A Diary Study of Rendezvousing: Implications forPosition-Aware Computi - PDF document

A Diary Study of Rendezvousing: Implications forPosition-Aware Computing and Communications for theGeneral PublicMartin ColbertKingston UniversityPenrhyn Road, Kingston-upon-Thames, Surrey, United Kingdom, KT1 2EE+44 20 8547 2000m.colbert@kingston.ac.ukThis paper presents a diary study of rendezvousing as performedby university students. The study suggests that rendezvousingfrequently does not occur exactly as planned, but this is notnecessarily problematic. It also reveals that ‘problem’ rendezvouswere attributed more frequently to modes of travel, over-runningof previous activities and lack of information about otherrendezvousers, than to lack of information about travel, or localgeography. These, and other, findings have implications for thedesign of position-aware computing and communications for thegeneral public.Categories and Subject DescriptorsH.1.2 [Models and Principles]: User/Machine Systems – humanH.4.3 [Information Systems Applications]: CommunicationsApplications.H.5.3 [Information Interfaces and Presentation]: Group andOrganization Interfaces – computer supported collaborative work,evaluation methodology, theory and models.General TermsDesign, Human Factors.KeywordsRendezvousing, performance, position-awareness, mobile,personal computing and communications, diary study.1. POSITION-AWARE COMPUTING ANDCOMMUNICATIONS FOR THE GENERALPUBLICCurrently, position-awareness for the general public ispredominantly provided by the satellite-based GPS (GlobalPositioning System). A GPS unit is about the same size andweight as a mobile telephone, is accurate to approximately 10m2,and displays current position in terms of longitude and latitude,and grid references (see www.garmin.com). Some units alsoprovide altitude data, and “derived” data, such as speed, headingand relation to predefined routes. Some units superimpose currentposition on electronic map displays. If a GPS is connected to alaptop computer loaded with appropriate software, currentposition data may be integrated with more detailed, interactivemaps and route planning (see www.microsoft.com/catalog). Otherpositioning systems based upon wireless network infrastructures,are also being introduced. However, these systems are currentlyless accurate than GPS, and their geographical coverage is far lessthan global.Current mobile computing and communications for the generalpublic essentially comprise personal organisers and mobiletelephones. The former support basic applications, such asnotepads, calendars, address books etc., accessed via a reduced-size keyboard and screen. The latter comprise real-time, two-wayaudio services and SMS (Short Messaging Service) delivered overlow bandwidth, wireless networks, such as GSM (Global Systemfor Mobiles) delivering 9.6kbps (kilobits per second). Somemobile phones deliver additional services, such as Internet access(via WAP (Wireless Access Protocol)), and may be connected tohigher bandwidth networks, such as GPRS (General Packet RadioServices) delivering 56-115kbps.Permission to make digital or hard copies of all or part of this workfor personal or classroom use is granted without fee provided thatcopies are not made or distributed for profit or commercialadvantage and that copies bear this notice and the full citation onthe first page. To copy otherwise, or republish, to post on serversor to redistribute to lists, requires prior specific permission and/or afee., Sept. 30-Oct. 3, 2001, Boulder, Colorado, USACopyright 2001 ACM 1-58113-294-8/01/0009…$5.00.. Forth-coming generations of position-aware mobile devices for thegeneral public are likely to be far smaller and lighter, and moreclosely integrated. At the time of writing, a GPS phone for GSMhas recently been launched in Europe (www.benefon.com). GPScapabilities may be embedded in a range of forth-coming mobiledevices (Smart Phones, Communicators, Personal DigitalAssistants etc.) connected to higher capacity wireless networks,such as UMTS (Universal Mobile Telephone System) deliveringup to 2mbps.The press speculates about the services that the technology of thenear future will support:“Companies will be able to keep track of their sales peopleand engineers, and see how long they spend with specificcustomers … and be able to analyse who has dealt with themost customers and been the most efficient”“A restaurant could identify people passing by who areknown to like the kind of food it serves. The restaurantcould then ‘push’ a message perhaps advertising aparticular menu to that person’s mobile handset as theyare passing”Computer Weekly, June 2000, p.36Such speculation suggests that it is not easy to design services thatlead to real performance benefits for users. To achieve suchbenefits, we need, as always, user-centred design that attends toperformance, from the outset, and that uses performance to guidethe iterative design of the system as a whole [10]. Performance,here, is the trade off between quality of outcomes achieved forcosts incurred.2. AIMSThis paper reports an empirical study of a task, and its currentlevel of performance, that is of relevance to the design of position-aware, mobile computing and communications for the generalpublic. By reporting the study, the paper hopes to provokefurther research and development work in the area. The taskstudied is rendezvousing - i.e. meeting family, friends or teammates at an agreed time and place.Rendezvousing was selected for study for a number of reasons:? it is implied by many aspects of home and work life, fromkeeping the whole family happy on holiday, to engineeringfieldwork [9];? rendezvousing is performed by a group;? rendezvousing inherently involves user mobility. Studies ofmobile tasks will enable us to develop groupware that enhances“being mobile”, rather than enabling “tele-work” (increasingaccess to remote resources and group members, so that it is nolonger necessary to be mobile).? rendezvousing is relevant to a wide range of position-awareservices for mobile devices. Some of these services meet thecomputing needs of individual users (e.g. alerting, navigation,information seeking). Others support individuals as members ofa group (eg communication).Rendezvousing per se has yet to be studied as a subject for user-centred research. The most closely related work are ergonomic‘transit’ studies of user movement within interior spaces [5], andtime-geographic studies of daily routines [3]. Recently, Grinterconsidered the geography of coordination [6]. ConversationAnalysis has also studied the content of group communications,and sometimes this content concerned rendezvousing [4]. Ofparticular interest, are studies that show how personal technologyblurs traditional distinctions between technology for theworkplace and technology for the home. For example, Palen et al.studied the evolving communicative practices of new, mobilephone users [12]. Phones initially envisaged to be for work use,were increasingly used for non-work activity.The paper focuses upon rendezvousing performance to respond tothe suggestion that performance does not figure in user-centreddesign practice as prominently as might be expected, given itsaccepted fundamental importance [10]. Although some empiricalevaluations of prototype systems do address overall taskperformance [1], much work has focussed upon novel userinteractions for performing familiar on-line sub-tasks with mobiledevices, for example, manual controls for accessing structuredaudio [14] or semi-transparent widgets for accessing structuredtext [7].This paper comprises three sections. In the first section,rendezvousing goals, behaviour and performance are brieflycharacterised. In the second section, a diary study of 34university students is presented. Empirical findings aboutrendezvousing goals, behaviour and performance are reported.The study used a diary method, because the events of interest aretoo rare, private and geographically dispersed for directobservation, and users too easily forget important details of theirbehaviour to report them accurately long after the event. Thediary method used here also generated, not only qualitative dataabout the nature of rendezvousing, but also quantitative data aboutselected features of users, the task and performance. A diarymethod was recently used to study potential applications offuture media capture technology [2] and is generally useful in user-centred development [13]. In the third section, the implications ofthe findings for position-aware computing and communications,and some future research topics, are discussed.3. RENDEZVOUSING Rendezvousing GoalsRendezvousing refers to the informal, geographical co-ordinationof small groups of friends, family and team mates. The purposeof a rendezvous is for individuals to come together to participatein a subsequent group activity. They may be conducted in thecontext of home or work life. Example rendezvous include“meeting a friend for lunch”, “collecting the kids from school” and “pausing at an intermediate way-point to re-stock and plan thenext phase of activity”. Rendezvousers have personalrelationships – they are not impersonal embodiments oforganisational roles. Consequently, rendezvous do not include:formal, or anonymous attendance at institutions, such as“reporting to the Tax Office for interview” and “going to myElectronics lecture”; business fora, such as Annual GeneralMeetings; or receipts of service, such as “Pizza delivery”.Rendezvous are also physical encounters, and not virtual ones,such as international telephone calls, and internet chats.3.2 Rendezvousing BehaviourThe following scenarios based upon diary entries (see later)illustrate rendezvousing behaviour. Rashid and his girlfriend, Urshana, have agreed to meet in theevening to talk about coursework and other things. Thatafternoon, Rashid happens to be passing close to Urshana’shouse, and decides to drop in. As he thought, Urshana is athome, so they talk about the coursework, and make otherarrangements for the evening.? Senga and Jim are old friends. As first year undergraduates,they used to work together, but now they are taking differentcourses at different sites. It is Jim’s first visit to Senga’s newsite, and Senga is due to collect him from the train station. Sengais a bit early, and there is no information about the arrival ofJim’s train, so she makes a short visit to the shops, rather thanwait around. Unfortunately, Senga gets caught in a check-outqueue. When she returns to the station, Jim is waiting at theentrance. Luckily, it is the entrance that Senga uses.? Sue leaves her husband Bob a telephone message, asking himto collect their daughter Rita from school, because Sue’sworkshop is late starting, so Sue is no longer sure to make thecollection. Sue emerges from the workshop to find that Bobhasn’t confirmed the new arrangement. There is no answer fromBob’s mobile (he’s driving), so Sue rushes to the school. Shearrives just in time only to find Bob at the gates.? David has been working all morning in the library to meet animminent deadline. Rather than taking the lift directly out of thebuilding, he walks around the library floors, hoping to find oneof his mates to have a coffee.The diaries suggested that rendezvousing exhibits thefollowing structure. Having jointly negotiated and agreedthe plan to meet, and assuming the meeting is recalled, eachrendezvouser journeys to the rendezvous point (ifnecessary), where a group activity then occurs. During therendezvous, a number of ‘enabling’ tasks may be performedto ensure that rendezvousing is performed well. Thesetasks are: re-planning the rendezvous, seeking information(about travel, local geography, other rendezvousers, therendezvous plan or the subsequent group activity); andcommunicating with other rendezvousers. These tasks areoptional i.e. performed as required, possibly repeatedly,and, typically at present, over mobile phones.Rendezvousing tasks are also often interleaved with othertasks of home and work life in order to ‘make the most ofone’s time’.3.3 Rendezvousing Performance“The people who are late are often so much jollier than the peoplewho have to wait for them”E.V. Lucas"The trouble with being punctual is that nobody's there toappreciate it."Franklin P. Jones3.3.1 OutcomesFor this study, rendezvous outcomes comprise the additionalstress and lost opportunity associated with attempts to meet at anagreed time and place.Figure 1. Rendezvousing behaviourJamie’s HouseSue’s Law CourtBob’s Work ( 23 Ken goes to work and Babs goes to the University.Ken drops off their daughter Rita at her primaryschool on the way.Ken goes to the Law Courts to pay for hismotoring offence.1. Babs meets her mates to discuss the group coursework and have lunch. 2. Babs collects Rita and Rita’s best friend Jamie from school, and they all go to Jamie’s house. The kids play and the mum’s talk.3. Ken works a bit late, and collects Babs and Rita Rendezvous outcomes do not comprise ‘success at realising theplan as initially agreed’, because failure to meet as agreed is notnecessarily problematic (see first scenario). Similarly, meeting asagreed does not necessarily imply unproblematic behaviour (seethird scenario). The problematic aspects of rendezvousing are:? stress - worrying about how well a group will meet up, andthe consequences of this; and? lost opportunity – what rendezvousers would otherwise havedone, had they met up well.Physical discomfort, such as getting tired, cold, wet or hungry, isalso problematic, but information was not specifically requested inthis study. The diaries suggested physical discomfort does occur,but rarely.The following kinds of lost opportunity were sufficient tocharacterise the events reported in this study.? re-structuring – the group activity (the purpose of therendezvous) is conducted, but the tasks that comprise it are re-structured, re-sequenced, re-located or re-conceived.? delay – the group activity proceeds, but the start time of thewhole activity is later than planned.? reduced participation – the group activity proceeds, but one,more or all individuals perform less than originally planned,because they join in late.? non-participation – the group activity proceeds, but one ormore of the planned participants does not attend i.e. does notperform any tasks in the group activity.? non-occurrence – the group activity does not proceed, and isabandoned. individual sacrifice – an individual may re-structure, delay,cut-short or abandon a task from their own life, in order to meetas agreed.All categories (apart from non-occurrence) are mutually inclusive.For example, a rendezvous associated with a delay, may also beassociated with a restructuring, reduced participation, non-participation, and individual sacrifices.Rendezvous outcomes comprise stress and lost opportunity, because daily life inevitably involves a certainamount of each. For example, the mere act of making a journeymay be stressful, if you are prone to wondering whether the trainis about to crash, and inherently loses the opportunity to not travel and do something else.3.3.2 CostsFor this study, the ‘costs’ of rendezvousing comprised the effortand dissatisfaction incurred by enabling tasks i.e. recalling,navigating, seeking information and communicating. ‘Costs’ areexpressed in these, somewhat general terms, to accommodate thediverse implications of a wide range of tasks in a limited number ofconcepts. This helps to compare and contrast different types ofposition-aware service. Relevant implications include: the“nuisance” of putting the baby down again to answer the phone,only to be told that a visitor is going to be trivially late; the“frustration” of attempting to give directions to somebody who islost but cannot describe where they are; or the “danger” of talkingon the telephone whilst driving. They also include the “pleasure”of seeing the dot that represents yourself pass the symbol for‘bridge’ on the map, as you cross the bridge in real life, or the“social value” of knowing you are special enough for someone elseto confirm your attendance. Future studies may wish to isolatedifferent aspects of cost and address them separately.4. THE DIARY STUDY4.1 Method ParticipantsThe participants in this study were 17 male and 17 femalestudents from the School of Computing and Information Systems,Kingston University.All second and final year undergraduates registered for Human-Computer Interaction modules, and post-graduates takingconversion courses, completed a diary as part of courseworkexercises. The diary keeping procedure was identical for allstudents.The aim of selecting participants was to obtain a sample that wasbalanced in terms of sex, and also large, despite the fact that thevast majority of potential participants were male. Drawing on allmodules, 17 female students completed a diary (10 second years,2 final years and 5 post-graduates) and consented to itsanonymous use here. The diaries of 17 male students were thenselected from 43 final year undergraduates, to match the femaleparticipants as closely as possible in terms of age, marital status,number of children, and mobile phone ownership. Final yearstudents were thought to have produced diaries that were easier tointerpret than students on other modules. Of those eligible toparticipate in this study i.e. females on all modules and final yearmales, 13 individuals completed a diary but did not consent to itsanonymous use here, and 10 did not attach a consent form to theircoursework. This represents a drop out rate of 38% - somewhathigh, but to be expected given the sensitivity of the information,and the fact that diary completion was compulsory.Participants had a mean age of 24 years 6 months. 53% weremarried, or with the same partner for more than one year, and 47%were single. 15% had one or more children, and 44.5% did morethan 10 hours per week paid work in addition to their Universitystudies. Most participants owned a mobile telephone (79.5%), allhad access to a fixed line telephone (100%), and almost all had aprivate e-mail account in addition to their University account(88%). Where applicable, 74% used their mobile phone more than10 times per week, and 67% used their fixed phone more than 10times per week. All participants lived within commuting distanceof Kingston-upon-Thames, a suburb of south west London, UK,and 76% of participants are non-White in ethnic origin. All participants had access to hard copy maps and route-planningweb sites. No participant reported use of a GPS unit.This population was convenient for data collection, and is ofparticular interest, because of the number of early adopters andfrequent users it includes. It also approximates to ‘youthful,urban knowledge workers’. However, the population is differentto the general public, in terms of potentially relevantcharacteristics, such as age, education, and occupation. That said,the importance of such differences for rendezvousing remains tobe established. Alas, user-centred studies are often difficult togeneralise. Daily LifeParticipants kept diaries about their own rendezvousing behaviourfor 14 days during the month of February, 2000. This period isearly in Kingston University’s Spring semester, and coincidedwith the one-week, ‘half-term’ break of many local schools(important for participants with children). As such, this sampleof daily life approximates to a ‘term-time routine’. However, thissample of daily life differs from a full calendar year, because itdoes not include vacations, or exam-time. Despite the increasingadoption of flexible work practices, the pattern of student life isperhaps more unstructured than that associated with other kindsof knowledge-intensive work. Again, the importance of suchdifferences for rendezvousing remains to be established, but theresults of this study will be difficult to generalise. ProcedurePrior to the main study, the author kept his own rendezvous diaryfor an eight week period.At the outset of the study, all students were given an overview offuture position-aware, computing and communications for mobileusers and introduced to the aims of the study, and the obligationsof diary keeping. To illustrate the kind of services that could bedeveloped, participants examined existing Web sites (with fixedaccess) that provide map, transport and venue information. Apossible future service was also described, which enabled eachmember of a small group to display on their mobile telephone thepositions of other group members, superimposed upon anannotated map (see Figure 2).One diary entry was made for each rendezvous event. Each entrycomprised: () an open-ended, narrative description in theparticipant’s own words of what happened, and why; and (ii) theparticipant’s responses to a rendezvous questionnaire, whichasked for specific details of each event. Incomplete or ambiguousentries were not included in subsequent analysis.(i) select group member (ii) obtain positionFigure 2. A possible future position-aware, communicationserviceAt the end of the diary-keeping period, participants alsocompleted a further form, which summarised their diary and itscompleteness. Findings RendezvousingOver the 14-day period, participants reported a total of 415rendezvous - an average of 6 rendezvous per week. The averagesize of rendezvous (including the diary-keeper) was 2.9 people.Rendezvous involved 2 people (including the diary-keeper) on60% of occasions, 3 people on 20% of occasions, 4 or 5 on 11%of occasions, and 6 or more on 9% of occasions.55% of rendezvous included close friends, 31% included theimmediate family (partners, children, brothers or sisters), 12%included acquaintances and 11% included the extended family(aunts and uncles, cousins, in-laws). Only 1% included strangers.10% of rendezvous involved more than one category of ‘otherrendezvouser’.The diary-keeper had rendezvoused before with the group on 95%of occasions. The diary-keeper was familiar with the area of therendezvous on 80% of occasions, knew landmarks on 8%, and wasunfamiliar with the area on 12%. The diary-keeper hadrendezvoused at the same location before on 78% of occasions.4.2.2 OutcomesOn 216 occasions (52%), the rendezvous took place successfullyi.e. at the time and place agreed before the first rendezvouserdeparted for the rendezvous point). On the other 199 occasions(48%), at least one individual failed to arrive as initially agreed.Of the total 415 rendezvous reported, 14 (3%) were tooincomplete or ambiguous to be included in subsequent analysis.Of the 401 remaining, 176 (44%) were found to be problematic(led to stress or lost opportunity), and 225 (56%) were notproblematic. Of the 176 problematic rendezvous, all (100%) led to stress. Themean stress rating on a scale from 1(low) to 5(high) was 2.83.24% of problematic rendezvous led to stress rated at levels 4 or 5.Table1. Number of problematic rendezvousTotal No.Rendezvous 401176 (44%) Of the 176 problematic rendezvous, 125 (71%) led to a lostopportunities. The mean overall value of these lost opportunitiesin social and financial terms and rated on a scale from 1(low) to5(high) was 2.48. 14% of problematic rendezvous led to lostopportunity rated at levels 4 or 5.Lost opportunities frequently took the form of delay (50%) andreduced participation (40%). Other types of lost opportunitywere reported less frequently – individual sacrifices (11%), non-participation (10%), re-structuring (6%) and non-occurrence (4%).Table 2. Number of different types of problemNumberwith StressMeanwith LostOpportunityMean LostOpportunity 2.831252.48 Table 3. Distribution of stress and lost opportunity ratings(%)2345 3329177 251995 CostsEffort associated with enabling tasks was indicated by the numberof times communications of all types were successfully performedduring the rendezvous.Rendezvous were conducted without any communication onapproximately one third of occasions (30%). A singlecommunication occurred on 44% of occasions, twocommunications on 18%, three communications on 5%, and morethan three communications on 3%.Dissatisfaction associated with enabling tasks was measured interms of the diary-keeper’s activity that was interrupted bycommunication.Communication did not disrupt any activity on 27% of occasions- the diary keeper may have been waiting at the rendezvous site,or relaxing in front of the television. Communication disrupted ajourney (driving, sitting on a train or bus, cycling, walking) on25% of occasions, and a standing activity (doing the laundry,browsing in a shop) also on 25%. Communication disruptedindividual seated activity, such as studying or programming, on11% of occasions, and disrupted group activity, such as playingwith the kids, or talking over dinner, on 12%.Information about unsuccessful attempts to initiatecommunication (eg no answer, beyond network coverage) andactive steps to regulate communication (eg turning the phone off)was not specifically requested. The diaries confirm that theseevents happen sufficiently often to be of interest.4.2.4 Causes of Problematic RendezvousEight possible causes of problematic rendezvous were identifiedfrom the diaries. The types of cause identified are related to thedecomposition of rendezvousing outlined in section 2.? mode of travel. Transportation is less adequate than expectede.g. trains are delayed or cancelled, traffic is heavy, cars do notstart; over-running/knock-on effects. Previous, planned activitiesdo not finish on time, and have adverse consequences e.g. adental appointment may take longer than expected, if urgentwork is discovered.? poor planning. The original plan for the rendezvous isincomplete, inaccurate, never agreed or forgotten e.g. arendezvous at a cinema states a time after the film actuallystarts, or does not state which entrance;? lack of travel information. Rendezvousers are not aware of,or misunderstand, service routes or timetables;? lack of geographical information. Rendezvousers becomedisorientated in unfamiliar surroundings, or fail to find themeeting place where they expect it;? lack of information about other Rendezvousers are unaware how others are following the plane.g. some rendezvousers decide not to come after all, or arewaiting in a different place.? spontaneous additional tasks. Rendezvousers take theopportunity to do something else. For example, on learning thatone flat-mate is about to drive into town, another flat-mate asksfor a lift;? success not valued. Rendezvousers place insufficient valueon meeting as agreed. For example, an individual might be anotoriously bad time-keeper, or contrive to be late, so that theirarrival is more noticeable.These categories are not mutually exclusive. For example,problems in the third scenario in section 3.2 are attributable toboth over-running of previous activities (Sue’s workshop), and tolack of information about other rendezvousers (Bob’s actualmovements). Diary-keepers attributed rendezvous problems to the modeof travel on 37% of occasions, to the over-run of previousactivities on 25%, and to lack of information about otherrendezvousers on 21%. Problems are attributed to lack oftravel information on 7% of occasions, to lack of geographicinformation on 10%, the plan on 8%, the performance ofadditional tasks on 12%, and the failure to value success on11%.Table 4. Causes of rendezvous problems (number ofattributions) Mode of Tavel Overrun Previous Actvity Info About Others Travel Info Geographic Info Additional Tasks Poor Planning Success not Valued 37252171012811 N.B. These categories are not mutually exclusive.5. DISCUSSION Design ImplicationsA single study, particularly one based upon a single method and inwhich users only participate as ‘subjects’, is no basis forsuggesting user requirements. However, the findings of this studydo have implications for the design of position-aware computingand communications.5.1.1 Is There a Performance Deficit?The presence or absence of a performance deficit is starting pointfor many studies of user requirements.With respect to outcomes, the diaries suggest that rendezvousingfrequently does not occur exactly as planned (48% of occasions).However, this ‘failure’ is not necessarily problematic. Meanratings for stress and lost opportunity are only medium (2.83 and2.48 respectively), the problem rendezvous sometimes involveonly stress and no lost opportunity (29%), and the lostopportunity is very often of a “less severe” kind (only 10% led tonon-participation, 4% to non-occurrence and 11% to unnecessaryindividual sacrifice). The effort required to achieve this level ofoutcome also seems reasonable - 74% of rendezvous wereexecuted with only one communication, or none at all. Theoutcomes of a few rendezvous, however, are extremely poor (24%of problematic rendezvous were rated 4 or 5 for stress, and 14%lost an opportunity rated at 4 or 5)(see the third scenario insection 3.2).In the author’s view, these ‘extremely poor’ ratings occur toofrequently for current mobile devices to be wholly satisfactory.That said, it is worth recalling the alleged “bloat” of desk-topsoftware for the general public in the 1990’s - a condition in whichthe usability of core functions, frequently used by most users,became impaired (allegedly) by the addition of further functions,rarely-used by only a few users (see [8]). Although there isevidence that the adverse effects of “bloat” may be handled by anappropriate trade-off between breadth and depth in the userinterface [11], current mobile devices are so popular that it will beeasy for future ‘enhancements’ to disappoint.With respect to user costs, the majority of communications duringthe rendezvous were potentially disruptive – 25% were made orreceived in transit, 25% when doing something standing up, and11% when engaged in a group activity. However, it must be notedthat the measures of cost applied under-estimated the actual rangeof relevant daily activities and information services. For example,the measure of ‘diary-keeper activity’ was selected, to indicate asingle aspect of dissatisfaction (disruption). However, the diariessuggested that many aspects of dissatisfaction are, in fact,important to students (see Section 3.3.2). Further, relationship between categories of ‘diary keeper activity’ anddissatisfaction was confounded by actual student life and the rangeof communication channels actually employed. The category ‘noactivity’ was intended to indicate ‘not-very-dissatisfying’communication. However, this category included telephone callsto students who had not arrived at the rendezvous, because theywere still asleep in bed (highly dissatisfying communication, Iwould have thought!). The category ‘seated activity’ wasintended to indicate disruption of work (and so dissatisfaction),but in this context, a communication by mobile phone was oftenmuch more disruptive than an equivalent communication by e-mail. Diverse, social implications of the form of communicationwere also at work, but were not captured. For example, in somecontexts, some students would go to great effort to ensure theyused a particular channel of communication e.g. to use thetelephone, rather than e-mail, if they were going to stand up adate. Others failed to communicate, or communicated in aninappropriate way, and were disapproved of, accordingly. Futurestudies may wish to use more general indicators of effort anddissatisfaction, such as overall subjective ratings, and requestspecific information about the reasons for dissatisfaction. Theymay also wish to prepare for diary studies with interviews, aswell as pilot diaries.Less-disruptive means of communication during rendezvous maybe desirable, but existing channels of communication (and otherpotentially relevant services, such as maps etc.) have relativestrengths and weaknesses for all rendezvousers for which thisstudy did not fully accommodate.5.1.2 What Kind of Position-Aware Service isLikely to be Most Frequently Useful?The possibility of a deficit in current performance begs thequestion, “What kind of position-aware service is likely to bemost frequently useful?” Many kinds of position-aware service are possible, but let usconsider position-aware communication, position-awarenavigation and position-aware information-seeking. Position-aware communications (including services that communicateposition), for example, enable users to send their current positionas co-ordinates, vectors, gazetteer entries or maps etc. to otherrendezvousers, either as distinct messages or e-mail attachments.Alternatively, a user may permit other rendezvousers (or a sub-setof the whole group) to remotely access information about his orher current, or last known position, speed and direction, with orwithout explicit authorisation at the time (so-called “Friend-Finder” features). Such services have the potential to enhanceperformance by supporting the exchange of relevant, personalinformation.Position-aware navigation services display, for example, a user’scurrent position with respect to planned routes or way-points,and alert users to unintended deviations. The web sites of popularrendezvous points may also provide their exact co-ordinates tomobile navigators. Such services have the potential to enhanceperformance by preventing users from becoming disoriented –users make fewer ‘map-reading’ errors, because they are more ableto relate what they see around them to what appears on a map.Position-aware information seeking services, for example, suggestpossible rendezvous sites, given rendezvousers’ current positions.They also estimate journey times between current positions andrendezvous points, taking current transport conditions intoaccount (www.connexwap.net/cx.wml). Such services have thepotential to enhance performance by helping rendezvousers tomake good plans, and to relate their behaviour to these plansappropriately, in the light of unfolding events.The causes of rendezvous problems are diverse - all eight causeswere cited for more than 5% of problems. Also, the likelihood ofa particular cause actually leading to a problem is relatively low -only one cause (mode of travel) was cited for more than 25% ofproblems. Consequently, only position-aware communicationseems sufficiently flexible to be frequently useful, at least for thepopulation studied here. For problems attributed to the mode oftravel (37%), the communication of position information may, forexample, enable rendezvouser ‘A’ to infer that rendezvouser ‘B’ isen route. For problems attributed to overrunning of previousactivities (25%), position information may, for example, enablerendezvouser ‘A’ may infer that rendezvouser ‘B’ is still at aprevious engagement. For problems attributed to additional tasks(12%), position information may enable rendezvouser ‘A’ to inferthat rendezvouser ‘B’ has, for example, made a quick visit to theshops, rather than wait around. For problems attributed to lack ofpersonal information (21%), position information may enablerendezvouser ‘A’ to infer that rendezvouser ‘B’, for example, hasreceived an earlier message, or has overslept.In contrast, position-aware navigation and information-seekingmay provide information that is likely to be relevant lessfrequently. Lack of geographical information and travelinformation is only cited as a cause of rendezvous problems on7% and 11% of occasions respectively. Poor planning is cited on5% of occasions. Such low frequencies are to be expected,perhaps, given participants’ knowledge of the locality (see later)).Mobile devices already comprise various means of communication(telephony, voice mail, SMS, even e-mail), so it is perhapssurprising that the additional service likely to be most frequentlyuseful is yet another channel of communication. It might beinteresting to re-examine the diaries to see how many rendezvousproblems could have been avoided, had certain services beenavailable, or to see why existing channels of communication werenot used, or insufficient to successfully exchange the relevantinformation. Friend Finding – but that’s Spying, isn’t it?Position-aware communication is intriguing, because it suggeststhe kind of service (remote access to sensitive information aboutanother individual) that appears to be an obvious invasion ofprivacy, even though rendezvousers are likely to have closerelationships (86% involved immediate family or close friends)and need to share position information within only a small group(80% involved 2 or 3 individuals). Unfortunately, privacy is noteasily addressed using a diary method, because the participant isasked a hypothetical question about how they would react weretheir position information obtained, albeit contextualised in anactual rendezvous. Our diary asked participants to state, for eachrendezvous they reported, whether they would have been willingfor at least one other rendezvouser to obtain their current positionto make the rendezvous more successful. It also askedparticipants to estimate the number of other individuals withwhom they would have long-standing agreements to obtainposition information, or to permit their position to be obtained.Responses to these questions were so positive as to bring theiraccuracy into doubt - many students had ten or more others ontheir ‘positioning’ lists. Perhaps there was an experimenter effect,or participants had different or unstated expectations about thepossible, future communications service. For these reasons, thedata are not reported here. Further work may address privacydirectly, for example, by interviews following long-term use ofprototypes.The findings we have suggest that position information alone islikely to be informative, at least for the population studied here.For example, if rendezvouser ‘A’ learns that rendezvouser ‘B’ iscurrently located in a particular street, to avoid rendezvousproblems, ‘A’ needs to infer that ‘B’ has been delayed by an over-running activity (‘A’ knows that ‘B’ habitually takes classes thatare held in this street at about this time). ‘A’ also needs to knowapproximately how long it will take ‘B’ to get to the rendezvouspoint, given the current position. Our participants are likely to beable to make such inferences, because of their knowledge of thelocality (80% are familiar, and 78% have rendezvoused at thatpoint before), and because of their knowledge of the other rendezvousers (on 95% of occasions the participant hadrendezvoused with the group before).Further work is required to discover whether position informationis additional (not available with current communication tools) andsufficient (resolves the problem without any further informationor discussion).5.2 Topics for Research5.2.1 Problematising Outcomes and ExplainingDissatisfactionThe practical difficulties encountered by this study, such as howto conceive of ‘user costs’ and exactly what questions to includein the diary questionnaire, suggests that future research needs toaddress a couple of issues, specifically, the problematisation ofoutcomes and the explanation of dissatisfaction. This studyreported how rendezvous’ became problematic (because the trafficdelayed some en route, because of knock-on effects when previousactivities overran etc). However, it did not explain why being 10minutes late was sometimes rated as extremely problematic andsometimes not at all problematic. Problematisation may reflect:(I) temporal or participatory constraints of the group activity,that is, hard or soft (precise or approximate) start-times, durationsor end-times; (II) the kind of group activity that follows therendezvous, such as discourse (eg talking, eating and drinking), anobjective task (watching a film, playing sports), parental duties(collecting the kids), or logistic tasks (giving someone a lift); and(III) the qualities of relationships between rendezvousers thatexist or are being constructed (status, intimacy etc).Similarly, dissatisfaction with alternative rendezvousing servicesneed to be explained, for example, with reference to the cognitiveabilities of different types of users, and (III) above.5.2.2 Effects of Position-Aware Services UponRendezvous PerformanceIt will be important for user-centred research to document andexplain any effect of position-aware services upon rendezvousingperformance. New technology does not necessarily make lifebetter, so we ought to check to see what its effect was (if any).Moreover, there are alternative fundamental views of userbehaviour, and these views make different predictions about theeffects of position-aware services, so there may be an opportunityto test out the views. For example, if users are maximising thesubjective expected utility of alternative rendezvousing services(utililitarian views of user behaviour), then position-awarecommunication may help them to reduce or avoid the problematicconsequences of failing to arrive at the time and place initiallyagreed. In terms of the third scenario in Section 3.2, if Sue is ableto see that Bob is on his way to collect the kids, Sue may not getso stressed when her meeting overruns, or leave in such a hurry.A utililtarian view, then, predicts that the proportion ofrendezvous that do not occur as planned but are nevertheless not aproblem will increase, and those which do not occur as plannedand which are a problem will decrease. Alternatively, if users aresimply maintaining an acceptable level of risk of ‘havingproblematic rendezvous’ (risk homeostasis views of userbehaviour), then users may compensate for the reduction in riskperceived to result from position-aware communication, byincreasing the risk perceived to result from other behaviour, suchas spontaneously performing additional tasks. In terms of thesecond scenario in Section 2.3., Senga is more likely to ‘pop’ tothe shop, to compensate for her perception that Jim is more ableto find her. In this case, the proportion of rendezvous that did notoccur as planned and were problematic would be expected toremain unchanged, or even increase.Current performance, then, is a baseline reading - to be comparedwith levels of performance achieved in the future, and perhapswith respect to desired or predicted values.6. CLOSING REMARKSThis paper has attended to the performance of a user task that isrelevant to the design of position-aware, mobile computing andcommunications, and which has not previously been the subject ofuser-centred research. It has also noted opportunities for futureresearch studies, and service development. We are currentlydeveloping a prototype service for positioning members of a smallgroup within an area covered by a wireless LAN, in order to studyprivacy issues more satisfactorily.It is appropriate, however, to close on a cautionary note – thatcurrent GPS units do not have the accuracy or coverage to assistall of the problematic rendezvous reported in this study.7. ACKNOWLEDGEMENTSGreat thanks are due to Francis Djabri and Timo Bruns ofSymbian, UK for their support and encouragement, and toanonymous referees for their critical comments.8. REFERENCES Baber, C., Arvanitis, T. N., Haniff, D. J. and Buckley, R.A Wearable Computer for Paramedics: Studies inModel-based, User-Centred and Industrial Design. inProceedings of INTERACT’99 (Edinburgh UK,September 1999), IOS Press, 126-132.[2] Brown, B.A.T., Sellen, A.J. & O’Hara, K.P. A DiaryStudy of Information Capture in Working Life: Glimpsesof the Future. in Proceedings of CHI’2000 (The HagueNetherlands, April 2000), ACM Press, 438-445.[3] Carlstein, T., Parkes, D & Thrift, N. Human Activity andTime Geography: Timing Space and Spacing Time (vol2). Edward Arnold, London, 1978.[4] Frohlich, D.MChilton, K. & Drew, P. Remotehomeplace communication: what is it like and howmight we support it? In People and Computers XII,Proceedings of HCI’97 (Bristol UK, August 1997),Springer-Verlag, London, 133-153. [5] Grandjean, E. The Ergonomics of the Home, TaylorFrancis, London, 1973.[6] Grinter, R. E., Herbsleb, J. D. & Perry, D. E. TheGeography of Coordination. in Proceedings ofGROUP’99 (Phoenix AZ, November 1999), ACM Press.[7] Kamba, T., Elson, S., Harpold, T., Stamper, T. &Sukaviriya, P. Using Small Screen Space Efficiently. inProceedings of CHI’96 (Vancouver Canada, April 1996),ACM Press, 383-390.[8] Kaufman, L. & Weed, B. Too Much of a Good Thing?Identifying and Resolving Bloat in the User Interface:A CHI 98 Workshop. SIGCHI Bulletin, 30(4), 1998, 207-208. Kristoffersen, S. & Ljungberg, F. Making Place to MakeIT Work: Empirical Explorations of HCI for MobileCSCW. in Proceedings of GROUP’99 (Phoenix AZ,November 1999), ACM Press.[10] Newman, W.M. & Taylor, A., Towards a MethodologyEmploying Critical Parameters to Deliver PerformanceImprovements in Interactive Systems. in Proceedings ofINTERACT’99 (Edinburgh UK, September 1999), IOSPress, 605-612.[11] Nyberg, M., Bjork, S., Goldstein, M. & Redstrom, J.Handheld Applications Design: Merging InformationAppliances with Affecting Usability. in Proceedings ofINTERACT2001 (Tokyo Japan, July 2001), IOS Press,391-398. Palen, L., Salzman, M. & Youngs, E. Going Wireless:Behaviour & Practice of New Mobile Phone Users. inProceedings of CSCW’2000 (Philadelphia PA,December 2000), ACM Press.[13] Rieman, J. The Diary Study: a Workplace-orientedResearch Tool to Guide Laboratory Efforts. inProceedings of INTERCHI’93 (AmsterdamNetherlands, April 1993), IOS Press, 321-326.[14] Roy, D. & Schmandt, C. NewsComm: A Hand-HeldInterface for Interactive Access to Structured Audio. inProceedings of CHI’96 (Vancouver Canada, April 1996),173-180.