Fugitive utilises real time analysis and interpretation of infra-red video images of the interaction space as an unintrusive interaction technology. The goal of the interaction design is to allow unencumbered, intuitive, bodily interaction. The user is not required to handle devices, don special hardware or learn an abstract symbolic language in order to interact. Rather the user interacts with the system by normal bodily behavior, the same kind of bodily behavior they might employ on the street, the dancefloor or the sportsfield. This conception of interaction is in stark contrast to the mainstream of computer interfaces, which still require a fixed sitting body posture, and the learning of a highly contrived set of finger movements in order to enter data. Note that interaction is, essentially conceived as ‘data entry’. Any aspects of human experience which cannot be reconformed as ‘entering alphanumeric data’ is incompatible with the machine. Thus vast aspects of human intelligence and the diverse ways that intelligence is communicated in human interaction are excluded. The interface of Fugitive says ‘it doesn’t have to be that way’.
First order interactivity employs the analogy of switches, whether those switches are physical switches, buttons and hotspots on a screen, or sensors with binary output. Even with relatively sophisticated arrays of such sensors, one can only know ‘user is here’ and by extrapolation, ‘user was there’. The user may have hopped, or crawled there, the system remains oblivious. The goal of the Fugitive system is to discern more subtle, rich and informative aspects of the users behavior, by capturing the temporal dynamic of their ongoing movements. Tongue-in-cheek, this aspect of the system was originally called the ‘mood analysis engine’. It is the dynamics of their ongoing behavior, rather than their instantaneous position, which drives Fugitive. This is done by calculating acceleration and velocity of the user, along with other qualites, over various timeframes.
When Fugitive was first developed, in 1996-7, machine vision was regarded as a hard problem in computer science. The notion of an artwork which utilised analysis of realtime video as a source for sensing and interpretation of behavior was rare, though pioneering work had been done by Myron Kreuger and by David Rokeby. Now, vision systems with some of the capabilities of the original Fugitive system are available commercially for the domestic market.
Frantic pursuit of the image will elicit similarly frantic behavior on the part of the system. A more considered interaction will reward the user with calm behavior.
The images and their position are the expressions of the system, the user understands the system by the dynamics of the image, not by its content. The user is invited to look, not at the image, but through the image, to discern the behavior of the system.
as a cultural artifact, fugitive responds to the condition of consumption of pre-interactive cultural artifacts. Fugitive actively dissuades ‘passive contemplation of a fetishised object’. Fugitive rewards movement, active engagement. The image evades capture, both because it physically changes location and because the image content changes in response to the users behavior. The user who enacts passive contemplation is unrewarded. The users who pursues closure is frustrated.
Fugitive, while screenal, is emphatically not cinema. Like all interactive media, in Fugitive there is no pregiven narrative. Rather a unique experience unfolds for the user as a result of her interaction with the system.
If the user moves circumferentially, the scene that is triggered is a pan. As long as she circles, the image also circles, unfolding successive frames of the pan in successive positions around the wall. If the user moves radially, the shot triggered is a zoom, corresponding to the position in the pan. Fugitive, in a sense, ‘undoes’ cinema, since the image is aligned, (relatively) to the original position of the camera. As the user moves toward the image, the image zooms. The system can be understood as a kinesthetic video editor. Each user makes a different movie, depending on her behavior.
The scenes are chosen specifically to be unremarkable. Yet the sensuousness, almost eroticism of the zoom to telephoto, is undeniable. Fugitive exists in this conflicted relationship with the image, acknowledging its emphemerality, yet revelling in its capture.
While the tradition of the narrow perspectival window was triumphant in C20th photography, cinema and video, its unrequited other has been the tradition of the panorama. This was the case from the outset, Daguerre made his fortune constructing dioramas, large immersive visual spectacles. The panoramic drive has resurfaced in various panoramic cinematic specacles such as omnimax, in digital tools such as ‘quicktime VR’ and in the work of digital media artists such as Michael Naimark and Jeffrey Shaw. But fugitive is only quasi-panoramic. Kinesthetically driven, the image remains a perspectival window.
The premise of virtual worlds is that the user is able to navigate a stable, objective, albeit virtual location. The environment has architectonic stability, the user is a wandering subjectivity. Fugitive can present the suggestion of a navigable virtual world, yet that illusion is broken by the ongoing dynamics of the user. The central continuity of conventional virtual worlds is the stability of the virtual architeture. In Fugitive, the central continuity is that of the users’ embodied temporality.
As scientised westerners, we are conditioned to believe in the permanence of a world which is ‘out there’, at arms length, so to speak. This is the triumph of objectivism, a principle on which the scientific method is premised. Yet that principle, like the Cartesian concept of disembodied mind, is, at root, an assumption unsupported by experience. Critiques of this objectivism, such as those coming from phenomenological philosophy, contend that our everyday experience is of temporal and spatial immersion in a world which surrounds us, a world which is constituted by our moment to moment action and perception. Approaching lived experience form this vantage point, there is no objective world ‘out there’.
Vision is fugitive. While we naively believe that our vision presents to us a detailed and accurate representation of the world, that is illusory. We have large blind spots in each eye, which we are are oblivious to. Only a tiny sector of our visual field is in focus at any one time. Our eye moves rapidly to pass that sector over different parts of our visual field. While the eye is in motion, we are effectively blind, otherwise we would see motion blur. We only have color vision in the central part of the visual field. The conception of a stable, continuous visual field, confirming the existence of an objective world, does not reflect immediate visual experience. In much the same way, it is the neurological quirk of persistence of vision which allows for the perception of smooth motion, rather than sequential still, in cinema.
In the visual arts, the notion of ‘intervention’ has become well established: the positioning of an ‘artwork’ as an activist gesture. In media art the notion of Tactical Media is undertood in a similar way. Fugitive is, a heart, an intervention into certain prevailing attitudes regarding embodiment and interaction within the discipline of computer science and more particularly within AI. It is a demonstration, a proof of concept. While it is not particularly comon for artworks to function in this way, as interdisciplinary interventions in related academic fileds, it is not unknown.
The intervention of fugitive is to present a mode of interaction which is predicated on the system interpreting a person engaged in normal human bodily behaviors. This is in stark contrast to the conventional notion of interface, in which ideas and concerns must be encoded, usually as alphanumeric data, demanding the sequential pressing of little buttons on a board. This uterly impoverished interface functions, in fact, as a filter, excluding all rich and diverse aspects of human intelligence which cannot be encoded alphanumerically.
The term Deixis was adopted by Agre and Chapman to describe a mode of computer programming for artificial intelligence in which they developed which did not require totalising universal knowledge of the ‘world’ by the system. The Deictic approach is rooted in the commonsense recognition that, for instance, when I sit at a restaurant table, the specific identity of the fork at my place is irrelevant, any fork will do in the social function of dining. A deictic system does not, therefore confuse a ‘subjective’ perception (by the agent) of the world, with a ‘gods eye view’ of the resaercher external to the system, andall knowing. While it is not programmed according to deictic principles, Fugitive is sympathetic with a deictic approach, as it wraps a world around a user, and also contest the requirement of objective and global knowledge.