The Pliant Story

Our human world is held together by semantically rich interactions. For example, people in conversation continually try to figure out what others mean, check to see if they are understood, correct misunderstandings, act in ways that depend on immediate context, past history and shared experiences, and so forth. Such interactions may last minutes, hours, days, years or even centuries; and they may be embodied in spoken conversation, written correspondence, works of art, buildings, or other media. Such semantically rich interaction, which we also call discourse, typically involves the gradual development of a common ground: a set of shared understandings about both the content of the interaction, and about the process of conducting it. We aim to get digital systems to support and participate in discourse, at least in limited ways.

Our Approach: Sociotechnical Analysis and Design

One of our basic assumptions is that the proper unit of analysis and design is neither the technology, nor the person, nor the group, but rather people in groups using technology: what we call a sociotechnical system. Another way of expressing our our goal is that we aim to understand how to design sociotechnical systems in which the boundary between the technical and the social is gradual, fluid, and dynamic.

Three Areas of Investigation

It's useful to divide our concerns into three areas: Human-Machine Interaction; Social Interaction; and Foundations.

1. Human-Machine Interaction: Action as Appropriation

Although today's digital systems do many things well, they don't know what their activities mean to their users, they have no coherent view of their past interactions, and they have no model of their own interaction process. Therefore, they cannot explain what they're doing, discuss alternatives with users, or adjust what they are doing based on their users' understanding or goals. In short, they cannot participate in rich interaction, because there is no common ground between the system and its users.

People rarely begin from a blank slate, following carefully thought out plans about how to achieve well-specified goals. Rather, human action is improvisational and opportunistic, and people's goals are often vague. People look around for useful stuff, and gradually reshape it to their own ends, specifying and redefining their goals in the process. We want to design computational systems that support this process, which we call appropriation. We are investigating two (entwined) ways of supporting appropriation: accounts and coproduction. Systems that provide accounts reveal, as an intrinsic part of their behavior, what they're doing, how they're doing it, and why they're doing it. This seems essential to making computational systems appropriable. Coproduction is process of the user and the system working together, each applying its competencies to ease the process of appropriation.

2. Social Interaction: From Data to Meaning

This perspective also applies at a higher level: supporting rich semantic interaction among people through the media provided by digital systems. Groups of people interacting over time (discourse communities) are very effective at storing, retrieving, filtering and distributing information, supporting learning and problem solving, and coordinating distributed activity. However, today digital systems provide little support for such activities; the current paradigm is that computers contain data, disembodied information that lacks any connection with the social context in which it was produced.

Our goal here is to help users participate in the web of discourse that connects community members, helping them understand the social context of a particular piece of information, interact with those who produced it and others who are using it, and continue the discourse by commenting on, adding to, or otherwise modifing it. Our investigations in this area include studying and analyzing examples of on-line interactive genres, exploring ways in which pattern languages might provide common grounds for such genres, and developing scenarios of how a discourse friendly systems might support social interaction.

3. Foundations: From Rigid to Pliant Media

Computing technology supports responsive media that are fundamentally different from the mostly passive physical media that have been used throughout history. Such media can respond to their context, history, and users. However, so far computer based responsive media are exceptionally rigid and fragile. A primary goal of our research to create computational media that are flexible, resilient and responsive--in a word, pliant. Specific areas of investigation here include design approaches based on ways that people express meanings through material artifacts, and virtual machines that directly support computing with richer semantics.