JOURNALOFECONOMICBEHAVIORREVISED23JUNE2000ACCEPTED15SEPTEMBER2000AVAILABLEONLINE28MARCH2002ABSTRACTTHISPAPERISANATTEMPTTORAIDBOTHTHELITERATUREONMODULARDESIGNANDTHELITERATUREONPROPERTYRIGHTSTOCREATETHEOUTLINESOFAMODULARITYTHEORYOFTHEFIRMSUCHATHEORYWILLLOOKATFIRMS,ANDOTHERORGANIZATIONS,INTERMSOFTHEPARTITIONINGOFRIGHTSUNDERSTOODASPROTECTEDSPHERESOFAUTHORITYAMONGCOOPERATINGPARTIESANDITWILLASSERTTHATORGANIZATIONSREFLECTNONMODULARSTRUCTURES,THATIS,STRUCTURESINWHICHDECISIONRIGHTS,RIGHTSOFALIENATION,ANDRESIDUALCLAIMSTOINCOMEDONOTALLRESIDEINTHESAMEHANDSAUTHORKEYWORDSMODULARITYPROPERTYRIGHTSTRANSACTIONCOSTSCAPABILITIESSTANDARDSJELCLASSIFICATIONCODESD23L22ARTICLEOUTLINE1INTRODUCTION2MODULARITYANDCOMPLEXITY3DESIGNPROCESSES4ENCAPSULATIONBOUNDARIES5MODULARITY,PROPERTYRIGHTS,ANDORGANIZATION6MODULARITY,ORGANIZATION,ANDTECHNOLOGY7AMODULARITYTHEORYOFTHEFIRMREFERENCES1INTRODUCTIONMODULARITYISAVERYGENERALSETOFPRINCIPLESFORMANAGINGCOMPLEXITYBYBREAKINGUPACOMPLEXSYSTEMINTODISCRETEPIECESWHICHCANTHENCOMMUNICATEWITHONEANOTHERONLYTHROUGHSTANDARDIZEDINTERFACESWITHINASTANDARDIZEDARCHITECTUREONECANELIMINATEWHATWOULDOTHERWISEBEANUNMANAGEABLESPAGHETTITANGLEOFSYSTEMICINTERCONNECTIONSSUCHIDEASARENOTNEWINTHELITERATUREOFTECHNOLOGICALDESIGNSIMONANDALEXANDER,EVENIF,ASSOMECLAIMBALDWINANDCLARK,1997,MODULARITYISBECOMINGMOREIMPORTANTTODAYBECAUSEOFTHEINCREASEDCOMPLEXITYOFMODERNTECHNOLOGYWHATISNEWISTHEAPPLICATIONOFTHEIDEAOFMODULARITYNOTONLYTOTECHNOLOGICALDESIGNBUTALSOTOORGANIZATIONALDESIGNSANCHEZANDMAHONEY1996GOSOFARASTOASSERTTHATMODULARITYINTHEDESIGNOFPRODUCTSLEADSTOORATLEASTOUGHTTOLEADTOMODULARITYINTHEDESIGNOFTHEORGANIZATIONSTHATPRODUCESUCHPRODUCTSFROMANOTHERANGLE,HOWEVER,THEPRINCIPLESOFMODULARITYHAVEANEVENLONGERPEDIGREEINTHESOCIALSCIENCESWECANTHINKOFADAMSMITHS“OBVIOUSANDSIMPLESYSTEMOFNATURALLIBERTY“SMITH1976ASAMONGTHEEARLIESTPROPOSALSFORHOWACOMPLEXMODERNSOCIETYMIGHTBEMADEMOREPRODUCTIVETHROUGHAMODULARDESIGNOFSOCIALANDECONOMICINSTITUTIONSINSEPARATINGMINEFROMTHINE,RIGHTSOFPRIVATEPROPERTYMODULARIZESOCIALINTERACTION,WHICHISTHENMEDIATEDTHROUGHTHEINTERFACEOFVOLUNTARYEXCHANGE,ALLUNDERTHEGOVERNANCEOFTHESYSTEMSARCHITECTUREOFCOMMONLAWIWILLTRYTOSUGGESTTHAT,DESPITEITSHEAVYEMPHASISONINCENTIVEISSUES,THEECONOMICSOFPROPERTYRIGHTSTHATEMERGEDFROMTHEWORKOFCOASEANDCOASEINBOTHITSORIGINALFORMFURUBOTNANDPEJOVICH,1972ANDITSMOREFORMALRECENTINCARNATIONHART,1989PROVIDESAGOODDEALOFMATERIALUSEFULFORUNDERSTANDINGTHENATUREANDROLEOFMODULARITYINSOCIALINSTITUTIONSAVENNDIAGRAMWOULDREVEALONLYTHETHINNESTOFOVERLAPSBETWEENTHEMANAGEMENTLITERATUREONMODULARITYANDTHEECONOMICSOFPROPERTYRIGHTSTHEMANAGEMENTLITERATUREASKSHOWANDWHYFIRMSSHOULDDESIGNMODULARPRODUCTSANDHOWTHEYSHOULDORGANIZETHEMSELVESINTERNALLYTOCOPEWITHTHOSEMODULARPRODUCTSTHELITERATUREONWHATISIMPLICITLYMODULARITYINSOCIALINSTITUTIONSASKSHOWTHEINTERACTIONSAMONGORGANIZATIONALUNITSSHOULDBESTRUCTUREDTOINCREASEWEALTHANDPROMOTEECONOMICGROWTHTHEOVERLAPPINGQUESTION,OFCOURSE,ISCOASESQUESTIONWHATDETERMINESTHEBOUNDARIESOFORGANIZATIONSWHYARESOMEMODULARSOCIALUNITSGOVERNEDBYTHEARCHITECTUREOFTHEORGANIZATIONANDSOMEGOVERNEDBYTHELARGERARCHITECTUREOFTHEMARKETTHISPAPERISANATTEMPTTORAIDBOTHTHELITERATUREONMODULARDESIGNANDTHELITERATUREONPROPERTYRIGHTSTOCREATETHEOUTLINESOFAMODULARITYTHEORYOFTHEFIRMTHISTHEORYWILLLOOKATFIRMS,ANDOTHERORGANIZATIONS,INTERMSOFTHEPARTITIONINGOFRIGHTSUNDERSTOODASPROTECTEDSPHERESOFAUTHORITYAMONGCOOPERATINGPARTIESANDITWILLASSERTTHATORGANIZATIONSREFLECTNONMODULARSTRUCTURES,THATIS,STRUCTURESINWHICHDECISIONRIGHTS,RIGHTSOFALIENATION,ANDRESIDUALCLAIMSTOINCOMEDONOTALLRESIDEINTHESAMEHANDSTHEREASONSBEHINDALTERNATIVEPARTITIONSARENUMEROUSANDCOMPLEX,HOWEVER,CALLINGFORSUBTLETYINTHEAPPLICATIONOFTHEIDEAOFMODULARITYTOTHEFIRM2MODULARITYANDCOMPLEXITYTHEWORLDISFULLOFCOMPLEXSYSTEMSNATUREPROVIDESANABUNDANCEOFCOMPLEXORGANISMSANDECOSYSTEMS,ANDHUMANSHAVECONSTRUCTEDCOMPLEXMECHANICAL,INTELLECTUAL,ORGANIZATIONAL,ANDSOCIALSYSTEMSBUTWHATEXACTLYDOESITMEANFORASYSTEMTOBECOMPLEXFORHERBERTSIMON,ACOMPLEXSYSTEMIS“ONEMADEUPOFALARGENUMBEROFPARTSTHATINTERACTINANONSIMPLEWAYINSUCHSYSTEMS,THEWHOLEISMORETHANTHESUMOFTHEPARTS,ATLEASTINTHEIMPORTANTPRAGMATICSENSETHAT,GIVENTHEPROPERTIESOFTHEPARTSANDTHELAWSOFTHEIRINTERACTION,ITISNOTATRIVIALMATTERTOINFERTHEPROPERTIESOFTHEWHOLE“1SIMON,1962,P195COMPLEXITYISTHUSAMATTERBOTHOFTHESHEERNUMBEROFDISTINCTPARTSTHESYSTEMCOMPRISESANDOFTHENATUREOFTHEINTERCONNECTIONSORINTERDEPENDENCIESAMONGTHOSEPARTSONEWAYTOMANAGECOMPLEXITY,THEN,ISTOREDUCETHENUMBEROFDISTINCTELEMENTSINTHESYSTEMBYGROUPINGELEMENTSINTOBYHIDINGELEMENTSWITHINASMALLERNUMBEROFSUBSYSTEMSSIMONARGUESFORTHECRITERIONOFDECOMPOSABILITYINMODULARDESIGN,WHICHHEOFFERSBOTHASAPRESCRIPTIONFORHUMANDESIGNERSANDASADESCRIPTIONOFTHESYSTEMSWEFINDREADYMADEINNATURETOMAKETHELATTERPOINTHEOFFERSTHEPARABLEOFTHEWATCHMAKERSTEMPUSANDHORABOTHMAKECOMPLICATEDWATCHSYSTEMSFROMMYRIADPARTS,ANDBOTHAREINTERRUPTEDFREQUENTLYINTHEIRWORKTEMPUSDOESNOTDESIGNHISWATCHESASDECOMPOSABLESYSTEMS,SOEVERYTIMEHEISINTERRUPTEDANDFORCEDTOSETASIDEHISWORK,THEENTIREUNFINISHEDASSEMBLYFALLSTOPIECESBYCONTRAST,HORAFIRSTBUILDSSTABLESUBASSEMBLIESTHATHECANTHENPUTTOGETHERINHIERARCHICFASHIONINTOLARGERSTABLESUBASSEMBLIESTHUS,WHENHORAISINTERRUPTED,ONLYTHELASTUNFINISHEDSUBASSEMBLYFALLSAPART,PRESERVINGMOSTOFHISEARLIERWORKINANEVOLUTIONARYSELECTIONENVIRONMENT,SUCHSTABILITYWOULDBEREWARDEDWITHSURVIVALSIMON,1962,PP200205LOASBY,1976INTHEEND,HOWEVER,WHATMAKESTEMPUSSUNFINISHEDWATCHESSOUNSTABLEISNOTTHESHEERNUMBEROFDISTINCTPARTSINVOLVEDRATHER,ITISTHEINTERDEPENDENCYAMONGTHEPARTSINHISDESIGNTHATCAUSETHEWATCHESTOFALLAPARTINANONDECOMPOSABLESYSTEM,THESUCCESSFULOPERATIONOFANYGIVENPARTISLIKELYTODEPENDONTHECHARACTERISTICSOFMANYOTHERPARTSTHROUGHOUTTHESYSTEMSOWHENSUCHASYSTEMISMISSINGPARTSBECAUSEITISNOTFINISHED,FOREXAMPLE,ORBECAUSESOMEOFTHEPARTSAREDAMAGED,THEWHOLECEASESTOFUNCTIONANDTHESYSTEMBECOMESEVOLUTIONARYSHARKBAITINADECOMPOSABLESYSTEM,BYCONTRAST,THEPROPERWORKINGOFAGIVENPARTWILLDEPENDWITHHIGHPROBABILITYONTHECHARACTERISTICSOFOTHERPARTSWITHINITSSUBASSEMBLYBUTWILLDEPENDWITHRELATIVELYLOWERPROBABILITYONTHECHARACTERISTICSOFPARTSOUTSIDEOFTHATSUBASSEMBLY2ASARESULT,ADECOMPOSABLESYSTEMMAYBEABLETOLIMPALONGEVENIFSOMESUBSYSTEMSAREDAMAGEDORINCOMPLETEINORGANIZATIONALANDSOCIALSYSTEMSANDPERHAPSEVENINMECHANICALONESASWELLITISPOSSIBLETOTHINKOFINTERDEPENDENCYANDINTERACTIONAMONGTHEPARTSASAMATTEROFINFORMATIONTRANSMISSIONORCOMMUNICATIONCONSIDER,WITHERICVONHIPPEL1990,THEPROBLEMOFORGANIZINGPRODUCTINNOVATIONHERE,THEISSUEISHOWTODECOMPOSETHEORGANIZATIONOFARESEARCHANDDEVELOPMENTPROJECTBYPARTITIONINGTASKSAMONGDEVELOPMENTTEAMSASVONHIPPELPOINTSOUT,INORDERTOSOLVETHISDECOMPOSITIONPROBLEM,ONEHASTOFOCUSONTHEINTERDEPENDENCIESAMONGTHEVARIOUSTASKSTHEPROJECTCOMPRISES3IFTHEPROJECTISORGANIZEDINANONDECOMPOSABLEWAY,THENINTERDEPENDENCYWILLBEHIGH,MEANINGTHATEACHDEVELOPMENTTEAMWILLNEEDCONSTANTLYTORECEIVEANDUSEINFORMATIONABOUTWHATALLTHEOTHERDEVELOPMENTTEAMSAREDOINGFOREXAMPLE,THEDEVELOPMENTOFTHEOS/360OPERATINGSYSTEMFORTHEORIGINALIBM360LINEOFCOMPUTERSWASEVIDENTLYORGANIZEDINARELATIVELYNONDECOMPOSABLEWAYTHEMANAGEROFTHEPROJECT,FREDERICKBROOKS,INSISTEDONACONSCIOUSATTENTIONTOINTERDEPENDENCIESANDAHIGHLEVELOFCOMMUNICATIONAMONGALLPARTICIPANTSTHISINCLUDEDTHECREATIONANDMAINTENANCEOFAFORMALPROJECTWORKBOOKTHATDOCUMENTEDEVERYASPECTOFTHESYSTEMSOTHAT,INPRINCIPLEATLEAST,EVERYWORKERCOULDDETERMINEHOWCHANGESELSEWHEREWOULDAFFECTHISORHERPARTOFTHEPROJECTBROOKSDECIDED“THATEACHPROGRAMMERSHOULDSEEALLTHEMATERIAL,IESHOULDHAVEACOPYOFTHEWORKBOOKINHISOWNOFFICE“BROOKS,1975,P76BUTWITHIN6MONTHS,THEREWASONESMALLPROBLEM“THEWORKBOOKWASABOUT5FTTHICKIFWEHADSTACKEDUPTHE100COPIESSERVINGPROGRAMMERSINOUROFFICESINMANHATTANSTIMELIFEBUILDING,THEYWOULDHAVETOWEREDABOVETHEBUILDINGITSELFFURTHERMORE,THEDAILYCHANGEDISTRIBUTIONSAVERAGED2IN,SOME150PAGESTOBEINTERFILEDINTHEWHOLEMAINTENANCEOFTHEWORKBOOKBEGANTOTAKEASIGNIFICANTTIMEFROMEACHWORKDAY“BROOKS,1975,P77THETEAMSOONSWITCHEDTOMICROFICHEAND,CLEARLY,WITHMODERNTECHNOLOGY,THEWORKBOOKCOULDRESIDEONLINEANDBEUPDATEDRAPIDLYBUTTHEPOINTREMAINSTHATANONDECOMPOSABLESYSTEMINCURSHIGHCOMMUNICATIONSCOSTINDEED,ITISANINSIGHTFORWHICHBROOKSISWELLKNOWNTHAT,INTHEDESIGNOFCOMPLEXSYSTEMS,THECOSTSOFCOMMUNICATIONAMONGWORKERSWILLEVENTUALLYOUTWEIGHTHEBENEFITSOFTHEDIVISIONOFLABORASMOREANDMOREWORKERSAREADDEDTOAPROJECTBROOKS,1975,PP1819ATONEPOINT,BROOKSBRIEFLYCONSIDERSA“RADICAL“ALTERNATIVEPROPOSEDBYDLPARNAS,WHOSE“THESISISTHATTHEPROGRAMMERISMOSTEFFECTIVEIFSHIELDEDFROM,RATHERTHANEXPOSEDTOTHEDETAILSOFCONSTRUCTIONOFSYSTEMPARTSOTHERTHANHISOWN“BROOKS,1975,P78THISRADICALALTERNATIVEISINFACTTHESTRATEGYOFSEEKINGDECOMPOSABILITYINTHEDESIGNOFTHEDEVELOPMENTPROJECTANDOFTHEUNDERLYINGSOFTWAREPARNAS1972ISTHEINVENTOROFTHENOTIONOFINFORMATIONHIDING,AKEYCONCEPTINTHEMODERNOBJECTORIENTEDAPPROACHTOCOMPUTERPROGRAMMINGPROGRAMMERSHADLONGUNDERSTOODTHEIMPORTANCEOFMODULARITY,THATIS,OFBREAKINGPROGRAMSINTOMANAGEABLEPIECESBUTAMODULARSYSTEMISNOTAUTOMATICALLYADECOMPOSABLEONE,SINCEONECANBREAKTHESYSTEMSINTOMODULESWHOSEINTERNALWORKINGSREMAINHIGHLYINTERDEPENDENTWITHTHEINTERNALWORKINGSOFOTHERMODULESPARNASARGUEDTHAT,ESPECIALLYINLARGEPROJECTS,PROGRAMMERSSHOULDABANDONMODULARIZATIONBASEDONSIMPLEFLOWCHARTSANDPAYATTENTIONINSTEADTOMINIMIZINGINTERDEPENDENCIESIFKNOWLEDGEISHIDDENORENCAPSULATEDWITHINAMODULE,THATKNOWLEDGECANNOTAFFECT,ANDTHEREFORE,NEEDNOTMUSTNOTBECOMMUNICATEDTOOTHERPARTSOFASYSTEMUNDERTHISSCHEME,EVERYMODULE“ISCHARACTERIZEDBYITSKNOWLEDGEOFADESIGNDECISIONWHICHITHIDESFROMALLOTHERSITSINTERFACEORDEFINITIONWASCHOSENTOREVEALASLITTLEASPOSSIBLEABOUTITSINNERWORKINGS“PARNAS,1972,P1056RECENTLY,BALDWINANDCLARK1997,P86HAVEDRAWNONSIMILARIDEASFROMCOMPUTERSCIENCETOFORMULATESOMEGENERALPRINCIPLESOFMODULARSYSTEMSDESIGNTHEDECOMPOSITIONOFASYSTEMINTOMODULES,THEYARGUE,SHOULDINVOLVETHEPARTITIONINGOFINFORMATIONINTOVISIBLEDESIGNRULESANDHIDDENDESIGNPARAMETERS4THEVISIBLEDESIGNRULESORVISIBLEINFORMATIONCONSISTOFTHREEPARTSANARCHITECTURESPECIFIESWHATMODULESWILLBEPARTOFTHESYSTEMANDWHATTHEIRFUNCTIONSWILLBEINTERFACESDESCRIBEINDETAILHOWTHEMODULESWILLINTERACT,INCLUDINGHOWTHEYFITTOGETHERANDCOMMUNICATEANDSTANDARDSTESTAMODULESCONFORMITYTODESIGNRULESANDMEASURETHEMODULESPERFORMANCERELATIVETOOTHERMODULESTHESEVISIBLEPIECESOFINFORMATIONNEEDTOBEWIDELYSHAREDANDCOMMUNICATEDBYCONTRAST,THEHIDDENDESIGNPARAMETERSAREENCAPSULATEDWITHINTHEMODULES,ANDTHEYNEEDNOTINDEED,SHOULDNOTBECOMMUNICATEDBEYONDTHEBOUNDARIESOFTHEMODULEASBALDWINANDCLARKPOINTOUT,THELITERATUREONMODULARSYSTEMSTENDSTOCOLLAPSETHESETHREEKINDSOFINFORMATIONTOGETHER,CALLINGTHEMALLEITHER“THEARCHITECTURE,“THEINTERFACES,“OR“THESTANDARDS“INORDERTOHAVEASINGLETERMFORALLTHREE,IWILLREFERTOASETOFARCHITECTURE,INTERFACES,ANDSTANDARDSASAMODULARIZATION3DESIGNPROCESSESCLEARLY,MODULARITYISADESIGNSTRUCTUREWITHAGREATMANYADVANTAGESBYREDUCINGTHEDEGREEOFINTERDEPENDENCYAMONG,ANDTHUSTHECOSTSOFCOMMUNICATINGACROSS,THEPARTSOFASYSTEM,ITGIVESFULLREINTOTHEMANYBENEFITSOFTHEDIVISIONOFLABORAMONGCOMPONENTSASUSUAL,HOWEVER,THEREISNOFREELUNCHITTURNSOUTTHATMODULARSYSTEMSAREMUCHMOREDIFFICULTTODESIGNTHANCOMPARABLEINTERCONNECTEDSYSTEMSTHEDESIGNERSOFMODULARSYSTEMSMUSTKNOWAGREATDEALABOUTTHEINNERWORKINGSOFTHEOVERALLPRODUCTORPROCESSINORDERTODEVELOPTHEVISIBLEDESIGNRULESNECESSARYTOMAKETHEMODULESFUNCTIONASAWHOLETHEYHAVETOSPECIFYTHOSERULESINADVANCEANDWHILEDESIGNSATTHEMODULARLEVELAREPROCEEDINGINDEPENDENTLY,ITMAYSEEMTHATALLISGOINGWELLPROBLEMSWITHINCOMPLETEORIMPERFECTMODULARIZATIONTENDTOAPPEARONLYWHENTHEMODULESCOMETOGETHERANDWORKPOORLYASANINTEGRATEDWHOLEBALDWINANDCLARK,1997,P86AWELLDECOMPOSEDMODULARSYSTEMMUSTPAYAKINDOFFIXEDCOSTTHATANINTERTWINEDSYSTEMNEEDNOTPAY,ANDTHISISTRUEWHETHERTHESYSTEMISBROUGHTTOUSBYTHEEVOLUTIONARYPROCESSORBYADESIGNTEAMATSONYUNDERSOMECIRCUMSTANCES,THEBENEFITSOFMODULARIZATIONMAYNOTBEWORTHTHECOSTFOREXAMPLE,ASYSTEMWHOSEENVIRONMENTNEVERCHANGESMAYNOTHAVETOWORRYMUCHABOUTMODULARIZATIONTEMPUSWILLDOASWELLASHORAIFNEITHERISEVERINTERRUPTEDSYSTEMSTHATDEVELOPSLOWLYINASLOWLYCHANGINGENVIRONMENTMAYNOTACQUIRE,ORREQUIRE,MUCHMODULARITYASWESAW,ANONDECOMPOSABLECOMPLEXSYSTEMTHATISCONSCIOUSLYDESIGNEDASBYADEVELOPMENTTEAMREQUIRESAGREATDEALOFCOMMUNICATIONAMONGTHEDESIGNERSINANOTHERSENSE,THESYSTEMITSELFALSODISPLAYSAGREATDEALOF“COMMUNICATION“AMONGTHEPARTS,INTHATTHEFUNCTIONINGOFONEPARTISHIGHLYDEPENDENTONTHEFUNCTIONINGOFPARTSELSEWHEREINTHESYSTEMBUTWHENASYSTEMISNOTCONSCIOUSLYDESIGNEDANDINSTEADDEVELOPSSLOWLYOVERTIME,LEARNINGCANTAKEPLACEINAWAYTHATINVOLVESFARLESSEXPLICITTRANSMISSIONOFINFORMATIONASALEXANDERSUGGESTSINTHECONTEXTOFARCHITECTUREANDURBANDESIGN,MANYOFTHEMOSTATTRACTIVEANDDURABLESYSTEMSARETHOSEOFTENTRADITIONALONESTHATDEVELOPTHROUGHAN“UNSELFCONSCIOUS“PROCESSWHETHERTHEOBJECTISABUILDINGORACOMPUTEROPERATINGSYSTEM,ADESIGNERMUSTSETFORTHDESIGNRULESEXPLICITLYANDCOMMUNICATETHEMCLEARLYINTHEUNSELFCONSCIOUSPROCESS,BYCONTRAST,“THERULESARENOTMADEEXPLICIT,BUTARE,ASITWEREREVEALEDTHROUGHTHECORRECTIONOFMISTAKES“ALEXANDER,1964,P35INCONSISTENCIESANDINTERDEPENDENCIESAREREVEALEDNOTBYEXANTECOMMUNICATIONBUTBYTRIALANDERROR5ALEXANDERSEEMSTOTHINKTHATTRADITIONALDESIGNSDEVELOPEDBYTHEUNSELFCONSCIOUSPROCESSWILLALSOTENDTOBEMODULARTHISCERTAINLYCANBESOTHEMODULARDESIGNOFTHEEARLIESTPCS,WHICHINFLUENCEDTHEDESIGNOFTHEIBMPC,WASNOTTHERESULTOFANYCONSCIOUSDESIGNLANGLOIS,1992ABUTITISBYNOMEANSOBVIOUSTHATUNSELFCONSCIOUSDESIGNMUSTALWAYS,OREVENUSUALLY,RESULTINMODULARITY6UNDERSTABLECONDITIONS,INDEED,ANONMODULARSYSTEMMAYPRESENTCERTAINADVANTAGESTOANUNSELFCONSCIOUSPROCESSOFTRIALANDERRORASBROOKSPOINTSOUTINCRITICIZINGTHEPARNASIANAPPROACHTOSOFTWARE,“AGOODINFORMATIONSYSTEMBOTHEXPOSESINTERFACEERRORSANDSTIMULATESTHEIRCORRECTION“BROOKS,1975,P78TOTHEEXTENTTHATANONDECOMPOSABLESYSTEMTENDSTOREVEALERRORSMOREQUICKLYANDMOREVISIBLY,SUCHASYSTEMMAYSTIMULATELEARNINGBYDOINGINAWAYTHATMODULARSYSTEMSDONOTTHISISINDEEDTHEBENEFITOFNONDECOMPOSABILITYTHATJAPANESEMANUFACTURINGSYSTEMSINCLUDINGJUSTINTIMEINVENTORYSYSTEMSARESAIDTOTAKEADVANTAGEOFITSINCETHEFAILUREOFANYONEPARTCANCAUSETOTALSYSTEMBREAKDOWN,NONDECOMPOSABILITYRAISESTHECOSTOFMISSINGORPOORLYFUNCTIONINGPARTS,WHICHINTURNRAISESTHEINCENTIVETOMAKESURETHATEACHPARTISOFHIGHQUALITYNONDECOMPOSABILITYALSOHIGHLIGHTSBOTTLENECKSANDINCONSISTENCIESONEWOULDEXPECT,HOWEVER,THATTHEBENEFITSOFNONDECOMPOSABILITYWOULDDOMINATETHOSEOFMODULARITYONLYFORSOMEKINDSOFRELATIVELYSTABLESYSTEMSWHEREFREQUENTCHANGEISNOTIMPORTANTALTHOUGHITDIDEVENTUALLYHAVESOMEMODULARASPECTS,HENRYFORDSPRODUCTIONSYSTEMFORTHEMODELTWASONTHEWHOLEAFINELYTUNEDANDNONDECOMPOSABLESYSTEMFORCRANKINGOUTLARGEVOLUMESOFANUNDIFFERENTIATEDPRODUCTINTHE1920S,THISSYSTEMCOULDNOTADAPTWELLTOTHECHANGEDMARKETCONDITIONSANDTHEMOREMODULARSTRATEGIESOFCOMPETITORS,NOTABLYGENERALMOTORSLANGLOISANDROBERTSON,1989HISTORYOFFERSMANYOTHEREXAMPLESOFNONDECOMPOSABLESYSTEMSTHATDEVELOPEDBYTRIALANDERRORINSTABLEENVIRONMENTS4ENCAPSULATIONBOUNDARIESINAWORLDOFCHANGE,MODULARITYISGENERALLYWORTHTHECOSTSTHEREALISSUEISNORMALLYNOTWHETHERTOBEMODULARBUTHOWTOBEMODULARWHICHMODULARIZATION,WHICHSTRUCTUREOFENCAPSULATIONBOUNDARIES,WILLYIELDTHEBESTSYSTEMDECOMPOSITIONTHEGOALISCLEARLYTOFINDTHEMODULARIZATIONTHATMINIMIZESINTERDEPENDENCIESANDMOSTCLEANLYDECOMPOSESTHESYSTEMBUTHOWTODOTHISHOWDOWEFINDTHE“NATURAL“ENCAPSULATIONBOUNDARIESWEWOULDTHINKITODDINDEEDTOASSIGNTWOINTERIORDESIGNERSEACHHALFOFAROOMVONHIPPEL,1990,P410ITMAKESAGOODDEALMORESENSEEITHERTOGIVEEACHDESIGNERAWHOLEROOMORTOGIVEUPENCAPSULATIONENTIRELYANDLETTHETWODESIGNERSCOMMUNICATEEXTENSIVELYTHEPROBLEMOFDEFININGBOUNDARIESOFENCAPSULATIONBECOMESEVENMORECHALLENGINGINADYNAMICSETTINGFOREXAMPLE,THETASKSINANINNOVATIVEDEVELOPMENTPROJECTCANNOTBEPARTITIONEDINADVANCE,SINCEKNOWLEDGEISCONTINUALLYCHANGING7INSUCHACASE,THEMODULARIZATIONOFTHESYSTEMTHEDEVELOPMENTPROJECTHASTOCHANGECONTINUALLYMOREOVER,THEMODULARIZATIONATANYPOINTHASTOTAKEINTOACCOUNTTHEINEVITABILITYOFREMODULARIZATIONASLEARNINGTAKESPLACEINROUTINEPROJECTSLIKESETTINGDESIGNERSTOWORKONAROOM,ITMAYBEPOSSIBLETOPREDICTWHICHTASKSWILLBECOMETHESOURCESOFNEWINFORMATIONBUTINGENUINELYNOVELPROJECTS,PREDICTIONBECOMESALLBUTIMPOSSIBLE8INDEED,UNPREDICTABLENOVELTYMAYMAKEANYHARDENCAPSULATIONUNDESIRABLE,CALLINGINSTEADFOR“LOOSELYCOUPLED“DEVELOPMENTTEAMSOFTHESORTEVIDENTLYFAVOREDBYJAPANESECOMPANIES9IMAIETAL,1985OFCOURSE,INNOVATIONNEEDNOTALWAYSREQUIREORIMPLYREMODULARIZATIONINSOMECASES,INDEED,AFIXEDANDPREDICTABLEMODULARIZATIONCANSPURINNOVATIONTHEPERSONALCOMPUTERISACASEINPOINTLANGLOISANDROBERTSON,1992,PP301302WITHTHEADVENTOFTHEIBMPCINTHEEARLY1980S,THEMICROCOMPUTERBECAMEAMODULARSYSTEMWITHARELATIVELYFIXEDORATLEASTSLOWLYCHANGINGARCHITECTURE,INTERFACES,ANDSTANDARDSDESPITEOR,RATHER,BECAUSEOFTHESLOWLYCHANGINGCHARACTEROFTHEPCSBASICSTRUCTURE,THEINDUSTRYEXPERIENCEDAPHENOMENALINCREASEINVALUEANDREDUCTIONINCOSTARISINGALMOSTENTIRELYFROMIMPROVEMENTSINMODULESMICROPROCESSOR,SOFTWARE,MODEMS,PERIPHERALS,ETCRATHERTHANFROMIMPROVEMENTSINTHEWAYTHEMODULESCONNECTTOGETHERINNOVATIONTHATTAKESPLACETHROUGHCHANGEINTHEMODULESWECANCALLMODULARINNOVATIONLANGLOISANDROBERTSON,1992,PP301302SANCHEZANDMAHONEY,1996,PP6869THISISINCONTRASTTOWHATHENDERSONANDCLARK1990CALLARCHITECTURALINNOVATION,INWHICHTHEPARTSREMAINTHESAMEBUTTHEARCHITECTURECONNECTINGTHEMCHANGESNOTICE,HOWEVER,THATARCHITECTURALINNOVATIONNEEDNOTALWAYSIMPLYACHANGEINTHESYSTEMSVISIBLEDESIGNRULESLEGOSANDTINKERTOYSARECLASSICMODULARSYSTEMSDESIGNEDFORARCHITECTURALINNOVATIONHERETHEARCHITECTURETHEWAYTHEUNCHANGINGPARTSARERECOMBINEDCANCHANGEWITHOUTAFUNDAMENTALCHANGEINTHEOVERALLMODULARIZATION10AND,INFACT,PERSONALCOMPUTERSALSOBENEFITFROMTHEMIXANDMATCHCAPABILITIESOFAMODULARSYSTEMTHATALLOWONETOCONFIGURETHESYSTEMTOTASTEASMUCHASTHEYDOFROMIMPROVEMENTINTHECONSTITUENTMODULESLANGLOISANDROBERTSON,1992SOMETIMES,HOWEVER,IMPROVINGTHEFUNCTIONINGOFASYSTEMCALLSFORREMODULARIZATIONRATHERTHANRECOMBINATIONSINCEAREMODULARIZATIONWILLRENDERTHEEXISTINGSTOCKOFMODULESINCOMPATIBLEORIRRELEVANT,SYSTEMICINNOVATIONCOMESATACOSTTHISINDEEDISONEOFTHECENTRALINSIGHTSOFTHEMUCHDISCUSSEDLITERATUREONPATHDEPENDENCYANDTECHNICALSTANDARDS11CANSYSTEMSGETSTUCKINANINFERIORMODULARIZATIONBECAUSEOFTHECOSTSOFREMODULARIZATIONTHEANSWERWILLNODOUBTDEPENDONTHECONTEXTARELATEDISSUEARISESINADYNAMICSETTINGASWEHAVESEEN,WELLDECOMPOSEDMODULARITYREQUIRESVISIBLEDESIGNRULESTHATAREFIXEDANDUNAMBIGUOUSIFTHEDESIGNRULESAREINFLUX,THEPOSSIBILITIESFORMODULARINNOVATIONANDMIXANDMATCHRECOMBINATIONWILLBELIMITEDONTHEOTHERHAND,FREEZINGTHEDESIGNRULESTOOEARLYMAYRESULTINANINFERIORMODULARIZATI