32
temporalcontextmayhelpindeterminingtowhatextenttheyhavebeenexecutedorig-
nored.
Theutilityofcross-versioncomparisonsalsoimpliesaneedtobeabletotemporallyfor-
wardlinks,sothatasystemcanprovidethemostaccuratepossibleviewofthepastandpre-
sentstateofalinktarget.Eventhedeletionofatargetmaybehighlysignificant,aswhena
paragraphisdeletedinsteadofbeingrewritten,orhavingitsargumentupdated.Thisisan-
otherreasonforthesystemtoprovidelocalizedinformationaboutchangesaffectingdocu-
menthistoryinsomepublishedhypertexts,inadditiontotheauthoringconsiderationsad-
vancedearlier.
Naturally,itmustalsobepossibletodeletesuchhistoricalinformationasappropriate.
Evenapublicationthatcommitted,asamatterofeditorialpolicy,toupdatingitsstoriesfor
accuracyandpreservingthefullrecordofpublication,wouldbehighlyunlikelytowantto
exposeinternaleditorialprocesses,drafts,andeditingchanges.Thiswouldbetrueevenif
retentionofthisinformationwereanimportantpartofitsinternaleditorialprocess.
Evenifeditorialhistoryiseffaced,thecombinationofversion-numberinformationand
temporalforwardingallowsforanelegantwaytoupdatelinksautomaticallywhenthedesti-
nationdocumentischanged.Furthermore,itsimultaneouslysatisfiesthelink-follower’sde-
siretoseethatthelinkeddocumenthaschanged(thusenablingacriticalappraisalofpo-
tentialchangesinthelink’smeaning),andtheirdesiretoseetheactualdestination,evenif
ithaschangedsincethelinkwasoriginallymade.Thiskindofscenarioisonereasonthat
Internetstandardsgroupsareattemptingtostandardizeeditingandversionmanagement
protocolsfortheWorldWideWeb(Goland,Whiteheadetal.1998;Slein,Vitalietal.1998).
1.10 Requirementsandgoals
Inthissection,weshallbrieflyreviewsomeofthemajorpointsoftheforegoingsurvey
ofsystemsandideas,andextractthegeneralprinciplesthathaveguidedtheevolutionof
33
thePalimpsestmodelandsystem.Theserequirementsforcollaborativeeditinginfrastruc-
ture,representthekeyissuesaddressedinthisdissertation.Theoverridinggoalhasbeento
eliminateconstraintsonparallelandsimultaneousworkimposedinordertoensuresystem
anddataconsistency.Avarietyofstudiesofwritingandcollaborationhaveshownthene-
cessityofcollaborationtoolstosupportauthorsinwhattheywanttodo,ratherthantoex-
plicitlycontroltheprocessthattheyfollow(BeckandBellotti1993).Awidevarietyofdif-
ferentstrategiesforwritinghavebeenobservedinbehavioralstudies,(GalegherandKraut
1990;Beck1993;Sharples1993),suggestingthatthebestwritingsupporttoolsaregeneric
onesthatcansupportauthors’chosenstrategiesforwritingbutthatdonotcompelparticu-
larstrategies.Experiencewithstructuredmethodsofcreatinghypertext(Collier1987;Conk-
linandBegeman1987;Conklin1988;Marshall,Halaszetal.1991),alsosuggeststhatitis
counterproductivetoputapriorirestrictionsondataandprocessstructure(Marshalland
Shipman1993;Marshall,Shipmanetal.1994).
Thus,consideringonlytheissuesofsynchronizationandconcurrencycontrolalreadydis-
cussedinthischapter,thefollowingareallfeaturesofanidealframework:
1. Itshouldsupportflexibletransitionsfromfully-synchronoustofully-asynchronous
work.Authorsmayneedtomakechangesasynchronouslybecausetheymaydiscover
theneedtomakeachangeatanytime.Conversely,theymaydecidetocollaborate
interactivelyatanytime,iftheyarebothonlineandhavesomethingtodiscuss.To
theextentthattheunderlyingtechnologyfailstoaccommodatesuchflexibility,
authorswillbepreventedfromtakingadvantageof(orevenfindingoutabout)such
opportunities.
2. Itshouldsupportbothsynchronousandasynchronousediting.Someapproachesto
collaborationinfrastructurearestronglyorientedtoeithersynchronousorasynchro-
nouscollaboration.Sincebothsynchronousandasynchronousworkareusuallyin-
Previous Page Next Page