The way we interact with computers is unsophisticated. Our dialog with them is one of yeses and nos, degrees of up or down, left or right – simple directions, immediate answers, and cascading responses. Despite this, computational systems have allowed an explosion of incredibly complex human interactions. Social networks have become fountains of complexity, providing for emergent behaviours within and around our interactions with each other, our screens and ourselves. As these interactions mature and our understandings from their outputs grow, we will likewise be seeing an evolution in what it means to interact with the network versus the computer. Ambient Interaction is a shift from the one-to-one or one-to-many model of interaction that has become custom: we click a mouse and something happens on the screen. The feedback loop is tight, immediate, and provides instant gratification. Ambient Interaction reverses this model, providing a many-to-one relationship between human interactions with computational devices and subsequent system provided feedback. The feedback loop is not as tight, but because of its inputs, proves much more nuanced. The computational feedback to the user becomes an aggregate interpretation rather than a immediate response, and so provides a degree of scalability and emergent complexity that is not fully realized in a direct interaction scenario. In our day to day, ambient interaction is the subtle gesture, the shifting of weight, and the tone of voice which tell your friend that something is wrong. Within computation, it is a system's ability to understand its environment, instead of forcing its own environment on anyone who might approach it. The foundation of ambiance as a material in interaction design lies in a handful of core technological concepts: wirelessly networked devices, ubiquitous and cheap sensors, and the read/write environment. When discussing these challenges, I assume a simplicity and ease of implementation on a scale that is not quite there, but is just around the corner. On a do-it-yourself and research level, tools like the arduino, zigbee devices, and providers like Sparkfun and Adafruit give the accessibility for artists, designers, and coders to knock down conceptual barriers of what is and isn’t possible. However, the implementation of these designs at scale is lagging behind for lack of infrastructure and supply chain. The technological challenges are ones that are broadly speaking, already addressed. The issues ambient interactions face are more around the behavioural and cultural approaches to space and objects than anything else. Malcolm McCullough offers an argument on the intention of a place in relation to technologically augmented spaces: "Our very presence in one kind of space must serve as consent to take part in its technical environment, but in another space should indicate our desire for anonymity." (Digital Ground, 101) Designing for ambient interactions requires one of two things: an explicit declaration of intent on the part of the system, or an implicit understanding of the role of the place in question. A common example, our expectations around a space prompting us about unanswered work emails would be entirely different in our home office than in a cafe or bar. In determining the construction of ambient interactive systems, we need to develop a taxonomy of place, and the sub-taxonomies of the interactions appropriate therein.
[Warren Ellis, Matt "D'Israeli" Brooker & BERG / SVK / 2011] Differentiation becomes the challenge presented by the varied intent of these software systems. Ambient systems, by their very nature, are often passive in their absorption of information. Not interacting with these systems becomes the purview of the sousveiller, weaving inbetween sensor grids and the invisible cones of detection cast out by networked cameras. For the rest of us, we must then rely on either the overt display of identity to remove us from the context of a system through the policy of the system in question, or its shrouding so to make such data collected worthless. In the latter case, Warren Ellis and BERG London's SVK proposed such a device: wearable infrared emitter which shrouded the wearer's face in a static glow, imperceptible to humans but blinding to improperly filtered cameras. Ambiently interactive systems are an inevitability: built on ubiquitous, cheap sensors and networked devices, these systems have numerous applications of interest to marketers, home owners, retailers, and governments. The quantification of all that we, or the system, survey becomes the purview of those who design the system, their clients, and the requirements for any particular physical sensor to be present. Examples of these devices are everywhere in the commercial space, laying an initial groundwork for how broader implementations can be approached and understood by the public. One such example is the Fitbit. The Fitbit is a wireless accelerometer worn to monitor and record movement. The resolution of the signal from an physical standpoint is low: what the Fitbit understands is magnitude of movement in a particular direction. From an algorithmic and ambient perspective, this information is rich, relatively effortless, and nuanced. Types of movement: via bicycle, car, walking, or running, are all distinguishable. Quality of sleep, intensity of movement, and similar can all be tracked and then wirelessly transmitted to the dock. For under a hundred dollars, we have access to an early example of an ambiently interactive device with the purpose of behaviour modification through a more diffused feedback loop. The Fitbit's role and purpose is framed around fitness and self awareness: if you can guess at how many calories you've burned in a day through movement and activity, then your eating, sleep, and exercise habits are meant to change from the greater self-awareness around this information.
[Fitbit Personal Tracker / photo: bfishadow] The difficulty is that having a movement sensor attached to yourself twenty four hours a day, seven days a week isn't enough. It's not enough information to accurately paint the full image of human behaviour to the system in question, and consequently, insufficient for the goals that the Fitbit sets out to accomplish. For the Fitbit to be fully realized as an ambiently interactive system, it must become necessarily divorced from the wearable object it is identified with now. In so doing, the Fitbit is instead recognized as the software system within which a single Fitbit becomes only one input. Other objects, such as scales, galvanic skin sensors, home depth sensors, networked dumbbells with their own accelerometers, even gym memberships with open APIs connecting their own physical sensor networks and usage statistics; all these become components feeding data into the broader system that is the Fitbit. These tools must become ecosystems of both physical sensors and linked services in order to provide their true value and accomplish their goal: a complete quantitative picture of ones physical health, tracking change around day-to-day habits, behaviours, and decision making processes. Having considered the future steps of a current technology and product, it becomes necessary to imagine something more mature and fully integrated into an environment supportive and amenable to ambiently interactive devices. In the following paragraphs, we will take a look at two hypothetical applications: a learning tool and a mobile service. |
|||
