3

Media Spaces:Envir onmentsforInformal Multimedia Interaction

WENDY E. MACKAY

AarhusUniversity

ABSTRACT

Distributedorganizations,with distributedcooperative work, area fact of life. How cannew technologieshelp?Distributedvideo is an appealingchoice,carryingmorecontex-tual informationthanvoicealoneand,arguably, betterat conveying subtlecues,suchastheemotionalstates.Althoughnew commercialsystemsarebeingintroduced,they focusprimarily on providing new technology. Most arebasedon relatively simpleextensionsof two existingmodelsof communication:formalmeetingsbecomevideoconferencesandtelephonesbecomevideophones.However, researchin computer-supportedcooperativework hastried to emphasizetheuser, with modelsbasedon SharedWorkspaces(to sup-port sharedwork on a commontask),CoordinatedCommunication(to supportstructuredcommunicationto serve a specifiedpurpose),andInformal Interaction(to supportinfor-mal, unplannedandunstructuredinteractions).Althoughmediaspacescanincorporateallthree,they emphasizeinformal communication,providing peopleworking togetherat adistancewith interactionsthatthey takefor grantedwhenthey areco-located.Thischapterdescribessomeof the pioneeringwork in mediaspaces,with moredetaileddescriptionsof our own work at RankXerox EuroPARC (RAVE for our own usein thelaboratoryandWAVE, to supportengineersworkingcollaboratively betweenfacilitiesin EnglandandtheNetherlands),concludingwith adiscussionof thetechnical,userinterfaceandsocialissuesinvolvedin designingmediaspaces.

3.1 INTR ODUCTION

Telephones,faxes,electronicmail andtheWorld WideWebhave transformedwork,enablingpeopleto work together, evenwhenthey livein differentcountriesandin differenttimezones.Yet long-distanceprojectsarestill difficult, evenwhenculturalandorganizationaldifferences

ComputerSupportedCooperativeWork, Editedby Beaudouin-Lafonc

�1999JohnWiley & SonsLtd

56 MACKAY

are takeninto account.Why? One importantreasonis the lack of informal social contactthat peoplehave when they work in the samephysicallocation [Hea91].Peoplewho areco-locatedbenefitfrom chanceencountersin hallwaysor chatsbeforeandafter meetings,resolvingproblemsbeforethey becomecritical. Working in the samephysicalenvironmenthelpspeoplediscoversharedinterestsanddevelopasenseof community. Implicit knowledgeaboutthe stateof eachother’s work canprevent misunderstandingsor resentment:If I seethat my colleague’s report is sitting in her “out” basket,readyto send,I canavoid askingheraboutit andthusavoid offendingher. Whenpeopleareseparatedgeographically, muchof their informal knowledgeabouteachotherdisappearsandcommunicationbecomesmuchmoreformal. Attemptsto addressthis with additionalmeetingsandreportsoften serves toexacerbatethesituationandemphasizesthedifferencesbetweengroups.

Moran and Anderson[Mor90] identify threefundamentalapproachesto supportingco-operative work at a distance:SharedWorkspaces[Tan90, Min91, Ish92, Ols91a,Ols91b],which emphasizepeopleworking cooperatively on a commontask,CoordinatedCommuni-cation[Win89, Ell99], in whichpeoplecommunicatein astructuredfashionfor somepurpose(suchasdecision-making),andInformal Interaction,in which peopleengagein unplannedandunstructuredinteractions.Chapters4 and5 in this book [Ish99, Pra99]addresssharedworkspacesandChapters1 and2 [Ehr99, Ell99] addresscoordinatedcommunication.Thischapteris mostconcernedwith informal interaction,providing peopleworking at a distancewith thekindsof informal interactionsthey enjoywhenworking in closeproximity to eachother.

Theexplosionof networkedcomputingthroughthe World Wide Webandthe decreasingcostof video technologyhave madedistributedvideo a popularchoicefor addressingtheproblemof providing distributedsocialcontext. However, whenever new technologyis cre-ated,it usuallybeginsasanimitationof somethingthatalreadyexists.Not surprisingly, then,mostcommercialdistributedvideo systemsaremodeledafter oneof two familiar forms ofcommunication:telephonecallsandformalmeetings.Videophonesarebasedonthemodelofa telephonecall in which a callerestablishesa videoandaudiolink to a secondparty. Whenthecall is initiated,thephoneringsin theotherlocation;if theotherpersonis available,heorshedecideswhetheror not to acceptthecall andcompletetheconnection.Thecall continuesaslongasbothpartiesparticipate;whenonehangsup,theconnectionis broken.Videoconfer-encingis theothercommonmodel,usuallyinvolvingspecially-designedconferencerooms.Acommonarrangementusesonevideocamerato capturepeoplesittingata tableanda second,overheadcamera,to capturedocumentsor slides.Live videoimagesaresentto oneor moreremotevideo conferencerooms,via telephoneor satelliteandprojectedonto wall screens.Often,a separatespeakerphoneis usedto enhancethequality of thevoice.Desktopvideo-conferencingis a low-costalternative, designedto be usedwith computersin the office. Asmall videocamerais usuallyplacedon top of themonitoranddigitized imagesaresenttoa window on anotherparticipant’s screen.Someof thesesystemscanhandleseveral videoimagesat once,althoughthecomputermonitorquickly runsout of screenspace.Audio canbea problemif it is delayedandpeopleoftenusetelephonesor speakerphonesin additiontotheon-linevideo.

Telephoneandconferencingmodelsrepresenta limited subsetof thewaysin which videocansupportdistributedcooperative work. Thepurposeof this chapteris to describethecon-ceptof amediaspace,whichattemptsto extenddistributedvideoto includeavarietyof formsof communication,rangingfrom informal encountersandperipheralawarenessto focused,formalmeetings.Thedifferencebetweenmediaspacesandmostcommercialdistributedvideo

MEDIA SPACES 57

systemshaslittle to dowith technologyandeverythingto dowith theway in which thetech-nologyis embeddedinto thesocialenvironment.Understandingthesesocialissuesis essentialfor understandinghow to designandintroduceeffective mediaspaces.As MoranandAnder-son[Mor90] explain in theirdescriptionof EuroPARC’sRAVE mediaspace:

“EuroPARC’sconcernis not simplywith artifactsandtheir enablingtechnologies,but withunderstandingtheprocessesandrelationshipswhichsuchartifactssupport,includingthepro-cessesby whichthey aredesigned.Thedisciplineof designmustinvolveaconstantmovementbackandforth betweenthedesignanduseof technologiesandreflectionuponthoseactivities.”

3.2 EARLY MEDIA SPACES

Theideaof avideophonehasbeenaroundfor a long time.In theearly1960s,AT&T demon-strateda prototype“PicturePhone”at theSeattleWorld’sFair, which allowedcallersto vieweachotheron small video monitors,setup in expandedtelephonebooths.The set-upopti-mizedlighting conditionsandvideocamerapositionto simulateface-to-facecontact.(Notethatcallerscouldnot actuallycall someonethey knew; they hadto wait for a strangerat an-otherPicturePhoneboothto arrive beforehaving a conversation.)Althoughtoutedto bethephoneof the future, it never really caughton andwasultimatelydeemeda failure [Nol92].Anotherinterestingexperimentwasthe Hole-In-Spaceby videoartistsGalloway andRabi-nowitz [Gal80]. They createda real-timevideo/audioconnectionbetweentwo sitesin LosAngelesandNew York. Pedestrianswalkingby couldseefull-size imagesof peoplewalkingin thecorrespondinglocation3000milesaway. Peoplenot only stoppedandstared,but oftenwould respondto theremoteconversationandbegin talking to thepassersbyat theotherend.

Theterm“Media Space”wascoinedby Stultsandhiscolleaguesat Xerox PARC [Stu86],who developedwhat wasprobablythe first real mediaspace.The costof video hadbegunto drop in the 1980s,makingit possibleto link a laboratoryin Palo Alto, California with arelatedlaboratoryin Portland,Oregon.Stultsdefinedamediaspaceas:

“An electronicsettingin whichgroupsof peoplecanwork together, evenwhenthey arenotresidentin the sameplaceor presentat the sametime. In a mediaspace,peoplecancreatereal-timevisualandacousticenvironmentsthatspanphysicallyseparateareas.They canalsocontroltherecording,accessingandreplayingof imagesandsoundsfrom thoseenvironments.”

TheXeroxMediaSpacewasoriginally designedto modeltheinformal typesof communi-cationthatoccurin hallwaysandin commonareas,re-establishingthepossibilityof informalcommunicationfor peoplelocatedapartfrom eachother. Thegoalwasto createatechnology-supportedanalogto themailroomor cafeteria;placeswherepeoplenaturallycongregatein-formally andchat,with oneconversationleadinginto another. An importantaspectof thismediaspacewasthat the connectionswerealwaysthere:only the peoplecameandwent.Conversationsor meetingsdid not have a formal startor stop; they simply representedon-going interactionsamongpeople.Subsequentmediaspaceresearchhasemphasizedtheroleof informal interactionasits key goal,althoughmany have beenextendedto includefacilitiesfor sharedworkspacesandcoordinatedcommunication.

Theperiodof thelate1980sandearly1990swasanactive periodin mediaspaceresearch.Several laboratoriesembarkedon major long-termprojectsin which membersof the labo-ratoriesboth developedand lived in their mediaspaces.Although they sharea numberof

58 MACKAY

characteristics,eachmediaspacewasclearlyshapedby theparticularsocialandphysicalen-vironmentsin which they wereestablishedandreflectdifferentresearchgoals.Someof thegroupscollaboratedclosely, with researchersmoving back and forth betweenlaboratories.In particular, Xerox PARC, EuroPARC, University of TorontoandUniversite de Paris-Sudsharedresearchers,softwareandnumerousdesigndiscussions.

Bly et al [Bly93] describetheir experienceswith theXeroxPARC MediaSpace,includingthe three-yearexperimentlinking their laboratorywith Portlandandthe on-goingevolutionof the Media Spaceeven after the Portlandlaboratorywas closed.In the beginning, theyusedphysicalbuttonsto establishor replacevideoconnections.Lab membersexperimentedwith differentconfigurationsandhandledprivacy in a very mechanicalway, by turning offthemicrophoneor turningthevideocameratowardsthewall. They examinedhow their ownsocial and work relationshipschangedas they usedthe Media Spaceand highlightedtheneedfor additionalresearchin usertailorability of the interfaceandsupportfor managingprivacy issues[Ols91b]. Over time,theMediaSpacewasexpandedfrom four offices,severalpublic areasandthe link to Portland,to includemultiple officesin both sitesanda varietyof video devicesattachedto the network.This pioneeringwork at Xerox PARC influencedthe developmentof the next major mediaspace,createdat PARC’s sister lab in England,EuroPARC.

3.3 RAVE: EUROPARC’S MEDIA SPACE

Rank Xerox EuroPARC was foundedin 1987 as a laboratoryof Xerox PARC, locatedinCambridge,England.The building, RavenscroftHouse,has27 roomswith five openareasspreadover four floors. Eachfloor hastwo “pods” separatedby a centralstairwell, whichcausesa surprisingdegreeof isolationamonglab members.The layout of the lab simulatessomeof theproblemspeoplefacewhenthey mustwork together, but arephysicallyseparated.The lab decidedto encouragecooperative work andfostersocialinteractionby offering labmembersubiquitousaudio,videoanddatainterconnectivity within thebuilding [Bux90]. Thesmall size of the laboratory(approximately30 staff and researchers)madeit possibleforeveryoneto have a mediaspacenode:everyonelivedandworkedin boththephysicalspaceand the mediaspace.This global participationenabledEuroPARC to explore a variety ofsocialaswell asuserinterfaceandtechnicalissuesandprovidedinsightsinto how to providesimilar levelsof socialcontactfor peopleworkingtogetherbut ata distance.

RAVE (the RavenscroftAudio VisualEnvironment)wasnot designedto replaceface-to-facecommunicationbut ratherto supportwork andsocialinteractions,rangingfrom informalcasualencountersto formal plannedcooperative tasks[Gav92b]. RAVE wasbuilt with off-the-shelfanalogaudio and video technology, using several kilometersof coaxial cabletoconnectall analogdevices to a computer-controlled64 � 64 analogswitch. This approachprovidedveryhigh-qualityvideoimagesandstereosound;butwaslimited toasinglebuilding(extendingit furtherwouldhave beenprohibitivelyexpensive)andrequiredamajorrecablingeffort whenever nodesweremoved.Figure3.1shows thebasicset-upof EuroPARC’sRAVEmediaspace.

Eachoffice andmany of thecommonareaswereequippedwith mediaspace“nodes”withaPAL videocamera,amonitor, amicrophone,amixer to handlemultipleaudioinputs,stereospeakersandanoptionalfoot pedalfor controllingaudio(Figure3.2).Audio andvideocon-nectionsweremanagedfrom client applicationsrunningon eitherLISP machinesor, later,

MEDIA SPACES 59

Ethernet

coaxial cable

64x64analogswitch

ifff server

RS232

Figure 3.1 The RAVE mediaspaceconsistsof individual nodesconnectedvia coaxial cable to acomputer-controlledanalogswitch

Figure3.2 A typicalconfigurationfor ananalogmediaspacenode,with avideocamera,microphone,videomonitorandworkstation

Unix workstations.Usershadcompletecontrolover thepositionof theequipment,includinglocationof camerasandmicrophones.They couldturn equipmenton or off, eitherelectron-ically or physically(e.g.puttingon a lenscapor unpluggingit). Somenodeswereequippedwith additionalvideopicture-in-picture(PIP)hardware,whichpermitssimultaneousconnec-tionswith up to four videosourceson thesamemonitor. Connectionsto remotemediaspaceswerecreatedby connectingdigital codecsto theanalogswitchvia ISDN lines.

The iiif server [Bux90], runningonUnix, controlledtheanalogswitchandmanagedaudioandvideoconnectionsamongmediaspacenodes,asrequestedby client applications.In ad-dition to allowing easypoint-to-point connectionswithin the building, iiif alsoguaranteedprivacy andsecurity. Dif ferentversionsof the iiif server wereusedto setup mediaspacesatXerox PARC, theUniversityof TorontoandtheUniversitedeParis-Sud,with individualizedclientapplicationsdesignedaccordingto thetechnicalandsocialneedsateachlocation.

60 MACKAY

Figure3.3 Labmemberscanglanceateachotheror maintainlong-termconnections,suchasthisonebetweentwo administrativestaff members

3.3.1 The RAVE User Interface

TheRAVE userinterfaceenabledusersto displayviewsfrom differentvideocameraslocatedaroundthebuilding and,if givenpermission,setup a two-wayaudio-videoconnectionwithany othernodeconnectedto theanalogswitch.Figure3.3 shows oneof thelongest-runningconnections,betweenthe receptiondeskon the first floor andthe personalassistantto thedirectoron the top floor. They establisheda permanent“Office Share”,with continuously-availablevideoof theotherperson.Audioconnectionsweremadeonly whenafoot pedalwaspressed,to increaseprivacy. Whentheparticipantswantedto glanceat othersin thebuildingor makeothervideoconnections,they did sodirectly andthenreturnedto thedefaultOfficeShare.Othermembersof the lab cameto rely on their Office Shareconnection,usingit totalk to eachotherandavoid runningup anddown thestairs.Theeffect wasof a permanent“hole-in-space”,changingeveryone’s psychologicalperceptionof the physicallayoutof thelab.

The userinterfaceto RAVE evolvedover the years,asusersrequestednew functionalityandwhentheentirelabshiftedfromaLISP-machine-basedto aUnix-basedenvironment.Theoriginal RAVE interfacewasbasedon user-tailorableon-screenbuttonsthataccesseddiffer-entfunctionality. Tailorability wasparticularlyimportant,allowing usersto exploredifferentkindsof connectivity andexpressindividual differences,ensuringeveryonea choicein howthey wererepresentedwithin themediaspace.

The Buttons interfacegrew out of researchat Xerox PARC [Hen86] and EuroPARC[Mac90]. Insteadof typing commandsor selectingfrom a menu,userscouldinteractwith anon-screengraphicalobjectthatranrelevantcommands.They couldlook insidethebuttonandtailor its functionality, aswell aschangeits appearance,copyor evene-mailit to otherusers.Sincebuttonscouldbeparameterized,userscouldchangeapplication-specificvariablesandedit theencapsulatedcode.Thisflexibility allowedlab membersto exploretheRAVE mediaspaceanddevelop theservicesthatweremostusefulto them.Theearliestbuttonsprovidedrelatively low-level functionality, suchas makingor breakinga specificconnection.Over

MEDIA SPACES 61

Figure3.4 RAVE buttonsprovideuserswith servicesthatreflectvaryingdegreesof engagement

time, buttonsevolved higher-level functions,providing encapsulatedservices,with built-inassumptionsabouthandlingissuessuchasprivacy.

Oneof themostinterestingfeaturesof RAVE is theability to shift easilyfrom peripheraltofocusedviews.Thefivebuttonsshown in Figure3.4offer differentlevelsof interactionbasedon the level of engagementrequired:from Backgroundviews operatingat the peripheryofattention,to theunobtrusivepresenceof aSweepthroughthebuilding or aninformalGlance,to thesharedawarenessenjoyedby participantsin anOfficeShare, to thefull engagementofa Vphoneconversation.

Vphone:A highly-focusedform of interactionwith two-wayaudioandvideoconnections.Like telephonecalls,onepartymustexplicitly initiate thecall andtheothermustexplicitlyacceptthe connection.The call endswhenoneparty hangsup. Participantsusedthis whenthey wantedto discussa specifictopic.

Office Share: The physicalconnectionis technicallyidentical to a Vphonecall; the dif-ferenceis that theparticipantsdecideon theconnectionin advance,afterwhich they do notexplicitly initiate or terminateindividualcalls.Participantschoosewhetheror not to includepermanentaudioaswell asvideoconnections.OfficeSharesfacilitateda rangeof communi-cation,from passive awarenessto highly-focusedinteraction.Long-termOffice Shareusers,with connectionslastingfor monthsor evenyears,claimedit waslike sharingaphysicalofficewithoutmany of theannoyances.

Glance:A brief (three-second)one-wayvideoconnectionto anothernode.Thepersonbe-ing glancedat first hearsan audiocue(or the nameof the person),thenthe connectionisestablished,afterwhich anotheraudioclue indicatesthat theGlanceis complete.Lab mem-bersdefinein advancewho haspermissionto glanceat them.Glancesprovideda quick andunobtrusive methodof determiningwhetheror not someonewasaroundor currentlybusy,similar to glancinginto someone’sphysicaloffice.

Sweep:A brief (one-second)one-wayview of a seriesof pre-authorizednodes,local ordistant.Userscouldcustomizetheir sweeppatternsto includethemostusefulpublic (and,ifauthorized,private)nodes,in orderto find outwhowasaroundandgenerallywhatwasgoingon in thelab.

Background:A long-termview of a particularlocationthatactsasthedefaultview of themediaspace.Technically, Backgroundis indistinguishable from an Office Share.However,most Backgroundconnectionsare of public areasthat do not requirespecificpermission,unlikethepre-arrangedconnectionsto aparticularoffice.Althoughtheview fromtheroofwaspopularwith peoplein windowlessoffices;themostpopularBackgroundwastheEuroPARC

62 MACKAY

Figure3.5 Userscanselectwhohastheright to glanceat themthroughtheRAVE mediaspace

commons.Sinceeveryonein thelabvisitedthecommonsregularly to checkmail or getacupof coffee,it providedeveryonewith a low-level consciousnessof whowasin thebuilding andwheninformal gatherings(suchasafternoontea)hadstarted.

Godard[Dou91a] definedandcontrolledtheseandotherservices,directingall connectionsmadeby theiiif server. Basedonpreferencesexpressedby users(suchaswhohadpermissionto glanceat them),Godardwould determinewhetheror not to performspecificrequestsforservices,requestingadditionalinput whennecessary. Beforeperforminganauthorizedcon-nection,Godardwould recordthe previousconnection(suchasanOffice Shareor Vphonecall), ensuringthattheuserwould returnto thecorrectstateaftertheservicewasperformed.

Oneof theprimarybenefitsof thisarchitectureis thatparticipantscaninteractwith RAVEin termsof high-level servicesratherthanlow-level physicalconnections.Eachservicein-cludesa representationof the user’s intentionsandmakesit possibleto embedinformationusedto protectprivacy. Participantscandecidein advancewho hastheright to performeachservice,asopposedto makingon-goingdecisionsaboutlow-level connections,which helpsto balancethetradeoff betweenprivacy andaccess.Figure3.5showsanexampleof a servicecontrol panel.In this case,the userlooks at a list of lab membersandindicatesthosewhohave theright to establishaGlanceconnection.

In practice,thedefaultsettingsof suchlistsareveryimportant.If thedefaultis thateveryonehastheright to glanceunlessexplicitly deletedfrom thelist, a refusedGlancerequestcanbetakenasanexplicit, personalrejection.However, if thedefaultis thatno onehastheright toglanceunlessspecificallyaddedto the list (asat EuroPARC), thenrefusedGlancerequestsmaysimply indicatea non-updatedGlancelist, which is lesslikely to beviewedasaninsult.This posesparticularproblemsfor new membersof thelab. SincepeoplerarelyupdatetheirGlancelists, evenduringprojectandgroupchanges,newcomerswill find themselvesunableto glanceatmany of theircolleaguesandfeelexcludedfromthemediaspace.Theresultis thatlong-termmembersmayhave muchgreateraccessto themediaspacethannewer members,often without realizing it. Seemingly-innocuousdecisionsaboutdefaultsettingsmay havelong-termeffectson theultimateacceptanceof a mediaspacewithin anorganization.

MEDIA SPACES 63

3.3.2 Notification with Auditory Cues

Lab membersnot only wantedcontrol over the typesof connectionsthey made,but alsowantedfeedbackaboutwhenandwhat kinds of connectionswere in use.BecauseGodardhadsomeunderstandingof the participants’intentions,it couldnot only distinguishamongphysicallyidenticalconnectionssuchasVphone,Office ShareandBackground,but it couldalsodetermineanddeliver the appropriatefeedback.Userscould requesta varietyof notifi-cationtypes,suchaspresentinga messageon their workstations.However, themostpopularnotificationsweremoresubtle.Gaver’s work on auditoryicons[Gav86] andtheaffordancesof audio[Gav91b] wasincorporatedinto RAVE, providing real-worldauditorycuesthat in-dicatedwhatwasgoingon. For example,whensomeoneglancedat anotherperson,Godardtriggereda sound(thedefaultwasthatof a dooropening)astheconnectionwasbeingmade.Threesecondslater, whenthe Glancewasterminated,anothersoundwastriggered(usuallya door closing).Other soundswereassociatedwith otherkinds of connections,indicatingthe correspondingintent.A knock or telephonering signaleda Vphone,footstepsindicateda sweepand a camerawhir flaggedwhen a single-framesnapshothad beentaken.Gaver[Gav92a]explainswhy real-worldsoundsareparticularlyeffective:

� Soundindicatesthe connectionstatewithout requiringsymmetry;providing informationwithoutbeingintrusive.

� Soundsdonot requirethekind of spatialattentionthata writtennotificationwould.� Non-speechaudiocuesoftenseemlessdistractingandmoreefficient thanspeechor music

(althoughspeechcanprovidedifferentsortsof information,e.g.,who is connecting).� Soundscanbe acousticallyshapedto reduceannoyance[Pat89].Most soundsinvolve a

gradualincreasein loudnessto avoid startlinglisteners.� Finally, caricaturesof naturally-occurringsoundsareanintuitiveway to presentinforma-

tion. Thesoundof anopeningandclosingdoor reflectsandreinforcesthemetaphorof aglanceandis thuseasilylearnedandremembered.

A numberof otherresearchershave exploredtherole of audioin distributedcollaborativework andmediaspacesettings.ResearchersatPARC [Mor97] andMIT [Kob97] exploredtheproblemof browsingfor audiodata.Seligmannet al [Sel95] examinedthe cuesthat peopleuseto understandordinarytelephonecallsandthenlookedat themorecomplex informationneedsrequiredin multi-party, multimediaconversations.They found that needsfor “assur-ance”becamemorecomplex andthatusersneededinformationaboutconnectivity, presence,focusandwaysof distinguishingbetweenrealandvirtual activities.

Godard,with auditorycues,providedcontrol,feedbackandintentionality, threeprerequi-sitesfor privacy, at very little cost in termsof intrusiveness.Godarduseda systemcalledKhronika[Lov91] to handleauditoryevents.Khronika is an“eventnotificationservice”thatsupportsselectiveawarenessof plannedandelectronicevents,announcingwhenavideocon-nectionhasbeenmade,remindingpeopleaboutupcomingmeetings,providing informationaboutvisitorsandevengatheringpeopleto go to thepub.

Khronikawasbasedon threefundamentalentities:events,daemonsand notifications, asshown in Figure3.6.Eventswereorganizedwithin a classhierarchy, eachwith a class,starttime and duration.Specificevents includedmeetings,visitors, arriving e-mail and RAVEGlances.Eventscouldalsobemanipulatedasmoreabstractclasses,suchas“professional”,“electronic”and“entertainment”.Eventdaemonsproducednotificationeventswhenthey de-tectedspecifiedeventtypes.Userscouldconstraindaemons,enablingthemto selectonly the

64 MACKAY

events

daemons

eventsnotifications

queryof events

of events

even

ts

enter/query

daemons

Senders Khronika Recipients

Figure 3.6 Khronika maintainsan event databaseenteredboth by people and on-line systems.Daemonswatchfor specifiedeventsandpostnotificationswhenthey aredetected

eventsthey wishedto hearabout.For example,a usercouldcreatea daemonto look for allEuroCODEprojectmeetingsin theconferenceroomandgeneratea notificationfive minutesprior to themeeting.Userscouldalsospecifydifferentnotifications,suchasane-mailmes-sagea dayprior to animportantmeeting,with a pop-upscreenmessageanhourbeforeandasynthesizedspeechmessagefive minutesbeforethestarttime.

General-purposenon-speechaudiocueswerealsopopular. For example,the soundindi-catingthestartof ameetingbeganwith soundof peoplemurmuringat low-volume,followedby a gavel soundto indicatethe precisestart time. Suchsoundsprovided low-level periph-eralawarenessof events,enablingpeopleto shift their attentionto themwhennecessaryandignorethemotherwise.Gaver et al [Gav91a] foundthatnon-speechaudiofeedbackchangedbothparticipantsperceptionof thesystemandtheir tendency to collaboratewhile usingit.

A button interfaceto Khronika let usersbrowsethe event databaseaswell ascreateneweventsanddaemons.Anotherinterface,xkhbrowser, showeda calendarwith eventsspanningdifferentperiodsof time. Eventscould be displayedaccordingto their level of specificity,enablingusersto quickly view thekindsof eventsthey wereinterestedin. We comparedtheuseof two EuroPARC calendarsystems[Dou93]: Khronika’s and the paper-basedsystemmanagedby theadministrativestaff. Wediscoveredthateventhoughtheinformationwasos-tensiblyidentical,userswereinfluencedby their knowledgeof thesourceof theinformation,with correspondinglydifferentlevelsof trust in differentkindsof information.If therewasaconflict betweenthe two systems,userswould try to determinewhich calendarsystemwasmorelikely to beaccurate,giventheparticularpieceof information.(For example,thepersonin chargeof thebrownbaglunchseminarserieswasknown to input theeventsinto Khronika,souserstendedtobelievetheKhronika-basedinformation.In contrast,theadministrativestaffwouldtrackpeople’s travel schedules,thusthepaper-basedcalendarwasassumedto bemoreaccuratefor thatkind of information.)Thusthe socialcontext playedasimportanta role astechnologicalfunctionalityin determininghow usersinteractedwith eachsystem.

3.3.3 Long-DistanceAwareness

EuroPARC maintaineda closeassociationwith her sister laboratoryin the United States,Xerox PARC. Maintaininglive videoconnectionswastoo expensive, somembersof thelabinvestigateddifferentwaysof linking thetwo mediaspaces.

Polyscope[Bor91] distributedlow-resolution(200 � 150 bits) digitized imagescapturedapproximatelyevery five minutesfrom eachmediaspace(assumingthe owner had given

MEDIA SPACES 65

Figure 3.7 Portholespresentsa collectionof recently-capturedstill imagesfrom themediaspacesatEuroPARC andXeroxPARC

permission).A simple animationfacility loopedseveral imagesin sequence,to provide ajerky, but usuallyeffective, senseof motion anda way to disambiguatescenes.The buttoninterfaceallowed usersto selectan imageand immediatelyestablisha Glanceor Vphoneconnection.

Portholes[Dou91b] wasdevelopedto help lab membersstay in touchby sharingmore-frequentlycapturedstill imagesfrom therespective mediaspaces(Figure3.7).

Both Polyscopeand Portholesallowed multiple nodesfrom several remotelocationstobe presentedsimultaneously, providing passive awarenessof distributedworkgroupswith-out makingexplicit video connections.They offeredspatially-distributedbut asynchronousfunctionality,whichcomplementedthesynchronousbutsingle-channeledvideoservicesfromeachmediaspace.

3.3.4 Observationsof Use

Usersdefinethe social protocolssurroundingthe useof their electroniccommunications.For example,peopletry to avoid annoyingeachother. In the earlydaysof electronicmail,outsidersweresurprisedthat peopleworking physicallynext to eachotherwould still sendeachotherelectronicmail. Thiswasnot dueto a preferencefor technology-basedratherthanhuman-basedcommunication;it wassimply a matterof courtesy. Calling over the wall ortelephoningis aninterruption,whereassendinge-mailallowstherecipientto respondat theirconvenience.

Similarsocialprotocolsdevelopedin responseto themediaspace.Over time,mediaspaceusersbegan to changeit, creatingusesnot originally anticipatedby the system’s creators.Subtlecharacteristicsin the technologysuggestednew usesor resultedin changesin howthesystemcameto beunderstood[Suc91].An interestingexamplewastheuseof theRAVE

66 MACKAY

mediaspacefor “projecting” presence,taking advantageof theknowledgethat membersofthelabhada sharedperipheralawarenessof theEuroPARC commons.Peopleusuallywaiteduntil a critical massof peoplehadassembledin the commonsbeforeappearingin person.Peoplethusshowed up in two waves: the first few arrived.Whenthreeor four werethere,suddenlyeveryoneelseappeared.Peopletook advantageof their ability to work until thelastminute(andavoided“wastingtime”) andwerestill assuredof arriving whenthemeetingactuallystarted.

Sitting in the commons,in view of the camera,is a way for a researcherto broadcasthisor her availability, letting colleaguesknow that it is acceptableto comeup and chat.Forexample,I hadaboutan hour beforeI neededto catcha planeandfive peopleI neededtotalk to beforeI left. I decidedto sit in thecommonsandfoundthat,one-by-one,all five cameto talk. Eachpersonmonitoredmy meetingson their monitorsandcameupwhenthey couldseeI wasreadyto talk to someoneelse.Everyonecoordinatedtheir activities, managingtofind appropriatetimesto meet,withoutwastingtheir timewaiting for meor wastingmy timewaitingfor them[Mac92].

Interestingly, whenmembersof theadministrativestaff satin thecommons,their messagewastheopposite.They wereonabreakandit wasnotacceptableto askthemto dosomething(althoughit wasfine to have a chat).Thecommonshadanexplicitly “video free” section,sothatpeoplewhowantedto avoid beingseencoulddosoeasilyandnaturally. Theoveralleffectwasthatof having thecommonarearight outsideone’sdoor, but withoutthenoise.Wheneverthelink to thecommonsbrokedown,suchaswhentheequipmentwasbeingupgradedor therewaswork on thebuilding, membersof thelab reportedthesenseof beingslightly disorientedandfeelingoutof touch.

Sharingthe samephysicaloffice with someonecan be annoying,especiallyif you havedifferenttastesin musicor areproneto talking all the time. Office Shares,particularly forpeoplewhoworkedlateatnight,provedto beextremelycomforting.Without listeningto eachother, we couldstill senseeachother’s presence(andwhentheotherpersonwasreadyfor abreak).To a somewhat lesserextent, the Portholesconnectionto PARC provided a similarsenseof comfort,sincesomeonewasalwaysthereno matterhow late theEuroPARC crowdworked(giventhenine-hourtimechangefrom Englandto California).

HeathandLuff [Hea91]observed lab membersusing long-termRAVE connectionsandfoundthatvideosometimesunderminedtheeffectivenessof subtlecommunicative gestures.For example,sincethe cameraandmonitor areoffset,a personlooking at the monitor willappearto belooking down slightly whendisplayedon theothermonitor. Experiencedmediaspaceuserslearnedto shift theirgazebackandforth betweenthe“natural” view for them(i.e.looking at the personon the monitor)andthe “effective” view (i.e. creatingthe appearanceof eye-contactby looking directly into thecamera).Visitorsenteringanofficewith anOfficeSharewouldsometimesbeconfused,thinking thepersonon theotherendof themediaspaceconnectionwaslookingat themwhenthey werein fact looking elsewhere.

Gaveretal [Gav93]examinedtheeffectof givingusersachoiceamongfourdifferentviews,ratherthanthe usualsingle-camera,face-to-faceview. Theadditionalviews includedan “incontext” view, showing peopleandobjectsin relationshipto their workspace,a “deskview”,usingahigh-resolutionmonochromecamerato view documentsandeithera“dollhouse”viewspecificto theexperimentaltask,or a“birds-eye” view showing mostof theroom.They foundthatface-to-faceviewswererarelyusedwhenpeoplewereactively involvedin acollaborativetask; the exceptionwaswhenparticipantsengagedin negotiationabouttaskstrategy. This

MEDIA SPACES 67

study suggeststhat the camerasetupfor mediaspacesshouldchangewhenuserswant toengagein collaborative tasks.

3.4 OTHER MAJOR MEDIA SPACES

Severalothermediaspacesweredevelopedin thesametime frameasRAVE. TheUS WestAdvancedTechnologiesTelecollaborationproject[Bul89] wassimilar to thePARC/Portlandlink, supportinga smallgroupof peoplesharingseveralprojectswhowerelocatedin DenverandBoulder, Colorado.Themediaspaceincludedseveraloffices,aconferenceroomandpub-lic areasat bothsites.Userscould“call” to geta privateoffice-to-officeaudio/visualconnec-tion, “look around”to getavideo-onlyconnection,and“videoconference”to supportmultipleparticipantsin theconferencerooms.Publicareaswerecontinuouslyconnected,asatPARC.

BellCORE was also very active in mediaspaceresearchat this time, creatingCruiser[Roo88] and the VideoWindow [Fis90], both controlled by a systemcalled Rendezvous[Hil94]. In contrastto theXerox PARC approach,which emphasizedletting theusersevolvethecharacteristicsof themediaspace,BellCOREresearchersfollowedatheoreticalapproach,focusingon the role of informal communication[Kra88, Fis93]. Cruiserwasbasedon themodelof walking down a hallwayandglancinginto openofficesto seewho wasthere.Allconnectionswerereciprocal,in that the persondoing the glancingwasalwaysseenby thepersonbeingglancedat.Participantscouldcontrol theaccessto their imagesandcouldalsoestablishtwo-wayconnectionsin thecourseof a “cruise”.Cruiserwasdesignedto encouragespontaneous,informalcommunication,but oftenresultedin longer-termOffice Share.Coop-erstock[Coo92] reportsontheiterativedesignof Cruiser, describinghow usersanddesignersinfluencedthedesignof thesystemover four iterations.

VideoWindow wasmoresimilar to thePARC/PortlandMediaSpacelink, with two large-screendisplayslocatedin two public areason differentfloorsof the researchbuilding. Thelink wasavailablecontinuouslyfor threemonthsandwasdesignedto supportinformal com-municationamongthe 50 researchersandstaff in the area.Peoplewould arrive to get theirmail or have a cupof coffeeandengagein conversationwith thepeoplephysicallypresentaswell asthepeoplelocatedat adistance.

TheMontagesystemfrom SunSoft[Isa93]exploredhow to usevideoto helpmembersofdistributedgroupsdevelop a senseof “teleproximity”, helpingcollaboratorsfind opportunetimesto interactwith eachotherby usingreciprocalglancesto “peekinto someone’s office”.Thesystemalsoprovidedaccessto anon-linecalendar, e-mailandon-screennotefacility. Aswith Cruiser, videoconnectionsappearedin a smallwindow on thecomputerscreen,ratherthanon a separatemonitor. Researchersfoundthatmostglancesdid not resultin interactivecommunication[Tan94].Issacs[Isa95]reportsontheirexperiencesusingvideoin abroadcastsetting,asopposedto in asmallerforumwith a localaudience.They foundthattheaudiencespreferredwatchingthemultimediapresentationsby speakersin the broadcastsetting,whilethespeakersthemselvespreferredtheintimacy of local talks.

The University of Toronto engagedin two major mediaspaceprojects,in collaborationwith researchersat Xerox PARC and EuroPARC. CAVECAT [Man91, Gal91] was imple-mentedwith softwarefrom EuroPARC’s RAVE systemand supportedapproximatelytenoffices within a single building. In addition to the mediaspacefeaturesfrom RAVE, re-searchersexploredtheproblemof integratingshareddrawing facilities with sharedpresence[Pos92]. Sellen[Sel92a] studiedthe speechpatternsin video-mediatedconversations.She

68 MACKAY

Figure 3.8 Hydra incorporatesa video camera,monitor and directionalmicrophoneinto a smalltable-topunit thatcanactasaproxy for distantparticipantsof ameeting

comparedsame-roomandvideo-mediatedconversations,using two interfaces,including asystemcalledHydra [Sel92b]. Figure3.8 shows threeHydra units,eachwith a small videocamera,monitoranddirectionalmicrophone,whichactasproxiesfor distantmeetingpartici-pants.Realandvideo-mediatedparticipantscouldreactto eachotherasthey wouldif all wereco-presentin thesameroom.Sellenfoundquantitativedifferencesbetweenface-to-facecon-versationsandthetwo video-mediatedinterfaces.Althoughtherewerenosignificantquantita-tive differencesbetweenthetwo video-mediatedinterfaces,thereweresignificantqualitativedifferences,with userspreferringthe Hydra interface.OlsonandOlson[Ols95] studiedtherole of addingvideo to remote-collaborations.They found that usersof audio-onlyconnec-tionshadmoredifficulty communicatingbut thattherewerenobasicdifferencesbetweenthequalityof work performedin face-to-facesettingsandin settingswith bothhigh-qualityaudioandhigh-qualityvideo.

TheUniversityof Toronto’sfollow-onproject,calledTelepresence,experimentedwith me-dia spacesoutsideof a laboratorysettingandincludedstudiesof the waysin which mediaspaceschangedthe social relationshipsamongpeopleworking at a distance[Har94]. BothTorontomediaspacesusedaniconof a door, displayedin variousstatesto indicatetheuser’slevel of accessibility(Figure3.9).A fully-opendoor indicatedthatanyonecouldmakea fulltwo-wayaudio/videoconnection,whereasa doorajarenabledpeopleto glance,but requireda ring or further interactionfrom theuserin orderto makea full two-wayaudio/videocon-nection.Whenthe door wasshut,glanceswerenot authorizedand further interactionwasrequiredto establisha two-wayaudio/videoconnection.Whenthedoorwaslocked,novideoconnectionswerepossible.Yamaashietal [Yam96]describeanotherextensionin whichusersweregiventwo views:awide-angleview to show thecontext of theofficeandamoredetailedshotlinkedseamlesslytogether. They alsoexploredtheuseof sensorsplacedin thephysicalenvironmentto providecontextual cuesto remoteusersof themediaspace.For example,thestateof thephysicaldoorto theoffice(i.e.open,ajar, closedor locked)controlledthestateoftheon-screendooricon.

A laterXerox PARC mediaspace,calledKasmer[Bly93], wascreatedto supporta much

MEDIA SPACES 69

Figure 3.9 Theoriginal doorsinterfaceallowedusersto selectfrom four differentdoorstates:open,ajar, closedandlocked

Figure 3.10 Theinterfaceto theKasmermediaspace.Theuppercontrolpanelallows usersto selectservices.Thelower displaypanelshowsotherusersandtheir currentlevel of accessibility

larger groupof peoplein differentgroupswithin the laboratory, as well asoffering codeclinks to externalsites(Figure3.10). Theunderlyingsoftwarewasborrowedfrom CAVECAT,University of Toronto’s system,andRAVE, from EuroPARC. The systemwasdesignedtobalancefrequent,easycommunicationwithin groups,while alsoproviding lessfrequentcom-municationto distantor external groups.Eachworking group included10–25participantsandmediaspacenodes;eachwith their own socialconventionsandmodelsof use.Adler andHenderson[Adl94] describetheirexperienceswith a9-monthOfficeShareconnectionwithinthis environment.Mynattet al [Myn97] explorethedifferencesbetweenon-lineandphysicalspace,arguing thatmediaspacesreinterpretphysicalspacethroughthepositioningof audioandvideoelementsandarguethatactivities derivedfrom onespacedonot translatewell intootherspaces.

Anotherextensionof the mediaspacework is on-goingat the Universite de Paris-Sud,basedon work theauthorsdid at EuroPARC, PARC andUniversityof Toronto.Mediascape

70 MACKAY

Figure 3.11 The Mediascapemediaspace,with the standarduserinterface(left) andan electronicPost-Itnote(right)

[Rou98] wasoriginallybuilt with theiiif server, but hassincebeenreimplementedwith asetofcustomHTTPservers,enablingusersto embedaccessto themediaspaceinto any documenton theWorld Wide Web.Unlike thenumerous“webcams”thathave appeared,whichshow asingleview fromafixedcamera,Mediascapeis afull mediaspace,with facilitiesfor managingconnectionsamongmultiple sites,notifying usersandcontrollingaccess.Theuserinterfaceis highly customizablesinceit is aplainHTML document.Thestandardinterfaceis shown inFigure3.11(left). Imagesareupdatedevery few minutesasin Portholes.Passingthecursorover animageinitiatesa Glance.A double-clickestablishesa Vphoneconnection,accordingto bothusers’expressedavailability. MediascapeusesthesamedoormetaphorasCAVECAT.In the figure, oneuserhaslockedhis door (bottomleft) andanotherhasleft his door ajar(bottomright).AdditionalservicesincludePost-It,to leaveamessageonsomeone’scomputerscreen,Grab,to graba still imagefrom the mediaspace,andDvideo to sendpre-recordedor live digitizedvideo.ElectronicPost-Itnotesareimplementedby remotelycontrolling therecipient’sWebbrowser(in thiscaseNetscapeNavigator).Theimagein thenotehasthesamecapabilitiesasin theinterface:it updatesevery few minutesandcanbeusedto establishothermediaspaceconnections.

3.4.1 Building Upon the RAVE Media Space

The RAVE mediaspaceprovided an infrastructurefor other researchprojectsas well. Aspartof a EuropeanESPRITproject,calledEuroCODE[Mac95a,Mac98], wewereresponsi-ble for designinga multimediacommunicationsystemfor engineersbuilding a bridgeacrossDenmark’s GreatBelt (Storebaelt)Waterway. This projectdevelopeda radically-differentin-terfaceto the mediaspace:a paperengineeringdrawing. We developedAriel (Figure3.12)whichdetectsindividualengineeringdrawingsvia theirbarcodes.In theprototypeshown, thedrawing wasplacedover a large (A0 size)graphicstablet.A projectoron the ceiling pro-jectedmenusandothercomputer-generatedinformation,includingmediaspaceimages,ontothepaper. Here,theuseris establishinga Glanceconnectionwith theauthorof thedrawing,in orderto discusspossiblechanges.The usercould alsoassociateany audio,video or textinformationwith any partof thedrawing andcapturehandwrittennoteswhich couldbesentto colleagues.

MEDIA SPACES 71

Figure 3.12 Ariel lets constructionengineersaccessthe mediaspaceanda hypermediaannotationsystemvia paperengineeringdrawings.Theuserselectsthemediaspaceoptionfrom thecontrolsectionof thepaperengineeringdrawing (upperleft). Ariel projectsamenuandtheuserselectsGlancewith thegraphicstabletpen,whichestablishesa three-secondconnection

3.5 WAVE: A DETAILED CASE STUDY

WAVE [Pag93] wasan attemptto test the mediaspaceconceptslearnedfrom RAVE in areal-world setting:a distributedproductdevelopmentorganizationwithin a large multina-tionalcorporation.WAVE differedin severalimportantrespectsfrom theRAVE mediaspace.RAVE existedwithin asinglebuilding andwasdesignedto encouragecommunicationamongpeoplewhohadotherformsof informalandformalcommunicationavailableto them.WAVEwasmoresimilar to the original PARC Media SpaceandPortholes,in that the participantsweredistributedin both spaceandtime. (However, Portholeshadto spaneight,sometimesnine, time zones,whereasWAVE involveda one-hourdifferencebetweenEnglandandtheNetherlands.)

RAVE wasableto takeadvantageof point-to-pointanalogvideoconnections,with anana-log videoswitchandkilometersof coaxialcable,to provide high-qualityimagesandsoundwith no delay. Portholeswasrestrictedto still imagesdisplayedon a computerscreen,withoccasionaldial-up links with a low-resolutionvideo link anda ratherannoyingaudiodelay.The distancesbetweenthe WAVE sitescausedus to considerdifferent technicalsolutionsfor distributingvideo,whichhada correspondingimpacton theuserinterfaceandsocialuseof the mediaspaces.Oneresearchgoal wasto find out the acceptablethresholdsfor videoqualityundervariousmediaspaceconditions,giventhebandwidthandcostconstraintsof in-ternationallong-distancelinks.Anothercritical differencefrom RAVE andPortholeswasthattheparticipantswerenot researchers,but engineerscreatinga product.For them,the mediaspacewaslike a telephoneor fax, a technologyto be usedonly if it supportedthe work athand.

Westudiedanengineeringdesigncenterin England,whichtookdesignscreatedin JapanortheUnitedStatesandlocalizedthemfor theEuropeanmarket.Theorganizationwassubject

72 MACKAY

to thetypicalpressuresof any high-techcompany in a highly innovative,competitiveandtur-bulentmarket:They hadto increasecustomersatisfaction,maintainatechnologicaledgeandimprove quality while decreasingcosts.Productdevelopmenthadto reducetime-to-market,streamlineprocesses,adaptto rapid technologicalchange,while makingefficient useof re-sources.

Theorganizationhada sophisticatedtelecommunicationsinfrastructure:a corporatetele-phonesystembasedon leasedlines (usersneedonly dial anextensionto reachothersites);voiceconferencecalls;answeringmachines;beepers;electronicmail; fax; andsophisticated(andexpensive)satellitevideoconferencefacilities.All engineersandadministrativestaff hadeithera workstationor a computerterminalon their desks.Yet, in spiteof this infrastruc-ture,engineersspenta greatdealof their time traveling; travel accountedfor over 10%of theproductdevelopmentbudget.At the time of our study, the travel budgetwasbeingcut andmanagerswereinterestedin finding waysto reducethe needfor face-to-facemeetings.Wemet with the directorof the division and interviewedall of his seniormanagersandmanyof their staff, who often took us on toursof their work areasafter the interview. Interviewsweregenerallyopen-ended,althougheachbeganwith a setof standardquestions,includingthe person’s role in the organization,a descriptionof his or her work (eithera projector afunction),aswell asany communicationbreakdownsandstrategiesfor addressingthem.Wealsoattendedregularly-scheduledliveandvideo-mediatedmeetings[Pag93].

We chosea major productdevelopmentproject in a critical stagewithin its two-yearlifecycle. The productwas designedin Englandandwas being assembledin a factory in theNetherlands,requiringcomplex communicationandcoordinationbetweenthetwo sites.TheEnglishengineersunderstoodthe productdesign,whereasthe Dutch engineersunderstoodthemanufacturingproblemsandmaintainedtherelationswith thelocalsuppliers.We identi-fiedtwo situationswith seriouscoordinationandcommunicationneeds:cooperationbetweendesignandmanufacturingengineersandconfigurationmanagement.After furtheranalysisoftheirwork patterns,weinstalledtwo mediaspaceconnections:adial-upvideophonebetweenthe desktopof an engineerin Englandandthe shopfloor of the factory in the Netherlands,andanOffice Sharebetweenthedesktopsof two peoplesharingadministrative tasksacrossthetwo sites.

3.5.1 Analyzing the Existing VideoconferenceSystem

Early forecastsof the successof videoconferencingandvideo telephony werewildly opti-mistic. Egido [Egi88], in heranalysisof why videoconferencingsystemsfail, citesanearly1970spredictionthat a full 85% of meetingswould be conductedby videoconferencesbythe endof the decade.Yet, videoconferencinghasbeenslow to be accepted,despitemajorfinancialinvestmentsby corporations.Becausepeopleeasilyequatemediaspaceswith video-conferencing,we wereinterestedin how thepeoplein theorganizationwestudiedfelt abouttheirexisting satellite-basedvideoconferencingsystem.

Eachsite(in EuropeandtheUnitedStates)hadaspecialmeetingroomsetupto accommo-datesix peopleat a table,with two videocamerasto captureeachgroupof three.A ceiling-mountedcamerawasusedto transmitimagesof objectsor documents.Imagesof colleaguesat remotesiteswereprojectedon two large video monitorsoppositethe table,with imagesof documentspresentedon a third monitor in the middle.Oneuserlikenedit to beingon atelevisionquizshow, with opposingteamslined up,facingeachother.

MEDIA SPACES 73

Thevideoconferenceroomwasdesignedfor highly-stressfulprojectcheckpointmeetings,which hadpriority over everythingelse.Otherscouldschedulemeetingswhentheroomwasnot alreadybooked,usuallyto addresscritical problemsthatarose.Suchmeetingsgenerallyinvolvedtechnicalpeoplewhomadeuseof theceilingcamerasto discussdesigndocuments.(They hadno facilities for bringinghardwareprototypesinto themeetingroom,nordid theyhave facilities for sharingelectronicdocuments.)

Like Egido,we foundmixed,mostlynegative reactionsto the videoconferencingsystem.High-statusmanagerswere most likely to find it useful: they controlledthe meetingsandappreciatedthereductionin travel costs.For example,duringaperiodin whichcross-Atlantictravel waseliminated,onemanagersaid“It reallycameinto its own duringtheGulf war; [its]usehasreally increasedsincethen.” Anotherfoundit “goodfor sharingproblemsandprojectstatus;[althoughnot] for generalinformationexchange”.

Interestingly, mostothersfound it to be divisive, increasingthe adversarialnatureof therelationshipamongthe participants.Theseuserswereindividual contributorswho usedthesystemto negotiateissuesand solve problems.Several peopledescribedtheir concernsasfollows: “There is lots of friction. If people[already]have positions,beingableto seethemdoesn’t helpto bridgethegap.You seea panelof people;it’ sa stand-off situation.It encour-agesantagonism”andit “is not good for problemsolving . . .you reactdifferently to bodylanguageon it versusface-to-face.Theetiquettechanges.” Onepersondescribedhis weeklyFridaymeetings:“It wouldendtheweekhorribly . . . it wasprettybloody. Emotionsfly acrosstheairwaves.”

In summary, mostpeopleviewed face-to-facemeetingsas the optimal form of commu-nication.Telephoneswereuseful,but only for certainkinds of communication.The videoconferencesystemwasviewedasusefulby upper-level management,but createdadversarialrelationshipsamongthe participants.We were interestedin whetheror not a mediaspace,with its emphasison informal interaction,would provide bettercommunicationandreducetheadversarialqualityof theinteractionsfoundwith thevideoconferencesystem.

3.5.2 DesignCenter – ShopFloor Link

Sincestoppingthe productionline was very expensive, the engineershad a basicrule ofthumb:if aproblemarosethatthey did not think couldbesolvedwith telephoneor fax in lessthanfour hours,thedesignengineergot on a planeandflew to the Netherlands.The mediaspacewasseenasa wayto reducethelatter.

Oneendof the link wason the desktopof a systemintegrationengineerin England,re-sponsiblefor ensuringthatall sub-systemsworkedtogether. He knew mostof thedesignerson the projectandcould quickly contactthe appropriatepersonwhenever a problemaroseon theshopfloor. Theotherendof thelink wastheshopfloor itself. We installedequipmenton a cartwhich couldbemoved to any partof themanufacturingline. We usedtwo codecs,basedontheH.272standard(designedfor desktopvideoconferencingusingpublicISDN net-works),connectedby a 64Kb/sdataline (Figure3.13).Thecodecin Englandwasconnectedto anISDN telephonevia anX.21 interface;theISDN telephonewasusedfor dialingandfordisplayingline statusmessages.Unfortunately, theNetherlandsdid not have ISDN availableat thetime,soweuseda switched64 Kb/s IDN line. This madetheset-upon theDutchsidea bit morecomplicated.We connecteda 64 Kb/s modemto an X.21 controller, which wasconnectedto thecodecandto a VT100 terminal.In orderto dial anddisconnecttheline, theDutchusershadto typeseveralcommandson theVT100terminal.

74 MACKAY

codec

X.21 interf.

64k modem

Dutch PTTIDN

BTISDN

codec

theNetherlands England

VT100terminal

Figure3.13 WAVE link betweenthedesigncenterin Englandandthefactoryin theNetherlands

Figure3.14 Imagecapturedfrom theWAVE link. Notetheuseof thesmallhand-heldcamerato showaclose-upview of theproblem

In England,weinstalledasinglecamera,a monitorwith built-in speakersanda directionalmicrophone.Wealsoinstalledavideotaperecorderto capturevideofrom theNetherlandsandaudioin bothdirections.OntheDutchside,weinstalledtwocameras:astandard-sizedcamerawasclampedto thecart to provide anoverview, anda miniaturecamera(1cm� 5cm)with aflexible cableto show smalldetails(i.e. 2mmsizetype).We alsoinstalledtwo monitors,forincomingandoutgoingvideo.Sincetherewasa greatdealof backgroundnoiseon theshopfloor, weprovidedheadphoneswith a built-in microphone.

Thelink on theshopfloor wasup for two weeks;duringthis period,we spenttwo daysateachsitesittingnext to theequipmentandobservingwhatusersdid. For theremainingtime,wecollectedvideotapesof thevideoandaudiogoingthroughthelink andlaterinterviewedthe

MEDIA SPACES 75

peoplewhousedthelink (Figure3.14).Userscomplainedaboutthepooraudioquality, poorvideoresolutionandthelack of reliability. Yet they wereelatedwhenthey wereableto solveproblemswithout traveling. As one managersaid, “This technologyis a pain . . .Howeveryesterdayweusedit andit savedusa trip, thereforeI will let my peoplespendmoretime onit. I would ratherhave this lousylink thannothing.Therearemany thingsherewhicharenotperfectthat we have to copewith, this will be anotherone.” The following examplesshowhow they usedthesystem:

Showa problem:A manufacturingengineershowssomethinggoingwrongon theproduc-tion line andasksdesignersfor explanations,solutions,or changes.Videois very important,not only to improvecommunication,but alsoto overcometheinitial skepticismandmistrust.Looking at theproblemtogetherandjointly working on thesolutionhelpsto overcomecul-tural differences(not only betweenDutchandEnglish,but alsobetweenmanufacturinganddesignroles),fosteringacooperativeattitudetowardssolvingtheproblemand“gettingthingsdone”,ratherthanarguinganabstractproblemover thephoneto “passtheball”.

For example,following a part change,the packagingalsohad to be changed,makinganew cut into thecardboardandassemblingthe piecesso they would fit together. A packag-ing engineerwasableto usethe video link to show, step-by-step,how to makethe cut andassemblethepieces,while theDutchengineersrepeatedeachstep.In half anhour, a problemwhichwouldhave requiredatrip wassolved.Theengineerswereenthusiasticaboutthevideolink, sayingthat it hadbeenparticularlyusefulto do eachstepof theassemblyat bothsites,watchingeachother’sactions.

Showa solution:A designengineerdemonstratesthe correctway of doingsomethingonthe shopfloor. The video link allows the engineersto go througheachstepof the process,performingit simultaneouslyat both sitesto ensurethat everyoneunderstandsthe solutionandits consequences.Thevideolink is muchfasterandmoredirect;allowing participantsatbothsitesto seethesolution,whichincreasesconfidencein thesolutionandtrustbetweenthetwo sites.

For example,amanufacturingengineershowedaprogrammerin Englandasoftwarebugbypointingthecameraat thedisplayandkeyboardof theproductsothattheprogrammercouldseewhatwasgoingwrong.Theprogrammertold themanufacturingengineerwhich keys topressandthey foundanother, relatedbug. Theproblemmight have beendescribedover thetelephone,but the manufacturingengineerfelt that theprogrammerwasskepticalaboutthebug andassumedthe manufacturingengineerwasdoingsomethingwrong.Using the videolink allowedhim to actuallyseetheproblemandtry thingsout; it alsoallowedhim to locatethebugprecisely.

Cooperativeproblem-solving:A problemis shown andengineersat bothendsof the linkbrainstormsolutions,discussideas,point at causesandtry out experimentson themachine.Unlike a standardvideoconference,theparticipantsrarelylook ateachother’s facesandcon-centrateon thetechnicalproblemto besolved.

For example,thepaperfeedmechanismworkedwell on theprototypes,but did not workreliablyon theunitscomingoff themanufacturingline. Althoughit seemedto bea manufac-turingproblem,themanufacturingengineerswereinterestedin suggestionsfrom thedesign-ers.Six Dutch engineersshowed the problemto threeEnglishengineers,who wereabletobrainstormandtestvarioussolutions.Becausetherewasonly onesetof headphoneson theshopfloor, the Dutch engineerspassedaroundthe miniaturecamerato show thingsandletonepersonhandletheaudiocommunicationto England.

76 MACKAY

Themanagementmadea cost/benefitevaluationof effectivenessof thelink andconcludedthat thesystemwasusefulfor solvingproblemson theshopfloor andthat the link savedatleasttwo tripsduringthecourseof theexperiment.They werenothappywith theunreliabilityof thelink, but said,“when it worksandis usedfor theright application,it is a verypowerfultool”.

3.5.3 Configuration ManagementLink

Thesecondmediaspacelink wasinstalledbetweentwo planneranalysts,locatedin Englandandthe Netherlands.The planneranalystswereresponsablefor configurationmanagement,includingtrackingdesignchanges,evaluatingthecostof partsandchangesandmaintainingthe inventoryof the thousandsof partswhich makeup a productactingasa bottleneckforall designchanges.They registeredeachchange,evaluatedthe costandsubmittedthemtothe weekly ChangeandControl Board (CCB), held at the videoconferencefacility. Seniormanagementat bothsiteswould review andapprove changerequests,basedon cost,timing,quality and technicalissues.The planneranalystsspenta greatdeal of time on the phone(10–30callsperday)andmanaged30–50changerequestsperweek.

We installed an Office Sharetype of link with a continuousvideo connection,activethroughoutthe day. Becausea continuously-availableISDN line would have beentoo ex-pensive, we usedthe corporateTCP-IPnetwork,usingexisting 128 Kb/s leasedlines. Weinstalleda Videopixvideodigitizing boardoneachof theirSunworkstations,usingtwo soft-warepackages:vfctool, whichcamewith theVideopixboard,andIVS, a publicdomainsoft-waredevelopedat INRIA (Figure3.15).

Vfctool grabbedframesfrom a video digitizing boardsharedover a LAN, without com-pression.With the network traffic betweenthe Netherlandsand England,it took up to sixminutesto updatean imagewith a sizeof 320� 240 pixels and8 gray levels. IVS [Tur93]wasdesignedto supportvideo andaudioconferencesover the Internetandachieved higherrefreshratesby compressingtheimagesaccordingto theH.261standard.IVS transmitedthecompresseddatastreamover an IP networkusing the UserDatagramProtocol(UDP) andtook about20 to 30 Kb/s of bandwidth.We usedQCIF images(176� 144pixels)with eightgray levelsandobtaineda refreshrateof oneframeevery two to four seconds,accordingtonetworktraffic. IVS wasvery robust to packetlossandnetworkoverload;the only problemwasthatsometimesthevideowindow wascloseddown.However thesoftwarenevercrashedandtheusercouldrestorethelink with a coupleof mouseclicks.

Thelink to supportconfigurationmanagementranfor a full six days,spreadover a periodof aboutonemonth.We spentseveralhoursobservinguserswhile thelink wasup andinter-viewedthemperiodically. We abandonedVfctool, becausetherefreshratewastoo slow andit wasunreliableasasourceof information.Althoughthey enjoyedputtingupmessagessuchas“Good morning”, “I’ll bebackat 3pm” or “I amon holiday today”, the planneranalystswerefrustratedthat the imagewasusuallyout-of-dateandthat thepersonin the imagewasoften no longerthere.We switchedto IVS, with a smallervideo window andlower imageresolution,but a muchhigher refreshrate.The planneranalystsregularly checkedthe IVSimagebeforecalling eachother (at leastten timesper day). They particularlyenjoyedtheOfficeShareonMondaynights,whenthey wouldoftenwork until midnightpreparingfor theChangeandControlBoardmeetingthenext morning.They saidthat thelink providedthemwith “remotesolidarity”, encouragingthemto drink coffee togetherandkeepworking untilthey weredone.Another, moresubtleaspectwashow they communicatedthat they did not

MEDIA SPACES 77

CorporateTCP/IP network

videowindow

theNetherlands

England

Sun sparkstation& VideoPix board

Videocamera

Headphones

Built-inspeakers

Sunmicrophone

Ethernet

Figure3.15 Office Sharebetweenplanneranalystslocatedin EnglandandtheNetherlands

wantto bedisturbed,by explicitly holdingthephone,actively workingthrougha largepile ofpapersor by moving out of therangeof thecamera.Their informal waysof communicatingwereencodedanddecodedwith noeffort or attention;in mostof thecasesthey wereunawarethatthey werecommunicating.In onecase,aplanneranalystsaid:“YesterdayI saw youweretalkingwith . . . ”, but waslaterunableto rememberthathehadseenthepersonvia theOfficeShare.Otherpeoplein thebuilding alsousedtheOffice Share.Passersbywould wave at theremoteplanneranalystandsometimesusedit to find someoneor talk to someoneat theothersite.Peopleadaptedeasilyto thelink; afteraninitial periodof self-consciousnessthey quicklyforgotaboutthecameraandrespondedto theotherperson.

3.6 ETHICAL ISSUES

Videois a very powerful medium,perhapstoopowerful. Oneof thebiggestissuesis privacy:how do we balancethebenefitsof a relatively openmediaspacewith individuals’needsforprivacy? Privacy issuesaremulti-dimensionalandaregreatlyaffectedby the cultureof theorganizationin which the mediaspaceis placedandthe purposesfor which it wascreated.Gaver [Gav92a]identifiesfour issuesthatmustbedisentangledwhenthinkingaboutprivacy:

� Control: Userswantto controlwhocanseeor hearthematany time.� Knowledge:Userswantto know whensomebodyis in factseeingor hearingthem.� Intention:Userswantto know whattheintentionof theconnectionis.� Intrusions:Userswantto avoid connectionsthatdisturbtheirwork.

Fishet al [Fis93] point out that the tradeoff betweenprivacy andfunctionality involvesaconflictbetweenthedesirabilityof controlandknowledgeandtheintrusionimpliedby activ-ities neededto maintainthem.Explicitly acknowledgingevery connectionprovidescontrol,but the requeststhemselves would be intrusive. Similarly, if every glanceresultsin seeingsomeone’s faceon themonitor, it demandssomesortof socialresponseandmaywell disruptpreviously-existingconnections.If usersmustspecifyandbeinformedof theexact intention

78 MACKAY

of every connection,themediaspaceis no longerlightweightandis boggeddown with con-tinuousdemandsfor theuser’s attention.Thechallengeis to provide userswith controlandnotification,but in a lightweightandunobtrusiveway.

Sinceprivacy issuesareaffectedby the social context in which the mediaspaceis em-bedded,it is not possibleto simply createan “ideal technology”that is appropriatein everysetting.The Xerox PARC MediaSpace,RAVE andWAVE mediaspacesworkedwithin anatmosphereof trust becausethe participantsknew eachotherandworkedtogether. From amanagementperspective, it wasalsovery importantto enforcethe ideathat“turning off themediaspace”wasacceptablebehavior; allowing peopleatall levelsof theorganizationto feelcomfortable.(Contrastthis to theexperiencesof usersof thevideoconferencingsystemin theWAVE study, in whichhigh-level managerswereverysatisfiedwith thesystem,but everyoneelsefound it to be disruptive.) Larger mediaspaces,suchasKasmer,have hadsomeprob-lems,whenpeoplesuddenlyfoundthemselvesbeingglancedat by peoplethey don’t know.Web-basedmediaspacessuchasMediascapehave world-widereach,with correspondinglylower levelsof trustamongtheusersandrequiregreaterlevelsof privacy protection.

Theorganizationsthat createdthe mediaspacesin this chaptereachdevelopedtheir ownsafeguardsto privacy, makingjudgmentsabouthow to balanceprivacy concernswhile stillmakingthemediaspaceworthwhile.In Cruiser[Roo88], all connectionshadto besymmetri-cal; suchthathearingor seeingsomeoneimplied that thatpersoncouldalsoseeor hearyou.TheMediaSpace[Bly93] took theoppositeextreme;all videoconnectionswerefully openwith bothaudioandvideolinks. This workedwell whenthemediaspaceinvolvedclose-knitmembersof a small work group,but erodedwhenothersfrom otherpartsof the organiza-tion joined themediaspaceandwereseento be “voyeurs”.RAVE wasbasedon thenotionof services,suchasGlance,in which the user’s intentionwasincorporatedinto theservice.Usersdecidedin advancewho hadpermissionto glanceat them;letting themavoid givingpermissioneachtime. RAVE specifiedaccesslevels perperson,which wererarely updatedover time, whereasCAVECAT usedthe door metaphorto establishdynamicaccesslevels,but did not distinguishwho requestedwhich type of access.Mediascapecombinesthe twomodels:userscancustomizeaccessrightsaccordingto thecurrentstateof thedooraswell astheorigin of thecall. This is especiallyimportantsinceMediascapeis accessiblethroughtheWorld Wide Web.

Media spacedesignersneedto explicitly considera setof ethical issueswhenhandlingvideo[Mac95b]. Peopleshouldbeinformedof thepresenceof live cameras.(Unfortunately,mostof usarelargely unawareof themyriadsecuritycamerasthatcaptureandrecordvideoof us every time we usea bank teller, shopin a storeor even walk down the street.Suchusesof video increaseour insecuritywhencontemplatingmediaspaces.)At EuroPARC, amannequinwith a signaroundhis neckwasa light-heartedway of letting peopleknow theywerein the rangeof the cameras.Displayingthecamera’s imageon an adjacentmonitor isalsoeffective.Peopleshouldbeableto easilydetectwhenacamerais left onall thetime,suchasin commonsareas.Peopleshouldbeableto figureout whenthey areon cameraandhavetheopportunityto avoid it by moving outof range.Recordingvideois especiallyproblematic,sincevideo takenout of context canbe usedin waysthatmaycauseviewersto completelymisinterpretwhathappened.Peopleshouldknow whenvideois beingrecordedandbegiventhe opportunityto stop.Oncerecorded,peopleshouldhave theability to view the recordedmaterialandconsentto any furtheruseof thematerial,by giving their informedconsent.

MEDIA SPACES 79

3.7 CONCLUSION

Distributedvideois nota single,unitaryphenomenonthatcanunderstoodsimplyat thelevelof thetechnologyit incorporates.Whatis importantis theway in which thevideo(andasso-ciatedtechnologies)aresetupandusedwithin a socialsetting.Mediaspaces,with their em-phasison informal andopen-endedaswell asformal communication,arean importantnewapproachfor supportingdistributedcooperative work groups.Media spacedesignersmustconsiderthe context in which their technologywill beusedandensurethat usescaneasilyadaptthemto meetthespecificneedsof theirusers.Mediaspacesarestill in theirinfancy withmuchresearchto bedone.However, asvideocostscontinueto dropandastheWebbecomesubiquitous,mediaspacespromiseto provide an effective meansfor supportingdistributed,collaborativework.

ACKNOWLEDGEMENTS

TheRAVE mediaspaceat EuroPARC andits extensionswerea collaborativeeffort of manypeople,including Bob Anderson,Victoria Bellotti, Michel Beaudouin-Lafon,Bill Buxton,KathyCarter, IanDaniel,Paul Dourish,Bill Gaver, ChristianHeath,LennartLvstrand,AllanMacLean,Tom Milligan, Mike Molloy, Tom Moran,Toby Morrill, DanielePagani,GaryOl-son,JudyOlsonandRandySmith.OurcolleguesatXeroxPARC in PaloAlto alsocontributedresearchideas,particularlyAnnetteAdler, SaraBly, SteveHarrison,AustinHenderson,ScottMinnemanandJohnTang.

REFERENCES

[Adl94] Adler, A. andHenderson,A., A roomof ourown: Experiencesfrom adirectoffice share.InProceedingsof HumanFactorsin ComputingSystems,CHI ’94 (Boston,MA), pages138–144. ACM Press,New York, 1994.

[Bly93] Bly, S.,Harrison,S. andIrwin, S., Mediaspaces:Bringing peopletogetherin a video au-dio andcomputingenvironment. Communciationsof theACM specialissue,36(1):29–47,January, 1993.

[Bor91] Borning,A. andTravers,M., Two approachesto casualinteractionover computerandvideonetworks.In Proceedingsof HumanFactorsin ComputingSystems,CHI ’91 (New Orleans,LA), pages13–19.ACM Press,New York, 1991.

[Bul89] Bulick, S.,Abel, M., Corey, D., Schmidt,J.andCoffin, S., TheUS WESTAdvancedTech-nologiesPrototypeMultimediaCommunicationsSystem.GLOBECOM’89: In Proceedingsof theIEEE GlobalTelecommunicationsConference(Dallas,Texas),1989.

[Bux90] Buxton,W. andMoran,T., EuroPARC’sintegratedinteractive intermediafacility (iiif): Earlyexperiences.In Proceedingsof the IFIP WG8.4Conferenceon Multi-User InterfacesandApplications(Herakleion,Crete),pages11–34.North-Holland,1990.

[Coo92] Cool,C. Fish,R.,Kraut,R., Lowery, C., Iterative designof a videocommunicationsystem.In Proceedingsof the Conferenceon Computer-SupportedCooperative Work, CSCW’92(Toronto,ON), pages25–32.ACM Press,New York, 1992.

[Dou91a] Dourish,P., Godard:A flexible architecturefor AV servicesin a mediaspace.EuroPARCworkingpaper, 1991.

[Dou91b] Dourish,P. andBly, S., Portholes:Supportingawarenessin a distributedwork group. InProceedingsof HumanFactorsin ComputingSystems,CHI ’92 (Monterey, CA), pages541–547. ACM Press,New York, 1992.

80 MACKAY

[Dou93] Dourish,P., Bellotti, V., Mackay, W. andMa, C., Informationandcontext: Lessonsfrom astudyof two calendarsystems. In Proceedingsof COCS’93 (SanFrancisco,CA). ACMPress,New York, 1993.

[Egi88] Egido,C., Videoconferencingasatechnologyto supportgroupwork: A review of its failure.In Proceedingsof the Conferenceon ComputerSupportedCooperative Work, CSCW’88(Portland,OR),pages13–24. ACM Press,New York, 1988.

[Ehr99] Ehrlich,K., Designinggroupwareapplications:A work-centeredapproach.In Beaudouin-Lafon,M. (Ed.),ComputerSupportedCooperativeWork, Trendsin SoftwareSeries7:1–28.JohnWiley & Sons,Chichester, 1999.

[Ell99] Ellis, C.A., Workflow technology. In Beaudouin-Lafon,M. (Ed.), ComputerSupportedCooperativeWork, Trendsin SoftwareSeries7:29–54.JohnWiley & Sons,Chichester,1999.

[Ens92] Ensor, J.,Ahuja, S.,Connaghan,R., Pack,M. andSeligmann,S., The Rapportmultimediacommunicationsystem. Demo in ConferenceCompanion:HumanFactors in ComputingSystems,CHI ’92 (Monterey, CA), pages49–59.ACM Press,New York, 1992.

[Fis90] Fish, R., Kraut, R. andChalfonte,B.L., The VideoWindows systemin informal commu-nications. In Proceedingsof the Conferenceon Computer-SupportedCooperative Work,CSCW’90 (LosAngeles,CA), pages1–13.ACM Press,New York, 1990.

[Fis93] Fish,R.,Kraut,R.,Root,R. andRice,R.E.,Videoasatechnologyfor informalcommunica-tion. Communciationsof theACM specialissue,36(1):48–61,January, 1993.

[Gal80] Galloway, J. andRabinowitz, S., Hole-In-Space:Mobile imagevideotape. SantaMonica,CA.1980.

[Gal91] Gale,S.,Addingaudioandvideoto anofficeenvironment.In Studiesin ComputerSupportedCooperativeWork. BowersandBenford(Eds.),pages49–62,Elsevier SciencePublishersB.V., 1991.

[Gav86] Gaver, W.W., Auditory icons:Usingsoundin computerinterfaces.Human-ComputerInter-action, 2:167–177.1986.

[Gav91a] Gaver, W.W.,Smith,R.B.andO’Shea,T., Effectivesoundsin complex systems:TheARKolasimulation.In Proceedingsof HumanFactorsin ComputingSystems,CHI ’91 (New Orleans,LA), pages85–90.ACM Press,New York, 1991.

[Gav91b] Gaver, W.W., Soundsupportfor collaboration. In Proceedingsof the EuropeanConfer-enceonComputerSupportedCooperativeWork,ECSCW’91 (Amsterdam,theNetherlands),pages293–308.ACM Press,New York, 1991.

[Gav92a] Gaver, W., Moran,T., MacLean,A., Lovstrand,L., Dourish,P., Carter, K. andBuxton,W.,Realizingavideoenvironment:EuroPARC’sRAVE system.In Proceedingsof HumanFac-tors in ComputingSystems,CHI ’92 (Monterey, CA), pages27–35.ACM Press,New York,1992.

[Gav92b] Gaver, W.W., The affordancesof mediaspacesfor collaboration. In Proceedingsof theConferenceonComputer-SupportedCooperativeWork,CSCW’92 (Toronto,ON),pages17–24. ACM Press,New York, 1992.

[Gav93] Gaver, W., Sellen,A., Heath,C. andLuff, P., Oneis notenough:Multiple views in a mediaspace.In Proceedingsof HumanFactorsin ComputingSystems,InterCHI ’93 (Amsterdam,theNetherlands),pages335–341.ACM Press,New York, 1993.

[Har94] Harrison,B., Mantei,M., Beirne,G. andNarine,T., Communicatingaboutcommunicating:Cross-disciplinarydesignof a mediaspaceinterface. In Proceedingsof HumanFactorsinComputingSystems,CHI ’94 (Boston,MA), pages124–130.ACM Press,New York, 1994.

[Hea91] Heath,C. and Luff, P., Disembodiedconduct:Communicationthroughvideo in a multi-mediaofficeenvironment.In Proceedingsof HumanFactorsin ComputingSystems,CHI ’91(New Orleans,LA), pages99–103.ACM Press,New York, 1991.

[Hen86] Henderson,D.A. andCard,S., Rooms:The useof multiple virtual workspacesto reducespacecontentionin a window-basedgraphicaluserinterface.ACM Transactionson Graph-ics, 5(3):211–243,1986.

[Hil94] RalphHill, TomBrinck, StevenRohall,JohnPattersonandWayneWilner. TheRendezVousarchitectureand languagefor constructingmultiuserapplications. ACM TransactionsonComputerHumanInteraction, 1(2):81–125,June1994.

[Isa93] Isaacs,E. andTang,J., What video canandcan’t do for collaboration:A casestudy. InProceedingsof ACM Multimedia’93 (Anaheim,CA), pages199–206,1993.

MEDIA SPACES 81

[Isa95] Isaacs,E., Morris, T., Rodriguez,T. andTang,J., A comparisonof face-to-faceanddis-tributedpresentations.In Proceedingsof HumanFactors in ComputingSystems,CHI ’95(Denver, CO),pages354–361.ACM Press,New York, 1995.

[Ish92] Ishii, H. andKobayashi,M., Integrationof interpersonalspaceandsharedworkspace:Clear-boarddesignandexperiments. In Proceedingsof the Conferenceon Computer-SupportedCooperativeWork, CSCW’92 (Toronto,ON), pages33–42.ACM Press,New York, 1992.

[Ish99] Ishii, H., Integrationof sharedworkspaceandinterpersonalspacefor remotecollaboration.In Beaudouin-Lafon,M. (Ed.),ComputerSupportedCooperativeWork, Trendsin SoftwareSeries7:83–102.JohnWiley & Sons,Chichester, 1999.

[Kob97] Kobayashi,M. andSchmandt,C., Dynamicsoundscape: mappingtime to spacefor audiobrowsing. In Proceedingsof HumanFactorsin ComputingSystems,CHI’97 (Atlanta,GA),pages194–201.ACM Press,New York, 1997.

[Kra88] Kraut, R. andEgido,C., Patternsof contactandcommunicationin scientificresearchcol-laboration. In Proceedingsof the Conferenceon Computer-SupportedCooperativeWork,CSCW’88 (Portland,OR),pages1–13.ACM Press,New York, 1988.

[Lov91] Lostrand,L., Being selectively aware with the Khronika system. In Proceedingsof theEuropeanConferenceonComputerSupportedCooperativeWork, ECSCW’91 (Amsterdam,theNetherlands),pages265–278.ACM Press,New York, 1991.

[Mac90] MacLean,A., Carter, K., Moran, T. and Lovstrand,L., User-tailorablesystems:Pressingtheissueswith Buttons. In Proceedingsof HumanFactorsin ComputingSystems,CHI ’90(Seattle,WA), pages175–182.ACM Press,New York, 1990.

[Mac92] Mackay, W.E., Spontaneousinteractionin virtual multimediaspace:EuroPARC’s RAVEsystem.Imagina’92, MonteCarlo,Monaco.1992.

[Mac95a] Mackay, W.E., PaganiD.S.,FaberL., InwoodB., LauniainenP., BrentaL. andPouzolV.,Ariel: Augmentingpaperengineeringdrawings. Video in ConferenceCompanion:HumanFactorsin ComputingSystems,CHI ’95 (Denver, CO), pages421–422. ACM Press,NewYork, 1995.

[Mac95b] Mackay, W.E., Ethics,liesandvideotapes.In Proceedingsof HumanFactorsin ComputingSystems,CHI ’95 (Denver, CO),pages138–145.ACM Press,New York, 1995.

[Mac98] Mackay, W.E., AugmentedReality:linking realandvirtual worlds. In Proceedingsof ACMConferenceon AdvancedVisual Interfaces,AVI ’98 (L’Aquila, Italy), pages13–21. ACMPress,New York. 1998.

[Man91] Mantei,M., Baecker, R.,Sellen,A., Buxton,W., Milligan, T. andWellman,B., Experiencesin theuseof amediaspace.In ProceedingsofHumanFactorsin ComputingSystems,CHI ’91(New Orleans,LA), pages203–208.ACM Press,New York, 1991.

[Min91] Minneman,S.L. andBly, S.A., Managinga trois: A studyof a multi-userdrawing tool indistributeddesignwork. In Proceedingsof HumanFactorsin ComputingSystems,CHI ’91(New Orleans,LA), pages217–224.ACM Press,New York, 1991.

[Mor90] Moran,T.P. andAnderson,R.J., Theworkadayworld asa paradigmfor CSCWdesign. InProceedingsof theConferenceon Computer-SupportedCooperativeWork, CSCW’90. (LosAngeles,CA), pages381–393.ACM Press,New York, 1990.

[Mor97] Moran,T., Palen,L., Harrison,S.,Chiu, P., Kimber, D., Minneman,S., van Melle, W. andZellweger, P., “I’ll get that off the audio”: A casestudyof salvagingmultimediameetingrecords. In Proceedingsof HumanFactorsin ComputingSystems,CHI’97 (Atlanta,GA),pages202–209.ACM Press,New York, 1997.

[Myn97] Mynatt,E.,Adler, A., Ito, M. andO’Day, V., Designfor networkcommunities.In Proceed-ingsof HumanFactorsin ComputingSystems,CHI’97 (Atlanta,GA), pages210–217.ACMPress,New York, 1997.

[Nol92] Noll, A., Anatomyof a failure: Picturephonerevisited. TelecommunicationsPolicy, pages307–316.May-June1992.

[Ols91a] Olson,G. andOlson,J., User-centereddesignof collaborationtechnology. Journalof Orga-nizationalComputing, 1:61–83,1991.

[Ols91b] Olson,M.H. andBly, S., ThePortlandexperience:A reporton adistributedresearchgroup.InternationalJournalof Man–MachineStudies, 34,1991.

[Ols95] Olson,JandOlson,G., Whatmix of videoandaudiois usefulfor smallgroupsdoingremotereal-timedesignwork? In Proceedingsof HumanFactorsin ComputingSystems,CHI’95

82 MACKAY

(Denver, CO),pages362–368.ACM Press,New York, 1995.[Pag93] Pagani,D. andMackay, W.E., Bringingmediaspacesinto therealworld. In Proceedingsof

the EuropeanConferenceon Computer-SupportedCooperativeWork, ECSCW’93 (Milan,Italy). ACM Press,New York, 1993.

[Pat89] Patterson,R.D., Guidelinesfor thedesignof auditorywarningsounds.In Proceedingsof theInstituteof Acoustics1989SpringConference, 11(5):17–24,1989.

[Pos92] Posner, I. andBaecker, R., How peoplewrite together. In Proceedingsof theTwenty-fifthHawaii InternationalConferenceon SystemsSciences(Kauai,Hawaii), volumeIV, 1992.

[Pra99] Prakash,A., Groupeditors.In Beaudouin-Lafon,M. (Ed.),ComputerSupportedCooperativeWork, Trendsin SoftwareSeries7:103–133.JohnWiley & Sons,Chichester, 1999.

[Roo88] Root, R.W., Designof a multi-mediavehiclefor social browsing. In Proceedingsof theConferenceon Computer-SupportedCooperative Work, CSCW’88 (Portland,OR), pages25–38.ACM Press,New York, 1988.

[Rou98] Roussel,N., Towardsa toolkit for building mediaspaces.LRI TechnicalReport, UniversitedeParis-Sud,Orsay, France,1998.http://www-ihm.lri.fr/˜roussel/Mediascape.

[Sel92a] Sellen,A., Speechpatternsin video-mediatedconversations. In Proceedingsof HumanFactors in ComputingSystems,CHI ’92 (Monterey, CA), pages49–59. ACM Press,NewYork, 1992.

[Sel92b] Sellen,A., Buxton, W. and Arnott, J., Using spatialcuesto improve videoconferencing.Videoin ConferenceCompanion:HumanFactorsin ComputingSystems,CHI ’92 (Monterey,CA), pages651–652.ACM Press,New York, 1992.

[Sel95] Seligmann,D., Mercuri,R. andEdmark,J., Providing assurancesin amultimediainteractiveenvironment. In Proceedingsof HumanFactors in ComputingSystems,CHI’95 (Denver,CO),pages250–256.ACM Press,New York, 1995.

[Stu86] Stults,R., Mediaspace.XeroxPARCtechnical report, 1986.[Suc91] Suchman,L. andTrigg, R., Understandingpractice:Videoasa mediumfor reflectionand

design.In Designat Work: CooperativeDesignof ComputerSystems, GreenbaumandKyng(Eds),LawrenceErlbaum,Hillsdale,N.J.,1991.

[Tan90] Tang,J. andMinneman,S., VideoDraw: A video interfacefor collaborative drawing. InProceedingsof HumanFactorsin ComputingSystems,CHI ’90 (Seattle,WA), pages313–320. ACM Press,New York, 1990.

[Tan94] Tang,J.andRua,M., Montage:Providing teleproximityfor distributedgroups.In Proceed-ingsof HumanFactorsin ComputingSystems,CHI ’94 (Boston,MA), pages37–43. ACMPress,New York, 1994.

[Tur93] Turletti, T., H.262Softwarecodecfor videoconferencingover theInternet.INRIATechnicalReportNo. 1834,SophiaAntipolis, France,1993.

[Wel93] Wellner, P., Interactingwith paperon the DigitalDesk. In Communicationsof the ACM,36(7):86–96,July 1993.

[Win89] Winograd,T. andFlores,F., UnderstandingComputersandCognition:A New Foundationfor Design. NJ:Ablex, 1986.

[Yam96] Yamaashi,K., Cooperstock,J.,Narine,T. andBuxton,B., Beatingthelimitationsof camera-monitormediatedtelepresencewith extra eyes. In Proceedingsof HumanFactorsin Com-putingSystems,CHI ’96 (Vancouver, BC), pages50–57.ACM Press,New York, 1996.