Paper Summary
Chalmers and MacColl challenge the assumption that seamless integration is a necessary design goal in ubiquitous computing.
Chalmers and MacColl challenge the assumption that seamless integration is a necessary design goal in ubiquitous computing. Drawing on Weiser’s own critique, they argue that seamlessness risks reducing systems to a lowest common denominator, sacrificing the distinct character of each component for bland compatibility.
Their alternative concept, seamfulness, treats system “seams” — sensor uncertainty, positional inaccuracy, representational limits — as potential resources rather than failures to be engineered away. They illustrate this through two systems: a mixed-reality museum co-visiting platform where ultrasonic positioning errors shaped how users coordinated, and a mixed-reality chase game (CYSMN) where runners learned to exploit GPS inaccuracy as a tactical advantage.
- Three visitors experienced the same exhibition through different media: one physically present with a PDA, one in VR, one via web browser. The PDA visitor’s position was tracked by an ultrasonic system with meaningful error (95% accuracy only within 1.83m). Rather than hiding this, the authors argue the system should have represented the PDA visitor as a spatial extent or probability distribution rather than a precise point — making the uncertainty visible and honest. Users in trials compensated spontaneously: they used talk to establish shared context when the map failed them, and a VR visitor developed an exaggerated “wiggle” gesture because the map scale made small movements invisible. The seam (positional uncertainty) shaped, and in some cases enriched, how people coordinated.
- Runners with GPS-tracked handhelds chased online players through city streets. GPS error ranged from 4m to 106m depending on urban geometry — open spaces had low error, narrow streets had high error. Online players ignored this, but runners, over two days of play, developed an explicit tactical awareness of it. By the second day they were deliberately herding players into open areas where GPS updates were fast and reliable, then closing in. The seam (GPS inaccuracy mapped onto urban space) became a strategic resource that runners appropriated through experience, substantially changing the game’s competitive balance.
The paper’s most theoretically interesting move is the concept of design for appropriation. Users routinely adapt their behavior to accommodate system characteristics in ways designers never intended, and the authors argue this should be anticipated and supported rather than suppressed. They propose that making system mechanisms “literally visible, effectively invisible” — accessible when users need them, ambient otherwise — enables this kind of flexible, emergent use.
Relevance
This is another radical design theory that flips common design goals into potential avenues for greater user engagement and pleasant surprises. It also encourages design to be honest with its shortcomings, and instead use the “seams” to its advantage, anticipating surprise in actual usage. For Machines That Let You Leave, seamful design suggests that exit-oriented AI should not hide dependence, uncertainty, friction, or loss of capability behind a smooth assistant interface. Those seams should become available to the user as resources: visible traces of when the system is substituting for judgment, where personalization is narrowing action, when automation is becoming habitual, and what offline capacities would need to be rebuilt in order to leave. In this sense, a machine that lets you leave may need to be deliberately less seamless over time, turning its own limits and the user’s reliance on it into interpretable material for appropriation, reflection, and eventual disengagement.