Abstract / Executive summary

This dissertation describes interaction around tangible media devices to provide an account of the sense making activity that couples artifacts and information exchange. This description aims to be general enough to evaluate, inform and guide the design of information devices that structure participation similarly to those examined. Tangible media, couples the richness of material objects to digital information, and as such is a strategy that has interested many designers in the last 15 years of rapid information technology development. However, market forces usually favor the lower production and distribution costs of software driven computational products that have no associated specific material interface. This conflict demands an explicit account of the value ‘tangibility’ and what material specificity contributes to computational interaction.

Method and Execution

This dissertation describes the behavior observed around and with a variety of Tangible Media (TM) interfaces. The main study is based in a museum exhibition called Experiments in the Future of Reading (XFR) which provided an informal learning context and naturalistic social setting in which to examine the interactions of diverse participants, both alone and in small groups. Direct observation (60 hours) and video analysis (40 hours of video data) of these interactions were used to produce qualitative descriptions of small groups (n) interacting, the spatial patterns of use, and the social and physical constraints of information exchanges between people in this context.

Qualitative results:

Information exchanges were not predicted by social roles— strangers interacted, discovery was often led by children, parents did not assume authority, and mastering, explaining and guiding others or watching, questioning or being guided were roles in which most participants took turns.

Action is information:

People rarely followed printed instructions yet consistent patterns of use emerged. The visible actions of others attendees played a more critical role in shaping the actions of those that followed, than the written instructions and diagrams. The actions and physical engagement provided a medium that coordinated attention and through which ideas were transmitted between visitors, as if an information conduit. The lower age ranges of the audience demonstrated a hands first exploration of the information world and uniformly across gender and age, the spatial modulation—stepping into and out of the vicinity of the exhibit--was used communicatively in the small groups gathered around each exhibit.

Shared social metronome:

Verbal and gestural exchanges between people were carried on at a remarkably consistent rate, although interpretations of the exhibits and the length of engagement varied widely. That is, although some people spent more or less time at different exhibits, the actual rate of exchanges was invariant—as if a metronome were running--refuting a common expectations that some interfaces provide "more" information, or could be described as more cognitively "efficient", or more effective. In contrast I observed people using the social and physical resource to allocate their attention and organize their time at exhibits.

Spatial relevance filter:

People in the immediate vicinity of the TM device exchanged information relevant to the device at hand. In this way, the hardware and its spatial dimension acted as an information filter, filtering for relevant comments inside various zones. The zones were characterized by activity and decisions people made therein: a distal zone in which people waited and decided to enter closer or to move on; a mid-zone in which more audible information was available, and the actions of the person using the exhibit were observable; an inner zone where the actual actions on the TM device took place and people exchanged commentary.

Quantitative Results:

To develop a feasible quantitative measure I used the variables that could be unambiguously counted and classified. Although I classified the roles people assumed and analyzed the content of their exchanges, the quantitative measure represents information exchanged, the rate of exchange, and where and when it is exchanged. For instance stepping into and out of particular zones and time spent therein, and group size, had less ambiguity than classification of an utterance. In this section I treated gestures, utterances, and actions (including proximity modulation) as packets of information the same way keystrokes are counted in traditional user studies, or click-throughs are counted on web-based interaction studies. The most salient variables those describing the group dynamics (groups size fluctuations) and the time people spent at the exhibits.

The primary finding of the quantitative analysis is that, holding the institutional context constant, the outcome most elastic to the exhibit (that varies with the exhibit) is not rate-of- exchange but group form, size and dynamics. Specifically, the important effect I show is an increase in the amount of time spent by a person at an exhibit for each additional person that joined the small group at the device—I call this the small group effect. That is, some exhibits allow information sharing with other people, and others do not. It made no difference (to the time a person expends) if there were one, or four people at the single channel video display (effectively a "control" exhibit with no interactivity). Likewise there was no small-group-effect on the Glyph-o-Scope exhibit which incorporated a microscope display preventing shared visual access by any proximal participants. In contrast, the Tilty Tables—table top displays controlled with highly visible actions of tilting the table allowing coordinated social attention—showed this small-group-effect most significantly. In other word’s a person at the Tilty Table, due in part to the physical design of the interface, was able to draw on the activity or commentary of nearby participants. Other people were as much information resource as the computational display.

Not just physics; not just social

The physical parameters of the exhibits themselves do not produce the effect—for instance, the size of the display had no systematic effect on the group size or dynamics. The largest display, the Time Line Wall, had one of the smallest average group sizes, while many people crowded around the small(est) screen of the Speeder Reader exhibit (yet highly visible control-actions). In addition, information per user per exhibit is not tightly related to an individual’s time expenditure per exhibit. Time spent (or "stickiness")--the traditional method for evaluating museum exhibits--does not differentiate between time spent waiting with no relevant information exchange (looking at the ceiling), or the time spent interpreting, sharing control or otherwise observably interacting.

I describe tangible devices in terms of their structures of participation (STROPs) that allow peripheral participation, facilitate shared control, and provide visibility and audibility of the actions and interactions. I show that open, tangible interfaces allow participation and maximize information exchange in small groups vis à vi other computational interfaces. Tangible media can be understood as effective in coupling social and technical resources, such that more people get more information, errors are corrected more quickly, and more information circulates around them because people allocate more time.

Summary of Contribution

To account for the effects of tangible media required changing the unit of analysis. The exchanges between people around the interfaces emerged as critical in characterizing the use of the devices rather than any of the particularities of individuals interacting with the devices. This challenges a longstanding assumption of Human Computer interaction literature that presumes the atomized "user" is the apt unit of analysis and representative of the important phenomena in interaction design. I have shown that consistencies apparent at the level of small group were not evident when observing individuals, yet dominate the information and usability experience of an individual.

Structures of participation (STROP) as a unit of analysis enables the analysis of the cumulative effects of interactivity, simultaneously accounting for the social dynamics and technical constraints. STROPS provide a socially grounded alternative to user-centric design approaches. STROPS are measurable and quantifiable and this provides a theoretical device to integrate over individual interactions, without obscuring the variability of individuals, yet account for the predictability of small groups behavior.

Intelligent Design:

Using structures of participation in designs can guide effective information exchange in and for small groups. Similarly STROPS can have design consequences for other technological interfaces and institutional contexts. If designers intend to design computational interfaces that can facilitate information exchanges and promote participation locally one can evaluate, develop and refine designs relative to the desired Structure of Participation.