cacm-ubicomp

8/13/2019 cacm-ubicomp

http://slidepdf.com/reader/full/cacm-ubicomp 1/34

6 2 December 2002/Vol. 45, No. 12 CO MM UN IC ATI ONS OF THE ACM



The next step in this evolution involves the move towardubiquitous computing, in which computers will be embedded inour natural movements and interactions with our environ-ments—both physical and social. Ubiquitous computing willhelp organize and mediate social interactions wherever and whenever these situations might occur. The idea of such an envi-ronment emerged more than a decade ago in Weiser’s [2] semi-nal article and its evolution has recently been accelerated by improved wireless telecommunications capabilities, open net- works, continued increases in computing power, improved bat-tery technology, and the emergence of flexible softwarearchitectures. Consequently, during the next five to ten years,ubiquitous computing will come of age and the challenge of developing ubiquitous services will shift from demonstrating thebasic concept to integrating it into the existing computing infra-structure and building widely innovative mass-scale applicationsthat will continue the computing evolution.

The movement into the ubiquitous computing realm willintegrate the advances from both mobile and pervasive comput-ing. Though these terms are often used interchangeably, they areconceptually different and employ different ideas of organizing and managing computing services (see the accompanying fig-ure). Mobile computing is fundamentally about increasing our

ILLUSTRATION BY RICHARD TUSCHMAN

A fun dament al measur e of pr ogr ess in comput inginvolves rendering it as an inseparable part of our every-day experience while simultaneously making it disappear [2]. Radical improvements in microprocessor cost-perfor-mance ratios have pushed this process forward while drastically reducing computing-device form factors,enabling us to embed computers in many parts of our environments. In 40 years this change has transformed the early large “computing machines” into compact devices that enable, mediate, support, and organize our daily activities.

Issues and Challenges inUbiquitous Computing

By Kalle Lyytinen and Youngjin Yoo

COM MUN IC AT IO NS OF THE ACM December 2002/Vol. 45, No. 12 6 3





rethinking of feasible architectures, design ontolo-gies and domain models, requirements and interac-tions scenarios, and analyzing new families of nonfunctional requirements (such as configurability and adaptability). Anticipated new ways of dynami-cally configuring services will also shift the line

between proactive design and tailoring during use.Previously, unexplored challenges will also emerge atthe border between the technical and the social:some issues are to be left outside the technical imple-mentation to be addressed by social negotiation anddue process; other issues should be addressed during technical design. Finally, the emergence of truly inte-grated sociotechnical systems will create a wide array of research and policy issues that deal with socialorganization, impact, and the future of work, orga-nizations, and institutions.

We conducted a workshop at Case Western

Reserve University in late 2001 to address theseissues with leading researchers in the ubiquitouscomputing realm from academia and industry. Thegroup included researchers with an interest either inthe technical or the social issues associated with thisemerging field. Our goal was to examine the statusof the field, to formulate new research directions, toexplore and refine an emerging research agenda, andto identify topics requiring particular attention inthe future. The articles in this special section arederived from material presented at the workshop.The authors provide rich explorations into pending issues surrounding ubiquitous computing anddemonstrate diverse and even conflicting ideas of how mobility and embedded computing can andshould be integrated, along with giving considera-tion to associated social ramifications.

The first two articles in this special section com-plement one another quite effectively. Davis focuseson the potential positive and negative impacts of mobility on individual knowledge work and its orga-nization. Grudin discusses the impact of increasedembeddedness on group dynamics and meeting behavior. The next two articles highlight specificdesign and business challenges affecting ubiquitouscomputing. Siewiorek’s article addresses design chal-lenges of integrating wearable computing withincreased mobility in the context of supporting office and factory work. The article by Fano andGershman explores potential business opportunitiesand challenges resulting from integrating mobility and embedded computing. The last two articles inthe section paint a broader picture of social and tech-nical challenges involved in making progress in ubiq-uitous computing. Jessup and Robey discusspotential impacts of ubiquitous computing, includ-

ing its unintended and contradictory outcomes, which require attention by practitioners of severalsocial and technical disciplines at multiple levels of analysis—individual, team, and organizational.Banavar and Bernstein identify key characteristicsand functionality of emerging ubiquitous comput-

ing environments. They also describe specific tech-nical challenges involved in building supporting infrastructure and applications.

The emergence of ubiquitous computing willprovide a rich and exciting opportunity for futureresearch. As evidenced by the articles here, studying ubiquitous computing itself presents many uniquechallenges—we close by noting three: First, ubiqui-tous computing is currently in an early stage of development. Therefore, studying it entails studying something that does not yet exist. Researchers in thisfield are still “dreaming” and “creating problems” as

much as they are solving problems and recording and theorizing about effects. Researchers need tofind ways to maintain the rigor of scientific research

without restraining their ability to imagine. In addi-tion, researchers need to find ways to study personalissues at a global level. For example, how wearablecomputers interact with environments and how thisaffects individuals’ satisfaction or productivity mustbe studied in the context of global diffusion of mobile technology. Finally, research in ubiquitouscomputing requires transcending the traditional bar-riers between social and technical as well as levels of analysis—individual, team, and organizational [1].

As technology becomes more embedded and inte-grated with mobility, the barriers between social andtechnical aspects become blurred. A paradoxical out-come of ubiquitous computing is that it is simulta-neously very personal and extremely global. Thus, a complete understanding of its impacts cannot begained at a single level of analysis. The articles thatfollow will stimulate the debate on how to addressthese compelling issues.

References1. Lyytinen, K. and Yoo, Y. The next wave of nomadic computing.Infor-

mation Systems Research 13, 4 (Apr. 2002).2. Weiser, M. The computer for the 21st century. Scientific American,

(Sept. 1991), 94–104.

Kal l e Lyyt inen ([email protected]) is a professor at Case Western Reserve University in Cleveland, OH.Youngjin Yoo ([email protected]) is an assistant professor of information systems at Case Western Reserve University in Cleveland,OH.

© 2002 ACM 0002-0782/02/1200 $5.00

c

COM MUN IC AT IO NS OF THE ACM December 2002/Vol. 45, No. 12 6 5



A dramatic increase in access to data and computing by knowl-edge workers can be achieved technically by mobile computing devices and/or by embedding computing devices in products andproduction technologies. When compared to current desktopaccess to data and computing, these technologies give greater avail-ability to transactions as they occur. Availability is achieved by embedded systems in products, processes, and buildings, and by allowing mobile computing input from personnel. The technolo-gies release knowledge work from the constraints of a fixed officelocation and fixed office hours. Knowledge workers can work withfull access to communication, data, and computing from any loca-tion at any time. Unlimited access and data availability as produc-tion and service events occur may change important aspects of knowledge work. For example, embedded computing devices may change the interface between physical workers and knowledge workers. In traditional information systems, there are usually information-handling delays between physical work or movementof physical goods in order to enter and process data about the work. With pervasive computing, tracking of physical work andmovement of goods is continuous and available on demand.Unlimited access to computing and communications networksalso changes the processes and dynamics of the knowledge work activities of communication, coordination, document sharing,knowledge exchange, and collaboration.

The possible effects of anytime/anyplace comput -ing on the productivity of an important part of the work

force known as knowledge workers are a significant area

of speculation and a subject worthy of further exploration.Since there is little experience with omnipresent comput-ing in knowledge work, the good and bad effects on pro-ductivity are anticipated based on the existing activities and behaviors of knowledge workers. Knowledge work is a significant portion of the work done in organizations,so productivity by knowledge workers is a matter of great concern to both organizations and individuals [3].

Anytime/Anyplace Computingand the Future of Knowledge Work

By Gordon B. Davis

Considering theimplications and

consequences of thealways-connected

lifestyle.

COMMUNICATIONS OF THE ACM December 2002/Vol. 45, No. 12 67



68 December 2002/Vol. 45, No. 12 COMMUNICATIONS OF THE ACM

A fundamental assumption in adopting new tech-nologies is that organizations and individuals are will-ing to invest in new technology and new applicationsif it results in improved performance and productiv-ity. However, it is difficult to foresee the results of new technologies—there may be unintended conse-quences. Organizations and individuals may makeunanticipated responses as they adapt to new tech-nology. They may create new structures to promoteor restrict its use. It is therefore useful to anticipateboth desirable and undesirable responses and conse-

quences. An organization can incorporate theseexpectations in its plans, procedures, systems, andtraining for using the technology.

Knowledge Work and Knowledge WorkersKnowledge work is inherently cognitive rather thanphysical [1]. Examples of outputs from knowledge work are analyses, evaluations, instructions, programs,plans, assurances, reasoning or arguments, decisions,and action plans. In other words, knowledge work ishuman mental work performed to generate usefulinformation and knowledge. In doing the work,knowledge workers access data, use knowledge, employ mental models, and apply significant concentrationand attention. A knowledge worker’s dominant activi-ties in terms of time, energy, or intensity are knowledge work. Examples are systems analysts, programmers,accountants, managers, analysts, and lawyers. Work may be done individually, in groups, or in teams. How do knowledge workers differ from clerical workers? Justas knowledge workers may engage in some clericalactivities in performing knowledge work, clerical work-ers may perform some knowledge work activities. Thedifference is the mix of work activities; for the knowl-edge worker, the dominant, most important activitiesare knowledge work activities.

Knowledge workers have value because of theirknowledge and their abilities to apply it in work activ-ities. They are expected to possess formal knowledgeconsisting of general principles, concepts, and proce-dures related to classes of problems and domains of work. They also have some procedural knowledgeabout typical procedures, forms, and rules governing a domain of work. In general, a knowledge worker hassignificant responsibility for structuring and manag-ing his or her work.

Knowledge Work Tasks and Activities. There arethree types of knowledge work tasks: job-specific,knowledge-building and maintenance, and work management [2]. The effect of unlimited access com-puting will depend on specific tasks and activities.

Job-specific tasks.Every knowledge worker has job-specific tasks that produce outputs of value to theorganization. Examples are preparing a budget, ana-lyzing results in terms of estimated and actual costs,planning and scheduling a project, eliciting and doc-umenting system requirements, and writing applica-

tions software.Knowledge-building and maintenance tasks.Knowl-edge workers are valued for their knowledge andexpertise, but this will decay over time. Therefore,knowledge workers need to engage in frequent knowl-edge building and knowledge maintenance. Examplesof this second type of knowledge work tasks are scan-ning and reading professional literature, attending professional meetings, learning new systems and tech-nologies, and building a network of colleagues.

Work management tasks.These support self-man-agement of knowledge work. Examples of work man-agement tasks are planning and scheduling work,allocating time and attention, and acquiring access toresources that enable effective work.

Productivity in Knowledge Work. There are very large differences in productivity among knowledge

workers. For example, using typical measures of per-formance, productivity of the best performing systemsanalysts and programmers can range up to three ormore times that of the lowest performers (who are pro-ductive enough to be retained in their positions). Thishigh ratio is not usually found in production work andclerical work because the organization provides work routines that reduce wasted time and effort and estab-lish an expected pace for the work. In knowledge work,there may be some organization standards and proce-dures, such as deadlines for reports, requirements forevidence of progress, and expectations about outputs.These factors provide some incentives for work com-pletion, but quality of work and timely completiondepend largely on self-management and self-pacing.

The most productive knowledge workers tend toemploy the most efficient work flow and work meth-ods. More important, they tend to be better at man-aging the use of their time, attention, and motivation.

Organizations and individuals may make unanticipatedresponses as they adapt to new technology.



Knowledge work productivity depends on good self-management. For example, a very productive knowl-edge worker will schedule for productivity (scheduleimportant, high-productivity work activities to occurduring times of high energy and attention), schedulefor motivation (create motivation by frequent, short-

term deadlines), and manage demands for attention(because a knowledge worker has limited attentionresources and an oversupply of inputs to process).

The value of unlimited computing access may depend on the knowledge work tasks. Knowledge workers whose tasks involve obtaining data from a variety of locations, activities, and people (such as a scheduler of production activities) are likely to benefitfrom anytime/anyplace computing and communica-tions facilities. Less-certain benefits may be achieved if the knowledge work being performed is dependent onconcentrated effort without interruptions.

Possible Beneficial Effects of Anytime/Anyplace Computing on Knowledge Work In unlimited access computing, a person has access atall times and all places to all information and commu-nication resources. The access mechanisms are soportable they move with the person or are found in theplaces in which the person moves. Four beneficialeffects of unlimited access that change the perfor-mance of knowledge work are: removal of time andspace constraints to communications, removal of timeand space constraints for doing knowledge work,improved access to decision makers, and increasedability to receive and process a rich stream of signalsabout the organization and its environment.

Enhanced Capabilities for Communications,Coordination, Collaboration, and Knowledge

Exchange. A significant characteristic of knowledge work is communication, coordination, and collabora-tion. These important activities may be within a work group, within a project team, with suppliers or cus-tomers, with production or clerical workers, or withmanagers. These activities usually involve sending andreceiving messages, holding physical and virtual meet-ings, or exchanging documents. With anytime/any-place computing, a knowledge worker is no longerconstrained by employment location and employmenttimes in performance of important communications,coordination, and collaboration activities. Communi-cation, coordination, and collaboration access are avail-able 24/7 from anywhere. This is especially importantfor those whose jobs require communication, coordi-nation, or collaboration with persons in other locationsand other time zones.

Exchanging messages, holding meetings, andexchanging documents may involve data (such as data

describing various attributes of events or transac-tions), information (data organized and analyzed inmeaningful ways), and knowledge (information syn-thesized into rules or guides for action). Pervasivecomputing enables prompt exchange of data becauseembedded computers record event data as events

occur and make the data available immediately or onrandom demand. It eliminates delays in recording physical work and physical movement of goods andmaking the data available.

Unlimited access computing enables exchange of information because the analyses and documents thatcontain information can be accessed at any time andfrom any place. Insights reflecting knowledge areaccessed by finding documents that explain the mean-ing and implications of past projects or by finding individuals who have developed knowledge by work-ing with conditions and situations similar to ones

being experienced. Unlimited access computing sup-ports search engines that find and retrieve relevantdocuments, analyses, and reports. It also supportsfinding and connecting with individuals who havedesired knowledge.

Removal of Time and Space Constraints for Doing Knowledge Work. The use of an office for work and the associated custom of set hours for theoffice to be open and in use have imposed both timeand space constraints on performing knowledge work.These constraints may be convenient for office supportservices, but they may not be optimal for productivity. Anytime/anyplace computing allows the office func-tions to move with the knowledge worker rather thanbeing tied to a physical office.

With unlimited access and mobile computing,knowledge work can be removed from the constraintsof office hours. Knowledge workers can take advan-tage of productive times outside of normal work hours. Access to computer and communicationsresources means knowledge workers who have pro-ductive periods at times other than normal work hours can take advantage of those times. An analyst who has an idea at midnight or wakes up early in themorning with an idea can perform knowledge work with access to all the data and computer support nor-mally available at the office. This also facilitates doing knowledge work with persons in other time zones.

Access to Critical Decision Makers at Any Time.Many times, knowledge work is delayed in order toreceive input or decisions from key stakeholders forthe work being performed. Anytime/anyplace com-puting increases the opportunity to get access to criti-cal decision makers at any time. A cell phone may provide voice access, but mobile computing makes theentire range of computing and communications capa-




bilities available at any time and any place. Spread-sheets, reports, analyses, and requests for commentsand decisions can be sent from any location at any time to relevant stakeholders and decision makers.

Increased Ability to Receive and Process Rich Streams of Signals about the Organization and Its

Environment. Periodic structured reports are com-mon to knowledge workers in organizations. How-ever, aggregated, structured reports may hideimportant information. Recipients of reports may miss important factors because the factors are notsummarized and reported. Readers need contact withthe richness of detail that is only contained in data describing individual transactions. Customer dissatis-faction, employee resistance, product failures, new competition, and other situations are usually signaledvery early by messages that are part of streams of transaction data. It may take months or years for the

early signals to become strong enough to emerge inthe standard summarized reports.Gaining access to transactions and picking up early

signals is one basis for the managerial practice of “walking around” to sample operations and opinionsand for meeting periodically with a sample of cus-tomers and suppliers. In addition, managers tend tobe more confident in their decisions if they can scandetailed transaction data (if they desire).

Mobile computing increases knowledge work access to transaction data. Managers, analysts, andother knowledge workers can sample transaction data to enrich their understanding of the events in theoperations of the organization and its environment.The mobile computing system expands the reach of managing by walking around. The availability of real-time streams of transaction data allows knowledge workers to monitor transactions as they occur. They can select and evaluate critical transactions or envi-ronmental signals.

Possible Undesirable Effects of UnlimitedAccess Computing Although computing with unlimited access may result in beneficial effects on knowledge work pro-ductivity, it may have unintended consequences andundesirable effects. The undesirable consequencesresult from organization and individual behaviors inresponse to the systems that implement anytime/any-place computing. The possible undesirable effects arenot certain; they are possible. Just as beneficial effectsmay not be achieved, undesirable effects may notoccur because new organization structures andprocesses may overcome unfavorable effects. The dis-cussion of undesirable consequences alerts organiza-tions to the need for dealing with these possibilities.

The undesirable effects of anytime/anyplace comput-ing may be found in individual management of knowledge work, weakening of desirable boundariesbetween work and personal life for knowledge work-ers, bias in decision making, and dysfunctional orga-nization behaviors,

Effects on Individual Management of Knowl-edge Work. Managing demands for attention requiresdiscipline in scheduling work and in responding tointerruptions. Computing with access available any time and any place may enable unnecessary andunproductive interruptions to knowledge work. If interruptions are easy, more may happen (similar tothe effect of email increasing the sending of multiplecopies of messages). In fact, pervasive access may change organization expectations, making interrup-tions and immediate responses to them appear to bethe norm. Individuals may be motivated to accept

and process interruptions. Productivity may drop asknowledge workers attend to interruptions and aredistracted from concentrating on important activities.

Interruptions may enable avoidance of some activi-ties. Some knowledge work activities require uninter-rupted periods of concentrated work. For example,long-range planning usually requires concentrationover an extended period. Doing work that needs unin-terrupted concentration requires cutting off communi-cations that interrupt. Mechanisms, such as “hiding out” or relocating to remote work sites, may be less suc-cessful with unlimited access computing.

Effects on Desirable Boundaries between Work and Personal Life. Humans need rest and renewal by changing from work to personal life activities. Also,some personal activities, such as those associated withfamily and friends, are important to a well-roundedindividual. Anytime/anyplace computing by its very nature has the potential for intruding into personaltime and space because the boundaries between work and personal time and space are removed. For mostprojects, it is probably destructive to long-run pro-ductivity to allow routine 24/7 communications thatintrude into personal time and space.

Effects on Decision Making. Some decisions bene-fit from real-time scanning of individual transactions.However, there are possible adverse effects. One is toomuch data, which may consume scarce attention. Typ-ically, managers do not lack for data; they lack insightinto what is important and needs attention. Also, man-agers, executives, and analysts, by being able to moni-tor individual transactions, may make decisions basedon a few recent events and take actions based on inad-equate evidence. In other words, they may be encour-aged by unlimited access to transaction data to fall into

well-known data analysis traps. Examples are recency




bias (undue weight given to recent events), smallsample judgment errors (imagining patterns from ran-dom data), and concreteness bias (the availability of data encourages its use and discourages seeking moreuseful data).

There are traditional organization mechanisms todeal with the tendency to give too much weight torecent events and make judgments on a very smallsample of transactions. These mechanisms include fil-tering transactions remove irrelevant data and sum-

marizing transactions in reports that aggregateadequately. Effects on Organization Behaviors. The behavior

of individuals and teams in an organization may beadversely affected by the increased communicationcapabilities of anytime/anyplace computing. Theseeffects may be increased communications and coordi-nation costs, tendency toward centralization, andreduced organization development.

Organizations are designed to provide communi-cations and coordination capacity, so individualsand teams receive information relevant to theirduties. Large projects are organized into groups orteams that handle well-defined activities required by the project. The boundaries of the groups aredefined by specification of inputs and outputs. Thisorganization of work reduces the need for commu-nication and coordination among groups. Sincecoordination and communication are costly in timeand attention, team organization is efficient. Unlim-ited access computing is relevant to organizationcommunication, as it enables increased communica-tions, often changing the boundaries of “need toknow” to “you may find this interesting.” The resultmay be an increase in communications withoutachieving increased productivity.

Limits to managerial attention, communication,and coordination require organization hierarchies. Workers at the lowest level deal with individual trans-actions. Middle managers supervise the lowest levels.Top managers receive summarized reports and focuson performance measures and changes in strategy. Sys-tems that enable managers at all levels to observe trans-actions in real time may lead to more centralization of control. Managers can track activities more easily andrequire many decisions to be decided at higher levels

rather than being delegated to personnel performing the transactions. In some cases, getting too muchinformation causes higher-level management to getinvolved, when it should be left to the people close tothe problem. The availability of information may pro-vide the illusion of control: higher-level managers may misinterpret what is going on and make serious mis-takes because of lack of local knowledge.

Another undesirable organization behavior from cen-tralization and illusion of control is the tendency to

move decision making higher in the organization andthus reduce the amount of training for lower-level man-agers (top executives in training). If the “chief” can anddoes make more decisions, then the second in com-mand (chief in training) will not be able to make asmany decisions. The propensity to want the highestauthority to make the decision will be supported. This will be destructive to the development of personnel whocan be future higher-level managers.

Maximizing the Benefits and Minimizingthe Dangers and Losses of UnlimitedAccess Computing Anytime/anyplace computing is a powerful concept,and may change the way some knowledge work isdone and require knowledge workers to adjust themanagement of their work. It might also affect thedesign of organizations. These considerations suggestnew challenges for the designers and implementers of unlimited access computing in organizations.

Unlimited Access Computing and the Way Knowledge Work is Done and Managed. Knowledge work typically involves using personal knowledge oraccessing knowledge from repositories and otherknowledge workers. The work can be either solitary orcollaborative. Anytime/anyplace computing signifi-cantly reduces the time and effort to locate and accessknowledge from repositories and locate and interact with other knowledge workers. The systems enableimproved collaboration and allow knowledge workersto monitor and capture transactions important toknowledge work tasks. Knowledge work can thereforebe performed with better information, better access tothe knowledge of others, and more effective collabora-tion. The result can be higher quality and greaterproductivity.


Since knowledge work is characterized by self-management,providing knowledge workers with new capabilities willcause them to innovate and change the way they work.



The current approach to knowledge work is forknowledge workers to obtain formal knowledgethrough education and training. They are expected,largely, to manage their own work. Self-management works well for tasks defined in formal education ortasks that have become routinized in an organization.

Self-management is often more difficult in scheduling work and managing the use of personal time andattention. Unlimited access computing introducesnew demands for attention and new capabilities forinterruptions. It provides increased capabilities forobtaining data and monitoring transactions with thedanger of being overloaded. Another potential dangeris a decrease in productivity, as knowledge workersallow unmanaged access and interruption that con-sumes their attention and time.

Two approaches may assist knowledge workers toself-manage in this environment. One is to have sys-

tems that assist personnel with self-management, as inavoiding data overload and managing access andinterruptions. The second is training and mentoring that transfers knowledge about self-managementfrom those who have learned to work productively ina high-access, high-interruption environment.

Unlimited Access Computing and the Design of Organizations. Traditional organization structureestablishes boundaries between organization units.They are reflected in responsibilities, hierarchies of authority, decision making, and access to data. Officelayouts reinforce the structure. Mobile computing weakens these boundaries. The design of organizationsystems for the new technology environment enablesknowledge workers to cross organization boundaries,ignore office layouts, and access data without regard tohierarchies. Instead of rigid structures, organizationsmay respond to unlimited access computing withguidelines, norms, and systems that support self-man-agement. This may reduce organizational friction andincrease innovation and productivity.

An organization without some boundaries is prob-ably not possible or desirable: boundaries improveperformance and productivity. Organization designshould consider boundaries between organizationunits, between project teams, and between individu-als who now have increased access and interrupt capa-bilities. An important boundary is between work timeand personal time. Establishing reasonable bound-aries, including norms and expectations, tends tohave a long-run effect on productivity, quality of work life, and retention.

Challenges for the Designers and Implementers of Anytime/Anyplace Computing for Knowledge Work. There are several challenges for designers andimplementers of unlimited access computing. One is

to identify the knowledge work tasks and activitiesthat will benefit and estimate the benefits. A relatedchallenge is to work with user groups to design thesystems and related organization changes. Usersshould be provided both technical and knowledge

work management training for the new systems.

Change procedures to respond to unanticipatedeffects and unexpected adaptations in system usesshould be implemented.

Identifying the knowledge work that will benefitfrom dramatically increased access computing is com-plicated by the anticipation of changes in the ways

work will be performed. Looking at existing work processes may not indicate the most fruitful applica-tions. The process of requirements discovery is there-fore a combination of identifying tasks that are likely to benefit without major changes in the work andcharacterizing those that will benefit only if the work

process is altered. The requirements process mustinvolve potential users and elicit both technical andsocial requirements, social requirements being thoserelated to satisfying work and quality of work life.

Even given a set of social and technical require-ments, the design and implementation of a system tosupport knowledge work still needs significant userparticipation. Knowledge work processes are not rou-tinized; they may vary somewhat based on the persondoing the work. Much of the work process is proba-bly not documented. The design of a successful sys-tem will probably involve changes in the way many tasks and activities are performed and may involveorganization changes. In this environment, it is espe-cially important to have a collaborative process in

which potential users participate.It is vital to train knowledge workers to manage

the capabilities provided by anytime/anyplace com-puting. The technical training is fairly simple. Themain problem is to provide training for knowledge

workers in how to employ the system to support pro-ductivity and timeliness in their knowledge work.This training should emphasize both the advantagesand benefits that can be achieved and the dangers thesystem will enable and even encourage dysfunctionalknowledge worker behaviors. Since many of theimportant principles and examples of productive andunproductive use will emerge through use, a training strategy that includes follow-up sessions may be themost effective.

Unanticipated effects occur with all systems, butthey are likely to be very important among knowledge

workers. Unlike clerical or production systems whereintended use can be reasonably enforced and desirablesystem changes documented and implemented care-fully, the knowledge worker environment typically




involves more independence in system use andundocumented adaptation. Given a process of adap-tation by knowledge workers and creation of new structures for productivity with the systems, a follow-up process to identify these changes may be very use-ful. Some of the adaptations may suggest future

system changes.ConclusionUnlimited access computing with anytime/anyplacecapabilities introduces powerful new technology intoorganizations. For some workers, the mobility andhigh level of access to data and personnel will improveproductivity. For the large segment of the work forceidentified as knowledge workers, the productivity andquality of work life effects are somewhat uncertain(but certainly interesting). This overview of some of the issues in unlimited access computing for knowl-

edge workers suggests opportunities and dangers.Since knowledge work is characterized by self-man-agement, providing knowledge workers with new capabilities will cause them to innovate and changethe way they work. At the same time, knowledge workers will need to learn to manage their attention,one of their scarce human resources, in an environ-

ment in which barriers to interruption are reducedand expectations of availability are increased. Tradi-tional boundaries and roles in organizations will bechallenged by increased access to data and personnel.Faced with powerful capabilities, individuals andorganizations will create new structures and adapt to

them. The system designer challenge is to understandthe new requirements and build new systems that willincrease productivity and, at the same time, maintainor improve the quality of working life.

References1. Davis, G.B. A research perspective for information systems and example of

emerging area of research.Information Systems Frontiers 1, 3 (1999), 195–203.2. Davis, G.B. and Naumann, J.D. Personal Productivity with Information

Technology.McGraw-Hill, New York, 1997.3. Drucker, P.F. The Age of Discontinuity . Harper & Row, New York, 1969.

Gordon B. Davis ([email protected]) is the HoneywellProfessor of Management Information Systems at the University of

Minnesota.Permission to make digital or hard copies of all or part of this work for personal or class-room use is granted without fee provided that copies are not made or distributed for profitor commercial advantage and that copies bear this notice and the full citation on the firstpage. To copy otherwise, to republish, to post on servers or to redistribute to lists, requiresprior specific permission and/or a fee.

© 2002 ACM 0002-0782/02/1200 $5.00

c




Human nature changes very little. Whether its constancy isviewed as reassuring or as a cause for concern, technology design is constrained by ourinnate capabilities and limita-tions. These innate character-istics are social as well asperceptual and cognitive.Humans have lived in groupsfor millions of years and socialbehaviors have been essentialto survival: group interactionskills are part of our nature. Incontrast, we are not designedto function in the organiza-tions, nations, and other large-scale aggregations that only emerged when consistent tem-perate climates permitted theexpansion of agriculture overthe past 10,000 years.

We accept that technology must accommodate human

perceptual and motor limita-tions. Less evident is the factthat, because group dynamicsare also ingrained in humannature, unsusceptible tochange and inaccessible toconscious awareness, technol-ogy must also accommodatethe way we naturally interact.It is not readily apparent how to make this happen. This may contribute to the slow uptakeof many forms of groupware.In contrast, because organiza-tional dynamics are not inte-gral to our nature, technology use, like management practice,can be more improvisational atthat level.

Technology and GroupsDigital technology designed tosupport group interaction fallsloosely into three phases: pre-

The vision of ubiquitous computing [10 ] hascentered on potential benefits of widely distributed input and output devices—sensors, effectors, and dis-

plays that will be carried, worn, or embedded in the environment. These will benefit us as individuals.Social implications of these technologies are acknowl-edged but are rarely considered carefully. In this article I focus on group dynamics in this new world. This requires expanding the focus to include the vast network that will link the devices and the digital repositories dis-tributed across that network.

Group Dynamicsand Ubiquitous Computing

By Jonathan Grudin

From “Here and Now”to “Everywhere and

Forever.”

7 4 December 2002/Vol. 45, No. 12 COM MU NI CAT IO NS OF THE ACM



ubiquitous support for face-to-face meet-ings (beginning in approximately 1970),proto-ubiquitous support for distributedmeetings (beginning in earnest inapproximately 1990 although visionary demonstrations appeared earlier), and

ubiquitous support for activity in distrib-uted locations over time, now getting under way. Here, I review discoveriesfrom the first two phases that provideinsight into the immutable aspects of group dynamics with which ubiquitoustechnology must coexist.

Pre-Ubiquity: Face-to-Face Meeting Support. For millions of years, ourancestors interacted directly. Complex social and emotional behaviors haveevolved that are tuned to this context.

“Same-time, same-place” technology support appeared as early as 1971 [3].Tools for brainstorming, idea organizing,and voting were refined over twodecades. Electronic meeting room suc-cesses included a widely publicized study that showed a 90% reduction in projecttime [8]. Three companies marketedthese types of products in the early 1990s. None did well (the Fortune 500company that reported a 90% saving dis-continued use of their system).

A Wake-Up Call. In 1990, the Insti-tute for the Future convened a group of experts to discuss the future of group-

ware. They used a commercial electronicmeeting room system for the day-long exercise, heralding its efficiency. In2001, the IFF again convened experts.No digital technology was used—instead, markers, paper, and colored tags

were used to record, distribute, andendorse ideas. Technology can supportmeetings effectively: blackboards,flipcharts, microphones, overhead pro-

jectors, and slides. The benefits may notalways be great—presentation expertsdiscourage its use, arguing that effectivespeakers should draw audience attentionto themselves—as speakers have donesince pre-history. Even when a digitalsystem does increase productivity, meet-ings are complex social situations thatmust also address other concerns,notably group cohesion and member

well-being [5]. Establishing and main-taining status, process, and morale can beas important as productivity. For exam-ple, a meeting was stopped by an execu-tive when no one would admit to having typed a critique that was displayed

anonymously by software that supportsbrainstorming [7]. This illustrates thatstatus and process concerns can outweighimmediate productivity goals.

A related reason we are reluctant toadopt new meeting-support technologiesis that unlike with a personal productiv-ity tool, experimentation occurs in pub-lic. We learn by making mistakes, butmistakes in this domain are often embar-rassing. To avoid problems, the use of trained facilitators and technology

experts is recommended, but they divertattention from the meeting “owner.”Process and status are altered.

We complain about meetings and real-ize that some could be more effective,but on the whole we are experienced

with and comfortable with meetings.Support technologies that have suc-ceeded (blackboards…) are those thatminimally alter the social dynamics. Themessage for ubiquitous computing is thatthese technologies, too, should meld

with human social dynamics. They mustnot focus exclusively on improving pro-ductivity if this requires us to changefundamental aspects of how we interact.

Proto-Ubiquity: Distributed Meet-ing Support. Consider a relatively sim-ple technology enhancement: two distantconference rooms linked by a micro-phone on each table and a camera andmonitor at one end of each room. With-out the system, the person at the head of the table (the organizer) is the center of attention and people less interested inthe meeting topic sit at the far edges.

With the technology in place, the boredpeople loom largest in the camera view,the distant organizer is most difficult todiscern. Attention in the room leaves theorganizer and focuses on the monitor,making the organizer uncomfortable.

When the organizer addresses remoteparticipants, visible as small imagesacross the room, the natural tendency is

COM MU NI CA TIO NS OF THE ACM December 2002/Vol. 45, No. 12 7 5



7 6 December 2002/Vol. 45, No. 12 COM MU NIC AT IO NS OF THE ACM

to raise the voice, which microphone placementtranslates into shouting. Lighting is difficult: distantparticipants often appear washed out, ill, even moreso if the video is delayed or jerky. The net effect onremote participants—being shouted at by a small,expressionless person who looks ill and is sur-rounded by inattentive colleagues—may not stop work from getting done, but can unconsciously undermine the experience, team cohesion, and

morale. And this is a scenario involvingsimple tech-nology intervention.Real-time distributed communication has been

examined, often in two- and three-person groups,sometimes in the field [4]. These studies indicateparticipants in distributed meetings lack importantcontextual information that is present in face-to-facemeetings: knowledge of who is coming and going atremote sites, how people are reacting, what objectsare around them, what they are paying attention to,and so forth. Seemingly small things make a differ-ence, such as whether remote participants are seatedin an auditorium or crowded in an office, have full-or half-duplex audio, or have a fax machine or com-puter nearby. The importance of audio quality isoften underestimated even by participants them-selves. For example, if everyone at one site laughs ata side comment that could only be heard in theroom, distant participants are not in on the joke andmay wonder whether they were its target, since in a face-to-face meeting people lower their voices pre-cisely to avoid being overheard.

The fact that participants in distributed meetingsare actively (though perhaps unconsciously) seeking contextual information is an opportunity for ubiq-uitous computing. If we identify the missing infor-mation, sensors may be able to deliver it. But thereare challenges. Identifying the important contextualelements may be difficult. When we succeed, how do we then deliver it in a manner that can be appre-hended as easily and accurately as in person, giventhat our sensory and social skills have evolved formillions of years to do the latter effectively? The evi-dence suggests this is difficult. For example, peopleeffortlessly signal emotion through facial expression,body posture, and gesture. Emoticons and mood

indicators do not work nearly as effectively. Another challenge is that representing contextual

information digitally transforms it in several ways[2]. Most radically, once on a network, informationcannot be controlled. Gestures, expressions, and

words in a face-to-face meeting are ephemeral. Butonce on a network, contextual information couldreappear anywhere in the world, at any time in thefuture. This aspect of ubiquitous computing is

explored later in this article. Asynchronous Group Support: The Realm of Ubiquitous Computing. Group activity that is dis-tributed in time and space includes individual activ-ities that must be coordinated, informal and formalcommunication, and information sharing. As com-puting diffuses throughout our environment, all of these activities could be supported. To identify opportunities and challenges, let’s first consider therecord of success and failure.

Personal productivity tools are a major success of computing. These tools have been improved by adapting them to innate human perceptual and cog-nitive characteristics. Efforts to coordinate individ-ual work have been far less successful. The dozens of

workflow management products, similar to theiroffice automation predecessors of the 1970s, areappealing in concept but not successful in practice,arguably because they have not accommodated theflexible ways people work together.

In contrast, many communication and informa-tion-sharing tools have succeeded, especially thosethat resemble preexisting approaches. Email andinstant messaging are similar to conversing; attach-ments resemble objects handed to people. However,digital representation alters them, and the changeshave interesting consequences. Unlike conversation,email and online messages do not disappear as they are consumed, they are not reliably ephemeral.

Attachments can be distributed effortlessly, unlikephysical objects.

Ubiquitous computing encompasses efforts torecord and archive formal meetings and sponta-neous interactions for subsequent review or viewing by those not present [1, 6]. Distributing informa-tion in time and place is the intent, not a

Ubiquitous computing is the ultimate cleavage of actionfrom the “here and now.” Once a digital representation

of an action reaches a network, it could surfaceanywhere on the planet at any future time.



by-product, of the digital representation of sensordata. There may be an intended audience for theinformation, but no guarantee that it will not spreadmore widely. Digital rights management efforts, if successful, might impede but probably could notprevent the spread of digital information.

From “He re and Now” t o “E verywhe reand Forever”Masanao Toda [9] describes the context in whichour social behaviors evolved and for which they areadapted. In the wilderness of pre-history, opportuni-ties and challenges were faced directly. Anger, empa-thy, fear, desire, sadness, joy—all were reactions toobjects, events, actions, or people in the immediatevicinity. Social relations and social status were estab-lished and maintained in face-to-face settings. Theseurges and activities became part of our nature

because they were effective in such environments.Toda notes that when larger-scale social organiza-tion appeared, for the first time people were oftenaffected by actions that took place not “here andnow,” but in distant locations and some time before.Distant events have real consequences. They triggerthe urges and emotions that were designed in and forthe immediacy of the wilderness state, but the perti-nent objects and people are no longer present. Ourreactions cannot be suppressed—it is humannature—but they are often inappropriate. Over thepast 10,000 years we have explored ways to controlor channel urges that evolved for the “here and now”in our less localized organizations and societies.

Memory and storytelling allowed limited excur-sions in time and place. But technologies havesteadily separated us from the things that affect us:transportation, writing and printing, telegraph andtelephone, radio and television. Through them weare affected by past and distant actions of others, andour own actions have echoes that we do not hear.

Ubiquitous computing is the ultimate cleavage of action from the “here and now.” Once a digital rep-resentation of an action reaches a network, it couldsurface anywhere on the planet at any future time. Itmight not, but it might. As a result, we do notalways know the contexts in which we act. A highschool student may create a Web page to be viewedby other students and 10 years later discover it isbeing viewed by prospective employers. Understand-ing our context-dependent social behaviors,designed through natural selection, is crucial inidentifying the limits, opportunities, and risks of context-traversing digital technologies.

When digital information appears in places notoriginally envisioned, it is unclear how it will be

interpreted. Online publication of prisoners being paroled lacks the background information that wentinto the parole decisions; online access to malprac-tice histories does not reveal which medical special-ties are most at risk; private email surfaces in publiccourt cases where the workplace vocabulary and style

are not understood; remarks are made to a smallnewsgroup that the writer never imagined would beread by colleagues; comments in a graduate student’s Web page ruin the student’s subsequent applicationfor a faculty position; a review written for people who had read a paper is made available to an entirecommittee; and so forth. Digital information is notinevitably exposed, but one cannot guess a priori thecontexts in which it will surface. The safest assump-tion is “this could show up on the evening news.”

Toda explores the means by which we establishand maintain status and social relations. Status is a

major concern in primate and human societies. A contemporary egalitarian ethos may cause us tooperate in this sphere more subtly and with less con-scious attention, but it remains supremely impor-tant. Toda analyzes the role of informal conversation(such as chatting and gossip) in social positioning. A key characteristic of such discourse is that it isephemeral, recorded only in imperfect memory.

In [2] I explore Toda’s observations in the contextof email and instant messaging. Email was welcomedas an informal, conversation-like alternative to writ-ten correspondence. At the time, disks offered very limited storage capacity and printing was expensive.Later, saving email became routine and emailbecame more formal. More recently, instant messag-ing was welcomed as an informal, conversation-likealternative to email. In both cases, an ephemeralmedium supported the casualness, imprecision, pos-turing, blustering, provoking, and bantering that ispart of social network maintenance but that canhave a negative effect when encountered out of con-text. However, features are being added to enablemessage sessions to be saved so that we can review them. Can we manage without the ephemeral com-munication that has served us for millennia? Ubiq-uity and perceived efficiency work againstephemerality.

Consider once again a meeting, recorded andarchived for later viewing. Someone is late; those waiting, mildly frustrated, joke at his or her expense.Seen by those present as good-natured defusing of frustration, it may appear less tasteful to a laterviewer who was not there; the late arrival checking tosee what was missed, perhaps. In the meeting, a caustic critique is delivered, the speaker not antici-pating who will later view it, similar to email com-




posed without considering how it will look whensubpoenaed by a competitor’s attorney. One person’ssense of humor leads to a risqué remark that anotherquickly deflects—harmless banter perhaps untiltaken from context to create an impression of a hos-tile work environment. And of course an “expiration

date” for a file is no more reliable than “please deleteafter reading” added to an email message.Context is critical to interpreting social behaviors,

context that is stripped away by digital mediation.Systems that do not mesh with natural social behav-ior will fail or lead to unforeseen outcomes. Frustra-tion that cannot be defused harmlessly may findother outlets. If a meetings can be viewed by poster-ity, will we avoid meetings and seek other places forephemeral expression? And if we leave no suchrefuge, if computing becomes truly ubiquitous, what will our ancestors say to us?

References1. Dey, A.K., Abowd, G.D., and Salber, D. A conceptual framework and

a toolkit for supporting the rapid prototyping of context-aware appli-cations. Human-Computer Interaction 16 (2001), 97–166.

2. Grudin, J. Desituating action: Digital representation of context.Human-Computer Interaction 16 , (2001), 269–286.

3. Kraemer, K.L. and King, J.L. Computer-based systems for cooperative work and group decision making. ACM Computing Surveys 20 , 2(1988), 115–146.

4. Mark, G., Grudin, J., and Poltrock, S.E. Meeting at the desktop: Anempirical study of virtually collocated teams. InProceedings of ECSCW 99 (1999), 159–178.

5. McGrath, J.E. Time, interaction and performance (TIP): A theory of groups.Small Group Research 22 , 2 (1991), 147–174.

6. Moran, T.P., Palen, L., Harrison, S., et al. I’ll get that off the audio: A case study of salvaging multimedia meeting records. InProceedings of CHI ‘97 , (1997), 202–209.

7. Nunamaker, J., Briggs, R., Mittleman, D., and Vogel, D. Lessons froma dozen years of group support systems research: A discussion of laband field findings. Journal of MIS 13, 3 (1997), 163–207.

8. Post, B.Q. Building the business case for group support technology. InProceedings of the 25th HICSS 4, (1992), 34–45.

9. Toda, M. The urge theory of emotion and social interaction. Unpub-lished manuscript. Chukyo University, 1999.

10. Wieser, M. The computer for the 21st century. Scientific American 9 ,(1991), 933–940.

Jonathan Grudin ([email protected]) is a seniorresearcher at Microsoft Research in Redmond, WA.

Permission to make digital or hard copies of all or part of this work for personal or

classroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full cita-tion on the first page. To copy otherwise, to republish, to post on servers or to redis-tribute to lists, requires prior specific permission and/or a fee.

© 2002 ACM 0002-0782/02/1200 $5.00

c




Table 1 briefly summarizes the past four decades of user interfaceevolution. While technologies show doubling of capability every few years, it takes more than a decade for a new user interface to become widely deployed. The extra time is required for working out tech-nology bugs, reducing costs, and adapting applications to the new user interfaces. During the current decade speech recognition, posi-tion sensing, and eye tracking should be common inputs. In thefuture, stereographic audio and visual output will be coupled with3D virtual reality information. In addition, heads-up projection dis-plays should allow superposition of information onto the user’senvironment.

There is no Moore’s Law for humans. Human evolution is a slow process and society-wide human adaptation takes substantial time.For example, the size and spacing between fingers has been essen-tially the same for approximately a millennium. Furthermore,humans have a finite and non-increasing capacity that limits thenumber of concurrent activities they can perform. Human effec-tiveness is reduced as humans try to multiplex more activities. Fre-quent interruptions require a refocusing of attention. After each

The goal of the merger of ubiquitous andwearable computing should be to provide “the right information to the right person at the right place at the right time.” In order for ubiquitous computing to reachits potential, the average person should be able to take advantage of the information on or off the job. Evenwhile at work, many people do not have desks and/or spend a large portion of their time away from a desk.Thus, mobile access is the gateway technology required tomake information available at any place and at any time. In addition, the computing system should be aware of the user’s context not only to be able to respond in an

appropriate manner with respect to the user’s cognitive and social state but also to anticipate needs of the user.

New Frontiers ofApplication Design

By Daniel P. Siewiorek

Applying thelessons learned from

wearable andcontext-aware

computers.





refocus of attention a period of time is required toreestablish the context prior to the interruption. Inaddition, human short-term memory can hold sevenplus or minus two (that is, from five to nine) chunks of information. With this limited capacity, today’s systemscan overwhelm users with data, leading to information

overload. The challenge to human-computer interac-tion design is to use advances in technology to preservehuman attention and to avoid information saturation.

The objective of wearable computer design is tomerge the user’s information space with his or her work space. The wearable computer should offer seamlessintegration of information processing tools with theexisting work environment. To accomplish this, the wearable system must offer functionality in a naturaland unobtrusive manner, allowing the user to dedicateall of his or her attention to the task at hand with nodistraction provided by the system itself. Conventional

methods of interaction, including keyboard, mouse, joystick, and monitor, all require some fixed physicalrelationship between user and device, which can con-siderably reduce the efficiency of the wearable system.The most recent research on wearable computing canbe found at the International Symposium of WearableComputing Web site: iswc.gatech.edu.

A three-tiered taxon-omy based upon the timerate of change of data canbe used to categorizemobile applications:

• Procedures. Mainte-nance and plant opera-tion applications arecharacterized by a largevolume of informationthat changes slowly over time. For example, evensimple aircraft will have over 100,000 manual pagesassociated with them. But due to operationalchanges and upgrades, half of these pages becomeobsolete every six months for even very mature air-craft. Rather than distribute CD-ROMs for eachmaintenance person and incur the risk of a mainte-nance procedure being performed using obsoleteinformation, maintenance facilities usually maintaina centralized database to which maintenance person-nel make inquiries for the relevant manual sectionsusing demand. A typical request consists of approxi-mately 10 pages of text and schematic drawings.Changes to the centralized information base canoccur on a weekly basis.

• Work Orders. The trend is toward more customiza-tion in manufacturing. In aircraft manufacturing notwo aircraft on an assembly line are identical. The

aircraft may belong to different airlines or be config-ured for different uses. Customization extends toother industries. One leading manufacturer pro-duces over 70,000 trucks per year, representing over20,000 different configurations. The customer canselect the transmission, the engine, and even the

stereo system. In the near future trucks will beaccompanied by their own documentation describ-ing “as built,” “as modified,” and “as repaired.” When personnel doing manufacturing or scheduledmaintenance arrive for work they receive a list of joborders describing the tasks and including documen-tation such as text and schematic drawings. Thus,this information can be expected to change on a daily or even hourly basis.

• Collaboration. There are times, however, when anindividual requires assistance from experienced per-sonnel. Historically, this assistance has been pro-

vided by an apprenticeship program wherein a novice observes and works with an experienced worker. Today, given downsizing and productivity improvement goals, teams of people are geographi-cally distributed yet are expected to pool theirknowledge to solve immediate problems. A simpleexample of this is the help desk scenario, wherein an

experienced person iscontacted for audio andvisual assistance in solv-ing a problem. The helpdesk can service many people in the fieldsimultaneously. Anextension of help desks

is a team of personnel such as police and firefighters who are joining together to resolve an emergency situation. Information can be expected to change ona minute-by-minute and sometimes even second-by-second basis.

When combined with ubiquitous computing, wear-able computers will provide access to the right infor-mation at the right place and at the right time.Distractions are even more of a problem in mobileenvironments than desktop environments, since theuser is often preoccupied with walking, driving, orother essential real-world interactions. A ubiquitouscomputing environment that minimizes distractionmust be context-aware [1]. Context-aware computing describes a situation in which a mobile computer isaware of its user’s state and surroundings, and modifiesits behavior based on this information. A user’s contextcan be quite rich, consisting of attributes such as phys-ical location, physiological state (such as body temper-ature and heart rate), emotional state (such as angry,

Year Input/Output/Information

Keyboard, alphanumeric display, text

Keyboard/mouse, graphics display, icons

Handwriting/speech recognition, speech synthesis, multimodal

Position sensing/eye tracking, stereo audio/video, 3D virtual reality

1970

1985

2000

2015

Table 1. User interfaceevolution.





boards, human evolution has not kept pace by shrink-ing our fingers. There are minimal sizes beyond whichobjects become difficult to manipulate—the humananatomy introduces minimal and maximal dimensionsthat define the shape of wearable objects. The mobilecontext also defines dynamic interactions. Attempting

to position a pointer on an icon while moving can betedious and frustrating. Wearability is defined as the interaction between the

human body and the wearable object. Dynamic wear-ability includes the human body in motion. Design for wearability considers the physical shape of objects andtheir active relationship with the human form. Theeffects of history and cultures, including topics such asclothing, costumes, protective wearables, and carrieddevices were explored in [2], which also studied physi-ology and biomechanics, and the movements of mod-ern dancers and athletes. The authors of [2] drew upon

their experiences with over two-dozen generations of wearable computers representing over 100 person-yearsof research, codifying the results into guidelines fordesigning wearable systems. These results are summa-rized in Table 2. By considering how the wearable prod-uct designer responded to these design guidelines inTable 2, the buyer can make a more informed purchase.

The long-term use of wearable computers at thispoint in time has an unknown physiological effect onthe human body. As wearable systems become increas-ingly useful and are used for longer periods of time, it will be important to test their effect on the wearer’sbody.

Conclusion and Future Challenges Wearable computers are an attractive way to deliver a ubiquitous computing system’s interface to a user, espe-cially in non-office-building environments. Thebiggest challenges merging ubiquitous and wearablecomputing deal with fitting the computer to thehuman in terms of interface, cognitive model, contex-tual awareness, and adaptation to tasks being per-formed.

User Interface Models. What is the appropriate setof metaphors for providing mobile access to informa-tion (such as, what is the next “desktop” or “spread-sheet”)? These metaphors typically require more than a decade to develop (the desktop metaphor began in theearly 1970s at Xerox PARC; it took more than a decadebefore it was widely available to consumers). Extensiveexperimentation working with end-user applications will be required. Furthermore, there may be a set of metaphors, each tailored to a specific application or a specific information type.

Input/Output Modalities. While several modali-ties mimicking the input/output capabilities of the

human brain have been the subject of computer sci-ence research for decades, the accuracy and ease of use(many current modalities require extensive training periods) are not yet acceptable. Inaccuracies produceuser frustrations. In addition, most of these modalitiesrequire extensive computing resources that will not be

available in low-weight, low-energy wearable comput-ers. There is room for new, easy-to-use input devicessuch as the dial developed at Carnegie Mellon Univer-sity for list-oriented applications.

Quick Interface Evaluation Methodology. Cur-rent approaches to evaluate a human-computer inter-face require elaborate procedures and involve scores of subjects. Such an evaluation may take months and isnot appropriate for use during interface design. Theseevaluation techniques should especially focus ondecreasing human errors and frustration.

Matched Capability with Applications. The cur-

rent common belief is that technology should providethe highest performance capability. However, this capa-bility is often unnecessary to complete an applicationand enhancements such as full-color graphics requiresubstantial resources and may actually decrease ease of use by causing information overload for the user. Inter-face design and evaluation should focus on the mosteffective means for information access and resist thetemptation to provide extra capabilities simply becausethey are available.

Context-Aware Applications. How do we developsocial and cognitive models of applications? How do weintegrate input from multiple sensors and map theminto user social and cognitive states? How do we antici-pate user needs? How do we interact with the user?These, plus many other questions, must be addressedbefore context-aware computing becomes possible.

References1. Dey, A.K., Salber, D., and Abowd, G.D. Context-based infrastructure for

smart environments. In Proceedings of the 1st International Workshop on Managing Interactions in Smart Environments (MANSE 99), Springer-Verlag,New York, 1999, 114–129.

2. Gemperle, F., Kasabach, C., Stivoric, J., Bauer, M., and Martin, R. Designfor wearability. In Proceedings of the Second International Symposium onWearable Computers , IEEE Computer Society Press, 1998, 116–122.

Daniel P. Siewiorek ([email protected]) is the BuhlUniversity Professor of Computer Science and Electrical and ComputerEngineering and serves as Director of the Human ComputerInteraction Institute at Carnegie Mellon University.

Permission to make digital or hard copies of all or part of this work for personal or class-room use is granted without fee provided that copies are not made or distributed for profitor commercial advantage and that copies bear this notice and the full citation on the firstpage. To copy otherwise, to republish, to post on servers or to redistribute to lists, requiresprior specific permission and/or a fee.

© 2002 ACM 0002-0782/02/1200 $5.00

c




The trend toward ubiquitous computing does not representsimply a change in the way people access and use information. Inthe end it will have a profound effect on the way people access anduse services, enabling new classes of services that only make senseby virtue of being embedded in the environment. Ultimately thesetechnologies will lead us to a world of ubiquitous commerce. Theprospect of ubiquitous computing, therefore, poses a fundamentalquestion to businesses: What will it mean to conduct commerce ina world where our physical environments are teeming with services?

Fundamentally, ubiquitous computing can and will change the way businesses and consumers are able to access each other. Gain-ing access to customers has been in the past a key challenge forbusinesses. What if accessing customers disappeared as a problem?Doesn’t the rise of ubiquitous computing promise businesses theability to deliver the right message to the right person at the righttime at extremely low cost? Yet, these new and improved ways of reaching customers raise a whole new set of challenges that inmany ways are far more complex than issues of cost. After all, whatis the right message? When is the right time? Who is the right per-son? (If we can even be sure that our customer will be a person.)Beyond simply reaching a customer with a message, what kinds of interactions will become possible? How do we deploy and interact with services in this new world? How will relationships betweenbusinesses and their customers evolve? These questions must be

Ubiquitous computing will change the way we

live with technology. As Mark Weiser stated: “The most profound technologies are those that disappear. They

weave themselves into the fabric of everyday life until they are indistinguishable from it” [3]. We don’t think of pencils or hinges or faucets as technology. They are just simply features of the world we take for granted and shape the way we act in the world. With ubiquitous computing, using information technology will progres-sively feel more like using these everyday objects thanusing personal computers.

The Future ofBusiness Servicesin the Age of Ubiquitous Computing

By Andrew Fano and Anatole Gershman

Redefining the keyaspects of the

business-customerrelationship.





addressed to fully realize the promise of ubiquitouscomputing, and answering them will become a fun-damental challenge for business strategy.

During the e-business boom businesses began touse the Internet to change the ways in which they reach out to their customers. This has primarily beenthrough Web sites people access from PCs at home orat work. The move to ubiquitous computing where we can interact with a service through a productrather than a PC or phone will radically change the

nature of customer relationships. As businesses striveto achieve ever more intimate customer relationships,it becomes evident that content and interactionmodes appropriate for past channels no longer suffice.If we take seriously the idea of a relationship with a customer, we must heed to the same characteristicsthat foster relationships in other areas of our lives:awareness , accessibility and responsiveness .

When you have a relationship with someone, youare highly aware of one another. You know each oth-er’s concerns and how they change over time. Thedeeper the relationship, the greater the awareness.Meanwhile, the less accessible one is in a relationship,the harder it is to maintain that relationship. One canbe perfectly aware and accessible, but if he or shedoesn’t respond in a way that addresses the other’sneeds, the relationship will fail. Interacting withsomeone is an investment in time, energy, and trustone makes with the expectation that it will lead toappropriate and desired responses—responses that would not be possible without this investment.

Most current CRM applications focus on identify-ing and targeting the right customers. Tools exist tohelp calculate the expected lifetime value of a cus-tomer and which products one should strive to cross-sell. But once you know with whom you would liketo have a relationship, what comes next? What can bedone beyond dedicated call support, self-service Websites, targeted ads, and other assorted inducementsthat are the current model of the day?

Emerging technologies associated with ubiquitouscomputing including the Web, email, mobile phones, wireless PDAs, pagers, instant messaging, collabora-tion environments, videoconferencing, and kiosksallow us to consider approaches that will expand andalter today’s CRM functionality. Simply, they offer

new ways for achieving awareness, new channels foraccessibility, and new techniques for responding.Increasingly affordable technologies are capable of sensing the world. The E911 laws will mandate thesale of location-aware mobile phones. Similar capabil-ities will be found on connected PDAs. Radio Fre-quency Identification tags (RFID) and tagging technologies increase supply chain efficiency and cus-tomer value. As these tags grow in sophistication anddrop in price they will enable a variety of new services

that provide awareness, access, and new ways of responding. Other technologies are also finding more widespread use. Some will support security concerns,such as biometrics, which will enable us to identify and verify individuals in a variety of situationsthrough fingerprints, voiceprints, signature verifica-tion, face recognition, and handprints. To addressthese questions in terms of their impact on businessand strategy we describe the following examples of how ubiquitous computing could transform cus-tomer relationships and services.1

Online Medicine Cabinet. Imagine walking intothe bathroom in the morning, beginning to brushyour teeth, and as you look into the mirror of yourmedicine cabinet, hearing a voice suggesting that,since it is a high pollen day, you should take yourallergy medicine. The cabinet recognized you andyour needs. Reaching for the medicine, you mistak-enly choose the wrong drug. The Online MedicineCabinet gently corrects you and, since you are almostout of pills, orders a refill automatically [2].

Mobile Valet. Imagine entering an electronicsstore while carrying a wireless-enabled PDA. It recog-nizes your location and presents you with service cat-egories appropriate for shopping, such as productinformation, customer service, warranties, financing,and so forth. You choose a product comparison ser-vice you’ve previously found useful and point yourdevice at inkjet printers you’re interested in. This givesyou a product comparison on your PDA. Unfortu-nately, with a frustratingly small screen on your PDA,you can’t see much information. Mobile Valet allowsyou to context-shift the service into a nearby kiosk,

where you can view the information in far greater

Ubiquitous computing can transform some key characteristicsof customer interaction: the role of their location, thescope of the service, and its duration and frequency.

1The prototypes described here were developed at Accenture Technology Labs.



detail [1]. You then point your PDA to one of theprinters and ask a different service provider for prod-uct reviews that are also viewed through the kiosk. Youcheck financing and warranty options. You ask yourpersonal customer service provider if the printer is a good choice to use with your digital camera. Within a

few moments, a customer service representative fromthe store arrives to address your remaining questions.Both examples illustrate the three characteristics—

awareness, access, and responsiveness—of a relation-ship enabled by ubiquitous computing. The servicesare aware of the customer and his or her needs. They can access the customer and provide easy and naturalaccess to the customer at exactly the right time—whenthe customer needs it most. The services are responsiveto the specific needs of the customer and take advan-tage of the resources available at the customer’s loca-tion. The examples also illustrate how ubiquitous

computing can transform some key characteristics of customer interaction: the role of their location, thescope of the service, and its duration and frequency.

The location of your customer becomes the loca-tion of your business. Technology enables serviceproviders to make the location of their customers thelocation of their business. This is the fundamentalprinciple of anchoring cyberspace back to a physicalcontext. I can take a cell phone equipped with a bar-code scanner to a bookstore, use the store to select thebooks I like and then buy them from a different storeby simply scanning the barcodes on the back cover.My cell phone is thereby transformed into a self-service portable cash register for my favorite onlinebookstore. But the online and physical worlds do nothave to be viewed as adversarial. By creating new andinnovative service delivery channels integrated intothe locations we inhabit and the accessories we carry,business will be able to meet people on their ownterms—in the physical world. The Online MedicineCabinet is a good example of how a business can takeits services directly to the most appropriate locationfor its customers. The Online Medicine Cabinet alsoclearly illustrates the competitive importance of phys-ical points of presence.

A physical point of presence wherever your prod-ucts and services are used will become a competitive necessity. Today, an e-commerce-driven Web site isconsidered critical for many businesses. Yet, having such a site hardly means you’ve reached the pinnacleof customer interaction. Consider, for example, twopharmacies. One has the world’s greatest e-commerce Web site, featuring easy ordering, wondrously effi-cient fulfillment, self-service support, order-tracking,advice, account management, and all flavors of per-sonalization. The other pharmacy has this cabinet in

your bathroom. You’ll probably never get to that won-derful Web site. Why would you? Accessibility at thepoint of need to services that are aware of your imme-diate needs enables a far different class of interactionsand consequently customer relationships than dissoci-ated contact points. It would be meaningless even if it

were possible, for example, to receive an email mes-sage on your laptop hours later that you took the wrong pill.

Businesses will need a point of presence at the loca-tion where their customers use products and services. While Web sites are important, and will remain so,they are, in the end, just one of many points of con-tact with customers. The examples also show that it isnot enough for a service to have a physical point of presence: critical elements include awareness, access,and responsiveness at that site. A passive kiosk is notaware of the customer’s needs and cannot be very

responsive. Mobile devices and appliances become the eyes and ears of remote service providers. Knowledge of the customer is mostly historical: what the customerhas done in the past, not what the customer is doing now. This means businesses can apply a variety of data mining tools to estimate a customer’s expected value,select appropriate marketing campaigns, and choosethe level of service. But it helps little to enable andimprove the service or product provided to the cus-tomer at a given moment, because the business does-n’t know the customer’s current situation. Firedepartments don’t rely on data mining records frompast fires to tell them where to go next. They rely onsmoke detectors to tell them what is happening now.

Mobile devices have the promise to provide similarcontext-sensing capabilities resulting in awareness andresponsiveness. Slowly but surely our phones, PDAs,and other more specialized devices (such as digitalcameras) will become aware of their surroundings.They will soon know their locations—often a criticalindicator of the user’s task. For example, if you are ata gas station you are probably buying gas. If you are atthe bank you are probably banking, and, if you’re in a store you are probably shopping. Furthermore, if a service provider knows something about the specificlocation, it can deduce what resources are availablethere. For example, the service provider may know thetype of products and services available locally, whoelse is present, and/or the availability of resources suchas kiosks. Finally, as illustrated by the Mobile Valet,the customer can use the mobile device to inform ser-vice providers by pointing to objects of interest. Thismay be a product they would like to buy, a brokenappliance they would like to fix, or a house for sale for which they would like more information. The mobile




device is, in essence, their remote control to the world.

Services we associate with locations become attached to people. To gain competitive advantageremote services must make the best use of theresources available at the customer’s location. In theMobile Valet scenario the customer accesses his ser-vice provider, and considers financing and warranty options from several competing sellers. The mobiledevice, in essence, enables the customer to access and

bring with him a veritable army of service providersand information resources. Businesses can no longerprey on customers’ ignorance. They can no longerhide the fact that someone else has a cheaper price. Yet, if your customer is in your store you can be in thebest position to match the price or sweeten the deal insome other way.

By working with customer’s service providers, busi-nesses can offer more responsive and efficient services.Remote service providers face a serious constraint: they are, in the end, remote . They probably have no staff atthe location, which greatly limits the nature of the ser-vice they can deliver. The remote service provider can,however, collaborate with the local staff and togetherprovide a better service that is in both of their interests. With the Mobile Valet, the remote customer serviceprovider can inform the local store which model thecustomer is interested in, the accessories desired, andthe concerns he may have. The store is then in a goodposition to select the right service person and providedetailed help, increasing the chances for a sale and a satisfied customer. This results in richer interactions in which the staff of the location is made aware of specificcustomer needs and provided with the opportunity torespond specifically rather than approaching the cus-tomer anonymously. The capability to dissociate spe-cific functions such as customer service from a particular location creates new services such as per-sonal service providers that work across locations tosupport particular needs (such as financing, insurance,technical support, travel, and so forth). This intro-duces a different kind of customer relationship thanone based on a just a specific product or episode.

Services will use the customer’s locationresources to provide the best possible service.Thereis a tremendous amount of hype around today’s

mobile commerce. But when all is said and done, thecurrent race for m-commerce is a race to deliver ser-vices onto what is at any given time the worst display in the room: the customer’s phone, or PDA. This lim-itation prevents the widespread use of m-commercefor many services. The Mobile Valet illustrates anapproach that helps reach beyond these limitations by incorporating and exploiting the resources of the cus-tomer’s location. If the phone is the poorest screen inthe room, perhaps we can make a nearby better screen

available. This technology allows rapid response to a remote service request by first establishing the cus-tomer’s context, identifying available service channelssuch as the customer’s mobile device, nearby screens,audio systems, kiosks, and even human staff (throughRFIDs)—and then deliver the highest fidelity servicethrough the available channels. This is feasiblebecause the locations we inhabit want customers tosucceed and should therefore be motivated to maketheir resources available. The challenge, and opportu-nity, is to transform what is today just a screen in a store to a service channel that can be used by multi-ple service providers.

Service providers must pay continuous attentionto their customers. We normally think of medical careas a service we avail ourselves of a few times a year. Yet,through dedicated service appliances like the OnlineMedicine Cabinet, every visit to the bathroom poten-tially becomes a visit to your health care serviceprovider. Mobile Valet services have to be always on,knowing where the customer is, and what his or herneeds might be and enable any type of service throughthis awareness. How many companies are ready tooffer this level of customer attention today?

Service providers will have to be very selective and precise in their interactions with their cus-tomers. We walk by the thermostat in our residenceshundreds of times without ever adjusting it. In a sim-ilar way, the online medical cabinet will probably besilent most of the time, with only the occasionalminor interaction. It is important, however, that ourrelationship with the service provider has changedfrom one of a few long, intense interactions (such ascheckups) to one characterized by frequent, brief interactions in which “microservices” are delivered. Inthis world, we are in an almost constant conversation


The challenge, and opportunity, is to transform what istoday just a screen in a store to a service channel that can

be used by multiple service providers.





Each issue that follows is illus-trated by an anecdote demon-strating not only the socialpossibilities afforded by ubiqui-tous computing, but also theresidue of past and presentnotions of appropriate behavior.

As a result of the interplay between what is possible and

what people already understand,each social issue described here isframed as a somewhat contradic-tory outcome.1

Individual Behavior The Pied Piper of ConcourseC. Given the fixed positions of computers, furniture, and per-sonnel at the check-in countersin most airports, people havedeveloped expectations andclosely follow norms of checking in to get their boarding passes.They get in the back of the lineand slowly make their way up tothe person at the counter withthe computer. In a European

Social issues include individual, group, and

organizational behaviors that are affected by ubiquitous computing. Our discussion of these issues is prompted by the following questions. What if technology was literally untethered by any physical connection to a network, toa workspace, or to an organization? What new ways tocommunicate, collaborate, coordinate, organize, and manage would we see? Answers to these questions invite

fresh approaches to studying the social consequences of technologies. Ubiquitous computing technologies not only enable new ways of acting and interacting, but alsostimulate fundamental reassessments of the meaning of human action and interaction. In some cases, social actions will occur in entirely new ways, and in other cases completely new social actions will appear.

The Relevance ofSocial Issues

in Ubiquitous Computing Environments

By Leonard M. Jessup and Daniel Robey

New forms of

social interaction andorganization requiremodifying existing

models.

1 An analysis of contradictory organizational outcomes of information technologies can be fou nd in [8].




airport that had recently converted to wire-less computers, airline personnel roamedfreely throughout the concourse, checking in passengers with mobile computing devices. This posed a problem for passen-gers who did not understand how to

behave. In an attempt to get checked in,people lined up behind the roving employee. The scene quickly took on theappearance of the Pied Piper as theemployee with wireless computer walkedaround the concourse with a growing,snaking line of travelers desperately trying to follow the only norm they knew for thatcontext, forming a line behind her.

This example illustrates that ubiquitouscomputing challenges individuals torethink their behavior. However, it also

illustrates that old habits are often difficultto break. Thus, the handheld computersintroduced to remove queues at the airportresulted in mobile queues snaking throughConcourse C.2

More serious issues are raised when wetry to rethink how employees in organiza-tions should be supervised. Should notubiquitous computing simplify andenhance supervisory activity by steering ittoward an evaluation of work output ratherthan behaviors and appearance? Clearly, a mobile and distributed work force that isenabled by portable technologies cannot be

watched physically as office managers once watched workers through glass office walls.However, a manager (or robotic assistant)may more easily monitor the digital tracesof mobile workers’ activities in a ubiquitouscomputing environment. Performanceevaluation might be based on mountains of more detailed information, perhaps con-verted into performance indexes that mea-sure not only end results but alsointermediate activities. How long were youconnected? How many messages did yousend and receive? What was your total doc-ument production? In such a scenario,roaming employees may be less empoweredthan co-located office workers, and thevalue of their contributions might beobscured in the digital representation of their performance.

Although some managers may exercisesuch compulsive monitoring in at attemptto replicate their previous supervisory activ-ities in a ubiquitous computing environ-ment, others may abandon attempts tosupervise. As a result, roaming employees

may receive little guidance or development.Left on their own, without supervision,employees may not learn necessary skills ordevelop organizational commitment.

The search for new ways to superviseindividuals is comparable to the search fornew ways to handle passenger check-in.Because the new possibilities enabled by ubiquitous computing do not carry theirown prescriptions, people must discovernew behaviors on their own. In some cases,those behaviors simply replicate with wire-

less connections what was done previously with wires. In other cases, people abandonexisting practices without finding viablesubstitutes. Airline passengers, for example,may become so confused by the moving check-in counter that they lose their way,miss flights, bypass security, and disruptpassenger boarding in other ways.

Team Behavior It Is Rocket Science! In a recent study of virtual teamwork, Ann Majchrzak and hercolleagues described a team of rocketdesigners from different companies using a dedicated “Notebook” technology to coor-dinate their work [5]. In a few cases, teammembers were able to meet face-to-face,but the team insisted all voice communica-tions be logged into the discussion data-bases in the portable notebooks.Unfortunately, the requirement to writeeverything into textual databases placed toomuch pressure on team members. Not only

was data entry laborious, but members also were unable to express the complex ratio-nales underlying design recommendations.Eventually, the team introduced regularly scheduled telephone conferences to addverbal communication to the electronictext. This adjustment helped the team suc-ceed in its design task.

This example suggests virtual teams may need to employ older technologies, such astelephones, or even face-to-face meetings,to complement their dependence on ubiq-2 We are indebted to Anna Sidorova for this anecdote.





uitous computing technologies. Several recent studies

support the notion that teams are more effective whenthey intersperse face-to-face meetings with remotecommunication [6, 9]. Although ubiquitous comput-ing allows teams to form as needed, vary their compo-sition as tasks change, and operate independently of time and space, we know little about how effective vir-tual teams interact. How do they vary their work rhythms, distribute workloads, and pace themselves?Does each team develop “local” norms for group prac-tice? Are some coordinating technologies better thanothers? Despite the ability of technology to accommo-date any number of members, are there limits on

group size? As with the study of individual behavior,team behaviors in ubiquitous computing environ-ments are likely to reveal contradictory mixtures of oldand new practices.

Behavior of Organizations Virtual Organizations, Virtually Nonexistent. Ubiq-uitous computing has enabled organizational formsthat are ephemeral at best and illegal at worst. Forexample, the “paper” corporations alleged to have beenformed at Enron Corporation were supposedly createdsimply to move, hold, hide, and/or create the appear-ance of corporate assets [2]. This kind of corporateshell game is easier to accomplish with ubiquitous tech-nologies that so easily enable virtual corporations with-out any material basis.

Similarly, ubiquitous computing technologies havemade it possible to operate legitimate virtual compa-nies that merely coordinate the activities of other firms.For example, products may be designed by engineers inCalifornia, manufactured by contract employees inMexico and Malaysia, distributed by an internationalthird-party logistics carrier, and marketed by indepen-dent e-commerce companies and retailers. Theaccounting and information systems functions couldbe outsourced to companies in India, and independentcontractors working from a call center in Nevada couldhandle customer service. In such a “hollow” corpora-tion, no core competency is needed other than thestrategic imagination required to build and coordinatethe partners in the virtual alliance [3].

A related organizational issue concerns the creationand maintenance of social boundaries between work and non-work. Prior to the fourth technological wave,social boundaries specified locations where work wasdesigned to take place, and locations such as the home

lay outside of those spatial boundaries. Temporal

boundaries also specified when a person should work.The 40-hour, five-day workweek and the practice of moving oneself physically from the home to the work-place each day became institutionalized. With theadvent of ubiquitous computing, people can work out-side of traditional spatial and temporal boundaries.

Many households have become primary workplaces[1]. Thus, work can be performed anytime, anywhereas long as workers can maintain contact with otheremployees, customers, and share data via ubiquitouscomputing technologies. The main issue with working anytime, anywhere is having work becomeall the time,everywhere . As individuals and organizations interactmore frequently with portable computing devices, they

will need to establish their own boundaries between work and non-work.

ConclusionThe table appearing here offers a sample of specificresearch issues at each of the levels of social analysis.

We conclude with the following challenges to socialscientists. First, ubiquitous computing enables innova-tive forms of social action, novel organizational forms,and new business models. We can rely only partially upon what we know about people and organizations inorder to better understand how ubiquitous computing is used and managed. The challenge is to exploit andmodify existing social theory to explain, for example,

Level of SocialAnalysis

ResearchIssues

What prevailing social norms are challenged by theadvent of ubiquitous computing?

How can employees be supervised in technology-rich, mobile working environments?

How are definitions of action and work redefinedby ubiquitous computing?

How do work teams adopt and adapt ubiquitouscomputing technologies?

How can virtual teams be most effective?

How are social interactions redefined byubiquitous computing?

What new organizational forms and businessmodels can be realized with ubiquitous computing?

How can reasonable and effective social boundariesbe created and maintained in technology-rich environments?

How are organizations redefined by ubiquitous computing?

Individual

Team

Organization

Social research issues in ubiquitous computing.

Theories of organization should be revised with the thoughtthat organizations are not tied to particular places or times.





Consider the following sce-nario: Jane is the CIO of anorganization that relies oncomputer services and isattending an important meet-ing in her organization’s head-quarters in New York City. Sheis in a conference room withthree coworkers and tworemote participants who have

joined via a computer-basedvideoconferencing system. Thesystem allows her team to seeand hear the remote partici-pants as well use a shared

whiteboard space to jointly edit documents and exploredata. Unfortunately, Jane hasto leave the meeting early, asshe has to catch a plane to visita supplier in Paris. Luckily, shecan continue participating in

the meeting via her smartPDA. As soon as it detects thatshe left the conference room, itroutes the audio part of themeeting to her cell phone. Asshe enters the limousine that istaking her to the airport, thescreen built into the back of the driver’s seat displays thevideo stream of the meeting,including the people as well asthe shared whiteboard space.Furthermore, the audio por-tion of the meeting is trans-ferred to the limousine’s in-carspeakerphone system. As thecar starts to move, the videoadapts itself automatically tochanging network quality by changing its resolution.

As the meeting progresses,an instant message pops up on

In traditional computing environments, usersactively choose to interact with computers. Ubiquitous computing applications are likely to be different—they

will be embedded in the users’ physical environments and integrate seamlessly with their everyday tasks. This vision leads to a set of defining characteristics, require-ments, and research challenges for ubiquitous applica-tions. This article identifies some of the key characteristics via a possible real-world scenario and derives the important application design and software infrastructure challenges that must be addressed by the computing research community.

Software Infrastructureand Design Challengesfor Ubiquitous Computing Applications

By Guruduth Banavar and Abraham Bernstein

Striving to integratecomputing into

everyday activities ina seamless manner.




the screen informing her of a fire at theorganization’s data center in ColoradoSprings. She taps the message on thetouch-sensitive screen in order to dial thesender’s phone number. The sender, thedata center manager, reports that the fire

at the data center severely damaged someimportant servers, but fortunately thebackups are safely located off site. Janedecides to change her plans and fly toColorado Springs immediately to assist

with the repairs. She asks the data centermanager to send her a detailed damagereport in order to plan the repair. Sheactivates her software personal agentusing the in-car keyboard and tells theagent to change her flight reservationsfrom Paris to Colorado Springs. She also

requests that all the files regarding thedata center configuration be transferredto her PDA as soon as the required net-

work bandwidth is available. The agentrebooks the flights. In addition, it infersthat she will not be able to fulfill herappointments in Paris, so it cancels thehotel room reservation using the hotel’sonline services and the meetings with thesupplier by sending the appropriateemail notification. As Jane leaves thelimo, the personal agent “hops” from thecar’s computer to the PDA and scans thearea for a better network connection. Asshe enters the airport terminal, the agentdetects a wireless LAN to which itauthenticates itself and starts download-ing the requested data about the data center.

While waiting at the terminal, Janecalls the data center manager to get thelatest update. Once she boards the plane,her PDA again scans for possible servicesin her environment and detects that itcan connect to the screen and keyboardbuilt into the plane’s seat using an in-airplane wireless network. It displays

Jane’s desktop environment on the screenand highlights the data center informa-tion. After the plane takes off, Jane care-fully plans the necessary repairs, puts inexpress orders for the needed parts, andenters requests for freelancers with theappropriate skills to relieve her ownemployees who are working around the

clock. The PDA’s personal agent auto-matically prioritizes her requests andtransfers some files over the slower air-plane network. As she lands in ColoradoSprings, it can transfer the remaining files using the terminal’s wireless LAN.

By the time she arrives at the data center,the first bids by freelancers have already arrived, and the computer equipmentproviders have received the orders for thereplacement parts.

This scenario illustrates many aspectsof ubiquitous computing, three key char-acteristics of which we highlight here.

Task Dynamism. Ubiquitous com-puting applications, by virtue of being available everywhere at all times, willhave to adapt to the dynamism of users’

environments and the resulting uncer-tainties. In these environments, usersmay serendipitously change their goals oradapt their actions to a changing envi-ronment [7]. In our scenario, for exam-ple, new information about the data center fire arrived unexpectedly, thuschanging the actions Jane wanted totake, such as choosing an alternative des-tination. This requires programs thatdynamically adapt to changes in eitherthe goals or the plan structure by whichthose goals were to be achieved [3].Sometimes the user might actively recon-figure the system to adapt to the new task settings; at other times the system mighthave to infer from its sensory input thatthe user changed his or her mind. Appli-cations will, furthermore, have to be ableto explain why they inferred those task changes and learn from their right and

wrong inferences.Device Heterogeneity and Resource

Constraints. The omnipresence of ubiquitous applications is typically achieved by either making the techno-logical artifacts (devices) move with theuser or by having the applications movebetween devices tracking the user. Inboth cases, applications have to adapt tochanging technological capabilities intheir environment.

If the device itself is mobile (following the user or being carried around by him/her) then it usually has some con-





straints in terms of physical dimensions. Thesephysical constraints limit resources, such as battery power, screen size, networking bandwidth, and soforth. A PDA, for example, has relatively little usablescreen area and limited battery power; a cell phonehas an even smaller screen size but typically a longerbattery life and is at least connected to a network.Furthermore, applications might also experiencevariability in the availability of resources, as demon-strated in our scenario. Obviously those limitations

influence the development of applications and theircapabilities.The second approach to mobility is having the

application follow the user and move seamlessly between devices. Applications will thus have toadapt to changing hardware capabilities (differenttypes of pointing devices, keyboards, network types,and so on) and variability in the available softwareservices. In our example scenario, Jane moved fromthe office (with a PC) to the car (with a more lim-ited handheld and in-car computer), along with dif-ferent user interfaces and networking capabilities inthese two locations. Some resources might evenbecome completely unavailable, for example,because the limo might not have a printer. Themeeting software will have to dynamically adapt tochanging technological capabilities by adapting thecompression rate and size of transmitted pictures ordetermining an alternative output to a printout.

Computing in a Social Environment. Anothermajor characteristic of ubiquitous computing tech-nology is that it has a significant impact on thesocial environments in which it is used. Any intro-duction of a ubiquitous computing environmentimplies the introduction of sensors, which irrevoca-bly have an impact on the social structure, no mat-ter how unobtrusive they seem to be.

Imagine, for example, that your residence is out-fitted with all kinds of sensors to provide informa-tion to a ubiquitous computing system. Would you want your neighborhood police station to be able tomonitor which room you currently occupy (as indi-cated by the alarm system’s motion detectors) andhow much alcohol you are consuming (as inferredfrom your food inventory system)?

There are also policy questions: Who owns the

data from a ubiquitous computing system? How can we avoid making people feel like they are in an infor-mation panoptikon [9]? Can one subpoena the data collected by ubiquitous computing systems? Sincethe answer is probably yes, there might be demandfor ubiquitous computing systems in which the raw sensor data cannot be accessed at all, but processedinferences from the data, such as “burglar entry,” can.

Research Challenges

Given these characteristics, we identify a number of challenges for the computing research community,organized into four broad categories.

Semantic Modeling. A fundamental necessity foran adaptable and composable computing environ-ment is the ability to describe the preferences of users and the relevant characteristics of computing components using a high-level semantic model. Forexample, if we have information that a user prefersaisle seats on an airplane and that the user’s devicesupports voice interaction, then a travel applicationcan ask whether the user would like to have an aisleseat as usual. Ontologies [4] can be used to describeusers’ task environments, as well as their goals, toenable reasoning about a user’s needs and thereforeto dynamically adapt to changes. Furthermore,descriptions of the device capabilities and theirappropriate use will allow applications to reasonabout how to best support users in any given con-text. For example, the system could down-samplepictures when network bandwidth decreases on a device. Or, the system could determine how to mostappropriately interact with users given a social con-text by, for example, deciding not to produce anaudible ring on the arrival of a new message during an opera but only to vibrate.

The research challenges in semantic modeling include developing a modeling language to expressthe rich and complex nature of ontologies, develop-ing and validating ontologies for various domains of user activity, and finally, agreeing on shared ontol-ogy parts or translations for each applicationdomain [2].

Building the Software Infrastructure. An effec-tive software infrastructure for running ubiquitousapplications must be capable of finding, adapting,

Once instantiated, applications may have to moveseamlessly from one device to another and from oneenvironment to another based on the user’s activity.



and delivering the appropriate applications to theuser’s computing environment based on the user’scontext. Semantic models for users, applications,and tasks will be used for this purpose. Such aninfrastructure can support task dynamism as well asheterogeneous and resource-constrained devices.

The first challenge here is forthe system to determine whichuser tasks are most relevant to a user in a particular context.They may be determined basedon history, on preferences, orother knowledge of the user’sbehavior, as well as the environ-mental conditions. Once theuser has selected a task fromthe list of relevant tasks, anapplication must be synthesized

to suit the task, by finding andcomposing the appropriatecomponents and services. Theapplication should then beinstantiated by potentially parti-tioning application componentsamong various devices to takeadvantage of the relative strengths.

Once instantiated, applica-tions may have to move seam-lessly from one device toanother and from one environ-ment to another based on theuser’s activity, as described inour scenario. Applications mustprovide reasonable functional-ity, even when network connec-tivity is intermittent orunavailable, and must recovergraciously from failures. Finally,the software infrastructure mustbe scalable to support the largenumber of devices and applica-tions that ubiquitous comput-ing will enable.

Developing and Configuring Applications.The current trend in the software industry is for ser-vice providers to provide reusable functions via application components called services. These ser-vices are described using a standard description lan-guage, and in the future, using standard ontologies.Such semantic descriptions could enable automaticcomposition of services, which in turn enables aninfrastructure that dynamically adapts to tasks.Building applications will involve specifying theright composition of services, building a user inter-

face, and orchestrating the data flow among the var-ious components.

The requirements of pervasive computing imposea significant shift in the developer’s mindset whilebuilding pervasive applications [1]. Applications arenot pieces of software targeted to a particular device

or a particular environment but rather, high-leveldescriptions of the task a user needs to perform. Thechallenge is to be able to specify the interaction logicat an “intent-level,” and the application’s require-ments on data and computation. This kind of spec-ification allows the infrastructure to adapt theapplication to heterogeneous devices with con-strained resources.

Building reusable services will also be different for


Project Vision/Goals

To develop the requisite technologies tocreate a home environment that can bothperceive and assist its occupants.

To fundamentally rethink system designto address the problem of reduced usereffectiveness due to explicit and implicitdistraction from computers. To provideeach user with an invisible halo of com-puting and information services thatpersists regardless of location.

A technology future where people, places,and things are first-class citizens ofthe connected world, wired and wireless—

a place where e-services meet the physicalworld, where humans are mobile,devices and services are federated andcontext-aware, and everything has aWeb presence.

An open architecture that enablesdevelopers to create network-centricservices—whether implemented inhardware or software—that are highlyadaptive to change.

To bring an abundance of computationand communication to users throughnatural spoken and visual interfaces,making it easy for them to collaborate,access knowledge, and automaterepetitive tasks.

An application model and tools fordeveloping applications in aplatform-independent manner andfor adapting them to multipleheterogeneous device platforms.

To create a testbed for investigationinto ubiquitous task-specific computingdevices, which are so highly optimized toparticular tasks they blend into the worldand require little technical knowledgeon the part of their users.

An extension of the current Web inwhich information is given well-definedmeaning, better enabling computers andpeople to work in cooperation.

Challenges Addressed

User experienceenhancement andvalidation

Software infrastructure;Service constructionand composition; andUser experiencevalidation

Service constructionand composition;User experience

enhancement andvalidation

Context-basedadaptation

User experienceenhancement andvalidation

Applicationdevelopment anddeployment;Context-basedadaptation

Software infrastructure;Service constructionand composition; Userexperience enhance-ment and validation

Semantic modeling

Aware home(Georgia Tech)www.cc.gatech.edu/fce/ahri

Aura(CMU)www.cs.cmu.edu/~aura/

Cooltown(HP)www.cooltown.hp.com

Jini(Sun)www.sun.com/jini

Oxygen(MIT)oxygen.lcs.mit.edu/

PIMA(IBM Research)www.research.ibm.com/PIMA/

Portolano(University of Washington)portolano.cs.washington.edu/

Semantic Web(W3C) www.semanticweb.org

Ongoing projects addressing challenges.



ubiquitous environments. Not only will servicesencapsulate business logic and data but also “infor-mation interfaces” to physical artifacts, such as sen-sors and actuators. Finally, configuring services andapplications so they can be easily and reliably reusedby other developers and composed into larger appli-

cations will be a major challenge. Validating the User Experience. As argued pre-viously, pervasive computing has the potential tofundamentally change the way people use comput-ing devices to perform their tasks. The utility of some computing advancements cannot be evaluated without performing significant user studies and insome cases, widely deploying it. Consequently, thedevelopment of effective methods for testing andevaluating the usage scenarios enabled by pervasiveapplications is an important area that needs moreattention from researchers. The dynamic nature of

some of these usage situations, the device hetero-geneity with its varying resource availability, and the widely varying social environments in which mobilecomputing solutions are being used are, however,complicating the development of such evaluationmethods (see the article “Beyond Prototypes” in[6]). Traditional laboratory experiments might not work in many cases, since they don’t capture the richnature of the usage environments. This will forceresearchers to use more complicated approaches,

such as field-based quasi-experiments or ethnogra-phies. There are several ongoing projects that deal

with various challenges discussed here—a sampling is shown in the accompanying table.

Conclusion

As Mark Weiser described in his seminal article [8],ubiquitous computing is about interconnected hardware and software that are so ubiquitous that no onenotices their presence. This will enable people tofocus on their tasks and on interacting with otherpeople. This far-reaching vision is still far from ourreach [6], and will require fundamental advances insemantic modeling, context-aware software infra-structure, application modeling and tools, and userexperience validation. Going back to our scenario,in order to help Jane manage the fire, we need majoradvances in each of the challenge areas we identified

(as well as in others we have not mentioned). Mostimportant, all these advances must be integrated ina seamless manner into Jane’s life so she can usethem without constantly worrying about either theunderlying mechanics or social appropriateness.

References1. Banavarf, G. et al. Challenges: An application model for pervasive com-

puting. In Proceedings of the ACM Conference on Mobile Computing and Communications (Mobicom), Boston, August 2000.

2. Berners-Lee, T., Hendler, J., and Lassila, O. The Semantic Web.Scien-tific American(May 2001).

3. Bernstein, A. How can cooperative work tools support dynamic groupprocesses? Bridging the specificity frontier. InProceedings of the Com-

puter Supported Cooperative Work Conference. ACM Press, 2000.

4. Gruber, T. R. A translation approach to portable ontologies.Knowledge Acquisition 5 2 (1993), 199–220.5. Mobicom Challenges 1999/2000/2001. In Proceedings of the ACM Con-

ference on Mobile Computing and Communications (Mobicom), 1999,2000, 2001.

6. Reaching for Weiser’s Vision.IEEE Pervasive Computing Magazine 1(Jan–Mar. 2002)

7. Suchman, L.A.Plans and Situated Actions: The Problem of Human-Machine Communication.Cambridge University Press, Cambridge, UK, 1987.

8. Weiser, M. The computer for the twenty-first century.Scientific Ameri-can (Sept. 1991).

9. Zuboff, S. In the Age of the Smart Machine: The Future of Work and Power.Basic Books, New York, 1988.

Guruduth Banavar ([email protected]) is a research staff member manager at the IBM T.J. Watson Research Center inHawthorne, NY.Abraham Bernstein ([email protected]) is an assistant professor atNew York University.

This article expands on the work of a working group, which also in cluded: Mark Ack-erman (University of Michigan), Mark Adler (Nokia Research), Prasert Kanawat-tanachai, Kalle Lyytinnen, Sean McGann, Frank Merat, and Lin Zhao (all from Case

Western Reserve University).

Permission to make digital or hard copies of all or part of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full cita-tion on the first page. To copy otherwise, to republish, to post on servers or to redis-t ib t t li t i i ifi i i d/ f

c

Documents

cacm-ubicomp