2
ports.Finally,Chapter8brieflysummarizestheresultsoftheearlierchaptersandevaluates
PalimpsestintermsofthesetofrequirementsdevelopedinChapter1.
1.2 Abasicapplicationscenario
WhilePalimpsesthasusesforcollaborativeapplicationsotherthanjusttextediting,its
originandspecialfeaturesreflectthespecializedcollaborationsupportneedsofdocument
editors.Thisisinharmonywithahistoricalbiastowardstexteditinginothercollaboration
research,aswellasapersonalconvictionthatthewrittenwordhasaprivilegedplaceasthe
primarymediumforthecommunicationofideas.Thestudyofcollaborativewritingand
documentpreparationisoneoftheoldestsub-areasofcollaborationresearch,witharich
literaturecoveringawiderangeofindividualandcollaborativewritingstrategies,user-
interfaces,socialissues(Sharples1993;Sharples,Goodletetal.1993),andsystemdesigns.
Abasicscenarioforthiskindofworkisthefollowing:severalauthorsareworkingonthe
same(hypertextorstructured)document.Theauthorsworkonseparatecomputers,which
areintermittentlyincontact.Anyauthormaywanttochangeanyaspectofthedocument,
withoutbeingforcedtowaitforinter-computercommunicationtobeestablished,whilestill
allowingtheresultingchangestobeintegratedandharmonizedatalaterpoint.Givencur-
rentadvancesinstructureddocumentrepresentation,multimediaandhypertext,such
documentsmaycontainintricatedata(sound,structuredtext,graphics,video)aswellas
complexcross-referencestructures.
Thisscenariohighlightssomeofthetechnicalneedsandrealitiesofcollaborativecom-
puting.Becausethestructureofacollaborationisultimatelycontrolledbytheauthorand
notbyasystem,thevarietyofsystemenvironmentsandcommunicationmediatobesup-
portedislarge.Infrastructureswhichmayneedtosupportcollaborativeeditingmayinclude:
FastLANsconnectinglocalworkstations
Medium-speedWANconnections(e.g.directinternetaccess)
3
Intermittentconnections(modems,E-mail,“sneakernet”)
Afastlocalnetworkisidealforcooperativeediting,andallowstheuseoftraditional
synchronizationtechniqueswithoutcreatingbottlenecks.Thiswillremainpossiblegivenany
continuouslyavailablecommunicationsaccess(thoughevenWANconnectionsarefarfrom
perfectinthisrespect).Whencommunicationsareintermittent,suchstrategiesareinsuffi-
cient;inconsistenciesinthecollaborators’viewsoftheshareddataareboundtooccur(Greif
andSarin1987).
Agreatnumberofhumanfactorsarecriticalineffectivelysupportingthewritingtask.A
completelist,evenoftheknownfactorswouldbehardtogivehere,butthefollowinglistof
generalgoalshasguidedalltheworkinthisdissertation:
Itsmodelshouldbenaturaltousers,atleasttotheextentthatitisexposedtothem
throughtheinterface.Thisiseasiestiftheunderlyingdatamodelisnaturalinitself.
Thebasicobjectsinasystemshouldcorrespondtoentitiesthatareintheuser’sontol-
ogy.
Itmustallowmeaningfulinformationfilteringandselectionbyeachcollaborator.Col-
laborationintroducesawholenewsetofproblems,andusersmusthaveeasy,natural
accesstotheinformationtheyneedtosolvethoseproblems.
Itshouldassistuserstoevolvetheirownworkprocess,ratherthanenforcingoneon
them.Whilespecializedsituationsmayrequirevarioussortsofpolicyenforcement,such
policiesareapplication,organization,andsometimespersonneldependent.Generictool-
kits,especially,shouldstriveforgeneralityandpolicyneutrality,providingdefaultpoli-
ciesonlyforthoseapplicationsthatdonotneedtheirown.
Palimpsestismyattempttobalancehumanrequirementsagainsttechnicalones,while
makingminimalcommitmentstounderlyingnetworktechnologyandapplication-specific
policydecisions.
Previous Page Next Page