location as seen by the camera versus projected by the computer. In realtime, the computer processes the camera image to extract the pointer locations and modifies the state of the shared world based on this input. Most of the tasks we have considered do not require knowledge of the location of individual pointers, but can operate entirely on a density image, lowering the computational cost of the update algorithm. A tuned filter can be used in front of the camera to separate out the light from the pointers, further lowering computational cost. Open Issues ----------- Safety - The pointers we would use are of limited power, less than 5 milliwatts. These do not present a risk of accidental injury, due to normal aversion responses. Like any light source, if someone stares directly into the pointer for a prolonged time, overriding their blink response, damage could occur. Politeness - In any large group, it is possible that some individuals will have a poor sense of propriety, and may use a pointer in situations where it is unacceptable, such as during the rest of ET or a papers session. It is our opinion that the SIGGRAPH community is for the most part very mature and well-behaved, as well as self-policing. In one 1999 ET showing, several individuals were playing with pointers before the show. When the title of the first piece appeared on the side screens, one person left their pointer on, quite possibly inadvertently. Boos started up immediately, and the pointer was shut off almost immediately. If non-ET abuse is a significant concern, collecting the pointers as viewers exit is an option. Funding - Laser pointers cost between $1 and $2 in large quantities. If necessary, we feel that it would be very easy to convince vendors to sponsor this activity by paying for large numbers of pointers in exchange for a small sticker on the pointer and some acknowledgement during the program. Workability - As noted above, we are considering activities that do not involve expensive computations. We are implementing a prototype to verify the feasibility of the system. Registration and removing the distortion from the camera's input will probably be the most expensive operation, but fast algorithms for performing this mapping are well-known.