The vision dissolved into everyday life, and what arrived was not calm

Ubiquitous computing made two predictions at once. Computing would spread into the fabric of daily life. The spread would be calming, because technology at the periphery of attention informs without overburdening (Weiser, 1991; Weiser and Brown, 1997). The first prediction held almost literally. The second did not hold. The field gave initiative to its systems, and the industry that carried the devices to scale made attention its product. This first of ten units argues that the devices arrived while the calm did not, and that the field's own retrospectives say so.

The vocabulary it settles is the one the nine units that follow inherit. A team that believes calm follows from ubiquity buys more sensing. A team that treats attention as the scarce resource decides when a system may speak, and accepts that a competent system usually stays quiet.

The vision was a claim about attention, not about hardware

Weiser's target was not portability. The most profound technologies, he wrote, are those that disappear. He took writing as his model, because its products surround a literate person without demanding attention (Weiser, 1991). A personal computer failed the test. It was not immovable; like a single important book, it kept attention fixed on one box. His alternative was embodied virtuality, which drew computing into the world at three scales. He named them tabs, pads and boards, inch-scale, foot-scale and yard-scale. Location let a system act on where it was without artificial intelligence. Scale let it be present without competing for one screen (Weiser, 1991).

The attention claim was made explicit later. Calm technology engages both the centre and the periphery of attention and moves between the two. The periphery is what we are attuned to without attending to explicitly (Weiser and Brown, 1997). A technology is calming when it informs at the periphery without overburdening, and can still be brought to the centre on demand.

The spread arrived, and it arrived as infrastructure

The device prediction held. By 2001 Satyanarayanan could describe pervasive computing as an established field with a defined agenda. Its four named challenges were already about deployment rather than invention (Satyanarayanan, 2001). The clearest early evidence is a field deployment rather than a laboratory prototype. Mainwaring and colleagues instrumented seabird nesting habitat on an island off Maine with a network of 32 nodes streaming live data to the web. Their contribution was as much about retasking and node health as about sensing (Mainwaring et al., 2002). Ubiquitous sensing arrived as a maintenance problem.

Abowd and Mynatt mapped a decade of work. They added a fourth theme to the three they found. They called it everyday computing, defined by scaling interaction with respect to time (Abowd and Mynatt, 2000). Continuous interaction turns a tool into a companion, and a companion must handle interruption and resumption. Interruption became a core problem precisely because presence did not remove the demand for attention.

The goal was quietly redefined

One reason the vision looks unfulfilled is that the field narrowed it in the act of pursuing it. Satyanarayanan took Weiser's ideal to be complete disappearance from a user's consciousness. He then set a different target: in practice, a reasonable approximation is minimal user distraction (Satyanarayanan, 2001). He conceded that complete invisibility may be impossible while reduced variability is within reach. The field's most cited engineering statement therefore replaced calm with something weaker, the absence of avoidable interruption. Suppose the goal is minimal distraction. Then a system that interrupts only when it judges a message important has met it, and the user's total attentional load leaves the specification.

The same paper shows what minimal distraction looks like in a built system. Satyanarayanan's worked scenario follows a traveller, Jane, waiting at Gate 23 of Pittsburgh airport, whose wireless link is too thin to send her documents. Aura reads the bandwidth. It consults the airport's network weather and flight schedules. It proposes that she walk three minutes to Gate 15, where the link is good and no flights are due. She moves, sends the mail there, and walks back before boarding (Satyanarayanan, 2001). The scenario assumes a link that is thin rather than absent and a passenger who can read the advice in English, not one who cannot walk. Aura was a research initiative at Carnegie Mellon rather than a product, and the scenario is hypothetical. It therefore settles nothing about how often such advice helps. What it does show is the mechanism the field chose. The system watches the situation and speaks first. The user's saving is a shorter delay rather than a lighter load of interruption.

The initiative went to the system

Rogers supplied the diagnosis this unit builds on. Three programmes dominated the decade after Weiser: context awareness, ambient intelligence, and monitoring and tracking. Their accomplishments did not match his world (Rogers, 2006). Her reason was initiative: the field tried to make computers act for people, and proactive computing is the goal it fell short on. Her reversal was proactive people rather than proactive computers, with technology that provokes learning rather than quietly removing decisions.

Read against the vision, the point is precise. Calm technology keeps a person in the periphery of a system's action. A system that acts on inferred context takes decisions on the person's behalf. The more it does, the more often it must check whether it acted rightly, and every check is a demand for attention. Proactivity and calm pull in opposite directions, and the field chose proactivity.

Commerce chose the other half

The devices reached scale through a market, and the market did not reward calm. Greenfield argued that the likely outcome was not a benign embedding but an everywhere-computing shaped by commerce and default under weak public oversight. He argued that the choices deciding which version arrived were largely industry choices (Greenfield, 2006). His book asks who gets to determine the shape of everyware. It answers, in effect, that it is not the people who live in it.

The field's own historians reached a compatible conclusion. Bell and Dourish observed that ubicomp is unusual in being defined by a shared vision rather than technical problems. By the mid-2000s the field already inhabited the future its pioneers had imagined, without that future working out as predicted (Bell and Dourish, 2007). Dourish and Bell then distinguished the mythology that animates the field from the mess of practical reality. A vision defined by attention was carried to scale by an industry whose returns came from holding attention. The interruption problem Abowd and Mynatt identified in 2000 was still the field's central problem when the reviews below appeared.

The contested question: did ubicomp fail, and against which test?

Three positions are held, and they disagree about what would count as an answer. The first holds that ubiquitous computing failed on its own terms. Rogers (2006) is the clearest statement: the dominant programmes did not produce Weiser's world, and the error was to give initiative to systems. The test is whether the field produced calm.

The second holds that the question is malformed. Bell and Dourish (2007) and Dourish and Bell (2011) argue that ubicomp is held together by a myth rather than a specification. A myth cannot be falsified as a prediction can. We already live inside the imagined future in messier form, and the productive question is what people do with the technology.

The third holds that the goal was always approximate. Satyanarayanan (2001) treats complete invisibility as an ideal and minimal distraction as the target. Devices that occasionally demand attention are therefore the expected shape of success. Chalmers (2003) sharpens this: Weiser himself called smoothness misleading and argued for systems with visible seams. To measure ubicomp against connection that never breaks is to measure it against something its author did not ask for.

The evidence that would settle the dispute is an instrument the field has not built. One would measure attention over time in deployed settings against a stated baseline. Chalmers (2003) shows seams are part of the world, and Satyanarayanan (2001) shows uniform coverage was never expected, so a single broken connection proves nothing. Ntoa (2024) shows the gap directly. A systematic review of 46 usability and user experience frameworks for intelligent environments found most remained conceptual. It found 21 specified no metrics, and only three moved to concrete metrics and methods. A field that cannot measure an environment that does not hold attention cannot say whether its founding promise was met.

Rated by kind, the sides carry different evidence. Rogers (2006) argues from a decade of programme review rather than from a measured outcome. Bell and Dourish (2007) and Dourish and Bell (2011) argue from the field's own rhetoric, so their test is interpretive. The approximate-target reading rests on one design statement (Satyanarayanan, 2001) and on Chalmers's (2003) report of Weiser's own words. Only Ntoa (2024) surveys instruments, and it measures the field's tools rather than its effects. While the dispute is open, a designer sets an explicit attention budget for a product and records what it spends. The strongest objection is that no measure of calm has been built, so the unit's own test cannot be run.

Where this is heading

The successor programme keeps the vision's ambition and drops its calm. Pavlov and colleagues published the first systematic review of routine computing, synthesising 203 studies up to August 2025. They classify those studies by the temporal granularity, behavioural and social context and cognitive factors they sense (Pavlov et al., 2026). Routine computing senses daily life across days and weeks, exactly the scale at which the everyday-computing theme said interruption and resumption would matter. The review names the field's open problems as the gap between activity recognition and intent, the tension between personalisation and generalisation, and unresolved privacy concerns. Ntoa (2024) argues alongside it that evaluation must move into the environment the system occupies. The direction is a more honest accounting of attention.

Sources

• Weiser, M. (1991). The Computer for the 21st Century. Scientific American. doi:10.1038/scientificamerican0991-94 Argued that the most profound technologies disappear into everyday life and proposed embodied virtuality at three scales, tabs, pads and boards, with location and scale decisive. Its claim is about attention, not hardware.
• Weiser, M. and Brown, J. S. (1997). The Coming Age of Calm Technology. In Beyond Calculation. doi:10.1007/978-1-4612-0685-9_6 Defined calm technology as engaging the centre and the periphery of attention and moving between them, with the periphery informing without overburdening. It states the vision as a testable claim about attention that later work failed to measure.
• Satyanarayanan, M. (2001). Pervasive computing: vision and challenges. IEEE Wireless Communications. doi:10.1109/98.943998 Set out an engineering agenda with four thrusts, smart spaces, invisibility, localised scalability and masking uneven conditioning, and replaced Weiser's complete disappearance with minimal user distraction. The field weakened its own goal in the act of pursuing it.
• Mainwaring, A., Culler, D., Polastre, J., Szewczyk, R. and Anderson, J. (2002). Wireless sensor networks for habitat monitoring. In Proceedings of the First ACM International Workshop on Wireless Sensor Networks and Applications. doi:10.1145/570738.570751 Deployed 32 nodes on an island off Maine to monitor seabird nesting habitat, streaming live data to the web, and reported the sampling, retasking and node health problems that followed. Ubiquitous sensing arrives as a maintenance problem, not a laboratory result.
• Abowd, G. D. and Mynatt, E. D. (2000). Charting past, present, and future research in ubiquitous computing. ACM Transactions on Computer-Human Interaction. doi:10.1145/344949.344988 Reviewed a decade of ubicomp research and added everyday computing to natural interfaces, context-aware applications and capture and access, defining it by scaling interaction with respect to time. A companion must handle interruption and resumption, so presence does not remove the demand for attention.
• Chalmers, M. (2003). Seamful Design and Ubicomp Infrastructure. In Proceedings of the Ubicomp 2003 Workshop on Human-Computer Interaction with Ubiquitous and Pervasive Systems. https://www.dcs.gla.ac.uk/~matthew/papers/ubicomp2003HCISystems.pdf Argued that seams such as coverage gaps and variable positioning accuracy are resources for design rather than problems to hide. He records that Weiser himself called smoothness misleading, so the vision was never a demand for connection that never breaks.
• Rogers, Y. (2006). Moving on from Weiser's Vision of Calm Computing: Engaging UbiComp Experiences. In UbiComp 2006: Ubiquitous Computing. doi:10.1007/11853565_24 Argued that context awareness, ambient intelligence and monitoring and tracking had been fruitful but had not produced Weiser's world, and located the failure in proactive computing. Her reversal is proactive people and engaged living.
• Greenfield, A. (2006). Everyware: The Dawning Age of Ubiquitous Computing. New Riders. https://www.informit.com/store/everyware-the-dawning-age-of-ubiquitous-computing-9780321384010 Argued that ubiquitous computing arrived as an everywhere-computing shaped by commerce and default under weak public oversight, and asked who gets to determine its shape. It explains why the spread of devices did not by itself deliver calm.
• Bell, G. and Dourish, P. (2007). Yesterday's tomorrows: notes on ubiquitous computing's dominant vision. Personal and Ubiquitous Computing. doi:10.1007/s00779-006-0071-x Argued that ubicomp is held together by a shared vision rather than by technical problems, and that the field already inhabited the imagined future without it working out as predicted. Their ubicomp of the present is the strongest statement that the vision is a genre.
• Dourish, P. and Bell, G. (2011). Divining a Digital Future: Mess and Mythology in Ubiquitous Computing. MIT Press. doi:10.7551/mitpress/9780262015554.001.0001 Distinguished the mythology that animates ubicomp from the mess of practical reality, and argued that locating the field in a proximate future lets it ignore the mess in the present. It is the book-length treatment of the gap between imagined and practised systems.
• Ntoa, S. (2024). Usability and User Experience Evaluation in Intelligent Environments: A Review and Reappraisal. International Journal of Human-Computer Interaction. doi:10.1080/10447318.2024.2394724 Reviewed 46 usability and user experience frameworks for intelligent environments and found most remained conceptual, 21 specified no metrics and only three reached concrete metrics and methods. The field lacks the instrument to decide whether its attention promise was met.
• Pavlov, B., Li, J. J., Fang, J., Wang, Y. T. and Shi, Y. C. (2026). Routine Computing: A Systematic Review of Sensing Daily Life Dimensions Towards Human-Centred Goals. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. doi:10.1145/3772318.3791269 Synthesised 203 studies up to August 2025 into the first systematic review of routine computing, with a taxonomy covering temporal granularity, behavioural and social context and cognitive factors. The everyday-computing theme has become a programme whose open problems are intent, personalisation and privacy.