A Digital Photography Framework EnablingAffective Awareness in Home Communication

Olivier LiechtiInformation Systems Laboratory

1-4-1 KagamiyamaHigashi-Hiroshima 739, Japanolivier@isl.hiroshima-u.ac.jp

Tadao IchikawaInformation Systems Laboratory

1-4-1 KagamiyamaHigashi-Hiroshima 739, Japanichikawa@isl.hiroshima-u.ac.jp

ABSTRACT

By transforming the personal computer into a communication appliance, the Internet has initiated the truehome computing revolution. As a result, Computer Mediated Communication (CMC) technologies areincreasingly used in domestic settings, and are changing the way people keep in touch with their relativesand friends. This article first looks at how CMC tools are currently used in the home, and points at someof their benefits and limitations. Most of these tools support explicit interpersonal communication, byproviding a new medium for sustaining conversations. The need for tools supporting implicit interactionbetween users, in more natural and effortless ways, is then argued for. The idea of affective awareness isintroduced as a general sense of being in touch with one's family and friends. Finally, the KAN-Gframework, which enables affective awareness through the exchange of digital photographs, is described.Various components, which make the capture, distribution, observation and annotation of snapshots easyand effortless are discussed.

Keywords

Affective awareness, ambient media, awareness, calm technology, Computer Mediated Communication, digitalcamera, home photography, implicit social interaction, social role of home photography, ubiquitous computing,

World Wide Web

1 INTRODUCTION

The so-called "personal" computer has made itsapparition in the home more than two decadesago. It seems, however, that the true revolutionhas only been initiated much more recently[Venkatesh 1996]. Among other factors, it is theaddition of communication features and the widepopularization of the Internet that can explain theprofound social impact of home computing.Access to the Net has indeed transformed adevice that had no real domestic function into anappliance extensively used for informationaccess, entertainment and interpersonalcommunication. As a result, the personalcomputer is becoming more meaningful for thehousehold, used both by the entire family and forhome-related tasks (as opposed to work-relatedtasks performed at home). In this light, it mightvery well have effects as dramatic as those fromprevious technologies, such as the television orthe telephone.

Emerging communication technologies aredeeply affecting the way people "socialize", inother words the way people meet, keep in touchand interact with each other. While this is true atwork, it is also true at home. People are offered

new ways to maintain relationships with theirfamily and friends, new ways to establishcontacts with local and global communities, etc.But while the social impact of ComputerMediated Communication (CMC) technologiesis generally acknowledged, it is not clearwhether it is mostly positive or negative. Somepraise the Internet as a medium that removeseconomical, spatial and temporal constraints andthus allows people to extend their socialnetworks. Others fear that mediocre computer-based communication may lead to socialisolation and reduce well-being.

In any way, it is very important that engineersand computer scientists take into account theobservations and recommendations ofethnographers and sociologists studying theimpact of new technologies in the field.Adopting a multi-disciplinary approach shouldhelp make the best use of a very unique andpervasive communication infrastructure. As amatter of fact, the Internet is a very flexibleplatform, on top of which it is possible to build awide diversity of tools supporting socialinteraction. The challenge is to figure out whatthese tools could and should be, and how they

could and should be used by the generalpopulation. It will take time, imagination andeffort to address this challenge, and to trulyleverage the potential of global and ubiquitousnetworking.

In this article, our goal is more modestly to lookat existing CMC technologies used in the home,to point at some of their limitations, and todiscuss a number of ideas that would overcomethese limitations. While recognizing the merits ofcurrent technologies, we argue for the need toaugment the functional spectrum they cover. Weintroduce the idea of affective awareness, anddiscuss the envisioned benefits of toolssupporting lightweight, effortless and emotionalinterpersonal communication through sharedexperiences and non-verbal communication. Weexplain how such technologies require to gobeyond the traditional personal computer, andbeyond traditional human-computer interaction(seen as a one-to-one, exclusive dialogue).

An illustration of these ideas is then proposedwith the design of KAN-G, an interpersonalcommunication framework based on theexchange of digital photographs. The KAN-Gframework consists of a variety of hardware andsoftware components, which support the capture,the distribution, the observation and theannotation of photographs. The foremostmotivation for designing the framework has beenan observation of the social role played by"traditional" home photography, and a reflectionon how emerging digital photographytechnologies may affect this role. Familysnapshots are seen as very powerful artifacts thatcreate and strengthen affective and emotionallinks between people. Digital imagingtechnologies have the potential to make it easierto create, share and observe photographs, andthus to amplify the interpersonal communicationfunction of home photography. We believe,however, that achieving this goal requires thecareful design of "calm" systems [Weiser andBrown July 1996], that should be natural to use,that should be merged with architectural spaces[Junestrand and Tollmar 1998, Rijken 1999,Wisneski, et al. 1998], and that should privilegeaesthetic and artistic quality [Padula andAmanda 1999].

The remaining sections of this article areorganized as follows. In Section 2, the currentuse of Computer Mediated Communication(CMC) technologies in domestic environments isdiscussed. The functions and interaction modessupported by existing tools are described, and

some of their problems and limitations arepointed out. A number of field studies on thesocial impact of home computing are thenmentioned, and important implications for thedesign of communication technologies arehighlighted. The idea of affective awareness isfinally introduced and placed in the context ofrelated work. The discussion stresses that toolsenabling affective awareness should not replace,but augment traditional technologies such aselectronic mail.

Section 3 illustrates the previously introducedideas with the KAN-G framework, designed tosupport affective awareness throughelectronically mediated home photography. Thesocial role of "traditional" home photography isdiscussed first, with references to visualanthropology. It is then explained how a numberof different emerging digital imagingtechnologies are likely to affect and amplify thisrole. The components of the framework arereviewed, and it is explained how they supportthe capture, distribution, viewing and annotationof digital photographs. It should be noted that theimplementation of the KAN-G framework is stillin progress. Prototypes for the componentsdescribed here have been implemented, but theirintegration has not been fully completed yet. Themain objective of this article is thus not to reporton the implementation, but rather on the designof the framework. More importantly, it is to raiseattention on emerging domestic applications fordigital photography and to foster research in thisarea.

2 COMPUTER MEDIATED COMMUNICATIONIN THE HOME

Computer Mediated Communication covers abroad spectrum of technologies, from electronicmail to the WWW, from real-time video-conferencing to MUDs and MOOs. Suchtechnologies support very different functions,and enable very different interaction modes.Most of them were developed to support workrelated processes, but are now also used tosupport tasks within households. But because therequirements of work and domesticenvironments are very different, it is not clearwhether current CMC tools really fit the needs ofhome users. On the contrary, it is likely thatCMC performed in the home will raise a need oftools that do not have any equivalent in workenvironments. This is particularly the case fortechnologies that put an emphasis onentertainment and fun criteria.

In this section, we first review some of the mostpopular CMC technologies used in the home andpoint at some of their limitations. We then lookat a number of field studies on the social impactof home computing and highlight importantimplications for technology designers andcomputer scientists. We finally introduce theidea of affective awareness and review somerelated work in the Computer SupportedCooperative Work (CSCW) and Human-Computer Interaction (HCI) literature.

2.1 Current use of CMC technologies

The applications and services that have beendeveloped on top of the Internet infrastructurecan be categorized in different ways. The firstway is to distinguish between interpersonal andmass communication tools. Interpersonalcommunication tools support the direct exchangeof information between two (or a small number)of participants. They are thus electronicsurrogates to face-to-face, paper mail ortelephone communication. Mass communicationtools allow the dissemination of informationfrom one publisher to a possibly large number ofconsumers. In that sense, they are comparable totraditional media such as the television, the radioand the press. A fundamental difference withInternet technologies, however, is that thebarriers to entry are extremely low. Therefore,and for the first time, virtually everyone hasaccess to mass media as an informationpublisher.

The distinction between interpersonal and masscommunication technologies is in fact quiteloose. The WWW, for example, is generally seenas a mass communication medium. But when asite is designed to be visited by a small numberof people (e.g. a Web site published by a family,mainly visited by the members and friends ofthat family), its main role is to supportinterpersonal communication. Conversely, whenelectronic mail (primarily seen as aninterpersonal technology) is used in conjunctionwith large distribution lists, it becomes similar toa mass communication tool.

2.1.1 Electronic mail

At home like at work, electronic mail is certainlythe most common CMC technology. In domesticenvironments, it is essentially used to sustainconversations (with family, friends and peoplemet on-line) and interactions with commercialand administrative entities (e.g. with customersupport services or tax office). Several factorscan explain the wide popularity of email. First, it

is relatively easy to use and does not require along learning curve. Everybody in the family hasthus access to the medium, without anyparticular requirement of technical expertise ortime. Second, email is already ubiquitous, asalmost everybody now has an email address. Thetechnology can thus be used to interact with avery large number of people. Third, email is aninterest ing alternative to tradit ionalcommunication tools (e.g. paper mail,telephone), which among other benefits offerslow cost and immediate delivery.

One particularly interesting application for emailis exemplified by the "kids-at-college" scenario,where parents can more easily keep in touch withtheir children when they leave home. Writing anemail requires much less effort than writing aletter, buying a stamp and finding a mailbox.Writing an email is also cheaper than making along-distance phone call. These factors andothers contribute to increase the level ofcommunication between the members of afamily who do not live together. This is seen as apositive social impact of the technology, as itcontributes to strengthen social relationships.

This, however, does not mean that electronicmail is the ultimate interpersonal communicationtool and that no alternative or extension shouldbe searched for. For instance, one problem is thatkeeping up with electronic mail is often verytedious and requires a lot of time. As aconsequence, people often reduce either thefrequency (e.g. from daily to weekly) or thequality (e.g. limit to state weather information)of their messages. But these apparentlyinsignificant messages are far from beinguseless. In many cases, what is important is notso much the content of the message, but ratherthe simple fact that the message has been sent.The message, by its mere transmission, connectsthe sender to the receiver. Sending an email is insome cases very similar to waving or smiling atsomeone.

In other words, the real function of email issometimes not to support the exchange ofexplicit messages (i.e. conversations), but ratherto convey a general sense of being connected toeach other (what we will define later as affectiveawareness). But if this is the case, then othertools should be developed to support thisfunction more appropriately. The KAN-Gframework, described in Section 3, was preciselydeveloped with that idea in mind: instead ofexchanging mediocre textual messages, peoplecould maintain mutual affective awareness by

exchanging i) photographs and ii) impressions onthese photographs. This would present theadvantage of being at the same time lessconstraining (taking a snapshot takes less effortthan writing a few paragraphs) and emotionallyricher (looking at someone's smile conveys astronger impression than reading a weatherreport).

Electronic mail should of course not beabandoned, as there will always be a need toexchange factual information. But electronicmail should be augmented and used inconjunction with other tools that focus onlightweight and informal communication.

As a matter of fact, there are already examples oftechnologies which have roots in electronic mail,but which are particularly targeted at non-professional users. One example is the PostPetmail software1, tremendously popular in Japan.Although it can be used as a standard POP3client, PostPet has a unique user interface andfunctions inspired from the famous "Tamagochi"artificial creatures. The main window representsa room, in which lives a small animal (e.g. a pinkbear). When the user sends an email to a friend,it is the cute creature who gets the envelope,travels across cyberspace and enters thereceiver's living room. At this point, the twocreatures play with each other for a while, for thegreat enjoyment of the user. This applicationillustrates that an important aspect of CMCcommunication, when it performed at home, isentertainment and fun.

Another example is the electronic greeting cardservice offered by several Internet companies2.Users can choose among various categories(birthday, mother day, friendship, etc) and sendannotated drawings to their friends, who canview them with a standard Web browser.Sending an electronic postcard is very similar tosending a paper postcard. The main function isnot the communication of factual information,but the expression that the sender is thinking ofthe receiver. One limitation of electronicpostcards is that they require the receiver to use aPC to look at them. This requires too much timeand is too difficult for some people (e.g. olderpeople who have and will never use a computer).A better idea would be to have special displays(maybe on the fridge, or on a screen phone),where the postcards would automatically be

1 http://www.sony.com.sg/postpet/2 for example: http://www.bluemountain.com

shown after delivery. This would make thewhole system much easier and natural to use.

2.1.2 The World Wide Web

The World Wide Web is another very popularapplication with home users, that can be used intwo different ways. First, households can act asinformation consumers. This is the case whenkids search information for doing theirhomework, when parents do on-line shopping,and when the whole family surfs the Web forentertainment purposes. The Web has often beendescribed as a potential substitute for television.One difference, however, is that watchingtelevision can be a group activity (several peoplecan easily watch the same TV screen, share anexperience and exchange comments), as surfingthe Web is more often an individual activity. The"Let's browse" collaborative agent [Lieberman,et al. 1999] has been an attempt to make Webbrowsing a collaborative and co-located activity.

Second, households can use the WWW asinformation producers. Increasingly, home userscan design and maintain their own Web site. Thishas become both easy and cheap, with theavailability of GUI design tools and thepossibility to get free storage space at variousInternet portal sites. Very often, people take thisas an opportunity to introduce their family to theworld, to give information about their hobbiesand almost always to share a few photographs.

Sometimes, the family Web site is also used tosupport interaction around a specific event. Forexample, Web sites are often created to announceand remember a wedding celebration. Before thewedding, useful information is offered to theguests (maps, accommodation, schedule, etc).After the wedding, pictures, short comments andstories are added to the site.

When a Web site is created by a family, thepopular term "home page" seems particularlyappropriate. The Web site appears to be a digitalextension of the physical dwelling. As such, itcould become a social place where the familycould interact with visitors. This is however stillrarely the case, for several reasons. The firstreason is that it is difficult to grasp the activityoccurring within a Web site, for example to trackvisitors in real-time. The publishers of a Web sitetherefore do not know when visitors are arriving,and what they are doing in the site. Sometimesthey do not even know whether visitors arecoming at all. The second reason is that thephysical and electronic worlds are still verymuch separated from each other. As a

consequence, when people are engaged inactivities in the physical world (cooking, readingbooks, etc), they are not aware of what ishappening in the digital world. The visitors ofthe digital home are therefore not perceptible tothe inhabitants of the physical home, whichprevents much interaction between them. Wehave discussed these problems in a previousarticle [Liechti, et al. 1999], and introduced themetaphor of a window bridging the physical andcomputational worlds. When this window isopen, users hear different sounds when differentevents are detected on the site (e.g. they hear birdchirps or dog barking when particular pages areaccessed by particular people). A relatedapproach has been proposed in [Benford, et al.1996], with the idea of a virtual foyer (on theWWW) connected to the real foyer of anorganization.

Finally, it should be noted that some familiesalready have attempted to create a bridgebetween their real and virtual homes, by puttinga Webcam in their living room (i.e. a camerawhich periodically takes pictures that are thenvisible on a Web site). This however only createsa link from the real world towards the virtualworld, and not in the opposite direction. Whilethe use of Webcams is interesting, it raisesobvious privacy issues.

2.1.3 On-line communities

Another popular CMC application is the supportfor online communities. Newsgroups, multi-usersvirtual worlds (e.g. MUDs and MOOs), andmore recently WWW portal sites are some of thetechnologies that are used for this purpose.Although some are, many online communitiesare not work-oriented. They can be interest-focused (e.g. pregnant women or guitar players),but can also be bound to geographicalcommunities (e.g. foreigners living in Tokyo orresidents of the San Francisco bay area).

The first category offers a good example of howthe Internet offers new possibilities to meetpeople. They remove not only spatial andtemporal barriers, but also psychologicalbarriers. For example, shy people find it easier toinitiate interaction when they are on-line andprotected by some anonymity. Of course, onemight argue that this kind of interaction isartificial, shallow and not as gratifying as "real"interactions. The image of the asocial teenager,spending all his time in front of his computereasily comes in mind. But there are cases wherepeople really get a benefit from on-lineinteraction, as reported in [Rheingold 1993]. On-

line communities sometimes allow people toshare their problems with others and to get somecomfort. This interesting aspect has beendiscussed in [Preece 1998], with the descriptionof empathic communities. In these, participantsare primarily interested in sharing feelings andsupporting each other, as opposed to exchangingfactual information. According to the author,current CMC tools are not very well suited tosupport empathic communication. The usualinteraction mechanism is often limited to text,which lacks many channels used to conveyfeelings and compassion in face-to-facecommunication. Accordingly, there is a need fordesigning better user interfaces, with thepotential to improve the quality of life forthousands of people.

An interesting aspect of the geographically-bound communities is that they make it easier toaugment on-line interaction with face-to-faceinteraction at some later point. For example, afterhaving met interesting people in a discussionforum, it is possible (and cost-effective) to meetthem "in real". This is not as easy when peoplefrom Europe and Japan meet in a subject-focusedcommunity.

2.1.4 Internet presence and instant messaging

Finally, two related technologies that haverecently received a large echo in the press areInternet Presence (IP) and Instant Messaging(IM). Although they have existed for a very longtime (on time-sharing and UNIX systems), thesemechanisms are now used by millions of peopleon personal computers connected to the Internet.

Internet Presence makes it possible to knowwhen a group of people (often called the buddylist) are connected to the Internet, and to have anidea of their status (e.g. busy, available, etc.).Instant Messaging then allows to send shortmessages to these people. Instant messagesusually appear immediately on the receiver'sdisplay (unlike email, the messages are notstored in a mailbox).

An interesting aspect of IM is that they supportsome level of awareness between users. Awindow dynamically reflects the status of theuser's contacts, which conveys a sense ofapproachability. To some extent, it gives theimpression that one's "buddies" are right therebehind the screen, almost reachable. This isparticularly interesting for home users. Themembers of a family who do not live togethercan "feel" closer to each other, on a continuousbasis. There seems however to be room for much

development of IP and IM. For instance, statusinformation is now very limited (e.g. offline,online, available, busy, away). If IM is to be usedto create an artificial sense of "living together",then additional information could be beneficial(e.g. mood, an idea of the current activity, etc.).

2.2 Field studies and their implications

The social implications of home computing haveinterested a number of scholars for many years.Behaviors and usage patterns have beenexamined in field studies, and importantobservations have been made. The implicationsfor industrials, policy makers and technologydesigners are important. Some studies are nowreviewed, and important observations for thedesign of digital communication technologies arehighlighted.

In 1985, Vitalari, et al. [Vitalari, et al. 1985]examined the impact of home computers on timeallocation patterns, by analyzing the behavior ofsome 282 users. They observed that theintroduction of PC in the home increased thetime spent alone and decreased the time spentwith family and friends. It was however not clearwhether this was a short-term or long-termimpact, in which case it might be hypothesizedthat computers can lead to social isolation. Theauthors also observed a shift from pleasure-oriented to task-oriented activities. But it shouldbe noted that at the time of the study, there werestill very few applications targeted at home usersand that the home computer was merely a work-oriented tool used at home.

A very interesting point raised by the authors isthat the traditional human-computer interactionparadigm plays a key role in the timereallocation process:

"The time allocation phenomenon in computingis especially poignant with respect to personalcomputers, where personal, in operational terms,means a one-to-one real-time interaction withthe machine. [...] The popular example of thepersonal computer "widow" exemplifies thisdilemma."

The problem we see here is that interacting witha personal computer generally requires the userto be isolated from the rest of the household,both physically and mentally. The first reason isthat the computer is often in a room used byindividuals (an office, a room) and not in a roomused by groups (the kitchen, the living room). Inother words, the PC user is often in a roomwhere others don't spend much time, and is thusoften alone. The situation would be quite

different if the computer was in public space, forexample in the living room. The user would thenshare the physical environment with the otherfamily members and be less isolated. There areseveral reasons why PCs are not placed in publicspaces more often, including factors such as pooraesthetic quality, unavailable space and noisedisturbance.

Observation 1. Placing home computingtechnologies in public spaces, such as thekitchen or the living room, could reduce the riskof social isolation for users. This requires acareful design of the devices, which have tonicely integrate in the architectural space.

The second reason is that when using acomputer, it is difficult to do anything else at thesame time. The hands, the eyes, the whole bodyand the entire attention of the user is generallymobilized by the interaction. A good illustrationfor this problem is that many people feel verytired after using a computer for several hours.For some tasks (e.g. writing a report with a wordprocessing software), it is quite difficult toovercome this problem. But in some cases, itseems that the interaction between the user andthe computer could be achieved more naturally,not necessary requiring the use of a keyboardand a mouse.

The distribution of electronic postcards,mentioned earlier, provides a good example forthis issue. It seems that having the postcardsautomatically shown on displays scattered withinthe home (on walls, desks, phones, etc) is anadvantageous alternative to the current situation.In this case, people would not have to dedicate asignificant amount of time to the task ofconsulting incoming mail (going to the PC room,switching it on, launching the browser, fetchingthe URL, turning everything down, comingback), but would rather continuously andseamlessly be using the technology, withoutreally noticing it.

Observation 2. Human-computer interactionmechanisms that do not monopolize the mentaland physical attention of users could reduce therisk of social isolation. This requires the designof multi-modal user interfaces, where interactionis supported through voice, gesture and othermeans.

In another study published in 1996, Venkateshcompared the situation of home computing in theearly 1980s and the 1990s. He explained how anumber of factors had contributed to havedelayed the "home computer revolution"announced with the first apparition of computers

in the home. The addition of communicationfeatures and the availability of softwaresupporting household tasks (entertainment,education, financial management) are two factorsthat explain the greater acceptance andsignificance of home computers in the 1990s.Another critical factor is the understanding thattechnological innovation can happen in thehome, and does not have to happen at work andthen be transferred in the home:

Observation 3. Computing technologies do nothave to be transferred from work to domesticenvironments. Instead, they should be developedwith a consideration of the particular needs ofhome users.

Another observation made by the author was thatto gain acceptance within the home, computershad either i) to compete with existingtechnologies (e.g. phone) and to support existingtasks more efficiently or ii) to enable newactivities that were not possible withoutcomputers. For the designers of digitalcommunication technologies, it means thatradically different ways to support interactionbetween people should be searched for:

Observat ion 4. Digital communicationtechnologies should not only mimic traditionalcommunication media. Instead, they shouldenable functions that were not possible without apervasive communication infrastructure.

Another study with important implications is theHomeNet field trial. A research group atCarnegie Mellon University observed the onlinebehavior of more than 70 households during theirfirst years of Internet use.

The first question, addressed in [Kraut, et al.1998], was to find out what home users reallywant to use the Internet for. Is it for informationacquisition and entertainment, or is it forinterpersonal communication? To answer thisquestion, the respective use of a Web browser(used for access to information andentertainment) and of an electronic mail client(used for interpersonal communication) werecompared over time. The collected data suggeststhat users generally prefer email over the Web.In other words, the use of the Internet at homeseems to be motivated first by interpersonalcommunication, and not by information access.In the opinion of the authors, however, there hasbeen a tendency so far to focus primarily oninformation access and entertainment services:

"The commercial development of Internetservices and public policy initiatives to date have

probably over-emphasized information andentertainment and under-emphasizedinterpersonal communication."

The implication for computer scientists is that aneffort should be made to improve existinginterpersonal communication applications. It alsomeans that it is worth investigating new forms ofinterpersonal CMC, as there is a strong demandfor it:Observation 5. Interpersonal communication iswhat people really want to use the Internet for.Therefore, a special effort should be made toimprove existing interpersonal communicationtechnologies, and to propose new ones.

Another question examined in the HomeNetstudy is also the title of an article [Kraut, et al.1998]: "Internet paradox: a social technologythat reduces social involvement andpsychological well-being?". Because it is notclear whether the social impact of the Internet ismostly positive or negative, the authors collecteddata to identify a possible correlation betweenInternet use, social involvement andpsychological well-being. Social involvementwas measured by family communication, size oflocal social network, size of distant socialnetwork and social support. Psychological well-being were measured in terms of loneliness,stress and depression. The findings of theresearch indicate that greater use of the Internetwas associated with declines in socialinvolvement, and increases in loneliness anddepression.

In order to explain these results, the authorsmade two hypothesis. First, the negative impactcould be due to the fact that Internet use requiresto abandon other activities, such as readingbooks or talking with family members. This isthe same problem as mentioned before, with theanalysis of Vitalari et al. on the impact of homecomputing. Second, the authors argue thatcurrent CMC tools do not allow socialinteractions that are as good as face-to-face ortelephone interactions. They thus suggest that aconsiderable effort should be made to developbetter interpersonal communication technologieson the Internet.

2.3 Affective awareness

A particularly interesting application for CMC inhome settings is to offer efficient ways forpeople to keep in touch with their families andfriends. The notion of "keeping in touch" is verygeneral and has very different aspects.

It seems that current technologies primarilysupport one of these aspects, which is idea ofconversations. In other words, they support theexplicit exchange of factual information. Thismeans that to keep in touch with someone, it isnecessary i) to explicitly initiate an interaction,ii) to encode a message (usually by typing text),and finally iii) to transmit this message. The paceat which messages are sent back and forthbetween the participants depends very much onthe technology. With email they might beexchanged once a day. With synchronous text-based communication (e.g. IRC, MUD, chatrooms), they might be exchanged every fewseconds. In any way, this kind of interactionalways requires some effort from theparticipants, at least to establish thecommunication. This effort, as small as it couldbe, represents a barrier to communication andpotentially reduces the amount of socialinteraction.

When people share the same house, the situationis very different. For example, consider whathappens in the kitchen, when a family is havingbreakfast. Even when they are not talking, peoplehave many ways to interact with each other.They can easily guess what is the emotionalcondition of each other, have an idea of theirdaily schedule. Body language, dress code, facialexpressions are some of the many naturalchannels that convey this information. In thissituation, people are interacting tacitly: they donot have to explicitly exchange messages (i.e.talk) to feel "in touch" with each other. Theimportant thing is that they do not have to do anyeffort to interact with each other: sharing thesame physical space is sufficient.

A recent survey [Philips 1998] was conducted toget an understanding of communication patternsbetween middle-school-age children and theirparents. The results showed that 58% of theparents and 73% of the children said that theyspend less than an hour a day in conversation.Moreover, 27% of the parents and 46% of thechildren said that they spend less than 30minutes a day in conversation. This wasinterpreted negatively by the authors of thestudy, who then proposed different ways toincrease the level of communication betweenparents and kids. But a more encouraging way tolook at these results is to point out thatconversation is not the only mechanism throughwhich parents and kids build affectiverelationships. Therefore, digitally mediated

social interaction should not only be supportedby conversation, but also by other means.

What we call affective awareness is a generalsense of being close to one's family and friends.It seems that affective awareness is best achievedwhen people are engaged in shared experiences,especially when these experiences affect theiremotional state. Watching a funny TV show,observing paintings at an art gallery, enjoying agood meal, playing games are all simple ofactivities that, when performed together, makepeople feel close to each other. It seems thatCMC technologies that would recreate this kindof interactions would be very beneficial to users.First, because they would not require mucheffort, they would be used frequently and wouldincrease social interaction within distributedfamilies. Second, because they would not requirethe explicit interaction between a user and acomputer, they would avoid many of theproblems mentioned in the previous section (inparticular the risk of social isolation caused byextended periods of computer use).

As a matter of fact, similar issues have alreadybeen studied in work environments. Distributedworking groups have very similar problems, inparticular the difficulty to maintain a context forcommunication. The notion of awareness hasbeen extensively discussed in the ComputerSupported Cooperative Work (CSCW) literature[Dourish and Bellotti 1992, Dourish and Bly1992, Gutwin and Greenberg 1998, Isaacs, et al.1996, Tollmar, et al. 1996]. More specifically,peripheral awareness has been identified as theawareness that people can maintain about eachother, seamlessly and without effort, when theyare close to each other. A number of systemshave been developed to artificially recreateperipheral awareness within distributed groups.Media spaces [Bly, et al. 1993, Pedersen andSokoler 1997], ambient user interfaces[Wisneski, et al. 1998], and calm technologies[Weiser and Brown July 1996] are all related tothis idea and have been an inspiration for ourwork.

An important aspect is that in these approaches,human-computer interaction is not achievedthrough a conventional keyboard-displayinstallation. Instead, the computing environmentis more closely integrated with the architecturalspace, and is surrounding the users. Sounds,images, but also temperature and tangibleartifacts [Ishii and Ullmer 1997] are used tocreate an interface between the physical anddigital environments. Once again, this seems to

be particularly valuable to address the issuesraised in the previous section, as it does notrequire users to be cut out from the socialenvironment when they are using the CMCtechnology.

It is interesting to note that because media spacetechnologies rely on high-speed and continuousnetwork connections, they have long beenconfined to office settings. But continuous high-speed Internet connections are becoming a realityfor home users, which will make it possible todevelop similar systems for domestic use.

3 AFFECTIVE AWARENESS THROUGHDIGITAL HOME PHOTOGRAPHY

The previous section has argued for the need toaugment existing CMC technologies with toolsthat support affective awareness. The benefits ofapplications that do not support explicit, butimplicit social interaction, have been suggested.Engaging remote participants in sharedexperiences, that have an emotional effect onthem, seems to be a promising way to achievethis goal. In this section, it is explained howhome photography provides a context forenabling affective awareness in digitalenvironments.

The social role of traditional home photographyis discussed first, with references to visualanthropology. A description is then made of theKAN-G framework, which integrates hardwareand software components supporting the capture,distribution, observation and annotation ofphotographs. The framework supports affectiveawareness between two categories of users:photographers and watchers. An affective link iscreated between them, in two directions. First,watchers are connected to photographers whenthey receive their snapshots. At this point, thereaction of watchers is captured and notifiedback to the photographers, who in turn feelconnected to them. This two-way process is bestachieved when the components that display thephotographers snapshots and the watchersfeedback are integrated in the livingenvironment, and do not require the users tointeract with personal computers.

The description of the KAN-G framework isillustrated with a number of emerging digitalimaging technologies, such as digital cameras,on-line services and innovative displays.

3.1 Social role of home photography

Photography can been considered both as avisual medium, as an art and as scientific

instrument. Furthermore, photography supports arange of social functions. For example,photographs allow people to keep a memory ofhistory, they allow them to discover othercultures. The study of these functions has beenof foremost interest for many scholars,particularly for visual anthropologists [Ruby1981].

Home photography is a particular type ofphotography, that is concerned with the creationand utilization of pictorial artifacts in the contextof the home. In a broad definition, it includessnapshots, family albums, slides, home moviesand home videotapes [Chalfen 1987].Increasingly, it also includes digital artifacts andsupporting technologies, such as digital camerasand on-line photo albums. Richard Chalfen is avisual anthropologist who has been particularlyinterested in studying the functions of homephotography. His ethnographic studies are usefulto get a better understanding of how and whypeople take, display and share pictures. Theyhave also important implications for the designof digital imaging technologies, particularlywhen they are used to support social interaction.

Looking at Japanese home media as a popularculture [Chalfen 1997], Chalfen describesseveral uses of photographs and highlightsdifferences between Japanese and occidentalcultures. Questions that have interested himwere, in his own words:

"When and where, under what conditions haveJapanese people made photographs and, in turn,looked at them, displayed them or otherwiseshowed them to other people? In turn, how doJapanese people understand and value their ownphotograph collections?"

In his discussion, Chalfen describes fiveexamples of photograph use. The first example isreferred to as shared photographs, and highlightsthe fact that people often like to share pictureswith their relatives, friends and colleagues. Thesecond example is cal led householdphotography, and relates to the display ofphotographs within the home (on walls,furniture, etc). An interesting observation is thatthere are generally few pictures displayed inJapanese homes, unlike in occidental homes.This illustrates the fact that the functions ofphotography depend very much on culturalfactors. The third example is work photography,which is about snapshots both i) taken and ii)displayed at work. Again, there are significantdifferences between Japan and occidentalcountries: Japanese tend to take more pictures at

work, but at the same time tend to displaysnapshots at work less often (one reason is thatthere is a greater separation between private andpublic life). The forth example is wa l le tphotography, which indicates a common practiceto always carry a few snapshots (usually familysnapshots). Finally, the fifth example is touristphotography.

The observations made by Chalfen areinteresting for several reasons. First, they clearlyshow that photographs are social artifacts thatcan trigger affective processes. This is whypeople display photographs at home and at work,this is why people carry photographs in theirwallets. Looking at someone's photographicportrait naturally leads to think about this person.It often recalls memories and emotionally affectsthe viewer, provoking for example happiness,sadness or anger, depending on the situation.

Second, they show that sharing photographs is asocial process that connects people to each other.Particularly in Japan, there is a social obligationto regularly exchange snapshots with relatives,friends and coworkers. After a party or a trip, forexample, people often exchange reprints wherethey appear together. In this case, snapshots havea function of "social currency" and serve tostrengthen relationships and to reinforce groupcohesion.

There are many other situations that illustrate thesocial function of snapshot exchange. Think forexample of grandparents receiving pictures oftheir grandchildren. In this situation, they arelikely to feel in touch with them for two reasons.First, looking at the snapshots naturally leadsthem to think about their children andgrandchildren. Second, the fact that thephotographs were sent tells them that theirchildren are thinking about them, that they arenot forgetting them. This is very comforting forthem, and is an important positive outcome ofthe exchange. As a matter of fact, thisobservation has been a foremost motivation forthe design of the KAN-G framework.

A related phenomenon that should be mentionedis the tremendous success of print clubs(pronounced "puricura") in Japan. Located onevery street corner, these machines are used byhigh school students to take photographs of theirsmiling faces, later decorated with cute drawingsand printed on small stickers. These stickers arethen exchanged among friends and organized inlarge collections. What is interesting about printclubs, is that they are computing devices thatstress the social function of photographs.

In another study [Chalfen 1998], Chalfen lookedat the differences between still and video images.More precisely, he examined first how effectiveeach of them is to recall memories. Second, helooked at how each of them is qualitativelyperceived by home users. These questions wereanswered on the basis of interviews conductedwith 30 teenagers. The participants were asked toevaluate the relative merits of using still andvideo imaging media for home photography.

The first observation was that video couldrecreate the past with greater fidelity. Theamount of information is much larger (there isalso sound), and thus is better suited to reproducean ambiance. In other words, watching a videogives a sense of "being there". Participants to thesurvey generally agreed that video brought morememories than still pictures, that it made itpossible to duplicate and live experiences of thepast again.

But more interestingly, the study also suggestedthat this did not mean that video tapes werepreferred over snapshots and photo albums, inthe contrary. Clearly, there was an idea that inmany cases, less information is better. Peopleexplained that they preferred still pictures,because they had to use their imagination toremember the people and the events captured onthem. When watching a video, the past isrecreated but the viewer is passive and does notthink about anything that is not in the video. Inthe contrary, observing a still picture requires tomake an imagination effort, and triggers aprocess that is not bounded to what isrepresented in the picture. This idea wasmentioned several times, for example in thefollowing interview excerpts:

"When I look at a [still] picture of somebody,let's say someone I haven't seen in a long time orsomething, I think about everything I did withthat person and everything that person saidinstead of that one movement which on video isusually quite unimportant in a home video."

"Like still pictures, like they're like, you can lookat them and you can like gather what you want.But videotape, like, you can't like, you can'treally imagine anything about the situation."

Another reason why people prefer still picturesover videotapes has to do with the socialinteractions they generate when they are watchedby a group of people. In the case of video, peopleare usually watching a TV screen in silence, andthe level of communication between them isquite low. In the contrary, when a group ofpeople watch still pictures and photo albums,

they very often use them as a context to tellstories and exchange comments. In other words,it seems that still pictures are better suited togenerate interaction between people.

This has interesting implications for the designof visual CMC technologies. At first, it couldseem that video-conferencing is the ultimateinterpersonal communication tool. Indeed, itprovides rich interaction channels, and almostrecreates the conditions of a face-to-facemeeting. But maybe, as it is the case withmemory and still vs. video images, there aresituations where "less is better". When peoplejoin a video-conference, they h a v e tocommunicate, they have to talk to each other.This means that when they don't have anythingparticular to say, or when they have dosomething that would not allow them to talk atthe same time, then the communication isinterrupted. On the other hand, a CMC tool thatis based purely on the exchange of stillphotographs can generate interesting intellectualand affective processes. The person who receivesa snapshot remembers and imagines people andevents, and is emotionally affected by that.Another advantage of this approach is that itdoes not raise the privacy issues of video-basedawareness systems.

The point we are trying to make is that it is notnecessary to have an explicit, synchronouscommunication medium to create affective linksbetween people. Very often, the only thing that isneeded is to trigger a mental process with anartifact (e.g. a photograph) or a tangible event(e.g. a sound indicating that a person has smiledwhen looking at one of your photographs). Thisidea provides a foundation for the KAN-Gframework, that we now introduce.

3.2 The KAN-G framework

KAN-G is best defined as an interpersonalcommunication framework supporting affectiveawareness through home photography. It hasbeen named after the Japanese word kanji,meaning impression or feeling. This suggests thatthe goal of the framework is not to support theexchange of explicit messages, but rather toconvey a general sense of being "in touch"among users. This goal is achieved through theexchange of i) photographs and ii) reactions ofwatchers on these photographs. The capture,distribution and observation of snapshots areperformed with a number of digital imagingdevices, distributed across the Internet.

The interaction between users of the system isbased on the following observations:

• "Receiving and watching a photograph from aperson connects me to that person."

• "Knowing that a person is watching myphotographs, and knowing what this personthinks of them, connects me to that person."

Accordingly, as illustrated in Figure 1, there aretwo categories of KAN-G users: photographersand watchers. Photographers use digital camerasto take snapshots that they distribute to watchers(usually their relatives and friends). Watchersobserve the snapshots, their reaction is capturedand notified back to the photographers.

KAN-G supports the distribution of photographsvia publication channels (i.e. buffers of limitedsize, to which pictures are successively pushed).Photographers can define several channels (e.g."Family", "Funny pics", "Food"), and can decideto distribute a particular photograph to one orseveral of these channels. Symmetrically,watchers can subscribe to the channels they areinterested in. One motivation for using channels,as opposed to large photo collections thatcontinuously grow, is to make the system moredynamic. Because channels are limited buffers, itis likely that changes will be noticed more easily.This should make the system more interesting forwatchers, and reinforce the impression that thereis activity in the system. This is important,because the activity is interpreted as themanifestation of social interactions.

Photograph

Photographer Watcher

Feedback

AffectiveAwareness

capture &distribution display

feedbackcapture

feedbackdisplay

Figure 1. Affective awareness in KAN-G.

The architecture that supports these functions isdepicted in Figure 2. First, photographers takepictures with a KAN-G digital camera, andpossibly add comments to them (title, shortdescription, etc). Some contextual information(e.g. time, location if a GPS sensor is used) isautomatically added by the camera.Photographers then use the digital camera toinitiate the publication process (i.e. to specifywhich pictures should be published to whichKAN-G channels).

When this is done, the pictures have to betransferred to a KAN-G server, which maintainsthe state of publication channels. If the camera isequipped with wireless Internet access, this stepcan be performed directly. If not, the user has togo through a KAN-G kiosk. This means that thestorage media (e.g. Picture Card or Smart Media)has to be removed from the camera and insertedin a device connected to the Internet (e.g. a PCwith a PCMCIA card reader). This device willthen establish the connection with the KAN-Gserver and transfer the pictures.

Watchers receive and observe these snapshotswith different tools (e.g. Web-based interfacesand ubiquitous displays), which get thephotographs from a KAN-G server. These toolsare also used to make comments and express

emotions (e.g. laugh at a photograph). Thefeedback from watchers is finally notified to thephotographers with various tools, calledawareness monitors. They make it possible forphotographers to virtually hear their friendslaugh or cry when they observe their pictures.

In this scenario, there is no direct, explicitcommunication between photographers andwatchers. We nevertheless believe that anaffective link has been created between them.

3.2.1 Digital cameras

A very important component in the KAN-Gframework is the device that is used to create thesnapshots: the digital camera. Recent digitalcameras can truly be described as informationappliances: they integrate hardware components,an operating system, communication capabilitiesand a user interface. Moreover, their purpose isto produce, process and (temporarily) storeinformation. Not only graphical information, butalso contextual metadata such as the time,location or title of a photograph. Some digitalcameras even integrate a programmingenvironment, which makes them a uniqueplatform for developing innovative photo-centricapplications. Real estate, insurance and medicineare some of the professional fields that benefit

DigitalCamera

KAN-GServer

Photographer

Kiosk

- take snapshots- annotate snaphots- initiate distribution (select channels)- if possible, distribute

AwarenessMonitor

- who is looking?- which picture?- what is the impression?

- maintain the state of channels- receive pictures from photographers- serve pictures to watchers- monitor access to channels

- distribute

accessnotification

photographs& metadata

Watcher

WWWInterface

UbiquitousDisplays

EmotionCapture

FeedbackCapturephotographs

& metadata

feed-back

Channels

Figure 2. The KAN-G framework architecture.

from these applications, particularly becausethey enable automatic classification and efficientretrieval.

A very interesting device is the Locatio3, recentlyintroduced by Epson in the Japanese consumermarket. This appliance is at the same time aPDA, a mobile phone, a digital camera and isequipped with location sensors (GPS). It is alsointeresting to note that many prototypes ofwearable computers integrate a digital camera,for example [Mann 1997] and [Healey andPicard 1998].

Such devices are likely to change the behavior ofhome photographers. Low cost, immediateavailability of the snapshots, ergonomicsallowing to always carry a camera (e.g. when itis embedded in one's glasses [Mann 1997])should increase the quantity of photographstaken. Besides, the electronic nature of themedium, combined with the Internet, makes itextremely cheap and easy to duplicate and sharethe pictures with others. But while it is alreadypossible to exchange digital photographs (e.g. byattaching them to email messages), there is aclear need for better solutions.

When designing the KAN-G framework, one ofour main requirements was that initiating thedistribution of photographs should be aseffortless as possible. Otherwise, people wouldnot do it regularly. This is similar to the problemof keeping up with one's email or updating one'sWeb site, that was discussed in Section 2.1.1.Therefore, it seemed a good idea to implementthis function directly on the camera, with twoadvantages. First, the task can be performed incontext, i.e. when the snapshot has been takenand with a single task-oriented tool. Second, thetask can be performed very rapidly, in a matterof seconds (no need to find a PC and to wait forit to boot up).

Most of the currently available digital camerasoffer a limited number of functions, accessiblevia a user interface generally composed of anLCD display, a few buttons and a few switches.More interesting are the cameras, currentlyincluding models from Kodak and Minolta, thatuse the Digita operating system [Digita ]. Digitais a proposed standard operating system forimaging devices. It includes a menu-driven userinterface, various sub-systems and mostinterestingly a scripting language. Using this

3 http://www.i-love-epson.co.jp/locatio/index.html

language, it is possible to extend thefunctionality of the camera and to implementinteresting applications. Digita scripts cancontrol the hardware (e.g. zoom in and out, takea shot), write text files (but unfortunately notread them yet) and control GUI widgets (e.g.option lists, informative messages, text input).They also have R/W access to a number of datafields storing metadata for each picture (e.g.location, title). The scripts are organized inhierarchical menus, which can easily be accessedby the user with a four arrow keys button.

As part of the KAN-G framework, weimplemented two Digita scripts that run on theKodak DC260 camera. These are used toindicate which pictures should be published, andto which channels they should be published.Screenshots of the digital camera screen arereproduced in Figures 3, 4 and 5. They representthe user interface at different steps during theexecution of the scripts.

The first script, "Set channel", lets the userdefine which pictures should be published. Thefirst step consists in going through the snapshots,and in marking those that are going to bepublished (Figure 3). When this is done, the userhas to select the "Set channel" script in the menu(Figure 4), and then to choose one of thepredefined channels in a scrolling list (Figure 5).The script then traverses the list of markedpictures. For each of them, it updates a metadatafield with the reference of the selected channel.A completion message is finally displayed on thescreen. This process can be repeated severaltimes before the actual publication ofphotographs.

When the user has performed this first task, thesecond script can be invoked. The "Export tochannels" script traverses the list of photographsstored in the camera and checks the content ofthe metadata field assigned by the "Set channel"script. During this process, the script generatesan XML document [Bosak 1997], whichspecifies how the photographs should bepublished. This document also contains themetadata (e.g. time, title, etc.) that was createdon the camera (either automatically or asspecified by the user). The XML document isused when the snapshots are actually transferredfrom the camera to a KAN-G server, eitherdirectly or via a KAN-G kiosk.

Figure 3. Marking snapshots for publication.

Figure 4. Selecting the Digita script.

Figure 5. Selecting the KAN-G channel.

3.2.2 Servers, channels and kiosks

In the KAN-G framework, pictures are not sentdirectly from photographers to watchers (as itwould be the case with email attachments, forexample). Instead, they are published onchannels managed by servers. Photographerspush information on these channels, watcherspull information from them. Therefore, KAN-Gservers essentially have two functions. On onehand, they must accept photographs and maintainthe state of channels. On the other hand, theymust handle requests from watchers and givethem access to the channels.

A KAN-G server has been implemented byextending the functionality of an HTTP serverwith a collection of Java servlets. Java servletsare one approach to dynamic content on theWeb. In other words, they are an alternative toCGI scripts, and have several advantages overthem. Because KAN-G servers are built asextensions of HTTP engines, they are directlyaccessible on the WWW. This seems to be veryimportant, as it makes the system readilyaccessible to almost everyone. Using a standardWeb browser is one possibility to browsethrough KAN-G channels. While it certainly is arequired one, we will later explain that it mightnot be the best one.

In a near future, many digital cameras will beable to establish wireless Internet connections.This will make it possible to send photographs

directly from a camera to a KAN-G server. Butbecause this is not the case yet, we decided tointroduce kiosks in the architecture. Afterinitiating the publication of photographs, i.e.after running the two Digita scripts, the usersimply has to extract the storage media from thecamera and to insert it in a kiosk. In ourprototype installation, the role of the kiosk isplayed by a PC with a PCMCIA card reader. Thesoftware running on the kiosk reads the XMLdocument generated by the script, whichspecifies what photographs to fetch. It also readsa special file stored on the picture card, thatdescribes on which KAN-G server the user ownsan account. Finally, a connection is establishedwith the server and the photographs listed in theXML document are pushed to the specifiedchannels. Although it introduces an extra stepand some delay, the kiosk still keeps thepublishing process simple and does not requiremuch effort from the user (all necessaryinformation has been gathered before, directly onthe camera).

A number of commercial services alreadysupport some of the functions enabled byKAN–G servers. Kodak PhotoNet Online[Kodak ] and FujiFilm.net [FujiFilm ], forexample, allow users to store digital photographsin on-line photo albums. These albums can thenbe browsed by family and friends. Besides, thevisitors can also order reprints and variousaccessories (mugs, tee-shirts, etc). An interestingpoint is that when people send a regular film tobe developed and printed, they can tick an optionto have them digitized and published on-line.This is very important, because it makes thepublication very easy for the user: no need tohandle a scanner, no need to use FTP, etc. On-line photo services have limitations that we havetried to overcome in KAN-G. First, it is difficultto know when new pictures are published by aphotographer. It is necessary to periodicallycheck the Web site, which requires a substantialeffort. Second, photographers don't knowwhether their family and friends are watching thepictures, nor what they think of them. Third, theonly mechanism for looking at the pictures is to

Figure 6. Web-based interface for browsing KAN-G channels.

use a Web browser (and thus a PC). It seems thatother ways to perform this task would be moreappropriate, as we suggested earlier with theelectronic postcards application (Section 2.2).

It should also be pointed out that digital photocollections, whether they are accessible on-lineor on CD/DVD, raise a number of researchissues in terms of efficient organization,annotation, browsing and retrieval. A veryinteresting system is the FotoFile system[Kuchinsky, et al. 1999], which makes themanual annotation of photographs easier. This isachieved by several means, in particular by theuse of a narrative structure to organize thephotographs. Because the collection organizationprocess is transformed into a storytelling process,it is more enjoyable to the users.

3.2.3 Tools for watchers

Because KAN-G servers are HTTP servers,standard Web browsers can be used to observephotographs published in KAN-G channels.Different user interfaces can be developed forthis purpose, from very simple ones (a Web pagethat shows the last photograph pushed on achannel), to very complex ones (whereJavaScript and DHTML are used to createinteractive behaviors).

An example for such a user interface isrepresented in Figure 6. On the left side,thumbnails of the pictures in a channel aredisplayed. When the user clicks on one of them,it appears at the center of the window, in large.On the right, a textual description of the picture(which has been defined on the digital camera) isdisplayed. Finally, the user can click on differentbuttons situated below the photograph. "Smaller"and "Larger" allow to change the size of thecenter photograph. More interestingly, "Laugh","Applause" and "Booh!" are used to get afeedback from the watcher. When one of thesebuttons is clicked, a message is sent to the KAN-G server, and the photographer who publishedthe photograph receives a notification (allowingfor example to virtually hear the watcher laughor applause). This user interface was onlydeveloped as a proof of concept, and more subtleways to collect feedback from the watchershould be examined.

Web browsers provide a practical solution towatch pictures published in KAN-G channels:access to the WWW is ubiquitous and does notrequire particular hardware or software. It thusmakes it possible for anyone to start using thesystem. Furthermore, a Web interface is

sometimes perfectly suitable. This is particularlytrue when the watcher wishes to browse throughcollections of photographs. In this case, it isacceptable to use a personal computer, akeyboard and a mouse to watch and comment thephotographs.

But only providing this kind of interface is notenough. In Section 2, we have seen how usingpersonal computers presents a risk of socialisolation. Two reasons for this were that using aPC often requires to be alone, and that traditionalhuman-computer interaction monopolizes all theuser's physical and intellectual faculties. Themotivation for designing the KAN-G frameworkwas to increase the level of social interactionamong people who do not live together. But ofcourse, this goal should be achieved bydecreasing the level of interaction among thepeople who do live together. A worse situationwould be achieved if a user was spending toomuch time exploring on-line photo collections,alone, and consequently unable to interact withthe members of the household.

One solution to this problem consists inproviding additional interfaces, through whichusers can interact with the system in moreseamless ways. These interfaces should alsoenable processes where the user can remainpassive, and does not have to give an input to thesystem (e.g. watching TV as opposed tobrowsing the Web). Whenever possible, theseinterfaces should be integrated in thearchitectural space of the home: in furniture, onwalls, etc.

Indeed, the home is already populated with awide range of displays. These includetelevisions, computer displays, but also screenphones, wall-mounted panels, etc. In the future,information will be displayed on digitalwallpaper, mirrors and windows. Very often, thedisplays found in domestic environments areused punctually for a specific purpose, but areinactive for long periods. This seems like a wasteof interactive display real-estate, which could forexample be used by CMC technologies, inparticular by systems supporting socialawareness between people. In our case, thesedisplays could be used to show the photographspublished in KAN-G channels.

As an example, Figure 7 shows a device that isused by one of the authors, in his kitchen. It is aToshiba Libretto palmtop computer, that is nevershut down (it is therefore always immediatelyavailable). There are a few interesting commentsto make about how this device is used on a daily

basis. First, it is in the kitchen, and is generallyused when people are at the table (before a meal,for example). This means that interaction withthe device is generally done when familymembers are together. Second, the keyboard israrely used (this requires user interfaces such asthe one described below), and there is no mouse.The pointing device is integrated to the palmtop,on the right side of the display. All this makes itpossible to use the device almost like a book, andin very natural body positions. Finally, itexclusively used as a communication device,mainly of (non-professional) email and Webbrowsing. The interaction with the device ofteninitiates social processes within the household. Agood example is the conversations between theauthor and his wife that are triggered byincoming emails (the activity of writing emails isalso sometimes done in a common).

Figure 7. KAN-G monitor used in the kitchen.

This device is perfectly suitable to be integratedin the KAN-G framework, where it could playthe role of a digital picture frame. Digital pictureframes are very similar to traditional pictureframes: they have different shapes and sizes, it ispossible to move them around, etc. Thedifference between the two is that the imagedisplayed in the frame is digital and cantherefore easily and automatically be changed. Itis interesting to note that Sony has already

introduced a digital picture frame4 in theconsumer market. While this device is stillexpensive and does not have a networkconnection (the pictures are transferred via aMemory Stick storage media), it is a promisingdevice for home photography applications.

Digital imaging devices used to displayphotographs require a careful user interfacedesign. The first thing to consider is that theirmain function is not to support the exploration ofa photo collection (as it was the case with theWeb-based interface). In the contrary, it is toautomatically display photographs, withoutrequiring the intervention of the user.Additionally, it is to gather some feedback fromthe watchers. Ideally, this feedback should becaptured automatically. This would require theuse of sensors (e.g. camera, microphone) andsoftware to recognize emotions. This is of coursea difficult problem, where many issues remainopen. Nevertheless, a lot of research is done inthis area [Moriyama and Ozawa 1999, Nakatsu,et al. 1999, Picard 1997]. While automaticemotion recognition is not possible, the userinterface should make it as easy as possible forthe user to give a feedback. In particular, itshould not require the use of keyboard.

A prototype was built to provide an example forsuch a user interface. A snapshot, given in Figure8, shows this tool. While it is difficult to get agood idea of its dynamic and interactive behaviorwith a static image, here is a brief description ofthe software:

• In the lower part of the display, photographspublished in KAN-G channels are scrollingfrom left to right. In this way, the user canremain passive and simply watch at theincoming photographs.

• In the upper left corner of the panel, the usercan express emotions inspired by thephotographs. This is done by moving acircular cursor (with the mouse, or anotherpointing device) within a pie chart that showssix different emotions (happiness, surprise,anger, disgust, fear and sadness). The intensityof the emotion is variable and is determinedby how long the user presses the mouse buttonbefore releasing it (a visual feedback is givenby an expanding-shrinking size of the cursor).

4

http://www.sel.sony.com/SEL/consumer/ss5/home/digitalimagingmavicartmcamera/digitalimagingmavicartmcamera/phd-a55_specs.shtml

• When an emotion has been expressed, a visualfeedback is given in the upper right portion ofthe screen, with vertically scrolling text. Thesize of the words indicates the intensity of thefeeling expressed by the user (large sizeindicates a strong feeling, small size indicatesa moderate feeling).

• When feedback has been gathered via thisuser interface, it is notified to the person whopublished the photograph. This function issupported by a messaging infrastructureintegrated with the KAN-G framework.

Visual feedback(scrolling vertically)

User feedback(mouse press-release)

Photographs (scrolling horizontally)

Strongfeeling

Weakfeeling

Figure 8. A KAN-G watcher monitor

3.2.4 Awareness monitors for photographers

The previous components of the KAN-Gframework support the flow of informationbetween photographers to watchers. Their goal isto make the capture, the distribution and thevisualization of photographs as easy and naturalas possible. This is important to guarantee thatusers will not perceive the picture sharingactivity as a burden, and will go through thisprocess regularly and lastingly. While sharingphotographs on the Internet is already possible(e.g. via email or Web sites), it requiresconsiderable time and effort. Moreover itrequires some technical knowledge that is notnecessary mastered by the average consumer. Allthis contributes to limit the amount of digitalphotographs exchanged by home users.

Another original aspect of the KAN-Gframework is that it enables a flow ofinformation in the opposite direction, fromwatchers towards photographers. Whenphotographers know that their relatives andfriends are watching their pictures, they can feelconnected to them. An even richer experience isachieved when they have some idea of whatthese people have felt when watching thepictures. Supporting this function first requires tocapture the feedback from the watchers, as weexplained before. It then requires photographersto use special tools to notify them about thesocial activity generated by their snapshots.These tools are called awareness monitors in theKAN-G architecture.

What was said about the tools used to displayphotographs also applies to awareness monitors.They should ideally not require the use ofstandard PC installations, but be more nicelyintegrated into the living environment. Hereagain, the idea of peripheral awareness is central:photographers should be able to interact with thesystem (i.e. to receive access and feedbacknotifications) while they are doing other tasks.

Awareness monitors can take many differentforms, and their design is only limited byimagination. In simple cases, they might useaudio and visual signals to notify the activityoccurring in the system. In more elaboratedscenarios, they might also involve tangible andambient channels (physical objects, temperature,light, etc.). We developed tools based on bothaudio [Liechti, et al. 1999] and visual [Liechtiand Ichikawa 1999] interfaces for the moregeneral case of notifying activity occurringwithin Web sites. These interfaces could easilybe integrated in the KAN-G framework, asaccessing and commenting a personal Web site isvery similar to accessing and commentingpersonal photographs.

Another prototype was also specificallydeveloped to fit the need of KAN-G awarenessmonitors. This tool is using the new 2D imagingJava API, which among other features supportssemi-transparency. When a notification isreceived by the awareness monitor, thephotograph that is being looked at, as well as acaption indicating the identity of the watcher,fade in on the display. If the notificationspecifies an emotion, a sound is also played bythe device. After a while, the photograph and thecaption fade out. Using this tool, it is thuspossible to hear that someone laugh acrosscyberspace. In a second step, it is possible to

look at the device to find out who is laughing,and why (i.e. what picture is being looked at).This seems like a very natural process, that doesnot require any effort from the user.

It should be noted that this particular userinterface is particularly suitable for synchronousnotifications. It seems however important to alsosupport asynchronous notifications. For example,it should be possible for the user to ask thesystem questions like:

• What was the activity during the last 3 days?

• What was the activity generated by this groupof photographs?

• What was the activity generated by this groupof people?

While we have many ideas for addressing theseneeds, they have not been implemented yet andhave been left for future work.

4 CONCLUSION

This article has examined the role of ComputerMediated Communication in domesticenvironments. Existing technologies have beenreviewed, some of their benefits and limitationshave been described. Related ethnographicstudies on the social impact of home computinghave been mentioned, and important implicationsfor technology designers have been highlighted.

The discussion has insisted on the fact that mostof the current technologies support explicit socialinteraction, by creating a new medium forconversations. While the merits of conversationsare recognized, the need to also support implicitsocial interaction has been raised. The notion ofaffective awareness has been introduced anddefined as a general sense of feeling in touchwith relatives and friends. It has also beenexplained how engaging people in sharedexperiences that emotionally affect them is onepossibility to create affective awareness betweenthem.

Finally, KAN-G has been described as aninterpersonal communication framework thatsupports affective awareness through digitalhome photography. For a motivation, the socialrole of home photography has been discussed,with references to visual anthropology. It hasthen been explained how the KAN-G frameworkintegrates a range of hardware and softwarecomponents that make the capture, distribution,observation and annotation of photographsnatural and effortless.

ACKNOLEDGEMENTS

This work was supported by the Japanesegovernment with a Monbusho scholarshipawarded to the first author. The authors arethankful to Eastman Kodak for providing themwith the Digital Science DC260 digital cameraused to implement the prototype system.

REFERENCES

1. S. Benford, C. Brown, G. Reynard, and C.Greenhalgh, “Shared Spaces: Transportation,Artificiality, and Spatiality”, in proceedings ofACM Conference on Computer supportedcooperative work (CSCW'96), Cambridge, MA,1996.

2. S. Bly, S. Harrison, and S. Irwin, “Media Space:Bringing People Together in a Video, Audio andComputing Environment”, in Communications ofthe ACM, vol. 36, 1993, pp. 28-47.

3. J. Bosak, “XML, Java and the future of the Web”,World Wide Web Journal, vol. 2, pp. 219-227,1997.

4. R. Chalfen, “Family Photograph Appreciation:Dynamics of Medium, Interpretation andMemory”, Communication and Cognition, 1998.

5. R. Chalfen, “Japanese Home Media as PopularCulture”, in proceedings of Japanse PopularCulture Conference, Victoria, British Colombia,CA, 1997.

6. R. Chalfen, Snapshot Version of Life: BowlingGreen State University: Popular Press, 1987.

7. Digita, Digita Operating System for imagingdevices, http://www.flashpoint.com/

8. P. Dourish and V. Bellotti, “Awareness andCoordination in Shared Workspaces”, inproceedings of CM Conference on Computersupported cooperative work (CSCW'92),Toronto, Ontario, 1992.

9. P. Dourish and S. Bly, “Portholes: SupportingAwareness in a Distributed Work Group”, inproceedings of ACM SIGCHI Conference onHuman Factors in Computing Systems (CHI'92),Monterey, CA, 1992.

10. FujiFilm, “FujiFilm.net”, .

11. C. Gutwin and S. Greenberg, “Design forindividuals, design for groups: tradeoffs betweenpower and workspace awareness”, in proceedingsof ACM Conference on Computer supportedcooperative work (CSCW'98), Seattle, WA,1998.

12. J. Healey and R. W. Picard, “StartleCam: ACybernetic Wearable Camera”, in proceedings ofIEEE International Symposium on WearableComputers, 1998.

13. E. A. Isaacs, J. C. Tang, and T. Morris, “Piazza:A Desktop Environment Supporting Impromptuand Planned Interactions”, in proceedings ofACM Conference on Computer supported

cooperative work (CSCW'96), Cambridge, MA,1996.

14. H. Ishii and B. Ullmer, “Tangible Bits: TowardsSeamless Interfaces between People, Bits andAtoms”, in proceedings of ACM SIGCHIConference on Human Factors in ComputingSystems (CHI'97), Atlanta, GA, 1997.

15. S. Junestrand and K. Tollmar, “The Dwelling as aPlace for Work”, in proceedings of FirstInternational Workshop on CooperativeBuildings (CoBuild'98), Darmstadt, Germany,1998.

16. Kodak, “Kodak PhotoNet Online”, .

17. R. Kraut, T. Mukhopadhyay, J. Szczypula, S.Kiesler, and W. Scherlis, “Communication andInformation: Alternative Uses of the Internet inHouseholds”, in proceedings of ACM SIGCHIConference on Human Factors in ComputingSystems (CHI'98), Los Angeles, CA, 1998.

18. R. Kraut, M. Patterson, V. Lundmark, S. Kiesler,T. Mukophadhyay, and W. Scherlis, “Internetparadox: a social technology that reduces socialinvolvement and psychological well-being?”,American Psychologist, vol. 53, pp. 1017-1031,1998.

19. A. Kuchinsky, C. Pering, M. L. Creech, D.Freeze, B. Serra, and J. Gwidzka, “FotoFile: AConsumer Multimedia Organization andRetrieval System”, in proceedings of ACMSIGCHI Conference on Human Factors inComputing Systems (CHI'99), Pittsburgh, PA,1999.

20. H. Lieberman, N. W. Van Dyke, and A. S.Vivacqua, “Let's Browse: A Collaborative WebBrowsing Agent”, in proceedings of InternationalConference on Intelligent User Interfaces,Redondo Beach, CA, 1999.

21. O. Liechti and T. Ichikawa, “A Visual InteractionMechanism for Increasing Awareness on theWWW”, in proceedings of IEEE InternationalSymposium on Visual Languages, Tokyo, Japan,1999.

22. O. Liechti, M. Sifer, and T. Ichikawa, “A Non-obtrusive User Interface for Increasing SocialAwareness on the World Wide Web”, PersonalTechnologies, vol. 3, pp. 22-32, 1999.

23. S. Mann, “An historical account of the'WearComp' and 'WearCam' inventionsdeveloped in 'Personal Imaging'”, in proceedingsof IEEE International Symposium on WearableComputers, 1997.

24. T. Moriyama and S. Ozawa, “EmotionRecognition and Synthesis System on Speech”, inproceedings of International Conference onMult imedia Computing and Systems(ICMCS'99), Florence, Italy, 1999.

25. R. Nakatsu, A. Solomides, and N. Tosa,“Emotion Recognition and Its Application toComputer Agents with Spontaneous InteractiveCapabilities”, in proceedings of International

Conference on Multimedia Computing andSystems (ICMCS'99), Florence, Italy, 1999.

26. M. Padula and R. Amanda, “Art Teams Up withTechnology through the Net”, in Interactions,vol. 6, 1999, pp. 40-50.

27. E. R. Pedersen and T. Sokoler, “AROMA:abstract representation of presence supportingmutual awareness”, in proceedings of ACMSIGCHI Conference on Human Factors inComputing Systems (CHI'97), Atlanta, GA,1997.

28. P h i l i p s , L e t ' s C o n n e c t S u r v e y ,http://www.philipsconsumer.com/letsconnect/

29. R. W. Picard, Affective Computing. Cambridge:MIT Press, 1997.

30. J. Preece, “Reaching Out Across the Web”, inInteractions, vol. 5, 1998, pp. 32-43.

31. H. Rheingold, The Virtual Community. Reading,MA: Addison-Wesley, 1993.

32. D. Rijken, “Information in Space: Explorations inMedia and Architecture”, in Interactions, vol. 6,1999, pp. 44-57.

33. J. Ruby, “Seeing Through Pictures: theAnthropology of Photography”, Camera-Lucida:The Journal of Photographic Criticism, pp. 19-32, 1981.

34. K. Tollmar, O. Sandor, and A. Schoemer,“Supporting Social Awareness @Work: Designand Experience”, in proceedings of ACMConference on Computer supported cooperativework (CSCW'96), Cambridge, MA, 1996.

35. A. Venkatesh, “Computers and Other InteractiveTechnologies for the Home”, in Communicationsof the ACM, vol. 39, 1996, pp. 47-54.

36. N. P. Vitalari, A. Venkatesh, and K. Gronhaug,“Computing in the Home: Shifts in the TimeAllocation Patterns of Households”, inCommunications of the ACM, vol. 28, 1985, pp.512-522.

37. M. Weiser and J. S. Brown, “Designing CalmTechnology”, PowerGrid Journal, vol. Version1.01, July 1996.

38. C. Wisneski, H. Ishii, A. Dahley, M. Gorbet, S.Brave, B. Ullmer, and P. Yarin, “AmbientDisplays: Turning Architectural Space into anInterface between People and DigitalInformation”, in proceedings of FirstInternational Workshop on CooperativeBuildings (CoBuild'98), Darmstadt, Germany,1998.