Sociable Objects

The Sociable Objects project aims to explore how clocks, plants, communications devices, toys, artworks and wearables might work together to enhance their utility to people. My thesis will investigate the ways in which objects can share their data, learn about each other and perform intelligent behaviors based upon each other's states. I call them “sociable objects.” Modern devices sense and generate information, but they tend to keep this information local. The information sharing potential between devices has not been fully exploited. Objects usually don't share a language. Most importantly, they aren’t asking each other the right questions. Sociable objects will behave differently. They will be well-informed, friendly and aware of their context through communication with each other. I’ll create my own sociable objects, and I’ll begin the process of networking and socializing objects created by others.

There is an ongoing debate in cognitive psychology about the sources of knowledge. Initially it was assumed that all the things we know about the world are learned, and though recent research has added a role for innate knowledge, direct learning is still vastly important. People see, hear, feel and smell things that teach them about the world. Furthermore, they react to those stimuli with behaviors, including some that directly affect the input stream. One thing that makes humans smart is that we integrate information from a wide variety of sources. In fact one of the largest sections of the brain (and one of the least well understood) is the parietal lobe’s sensory integration area. Here discrete information from the five senses is combined to create a comprehensive picture of the world that guides our reactions and behaviors. People with damage to the sensory integration area have tremendous difficulty comprehending the world and managing their behaviors. When diverse information is not available or combined, we have trouble coping.

Humans are also highly social. We rely upon information transmitted by others even when we cannot see with our own eyes. In fact, much of what we know about the world comes from what we are told by others. We know about the back of the moon, the bottom of the ocean and the interior of atoms, though nearly none of us have experienced these things first-hand. Intelligent behavior is rooted in the integration of the direct sensory input with remote sources that we at least implicitly trust.

Humans tend to anthropomorphize active objects, but these objects behave nothing like us. In fact our modern devices cannot serve even as a sketchbook model of the human cognitive and behavioral system. They lack integral structures. Devices tend to get all their information from one source. For example, a smoke detector simply evaluates a sensor module that looks for particulate matter in the air, often mistaking innocuous matter for toxic gas. Automatic flush toilets react to a single detector, which is sometimes fooled if the user leans forward. Even devices that take information from two or more internal sensor sources are unable to query the neighboring devices which hold pertinent information. Their behaviors are annoying and un-human because they lack access to the data that could help them act “smart.”

Modern artifacts seem poised on the brink of dramatically extended helpfulness. In the last century electrical power granted devices the power for autonomous motion, and clever mechanisms enabled those motions to follow simple ordered procedures. Washing machines knew to agitate before they spun dry. Toasters can (and still do) mechanically sense the transformation from bread to breakfast, then physically present us with the gloriously browned results. Of course, in the last 30 years, the computer revolution has eclipsed these mechanical computations with electrical ones that are both highly sophisticated and stunningly cheap. The average home contains 40-50 microcontrollers (Ganssle Group, ). These tiny chips bring the computer revolution to a tremendous variety of devices. There are 50 even in the cheapest automobile (Bauer, ). There are four in my wallet. These microcontrollers don’t quite bring sentience, but they certainly allow the incorporation of complex logical operations.

Despite the wide base of technological advances available to modern-day devices, most fall short of being anything like “smart.” Weiser’s seminal 1991 paper on ubiquitous computing (aka. ubicomp) describes a world in which there are “hundreds of computers per room.” (Weiser, ) However, in his 2005 book Everyware, Adam Greenfield notes, “As a culture, we have so far been unable to craft high-technology artifacts that embody an understanding of the subtlety and richness of everyday life." (Greenfield, 2006)[p3] What is going on? Weiser’s vision of technological advances has largely come true, but his descriptions of a seamless web of invisible computing that supports our every move has clearly failed to occur. Greenfield adds, “[despite] a coordinated suite of devices and user interfaces, sensor grids, software architecture, and ad hoc mesh network strategies...very few of the people working in ubicomp or its tributaries seem to have quite gotten how all these pieces would fit together.” Clearly this is the puzzle must be solved if devices are ever going to be smart (Greenfield, 2006)[p.16].

The Sociable Objects project will engage a lot of different groups. It will clearly affect the ITP community because they will be viewers, users and participants. Objects will be displayed on the floor. For devices with a utilitarian purpose, the ITP community will be the likely users. My goal is to involve as many makers as possible. Anyone with a project that obtains data and is willing to share it, or creates a behavior that they are willing to have remotely triggered, can be a bona fide research participant. Beyond the ITP community, I hope to influence researchers at other institutions to consider, test and extend my work. To this end I’ll participate in conferences and eventually submit the results of my research for post-graduate publication. In the interest of reaching the broadest possible audience, many of my Sociable Objects, and those of my fellow makers will be whimsical and entertaining, as well as technically creative, so that they are worth of inclusion in the popular press. While I’d like for everyone who has contact with the project to fully understand it, I’d be satisfied if most viewers understand the basics, with some inspired to consider and expand upon the deeper implications.

Botanicalls is one of my previous projects, created in partnership with Kati London, Rebecca Bray and Kate Hartman. It’s a system that allows plants to place phone calls for human help. When a plant on the Botanicalls network needs water or more light, it can call a person and ask for exactly what it needs. When people phone the plants, they orient callers to their habits and characteristics. While this may seem plenty sociable, it really isn’t in a Sociable Objects sense. The plants are contacting people, and the people are contacting the plants, but the system information is kept internal and never shared with any other devices. The plants will call for light even when it’s just been cloudy for a few days. They call day or night, without any regard to whether someone is awake or around to take the call. For this part of the project, I set out to change all that, and demonstrate how Sociable Objects can be better behaved objects.
Here’s the scenario, as it plays out in the Botanicalls installation set up in the ITP shared public areas:
Fiddle-leaf fig is low on light. Its light level for the last two days is only 30% of optimal and its Botanicalls event sensor is triggered. But is it time to make a phone call? As a Sociable Object device, its decisions don't need to rely on single data points. The fig begins making contact with the devices around it. First it checks with nearby Botanicalls plants. If they haven't gotten enough light either recently, the call for help will be postponed because it's probably been cloudy and that's not a reason to disturb humans with a useless phone call. However in this case the spider plant, cuban oregano and prayer plant all respond that their light levels have been close to optimal. So it's not cloudy. In fact the fiddle-leaf fig also makes contact with the light sensor in the solar panel on the roof, which is monitoring its battery charging levels, but sociable enough to share its data with whomever asks. Sure enough, there has been plenty of sun. Now it's appropriate to make a call. Is anyone there to answer? Fiddle-leaf fig contacts a real-time clock and discovers that it is 4 p.m. Someone is likely to be around. A quick check-in with some nearby art projects that happen to use motion detectors confirms this. People are moving around the space, and at 4 p.m. they're probably students who will answer the phone. A call is placed and answered by a student, who heeds the request for light and moves the fiddle-leaf fig out of its shady corner and into a sunny window.

The Sociable Objects Project has four major components:
1. The physical design and creation of the objects
2. The technical design of protocols, APIs and networking methodologies to support object interactions
3. Research into existing work regarding device-level communications.
4. Social networking to inform, influence and convince other makers to integrate sociable communications in their own creations. The end product will be a diverse menagerie of seemingly unrelated devices that can react to and interact with each other.

Moving forward I would like to continue building Sociable Objects. I think it’s very important to increase participation in the project. My fellow students have been very receptive to publishing the information latent in their creations for use by others. I believe that there is some critical level of available resources to be attained, at which point it will be easier and cheaper to utilize shared information than to create redundant data sources or ignore potentially useful behavioral requests. The Collaborative Mesh Networking class will make the fall of 2007 an important opportunity to jump-start this interaction. I’d also like to create project alliances with other classes, certainly with the New Interfaces for Musical Expression and Network Objects courses, but also with any other course that can use information and triggered behaviors to make a statement, tell a story or enable helpful actions.

Future research will include full mesh networking as a platform for sharing information interactively over larger distances and in complicated environments like outdoor urban spaces or between moving vehicles. There’s also a great opportunity in the Ubiquitous computing literature for research that generates findings about inter-object communications of any kind. Hopefully these ventures, and the growing participation of other Sociable Object enthusiasts will start making room in the world for devices that can share and language, ask each other the right questions and enhance their utility to people.