Electronic copy available at: http://ssrn.com/abstract=1577556 GOOGLE AND THE LIMITS OF ANTITRUST: THE CASE AGAINST THE ANTITRUST CASE AGAINST GOOGLE ...
1 downloads
25 Views
1MB Size
GOOGLE AND THE LIMITS OF ANTITRUST: THE CASE AGAINST THE ANTITRUST CASE AGAINST GOOGLE Geoffrey A. Manne, Lewis & Clark Law School Joshua D. Wright, George Mason University School of Law
Harvard Journal of Law & Public Policy, Vol. 34, No. 1, Winter 2011 George Mason University Law and Economics Research Paper Series
10-25 This paper can be downloaded without charge from the Social Science Research Network at http://ssrn.com/abstract_id=1577556
Electronic copy available at: http://ssrn.com/abstract=1577556
GOOGLEȱANDȱTHEȱLIMITSȱOFȱANTITRUST:ȱTHEȱCASEȱ AGAINSTȱTHEȱCASEȱAGAINSTȱGOOGLEȱ GEOFFREYȱA.ȱMANNE*ȱ&ȱJOSHUAȱD.ȱWRIGHT**ȱ
ABSTRACTȱ TheȱantitrustȱlandscapeȱchangedȱdramaticallyȱinȱtheȱlastȱdecȬ ade.ȱWithinȱtheȱlastȱtwoȱyearsȱalone,ȱtheȱDepartmentȱofȱJusticeȱ hasȱheldȱhearingsȱonȱtheȱappropriateȱscopeȱofȱSectionȱ2ȱofȱtheȱ ShermanȱActȱandȱhasȱissued,ȱthenȱrepudiated,ȱaȱcomprehensiveȱ Report.ȱ Duringȱ theȱ sameȱ time,ȱ theȱ Europeanȱ Commissionȱ hasȱ becomeȱ anȱ aggressiveȱ leaderȱ inȱ singleȬfirmȱ conductȱ enforceȬ mentȱ byȱ bringingȱ abuseȱ ofȱ dominanceȱ actionsȱ andȱ assessingȱ heavyȱ finesȱ againstȱ firmsȱ includingȱ Qualcomm,ȱ Intel,ȱ andȱ MiȬ crosoft.ȱInȱtheȱUnitedȱStates,ȱtwoȱofȱtheȱmostȱsignificantȱcharacȬ teristicsȱofȱtheȱnewȱantitrustȱapproachȱhaveȱbeenȱtheȱincreasedȱ focusȱonȱinnovativeȱcompaniesȱinȱhighȬtechȱindustriesȱandȱtheȱ diminishedȱconcernȱthatȱerroneousȱ antitrustȱinterventionsȱwillȱ hinderȱeconomicȱgrowth.ȱThisȱfocusȱonȱhighȬtechȱindustriesȱisȱ dangerous,ȱandȱtheȱconcernsȱregardingȱerroneousȱinterventionsȱ shouldȱnotȱ beȱdismissedȱtooȱlightly.ȱThisȱArticleȱoffersȱaȱ comȬ prehensive,ȱ cautionaryȱ taleȱ inȱ theȱ contextȱ ofȱ aȱ detailedȱ factual,ȱ legal,ȱandȱeconomicȱanalysisȱofȱtheȱnextȱMicrosoft:1ȱtheȱtheoretiȬ cal,ȱ butȱ perhapsȱ imminent,ȱ enforcementȱ againstȱ Google.ȱ Closeȱ scrutinyȱofȱtheȱcomplexȱeconomicsȱofȱGoogle’sȱdisputedȱtechnolȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ *ȱExecutiveȱ Director,ȱ Internationalȱ Centerȱ forȱ Lawȱ &ȱ Economics;ȱ Lecturerȱ inȱ Law,ȱLewisȱ&ȱClarkȱLawȱSchool.ȱ **ȱAssociateȱ Professor,ȱ Georgeȱ Masonȱ Universityȱ Schoolȱ ofȱ Lawȱ andȱ DepartȬ mentȱofȱEconomics;ȱDirectorȱofȱResearch,ȱInternationalȱCenterȱforȱLawȱ&ȱEconomȬ ics.ȱ Weȱ thankȱ Juddȱ Stoneȱ forȱ superbȱ researchȱ assistance.ȱ Weȱ alsoȱ gratefullyȱ acknowledgeȱaȱgrantȱfromȱtheȱInternationalȱCenterȱforȱLawȱ&ȱEconomics,ȱwhichȱ hasȱ previouslyȱ receivedȱ supportȱ fromȱ Google,ȱ amongȱ otherȱ companies.ȱ Theȱ analysisȱisȱourȱownȱandȱdoesȱnotȱnecessarilyȱreflectȱtheȱviewsȱofȱtheȱInternationalȱ CenterȱforȱLawȱ&ȱEconomics,ȱitsȱaffiliatedȱacademics,ȱorȱanyȱofȱitsȱsupporters.ȱAllȱ errorsȱareȱourȱown.ȱ 1.ȱUnitedȱ Statesȱ v.ȱ Microsoftȱ Corp.,ȱ 253ȱ F.3dȱ 34ȱ (D.C.ȱ Cir.ȱ 2001)ȱ (enȱ banc)ȱ (perȱ curiam).ȱ
Electronic copy available at: http://ssrn.com/abstract=1577556
2ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
ogyȱ andȱ businessȱ practicesȱ revealsȱ aȱ rangeȱ ofȱ procompetitiveȱ explanations.ȱ Economicȱ complexityȱ andȱ ambiguity,ȱ coupledȱ withȱ anȱ insufficientlyȱ deferentialȱ approachȱ toȱ innovativeȱ techȬ nologyȱandȱpricingȱpracticesȱinȱtheȱmostȱrelevantȱcaseȱlaw,ȱporȬ tendȱaȱpotentiallyȱerroneous—andȱcostly—result.ȱOurȱanalysis,ȱ byȱcontrast,ȱembracesȱtheȱcautiousȱandȱevidenceȬbasedȱapproachȱ toȱ uncertainty,ȱ complexity,ȱ andȱ dynamicȱ innovationȱ containedȱ withinȱtheȱwellȬestablishedȱerrorȬcostȱframework.ȱAsȱweȱdemȬ onstrate,ȱthoughȱthereȱisȱanȱabundanceȱofȱerrorȬcostȱconcernȱinȱ theȱSupremeȱCourtȱ precedent,ȱthereȱisȱaȱrealȱ riskȱthatȱtheȱcurȬ rent,ȱ aggressiveȱ approachȱ toȱ antitrustȱ error,ȱ coupledȱ withȱ theȱ uncertainȱeconomicsȱofȱGoogle’sȱinnovativeȱconduct,ȱwillȱyieldȱ aȱ costlyȱ intervention.ȱ Theȱ pointȱ isȱ notȱ thatȱ weȱ knowȱ thatȱ Google’sȱ conductȱ isȱ procompetitive,ȱ butȱ ratherȱ thatȱ theȱ veryȱ uncertaintyȱsurroundingȱitȱcounselsȱcaution,ȱnotȱaggression.ȱ ȱ I. INTRODUCTION ..........................................................3 II. INNOVATION,ȱERRORȱCOSTSȱANDȱTHEȱȱ LIMITSȱOFȱANTITRUST ...............................................8 III. THEȱUNCERTAINȱECONOMICSȱOFȱGOOGLEȇSȱ BUSINESSȱANDȱGOOGLE’SȱMARKET ........................19 A. SomeȱBasicsȱofȱOnlineȱSearch ........................22 B. Google’sȱMarket ..............................................24 C. TheȱImportanceȱofȱQualityȱScores.................33 D. NetworkȱEffects ...............................................36 IV. THEȱMONOPOLIZATIONȱCASEȱȱ AGAINSTȱGOOGLE ...................................................43 A. FirstȱPrinciplesȱofȱMonopolizationȱ Enforcement .....................................................44 B. MonopolyȱPower .............................................50 C. MarketȱDefinitionȱandȱȱ MonopolyȱPower .............................................50 D. TheȱQuestionȱofȱNetworkȱEffects ..................53 E. HasȱGoogleȱEngagedȱinȱExclusionaryȱ Conduct?...........................................................58 F. ExclusiveȱSyndicationȱAgreementsȱandȱ OtherȱForeclosureȬBasedȱArguments ...........59 G. SubstantialȱForeclosure...................................64 H. QualityȱScores ..................................................69 V. CONCLUSION ...........................................................74 ȱ
Electronic copy available at: http://ssrn.com/abstract=1577556
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ I.
3ȱ
INTRODUCTIONȱ
Muchȱhasȱchangedȱinȱtheȱmonopolizationȱlawȱlandscapeȱsinceȱ theȱwatershedȱMicrosoftȱdecisionȱoverȱaȱdecadeȱago.ȱInȱtheȱpastȱ twoȱyears,ȱtheȱDepartmentȱofȱJusticeȱhasȱissued,ȱandȱthenȱrepuȬ diated,ȱaȱcomprehensiveȱreportȱonȱSectionȱ2ȱofȱtheȱShermanȱAct,ȱ andȱ theȱ Europeanȱ Commissionȱ hasȱ risenȱ asȱ aȱ leaderȱ inȱ singleȱ firmȱ conductȱ enforcementȱ byȱ bringingȱ claimsȱ againstȱ firmsȱ inȬ cludingȱQualcomm,ȱIntel,ȱandȱMicrosoft.ȱMeanwhile,ȱChinaȱhasȱ passedȱitsȱownȱantitrustȱlawȱandȱhasȱbecomeȱanȱimportantȱparȬ ticipantȱ inȱ debatesȱ overȱ theȱ futureȱ ofȱ internationalȱ antitrust.ȱ Mostȱ recently,ȱ theȱ Federalȱ Tradeȱ Commissionȱ (FTC)ȱ controverȬ siallyȱinvokedȱitsȱauthorityȱunderȱSectionȱ5ȱofȱtheȱFederalȱTradeȱ CommissionȱActȱ(FTCȱAct)ȱtoȱchallengeȱIntel’sȱpricingȱpracticesȱ inȱtheȱmicroprocessorȱmarket.2ȱ Applyingȱantitrustȱlawsȱtoȱinnovativeȱcompaniesȱinȱdynamicȱ marketsȱ hasȱ alwaysȱ beenȱ aȱ perilousȱ proposition,ȱ andȱ despiteȱ significantȱadvancesȱinȱeconomicsȱandȱjurisprudence,ȱitȱremainsȱ so.ȱ Successfulȱ firmsȱ suchȱ asȱ Google,ȱ whichȱ competeȱ inȱ marketsȱ characterizedȱ byȱ innovation,ȱ rapidȱ technologicalȱ change,ȱ andȱ aȱ strongȱ relianceȱ onȱ intellectualȱ propertyȱ rights,ȱ areȱ especiallyȱ likely,ȱandȱespeciallyȱproblematic,ȱtargets.3ȱȱ Contemporaryȱ monopolizationȱ enforcementȱ inȱ theȱ USȱ isȱ foȬ cusedȱ substantiallyȱ onȱ innovativeȱ companiesȱ inȱ highȬtechȱ inȬ dustries,ȱcreatingȱsubstantialȱconcernsȱthatȱantitrustȱerrorȱinȱtheȱ formȱ ofȱ successfulȱ interventionsȱ againstȱ proȬcompetitiveȱ innoȬ vationsȱ andȱ businessȱ practicesȱ willȱ hinderȱ economicȱ growth.ȱ Givenȱtheȱfundamentalȱdifficultyȱofȱidentifyingȱtheȱcompetitiveȱ consequencesȱ ofȱ businessȱ practicesȱ generally,ȱ andȱ innovationsȱ especially,ȱconcernȱwithȱtheȱsocialȱcostsȱofȱtheseȱerrorsȱ(“errorȱ costs”)ȱhasȱbeenȱaȱmainstreamȱconsiderationȱinȱantitrustȱpolicyȱ discourseȱ forȱ theȱ lastȱ quarterȱ century.ȱ Unfortunately,ȱ currentȱ antitrustȱ enforcersȱ inȱ theȱ USȱ haveȱ minimizedȱ theseȱ errorȱ costȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 2.ȱComplaint,ȱInȱreȱIntelȱCorp.,ȱNo.ȱ9341,ȱFTCȱ(Dec.ȱ16,ȱ2009)ȱhttp://www.ftc.gov/ȱ os/adjpro/d9341/091216intelcmpt.pdf,ȱ Theȱ caseȱ wasȱ eventuallyȱ settled.ȱ ȱ Seeȱ Inȱ reȱ Intelȱ Corp.,ȱ No.ȱ 9341,ȱ FTCȱ (Oct.ȱ 29,ȱ 2010),ȱ http://www.ftc.gov/os/adjpro/ȱ d9341/101102inteldo.pdf.ȱ 3.ȱTheȱ Assistantȱ Attorneyȱ Generalȱ forȱ Antitrust,ȱ Christineȱ Varney,ȱ hasȱ sugȬ gestedȱ thatȱ Google,ȱ inȱ particular,ȱ isȱ ofȱ concernȱ toȱ theȱ government.ȱ Christineȱ Varney,ȱ U.S.ȱ Asst.ȱ Att’yȱ Genȱ forȱ Antitrust,ȱ Remarksȱ Beforeȱ theȱ Americanȱ AntiȬ trustȱ Instituteȱ (Feb.ȱ 11,ȱ 2008),ȱ availableȱ atȱ http://www.antitrustinstitute.org/files/ȱ 20080619_pv_aai061908holemanbreakout_020320091323.mp3.ȱ
4ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
concerns,ȱwithȱoneȱevenȱdeclaringȱthatȱ“thereȱisȱnoȱsuchȱthingȱasȱ aȱ falseȱ positive.”4ȱ Atȱ theȱ sameȱ time,ȱ enforcersȱ atȱ theȱ FTCȱ haveȱ broughtȱaȱcomplicatedȱandȱcontroversialȱcaseȱagainstȱIntelȱunȬ derȱSectionȱ5ȱofȱtheȱFTCȱAct,ȱpreciselyȱinȱorderȱtoȱmakeȱanȱendȬ runȱaroundȱShermanȱActȱjurisprudenceȱthatȱenshrinesȱerrorȱcostȱ concerns.5ȱ Lessȱ thanȱ aȱ yearȱ afterȱ theȱ Supremeȱ Courtȱ reinforcedȱ thatȱ errorȱ costsȱ wereȱ aȱ centralȱ componentȱ ofȱ monopolizationȱ doctrine,ȱantitrustȱenforcersȱinȱtheȱUnitedȱStatesȱhaveȱadoptedȱaȱ dramaticallyȱdifferent—andȱopposing—viewȱofȱtheȱroleȱthatȱanȬ titrustȱerrorsȱshouldȱplayȱinȱfutureȱenforcementȱdecisions.6ȱ Thingsȱ haveȱ alsoȱ changedȱ inȱ theȱ webȬbasedȱ economy.ȱ Asȱ isȱ toȱbeȱexpectedȱinȱdynamicȱmarkets,ȱitȱwouldȱhaveȱbeenȱdifficultȱ toȱ predictȱ inȱ 1998ȱ theȱ challengeȱ thatȱ Linuxȱ wouldȱ poseȱ toȱ MiȬ crosoft,ȱ theȱ growthȱ ofȱ Google,ȱ theȱ commercialȱ successȱ ofȱ theȱ iPod,ȱtheȱtransformativeȱroleȱofȱmobileȱandȱcellularȱcomputing,ȱ andȱ manyȱ otherȱ welfareȬenhancingȱ innovationsȱ overȱ theȱ lastȱ decade.ȱ Butȱ despiteȱ theseȱ apparentȱ changesȱ inȱ theȱ legalȱ andȱ economicȱ environment,ȱ theȱ antitrustȱ communityȱ findsȱ itselfȱ facingȱ theȱ sameȱ debateȱ thatȱ ragedȱ beforeȱ theȱ Microsoftȱ wars:ȱ Whatȱ isȱ theȱ appropriateȱ roleȱ ofȱ antitrust,ȱ andȱ monopolizationȱ lawȱinȱparticular,ȱinȱtheȱNewȱEconomy?ȱMuchȱhasȱbeenȱwrittenȱ onȱ thisȱ topic,ȱ withȱ virtuallyȱ everyȱ conceivableȱ policyȱ positionȱ havingȱbeenȱtakenȱinȱsomeȱformȱorȱanother.ȱSomeȱhaveȱarguedȱ thatȱ theȱ economyȱ movesȱ tooȱ fastȱ forȱ antitrustȱ remediesȱ toȱ beȱ fullyȱeffective.7ȱOthersȱhaveȱarguedȱthatȱantitrustȱrulesȱsimplyȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 4.ȱId.ȱ 5.ȱJanȱ Leibowitz,ȱ Chairman,ȱ Fed.ȱ Tradeȱ Comm’nȱ &ȱ J.ȱ Thomasȱ Rosch,ȱ Comm’r,ȱ Fed.ȱ Tradeȱ Comm’n,ȱ Statementȱ Regardingȱ Inȱ reȱ Intelȱ Corp.ȱ No.ȱ 9341ȱ (Dec.ȱ 16,ȱ 2009),ȱ availableȱ atȱ http://www.ftc.gov/os/adjpro/d9341/091216ȱ intelchairstatement.pdf.ȱ 6.ȱPac.ȱ BellȱTel.ȱCo.ȱ v.ȱLinklineȱCommc’ns,ȱInc.,ȱ 129ȱS.ȱCt.ȱ 1109,ȱ 1113–14ȱ (2009)ȱ (“Recognizingȱ aȱ priceȬsqueezeȱ claimȱ whereȱ theȱ defendant’sȱ retailȱ priceȱ remainsȱ aboveȱ costȱ wouldȱ inviteȱ theȱ preciseȱ harmȱ weȱ soughtȱ toȱ avoidȱ inȱ Brookeȱ Group:ȱ Firmsȱmightȱraiseȱtheirȱretailȱpricesȱorȱrefrainȱfromȱaggressiveȱpriceȱcompetitionȱtoȱ avoidȱpotentialȱantitrustȱliability”ȱandȱfindingȱitȱ“mostȱtroublingȱ[that]ȱfirmsȱthatȱ seekȱ toȱ avoidȱ priceȬsqueezeȱ liabilityȱ willȱ haveȱ noȱ safeȱ harborȱ forȱ theirȱ pricingȱ practices.”).ȱ 7.ȱRobertȱ W.ȱ Crandallȱ &ȱ Cliffordȱ Winston,ȱ Doesȱ Antitrustȱ Policyȱ Improveȱ ConȬ sumerȱWelfare?ȱAssessingȱtheȱEvidenceȱ17ȱJ.ȱECON.ȱ PERSP.ȱ3ȱ(2003).ȱȱInȱtheȱMicrosoftȱ context,ȱseeȱalsoȱWilliamȱH.ȱPage,ȱMandatoryȱContractingȱRemediesȱinȱtheȱAmericanȱ andȱ Europeanȱ Microsoftȱ Cases,ȱ 75ȱ ANTITRUSTȱ L.J.ȱ 787ȱ (2008);ȱ Williamȱ H.ȱ Pageȱ &ȱ SeldonȱJ.ȱChilders,ȱMeasuringȱComplianceȱwithȱCompulsoryȱLicensingȱRemediesȱinȱtheȱ AmericanȱMicrosoftȱCase,ȱ76ȱANTITRUSTȱ L.J.ȱ239ȱ(2009);ȱWilliamȱH.ȱPageȱ&ȱSeldonȱ J.ȱ Childers,ȱ Softwareȱ Developmentȱ asȱ anȱ Antitrustȱ Remedy:ȱ Lessonsȱ fromȱ theȱ EnforceȬ
ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
5ȱ
shouldȱ notȱ applyȱ whereȱ innovationȱ andȱ dynamicȱ competitionȱ areȱatȱstakeȱbecauseȱofȱtheȱpotentialȱchillingȱeffectsȱonȱinnovaȬ tion.8ȱȱStillȱothersȱhaveȱarguedȱthatȱanticompetitiveȱabusesȱareȱ evenȱ moreȱ likelyȱ toȱ stifleȱ innovationȱ andȱ harmȱ consumersȱ inȱ theȱ modernȱ economy,ȱ andȱ thusȱ antitrustȱ enforcersȱ shouldȱ beȱ especiallyȱactiveȱinȱtheseȱmarkets.9ȱ ThisȱArticleȱwillȱdiscussȱtheȱproblemsȱofȱantitrustȱenforcementȱ inȱtheȱInternetȱeconomy,ȱandȱtheȱtheoreticalȱcaseȱagainstȱtheȱantiȬ trustȱ community’sȱ contemporaryȱ bêteȱ noir,ȱ Google.ȱ Itȱ willȱ emȬ braceȱtheȱcautiousȱandȱevidenceȬbasedȱapproachȱtoȱuncertainty,ȱ complexity,ȱandȱdynamicȱinnovationȱcontainedȱwithinȱtheȱerrorȬ costȱ framework,ȱ aȱ mainstreamȱ andȱ wellȬdevelopedȱ methodȱ ofȱ evaluatingȱ legalȱ rulesȱ generally,ȱ and,ȱ inȱ thisȱ case,ȱ forȱ balancingȱ theȱfullȱsocialȱbenefitsȱandȱcostsȱofȱproposedȱantitrustȱintervenȬ tions.ȱ Thisȱ approachȱ isȱ wellȱ acceptedȱ inȱ theȱ antitrustȱ literatureȱ amongȱ lawyersȱ andȱ economists.10ȱ Butȱ manyȱ antitrustȱ enforcersȱ andȱaȱvocalȱsubsetȱofȱcommentatorsȱhaveȱshunnedȱtheȱapproach,ȱ becauseȱtheyȱviewȱtheȱerrorȬcostȱframeworkȱasȱanȱannoyingȱimȬ pedimentȱtoȱmoreȱvigorousȱenforcement.11ȱForȱexample,ȱatȱleastȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ mentȱ ofȱ theȱ Microsoftȱ Communicationsȱ Protocolȱ Licensingȱ Requirement,ȱ 14ȱ MICH.ȱ TELECOMM.ȱ&ȱTECH.ȱL.ȱREV.ȱ77ȱ(2008).ȱȱ 8.ȱDanielȱF.ȱSpulber,ȱUnlockingȱTechnology:ȱAntitrustȱandȱInnovation,ȱ4ȱJ.ȱ COMP.ȱL.ȱ &ȱECON.ȱ915ȱ(2008).ȱ 9.ȱSee,ȱe.g.,ȱCarlȱShapiro,ȱExclusivityȱInȱNetworkȱIndustries,ȱ7ȱGEO.ȱ MASONȱ L.ȱ REV.ȱ 673,ȱ674–75ȱ(1999).ȱȱ 10.ȱSee,ȱ e.g.,ȱ RICHARDȱ A.ȱ POSNER,ȱ ANTITRUSTȱ LAW,ȱ atȱ ixȱ (2ndȱ ed.ȱ 2001);ȱ C.ȱ FreȬ derickȱ Becknerȱ IIIȱ &ȱ Stevenȱ C.ȱ Salop,ȱ Decisionȱ Theoryȱ andȱ Antitrustȱ Rules,ȱ 67ȱ ANTITRUSTȱ L.J.ȱ41ȱ(1999);ȱFrankȱH.ȱEasterbrook,ȱTheȱLimitsȱofȱAntitrust,ȱ63ȱTEX.ȱ L.ȱ REV.ȱ1ȱ(1984);ȱDavidȱS.ȱEvansȱ&ȱA.ȱJorgeȱPadilla,ȱDesigningȱAntitrustȱRulesȱforȱAsȬ sessingȱ Unilateralȱ Practices:ȱ Aȱ NeoȬChicagoȱ Approach,ȱ 72ȱ U.ȱ CHI.ȱ L.ȱ REV.ȱ 73ȱ (2005);ȱ Lukeȱ Froebȱ etȱ al.,ȱ Verticalȱ antitrustȱ policyȱ asȱ aȱ problemȱ ofȱ inference,ȱ 23ȱ INT’Lȱ J.ȱ INDUS.ȱ ORG.ȱ639ȱ(2005);ȱKeithȱN.ȱHyltonȱ&ȱMichaelȱSalinger,ȱTyingȱLawȱandȱPolicy:ȱAȱDeciȬ sionȬTheoreticȱApproach,ȱ69ȱANTITRUSTȱL.J.ȱ469ȱ(2001);ȱGeoffreyȱA.ȱManneȱ&ȱJoshuaȱD.ȱ Wright,ȱInnovationȱandȱtheȱLimitsȱofȱAntitrust,ȱ6ȱJ.ȱCOMPETITIONȱL.ȱ&ȱECON.ȱ153ȱ(2010).ȱ 11.ȱThisȱ conflationȱ ofȱ activityȱ levelȱ withȱ successȱ hasȱ comeȱ fromȱ aȱ numberȱ ofȱ sources,ȱincludingȱthenȬpresidentialȱcandidateȱBarackȱObama.ȱSeeȱBarackȱObama,ȱ Senator,ȱ Statementȱ toȱ theȱ Americanȱ AntitrustȱInstituteȱ (Nov.ȱ5,ȱ 2009),ȱ availableȱ atȱ http://www.antitrustinstitute.org/archives/files/aaiȬ%20Presidential%20campaignȱ %20Ȭ%20Obama%209Ȭ07_092720071759.pdfȱ (promisingȱ toȱ “reinvigorateȱ antitrustȱ enforcement”ȱandȱassertingȱthatȱtheȱactivityȱlevelȱofȱenforcementȱduringȱtheȱBushȱ administrationȱcausedȱnegativeȱconsequencesȱforȱconsumers).ȱOfficialsȱalsoȱhaveȱ largelyȱ repudiatedȱ theȱ wellȬacceptedȱ errorȬcostȱ framework.ȱ Theȱ FTCȱ Chairmanȱ andȱ Commissionerȱ declaredȱ thatȱ “thereȱ isȱ noȱ suchȱ thingȱ asȱ aȱ falseȱ positive.”ȱ Seeȱ Varney,ȱ supraȱ noteȱ 3;ȱ seeȱ alsoȱ J.ȱ Thomasȱ Rosch,ȱ Comm’r,ȱ Fed.ȱ Tradeȱ Comm’n,ȱ Thoughtsȱ onȱ theȱ Withdrawalȱ ofȱ theȱ Sectionȱ 2ȱ Report,ȱ Remarksȱ Beforeȱ IBA/ABAȱ ConferenceȱonȱAntitrustȱinȱaȱGlobalȱEconomyȱ(Juneȱ25,ȱ2009).ȱ
6ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
oneȱFederalȱTradeȱCommissionerȱhasȱlamentedȱtheȱevolutionȱofȱ antitrustȱrulesȱthat,ȱinȱhisȱview,ȱsystematicallyȱunderȬdeterȱantiȬ competitiveȱ behaviorȱ becauseȱ ofȱ theȱ incorporationȱ ofȱ theȱ errorȬ costȱframeworkȱandȱconcomitantȱconcernsȱaboutȱfalseȱpositivesȱ intoȱShermanȱActȱjurisprudence.12ȱ Theseȱ recentȱ developments,ȱ whichȱ areȱ impelledȱ byȱ theȱ imȬ plicitȱ beliefȱ thatȱ antitrustȱ interventionȱ isȱ nearlyȱ alwaysȱ benefiȬ cialȱfromȱaȱlongȬtermȱconsumerȬwelfareȱperspective,ȱportendȱaȱ movementȱ awayȱ fromȱ competitionȱ policyȱ informedȱ byȱ errorȬ costȱ analysis.ȱ Thisȱ approachȱ standsȱ inȱ starkȱ contrastȱ toȱ theȱ erȬ rorȬcostȱframework,ȱwhichȱpresumesȱthatȱerrorsȱareȱanȱinevitaȬ bleȱ andȱ coreȱ featureȱ ofȱ theȱ antitrustȱ enterprise.ȱ Theȱ newȱ approachȱ impliesȱ thatȱ concernsȱ aboutȱ overȬdeterrenceȱ shouldȱ notȱaffectȱeitherȱenforcementȱdecisionsȱorȱtheȱdesignȱofȱliabilityȱ rules.ȱ Indeed,ȱ advocatesȱ ofȱ thisȱ approachȱ suggestȱ thatȱ errorȬ costȱ concernsȱ areȱ antiquatedȱ inȱ theȱ Newȱ Economy,ȱ andȱ thatȱ falseȱpositivesȱareȱnoȱlongerȱaȱconceptȱcapableȱ ofȱcontributingȱ toȱtheȱantitrustȱpolicyȱdebates.ȱThisȱisȱaȱproblematicȱstanceȱthatȱ isȱ contraryȱ toȱ modernȱ economicsȱ andȱ theȱ logicȱ ofȱ legalȱ rules,ȱ andȱitȱportendsȱaȱcostlyȱmistakeȱinȱtheȱperhapsȱinevitableȱantiȬ trustȱcaseȱagainstȱGoogle.ȱ Partȱ IIȱ willȱ argueȱ that,ȱ contraryȱ toȱ theseȱ recentȱ criticsȱ andȱ agencyȱ authorities,ȱ errorȬcostȱ analysisȱ isȱ notȱ onlyȱ helpful,ȱ butȱ essentialȱtoȱidentifyingȱandȱdesigningȱoptimalȱantitrustȱrulesȱinȱ theȱ Newȱ Economy.13ȱ Theȱ applicationȱ ofȱ theȱ errorȬcostȱ frameȬ workȱ inȱ antitrustȱ originatesȱ withȱ Judgeȱ Frankȱ Easterbrook’sȱ seminalȱ analysis,ȱ Theȱ Limitsȱ ofȱ Antitrust,ȱ whichȱ wasȱ builtȱ onȱ twinȱ premises:ȱ first,ȱ thatȱ falseȱ positivesȱ areȱ moreȱ costlyȱ thanȱ falseȱnegativesȱbecauseȱselfȬcorrectionȱmechanismsȱmitigateȱtheȱ latterȱbutȱnotȱtheȱformer,ȱandȱsecond,ȱthatȱerrorsȱofȱbothȱtypesȱ areȱ inevitableȱ becauseȱ distinguishingȱ procompetitiveȱ conductȱ fromȱ anticompetitiveȱ conductȱ isȱ anȱ inherentlyȱ difficultȱ taskȱ inȱ theȱsingleȬfirmȱcontext.14ȱAtȱitsȱcore,ȱtheȱerrorȬcostȱframeworkȱisȱ aȱsimpleȱbutȱpowerfulȱanalyticalȱtoolȱthatȱrequiresȱinputsȱfromȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 12.ȱRosch,ȱsupraȱnoteȱ11.ȱ 13.ȱJudgeȱRichardȱA.ȱPosnerȱdefinedȱtheȱ“NewȱEconomy”ȱtoȱdenoteȱtheȱriseȱofȱ threeȱindustries:ȱmanufactureȱofȱcomputerȱsoftware,ȱInternetȬbasedȱbusiness,ȱandȱ communicationȱservices.ȱRichardȱA.ȱPosner,ȱAntitrustȱinȱtheȱNewȱEconomyȱ2ȱ(Univ.ȱ ofȱChi.ȱLawȱSch.,ȱJohnȱM.ȱOlinȱLawȱ&ȱEcon.ȱWorkingȱPaperȱNo.ȱ106,ȱ2000),ȱavailȬ ableȱatȱhttp://www.law.uchicago.edu/files/files/106.Posner.pdf.ȱ 14.ȱȱEasterbrook,ȱsupraȱnoteȱ10.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
7ȱ
stateȬofȬtheȬartȱ economicȱ theoryȱ andȱ empiricalȱ evidenceȱ reȬ gardingȱtheȱcompetitiveȱconsequencesȱofȱvariousȱtypesȱofȱbusiȬ nessȱ conduct,ȱ andȱ thatȱ producesȱ outputsȱ inȱ theȱ formȱ ofȱ legalȱ rules.ȱ Althoughȱ legalȱ scholarsȱ typicallyȱ avoidȱ rigorousȱ atȬ temptsȱ toȱ workȱ throughȱ theȱ availableȱ economicȱ theoryȱ andȱ evidenceȱ whenȱ discussingȱ theȱ optimalȱ designȱ ofȱ legalȱ rules,ȱ economistsȱfrequentlyȱfailȱtoȱassessȱtheirȱanalysesȱinȱaȱrealisticȱ institutionalȱsettingȱandȱavoidȱincorporatingȱtheȱsocialȱcostsȱofȱ erroneousȱ enforcementȱ decisionsȱ intoȱ theirȱ analysesȱ andȱ recȬ ommendationsȱ forȱ legalȱ rules.ȱ Partȱ IIȱ outlinesȱ theȱ commonȱ sourcesȱ andȱ theȱ historyȱ ofȱ antitrustȱ error,ȱ settingȱ theȱ stageȱ forȱ anȱassessmentȱofȱtheȱcaseȱagainstȱGoogle.ȱȱ PartȱIIIȱwillȱdiscussȱtheȱmarkets,ȱbusinessȱconduct,ȱandȱecoȬ nomicsȱ ofȱ onlineȱ searchȱ advertisingȱ relevantȱ toȱ theȱ primaryȱ monopolizationȱ argumentsȱ leveledȱ againstȱ Google.ȱ Inȱ particuȬ lar,ȱ thisȱ Partȱ highlightsȱ theȱ indeterminacyȱ andȱ theȱ complexityȱ ofȱ theȱ economicȱ implicationsȱ ofȱ Google’sȱ marketȱ andȱ itsȱ conȬ duct—characteristicsȱthatȱcontributeȱsignificantlyȱtoȱtheȱriskȱofȱ anȱinefficientȱintervention.ȱȱ PartȱIVȱwillȱdiscussȱtheȱpotentialȱmonopolizationȱclaimsȱagainstȱ Google,ȱandȱweȱhighlightȱtheȱpitfallsȱofȱtheȱhypotheticalȱcase,ȱconȬ cludingȱthatȱtheȱsuitȱisȱaȱrecipeȱforȱaȱcostlyȱfalseȬpositiveȱoutcome.ȱ Ourȱ goalȱ inȱ thisȱ paperȱ isȱ toȱ harnessȱ theȱ powerȱ ofȱ theȱ errorȬcostȱ frameworkȱtoȱintroduceȱanȱEasterbrookian,ȱerrorȬcostȬminimizingȱ approachȱtoȱantitrustȱinterventionȱinȱGoogle’sȱprimaryȱactivities— areasȱwhereȱinnovationȱisȱaȱcriticalȱpartȱofȱtheȱcompetitiveȱlandȬ scape.ȱGivenȱrecentȱactivitiesȱinȱtheȱantitrustȱenforcementȱarena— identifyingȱ innovativeȱ firmsȱ inȱ highȬtechȱ marketsȱ asȱ likelyȱ antiȬ trustȱtargetsȱcombinedȱwithȱrecentȱdiscussionsȱofȱerrorȱcostsȱfromȱ leadingȱenforcers,15ȱatȱtheȱSectionȱ2ȱHearings,16ȱandȱelsewhere17— systematicȱanalysisȱofȱtheȱrelationshipsȱbetweenȱinnovation,ȱantiȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 15.ȱSeeȱVarney,ȱsupraȱnoteȱ3.ȱ 16.ȱU.S.ȱ DEP’Tȱ OFȱ JUSTICE,ȱ COMPETITIONȱ ANDȱ MONOPOLY:ȱ SINGLEȬFIRMȱ CONDUCTȱ UNDERȱ SECTIONȱ 2ȱ OFȱ THEȱ SHERMANȱ ACTȱ (2008),ȱ availableȱ atȱ http://www.usdoj.gov/ȱ atr/public/reports/236681.pdf.ȱ Theȱ reportȱ wasȱ withdrawnȱ theȱ followingȱ year.ȱ Seeȱ PressȱRelease,ȱU.S.ȱDep’tȱofȱJustice,ȱJusticeȱDepartmentȱWithdrawsȱReportȱonȱAntiȬ trustȱ Monopolyȱ Lawȱ (Mayȱ 11,ȱ 2009),ȱ availableȱ atȱ http://www.justice.gov/atr/public/ȱ press_releases/2009/245710.htm.ȱȱ 17.ȱSee,ȱ e.g.,ȱ Staceyȱ L.ȱ Doganȱ &ȱ Markȱ A.ȱ Lemley,ȱ Antitrustȱ Lawȱ andȱ Regulatoryȱ Gaming,ȱ87ȱTEX.ȱ L.ȱ REV.ȱ685,ȱ700–03ȱ(2009)ȱ(discussingȱtheȱcostsȱofȱfalseȱpositivesȱ andȱfalseȱnegatives).ȱ
8ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
trustȱerror,ȱoptimalȱliabilityȱrulesȱandȱtheȱbusinessȱandȱeconomicsȱ ofȱGoogleȱisȱtimelyȱandȱessential.ȱȱ PartȱVȱwillȱconcludeȱthatȱinȱlightȱofȱtheȱantitrustȱclaimsȱarisingȱ outȱofȱinnovativeȱcontractualȱandȱpricingȱconduct,ȱandȱtheȱapparȬ entȱ lackȱ ofȱ anyȱ concreteȱ evidenceȱ ofȱ anticompetitiveȱ effectsȱ orȱ harmȱtoȱcompetition,ȱanȱenforcementȱactionȱagainstȱGoogleȱisȱillȱ advised.ȱEnforcementȱonȱtheseȱgroundsȱcreatesȱsubstantialȱriskȱforȱ aȱ falseȱ positive,ȱ whichȱ wouldȱ chillȱ innovationȱ andȱ competitionȱ thatȱcurrentlyȱprovidesȱimmenseȱbenefitsȱtoȱconsumers.ȱ II.
INNOVATION,ȱERRORȱCOSTS,ȱANDȱTHEȱȱ LIMITSȱOFȱANTITRUST18ȱ
TheȱprimaryȱcontributionȱofȱJudgeȱEasterbrook’sȱTheȱLimitsȱofȱ Antitrustȱ wasȱ toȱ forceȱ theȱ antitrustȱ communityȱ toȱ thinkȱ muchȱ moreȱrigorouslyȱaboutȱtheȱrelationshipȱbetweenȱerrorsȱandȱanȬ titrustȱ liabilityȱ rules.ȱ Althoughȱ theȱ errorȬcostȱ frameworkȱ isȱ aȱ criticalȱconceptualȱtoolȱthatȱcanȱcomfortablyȱbeȱappliedȱtoȱanyȱ areaȱofȱtheȱlaw,ȱitȱisȱespeciallyȱusefulȱinȱantitrustȱgivenȱtheȱunȬ derappreciatedȱ difficultyȱ ofȱ theȱ taskȱ thatȱ antitrustȱ lawȱ assignsȱ toȱ judges:ȱ toȱ distinguishȱ anticompetitiveȱ behaviorȱ fromȱ proȬ competitiveȱ behaviorȱ givenȱ limitedȱ evidence,ȱ alongȱ withȱ anyȱ cluesȱ economicȱ theoryȱ mightȱ provide.ȱ Thus,ȱ theȱ problemȱ ofȱ dealingȱwithȱerrorȱinȱtheȱdesignȱofȱtheȱliabilityȱrulesȱthemselvesȱ isȱanȱimportantȱinnovationȱinȱantitrust.ȱFromȱsimpleȱlegalȱandȱ economicȱ assumptions,ȱ thisȱ frameworkȱ providesȱ aȱ coherentȱ structureȱwithinȱwhichȱjudgesȱcanȱthinkȱaboutȱtheȱoptimalȱdeȬ signȱ ofȱ antitrustȱ rulesȱ inȱ theȱ faceȱ ofȱ expectedȱ errors.ȱ Theȱ framework’sȱ assumptionsȱ areȱ asȱ follows:ȱ First,ȱ bothȱ Typeȱ Iȱ (falseȱpositive)ȱandȱTypeȱIIȱ(falseȱnegative)ȱerrorsȱareȱinevitableȱ inȱantitrustȱcasesȱbecauseȱofȱtheȱdifficultyȱinȱdistinguishingȱefȬ ficient,ȱ procompetitiveȱ businessȱ conductȱ fromȱ anticompetitiveȱ behavior.19ȱ Second,ȱ theȱ socialȱ costsȱ associatedȱ withȱ Typeȱ Iȱ erȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 18.ȱPartȱIIȱisȱdistilledȱfromȱourȱrecentȱwork,ȱManneȱ&ȱWright,ȱsupraȱnoteȱ10.ȱ 19.ȱThereȱareȱtwoȱseparateȱpointsȱhere.ȱTheȱfirstȱisȱtheȱinevitabilityȱofȱerrorsȱwithȱ decisionȱbyȱlegalȱruleȱgenerally.ȱSeeȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ14–15ȱ(reiteratȬ ingȱthatȱ“[o]neȱcannotȱhaveȱtheȱsavingsȱofȱdecisionȱbyȱruleȱwithoutȱacceptingȱtheȱ costsȱ ofȱ mistakes”).ȱ Theȱ secondȱ pointȱ isȱ thatȱ theȱ likelihoodȱ ofȱ antitrustȱ errorȱ deȬ pendsȱ cruciallyȱ onȱ theȱ developmentȱ ofȱ economicȱ scienceȱ toȱ produceȱ techniquesȱ andȱmethodsȱbyȱwhichȱweȱcanȱsuccessfullyȱidentifyȱconductȱthatȱharmsȱconsumȬ ers.ȱ Seeȱ Frankȱ H.ȱ Easterbrook,ȱ Workableȱ Antitrustȱ Policy,ȱ 84ȱ MICH.ȱ L.ȱ REV.ȱ 1696,ȱ 1712ȱ(1986).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
9ȱ
rorsȱareȱgenerallyȱgreaterȱthanȱtheȱsocialȱcostsȱofȱTypeȱIIȱerrorsȱ becauseȱ marketȱ forcesȱ offerȱ atȱ leastȱ someȱ correctionȱ withȱ reȬ spectȱtoȱTypeȱIIȱerrors,ȱbutȱnoneȱwithȱregardȱtoȱTypeȱIȱerrors.20ȱ Third,ȱ optimalȱ antitrustȱ rulesȱ willȱ minimizeȱ theȱ expectedȱ sumȱ ofȱ errorȱ costsȱ subjectȱ toȱ theȱ constraintȱ thatȱ theȱ rulesȱ beȱ relaȬ tivelyȱ simpleȱ andȱ reasonablyȱ administrable.21ȱ Thisȱ frameworkȱ givesȱ riseȱ toȱ aȱ numberȱ ofȱ simpleȱ filtersȱ thatȱ canȱ beȱ usedȱ toȱ minimizeȱerrorȱcosts.ȱPlaintiffsȱcanȱbeȱrequiredȱtoȱaffirmativelyȱ demonstrateȱ thatȱ theȱ firmȱ atȱ issueȱ actuallyȱ hasȱ marketȱ powerȱ andȱ thatȱ theȱ practicesȱ atȱ issueȱ areȱ substantiallyȱ likelyȱ toȱ harmȱ consumers.ȱ Courtsȱ canȱ questionȱ whetherȱ firmsȱ inȱ theȱ industryȱ useȱdifferentȱmethodsȱofȱproductionȱorȱdistribution,ȱwhetherȱtheȱ evidenceȱ isȱ consistentȱ withȱ aȱ reductionȱ inȱ output,ȱ andȱ whetherȱ theȱcomplainingȱfirmȱisȱaȱrivalȱinȱtheȱrelevantȱmarket.22ȱ JudgeȱEasterbrook’sȱanalysisȱofȱantitrustȱerrorsȱlendsȱitselfȱtoȱaȱ Bayesianȱ decisionȬtheoreticȱ framework,ȱ designedȱ toȱ addressȱ problemsȱ ofȱ decisionmakingȱ underȱ uncertainty.ȱ Economistsȱ haveȱappliedȱtheȱframeworkȱtoȱidentifyȱoptimalȱrulesȱforȱaȱrangeȱ ofȱpracticesȱincludingȱtying,ȱexclusiveȱdealing,ȱmergers,ȱȱandȱreȬ saleȱ priceȱ maintenance.23ȱ Applyingȱ theȱ Bayesianȱ approach,ȱ theȱ regulator,ȱcourt,ȱorȱpolicymakerȱholdsȱanȱearlierȱbeliefȱaboutȱtheȱ likelihoodȱ thatȱ aȱ specificȱ businessȱ practiceȱ isȱ anticompetitive.ȱ EarlierȱbeliefsȱareȱupdatedȱwithȱcaseȬspecificȱinformationȱorȱnewȱ evidenceȱ asȱ theȱ theoreticalȱ andȱ empiricalȱ understandingȱ ofȱ theȱ practiceȱevolves.ȱTheȱoptimalȱdecisionȱruleȱisȱthenȱbasedȱonȱtheȱ new,ȱupdatedȱlikelihoodȱthatȱtheȱpracticeȱwillȱbeȱanticompetitiveȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 20.ȱJudgeȱ Easterbrookȱ wellȱ articulatedȱ thisȱ phenomenon:ȱ “[T]heȱ economicȱ sysȬ temȱ correctsȱ monopolyȱ moreȱ readilyȱ thanȱ itȱ correctsȱ judicialȱ [Typeȱ I]ȱ errors.”ȱ Easterbrook,ȱsupraȱnoteȱ10,ȱatȱ15.ȱ 21.ȱThisȱ pointȱ isȱ mostȱ oftenȱ attributedȱ toȱ thenȬJudgeȱ Breyer’sȱ wellȬknownȱ adȬ monitionȱ thatȱ antitrustȱ rulesȱ “mustȱ beȱ administrativelyȱ workableȱ andȱ thereforeȱ cannotȱalwaysȱtakeȱaccountȱofȱeveryȱcomplexȱeconomicȱcircumstanceȱorȱqualificaȬ tion.”ȱTownȱofȱConcordȱv.ȱBos.ȱEdisonȱCo.,ȱ915ȱF.2dȱ17,ȱ22ȱ(1stȱCir.ȱ1990).ȱButȱtheȱ ChicagoȱSchoolȱofȱantitrustȱhasȱtraditionallyȱsharedȱwithȱBreyer’sȱHarvardȱSchoolȱ aȱ preferenceȱ forȱ usingȱ economicsȱ toȱ generateȱ simpleȱ andȱ administrableȱ rulesȱ ratherȱthanȱoverlyȱsophisticatedȱeconomicȱtests.ȱSeeȱWilliamȱE.ȱKovacic,ȱTheȱIntelȬ lectualȱ DNAȱ ofȱ Modernȱ U.S.ȱ Competitionȱ Lawȱ forȱ Dominantȱ Firmȱ Conduct:ȱ Theȱ ChiȬ cago/Harvardȱ Doubleȱ Helix,ȱ 2007ȱ COLUM.ȱ BUS.ȱ L.ȱ REV.ȱ 1,ȱ 32–35ȱ (2007);ȱ Joshuaȱ D.ȱ Wright,ȱ Theȱ Robertsȱ Courtȱ andȱ theȱ Chicagoȱ Schoolȱ ofȱ Antitrust:ȱ Theȱ 2006ȱ Termȱ andȱ Beyond,ȱ3ȱCOMPETITIONȱPOL’YȱINT’Lȱ25,ȱ37ȱ(2007).ȱ 22.ȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ18.ȱ 23.ȱSeeȱgenerallyȱBecknerȱIIIȱ&ȱSalop,ȱsupraȱnoteȱ10;ȱHyltonȱ&ȱSalinger,ȱsupraȱnoteȱ 10;ȱ Joshuaȱ D.ȱ Wright,ȱ Overshotȱ theȱ Mark?ȱ Aȱ Simpleȱ Explanationȱ ofȱ theȱ Chicagoȱ School’sȱInfluenceȱonȱAntitrust,ȱ5ȱCOMPETITIONȱPOL’YȱINT’Lȱ179ȱ(2009).ȱ
10ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
byȱminimizingȱaȱlossȱfunctionȱmeasuringȱtheȱsocialȱcostsȱofȱTypeȱ Iȱ(falseȱpositive)ȱandȱTypeȱIIȱ(falseȱnegative)ȱerrors.24ȱ TheȱfundamentalȱinsightȱfromȱtheȱLimitsȱofȱAntitrust,ȱtheȱinȬ sightȱthatȱdrivesȱaȱwedgeȱbetweenȱtheȱtreatmentȱofȱTypeȱIȱandȱ TypeȱIIȱerrors,ȱisȱthatȱTypeȱIȱerrorsȱareȱlikely,ȱonȱaverage,ȱtoȱbeȱ moreȱ costlyȱ toȱ societyȱ andȱ consumersȱ thanȱ Typeȱ IIȱ errorsȱ beȬ causeȱmarketȱforcesȱplaceȱsomeȱconstraintsȱonȱtheȱlatterȱbutȱnotȱ theȱ former.25ȱ Thisȱ insightȱ becomesȱ moreȱ importantȱ asȱ ourȱ colȬ lectiveȱ economicȱ wisdomȱ aboutȱ aȱ newȱ businessȱ practiceȱ deȬ creases—whenȱ aȱ challengedȱ practiceȱ orȱ settingȱ isȱ innovative.ȱ TheȱerrorȬcostȱframeworkȱcallsȱforȱaȱmoreȱinterventionistȱantiȬ trustȱruleȱonlyȱwhenȱTypeȱIIȱerrorȱcostsȱareȱsubstantial,ȱthereȱisȱ aȱlongȬstandingȱprecedentȱindicatingȱthatȱtheȱgivenȱpracticeȱisȱ anticompetitive,ȱandȱtheoryȱandȱevidenceȱsuggestȱaȱstrongȱlikeȬ lihoodȱthatȱtheȱpracticeȱisȱanticompetitive.ȱ ThereȱareȱseveralȱpotentialȱsourcesȱofȱerrorȱinȱantitrustȱanalyȬ sisȱandȱenforcement,ȱhowever,ȱthisȱArticleȱfocusesȱalmostȱexcluȬ sivelyȱonȱtheȱmostȱsignificantȱtypeȱofȱerror:ȱaȱcourtȱorȱregulator’sȱ erroneousȱ conclusionȱ thatȱ aȱ practiceȱ isȱ anticompetitiveȱ dueȱ toȱ theȱdifficultyȱofȱidentifyingȱanticompetitiveȱconductȱandȱdistinȬ guishingȱ itȱ fromȱ precompetitiveȱ conductȱ inȱ anyȱ specificȱ case.ȱ Judgesȱareȱoftenȱprematurelyȱledȱtoȱcondemnȱbusinessȱpracticesȱ asȱ anticompetitiveȱ becauseȱ antitrustȱ lawyersȱ andȱ economistsȱ tendȱ toȱ systematicallyȱ assignȱ anticompetitiveȱ explanationsȱ toȱ conductȱ thatȱ isȱ novelȱ andȱ notȱ wellȱ understood.26ȱ Moreover,ȱ judgesȱnotȱgenerallyȱtrainedȱinȱeconomicsȱareȱaskedȱtoȱmakeȱinȬ creasinglyȱ sophisticatedȱ economicȱ determinations,ȱ andȱ errorsȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 24.ȱSeeȱJamesȱC.ȱCooperȱetȱal.,ȱVerticalȱAntitrustȱPolicyȱasȱaȱProblemȱofȱInference,ȱ23ȱ INT’Lȱ J.ȱ INDUS.ȱ ORG.ȱ639ȱ(2005);ȱseeȱalsoȱKeithȱN.ȱHylton,ȱTheȱLawȱandȱEconomicsȱofȱ MonopolizationȱStandards,ȱinȱANTITRUSTȱ LAWȱANDȱ ECONOMICSȱ 82ȱ(KeithȱN.ȱHyltonȱ ed.,ȱ2dȱed.ȱ2010).ȱ 25.ȱMultipleȱacademicsȱreviewȱtheȱexistingȱtheoryȱandȱevidenceȱonȱverticalȱrestraintsȱ andȱsingleȬfirmȱconductȱmoreȱgenerallyȱandȱuniformlyȱconcludeȱthatȱtheȱpracticesȱatȱ issueȱ areȱ generallyȱ procompetitiveȱ andȱ thatȱ antitrustȱ rulesȱ shouldȱ “slant”ȱ towardsȱ requiringȱ plaintiffsȱ toȱ demonstrateȱ clearȱ anticompetitiveȱ effectȱ beforeȱ condemningȱ findingȱviolations.ȱSee,ȱe.g.,ȱCooperȱetȱal.,ȱsupraȱnoteȱ24;ȱFrancineȱLafontaineȱ&ȱMargaȬ retȱSlade,ȱExclusiveȱContractsȱandȱVerticalȱRestraints:ȱEmpiricalȱEvidenceȱandȱPublicȱPolicy,ȱ inȱ HANDBOOKȱ OFȱ ANTITRUSTȱ ECONOMICSȱ 391ȱ (Paoloȱ Buccirossiȱ ed.,ȱ 2008);ȱ Danielȱ P.ȱ O’Brien,ȱTheȱAntitrustȱTreatmentȱofȱVerticalȱRestraints:ȱBeyondȱtheȱPossibilityȱofȱTheorems,ȱ inȱTHEȱ PROSȱANDȱ CONSȱOFȱ VERTICALȱ RESTRAINTSȱ40ȱ(SwedishȱCompetitionȱAuth.ȱed.,ȱ 2008),ȱ availableȱ atȱ http://www.konkurrensverket.se/upload/Filer/Trycksaker/Rapporter/ȱ Pros&Cons/rap_pros_and_cons_vertical_restraints.pdf.ȱȱ 26.ȱSee,ȱe.g.,ȱinfraȱnoteȱ31.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
11ȱ
areȱinevitable,ȱnotȱbecauseȱtheȱeconomicȱtheoryȱisȱdemonstrablyȱ wrongȱorȱinadequateȱ(althoughȱsometimesȱitȱmayȱbe),ȱbutȱratherȱ becauseȱrelianceȱonȱtheȱtheoryȱmayȱbeȱinappropriate.27ȱ Consider,ȱ forȱ example,ȱ economicȱ knowledgeȱ concerningȱ theȱ relationshipȱ betweenȱ marketȱ concentrationȱ andȱ price.ȱ Duringȱ theȱlateȱ1950sȱandȱearlyȱ1960s,ȱeconomicȱanalysisȱviewedȱmarȬ ketȱ concentrationȱ andȱ oligopolisticȱ collusionȱ asȱ theȱ “principalȱ defectȱ ofȱ presentȱ antitrustȱ law.”28ȱ Scholarsȱ urgedȱ Congressȱ toȱ passȱ newȱ legislationȱ aimedȱ atȱ reducingȱ marketȱ concentrationȱ acrossȱtheȱeconomy,ȱandȱaȱWhiteȱHouseȱTaskȱForceȱReportȱonȱ Antitrustȱ Policyȱ endorsedȱ variousȱ formsȱ ofȱ suchȱ proposals.29ȱ Caseȱ lawȱ ofȱ theȱ eraȱ largelyȱ mirroredȱ thisȱ economicȱ analysis,30ȱ andȱasȱtheȱanalysisȱhasȱbeenȱdebunked,ȱtheȱcaseȱlawȱhasȱcomeȱ toȱbeȱuniversallyȱcriticized.31ȱ Evenȱatȱtheȱtime,ȱdissentingȱJusticesȱscrutinizedȱeconomicȱerȬ rorsȱandȱcontradictionsȱinȱtheȱCourt’sȱanalyses.ȱInȱhisȱdissentȱinȱ Von’sȱ Groceryȱ v.ȱ Unitedȱ States,32ȱ forȱ example,ȱ Justiceȱ Stewartȱ notedȱ thatȱ “evenȱ theȱ mostȱ superficialȱ analysisȱ ofȱ theȱ recordȱ makesȱ plainȱ theȱ fallacyȱ ofȱ theȱ Court’sȱ syllogismȱ thatȱ competiȬ tionȱisȱnecessarilyȱreducedȱwhenȱtheȱbareȱnumberȱofȱcompetiȬ torsȱhasȱdeclined.”33ȱNorȱdidȱtheȱrelationshipȱbetweenȱantitrustȱ errorȱ andȱ innovationȱ escapeȱ Justiceȱ Stewart,ȱ whoȱ admonishedȱ theȱmajorityȱinȱVon’sȱGroceryȱthatȱ“[theȱClaytonȱAct]ȱwasȱneverȱ intendedȱbyȱCongressȱforȱuseȱbyȱtheȱCourtȱasȱaȱcharterȱtoȱrollȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 27.ȱSeeȱgenerallyȱMichaelȱR.ȱBayeȱ&ȱJoshuaȱD.ȱWright,ȱIsȱAntitrustȱTooȱComplicatedȱ forȱ Generalistȱ Judges?ȱ Theȱ Impactȱ ofȱ Economicȱ Complexityȱ &ȱ Judicialȱ Trainingȱ onȱ ApȬ peals,ȱJ.L.ȱ&ȱECON.ȱ(forthcomingȱ2010)ȱ(findingȱaȱstatisticallyȱsignificantȱtendencyȱ forȱeconomicallyȱtrainedȱjudgesȱtoȱperformȱbetterȱinȱsimpleȱantitrustȱcases,ȱbutȱnotȱ thoseȱinvolvingȱsophisticatedȱeconomicȱevidence).ȱ 28.ȱCARLȱKAYSENȱ&ȱDONALDȱF.ȱTURNER,ȱANTITRUSTȱPOLICY:ȱANȱECONOMICȱANDȱ LEGALȱANALYSISȱ110ȱ(1959).ȱ 29.ȱPhilȱC.ȱNealȱetȱal.,ȱReportȱofȱtheȱWhiteȱHouseȱTaskȱForceȱonȱAntitrustȱPolicy,ȱ2ȱ ANTITRUSTȱL.ȱ&ȱECON.ȱREV.ȱ11,ȱ14–15,ȱ65–76ȱ(1968–69).ȱ 30.ȱFTCȱ v.ȱ Procterȱ &ȱ Gambleȱ Co.,ȱ 386ȱ U.S.ȱ 568ȱ (1967);ȱ Unitedȱ Statesȱ v.ȱ Pabstȱ BrewingȱCo.,ȱ384ȱU.S.ȱ546ȱ(1966);ȱUnitedȱStatesȱv.ȱVon’sȱGroceryȱCo.,ȱ384ȱU.S.ȱ270ȱ (1966);ȱ seeȱ alsoȱ Williamȱ E.ȱ Kovacicȱ &ȱ Carlȱ Shapiro,ȱ Antitrustȱ Policy:ȱ Aȱ Centuryȱ ofȱ Legalȱ andȱ Economicȱ Thinking,ȱ 14ȱ J.ȱ ECON.ȱ PERSP.ȱ 43,ȱ 51–52ȱ (2000);ȱ Williamȱ E.ȱ Kovacic,ȱTheȱInfluenceȱofȱEconomicsȱonȱAntitrustȱLaw,ȱ30ȱJ.ȱ ECON.ȱ INQUIRYȱ294,ȱ295– 96ȱ(1992)ȱ(describingȱtheȱfeaturesȱofȱtheȱU.S.ȱcompetitionȱpolicyȱsystemȱthatȱgiveȱ economistsȱaȱmajorȱroleȱinȱshapingȱantitrustȱrules).ȱ 31.ȱSeeȱgenerallyȱROBERTȱH.ȱBORK,ȱTHEȱANTITRUSTȱPARADOXȱ(1993);ȱPosner,ȱsupraȱ noteȱ10.ȱ 32.ȱ384ȱU.S.ȱ270ȱ(1966).ȱ 33.ȱId.ȱatȱ287ȱ(Stewart,ȱJ.,ȱdissenting).ȱ
12ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
backȱ theȱ supermarketȱ revolution”ȱ andȱ madeȱ theȱ obviousȱ ecoȬ nomicȱ pointȱ thatȱ “theȱnumericalȱ declineȱ inȱ theȱnumberȱ ofȱ sinȬ gleȬstoreȱ ownersȱ isȱ theȱ resultȱ ofȱ transcendingȱ socialȱ andȱ technologicalȱ changesȱ thatȱ positivelyȱ precludeȱ theȱ inferenceȱ thatȱ competitionȱ hasȱ sufferedȱ becauseȱ ofȱ theȱ attritionȱ ofȱ comȬ petitors.”34ȱButȱtheȱcostlyȱerrorsȱinȱVon’sȱGroceryȱandȱotherȱcasesȱ fromȱ thatȱ eraȱ wereȱ largelyȱ attributableȱ toȱ theȱ Court’sȱ relianceȱ on,ȱratherȱthanȱrejectionȱof,ȱthenȬcurrentȱeconomicȱscience.ȱ Withoutȱ aȱ seriousȱ methodologicalȱ commitmentȱ toȱ economicȱ science,ȱtheȱincorporationȱofȱeconomicsȱintoȱantitrustȱisȱmerelyȱ aȱ façade,ȱ allowingȱ regulatorsȱ andȱ judgesȱ toȱ selectȱ whicheverȱ economicȱ modelȱ fitsȱ theirȱ earlierȱ beliefsȱ orȱ policyȱ preferencesȱ ratherȱ thanȱ theȱ modelȱ thatȱ bestȱ fitsȱ theȱ realȬworldȱ data.ȱ Still,ȱ economicȱ theoryȱ remainsȱ essentialȱ toȱ antitrustȱ law.ȱ Economicȱ analysisȱ constrainsȱ andȱ harnessesȱ antitrustȱ lawȱ soȱ thatȱ itȱ proȬ tectsȱ consumersȱ ratherȱ thanȱ competitors.ȱ Itȱ isȱ alsoȱ responsibleȱ forȱ theȱ successfulȱ evolutionȱ ofȱ antitrustȱ fromȱ itsȱ economicallyȱ incoherentȱ originsȱ toȱ itsȱ presentȱ state.ȱ Thus,ȱ aȱ fundamentalȱ challengeȱ forȱ antitrustȱ isȱ theȱ existenceȱ ofȱ tooȱ manyȱ theoriesȱ withoutȱ methodologicalȱ commitmentsȱ fromȱ regulatorsȱ andȱ courtsȱonȱhowȱtoȱselectȱamongȱthem.ȱȱ Asȱaȱresultȱofȱtheȱproliferationȱofȱeconomicȱmodelsȱthatȱcameȱ withȱ theȱ riseȱ ofȱ postȬChicagoȱ economics,ȱ theȱ integrationȱ ofȱ gameȱ theoryȱ intoȱ industrialȱ organization,ȱ andȱ theȱ increasingȱ callsȱ toȱ incorporateȱ insightsȱ fromȱ behavioralȱ economicsȱ intoȱ antitrustȱ andȱ competitionȱ policy,ȱ regulatorsȱ andȱ courtsȱ nowȱ haveȱaȱbroadȱspectrumȱofȱmodelsȱtoȱchooseȱfromȱwhenȱanalyzȬ ingȱanȱantitrustȱissue.ȱAtȱtheȱsameȱtime,ȱantitrustȱlawȱdoesȱnotȱ provideȱ theseȱ decisionmakersȱ withȱ sensibleȱ criteriaȱ forȱ selectȬ ingȱ whichȱ modelȱ toȱ use.ȱ Takenȱ toȱ theȱ extreme,ȱ thisȱ modelȬ selectionȱ problemȱ threatensȱ toȱ stripȱ theȱ discipliningȱ forceȱ thatȱ economicsȱ hasȱ placedȱ onȱ antitrustȱ law.ȱ Thisȱ discipliningȱ forceȱ hasȱ playedȱ aȱ keyȱ partȱ ofȱ theȱ successfulȱ evolutionȱ ofȱ antitrustȱ lawȱ overȱ theȱ lastȱ fiftyȱ years.35ȱ Theȱ powerȱ ofȱ theȱ errorȬcostȱ frameworkȱisȱthatȱitȱallowsȱregulators,ȱjudges,ȱandȱpolicymakȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 34.ȱId.ȱatȱ288.ȱ 35.ȱSeeȱJoshuaȱWright,ȱCommissionerȱRosch,ȱRhetoric,ȱandȱtheȱRelationshipȱBetweenȱ Economicsȱ andȱ Antitrust,ȱ TRUTHȱ ONȱ THEȱ MARKETȱ (Juneȱ 12,ȱ 2009,ȱ 11:55ȱ AM),ȱ http://www.truthonthemarket.com/2009/06/12/commissionerȬroschȬrhetoricȬandȬ theȬrelationshipȬbetweenȬeconomicsȬandȬantitrust/;ȱ seeȱ alsoȱ Bayeȱ &ȱ Wright,ȱ supraȱ noteȱ27.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
13ȱ
ersȱtoȱharnessȱtheȱpowerȱofȱeconomicsȱ(andȱstateȱofȱtheȱartȱtheȬ oryȱandȱevidence)ȱthroughȱtheȱformulationȱofȱsimpleȱandȱsenȬ sibleȱ filtersȱ andȱ safeȱ harbors,ȱ ratherȱ thanȱ convertȱ themselvesȱ intoȱamateurȱeconometricians,ȱgameȱtheorists,ȱorȱbehaviorists.ȱ Innovationȱ hasȱ theȱ potentialȱ toȱ magnifyȱ theseȱ errorsȱ inȱ twoȱ importantȱ ways.ȱ First,ȱ innovation,ȱ byȱ definition,ȱ generallyȱ inȬ volvesȱ newȱ businessȱ practicesȱ orȱ products.ȱ Antitrustȱ authoriȬ tiesȱ historicallyȱ haveȱ notȱ treatedȱ novelȱ businessȱ practicesȱ orȱ innovativeȱproductsȱkindly,ȱandȱeconomistsȱhaveȱaȱlongstandȬ ingȱ tendencyȱ toȱ ascribeȱ anticompetitiveȱ explanationsȱ toȱ newȱ formsȱofȱconductȱthatȱareȱnotȱwellȱunderstood.ȱAsȱNobelȱLauȬ reateȱ Ronaldȱ Coaseȱ describedȱ inȱ lamentingȱ theȱ stateȱ ofȱ theȱ inȬ dustrialȱorganizationȱliterature:ȱ [I]fȱ anȱ economistȱ findsȱ something—aȱ businessȱ practiceȱ ofȱ oneȱ sortȱorȱother—thatȱheȱdoesȱnotȱunderstand,ȱheȱlooksȱforȱaȱmoȬ nopolyȱexplanation.ȱAndȱasȱinȱthisȱfieldȱweȱareȱveryȱignorant,ȱ theȱ numberȱ ofȱ ununderstandableȱ practicesȱ tendsȱ toȱ beȱ veryȱ large,ȱandȱtheȱrelianceȱonȱaȱmonopolyȱexplanation,ȱfrequent.36ȱ
Withȱ theȱ increasingȱ integrationȱ ofȱ economicȱ conceptsȱ intoȱ antiȬ trustȱlawȱandȱalmostȱuniversalȱagreementȱthatȱmodernȱeconomicsȱ containsȱusefulȱtoolsȱforȱincorporatingȱinnovationȱeffectsȱintoȱtraȬ ditionalȱantitrustȱanalysis,37ȱtheȱantimarketȱbiasȱdescribedȱbyȱProȬ fessorȱ Coaseȱ isȱ likelyȱ toȱ produceȱ evenȱ moreȱ significantȱ policyȱ consequencesȱinȱmodernȱantitrust.ȱFromȱanȱerrorȬcostȱperspective,ȱ theȱcriticalȱpointȱisȱthatȱantitrustȱscrutinyȱofȱinnovationȱisȱlikelyȱtoȱ beȱbiasedȱtowardȱassigningȱaȱhigherȱlikelihoodȱthatȱaȱgivenȱpracȬ ticeȱisȱanticompetitiveȱthanȱlaterȱliteratureȱandȱevidenceȱwillȱultiȬ matelyȱsuggestȱisȱreasonableȱorȱaccurate.ȱȱ Second,ȱ thisȱ biasȱ towardȱ committingȱ Typeȱ Iȱ errorȱ isȱ furtherȱ skewedȱ becauseȱ economistsȱ generallyȱ knowȱ muchȱ lessȱ aboutȱ theȱ relationshipȱ betweenȱ competition,ȱ innovation,ȱ andȱ conȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 36.ȱRonaldȱ H.ȱ Coase,ȱ Industrialȱ Organization:ȱ Aȱ Proposalȱ forȱ Research,ȱ inȱ POLICYȱ ISSUESȱ ANDȱ RESEARCHȱ OPPORTUNITIESȱ INȱ INDUSTRIALȱ ORGANIZATIONȱ 59,ȱ 67ȱ (VicȬ torȱ R.ȱ Fuchs,ȱ ed.,ȱ 1972).ȱ Forȱ moreȱ modernȱ critiquesȱ ofȱ theȱ industrialȱ organizationȱ literatureȱinȱtheȱsameȱvein,ȱseeȱEvansȱ&ȱPadilla,ȱsupraȱnoteȱ10,ȱatȱ73;ȱ BruceȱH.ȱKobaȬ yashi,ȱGameȱTheoryȱandȱAntitrust:ȱAȱPostȱMortem,ȱ5ȱGEO.ȱ MASONȱ L.ȱ REV.ȱ411ȱ(1997);ȱ TimothyȱJ.ȱMuris,ȱEconomicsȱandȱAntitrust,ȱ5ȱGEO.ȱMASON.ȱL.ȱREV.ȱ303ȱ(1997).ȱ 37.ȱANTITRUSTȱ MODERNIZATIONȱ COMM’N,ȱ REPORTȱ ANDȱ RECOMMENDATIONSȱ 32ȱ (2007)ȱ(recommendingȱthatȱ“inȱindustriesȱwhereȱinnovation,ȱintellectualȱproperty,ȱ andȱ technologicalȱ changeȱ areȱ centralȱ featuresȱ.ȱ.ȱ.ȱantitrustȱ enforcersȱ shouldȱ careȬ fullyȱconsiderȱmarketȱdynamicsȱinȱassessingȱcompetitiveȱeffects”).ȱ
14ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
sumerȱwelfareȱthanȱtheyȱdoȱaboutȱstandardȱpriceȱcompetition.38ȱ Theȱ antitrustȱ communityȱ appearsȱ toȱ endorseȱ enthusiasticallyȱ theȱpropositionsȱthatȱantitrustȱanalysisȱshouldȱmoreȱrigorouslyȱ incorporateȱ dynamicȱ efficienciesȱ andȱ innovationȱ effects,ȱ andȱ thatȱ itȱ couldȱ comfortablyȱ doȱ soȱ withinȱ itsȱ currentȱ analyticalȱ framework.ȱ Theȱ Antitrustȱ Modernizationȱ Commission,ȱ forȱ exȬ ample,ȱ hasȱ statedȱ that:ȱ “Currentȱ antitrustȱ analysisȱ hasȱ aȱ suffiȬ cientȱ groundingȱ inȱ economicsȱ andȱ isȱ sufficientlyȱ flexibleȱ toȱ reachȱ appropriateȱ conclusionsȱ inȱ mattersȱ involvingȱ industriesȱ inȱ whichȱ innovation,ȱ intellectualȱ property,ȱ andȱ technologicalȱ changeȱareȱcentralȱfeatures.”39ȱ Thisȱ assertionȱ isȱ almostȱ certainlyȱ overstated.ȱ Althoughȱ weȱ knowȱthatȱinnovationȱisȱcriticalȱtoȱ economicȱgrowth,ȱtheȱtheoȬ reticalȱ literatureȱ relatingȱ toȱ competitionȱ andȱ innovationȱ reȬ mainsȱ insufficientȱ toȱ instillȱ anyȱ greatȱ confidenceȱ inȱ ourȱ abilityȱ toȱdetermineȱwhichȱantitrustȱpoliciesȱwillȱencourageȱinnovationȱ andȱresultȱinȱnetȱconsumerȱwelfareȱgains.ȱSpecifically,ȱourȱabilȬ ityȱtoȱapplyȱantitrustȱstandardsȱdependsȱonȱourȱabilityȱtoȱpreȬ dictȱ howȱ aȱ ruleȱ willȱ impactȱ theȱ mixtureȱ ofȱ competitiveȱ formsȱ thatȱwillȱexistȱafterȱtheȱpolicyȱisȱimplementedȱandȱtoȱrankȱtheseȱ mixturesȱ onȱ consumerȱ welfareȱ orȱ efficiencyȱ criteria.40ȱ Onȱ thisȱ dimension,ȱcurrentȱeconomicȱtheoryȱisȱindeterminateȱatȱbest.41ȱ Moreover,ȱ itȱ isȱ enormouslyȱ difficultȱ toȱ identifyȱ whenȱ aȱ speȬ cificȱapplicationȱofȱtheȱtheoryȱcanȱbeȱrejected.ȱThisȱdifficultyȱisȱaȱ keyȱcauseȱofȱbothȱeconomists’ȱincentivesȱtoȱidentifyȱtheoreticalȱ possibilitiesȱofȱanticompetitiveȱbehaviorȱandȱtheȱ“inhospitality“ȱ traditionȱofȱenforcersȱtoȱtakeȱadvantageȱofȱit.ȱ“WheneverȱanȱantiȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 38.ȱSee,ȱe.g.,ȱRobertȱCooter,ȱInformation,ȱCapitalȱMarkets,ȱandȱPlannedȱDevelopment:ȱ Anȱ Essay,ȱ inȱ REGULATINGȱ INNOVATION:ȱ COMPETITIONȱ POLICYȱ ANDȱ PATENTȱ LAWȱ UNDERȱ UNCERTAINTYȱ(GeoffreyȱA.ȱManneȱ&ȱJoshuaȱD.ȱWrightȱeds.,ȱforthcomingȱ 2011),ȱ availableȱ atȱ http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1462489;ȱ JoshuaȱD.ȱWright,ȱAntitrust,ȱMultiȬDimensionalȱCompetition,ȱandȱInnovation:ȱDoȱWeȱ Haveȱ Anȱ Antitrustȱ Relevantȱ Theoryȱ ofȱ Competitionȱ Now?,ȱ inȱ REGULATINGȱ INNOVATION:ȱ COMPETITIONȱ POLICYȱ ANDȱ PATENTȱ LAWȱ UNDERȱ UNCERTAINTYȱ (Geoffreyȱ A.ȱ Manneȱ &ȱ Joshuaȱ D.ȱ Wrightȱ eds.,ȱ forthcomingȱ 2011),ȱ availableȱ atȱ http://papers/ssrn.com/sol3/papers/cfm?abstract_id=142489.ȱ 39.ȱAntitrustȱModernizationȱComm’n,ȱsupraȱnoteȱ37,ȱatȱ38.ȱ 40.ȱSeeȱWright,ȱsupraȱnoteȱ35.ȱ 41.ȱRichardȱ J.ȱ Gilbert,ȱ Competitionȱ andȱ Innovation,ȱ inȱ 1ȱ ISSUESȱ INȱ COMPETITIONȱ LAWȱANDȱ POLICYȱ 577,ȱ583ȱ(W.ȱDaleȱCollinsȱed.,ȱ2008)ȱ(“Economicȱtheoryȱdoesȱnotȱ provideȱ unambiguousȱ supportȱ eitherȱ forȱ theȱ viewȱ thatȱ marketȱ powerȱ generallyȱ threatensȱ innovationȱ byȱ loweringȱ theȱ returnȱ toȱ innovativeȱ effortsȱ orȱ forȱ theȱ Schumpeterianȱviewȱthatȱconcentratedȱmarketsȱgenerallyȱpromoteȱinnovation.”).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
15ȱ
trustȱcourtȱisȱcalledȱuponȱtoȱbalanceȱefficiencyȱagainstȱmonopoly,ȱ thereȱ isȱ trouble;ȱ legalȱ uncertainty,ȱ andȱ theȱ likelihoodȱ ofȱ error,ȱ soar.”42ȱ Certainlyȱ anȱ exclusionaryȱ innovationȱ isȱ notȱ alwaysȱ anticompetitive.ȱEvenȱanȱinnovationȱthatȱmightȱbeȱanticompetiȬ tiveȱ sometimesȱ willȱ unlikelyȱ beȱ anticompetitiveȱ allȱ theȱ time.ȱ Thus,ȱ aȱ keyȱ critiqueȱ ofȱ theȱ modernȱ industrialȱ organizationȱ litȬ eratureȱ andȱ itsȱ possibilityȱ theoremsȱ involvingȱ anticompetitiveȱ behaviorȱ hasȱ beenȱ thatȱ itȱ failsȱ toȱ produceȱ consistent,ȱ testableȱ implications.43ȱ Indeed,ȱ aȱ reviewȱ ofȱ theȱ leadingȱ modernȱ indusȬ trialȱ organizationȱ textbooksȱ andȱ literatureȱ surveysȱ revealsȱ gameȬtheoreticȱ modelsȱ identifyingȱ conditionsȱ underȱ whichȱ alȬ mostȱ everyȱ contractualȱ arrangement,ȱ productȱ innovation,ȱ orȱ businessȱactivityȱcouldȱresultȱinȱconsumerȱharm.44ȱȱ TheȱcriticalȱpointȱisȱthatȱinnovationȱisȱcloselyȱrelatedȱtoȱantiȬ trustȱ error.ȱ Becauseȱ innovationȱ involvesȱ newȱ productsȱ andȱ businessȱ practices,ȱ courts’ȱ andȱ economists’ȱ initialȱ misunderȬ standingȱofȱtheseȱpracticesȱwillȱincreaseȱtheȱlikelihoodȱthatȱtheyȱ viewȱtheȱinnovationȱasȱanticompetitiveȱandȱsubjectȱtoȱantitrustȱ scrutiny.ȱThatȱmodernȱantitrustȱanalysisȱreliesȱevenȱmoreȱheavȬ ilyȱonȱeconomicsȱexacerbatesȱthisȱproblem.ȱThisȱbiasȱisȱlikelyȱtoȱ doȱ evenȱ moreȱ damageȱ whenȱ economistsȱ haveȱ lessȱ systematicȱ theoreticalȱandȱempiricalȱknowledgeȱaboutȱtheȱrelationshipȱbeȬ tweenȱ competitionȱ andȱ innovationȱ onȱ policyȬrelevantȱ marginsȱ thanȱtheyȱdoȱaboutȱotherȱtraditionalȱformsȱofȱcompetition.ȱȱ Theȱstakesȱareȱalsoȱhigherȱinȱcasesȱinvolvingȱinnovationȱthanȱ inȱ aȱ regularȱ antitrustȱ case.ȱ Althoughȱ theȱ empiricalȱ literatureȱ doesȱ notȱ containȱ reliableȱ informationȱ onȱ theȱ relativeȱ magniȬ tudesȱ ofȱ Typeȱ Iȱ andȱ Typeȱ IIȱ errorȱ costs,ȱ theȱ wellȬestablishedȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 42.ȱPosner,ȱsupraȱnoteȱ10,ȱatȱ7.ȱ 43.ȱSeeȱgenerallyȱBayeȱ&ȱWright,ȱsupraȱnoteȱ27;ȱEvansȱ&ȱPadilla,ȱsupraȱnoteȱ10.ȱInȱ additionȱ toȱ aȱ generalȱ insensitivityȱ toȱ theȱ factsȱ andȱ marketȱ conditionsȱ ofȱ theȱ parȬ ticularȱcasesȱinȱwhichȱtheseȱtheoremsȱareȱtoȱbeȱapplied,ȱasȱnotedȱabove,ȱtheȱliteraȬ tureȱ isȱ particularlyȱ insensitiveȱ toȱ theȱ institutionalȱ andȱ politicalȱ economyȱ limitationsȱofȱenforcersȱandȱcourts.ȱ 44.ȱSee,ȱe.g.,ȱJEANȱ TIROLE,ȱ THEȱ THEORYȱOFȱ INDUSTRIALȱ ORGANIZATIONȱ(1988);ȱseeȱ alsoȱTimothyȱJ.ȱMuris,ȱCommentsȱonȱAntitrustȱLaw,ȱEconomics,ȱandȱBundledȱDisȬ countsȱ 7ȱ (Julyȱ 15,ȱ 2005),ȱ availableȱ atȱ http://govinfo.library.unt.edu/amc/ȱ commission_hearings/pdf/Muris.pdfȱ (“Inȱ theȱ sameȱ wayȱ thatȱ aȱ visitorȱ fromȱ Marsȱ whoȱreadsȱonlyȱtheȱmathematicalȱIOȱliteratureȱcouldȱmistakenlyȱconcludeȱthatȱtheȱ U.S.ȱeconomyȱisȱrifeȱwithȱmonopolyȱpower,ȱitȱwouldȱbeȱaȱmistakeȱtoȱinferȱthatȱtheȱ growingȱvolumeȱofȱtheoreticalȱpapersȱexaminingȱbundlingȱorȱbundledȱrebatesȱasȱ anȱ exclusionaryȱ deviceȱ impliesȱ thatȱ thereȱ isȱ anyȱ growingȱ orȱ significantȱ dangerȱ fromȱtheȱanticompetitiveȱuseȱofȱbundling.”).ȱ
16ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
empiricalȱ linkȱ betweenȱ innovationȱ andȱ economicȱ growthȱ tellsȱ usȱthatȱtheȱstakesȱofȱerrorȱareȱmuchȱhigher.45ȱAdditionally,ȱnewȱ goodsȱ areȱ generallyȱ quiteȱ valuable,ȱ andȱ theȱ costȱ ofȱ deterringȱ theȱintroductionȱofȱnewȱgoodsȱandȱexpendituresȱonȱinnovation,ȱ bothȱ ofȱ whichȱ areȱ potentiallyȱ costlyȱ andȱ extremelyȱ risky,ȱ isȱ high.46ȱ Forȱ theȱ sameȱ reason,ȱ regulatoryȱ interventionsȱ ofȱ allȱ sorts,ȱ especiallyȱ antitrustȱ ones,ȱ againstȱ aȱ productȱ innovationȱ areȱparticularlyȱrisky.ȱTypeȱIȱerrorsȱareȱlikelyȱtoȱbeȱsignificantlyȱ moreȱ costlyȱ thanȱ Typeȱ IIȱ errorsȱ becauseȱ marketȱ forcesȱ offerȱ atȱ leastȱaȱpartialȱcorrectiveȱinȱtheȱcaseȱofȱtheȱlatter.ȱInȱtheȱinnovaȬ tionȱ context,ȱ successfulȱ antitrustȱ challengesȱ ofȱ businessȱ orȱ productȱ innovationsȱ willȱ likelyȱ dampenȱ innovationȱ acrossȱ theȱ economy,ȱwhereasȱTypeȱIIȱerrorsȱareȱatȱleastȱmitigatedȱinȱpartȱ byȱentryȱandȱotherȱcompetition.ȱȱ Althoughȱ someȱ innovations—particularlyȱ technologicalȱ adȬ vances—areȱevident,ȱothersȱmayȱbeȱsomewhatȱmoreȱdifficultȱtoȱ identifyȱ butȱ nonethelessȱ generateȱ enormousȱ welfareȱ gainsȱ forȱ consumers.47ȱItȱisȱbecauseȱofȱtheseȱdynamicȱandȱoftenȱunanticiȬ patedȱ consequencesȱ ofȱ technologicalȱ innovationȱ thatȱ bothȱ theȱ likelihoodȱ andȱ socialȱ costȱ ofȱ erroneousȱ interventionsȱ againstȱ innovationȱ areȱ increased.ȱ Lessȱ obviously,ȱ butȱ ofȱ atȱ leastȱ equalȱ importance,ȱbusinessȱinnovations—innovationsȱinȱorganization,ȱ production,ȱ marketing,ȱ orȱ distribution—canȱ haveȱ similar,ȱ farȬ reachingȱconsequences.48ȱȱ Unfortunately,ȱ aȱ significantȱ portionȱ ofȱ importantȱ antitrustȱ casesȱ canȱ beȱ characterizedȱ asȱ interventionsȱ undertakenȱ underȱ uncertainty,ȱinȱtheȱfaceȱofȱaȱnovelȱbusinessȱpracticeȱorȱproduct,ȱ relyingȱ onȱ fundamentallyȱ flawedȱ orȱ misappliedȱ economicȱ analysis,ȱ laterȱ demonstratedȱ toȱ haveȱ beenȱ mistaken.49ȱ Inȱ someȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 45.ȱSee,ȱ e.g.,ȱ Charlesȱ I.ȱ Jonesȱ &ȱ Johnȱ C.ȱ Williams,ȱ Measuringȱ theȱ Socialȱ Returnȱ toȱ R&D,ȱ113ȱQ.J.ȱ ECON.ȱ1119ȱ(1998)ȱ(estimatingȱthatȱtheȱsocialȱreturnȱtoȱresearchȱandȱ developmentȱ investmentȱ farȱ exceedsȱ theȱ privateȱ return,ȱ meaningȱ existingȱ incenȬ tivesȱforȱinnovationȱareȱalreadyȱlowerȱthanȱoptimal).ȱ 46.ȱSeeȱ Jerryȱ A.ȱ Hausman,ȱ Valuationȱ ofȱ Newȱ Goodsȱ underȱ Perfectȱ andȱ Imperfectȱ Competition,ȱ inȱ THEȱ ECONOMICSȱ OFȱ NEWȱ GOODSȱ 209ȱ (Timothyȱ F.ȱ Bresnahanȱ &ȱ RobertȱJ.ȱGordonȱeds.,ȱ1997);ȱseeȱalsoȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ15ȱ(statingȱthatȱ theȱcostȱofȱdeterringȱbeneficialȱconductȱisȱhigh).ȱ 47.ȱSee,ȱ e.g.,ȱ Hausman,ȱ supraȱ noteȱ 46ȱ (discussingȱ theȱ consumerȱ welfareȱ gainsȱ fromȱnewȱproductȱintroductionsȱandȱproductȱlineȱextensions).ȱ 48.ȱSeeȱ OLIVERȱ E.ȱ WILLIAMSON,ȱ MARKETSȱ ANDȱ HIERARCHIES:ȱ ANALYSISȱ ANDȱ ANTITRUSTȱIMPLICATIONSȱ(1975).ȱ 49.ȱSeeȱsupraȱnotesȱ22–25ȱandȱaccompanyingȱtext.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
17ȱ
casesȱ theȱ courtsȱ correctȱ theȱ errorȱ ofȱ theȱ initialȱ enforcementȱ orȱ litigationȱdecision;ȱinȱmostȱcasesȱtheyȱdoȱnot.ȱ WhenȱviewedȱthroughȱtheȱerrorȬcostȱlens,ȱtheȱcombinationȱofȱ (1)ȱ theȱ antimarketȱ biasȱ inȱ favorȱ ofȱ monopolyȱ explanationsȱ forȱ innovativeȱ conductȱ thatȱ courtsȱ andȱ economistsȱ doȱ notȱ underȬ stand,ȱ andȱ (2)ȱ theȱ increasedȱ stakesȱ ofȱ antitrustȱ interventionȱ againstȱinnovativeȱbusinessȱpracticesȱisȱproblematicȱfromȱaȱconȬ sumerȱwelfareȱperspective.ȱȱ AȱproperȱapplicationȱofȱerrorȬcostȱprinciplesȱwouldȱdeterȱinȬ terventionȱinȱcasesȱmeetingȱtheȱtwoȱcriteriaȱuntilȱempiricalȱeviȬ denceȱcouldȱbeȱamassedȱandȱassessed.ȱButȱitȱisȱpreciselyȱwhenȱ theseȱ factorsȱ areȱ met,ȱ unfortunately,ȱ thatȱ interventionȱ isȱ moreȱ likely.ȱ Onȱ theȱ oneȱ hand,ȱ theȱ increasedȱ likelihoodȱ ofȱ intervenȬ tionȱmightȱbeȱbecauseȱinȱtheȱabsenceȱofȱinformationȱdisprovingȱ aȱpresumptionȱofȱanticompetitiveȱeffect,ȱthereȱisȱanȱeasierȱcaseȱ toȱ beȱ madeȱ againstȱ theȱ conduct—despiteȱ putativeȱ burdenȬ shiftingȱ rulesȱ thatȱ wouldȱ placeȱ theȱ onusȱ onȱ theȱ complainant.ȱ Onȱtheȱotherȱhand,ȱsuccessfulȱinnovationsȱareȱalsoȱmoreȱlikelyȱ toȱ arouseȱ theȱ ireȱ ofȱ competitorsȱ andȱ customers,ȱ andȱ thusȱ bothȱ theirȱexistenceȱandȱnegativeȱcharacterizationȱareȱmoreȱlikelyȱtoȱ beȱbroughtȱtoȱtheȱattentionȱofȱcourtsȱandȱenforcers—abettedȱinȱ privateȱlitigationȱbyȱtheȱlureȱofȱtrebleȱdamages.ȱ Thisȱhostileȱstanceȱtowardȱnovelȱeconomicȱbehaviorȱandȱtheȱ institutionalizationȱofȱlawsȱandȱprocessesȱthatȱtendȱtoȱcondemnȱ innovativeȱ behaviorȱ absentȱ clearȱ procompetitiveȱ justificationȱ areȱparticularlyȱproblematicȱbecauseȱthereȱmayȱbeȱonlyȱaȱweakȱ connectionȱ betweenȱ corporateȱ actors’ȱ actionsȱ andȱ theirȱ conseȬ quences.ȱJudgeȱEasterbrookȱhighlightsȱthisȱproblem:ȱ Wisdomȱlagsȱfarȱbehindȱtheȱmarket.ȱItȱisȱusefulȱforȱmanyȱpurȬ posesȱtoȱthinkȱofȱmarketȱbehaviorȱasȱrandom.ȱFirmsȱtryȱdozȬ ensȱ ofȱ practices.ȱ Mostȱ ofȱ themȱ areȱ flops,ȱ andȱ theȱ firmsȱ mustȱ tryȱ somethingȱ elseȱ orȱ disappear.ȱ Otherȱ practicesȱ offerȱ someȬ thingȱextraȱtoȱconsumers—theyȱreduceȱcostsȱorȱimproveȱqualȬ ity—andȱsoȱtheyȱsurvive.ȱInȱaȱcompetitiveȱstruggleȱtheȱfirmsȱ thatȱuseȱtheȱbestȱpracticesȱsurvive.ȱMistakesȱareȱburied.ȱ Whyȱ doȱ particularȱ practicesȱ work?ȱ Theȱ firmsȱ thatȱ selectedȱ theȱ practicesȱ mayȱ orȱ mayȱ notȱ knowȱ whatȱ isȱ specialȱ aboutȱ them.ȱ Theyȱ canȱ describeȱ whatȱ theyȱ do,ȱ butȱ theȱ whyȱ isȱ moreȱ difficult.ȱ Onlyȱ someoneȱ withȱ aȱ veryȱ detailedȱ knowledgeȱ ofȱ theȱmarketȱprocess,ȱasȱwellȱasȱtheȱtimeȱandȱdataȱneededȱforȱ
18ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
evaluation,ȱ wouldȱ beȱ ableȱ toȱ answerȱ thatȱ question.ȱ SomeȬ timesȱnoȱoneȱcanȱanswerȱit.50ȱ
Itȱmakesȱlittleȱsenseȱtoȱdemandȱthatȱeconomicȱactorsȱidentify,ȱ understand,ȱ andȱ assessȱ theȱ procompetitive,ȱ profitȬ maximizingȱ basisȱ forȱ theirȱ behavior,ȱ asȱ theyȱ areȱ hamperedȱ byȱ “imperfectȱ foresightȱ andȱ humanȱ inabilityȱ toȱ solveȱ comȬ plexȱ problemsȱ containingȱ aȱ hostȱ ofȱ variablesȱ evenȱ whenȱ anȱ optimumȱisȱdefinable.”51ȱȱ Aȱ fortiori,ȱ cognitiveȱ limitationsȱ applyȱ evenȱ moreȱ stronglyȱ toȱ regulatorsȱandȱcourts.ȱSuchȱlimitationsȱshouldȱcounselȱenforcersȱ againstȱassumingȱthatȱunexplained,ȱnovelȱeconomicȱbehaviorȱ(orȱ worse,ȱ theȱ anticompetitiveȱ intentionsȱ ofȱ economicȱ actors)ȱ hasȱ anticompetitiveȱconsequences.ȱYetȱsuchȱspeculativeȱassumptionsȱ areȱasȱfrustratinglyȱcommonȱasȱtheyȱareȱproblematic.ȱ Consider,ȱ forȱ example,ȱ theȱ Microsoftȱ case.52ȱ Microsoftȱ offersȱ theȱ standardȱ jurisprudentialȱ approachȱ forȱ highȬtechȱ monopolizationȱ cases.ȱTheȱ fundamentalȱ errorȱ inȱ Microsoftȱ wasȱ notȱ necessarilyȱ inȱ theȱ court’sȱ conclusion,ȱ butȱ ratherȱ inȱ itsȱ apȬ proachȱ toȱ assessingȱ theȱ complexȱ andȱ novelȱ economicsȱ ofȱ theȱ case.ȱAtȱfirst,ȱtheȱcourt’sȱapproachȱtoȱfairlyȱstandardȱcategoriesȱ ofȱ specificȱ exclusionaryȱ conductȱ undertakenȱ byȱ Microsoftȱ (forȱ example,ȱ itsȱ interactionsȱ withȱ Netscape),ȱ seemsȱ factȬspecific,ȱ difficultȱtoȱgeneralize,ȱandȱrelativelyȱuncontroversialȱforȱaȱcaseȱ involvingȱaȱfindingȱofȱmonopolyȱpower.53ȱ Theȱcourt’sȱapproachȱtoȱtheȱmonopolyȱpowerȱdeterminationȱ itself,ȱ however,ȱ isȱ moreȱ troubling.54ȱ Theȱ courtȱ basesȱ thisȱ determinationȱ onȱ tenuousȱ economicȱ assumptions,ȱ adȱ hocȱ resolutionȱofȱcomplexȱeconomicȱdisputes,ȱandȱaȱdearthȱofȱdirectȱ economicȱ evidence.55ȱ Predictably,ȱ theȱ courtȱ basedȱ itsȱ decisionȱ onȱtheȱpresumedȱimplicationsȱofȱaȱtheoreticalȱanalysisȱofȱaȱsetȱ ofȱmarketȱconditionsȱandȱbusinessȱconductȱthat,ȱtheȱcourtȱsays,ȱ grantsȱmarketȱpowerȱtoȱMicrosoft.56ȱThereȱwasȱanȱunfortunateȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 50.ȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ5.ȱ 51.ȱArmenȱ A.ȱ Alchian,ȱ Uncertainty,ȱ Evolution,ȱ andȱ Economicȱ Theory,ȱ 58ȱ J.ȱ POL.ȱ ECON.ȱ211,ȱ212ȱ(1950).ȱ 52.ȱUnitedȱStatesȱv.ȱMicrosoftȱCorp.,ȱ253ȱF.3dȱ34ȱ(D.C.ȱCir.ȱ2001)ȱ(enȱbanc)ȱ(perȱ curiam).ȱ 53.ȱId.ȱatȱ47–50.ȱ 54.ȱId.ȱatȱ50–58.ȱ 55.ȱId.ȱatȱ56–58.ȱ 56.ȱId.ȱatȱ50–58.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
19ȱ
lackȱ ofȱ evidenceȱ ofȱ harmȱ toȱ theȱ competitiveȱ processȱ fromȱ specificȱ actsȱ thatȱ resultedȱ inȱ monopolyȱ power,57ȱ whichȱ isȱ aȱ requirementȱofȱmodernȱShermanȱActȱjurisprudence.58ȱȱ Theȱcourt’sȱapproachȱshunnedȱMicrosoft’sȱofferȱofȱdirectȱeviȬ denceȱ onȱ marketȱ power,ȱ insteadȱ substitutingȱ aȱ setȱ ofȱ concluȬ sionsȱbuiltȱonȱcontroversialȱeconomicȱtheoryȱwithoutȱempiricalȱ support.59ȱ Suchȱ anȱ approachȱ isȱ particularlyȱ problematicȱ notȱ onlyȱ inȱ theȱ faceȱ ofȱ theȱ court’sȱ ignoranceȱ aboutȱ theȱ economicsȱ andȱ technologyȱ involved,ȱ butȱ alsoȱ theȱ economicȱ ramificationsȱ ofȱ overȬdeterringȱ investmentȱ inȱ innovativeȱ technologiesȱ andȱ businessȱpractices.ȱ Theȱ remainderȱ ofȱ thisȱ Articleȱ willȱ discussȱ theȱ applicationȱ ofȱ theȱprinciplesȱofȱtheȱerrorȬcostȱframeworkȱtoȱwhatȱmayȱbeȱtheȱ nextȱ greatȱ monopolizationȱ case:ȱ Google.ȱ Evenȱ ifȱ theȱ governȬ mentȱ neverȱ bringsȱ anȱ enforcementȱ actionȱ againstȱ Google,ȱ theȱ hypotheticalȱcaseȱpresentsȱaȱfascinatingȱsetȱofȱfacts.60ȱInȱparticuȬ lar,ȱalthoughȱmostȱcommentatorsȱseemȱtoȱviewȱGoogleȱlargelyȱ asȱaȱproductȱinnovator,61ȱitȱisȱinȱfactȱ bothȱaȱproductȱinnovatorȱ andȱ aȱ businessȱ innovator.ȱ Moreover,ȱ theȱ twoȱ categoriesȱ ofȱ inȬ novationȱareȱinextricablyȱintertwined,ȱperhapsȱmultiplyingȱtheȱ likelihoodȱofȱerroneousȱantitrustȱenforcementȱdecisions.ȱInȱtheȱ absenceȱ ofȱ obviousȱ exclusionaryȱ businessȱ practicesȱ orȱ innovaȬ tions,ȱGoogleȱmayȱwellȱescapeȱsignificantȱgovernmentȱantitrustȱ attentionȱ orȱ liability.ȱ Googleȱ presentsȱ aȱ uniqueȱ meldingȱ ofȱ businessȱ andȱ productȱ innovation,ȱ andȱ thusȱ isȱ anȱ interestingȱ andȱimportantȱcaseȱforȱstudy.ȱ III.
THEȱUNCERTAINȱECONOMICSȱOFȱGOOGLEȇSȱBUSINESSȱANDȱ MARKETȱ
Antitrustȱ agenciesȱ andȱ commentatorsȱ haveȱ alreadyȱ takenȱ sigȬ nificantȱinterestȱinȱGoogle.ȱTheȱUnitedȱStatesȱDepartmentȱofȱJusȬ ticeȱ (DOJ)ȱ investigatedȱ Google’sȱ proposedȱ jointȱ ventureȱ withȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 57.ȱSeeȱid.ȱatȱ56–58.ȱ 58.ȱVerizonȱCommc’nsȱInc.ȱv.ȱLawȱOfficesȱofȱCurtisȱV.ȱTrinko,ȱLLP,ȱ540ȱU.S.ȱ398,ȱ 407ȱ(2004).ȱ 59.ȱSeeȱMicrosoft,ȱ253ȱF.3dȱatȱ50–58.ȱ 60.ȱGoogleȱalreadyȱhasȱdefendedȱseveralȱprivateȱantitrustȱsuits—surelyȱthisȱwillȱ notȱbeȱtheȱlast.ȱSee,ȱe.g.,ȱKinderstart.com,ȱLLCȱv.ȱGoogle,ȱInc.,ȱNo.ȱC06Ȭ2057JF(RS),ȱ 2007ȱWLȱ831806ȱ(N.D.ȱCal.ȱMar.ȱ16,ȱ2007).ȱ 61.ȱSee,ȱ e.g.,ȱ Benȱ Elgin,ȱ Managingȱ Google’sȱ Ideaȱ Factory,ȱ BUSINESSWEEK,ȱ Oct.ȱ 3,ȱ 2005,ȱhttp://www.businessweek.com/magazine/contents/05Ȭ40/63953093.htm.ȱ
20ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
Yahoo!ȱ ,62ȱandȱtheȱFederalȱTradeȱCommissionȱ(FTC)ȱcontinuesȱtoȱ investigateȱ Googleȱ andȱ Apple’sȱ interlockingȱ boards.63ȱ Googleȱ facesȱ ongoingȱ scrutinyȱ (andȱ aȱ formalȱ DOJȱ investigation)ȱ forȱ itsȱ settlementȱofȱaȱsuitȱarisingȱfromȱitsȱGoogleȱBooksȱproject.64ȱMoreȬ over,ȱGoogleȱhasȱfacedȱaȱfullȱantitrustȱreviewȱinȱmultipleȱjurisdicȬ tionsȱofȱitsȱmergerȱwithȱDoubleClick,65ȱfullȱreviewȱbyȱtheȱFTCȱofȱ itsȱmergerȱwithȱAdMob,66ȱandȱhasȱalsoȱbeenȱtheȱtargetȱofȱprivateȱ litigation,ȱincludingȱantitrustȱclaims.67ȱȱ GoogleȱisȱlikelyȱtoȱfaceȱantitrustȱenforcementȱforȱseveralȱreaȬ sons.ȱFirst,ȱGoogleȱisȱaȱlarge,ȱdominant,ȱinteresting,ȱinnovative,ȱ andȱ highȬtechȱ company.ȱ Ifȱ Microsoftȱ isȱ anȱ analogue,ȱ thanȱ eachȱ ofȱtheseȱfactorsȱlikelyȱwillȱprovokeȱuntowardȱantitrustȱscrutiny.ȱ Additionally,ȱ Googleȱ facesȱ powerfulȱ competitorsȱ thatȱ feelȱ threatenedȱbyȱGoogle’sȱsuccess68ȱandȱtheȱubiquitousȱbrandȱthatȱ Googleȱ enjoys.ȱ Itȱ conductsȱ businessȱ inȱ theȱ highȬtechȱ world,ȱ fromȱ whenceȱ almostȱ allȱ modernȱ monopolizationȱ casesȱ come.ȱ Moreover,ȱ itsȱ businessȱ isȱ information,ȱ theȱ economicsȱ ofȱ whichȱ isȱpoorlyȱunderstoodȱandȱofȱpeculiarȱconcernȱinȱtoday’sȱworldȱ ofȱultraȬlowȬtransactionȱcostȱcommunication.69ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 62.ȱSee,ȱ e.g.,ȱ Stephenȱ Shankland,ȱ YahooȬGoogleȱ Dealȱ Doomed?ȱ No,ȱ Theyȱ Insist,ȱ CNET.COMȱ NEWS,ȱ Oct.ȱ 21,ȱ 2008,ȱ http://news.cnet.com/8301Ȭ1023_3Ȭ10071478Ȭ 93.html.ȱ 63.ȱPaulȱ Suarez,ȱ FCCȱ onȱ Apple,ȱ DOJȱ onȱ Googleȱ –ȱ theȱ Fedsȱ areȱ Tacklingȱ Tech,ȱ PCȱ WORLD,ȱ Sept.ȱ 19,ȱ 2009,ȱ http://www.pcworld.com/article/172295/fcc_on_apple_ȱ doj_on_google_the_feds_are_tackling_tech.html.ȱ 64.ȱMiguelȱHelft,ȱJusticeȱDept.ȱOpensȱAntitrustȱInquiryȱIntoȱGoogleȱBooksȱDeal,ȱN.Y.ȱ TIMES,ȱ Apr.ȱ 29,ȱ 2010,ȱ atȱ B5,ȱ availableȱ atȱ http://www.nytimes.com/2009/04/29/ȱ technology/internet/29google.html.ȱ Noteȱ theȱ DOJ’sȱ briefȱ inȱ theȱ case,ȱ theȱ court’sȱ grantȱofȱaȱdelayȱandȱongoingȱrenegotiations.ȱ 65.ȱSteveȱLohr,ȱGoogleȱDealȱSaidȱtoȱBringȱU.S.ȱScrutiny,ȱN.Y.ȱ TIMES,ȱMayȱ29,ȱ2007,ȱ atȱ C1,ȱ availableȱ atȱ http://www.nytimes.com/2007/05/29/technology/29ȱ antitrust.html;ȱ Pressȱ Release,ȱ Fed.ȱ Tradeȱ Comm’n,ȱ Fed.ȱ Tradeȱ Comm’nȱ Closesȱ Google/DoubleClickȱInvestigationȱ(Dec.ȱ20,ȱ2007),ȱavailableȱatȱhttp://www.ftc.gov/ȱ opa/2007/12/googledc.shtm.ȱ 66.ȱPressȱ Release,ȱ Fed.ȱ Tradeȱ Comm’n,ȱ FTCȱ Closesȱ itsȱ Investigationȱ ofȱ Googleȱ AdMobȱ Dealȱ (Mayȱ 21,ȱ 2010),ȱ availableȱ atȱ ȱ http://ftc.gov/opa/2010/05/ȱ ggladmob.shtm.ȱ 67.ȱSeeȱ Kinderstart.com,ȱ LLCȱ v.ȱ Google,ȱ Inc.,ȱ No.ȱ C06Ȭ2057JF(RS),ȱ 2007ȱ WLȱ 831806ȱ(N.D.ȱCal.ȱMar.ȱ16,ȱ2007).ȱ 68.ȱNicholasȱ Thompsonȱ &ȱ Fredȱ Vogelstein,ȱ Theȱ Plotȱ toȱ Killȱ Google,ȱ WIRED,ȱ Jan.ȱ 19,ȱ2009,ȱhttp://www.wired.com/techbiz/it/magazine/17Ȭ02/ff_killgoogle.ȱ 69.ȱOnȱ theȱ economicsȱ ofȱ information,ȱ economistsȱ haveȱ progressedȱ littleȱ sinceȱ ProfessorȱGeorgeȱStigler’sȱimportantȱaccountȱofȱtheȱeconomicsȱofȱinformationȱandȱ theȱimportanceȱofȱsearch—ofȱobviousȱsalienceȱhere.ȱSeeȱGeorgeȱJ.ȱStigler,ȱTheȱEcoȬ nomicsȱofȱInformation,ȱ69ȱJ.ȱPOL.ȱECON.ȱ213ȱ(1961).ȱAlthoughȱmuchȱofȱtheȱeconomicsȱ ofȱnetworksȱandȱinformationȱtechnologyȱhasȱmadeȱimportantȱadvances,ȱProfessorȱ
ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
21ȱ
Definingȱ relevantȱ antitrustȱ marketsȱ forȱ Google,ȱ therefore,ȱ posesȱ anȱ interestingȱ problem:ȱ Googleȱ performsȱ aȱ fairlyȱ tradiȬ tionalȱfunction,ȱbutȱinȱaȱnovelȱmedium,ȱandȱtheȱeconomicsȱofȱitsȱ businessȱ areȱ poorlyȱ understood.ȱ Asȱ aȱ result,ȱ whetherȱ Google’sȱ businesses—particularlyȱitsȱadvertisingȱbusinesses—existȱinȱtheȱ sameȱ economicallyȱ relevantȱ marketȱ asȱ moreȱ traditionalȱ formsȱ ofȱ advertisingȱ remainsȱ unclear.ȱ Manyȱ ofȱ Google’sȱ innovationsȱ haveȱ servedȱ toȱ differentiateȱ itsȱ productȱ fromȱ moreȱ traditionalȱ versions—offering,ȱforȱexample,ȱmuchȱmoreȱtargetedȱadvertisȬ ingȱthanȱisȱpossibleȱinȱmanyȱotherȱsettings.ȱThus,ȱnotȱonlyȱtheȱ mediumȱ butȱ alsoȱ Google’sȱ technologicalȱ andȱ businessȱ innovaȬ tionsȱdifferentiateȱitȱfromȱmoreȱtraditionalȱcompetitors.ȱȱ Googleȱ appearsȱ toȱ haveȱ aȱ largeȱ shareȱ ofȱ someȱ marketsȱ inȱ whichȱitȱparticipates.ȱWhetherȱtheseȱlargeȱsharesȱareȱeconomiȬ callyȱrelevantȱisȱaȱdifficult,ȱbutȱessential,ȱquestion.ȱTheȱanswerȱ turnsȱnotȱonlyȱonȱwhetherȱtheȱmarketȱisȱproperlyȱdefined,ȱbutȱ alsoȱ onȱ whetherȱ weȱ understandȱ theȱ implicationsȱ ofȱ highȱ conȬ centrationȱ andȱ theȱ natureȱ ofȱ competitionȱ andȱ contestabilityȱ inȱ theseȱ markets.ȱ Moreover,ȱ Google’sȱ particularȱ contractualȱ andȱ pricingȱ practicesȱ areȱ undertakenȱ inȱ anȱ uncertainȱ context,ȱ genȬ erallyȱ unanticipatedȱ byȱ currentȱ modelsȱ andȱ oftenȱ misunderȬ stoodȱbyȱbusinessmenȱandȱpolicymakersȱalike.ȱȱ Finally,ȱGoogleȱhasȱextremelyȱactiveȱcompetitorȬcomplainantsȱ andȱconfrontsȱaȱrelativelyȱhostileȱantitrustȱenforcementȱcommuȬ nityȱthatȱrejectsȱtheȱerrorȬcostȱconcernsȱraisedȱearlierȱinȱthisȱArtiȬ cle.70ȱ Whetherȱ enforcers’ȱ hostilityȱ isȱ rootedȱ inȱ ideologicalȱ predispositionȱ toȱ alternativeȱ models,ȱ orȱ aȱ stringentȱ preferenceȱ forȱmoreȱinterventionistȱantitrustȱpolicy,ȱitȱnonethelessȱthreatensȱ toȱcastȱaȱbroadȱshadowȱoverȱGoogle’sȱfutureȱbusinessȱdecisions.ȱ Theȱ difficultyȱ isȱ inȱ parsingȱ outȱ theȱ differenceȱ betweenȱ antiȬ competitiveȱconductȱundertakenȱbyȱaȱmonopolistȱandȱ“growthȱ orȱdevelopmentȱasȱaȱconsequenceȱofȱaȱsuperiorȱproduct,ȱbusinessȱ acumen,ȱorȱhistoricȱaccident.”71ȱThisȱdifferentiationȱisȱparticularlyȱ difficultȱtoȱmakeȱinȱtheȱenvironmentȱofȱuncertaintyȱandȱproblemȬ aticȱincentivesȱthatȱGoogleȱfaces.ȱTheȱremainderȱofȱthisȱPartȱoffersȱ anȱoverviewȱofȱGoogle’sȱbusinessȱandȱtheȱmarketsȱinȱwhichȱitȱopȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ Stigler’sȱ (andȱ Freidrichȱ vonȱ Hayek’s)ȱ fundamentalȱ pointȱ thatȱ informationȱ isȱ aȱ goodȱlikeȱanyȱother,ȱsubjectȱtoȱtheȱeconomicsȱofȱscarcity,ȱstillȱundergirdsȱourȱunȬ derstandingȱtoday.ȱ 70.ȱSeeȱsupraȱPartȱII;ȱseeȱalsoȱVarney,ȱsupraȱnoteȱ3.ȱ 71.ȱUnitedȱStatesȱv.ȱGrinnelȱCorp.,ȱ384ȱU.S.ȱ563,ȱ571ȱ(1966).ȱ
22ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
erates.ȱ Theȱdiscussionȱ isȱ intendedȱtoȱ highlightȱsomeȱ ofȱtheȱ comȬ plexitiesȱofȱGoogle’sȱbusinessȱstructureȱandȱsomeȱofȱtheȱbasicȱdyȬ namicsȱofȱGoogle’sȱmarkets,ȱincludingȱtheȱroleȱofȱnetworkȱeffectsȱ inȱitsȱprimaryȱsearchȱadvertisingȱbusiness.ȱȱ Althoughȱ Googleȱ engagesȱ inȱ aȱ wideȱ arrayȱ ofȱ businessesȱ andȱ sellsȱaȱwideȱarrayȱofȱproducts,ȱitsȱantitrustȱchallengesȱtoȱdateȱ(withȱ theȱexceptionȱofȱitsȱambitiousȱGoogleȱBookȱproject)ȱhaveȱcenteredȱ aroundȱ itsȱ searchȱ andȱ displayȱ advertisingȱ businesses.ȱ Indeed,ȱ Google’sȱcoreȱisȱinȱtheȱadvertisingȱbusiness.72ȱOurȱdiscussionȱcenȬ tersȱmostȱheavilyȱonȱGoogle’sȱsearchȱadvertisingȱbusiness,ȱwhereȱ Googleȱ hasȱ theȱ mostȱ significantȱ presence.ȱ Toȱ theȱ extentȱ thatȱ weȱ discussȱ Google’sȱ displayȱ advertisingȱ business,ȱ itȱ isȱ largelyȱ ancilȬ laryȱtoȱtheȱdiscussionȱofȱitsȱsearchȱadvertisingȱbusiness.ȱ A.
SomeȱBasicsȱofȱOnlineȱSearchȱ
WhenȱaȱuserȱentersȱaȱqueryȱonȱGoogle’sȱsearchȱwebsite,ȱtwoȱ typesȱ ofȱ resultsȱ areȱ generated:ȱ organicȱ orȱ naturalȱ resultsȱ andȱ sponsoredȱ orȱ paidȱ links.73ȱ Google’sȱ organicȱ searchȱ resultsȱ areȱ generatedȱatȱnoȱdirectȱcostȱtoȱtheȱwebsitesȱtoȱwhichȱtheyȱlink.74ȱ Google’sȱsearchȱengineȱreviewsȱwebpageȱcontentȱandȱproducesȱ aȱlistȱofȱtheȱpagesȱmostȱrelevantȱtoȱeachȱuser’sȱparticularȱsearchȱ query.75ȱ Theȱ searchȱ engineȱ alsoȱ assessesȱ howȱ relevantȱ aȱ webȬ site’sȱcontentȱisȱtoȱaȱuser’sȱqueryȱbyȱlookingȱatȱhowȱmanyȱotherȱ relevantȱ websitesȱ linkȱ toȱ it.76ȱ Theȱ finalȱ organicȱ resultsȱ areȱ loȬ catedȱonȱtheȱleftȬhandȱsideȱofȱGoogle’sȱsearchȱresultsȱpage.77ȱ Google’sȱ sponsoredȱ linksȱ areȱ producedȱ forȱ businessesȱ interȬ estedȱ inȱ advertisingȱandȱwillingȱ toȱ payȱGoogleȱwhenȱ usersȱclickȱ onȱ theirȱ ads.78ȱ Advertisementsȱ areȱ generatedȱ byȱ theȱ keywordsȱ aȱ userȱentersȱintoȱGoogle’sȱsearchȱengine.79ȱSponsoredȱlinksȱareȱloȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 72.ȱSeeȱ KENȱ AULETTA,ȱ GOOGLED:ȱ THEȱ ENDȱ OFȱ THEȱ WORLDȱ ASȱ WEȱ KNOWȱ ITȱ 16ȱ (2009)ȱ (quotingȱ Googleȱ CEOȱ Ericȱ Schmidtȱ asȱ saying,ȱ “Weȱ areȱ inȱ theȱ advertisingȱ business”).ȱ 73.ȱResultsȱ Page,ȱ GOOGLEGUIDE,ȱ http://www.googleguide.com/results_page.htmlȱ (lastȱvisitedȱNov.ȱ6,ȱ2010).ȱ 74.ȱAds,ȱ GOOGLEGUIDE,ȱ http://googleguide.com/ads.htmlȱ (lastȱ visitedȱ Nov.ȱ 6,ȱ 2010).ȱ 75.ȱId.ȱ 76.ȱResults,ȱsupraȱnoteȱ73.ȱ 77.ȱAds,ȱsupraȱnoteȱ74.ȱ 78.ȱId.ȱ 79.ȱId.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
23ȱ
catedȱ onȱ theȱ topȱ orȱ rightȬhandȱ sideȱ ofȱ theȱ searchȱ resultsȱ page.80ȱ Theȱ majorityȱ ofȱ Google’sȱ revenueȱ comesȱ fromȱ theȱ saleȱ ofȱ sponȬ soredȱlinksȱandȱotherȱsimilarlyȱgeneratedȱsearchȱadvertisements.81ȱ Google’sȱproducts—whetherȱorganicȱorȱpaid—areȱbuiltȱonȱitsȱ innovativeȱ PageRankȱ method,ȱ anȱ algorithmȱ developedȱ byȱ Larryȱ Pageȱ andȱ Sergeyȱ Brin,ȱ whoȱ foundedȱ Google.82ȱ Similarȱ rankingȱsystemsȱareȱnowȱusedȱonȱmostȱeachȱengines.ȱPageRankȱ remainsȱ anȱ essentialȱ partȱ ofȱ Google’sȱ searchȱ business,ȱ butȱ isȱ embeddedȱ withinȱ aȱ complexȱ setȱ ofȱ additionalȱ innovations,ȱ someȱofȱwhichȱfacilitateȱtheȱsaleȱofȱsearchȱresultsȱtoȱadvertisers.ȱ InȱadditionȱtoȱPageRank,ȱGoogle’sȱsearchȱresultsȱareȱbuiltȱonȱaȱ hostȱofȱinnovativeȱtechnologies,ȱincluding:ȱ [1.]ȱ languageȱ modelsȱ (theȱ abilityȱ toȱ handleȱ phrases,ȱ synoȬ nyms,ȱ diacritics,ȱ spellingȱ mistakes,ȱ andȱ soȱ on),ȱ [2.]ȱ queryȱ modelsȱ(itȇsȱnotȱjustȱtheȱlanguage,ȱitȇsȱhowȱpeopleȱuseȱitȱtoȬ day),ȱ[3.]ȱtimeȱmodelsȱ(someȱqueriesȱareȱbestȱansweredȱwithȱ aȱ30Ȭminuteȱoldȱpage,ȱandȱsomeȱareȱbetterȱansweredȱwithȱaȱ pageȱthatȱstoodȱtheȱtestȱofȱtime),ȱandȱ[4.]ȱpersonalizedȱmodȬ elsȱ(notȱallȱpeopleȱwantȱtheȱsameȱthing).83ȱ
TheȱamountȱthatȱGoogleȱchargesȱforȱsponsoredȱlinksȱisȱcalcuȬ latedȱ accordingȱ toȱ aȱ keywordȱ auctionȱ conductedȱ throughȱ Google’sȱ AdWordsȱ platform.84ȱ Theseȱ auctionsȱ areȱ automatedȱ basedȱonȱaȱsetȱofȱparametersȱspecifiedȱbyȱeachȱadvertiser,ȱandȱ theyȱoccurȱinstantaneouslyȱeachȱtimeȱaȱkeywordȱisȱenteredȱintoȱ Google’sȱsearchȱengine.85ȱAnȱadvertiserȱwhoȱplacesȱaȱhigherȱbidȱ forȱ aȱ keywordȱ willȱ receiveȱ betterȱ placementȱ ofȱ itsȱ advertiseȬ mentsȱwhenȱaȱuserȱentersȱthatȱkeywordȱasȱpartȱofȱhisȱsearch.86ȱ Additionally,ȱGoogleȱemploysȱanȱinnovativeȱqualityȱmetricȱthatȱ adjustsȱ theȱ placementȱ andȱ costȱ toȱ theȱ advertiserȱ ofȱ sponsoredȱ linksȱbasedȱonȱtheȱlinks’ȱrelevanceȱtoȱtheȱsearchȱqueryȱandȱtheȱ qualityȱ ofȱ theȱ underlyingȱ webpage.87ȱ Advertisersȱ payȱ Googleȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 80.ȱId.ȱ 81.ȱVIRGINIAȱSCOTT,ȱCOMPANIESȱTHATȱCHANGEDȱTHEȱWORLD:ȱGOOGLEȱ76ȱ(2008).ȱ 82.ȱAMYȱ N.ȱ LANGVILLEȱ &ȱ CARLȱ DEANȱ MEYER,ȱ GOOGLE’Sȱ PAGERANKȱ ANDȱ BEYOND:ȱTHEȱSCIENCEȱOFȱSEARCHȱENGINEȱRANKINGȱ25ȱ(2006).ȱ 83.ȱUdiȱ Manber,ȱ Introductionȱ toȱ Googleȱ Searchȱ Quality,ȱ THEȱ OFFICIALȱ GOOGLEȱ BLOGȱ (Mayȱ 20,ȱ 2008,ȱ 6:20ȱ PM),ȱ http://googleblog.blogspot.com/2008/05/introductionȬtoȬ googleȬsearchȬquality.html.ȱ 84.ȱHOWIEȱJACOBSON,ȱGOOGLEȱADWORDSȱFORȱDUMMIESȱ1–3ȱ(2dȱed.,ȱ2009).ȱ 85.ȱId.ȱ 86.ȱId.ȱ 87.ȱAds,ȱsupraȱnoteȱ74.ȱ
24ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
onlyȱ ifȱ aȱ userȱ actuallyȱ clicksȱ onȱ theȱ displayedȱ advertisement,ȱ andȱ theȱ amountȱ theyȱ payȱ isȱ generallyȱ aȱ functionȱ ofȱ theȱ nextȬ highestȱ bidȱ andȱ theȱ qualityȱ scoreȱ adjustmentȱ (whereȱ higherȱ qualityȱscoresȱmayȱresultȱinȱlowerȱcosts).88ȱ Googleȱ alsoȱ receivesȱ advertisingȱ revenueȱ byȱ sellingȱ advertiseȬ mentsȱthroughȱitsȱAdSenseȱapplication.ȱAdSenseȱallowsȱGoogleȱtoȱ placeȱ advertisementsȱ inȱ designatedȱ spacesȱ onȱ thirdȬpartyȱ webȬ sites.89ȱLikeȱAdWords,ȱGoogle’sȱAdSenseȱapplicationȱdisplaysȱadȬ vertisementsȱ basedȱ onȱ theȱ keywordsȱ aȱ userȱ entersȱ asȱ partȱ ofȱ hisȱ queryȱ (ifȱ thereȱ isȱ one)ȱ alongȱ withȱ aȱ qualityȱ scoreȱ adjustment.90ȱ Again,ȱadvertisersȱpayȱGoogleȱonlyȱwhenȱaȱuserȱclicksȱonȱtheȱdisȬ playedȱ advertisement.91ȱ Googleȱ thenȱ splitsȱ theȱ revenueȱ withȱ theȱ thirdȬpartyȱwebsiteȱthatȱhostedȱtheȱadvertisement.92ȱ B.
Google’sȱMarketȱ
Aȱ preliminaryȱ issueȱ forȱ assessingȱ Google’sȱ businessȱ isȱ deȬ terminingȱ theȱ relevantȱ marketȱ inȱ whichȱ itȱ operates.ȱ Althoughȱ colloquiallyȱitȱisȱunderstoodȱthatȱGoogleȱisȱtheȱdominantȱsearchȱ andȱ searchȱ advertisingȱ providerȱ inȱ anȱ onlineȱ searchȱ marketȱ comprisedȱofȱGoogle,ȱMicrosoft,ȱandȱYahoo!,ȱtheȱantitrustȱreleȬ vanceȱofȱthisȱassessmentȱisȱquestionable.ȱȱ TheȱcompetitiveȱlandscapeȱGoogleȱconfrontsȱisȱcomplex,ȱandȱ theȱcompanyȱplainlyȱfacesȱcompetitiveȱthreatsȱfromȱaȱrangeȱofȱ sources,ȱbothȱactualȱandȱpotential;ȱtheȱnotionȱofȱaȱwellȬcabined,ȱ “onlineȱ searchȱ advertisingȱ market”ȱ isȱ decidedlyȱ messy.ȱ Theȱ antitrustȬspecificȱ questionȱ isȱ whetherȱ thisȱ messinessȱ isȱ signifiȬ cantȱ enoughȱ toȱ castȱ doubt,ȱ absentȱ viableȱ econometricȱ data,ȱ onȱ theȱ antitrustȱ relevanceȱ ofȱ aȱ simplifiedȱ “onlineȱ searchȱ advertisȬ ingȱmarket.”ȱThereȱisȱreasonȱtoȱbeȱskeptical.ȱ Oneȱ setȱ ofȱ facts,ȱ takenȱ fromȱ Accuracast.com,ȱ aȱ websiteȱ reȬ countingȱ Europeanȱ searchȱ engineȱ statisticsȱ fromȱ 2006–07,ȱ sugȬ gestsȱaȱrangeȱofȱproblemsȱwithȱtheȱsimplifiedȱmarketȱstory.ȱTheȱ websiteȱnotes,ȱamongȱotherȱthings,ȱthat:ȱ “Onȱ averageȱ searchersȱ spentȱ 27ȱ minutesȱ onȱ searchȱ enginesȱ eachȱmonthȱandȱviewedȱ93ȱsearchȱpages,ȱaccountingȱforȱ3.4%ȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 88.ȱJACOBSON,ȱsupraȱnoteȱ84.ȱ 89.ȱDAVIDȱO’CONNELL,ȱMAKINGȱSENSEȱOFȱGOOGLEȱADSENSEȱ4–8ȱ(2008).ȱ 90.ȱId.ȱ 91.ȱId.ȱ 92.ȱId.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
25ȱ
ofȱtotalȱtimeȱspentȱonline,”ȱmeaningȱthatȱtheȱvastȱmajorityȱofȱ timeȱ spentȱ onlineȱ isȱ spentȱ onȱ websitesȱ otherȱ thanȱ searchȱ enȬ gines—eachȱofȱwhichȱpresentsȱaȱpossibleȱadvertisingȱoutlet.ȱ “Overȱ60%ȱofȱsearchȱengineȱvisitorsȱuseȱatȱleastȱtwoȱdifferȬ entȱ searchȱ engines,”ȱ meaning,ȱ asȱ Googleȱ soȱ oftenȱ asserts,ȱ thatȱ competitionȱ reallyȱ isȱ “justȱ aȱ clickȱ away”ȱ forȱ aȱ signifiȬ cantȱnumberȱofȱusers.ȱ “Manyȱ usersȱ useȱ aȱ searchȱ engineȱ asȱ theirȱ pointȱ ofȱ entryȱ toȱ theȱInternet,”ȱbutȱmanyȱdoȱnot—suggestingȱthatȱitȱisȱimporȬ tantȱtoȱinvestigateȱhowȱtheseȱalternativeȱinternetȱaccessȱporȬ talsȱ (includingȱ verticalȱ searchȱ engines,ȱ socialȱ mediaȱ sites,ȱ andȱdirectȱretailȱsites)ȱcompeteȱwithȱsearchȱengines.ȱ “Searcherȱbehaviorȱvariesȱonȱdifferentȱsearchȱengines,”ȱsugȬ gestingȱ thatȱ usersȱ mayȱ optimizeȱ forȱ differentȱ searchȱ engineȱ characteristics,ȱpermittingȱsuccessfulȱproductȱdifferentiation.ȱ “62%ȱ ofȱ advertisersȱ surveyedȱ saidȱ theyȱ planȱ toȱ increaseȱ searchȱmarketingȱspendȱ[sic]ȱoverȱtheȱnextȱ12ȱmonths,”ȱsugȬ gestingȱ increasingȱ advertisingȱ budgetsȱ devotedȱ toȱ onlineȱ outletsȱ andȱ increasingȱ pressureȱ onȱ theȱ tradeȬoffȱ betweenȱ traditionalȱandȱonlineȱadvertising.ȱ “Theȱmainȱobjectiveȱforȱmostȱsearchȱmarketersȱisȱtoȱgenerateȱ onlineȱsalesȱandȱbuildȱbrandȱawareness,”ȱhardlyȱsurprising,ȱ butȱ importantȱ toȱ noteȱ thatȱ onlineȱ searchȱ advertisingȱ isȱ notȱ onlyȱaboutȱgeneratingȱsales,ȱbutȱalsoȱaboutȱtheȱsortȱofȱbrandȱ marketingȱassociatedȱwithȱtraditionalȱadvertising.ȱ “TheȱmainȱproblemȱadvertisersȱfaceȱwasȱfoundȱtoȱbeȱincreasȬ ingȱ [sic]ȱ competition,”ȱ anȱ importantȱ andȱ unsurprisingȱ factȱ suggestingȱdiminishingȱreturnsȱtoȱscale,ȱtheȱabsenceȱofȱdirectȱ networkȱ effects,ȱ andȱ anȱ importantȱ opportunityȱ forȱ lessȬ congestedȱcompetitorsȱtoȱattractȱadvertisingȱrevenues.ȱ “Mobileȱ search,ȱ videoȱ search,ȱ andȱ payȱ perȱ callȱ areȱ areasȱ whereȱ advertisersȱ planȱ toȱ spendȱ moreȱ inȱ 2007.”ȱ Thisȱ indiȬ catesȱ thatȱ onlineȱ searchȱ advertisingȱ facesȱ competitionȱ fromȱ otherȱnonȬtraditionalȱsources,ȱasȱwell,ȱandȱtheȱsourceȱofȱfuȬ tureȱcompetitiveȱthreatsȱisȱuncertain.93ȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 93.ȱSearchȱ Engineȱ Statisticsȱ forȱ 2006–2007,ȱ ACCURACASTȱ (Feb.ȱ 28,ȱ 2007),ȱ http://www.accuracast.com/seoȬweekly/seȬstatistics.php.ȱ
26ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
Theseȱ areȱ hardlyȱ scientificȱ assessmentsȱ orȱ aȱ fullȱ blownȱ ecoȬ nomicȱ analysis,ȱ butȱ theseȱ factsȱ andȱ theirȱ implicationsȱ doȱ sugȬ gestȱ thatȱ anȱ assessmentȱ ofȱ theȱ trueȱ economicallyȬrelevantȱ marketȱ forȱ searchȱ advertisingȱ isȱ anȱ extremelyȱ complicatedȱ enȬ deavor.ȱ Theȱ assessmentȱ isȱ highlyȱ proneȱ toȱ errorȱ andȱ certainlyȱ moreȱnuancedȱthanȱcolloquialȱassessmentsȱwouldȱsuggest.ȱThatȱ colloquialȱ observationsȱ identifyȱ aȱ givenȱ marketȱ byȱ aȱ specificȱ productȱwithinȱthatȱmarketȱdoesȱnotȱobviateȱtheȱneedȱforȱcritiȬ cal,ȱ evidenceȬbasedȱ analysisȱ ofȱ theȱ relevantȱ productȱ marketȱ inȱ whichȱGoogleȱoperates.ȱ Inȱtheȱabsenceȱofȱspecificȱeconometricȱdataȱtoȱdetermineȱtheȱ crossȬelasticitiesȱ ofȱ supplyȱ andȱ demand,ȱ weȱ areȱ stuckȱ withȱ aȱ moreȱimpreciseȱandȱqualitativeȱassessmentȱofȱGoogle’sȱrelevantȱ market.ȱ Evenȱ aȱ qualitativeȱ assessment,ȱ however,ȱ revealsȱ thatȱ Google’sȱmarketȱisȱalmostȱcertainlyȱeitherȱbroaderȱorȱnarrowerȱ thanȱtheȱpresumedȱ“onlineȱsearchȱadvertising”ȱmarket.ȱ TheȱfirstȱquestionȱisȱwhetherȱGoogleȱoperatesȱinȱaȱbroadȱadȬ vertisingȱ marketȱ includingȱ allȱ orȱ mostȱ formsȱ ofȱ advertising,ȱ bothȱ onlineȱandȱ offline.ȱThereȱ isȱ actuallyȱ substantialȱ reasonȱ toȱ doubtȱ theȱ proprietyȱ ofȱ aȱ narrowȱ marketȱ definitionȱ limitedȱ toȱ merelyȱonlineȱsearchȱadvertising.ȱInȱtheȱfirstȱplace,ȱtheȱgeneralȱ defenseȱofȱtheȱpropositionȱisȱneitherȱeconomicȱnorȱeconometric,ȱ butȱanecdotal:ȱSomeȱadvertisersȱsuggestȱthatȱtheyȱdoȱnotȱviewȱ printȱandȱonlineȱadvertisingȱasȱcompetitors,ȱnorȱsearch,ȱcontexȬ tual,ȱ orȱ displayȱ advertisingȱ asȱ competitors.94ȱ Butȱ otherȱ adverȬ tisersȱ clearlyȱ seeȱ theȱ connection,ȱ especiallyȱ withinȱ theȱ constraintsȱ ofȱ limitedȱ advertisingȱ budgets.ȱ Oneȱ surveyȱ ofȱ 200ȱ onlineȱ retailersȱ foundȱ thatȱ “onlineȱ advertisersȱ doȱ inȱ factȱ perȬ ceiveȱ theȱ threeȱ channelsȱ ofȱ onlineȱ advertisingȱ [search,ȱ displayȱ andȱcontextual]ȱasȱsubstitutes.”95ȱAmongȱotherȱthings,ȱtheȱsurȬ veyȱfoundȱthatȱ“[i]nȱweightedȱterms,ȱrespondentsȱrepresentingȱ 83ȱpercentȱofȱallȱadȱspendingȱviewȱgraphicȱadsȱandȱsearchȱadsȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 94.ȱSeeȱ Fed.ȱ Tradeȱ Comm’n,ȱ Statementȱ Concerningȱ Google/DoubleClick,ȱ FTCȱ Fileȱ No.ȱ 071Ȭ0170,ȱ atȱ 3ȱ (Dec.ȱ 20,ȱ 2007),ȱ availableȱ atȱ http://www.ftc.gov/os/ȱ caselist/0710170/071220statement.pdfȱ (“Thus,ȱ searchȱ enginesȱ provideȱ aȱ uniqueȱ opportunityȱforȱadvertisersȱtoȱreachȱpotentialȱcustomers.ȱAdvertisersȱviewȱonlineȱ contentȱprovidersȱdifferently.”).ȱ 95.ȱRobertȱ W.ȱ Hahnȱ &ȱ Halȱ J.ȱ Singer,ȱ Anȱ Antitrustȱ Analysisȱ ofȱ Google’sȱ Proposedȱ Acquisitionȱ ofȱ DoubleClickȱ 5,ȱ 24–32ȱ (AEIȬBrookingsȱ Jointȱ Centerȱ forȱ Regulatoryȱ Studies,ȱ Workingȱ Paperȱ No.ȱ 07Ȭ24,ȱ 2008),ȱ availableȱ atȱ http://papers.ssrn.com/ȱ sol3/papers.cfm?abstract_id=1016189.ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
27ȱ
asȱ substitutes.”96ȱ Atȱ leastȱ oneȱ courtȱ hasȱ likewiseȱ determinedȱ thatȱallȱformsȱofȱatȱleastȱonlineȱadvertisingȱareȱinȱtheȱsameȱreleȬ vantȱmarketȱforȱantitrustȱanalysis.97ȱȱ Evenȱ whereȱ theȱ purposeȱ isȱ differentȱ betweenȱ theȱ differentȱ typesȱofȱadvertising—brandȱrecognitionȱforȱdisplayȱads,ȱeffortsȱ toȱ sellȱ forȱ searchȱ ads—thisȱ isȱ merelyȱ aȱ differenceȱ inȱ degree.ȱ Bothȱ areȱ formsȱ ofȱ reducingȱ theȱ costsȱ ofȱ searchȱ aȱ laȱ Stigler98— andȱ theȱ relevantȱ questionȱ isȱ whetherȱ theȱ differenceȱ isȱ signifiȬ cantȱenoughȱtoȱrenderȱdecisionsȱinȱoneȱmarketȱessentiallyȱunafȬ fectedȱbyȱdecisionsȱorȱpricesȱinȱtheȱother.ȱȱ Thereȱ isȱ someȱ additionalȱ anecdotalȱ evidenceȱ thatȱ thisȱ isȱ notȱ theȱ case.ȱ Oneȱ recentȱ exampleȱ isȱ Pepsi’sȱ decisionȱ notȱ toȱ buyȱ televisionȱ advertisingȱ duringȱ theȱ Superȱ Bowlȱ inȱ 2010ȱ toȱ focusȱ insteadȱonȱaȱparticularȱtypeȱofȱonlineȱcampaign.ȱ“Thisȱyearȱforȱ theȱfirstȱtimeȱinȱ23ȱyears,ȱPepsiȱwillȱnotȱhaveȱadsȱinȱtheȱSuperȱ Bowlȱ telecast.ȱ.ȱ.ȱ.ȱInsteadȱ itȱ isȱ redirectingȱ theȱ millionsȱ itȱ hasȱ spentȱannuallyȱtoȱtheȱInternet.”99ȱWouldȱPepsi’sȱdecisionȱhaveȱ beenȱ differentȱ ifȱ onlineȱ advertisingȱ wereȱ somewhatȱ moreȱ exȬ pensive?ȱ Oneȱ logicallyȱ hasȱ toȱ assumeȱ thatȱ itȱ wouldȱ (althoughȱ weȱdoȱnotȱknowȱhowȱinelasticȱitsȱdemandȱis)ȱgivenȱthatȱapparȬ entlyȱ financialȱ constraintsȱ impelledȱ Pepsiȱ toȱ forbearȱ fromȱ cerȬ tainȱexpensiveȱ(andȱhighlyȱsoughtȱafter)ȱ televisionȱadvertisingȱ atȱtheȱsameȱtimeȱasȱengagingȱinȱitsȱInternetȱstrategy.100ȱȱ Anotherȱ studyȱ suggestsȱ thatȱ thereȱ isȱ indeedȱ aȱ tradeȬoffȱ beȬ tweenȱ onlineȱ andȱ moreȱ traditionalȱ typesȱ ofȱ advertising:ȱ Aviȱ Goldfarbȱ andȱ Catherineȱ Tuckerȱ demonstrateȱ thatȱ displayȱ adȬ vertisingȱ pricingȱ isȱ sensitiveȱ toȱ theȱ availabilityȱ ofȱ offlineȱ alterȬ natives.101ȱ Companiesȱ haveȱ limitedȱ advertisingȱ budgets,ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 96.ȱId.ȱatȱ29.ȱȱ 97.ȱSeeȱ KinderStart.comȱ LLCȱ v.ȱ Google,ȱ Inc.,ȱ No.ȱ C06Ȭ2057JF(RS),ȱ 2007ȱ WLȱ 831806ȱ atȱ *6ȱ (N.D.ȱ Cal.ȱ Mar.ȱ 16,ȱ 2007)ȱ (notingȱ thatȱ “thereȱ isȱ noȱ logicalȱ basisȱ forȱ distinguishingȱtheȱSearchȱAdȱMarketȱfromȱtheȱlargerȱmarketȱforȱInternetȱadvertisȬ ing”);ȱPersonȱv.ȱGoogleȱInc.,ȱ456ȱF.ȱSupp.ȱ2d.ȱ488ȱ(S.D.N.Y.ȱ2006).ȱ 98.ȱGEORGEȱJOSEPHȱSTIGLER,ȱTHEȱORGANIZATIONȱOFȱINDUSTRYȱ201ȱ(Univ.ȱofȱChi.ȱ Pressȱ1983)ȱ(1968).ȱ 99.ȱLarryȱ D.ȱ Woodard,ȱ Pepsi’sȱ Bigȱ Gamble:ȱ Ditchingȱ Superȱ Bowlȱ forȱ Socialȱ Media,ȱ ABCȱNEWS,ȱDec.ȱ23,ȱ2009,ȱhttp://abcnews.go.com/print?id=9402514.ȱ 100.ȱMeanwhile,ȱ manyȱ advertisersȱ doȱ manageȱ comprehensiveȱ advertisingȱ budgetsȱ thatȱ allocateȱ spendingȱ betweenȱ onlineȱ andȱ otherȱ mediaȱ dependingȱ onȱ aȱ combinationȱofȱeffectivenessȱandȱprice.ȱThisȱisȱhardlyȱsurprising.ȱ 101.ȱAviȱGoldfarbȱ&ȱCatherineȱTucker,ȱSearchȱEngineȱAdvertising:ȱPricingȱAdsȱtoȱ Contextȱ 96ȱ (NETȱ Instituteȱ Workingȱ Paperȱ No.ȱ 07Ȭ23,ȱ 2007)ȱ availableȱ atȱ http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1021451&rec=1&srcabs=10084ȱ
ȱ
28ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
distributedȱacrossȱaȱbroadȱrangeȱofȱmediaȱandȱpromotionalȱefȬ forts.ȱAsȱoneȱcommentatorȱnotes:ȱ“Byȱ2011ȱwebȱadvertisingȱinȱ theȱUnitedȱStatesȱwasȱexpectedȱtoȱclimbȱtoȱsixtyȱbillionȱdollars,ȱ orȱ 13ȱ percentȱ ofȱ allȱ adȱ dollars.ȱ Thisȱ meantȱ moreȱ dollarsȱ siȬ phonedȱfromȱtraditionalȱmedia,ȱwithȱtheȱlargestȱsliceȱprobablyȱ goingȱtoȱGoogle.”102ȱ AtȱleastȱoneȱstudyȱconcludesȱthatȱonlineȱandȱofflineȱadvertisȬ ingȱ areȱ notȱ economicallyȬrelevantȱ substitutes,103ȱ andȱ noȱ doubtȱ theseȱ interactionsȱ andȱ crossȬelasticitiesȱ areȱ complicated,ȱ nuȬ anced,ȱ andȱ difficultȱ toȱ detect,ȱ isolate,ȱ andȱ identifyȱ withȱ cerȬ tainty.ȱNevertheless,ȱthisȱstudyȱwasȱlimitedȱtoȱlocalȱadvertising.ȱ Whileȱ theȱ reverseȱ dynamicȱ probablyȱ alsoȱ existsȱ (thatȱ is,ȱ onlineȬ onlyȱretailersȱprobablyȱfindȱofflineȱadvertisingȱlessȱeffectiveȱthanȱ onlineȱadvertising104),ȱtheȱpresenceȱofȱaȱsignificantȱenoughȱnumȬ berȱofȱnationalȱorȱinternationalȱadvertisersȱwithoutȱanȱonlineȬonlyȱ presenceȱcanȱexertȱaȱsignificantȱcompetitiveȱpressureȱonȱadvertisȬ ingȱpricesȱbothȱonlineȱandȱoffline,ȱevenȱifȱlocalȬonlyȱadvertisingȱ isȱnotȱaȱsignificantȱconstraintȱonȱonlineȱadvertising.105ȱ Advertisingȱ revenueȱ onȱ theȱ Internetȱ isȱ drivenȱ initiallyȱ byȱ theȱ sizeȱ ofȱ theȱ audience,ȱ withȱ aȱ significantȱ multiplierȱ forȱ theȱ likeliȬ hoodȱthatȱthoseȱconsumersȱwillȱpurchaseȱtheȱadvertisers’ȱprodȬ ucts106ȱ(basedȱonȱaȱviewer’sȱpropensityȱtoȱ“clickȱthrough”ȱtoȱtheȱ advertiser’sȱ site).ȱ Google’sȱ competitionȱ thusȱ comes,ȱ inȱ varyingȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ (determiningȱtheȱpriceȱofȱ“ambulanceȱchaser”ȱlawyerȱadsȱwasȱsignificantlyȱmoreȱ expensiveȱinȱstatesȱprohibitingȱdirectȱmailȱsolicitationȱbyȱattorneys).ȱ 102.ȱAULETTA,ȱsupraȱnoteȱ72,ȱatȱ16.ȱ 103.ȱLeonardȱN.ȱReidȱetȱal.,ȱLocalȱAdvertisingȱDecisionȱMakers’ȱPerceptionsȱofȱMediaȱ EffectivenessȱandȱSubstitutability,ȱ18ȱJ.ȱMEDIAȱECON.ȱ35ȱ(2005).ȱ 104.ȱAlthoughȱitȱisȱworthȱpointingȱoutȱthatȱGoogleȱitselfȱrecentlyȱadvertisedȱitsȱ onlineȱsearchȱonȱtelevisionȱduringȱtheȱSuperȱBowl.ȱȱ 105.ȱAnd,ȱnotȱsurprisingly,ȱlargeȱnationalȱadvertisersȱareȱamongȱAdWords’ȱtopȱ customers.ȱ Duringȱ oneȱ thirtyȬdayȱ periodȱ inȱ Septemberȱ 2009,ȱ theȱ topȱ fiveȱ AdȬ Wordsȱ spendersȱ includedȱ Progressive,ȱ Target,ȱ andȱ Geicoȱ (theȱ otherȱ twoȱ wereȱ ExpediaȱandȱAOL,ȱbothȱofȱwhichȱhaveȱonlineȬonlyȱpresences).ȱSeeȱBiggestȱOnlineȱ Adȱ Spendersȱ withȱ Adwordsȱ –ȱTopȱ PPCȱAdvertisers,ȱ FRUITIONȱ INTERNETȱ MARKETINGȱ BLOGȱ (Sept.ȱ 24,ȱ 2009),ȱ http://fruition.net/ppcȬmanagement/biggestȬonlineȬadȬ spendersȬwithȬadwordsȬtopȬppcȬadvertisers/.ȱȱ Itȱ shouldȱ alsoȱ beȱ notedȱ thatȱ theȱ Goldfarbȱ andȱ Tuckerȱ studyȱ refutesȱ evenȱ theȱ claimȱthatȱlocalȱadvertisersȱdoȱnotȱseeȱonlineȱandȱofflineȱadvertisingȱasȱsubstitutes.ȱ Seeȱ Goldfarbȱ &ȱ Tucker,ȱ supraȱ noteȱ 101,ȱ atȱ 4ȱ (“Thisȱ impliedȱ substitutabilityȱ ofȱ onlineȱ andȱ offlineȱ advertisingȱ suggestsȱ thatȱ whenȱ policyȱ makersȱ orȱ marketȱ anaȬ lystsȱ seekȱ toȱ defineȱ advertisingȱ markets,ȱ theyȱ shouldȱ considerȱ bothȱ onlineȱ andȱ offlineȱchannelsȱinȱtheirȱmarketȱdefinitions.”).ȱ 106.ȱDavidȱS.ȱEvans,ȱ Theȱ Economicsȱ ofȱ theȱ Onlineȱ Advertisingȱ Industry,ȱ 7ȱ REV.ȱ OFȱ NETWORKȱECON.ȱ359,ȱ35960ȱ(2008).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
29ȱ
degrees,ȱnotȱonlyȱfromȱotherȱsearchȱsites,ȱbutȱalsoȱfromȱanyȱotherȱ siteȱthatȱoffersȱaȱservice,ȱproduct,ȱorȱexperienceȱthatȱconsumersȱ mightȱotherwiseȱobtainȱbyȱfirstȱsearchingȱthroughȱGoogle.ȱȱ Socialȱ mediaȱ sitesȱ likeȱ Twitterȱ andȱ Facebookȱ areȱ thereforeȱ significantȱ competitorsȱ asȱ well,ȱ occupying,ȱ asȱ theyȱ do,ȱ aȱ conȬ siderableȱ amountȱ ofȱ Internetȱ “eyeball”ȱ time.ȱ Theȱ Pepsiȱ deviaȬ tionȱofȱadvertisingȱrevenueȱfromȱtheȱSuperȱBowlȱtoȱtheȱInternetȱ isȱnotȱlikelyȱtoȱinureȱmuchȱtoȱGoogle’sȱbenefitȱasȱtheȱstrategyȱisȱ aȱ “socialȱ mediaȱ play,”ȱ buildingȱ onȱ theȱ expressedȱ brandȱ loyalȬ tiesȱ andȱ peerȱ communicationsȱ thatȱpropelȱ socialȱ media.107ȱ Inȱaȱ worldȱofȱscarceȱadvertisingȱdollarsȱandȱeffectiveȱmarketingȱviaȱ socialȱmediaȱsites,ȱGoogleȱandȱallȱotherȱadvertisers,ȱonlineȱandȱ off,ȱ mustȱ competeȱ withȱ theȱ growingȱ threatȱ toȱ theirȱ revenueȱ fromȱ theseȱ stillȬnovelȱ marketingȱ outlets.ȱ “Ifȱ Facebook’sȱ comȬ munityȱ ofȱ usersȱ gotȱ moreȱ ofȱ theirȱ informationȱ throughȱ [theȱ Facebook]ȱnetwork,ȱtheirȱInternetȱsearchȱengineȱandȱnavigatorȱ mightȱbecomeȱFacebook,ȱnotȱGoogle.”108ȱ Mostȱ obviouslyȱ (andȱ perhapsȱ mostȱ significantly)ȱ Googleȱ facesȱcompetitionȱfromȱitsȱownȱ(andȱotherȱsearchȱengines’)ȱorȬ ganicȱ searchȱ results.ȱ Asȱ notedȱ above,ȱ Google’sȱ paidȱ searchȱ reȬ sultsȱappearȱonȱsearchȱresultȱpagesȱalongsideȱorganicȱresults.109ȱ Searchers—andȱ thusȱ advertisers—takeȱ advantageȱ ofȱ differentȱ characteristicsȱ ofȱ theȱ differentȱ typesȱ ofȱ searchȱ resultsȱ andȱ useȱ organicȱ andȱ paidȱ resultsȱ accordingly,ȱ andȱ theȱ twoȱ sourcesȱ ofȱ marketingȱplainlyȱvieȱforȱadvertisingȱdollars:ȱ AdvertisersȱhaveȱbeenȱgrapplingȱwithȱtheȱtradeȬoffsȱinȱeachȱ ofȱtheseȱtwoȱformsȱofȱreferrals.ȱ.ȱ.ȱ.ȱSomeȱanecdotalȱevidenceȱ suggestsȱ thatȱ thereȱ isȱ aȱ potentialȱ disconnectȱ betweenȱ theȱ perceptionȱofȱsponsoredȱlistingsȱbyȱbusinessȱandȱusers,ȱwithȱ consumersȱ havingȱ aȱ positiveȱ biasȱ towardsȱ organicȱ searchȱ listingsȱ.ȱ.ȱ.ȱ.ȱMoreover,ȱthereȱisȱalsoȱsomeȱanecdotalȱevidenceȱ suggestingȱthatȱpaidȱsearchȱmayȱleadȱtoȱhigherȱconversionsȱ thanȱorganicȱsearchȱ.ȱ.ȱ.ȱ.ȱTheseȱmixedȱfindingsȱthenȱmotivateȱ theȱ questionȱ regardingȱ thatȱ toȱ whatȱ extentȱ shouldȱ firmsȱ inȬ vestȱinȱsponsoredȱsearchȱadvertisementsȱwhenȱtheyȱalsoȱapȬ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 107.ȱSeeȱWoodard,ȱsupraȱnoteȱ99.ȱ 108.ȱAULETTA,ȱsupraȱnoteȱ72,ȱatȱ172–73.ȱ 109.ȱAds,ȱsupraȱnoteȱ74.ȱ
30ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
pearȱ inȱ theȱ organicȱ listingsȱ forȱ aȱ givenȱsearchȱqueryȱ inȱ thatȱ searchȱengine.110ȱ
Advertisingȱ firmsȱ andȱ theȱ companiesȱ thatȱ hireȱ themȱ spreadȱ theirȱmarketingȱbudgetsȱacrossȱtheseȱdifferentȱsourcesȱofȱonlineȱ marketing,ȱ andȱ “searchȱ engineȱ optimizers”—firmsȱ thatȱ helpȱ websitesȱtoȱmaximizeȱtheȱlikelihoodȱofȱaȱvaluableȱ“topȬofȬlist”ȱ organicȱsearchȱplacement—attractȱsignificantȱrevenue.111ȱ Theȱ relationshipȱ betweenȱ organicȱ andȱ paidȱ searchȱ resultsȱ isȱ nuanced.ȱBothȱorganicȱandȱpaidȱsearchȱresultsȱcompeteȱforȱusȬ ers’ȱeyeballsȱandȱclicks:ȱ Oneȱ distinctionȱ ofȱ theȱ coȬlistingȱ structureȱ isȱ thatȱ itȱ createsȱ twoȱ listsȱ competingȱ withȱ eachȱ otherȱ forȱ consumerȱ attenȬ tion.ȱ.ȱ.ȱ.ȱ[T]hoseȱ merchantȱ websitesȱ interestedȱ inȱ sponsoredȱ advertisingȱ mayȱ alsoȱ appearȱ inȱ theȱ organicȱ listȱ andȱ thusȱ couldȱgetȱsignificantȱattentionȱfromȱtheȱorganicȱlistȱwithoutȱ payingȱanything.ȱInȱthisȱsense,ȱtheȱorganicȱlistȱnotȱonlyȱcomȬ petesȱ forȱ consumerȱ attentionȱ butȱ alsoȱ playsȱ aȱ dominatingȱ roleȱinȱsuchȱcompetition.112ȱ
Theȱlastȱpointȱisȱparticularlyȱimportant,ȱgivenȱthatȱorganicȱsearchȱ resultsȱ areȱ availableȱ toȱ advertisersȱ atȱ noȱ costȱ paidȱ toȱ theȱ searchȱ engine,113ȱsuggestingȱtheyȱcouldȱhaveȱaȱstrongȱdiscipliningȱeffectȱ onȱaȱsearchȱengine’sȱabilityȱprofitablyȱtoȱchargeȱaȱmonopolyȱprice.ȱ Moreȱ importantly,ȱ perhaps,ȱ theȱ trueȱ interactionȱ betweenȱ paidȱandȱorganicȱsearchȱresultsȱisȱexceedinglyȱcomplex:ȱ Comparedȱ toȱ theȱ caseȱ withȱ noȱ organicȱ list,ȱ organicȱ listingȱ reȬ sultsȱinȱlowerȱrevenueȱforȱtheȱsearchȱengineȱinȱgeneral,ȱwhereasȱ itȱmayȱinduceȱaȱhigherȱlevelȱofȱsocialȱwelfareȱandȱsalesȱdiverȬ sity.ȱOnȱtheȱoneȱhand,ȱorganicȱlistingȱessentiallyȱsubsidizesȱtheȱ leadingȱ advertisersȱ inȱ prominenceȱ forȱ freeȱ toȱ diluteȱ theirȱ bidȬ dingȱincentiveȱforȱsponsoredȱslots,ȱwhichȱhurtsȱrevenueȱforȱtheȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 110.ȱShaȱYangȱ&ȱAnindyaȱGhose,ȱAnalyzingȱtheȱRelationshipȱBetweenȱOrganicȱandȱ SponsoredȱSearchȱAdvertising:ȱPositive,ȱNegativeȱorȱZeroȱInterdependence?,ȱ29ȱMARKETȬ INGȱSCI.,ȱ602,ȱ603ȱ(2010).ȱ 111.ȱSee,ȱe.g.,ȱBoȱXingȱ&ȱZhanghiȱLin,ȱTheȱImpactȱofȱSearchȱOptimizationȱonȱOnlineȱ Advertisingȱ Market,ȱ inȱ ICECȱ 2006ȱ PROCEEDINGSȱ OFȱ 8THȱ INTERNATIONALȱ CONFERENCEȱONȱELECTRONICȱCOMMERCEȱ519ȱ(2006).ȱ 112.ȱLizhenȱXuȱetȱal.,ȱTooȱOrganicȱforȱOrganicȱListing?ȱInterplayȱBetweenȱOrganicȱandȱ Sponsoredȱ Listingȱ inȱ Searchȱ Advertisingȱ 1,ȱ 2–3ȱ (McCombsȱ Researchȱ Paperȱ Seriesȱ No.ȱ IROMȬ13Ȭ09,ȱ 2009),ȱ availableȱ atȱ http://www.mccombs.utexas.edu/news/ȱ research_calendar/IROM_Xu_10_16_09.pdf.ȱ 113.ȱAdvertisersȱ mayȱ incurȱ costs,ȱ nevertheless,ȱ throughȱ theȱ processȱ ofȱ searchȱ engineȱoptimizationȱinȱanȱeffortȱtoȱmaximizeȱtheȱlikelihoodȱthatȱtheyȱwillȱsecureȱaȱ betterȱplacementȱinȱorganicȱsearchȱresults.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
31ȱ
searchȱengine.ȱOnȱtheȱotherȱhand,ȱwithȱaȱdilutedȱincentiveȱforȱ leadingȱadvertisers,ȱweakȱadvertisersȱhaveȱaȱgreaterȱchanceȱtoȱ winȱ aȱ prominentȱ sponsoredȱ slotȱ toȱ complementȱ theirȱ unsatisȬ factoryȱ prominenceȱ levelȱ inȱ theȱ organicȱ list,ȱ whichȱ leadsȱ toȱ aȱ higherȱ levelȱ ofȱ socialȱ welfareȱ andȱ salesȱ diversity.ȱ Fromȱ theȱ searchȱengineȇsȱperspective,ȱorganicȱlistingȱservesȱasȱaȱbalanceȱ betweenȱ shortȬtermȱ andȱ longȬtermȱ benefit—sacrificingȱ shortȬ termȱ revenueȱ toȱ enhanceȱ totalȱ welfareȱ andȱ salesȱ diversity,ȱ whichȱ couldȱ leadȱ toȱ longȬtermȱ prosperityȱ ofȱ theȱ onlineȱ comȬ munityȱandȱtheȱsearchȱadvertisingȱindustry.114ȱ
Althoughȱ theȱ extentȱ ofȱ competitionȱ alongȱ theseȱ dimensionsȱ variesȱ considerablyȱ byȱ searchȱ termȱ andȱ byȱ advertiserȱ andȱ inȬ dustryȱ characteristics,ȱ theȱ existenceȱ ofȱ suchȱ competitionȱ sugȬ gestsȱ thatȱ relevantȱ marketsȱ mayȱ beȱ narrowerȱ orȱ broaderȱ thanȱ presumedȱ andȱ marketȱ powerȱ difficultȱ toȱ determine.ȱ Inȱ eitherȱ case,ȱtheȱcomplexȱinteractionsȱbetweenȱorganicȱandȱpaidȱsearchȱ results,ȱ whichȱ areȱ atȱ onceȱ bothȱ complementsȱ andȱ substitutes,ȱ dramaticallyȱcomplicatesȱtheȱU.S.ȱDepartmentȱofȱJusticeȱMergerȱ Guidelinesȇȱtraditional,ȱnarrowestȬpossibleȬmarketȱtest.115ȱȱ Thisȱ raisesȱ anȱ interestingȱ caveatȱ toȱ theȱ facileȱ claimsȱ ofȱ wellȬ definedȱ advertisingȱ markets:ȱ manyȱ distinctȱ searchȱ termsȱ andȱ theirȱ searchȱ resultsȱ pages—eachȱ theȱ productȱ ofȱ aȱ particularȱ auctionȱ andȱ aȱ particularȱ setȱ ofȱ webȱ pagesȱ crunchedȱ throughȱ Google’sȱPageRankȱalgorithm—wouldȱconstituteȱseparateȱreleȬ vantȱmarketsȱunderȱaȱSmallȱbutȱSignificantȱandȱNonȬTransitoryȱ Increaseȱ inȱ Priceȱ (SSNIP)ȱ test.116ȱ Theȱ SSNIPȱ test,ȱ aȱ stapleȱ ofȱ Americanȱantitrustȱanalysisȱunderȱtheȱ1992ȱHorizontalȱMergerȱ Guidelines,ȱ determinesȱ theȱ appropriateȱ scopeȱ ofȱ aȱ productȱ marketȱbyȱimposingȱaȱsmallȱbutȱsubstantialȱnonȬtransitiveȱpriceȱ increaseȱuponȱaȱproductȱandȱmeasuringȱtheȱresultingȱelasticityȱ ofȱ demandȱ forȱ theȱ theoreticalȱ market.117ȱ Googleȱ doesȱ notȱ setȱ uniformȱ pricesȱ forȱ adȱ placementsȱ acrossȱ keywordsȱ andȱ aucȬ tions;ȱ rather,ȱ eachȱ keywordȱ isȱ pricedȱ inȱ itsȱ ownȱ repeatedȱ aucȬ tion.ȱNorȱisȱthereȱmuchȱtradeȬoffȱamongȱsearchȱtermsȱforȱscarceȱ spaceȱ onȱ searchȱ resultsȱ pages;ȱ rather,ȱ eachȱ searchȱ termȱ generȬ atesȱ itsȱ ownȱ resultsȱ page,ȱ andȱ thereȱ isȱ littleȱ competitionȱ beȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 114.ȱId.ȱatȱ6.ȱ 115.ȱDEP’TȱOFȱJUSTICEȱ&ȱFED.ȱTRADEȱCOMM’N,ȱHORIZONTALȱMERGERȱGUIDELINESȱ 7ȱ(2010),ȱavailableȱatȱhttp://www.ftc.gov/os/2010/08/100819hmg.pdf.ȱ 116.ȱIdȱ§§ȱȱ8–10.ȱ 117.ȱId.ȱ
32ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
tweenȱ keywordsȱ forȱ space.ȱ Consumersȱ mightȱ complicateȱ marȬ ketȱ determinationsȱ inȱ narrowerȱ marketsȱ evenȱ furtherȱ byȱ byȬ passingȱ searchȱ enginesȱ altogether.ȱ Forȱ example,ȱ Googleȱ mightȱ haveȱnoȱmarketȱpowerȱatȱallȱinȱtheȱmarketȱforȱonlineȱbookȱadȬ vertisingȱifȱconsumersȱgoȱstraightȱtoȱonlineȱbooksellers.ȱ Thus,ȱ Googleȱ competesȱ notȱ onlyȱ withȱ otherȱ generalȱ searchȱ enginesȱ (andȱ possiblyȱ allȱ otherȱ formsȱ ofȱ advertising)ȱ butȱ alsoȱ withȱsoȬcalledȱverticalȱsearchȱengines.ȱTheseȱareȱsearchȱenginesȱ andȱ eȬcommerceȱ websitesȱ withȱ searchȱ functionalityȱ thatȱ speȬ cializeȱ inȱ specificȱ content:ȱ Amazonȱ inȱ books,ȱ music,ȱ andȱ otherȱ consumerȱ goods;ȱ Kayakȱ inȱ travelȱ services;ȱ eBayȱ inȱ consumerȱ auctions;ȱ WebMDȱ inȱ medicalȱ informationȱ andȱproducts;ȱ SourȬ ceToolȱ inȱ businessȬtoȬbusinessȱ supplies;ȱ andȱ manyȱ others.ȱ Toȱ theȱ extentȱ thatȱ Internetȱ usersȱ bypassȱ Googleȱ andȱ beginȱ theirȱ searchesȱatȱoneȱofȱtheseȱspecializedȱsitesȱ(asȱisȱincreasinglyȱtheȱ case),ȱtheȱvalueȱtoȱtheseȱheavilyȬtraffickedȱwebsitesȱfromȱadverȬ tisingȱonȱGoogleȱdecreases.118ȱAtȱtheȱsameȱtime,ȱtheseȱsitesȱandȱ otherȱ aggregatorsȱ likeȱ themȱ offerȱ valuableȱ advertisingȱ outletsȱ forȱotherȱwebsitesȱandȱforȱmanufacturers.ȱȱ Competitionȱ fromȱ verticalȱ searchȱ enginesȱ isȱ intensifiedȱ beȬ causeȱclickȬthroughȱratesȱlikelyȱareȱhigherȱwhenȱconsumersȱareȱ activelyȱsearchingȱforȱsomethingȱtoȱbuy—justȱasȱsearchȱadverȬ tisingȱ targetsȱ consumersȱ whoȱ expressȱ someȱ interestȱ inȱ aȱ parȬ ticularȱsearchȱterm,ȱtheȱeffectȱisȱmagnifiedȱifȱtheȱsearcherȱcanȱbeȱ identifiedȱasȱanȱimmediateȱconsumer.ȱThusȱonlineȱretailersȱlikeȱ CDnowȱ thatȱ canȱ establishȱ theirȱ ownȱ brandsȱ andȱ theirȱ ownȱ navigationȱ channels119ȱ canȱ drawȱ searchers—andȱ advertisers— awayȱ fromȱ Google.ȱ Thatȱ aȱ consumerȱ goesȱ directlyȱ toȱ aȱ retailȱ siteȱwithȱaȱsearchȱitselfȱconveysȱimportantȱandȱvaluableȱinforȬ mationȱtoȱadvertisersȱthatȱisȱnotȱotherwiseȱavailableȱfromȱmostȱ undifferentiatedȱ Googleȱ searches—itȱ certainlyȱ increasesȱ theȱ chanceȱ thatȱ theȱ searcherȱ isȱ searchingȱ toȱ buyȱ aȱ CDȱ ratherȱ thanȱ learnȱsomethingȱaboutȱtheȱsinger.ȱBecauseȱthisȱ“readyȬtoȬbuy”ȱ trafficȱisȱtheȱmostȱvaluable,ȱthereȱisȱaȱpossibilityȱofȱaȱseparatingȱ equilibrium,ȱwithȱmostȱhighȬvalueȱtrafficȱbypassingȱsearchȱenȬ ginesȱ forȱ directȱ retailȱ sites,ȱ andȱ withȱ Googleȱ andȱ otherȱ searchȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 118.ȱForȱexample,ȱinȱtheȱthirtyȱdaysȱendingȱonȱFebruaryȱ23,ȱ2010,ȱlessȱthanȱtenȱ percentȱ ofȱ visitsȱ toȱ eBay.comȱ originatedȱ fromȱ aȱ searchȱ engine.ȱ Seeȱ eBay.comȱ Siteȱ Info,ȱALEXA,ȱhttp://www.alexa.com/siteinfo/ebay.comȱ(lastȱvisitedȱNov.ȱ6,ȱ2010).ȱ 119.ȱSeeȱDonnaȱL.ȱHoffmanȱ&ȱThomasȱP.ȱNovak,ȱHowȱtoȱAcquireȱCustomersȱonȱtheȱ Web,ȱHARV.ȱBUS.ȱREV.,ȱMay–Juneȱ2000,ȱatȱ3,ȱ5,ȱ7.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
33ȱ
enginesȱ servingȱ primarilyȱ nontargeted,ȱ lowerȬvalueȱ traffic.ȱ Amazonȱhasȱevenȱdevelopedȱitsȱownȱsearchȱengine,ȱspecializedȱ forȱconsumerȱsearches.120ȱTheȱimplicationȱisȱthatȱevenȱrelativelyȱ smallȬscaleȱcompetitionȱmayȱpresentȱaȱpotentiallyȱcatastrophicȱ threatȱtoȱGoogle’sȱsearchȱbusiness.ȱ Googleȱfacesȱcompetitionȱfromȱaȱnumberȱofȱsourcesȱandȱinȱaȱ numberȱofȱnuancedȱways.ȱAbsentȱreliableȱempiricalȱevidence,ȱitȱ isȱ difficultȱ toȱ identifyȱ Google’sȱ antitrustȬrelevantȱ market,ȱ butȱ thereȱisȱreasonȱtoȱdoubtȱtheȱtraditionalȱ“onlineȱsearchȱadvertisȬ ing”ȱmarket,ȱbecauseȱitȱignoresȱaȱnumberȱofȱimportantȱdynamȬ icsȱ withinȱ thatȱ proposedȱ marketȱ definition.ȱ Ultimately,ȱ anyȱ marketȱdefinitionȱshouldȱbeȱestablishedȱwithȱeconometricȱdataȱ andȱpertinentȱtheoriesȱofȱsupplierȱandȱconsumerȱbehavior.ȱAbȬ sentȱthis,ȱanyȱmarketȱpowerȱdeterminationsȱthatȱdependȱonȱinȬ tuitionȱandȱmarketȱshareȱcalculationsȱareȱsuspectȱandȱlikelyȱtoȱ exacerbateȱalreadyȱcostlyȱTypeȱIȱerrorȱproblems.ȱ Mostȱ troublingly,ȱ however,ȱ reliableȱ evidenceȱ ofȱ theseȱ (andȱ other)ȱ complicatedȱ marketȱ dynamicsȱ mayȱ neverȱ beȱ available,ȱ andȱasȱwithȱtheȱmodelsȱofȱanticompetitiveȱconductȱcriticizedȱinȱ PartȱIIȱabove,ȱcourtsȱmayȱendȱupȱforcedȱtoȱmakeȱmarketȱdefiniȬ tionȱdeterminationsȱbasedȱonȱincompleteȱevidenceȱandȱunsupȬ portedȱ theoriesȱ thatȱ failȱ accuratelyȱ toȱ captureȱ theȱ complicatedȱ economicsȱ ofȱ consumer,ȱ advertiser,ȱ andȱ supplierȱ conduct.ȱ EfȬ fortsȱtoȱglossȱoverȱtheseȱcomplicationsȱbyȱrelyingȱonȱdocumenȬ taryȱ proxiesȱ forȱ economicȱ relevanceȱ areȱ anotherȱ significantȱ sourceȱofȱerrorȱinȱantitrustȱcaseȱlaw.121ȱȱ C.
TheȱImportanceȱofȱQualityȱScoresȱ
TheȱheartȱofȱtheȱdominantȱtheoryȱofȱSectionȱ2ȱliabilityȱagainstȱ Googleȱrelatesȱtoȱ Google’sȱ useȱofȱ qualityȱscoringȱinȱ influencingȱ theȱ outcomeȱ ofȱ itsȱ AdWordsȱ auctions.ȱ Theȱ qualityȱ score— introducedȱ byȱ Googleȱ andȱ nowȱ usedȱ byȱ allȱ generalȱ searchȱ enȬ gines—isȱ anȱ importantȱ businessȱ innovationȱ employingȱ adȬ vancedȱ algorithmicȱ technologyȱ toȱ maximizeȱ theȱ relevanceȱ ofȱ searchȱ resultsȱ andȱ thusȱ theȱ valueȱ ofȱ theȱ searchȱ engineȱ toȱ users,ȱ theȱlikelihoodȱofȱrevenueȬproducingȱimpressionsȱtoȱadvertisers,ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 120.ȱINNOVATIONSȱ INȱ SEARCHȱ TECHNOLOGIES,ȱ http://www.a9.comȱ (lastȱ visitedȱ Nov.ȱ6,ȱ2010).ȱ 121.ȱSeeȱ Geoffreyȱ A.ȱ Manneȱ &ȱ E.ȱ Marcellusȱ Williamson,ȱ Hotȱ Docsȱ vs.ȱ Coldȱ EcoȬ nomics:ȱTheȱUseȱandȱMisuseȱofȱBusinessȱDocumentsȱinȱAntitrustȱEnforcementȱandȱAdjuȬ dication,ȱ47ȱARIZ.ȱL.ȱREV.ȱ609,ȱ610ȱ(2005).ȱ
34ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
andȱ revenueȱ toȱ Google.122ȱ “Googleȇsȱ introductionȱ ofȱ clickȬ throughȱweightingȱ[qualityȱscores]ȱinȱ2002ȱisȱregardedȱasȱanȱimȬ portantȱ competitiveȱ advantageȱ andȱ Yahoo!ȇsȱ introductionȱ ofȱ clickȬthroughȱ weightsȱ intoȱ itsȱ rankingȱ algorithmȱ inȱ earlyȱ 2007ȱ (“Panama”)ȱwasȱhighlyȱpublicizedȱasȱaȱcriticalȱimprovement.”123ȱ Theȱbasicȱideaȱbehindȱtheȱqualityȱscoreȱisȱtoȱpredictȱinȱadvanceȱ theȱ likelihoodȱ thatȱ aȱ particularȱ advertisementȱ willȱ generateȱ aȱ clickthrough,ȱandȱthenȱensureȱthatȱtheȱmoreȱrelevantȱadvertiseȬ mentsȱ (thoseȱ withȱ higherȱ clickthroughȱ rates)ȱ receiveȱ higherȱ placementsȱ inȱ theȱ paidȱ searchȱ resultsȱ (fromȱ whichȱ advertisersȱ canȱ expectȱ moreȱ clickthroughȱ andȱ Googleȱ thusȱ moreȱ revenue).ȱ Advertisersȱwithȱlowerȱqualityȱscoresȱareȱobligatedȱtoȱpayȱmoreȱ perȱ clickȱ toȱ winȱ higherȱ searchȱ resultȱ positionsȱ thanȱ advertisersȱ withȱbetterȱqualityȱscoresȱbecauseȱtopȱplacementȱofȱlessȬrelevantȱ adsȱ leadsȱ toȱ lowerȱ revenueȱ forȱ theȱ searchȱ engineȱ andȱ degradesȱ theȱoverallȱqualityȱofȱtheȱsearchȱengine’sȱrelevance.ȱ Theȱbasicȱintuitionȱisȱthis:ȱaȱsearchȱengineȱwantsȱtoȱsellȱeachȱadȱ impression—eachȱ placementȱ inȱ itsȱ paidȱ searchȱ results—toȱ theȱ advertiserȱwhoȱisȱwillingȱtoȱpayȱtheȱmostȱforȱit,ȱsoȱitȱcaresȱaboutȱ costȱ perȱ impression.ȱ Theȱ advertiserȱ caresȱ aboutȱ costȱ perȱ clickȱ (orȱ costȱperȱconversion,ȱforȱwhichȱcostȱperȱclickȱisȱaȱweakȱbutȱmoreȬ readilyȬmeasurableȱ proxy)ȱ whichȱ isȱ whatȱ theȱ advertiserȱ mustȱ pay.ȱTheȱtwoȱareȱrelatedȱbyȱtheȱfollowingȱsimpleȱequation:ȱ costȱperȱimpressionȱ=ȱcostȱperȱclickȱxȱclicksȱperȱimpressionȱ
“Clicksȱ perȱ impression”ȱ isȱ theȱ clickthroughȱ rateȱ (theȱ rateȱ atȱ whichȱusersȱactuallyȱclickȱonȱresultsȱtheyȱseeȱonȱaȱsearchȱresultsȱ page),ȱwhichȱisȱtheȱdominantȱcomponentȱofȱtheȱqualityȱscore.124ȱ Googleȇsȱ qualityȱ scoreȱ system—andȱ thatȱ ofȱ allȱ otherȱ generalȱ searchȱ engines,ȱ includingȱ Yahoo!ȱ andȱ Microsoft’sȱ Bing—thusȱ allowsȱ itȱ toȱ sellȱ impressionsȱ whileȱ advertisersȱ buyȱ clicks,ȱ perȬ mittingȱ eachȱ simultaneouslyȱ toȱ maximizeȱ theȱ relevantȱ metric.ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 122.ȱJACOBSON,ȱsupraȱnoteȱ84.ȱ 123.ȱSusanȱ Atheyȱ &ȱ Glennȱ Ellison,ȱ Positionȱ Auctionsȱ withȱ Consumerȱ Searchȱ 3ȱ (Harvardȱ Univ.ȱ &ȱ Mass.ȱ Inst.ȱ ofȱ Tech.ȱ Workingȱ Paper,ȱ No.ȱ D44,L86,M37,ȱ 2008)ȱ availableȱatȱhttp://kuznets.fas.harvard.edu/~athey/position.pdf.ȱ 124.ȱ “Inȱ Google’sȱ (andȱ moreȱ recentlyȱ inȱ others’)ȱ adȱ auctions,ȱ theȱ winningȱ bidȬ dersȱ areȱ notȱ theȱ firmsȱ withȱ theȱ highestȱ perȬclickȱ bids:ȱ advertisersȱ areȱ rankedȱ onȱ theȱbasisȱofȱtheȱproductȱofȱtheȱtheirȱ[sic]ȱbidȱandȱaȱfactorȱthatȱisȱsomethingȱlikeȱanȱ estimatedȱ clickthroughȱ rate.ȱ Theȱ roughȱ motivationȱ forȱ thisȱ isȱ straightforward:ȱ weightingȱbidsȱbyȱtheirȱclickȬthroughȱratesȱisȱakinȱtoȱrankingȱthemȱonȱtheirȱconȬ tributionsȱtoȱsearchȬengineȱrevenuesȱ(asȱopposedȱtoȱperȬclickȱrevenuesȱwhichȱisȱaȱ lessȱnaturalȱobjective).”ȱId.ȱatȱ26.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
35ȱ
Atȱtheȱsameȱtime,ȱbecauseȱadvertisersȱpayȱforȱclicksȱinsteadȱofȱ impressions,ȱ theyȱ haveȱ anȱ incentiveȱ toȱ increaseȱ theirȱ bidsȱ toȱ obtainȱbetterȱplacement,ȱknowingȱthatȱtheyȱwillȱnotȱhaveȱtoȱpayȱ forȱ theseȱ “excessive”ȱ impressions.ȱ Theȱ qualityȱ scoreȱ allowsȱ aȱ searchȱ engineȱ toȱ limitȱ impressions,ȱ enablingȱ itȱ toȱ weedȱ outȱ theseȱ lowȬrelevanceȱ ads—somethingȱ thatȱ wouldȱ otherwiseȱ beȱ impossibleȱ throughȱ anȱ unadjustedȱ auctionȱ process.ȱ Theȱ sameȱ analysesȱapplyȱtoȱdisplayȱadsȱplacedȱonȱotherȱwebȱpages.ȱThus,ȱ qualityȱscoresȱareȱalsoȱusedȱonȱtheȱAdSenseȱplatform.ȱȱ AllegationsȱofȱanticompetitiveȱconductȱsurroundingȱtheȱqualȬ ityȱscoreȱturnȱlessȱonȱitsȱexistence—allȱmajorȱsearchȱenginesȱuseȱ qualityȱ scoresȱ toȱ improveȱ theȱ relevanceȱ ofȱ searchȱ results—thanȱ onȱ itsȱ opacity.ȱ Theȱ specificȱ determinantsȱ ofȱ qualityȱ scoresȱ areȱ keptȱhiddenȱbyȱdesignȱandȱnecessity.ȱAsȱoneȱGoogleȱviceȱpresiȬ dentȱhasȱnoted:ȱ Weȱ are,ȱ toȱ beȱ honest,ȱ quiteȱ secretiveȱ aboutȱ whatȱ weȱ do.ȱ Thereȱareȱtwoȱreasonsȱforȱit:ȱcompetitionȱandȱabuse.ȱCompeȬ titionȱisȱprettyȱstraightforward.ȱNoȱcompanyȱwantsȱtoȱshareȱ itsȱ secretȱ recipesȱ withȱ itsȱ competitors.ȱ Asȱ forȱ abuse,ȱ ifȱ weȱ makeȱourȱrankingȱformulasȱtooȱaccessible,ȱweȱmakeȱitȱeasierȱ forȱpeopleȱtoȱgameȱtheȱsystem.ȱSecurityȱbyȱobscurityȱisȱneverȱ theȱstrongestȱmeasure,ȱandȱweȱdoȱnotȱrelyȱonȱitȱexclusively,ȱ butȱitȱdoesȱpreventȱaȱlotȱofȱabuse.125ȱ
Inȱ additionȱ toȱ improvingȱ theȱ qualityȱ ofȱ itsȱ searchȱ engine,ȱ Google’sȱuseȱofȱqualityȱscoresȱandȱitsȱcontrolȱofȱtheȱtermsȱofȱitsȱ auctionsȱlikelyȱgeneratesȱhigherȱrevenues.ȱAsȱnotedȱabove,ȱadȬ vertisersȱ haveȱ anȱ incentive,ȱ absentȱ aȱ deviceȱ likeȱ theȱ qualityȱ score,ȱ toȱ overȬbidȱ forȱ impressions,ȱ resultingȱ inȱ someȱ lessȬ relevantȱadsȱgainingȱbetterȱplacementsȱandȱthusȱyieldingȱfewerȱ revenueȬgeneratingȱ clicksȱ fromȱ theȱ moreȬrelevantȱ ads,ȱ pushedȱ furtherȱ downȱ theȱ searchȱ resultsȱ page.ȱ Forȱ example,ȱ bothȱ Intelȱ andȱAdvancedȱMicroȱDevicesȱ(AMD)ȱwouldȱlikeȱtoȱappearȱfirstȱ inȱpaidȱsearchȱresultsȱforȱtheȱkeywordȱ“Intel.”ȱOnȱaverage,ȱusȬ ersȱwouldȱfindȱIntel’sȱadvertisementȱtoȱbeȱmoreȱrelevant.ȱInȱtheȱ absenceȱ ofȱ qualityȱ scores,ȱ however,ȱ AMDȱ couldȱ outbidȱ Intelȱ andȱreceiveȱtheȱfirstȱposition.ȱAlthoughȱAMDȱhopesȱtoȱcaptureȱ someȱ clickthroughsȱ byȱ usersȱ lookingȱ forȱ Intel,ȱ itȱ alsoȱ knowsȱ thatȱ itȱ canȱ affordȱ toȱ bidȱ aȱ higherȱ amountȱ perȱ clickȱ becauseȱ itȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 125.ȱUdiȱ Manber,ȱ Introductionȱ toȱ Googleȱ Searchȱ Quality,ȱ THEȱ OFFICIALȱ GOOGLEȱ BLOGȱ(Mayȱ20,ȱ2008,ȱ6:20ȱPM),ȱhttp://googleblog.blogspot.com/2008/05/introductionȬ toȬgoogleȬsearchȬquality.html.ȱ
36ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
willȱ stillȱ receiveȱ fewerȱ clicksȱ thanȱ Intelȱ would.ȱ Intel,ȱ onȱ theȱ otherȱhand,ȱwillȱbidȱaȱrelativelyȱsmallȱamountȱforȱtheȱfirstȱposiȬ tion,ȱnotȱbecauseȱitȱdoesȱnotȱvalueȱit,ȱbutȱbecauseȱitȱknowsȱthatȱ itȱwillȱreceiveȱaȱlargeȱnumberȱofȱclickthroughsȱatȱwhatȱmayȱbeȱaȱ veryȱ highȱ aggregateȱ price.ȱ Butȱ thisȱ dynamic,ȱ potentiallyȱ leadȬ ingȱ toȱ AMDȱ inȱ theȱ firstȱ paidȱ searchȱ resultȱ position,ȱ isȱ subȬ optimalȱ notȱ onlyȱ forȱ Intelȱ butȱ alsoȱ forȱ Google’sȱ usersȱ andȱ forȱ Google’sȱ shareholders.ȱ Usersȱ willȱ findȱ aȱ nonrelevantȱ searchȱ resultȱinȱtheȱtopȱpositionȱandȱwillȱthusȱdevalueȱtheȱsearchȱenȬ gine.ȱ Andȱ Googleȱ willȱ receiveȱ smallerȱ revenueȱ becauseȱ ofȱ theȱ relativeȱ irrelevanceȱ ofȱ theȱ topȱ resultsȱ andȱ theȱ correspondinglyȱ smallerȱnumberȱofȱclicksȱ(evenȱatȱaȱslightlyȱhigherȱprice).ȱȱ Withoutȱ aȱ mechanismȱ qualitativelyȱ toȱ matchȱ searchȱ termsȱ withȱadvertisers,ȱendȱusersȱandȱadvertisersȱforegoȱqualityȱandȱ Googleȱ foregoesȱ profits.ȱ Theȱ nextȱ Partȱ willȱ discussȱ atȱ lengthȱ whetherȱ theseȱ orȱ otherȱ actionsȱ constituteȱ impermissibleȱ excluȬ sionaryȱconduct.ȱAtȱaȱminimum,ȱthereȱareȱclearlyȱprocompetiȬ tiveȱ justificationsȱ forȱ theȱ useȱ (andȱ secrecy)ȱ ofȱ qualityȱ scoresȱ includingȱbothȱensuringȱproductȱqualityȱandȱmaximizingȱreveȬ nue.ȱFurthermore,ȱtheȱfullȱeffectȱofȱtheseȱsortsȱofȱbusinessȱinnoȬ vationsȱisȱprobablyȱunknown,ȱevenȱtoȱGoogle.ȱAnyȱchallengeȱtoȱ theȱ useȱ ofȱ qualityȱ scoresȱ asȱ anȱ anticompetitiveȱ deviceȱ shouldȱ turnȱonȱaȱsetȱofȱspecificȱfactualȱallegationsȱandȱaȱdemonstrationȱ ofȱ aȱ cognizableȱ anticompetitiveȱ effect.ȱ Importantly,ȱ andȱ perȬ hapsȱprecludingȱsuchȱaȱfinding,ȱtheȱqualityȱscoreȱhelpsȱtoȱconȬ vertȱ quantityȱ intoȱ quality.ȱ Havingȱ moreȱ searchersȱ isȱ notȱ necessarilyȱ valuableȱ toȱ advertisersȱ perȱ se,ȱ butȱ havingȱ moreȱ searchersȱ findȱ anȱ advertiserȱ specificallyȱ whenȱ theȱ searcherȱ isȱ mostȱlikelyȱtoȱbuyȱsomethingȱisȱworthȱaȱconsiderableȱamount.ȱ Toȱ theȱ extentȱ thatȱ theȱ qualityȱ scoreȱ allowsȱ Googleȱ betterȱ toȱ matchȱ qualitativeȱ aspectsȱ ofȱ advertisersȱ andȱ endȱ users,ȱ itȱ inȬ creasesȱtheȱvalueȱofȱtheȱsystemȱtoȱallȱparticipantsȱinȱaȱwayȱthatȱ mereȱincreasesȱinȱscaleȱdoȱnot.ȱ D.
NetworkȱEffectsȱ
ManyȱclaimȱthatȱGoogle’sȱsearchȱengineȱandȱsearchȱadvertisȬ ingȱ representȱ aȱ multiȬsidedȱ platformȱ thatȱ benefitsȱ fromȱ theȱ presenceȱ ofȱ networkȱ effects,ȱ butȱ nearlyȱ allȱ suchȱ claimsȱ takeȱ theseȱeffectsȱforȱgrantedȱorȱassertȱthemȱwithoutȱempiricalȱbackȬ ingȱ andȱ deriveȱ legalȱ conclusionsȱ fromȱ theirȱ existenceȱ withoutȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
37ȱ
analysis.126ȱ Thereȱ mayȱ indeedȱ beȱ relevantȱ networkȱ effectsȱ inȱ Google’sȱ business,ȱ but,ȱ asȱ withȱ allȱ novelȱ andȱ innovativeȱ busiȬ nesses,ȱ theȱ facileȱ conclusionsȱ areȱ oftenȱ incorrect.ȱ Muchȱ moreȱ andȱbetterȱempiricalȱanalysisȱshouldȱbeȱbroughtȱtoȱbearȱbeforeȱ competitiveȱassessmentsȱareȱmade.127ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 126.ȱSee,ȱe.g.,ȱOrenȱBrachaȱ&ȱFrankȱPasquale,ȱFederalȱSearchȱCommission?ȱAccess,ȱ Fairnessȱ andȱ Accountabilityȱ inȱ theȱ Lawȱ ofȱ Search,ȱ 93ȱ CORNELLȱ L.ȱ REV.ȱ 1149,ȱ 1181ȱ (2008)ȱ (“Networkȱ Effectsȱ inȱ Improvingȱ Searchȱ Responsiveness.ȱ Theȱ moreȱ searchesȱ anȱ engineȱgets,ȱtheȱbetterȱableȱitȱisȱtoȱsharpenȱandȱperfectȱitsȱalgorithm.ȱTheȱresultȱisȱ thatȱeachȱadditionalȱuserȱdecreasesȱtheȱcostȱofȱaȱbetterȱqualityȱserviceȱforȱallȱusers.ȱ Thus,ȱincumbentsȱwithȱlargeȱnumbersȱofȱusersȱenjoyȱsubstantialȱadvantagesȱoverȱ smallerȱ entrants.”)ȱ (referencingȱ DAVIDȱ A.ȱ VISEȱ &ȱ MARKȱ MALSEED,ȱ THEȱ GOOGLEȱ STORYȱ215ȱ(2005));ȱKristineȱLaudadioȱDevine,ȱPreservingȱCompetitionȱinȱMultiȬSidedȱ InnovativeȱMarkets:ȱHowȱDoȱYouȱSolveȱaȱProblemȱLikeȱGoogle?,ȱ10ȱN.C.ȱJ.L.ȱ&ȱTECH.ȱ59ȱ (2008)ȱ (describingȱ Google’sȱ networkȱ effectsȱ byȱ analogyȱ andȱ conjecture,ȱ althoughȱ goingȱ intoȱ farȱ moreȱ detailȱ thanȱ anyȱ ofȱ theȱ otherȱ articlesȱ cited);ȱ Jamesȱ GrimmelȬ mann,ȱHowȱtoȱFixȱtheȱGoogleȱBookȱSearchȱSettlement,ȱJ.ȱ INTERNETȱ L.,ȱ Apr.ȱ2009,ȱatȱ1,ȱ 14ȱ(“Thus,ȱGoogleȇsȱfirstȬpastȬtheȬpostȱstatusȱhereȱcouldȱeasilyȱturnȱintoȱaȱdurableȱ monopoly.ȱThatȱmightȱbeȱtheȱinevitableȱresultȱanyway;ȱthisȱisȱaȱmarketȱwithȱsubȬ stantialȱeconomiesȱofȱscaleȱandȱpositiveȱnetworkȱeffects.”);ȱPeterȱS.ȱMenell,ȱKnowlȬ edgeȱAccessibilityȱandȱPreservationȱPolicyȱforȱtheȱDigitalȱAge,ȱ44ȱHOUS.ȱ L.ȱ REV.ȱ 1013,ȱ 1052ȱ(2007)ȱ(listingȱ“concentrationȱ(dueȱtoȱnetworkȱeffects)”ȱonȱaȱtableȱdescribingȱ competitiveȱ effectsȱ inȱ GoogleȱBookȱ Search);ȱ Vivaȱ R.ȱ Moffat,ȱ Regulatingȱ Search,ȱ 22ȱ HARV.ȱ J.L.ȱ &ȱ TECHȱ 475,ȱ489ȱ(2009)ȱ(“Additionally,ȱwhileȱGoogleȱmayȱnotȱbeȱaȱmoȬ nopoly,ȱitȱcertainlyȱhasȱaȱgreatȱdealȱofȱmarketȱpowerȱandȱnetworkȱeffectsȱalsoȱexistȱ inȱtheȱsearchȱworld.ȱTheseȱfactors,ȱtogetherȱwithȱtheȱarguablyȱsubstantialȱbarriersȱ toȱ entryȱ inȱ theȱ searchȱ engineȱ market,ȱ permitȱ theȱ analogyȱ toȱ commonȱ carriers.”ȱ (citingȱ aȱ Newȱ Yorkȱ Timesȱ articleȱ referringȱ toȱ “networkȱ advantages”));ȱ Williamȱ D.ȱ Rahm,ȱWatchingȱoverȱtheȱWeb:ȱAȱSubstantiveȱEqualityȱRegimeȱforȱBroadbandȱApplicaȬ tions,ȱ 24ȱ YALEȱ J.ȱ ONȱ REG.ȱ 1,ȱ 16ȱ (2007)ȱ (“Broadbandȱ technologyȱ exhibitsȱ twoȱ netȬ workȱeffects.ȱ.ȱ.ȱ.ȱCompaniesȱlikeȱGoogle,ȱVonageȱandȱAmazonȱwillȱdevelopȱmoreȱ servicesȱ whenȱ theyȱ areȱ confidentȱ thatȱ thereȱ areȱ moreȱ usersȱ whomȱ theyȱ canȱ reach.”);ȱ J.ȱ Gregoryȱ Sidak,ȱ Aȱ ConsumerȬWelfareȱ Approachȱ toȱ Networkȱ Neutralityȱ Regulationȱ ofȱ theȱ Internet,ȱ 2ȱ J.ȱ COMPETITIVEȱ L.ȱ &ȱ ECON.ȱ 349,ȱ 454ȱ (2006)ȱ (“Networkȱ effectsȱmakeȱtheȱmarketȱforȱInternetȱportalsȱhighlyȱconcentrated.ȱEntryȱisȱdifficultȱ becauseȱaȱcriticalȱmassȱofȱusersȱhasȱchosenȱaȱparticularȱportalȱ(Yahoo!ȱorȱGoogle)ȱ toȱ beginȱ theirȱ Internetȱ experience.ȱ Aȱ criticalȱ massȱ ofȱ advertisersȱ hasȱ followed.”);ȱ Kevinȱ Werbach,ȱ Onlyȱ Connect,ȱ 22ȱ BERKELEYȱ TECH.ȱ L.J.ȱ 1233,ȱ 1292ȱ (2007)ȱ (“NoneȬ theless,ȱ itȱ isȱ possibleȱ forȱ applicationsȱ toȱ becomeȱ exclusiveȱ platformsȱ withȱ antiȬ competitiveȱ effectsȱ similarȱ toȱ thoseȱ ofȱ exclusiveȱ physicalȱ broadbandȱ networks.ȱ Googleȇsȱ dominantȱ searchȱ engineȱ andȱ MySpaceȇsȱ massiveȱ socialȱ networkingȱ siteȱ mightȱbeȱcandidatesȱforȱsuchȱscrutinyȱatȱsomeȱpointȱinȱtheȱfuture.ȱBecauseȱtheseȱ areȱ networkȬcentricȱ applications,ȱ whateverȱ abilityȱ theyȱ haveȱ toȱ distortȱ competiȬ tionȱandȱinnovationȱarisesȱfromȱtheirȱabilityȱtoȱcaptureȱnetworkȱeffects.”).ȱȱ 127.ȱ“[Multisidedȱplatforms]ȱhaveȱbusinessȱmodelsȱthatȱareȱnotȱyetȱwellȱunderȬ stoodȱ andȱ engageȱ inȱ highlyȱ complexȱ businessȱ strategies;ȱ unusualȱ practicesȱ areȱ suspectȱpracticesȱinȱourȱexperience.”ȱDavidȱS.ȱEvansȱ&ȱMichaelȱD.ȱNoel,ȱDefiningȱ Marketsȱ Thatȱ Involveȱ MultiȬSidedȱ Platformȱ Businesses:ȱ Anȱ Empiricalȱ Frameworkȱ Withȱ anȱ Applicationȱ Google’sȱ Purchaseȱ ofȱ DoubleClickȱ 4ȱ (AEIȬBrookingsȱ Jointȱ Centerȱ forȱ
ȱ
38ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
Networkȱ effectsȱ occurȱ whenȱ theȱ valueȱ ofȱ aȱ goodȱ orȱ serviceȱ increasesȱ asȱ theȱ numberȱ ofȱ peopleȱ whoȱ useȱ itȱ grows.128ȱ Inȱ theȱ contextȱ ofȱ aȱ multisidedȱ operationȱ suchȱ asȱ Google’sȱ searchȱ enȬ gineȱ andȱ searchȱ advertisingȱ platform,ȱ indirectȱ networkȱ effectsȱ mightȱariseȱwhenȱanȱincreasedȱbaseȱofȱendȱusersȱforȱoneȱsideȱofȱ theȱplatformȱincreasesȱtheȱvalueȱofȱtheȱplatformȱtoȱadvertisersȱ onȱtheȱotherȱside.ȱȱ Asȱ notedȱ above,ȱ networkȱ effectsȱ areȱ generallyȱ beneficial,ȱ alȬ thoughȱ thereȱ isȱ someȱ disputeȱ overȱ whetherȱ andȱ underȱ whatȱ conditionsȱ theyȱ mightȱ alsoȱ raiseȱ exclusionaryȱ concerns.129ȱ Asȱ discussedȱ above,ȱ transactionsȱ involvingȱ complementaryȱ prodȬ uctsȱ (indirectȱ networkȱ effects)ȱ fullyȱ internalizeȱ theȱ benefitsȱ ofȱ consumingȱcomplementaryȱgoodsȱandȱdoȱnotȱpresentȱanȱexcluȬ sionaryȱconcern.130ȱInȱGoogle’sȱcase,ȱthisȱmeansȱthat,ȱwhileȱadȬ ditionalȱ endȱ usersȱ mayȱ increaseȱ theȱ valueȱ ofȱ Google’sȱ (orȱ anyȱ otherȱ searchȱ engine’s)ȱ platformȱ toȱ itsȱ advertisers,ȱ thisȱ increaseȱ inȱ valueȱ isȱ internalizedȱ byȱ theȱ platform,ȱ andȱ advertisersȱ areȱ chargedȱ accordingly.ȱ Typicalȱ “feedbackȱ effects”ȱ seenȱ inȱ manyȱ multisidedȱ platformsȱ areȱ attenuatedȱ orȱ absentȱ inȱ Google’sȱ businessȱbecauseȱtheȱeffectsȱareȱgenerallyȱunidirectional:ȱadverȬ tisersȱwantȱmoreȱendȱusers,ȱbutȱendȱusersȱcareȱlittleȱorȱnothingȱ aboutȱtheȱnumberȱofȱadvertisers.ȱ Moreover,ȱasȱinȱallȱanalysisȱofȱnetworkȱeffects,ȱtheȱstandardȱ assumptionȱthatȱ quantityȱ aloneȱ determinesȱ theȱ strengthȱ ofȱ theȱ effectȱisȱlikelyȱmistaken.131ȱRather,ȱtoȱtheȱextentȱthatȱadvertisersȱ careȱaboutȱendȱusers,ȱtheyȱcareȱaboutȱmanyȱofȱtheirȱcharacterisȬ tics.ȱAnȱincreaseȱinȱtheȱnumberȱofȱusersȱwhoȱareȱlookingȱonlyȱ forȱ informationȱ andȱ neverȱ toȱ purchaseȱ goodsȱ mayȱ beȱ ofȱ littleȱ valueȱtoȱadvertisers.ȱȱ Thus,ȱ becauseȱ onlineȱ searchȱ advertisersȱ targetȱ customersȱ andȱ salesȱ theyȱ careȱ aboutȱ theȱ sizeȱ ofȱ theȱ endȱ userȱ networkȱ onlyȱtoȱtheȱextentȱthatȱthisȱsizeȱcorrelatesȱwithȱincreasedȱsales.ȱ Toȱ aȱ firstȱ approximationȱ increasedȱ usageȱ shouldȱ leadȱ toȱ inȬ creasedȱ clickthroughs,ȱ andȱ increasedȱ clickthroughsȱ shouldȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ Regulatoryȱ Studies,ȱ Workingȱ Paperȱ 07Ȭ18,ȱ 2007),ȱ availableȱ atȱ http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1089073.ȱ 128.ȱStanȱJ.ȱLiebowitzȱ&ȱStephenȱE.ȱMargolis,ȱNetworkȱExternality:ȱAnȱUncommonȱ Tragedy,ȱ8ȱJ.ȱECON.ȱPERSP.ȱ133,ȱ135ȱ(1994).ȱ 129.ȱSeeȱinfraȱnoteȱ184.ȱ 130.ȱSeeȱLiebowitzȱ&ȱMargolis,ȱsupraȱnoteȱ128;ȱSpulber,ȱsupraȱnoteȱ8.ȱ 131.ȱSee,ȱe.g.,ȱSpulber,ȱsupraȱnoteȱ8.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
39ȱ
leadȱtoȱmoreȱsales.ȱButȱbecauseȱadvertisersȱpayȱperȱclick,ȱifȱtheȱ numberȱ ofȱ clicksȱ withoutȱ purchaseȱ increasesȱ withȱ increasedȱ usageȱ sufficientlyȱ fasterȱ thanȱ theȱ numberȱ ofȱ clicksȱ withȱ purȬ chase,ȱtheȱincreaseȱinȱsearchȱengineȱusageȱmayȱbeȱaȱcostȱratherȱ thanȱaȱbenefit.ȱThus,ȱtheȱabilityȱofȱtheȱsearchȱengineȱtoȱdeliverȱ notȱonlyȱscaleȱbutȱalsoȱquality—basedȱonȱtheȱcharacteristicsȱofȱ itsȱ usersȱ andȱ itsȱ abilityȱ toȱ matchȱ usersȱ withȱ advertisers132— determinesȱtheȱamountȱadvertisersȱareȱwillingȱtoȱpay.ȱForȱthisȱ reason,ȱ theȱ valueȱ ofȱ aȱ searchȱ engineȱ mayȱ notȱ increaseȱ asȱ theȱ numberȱofȱusersȱgrows,ȱand,ȱtoȱtheȱextentȱthatȱitȱdoes,ȱthisȱisȱaȱ directȱ functionȱ ofȱ theȱ qualityȱ score.ȱ Assessingȱ networkȱ orȱ scaleȱeffectsȱisȱextremelyȱdifficultȱinȱsearchȱengineȱadvertising,ȱ andȱ scaleȱ mayȱ notȱ evenȱ correlateȱ withȱ increasedȱ valueȱ overȱ someȱrangesȱofȱsize.ȱ Theȱ problemȱ forȱ thoseȱ whoȱ wouldȱ pointȱ toȱ indirectȱ networkȱ effectsȱ asȱ aȱ barrierȱ toȱ entryȱ forȱ Google’sȱ competitorsȱ isȱ thatȱ adȬ vertisersȱpayȱonlyȱwhenȱaȱuserȱclicksȱthroughȱitsȱpaidȱsearchȱreȬ sultȱtoȱtheȱadvertiser’sȱ landingȱ page.133ȱTheȱconsequenceȱisȱthatȱ theȱfullȱvalueȱofȱGoogle’sȱadvertisingȱplatformȱisȱinternalizedȱbyȱ theȱsystem,ȱwithȱadvertisersȱpayingȱaȱpriceȱthatȱreflectsȱtheȱfullȱ valueȱ ofȱ theirȱ useȱ ofȱ Google’sȱ platform—thereȱ areȱ noȱ externalȬ ities,ȱand,ȱasȱmentioned,ȱnetworkȱsizeȱmayȱnotȱbeȱrelevantȱtoȱadȬ vertisers.ȱIfȱhavingȱmoreȱusersȱmakesȱaȱclickȱmoreȱlikelyȱtoȱleadȱ toȱaȱconversion,ȱadvertisersȱwillȱpayȱmoreȱperȱclick,ȱinternalizingȱ theȱeffect.ȱIfȱhavingȱmoreȱusersȱmakesȱaȱclickȱmoreȱlikelyȱinȱtheȱ firstȱplace,ȱadvertisersȱalsoȱpayȱmoreȱbecauseȱtheyȱpayȱforȱeachȱ click.ȱInȱeitherȱcase,ȱtheȱeffectȱisȱfullyȱinternalized.ȱ Anyȱ claimȱ thatȱ Googleȱ possessesȱ marketȱ powerȱ protectedȱ byȱ anȱindirectȱnetworkȬeffectȱbarrierȱtoȱentryȱmustȱgrappleȱwithȱtheȱ problemȱthatȱtheseȱeffectsȱareȱinternalizedȱandȱofȱuncertainȱsign,ȱ andȱ thusȱ thatȱ theyȱ function,ȱ competitivelyȱ speaking,ȱ noȱ differȬ entlyȱ thanȱ anyȱ otherȱ measureȱ ofȱ qualityȱ (andȱ correspondingȱ price).ȱAȱcompetitorȱcanȱcompeteȱbyȱofferingȱlowerȱ“quality”ȱatȱ aȱlowerȱpriceȱifȱnecessary,ȱandȱbecauseȱnoȱbenefitsȱareȱleftȱexterȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 132.ȱSeeȱsupraȱnotesȱ124–26ȱandȱaccompanyingȱtextȱ(discussingȱtheȱroleȱofȱqualȬ ityȱscoresȱinȱincreasingȱrelevance).ȱ 133.ȱThereȱisȱconceivablyȱsomeȱbenefitȱtoȱanȱadvertiser,ȱparticularlyȱinȱtermsȱofȱ brandȱrecognition,ȱfromȱsimplyȱappearingȱonȱtheȱGoogleȱsearchȱresultsȱpage,ȱevenȱ ifȱanȱadȱisȱnotȱclicked.ȱWeȱassumeȱtheȱvalueȱofȱthisȱrecognitionȱisȱnegligible,ȱbutȱaȱ fullȱassessmentȱofȱtheȱeconomicsȱofȱGoogle’sȱbusinessȱwouldȱlikelyȱrequireȱsomeȱ assessmentȱofȱthisȱdynamic.ȱ
40ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
nalȱ inȱ Google’sȱ business,ȱ itȱ isȱ notȱ necessaryȱ toȱ compensateȱ adȬ vertisersȱ(orȱendȱusers)ȱforȱlostȱexternalȱbenefitsȱfromȱswitchingȱ toȱaȱcompetitor.ȱAȱcompetitorȱwithȱsmallerȱscaleȱbutȱbetterȱqualȬ ityȱ canȱ alsoȱ compete,ȱ evenȱ atȱ higherȱ prices;ȱ scaleȱ isȱ notȱ inherȬ entlyȱaȱbarrierȱtoȱcompetitionȱinȱsearchȱengineȱadvertising.ȱȱ Aȱ variantȱ ofȱ theȱ indirectȱ networkȱ effectȱ argumentȱ wasȱ proȬ poundedȱmostȱnotablyȱinȱtheȱTradeCometȱcomplaint.134ȱTradeȬ Cometȱ claimedȱ thatȱ theȱ valueȱ ofȱ anyȱ searchȱ advertisingȱ platformȱ increasesȱ asȱ theȱ popularityȱ ofȱ itsȱ searchȱ engineȱ grows.135ȱInherentȱinȱthisȱargumentȱisȱtheȱideaȱthatȱusersȱpreferȱ searchȱ enginesȱ thatȱ offerȱ theȱ newestȱ andȱ mostȱ functionalȱ freeȱ features.ȱ Suchȱ featuresȱ canȱ onlyȱ beȱ developedȱ withȱ considerȬ ableȱsurplusȱadvertisingȱrevenue,ȱmakingȱitȱdifficultȱforȱnascentȱ rivalsȱ toȱ gainȱ theȱ searchȱ trafficȱ necessaryȱ toȱ becomeȱ viableȱ alȬ ternativesȱtoȱGoogle.136ȱȱ Butȱ thisȱ isȱ notȱ anȱ argumentȱthatȱ turnsȱ onȱ networkȱ effectsȱ atȱ all;ȱ rather,ȱ itȱ isȱ simplyȱ anȱ argumentȱ aboutȱ financingȱ andȱ theȱ availabilityȱ ofȱ capitalȱ toȱ investȱ inȱ productȱ improvements.ȱ Itȱ isȱ anȱargumentȱthatȱthereȱmayȱbeȱsupplyȬsideȱeconomiesȱofȱscopeȱ andȱ scale,ȱ butȱ thisȱ isȱ neitherȱ aȱ uniqueȱ orȱ uniquelyȬinterestingȱ conclusion,ȱ norȱ oneȱ withȱ particularlyȱ interestingȱ antitrustȱ imȬ plications.ȱAndȱthoughȱGoogleȱperhapsȱgeneratesȱtheȱfundsȱforȱ itsȱcontinuedȱproductȱdevelopmentȱthroughȱitsȱsuccessfulȱbusiȬ ness,ȱ theȱ sameȱ businessȱ modelȱ needȱ notȱ beȱ adoptedȱ byȱ comȬ petitors.ȱ Inȱ fact,ȱ Microsoft,ȱ oneȱ ofȱ Google’sȱ primaryȱ competitors,ȱ hasȱ aȱ marketȱ capitalizationȱ substantiallyȱ largerȱ thanȱ Google’s,ȱ andȱ higherȱ profitsȱ generatedȱ byȱ itsȱ otherȱ busiȬ nessesȱ toȱ investȱ inȱ searchȱ engineȱ functionalityȱ improvements.ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 134.ȱComplaintȱ¶ȱ91,ȱTradeComet.ComȱLLCȱv.ȱGoogle,ȱInc.,ȱ693ȱF.ȱSupp.ȱ2dȱ370ȱ (S.D.N.Y.ȱ2009)ȱ(No.ȱ09Civ.1400(SHS)).ȱ 135.ȱId.ȱ¶ȱ81.ȱ 136.ȱThisȱargumentȱisȱanalogizedȱtoȱaȱclaimȱthatȱreceivedȱtractionȱinȱtheȱMicroȬ softȱ litigation.ȱ Seeȱ Unitedȱ Statesȱ v.ȱ Microsoftȱ Corp.,ȱ 253ȱ F.3dȱ 34,ȱ 54–55ȱ (D.C.ȱCir.ȱ 2001)ȱ(enȱbanc)ȱ(perȱcuriam)ȱ(findingȱmonopolyȱpowerȱasȱaȱresultȱofȱaȱ“‘chickenȬ andȬegg’ȱ situationȱ [that]ȱ ensuresȱ thatȱ [software]ȱ applicationsȱ willȱ continueȱ toȱ beȱ writtenȱforȱtheȱalreadyȱdominantȱWindows,ȱwhichȱinȱturnȱensuresȱthatȱconsumersȱ willȱ continueȱ toȱ preferȱ itȱ overȱ otherȱ operatingȱ systems”).ȱ Butȱ theȱ comparisonȱ isȱ inapposite.ȱInȱMicrosoft,ȱtheȱincentivesȱofȱplatformȱusersȱexternalȱtoȱtheȱplatformȱ itselfȱwereȱbeingȱdrivenȱbyȱtheȱfeedbackȱeffectȱbetweenȱsidesȱofȱtheȱnetwork.ȱHereȱ theȱ argumentȱ hasȱ theȱ platformȱ itselfȱ (Google)ȱ drivingȱ theȱ incentivesȱ throughȱ inȬ vestment.ȱ Inȱ theȱ presentȱ caseȱ theȱ networkȱ effectȱ isȱ endogenousȱ andȱ underȱ Google’sȱ(orȱaȱcompetitor’s)ȱcontrol.ȱConversely,ȱinȱMicrosoftȱtheȱeffectȱwasȱexogeȬ nousȱandȱthereforeȱdifficultȱforȱcompetitorsȱtoȱdisrupt.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
41ȱ
Thereȱisȱnoȱreasonȱwhyȱitȱmattersȱifȱthisȱinvestmentȱcomesȱfromȱ advertisingȱ revenue,ȱ theȱ saleȱ ofȱ operatingȱ systems,ȱ orȱ outsideȱ capitalȱsources.137ȱȱ Theȱ casualȱ invocationȱ ofȱ networkȱ effectsȱ inȱ searchȱ advertisȬ ingȱ isȱ seeminglyȱ refutedȱ byȱ theȱ moreȱ specificȱ realitiesȱ ofȱ theȱ searchȱ advertisingȱ market.ȱ Endȱ usersȱ areȱ insensitiveȱ toȱ theȱ numberȱ ofȱ otherȱ usersȱ inȱ theȱ network,ȱ andȱ thus,ȱ thereȱ areȱ noȱ directȱnetworkȱeffectsȱonȱthatȱsideȱofȱtheȱplatform.ȱExceptȱtoȱtheȱ limitedȱextentȱthatȱtheȱqualityȱofȱaȱsearchȱalgorithmȱmayȱbeȱafȬ fectedȱ byȱ theȱ numberȱ ofȱ usersȱ overȱ aȱ relevantȱ rangeȱ ofȱ users,ȱ endȱusersȱreceiveȱnoȱincidentalȱbenefitȱfromȱothers’ȱuseȱofȱtheȱ sameȱsearchȱengine.ȱAtȱtheȱsameȱtime,ȱanȱincumbentȱwillȱfindȱitȱ difficultȱ toȱ tradeȱ onȱ aȱ comparativeȱ costȱ advantageȱ toȱ staveȱ offȱ competitionȱgivenȱthatȱtheȱpriceȱchargedȱtoȱendȱusersȱisȱalreadyȱ zero.ȱ Althoughȱ thisȱ zeroȱ priceȱ alsoȱ makesȱ itȱ difficultȱ forȱ enȬ trantsȱtoȱattractȱendȱusersȱwithȱaȱlowerȱprice,138ȱthisȱtracksȱcomȬ petitionȱ inȱ aȱ perfectlyȱ competitive,ȱ nonnetworkedȱ industry,ȱ whereȱ incumbentsȱ areȱchargingȱ aȱpriceȱ equalȱ toȱ marginalȱcostȱ andȱentrantsȱareȱforcedȱtoȱsufferȱinitialȱlosses,ȱcompeteȱonȱnonȬ priceȱ dimensions,ȱ orȱ improveȱ productionȱ efficiency.ȱ Indeed,ȱ thisȱ compulsionȱ towardsȱ increasedȱ quantity,ȱ reducedȱ pricesȱ whereȱpossible,ȱorȱincreasedȱqualityȱ asȱanȱirreducibleȱ byprodȬ uctȱofȱcompetitionȱisȱtheȱveryȱpurposeȱofȱantitrustȱlaw.ȱ Itȱalsoȱisȱunlikelyȱthatȱendȱusersȱwillȱfindȱaȱlargerȱnumberȱofȱ advertisersȱ toȱ beȱ aȱ featureȱ ofȱ theȱ system.ȱ Newȱ entrantsȱ mightȱ actuallyȱ beȱ moreȱ attractiveȱ initiallyȱ forȱ havingȱ fewerȱ advertisȬ ersȱandȱads.ȱTheȱoneȱexceptionȱwouldȱbeȱwhereȱtheȱqualityȱofȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 137.ȱSeeȱDevine,ȱsupraȱnoteȱ126,ȱatȱ78–80.ȱThisȱArticleȱmakesȱtheȱsameȱargument.ȱ ThisȱisȱnotȱaȱnetworkȱeffectȱatȱallȱbutȱmerelyȱaȱdescriptionȱofȱaȱtwoȬsidedȱmarket,ȱ whereȱ revenueȱ isȱ obtainedȱ fromȱ onlyȱ oneȱ sideȱ ofȱ theȱ market.ȱ Thatȱ theseȱ profitsȱ mayȱ beȱ reinvestedȱ toȱ attractȱ moreȱ customersȱ seemsȱ incidentalȱ toȱ anyȱ networkȱ effectȱallegationȱandȱmerelyȱdescriptiveȱofȱaȱparticularȱbusinessȱmodel.ȱItȱisȱakinȱ toȱsayingȱthatȱaȱoneȬsidedȱmarketȱexhibitsȱnetworkȱeffectsȱifȱsellingȱmoreȱproductsȱ leadsȱ toȱ higherȱ revenueȱ whichȱ isȱ thenȱ usedȱ toȱ innovateȱ inȱ waysȱ thatȱ sellsȱ moreȱ products.ȱPlainly,ȱthisȱisȱneitherȱaȱnetworkȱeffectȱnorȱaȱbarrierȱtoȱentry.ȱ 138.ȱAndȱofȱcourseȱ“difficult”ȱdoesȱnotȱmeanȱ“impossible.”ȱEntrantsȱcouldȱpayȱ forȱnewȱusers,ȱandȱtheyȱdoȱsoȱinȱaȱvarietyȱofȱcreativeȱways.ȱMicrosoft,ȱforȱexamȬ ple,ȱintroducedȱitsȱ“cashback”ȱprogramȱonȱitsȱLiveȱSearchȱ(nowȱBing)ȱsearchȱenȬ gine,ȱ offeringȱ endȱ usersȱ whoȱ searchedȱ forȱ andȱ purchasedȱ productsȱ usingȱ Microsoft’sȱsearchȱplatformȱaȱdiscountȱonȱtheȱpurchasedȱproduct.ȱBingȱShopping,ȱ WIKIPEDIA,ȱ http://en.wikipedia.org/wiki/Bing_Shopping#Cashback_programȱ (lastȱ visitedȱNov.ȱ6,ȱ2010);ȱseeȱDanielȱF.ȱSpulber,ȱConsumerȱCoordinationȱinȱtheȱSmallȱandȱ inȱtheȱLarge:ȱImplicationsȱforȱAntitrustȱinȱMarketsȱwithȱNetworkȱEffects,ȱ4ȱJ.ȱ COMPETIȬ TIONȱL.ȱ&ȱECON.ȱ207,ȱ257–58(2008).ȱ
42ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
theȱsearchȱproduct—theȱsearchȱalgorithm—isȱitselfȱaffectedȱbyȱ theȱnumberȱofȱendȱusersȱorȱtheȱnumberȱofȱadvertisersȱthroughȱ aȱfeedbackȱeffect.ȱThisȱisȱaȱdifficultȱclaimȱtoȱassessȱfromȱoutsideȱ theȱindustry.ȱItȱseemsȱsuperficiallyȱ plausible,ȱbutȱitȱisȱnotȱnecȬ essarilyȱtheȱcase.ȱBasedȱonȱconversationsȱweȱhaveȱhadȱwithȱinȬ dustryȱ insiders,ȱ itȱ appearsȱ thatȱ algorithmicȱ resultsȱ areȱ onlyȱ weaklyȱaffectedȱbyȱtheȱnumberȱofȱendȱusersȱorȱsearches.ȱSearchȱ algorithmsȱ requireȱ aȱ minimumȱ scaleȱ toȱ establishȱ theirȱ effecȬ tiveness,ȱ butȱ thisȱ minimumȱ scaleȱ mayȱ beȱ easilyȱ reachedȱ (andȱ arguablyȱ hasȱ beenȱ reachedȱ byȱ allȱ ofȱ theȱ majorȱ searchȱ engineȱ competitorsȱ andȱ evenȱ smallȱ upstartȱ companies).ȱ Aboveȱ miniȬ mumȱscale,ȱthereȱisȱlimitedȱadvantageȱtoȱhavingȱmoreȱendȱusȬ ersȱ andȱ moreȱ searches.ȱ Efficientȱ searchȱ engineȱ managementȱ appearsȱ toȱ requireȱ assessmentȱ andȱ evaluationȱ ofȱ onlyȱ aȱ smallȱ fractionȱofȱtotalȱsearches,ȱandȱthereȱisȱrapidlyȱdiminishingȱmarȬ ginalȱreturnȱtoȱincorporatingȱmoreȱsearchesȱintoȱtheȱsearchȱalȬ gorithm.ȱ Asȱ aȱ result,ȱ viableȱ competitionȱ isȱ availableȱ atȱ fairlyȱ smallȱ scales,ȱ andȱ competingȱ searchȱ enginesȱ shouldȱ beȱ ableȱ toȱ produceȱ organicȱ searchȱ resultsȱ asȱ effectivelyȱ asȱ aȱ largeȬscaleȱ incumbent,ȱ subjectȱ onlyȱ toȱ theȱ limitationsȱ ofȱ theȱ searchȱ algoȬ rithm’sȱ designȱ andȱ execution.139ȱ Moreover,ȱ Googleȱ establishedȱ itsȱ industryȬleadingȱ positionȱ onȱ theȱ basisȱ ofȱ aȱ tinyȱ fractionȱ ofȱ theȱ volumeȱ thatȱ evenȱ theȱ smallestȱ searchȱ enginesȱ seeȱ today.ȱ Tellingly,ȱtheȱnumberȱofȱsearchesȱonȱYahoo!ȱtodayȱisȱaboutȱtheȱ sameȱasȱtheȱnumberȱonȱGoogleȱjustȱtwoȱyearsȱago.ȱ Theȱ issueȱ isȱ aȱ bitȱ moreȱ complicatedȱ fromȱ theȱ perspectiveȱ ofȱ potentialȱ advertisers.ȱ Again,ȱ thereȱ isȱ noȱ directȱ networkȱ effect,ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 139.ȱThisȱmakesȱespeciallyȱtroublingȱtheȱunfoundedȱandȱundocumentedȱclaimsȱ foundȱ inȱ Brachaȱ andȱ Pasquale’sȱ Federalȱ Searchȱ Commissionȱ paper.ȱ Inȱ explainingȱ whyȱ theyȱ thinkȱ “robustȱ andȱ dynamicȱ competitionȱ [is]ȱ unlikely,”ȱ andȱ whyȱ theȱ searchȱengineȱmarketȱisȱsubjectȱtoȱ“highȱbarriersȱtoȱentry,”ȱtheȱauthorsȱreportȱthatȱ “[s]earchȱalgorithmsȱmayȱbeȱanalogousȱtoȱtheȱhighȬcostȱinfrastructureȱrequiredȱforȱ entryȱ intoȱ theȱ utilityȱ orȱ railroadȱ markets.”ȱ Brachaȱ &ȱ Pasquale,ȱ supraȱ noteȱ 126,ȱ atȱ 1180–81.ȱAtȱtheȱsameȱtime,ȱtheȱauthorsȱargueȱthatȱ“[t]heȱmoreȱsearchesȱanȱengineȱ gets,ȱtheȱbetterȱableȱitȱisȱtoȱsharpenȱandȱperfectȱitsȱalgorithm.”ȱId.ȱTheȱlastȱpointȱisȱ notȱ true,ȱ orȱ atȱ leastȱ notȱ trueȱ beyondȱ aȱ minimumȱ scale.ȱ Andȱ consequently,ȱ theȱ claimȱ thatȱ anȱ algorithmȱ is,ȱ inȱ essence,ȱ aȱ “naturalȱ monopoly”ȱ orȱ essentialȱ facilityȱ (thusȱrequiringȱregulatoryȱorȱantitrustȱinterventionȱtoȱpryȱopenȱaccessȱtoȱcompetiȬ tors)ȱ isȱ equallyȱ fallacious.ȱ Whileȱ Google’sȱ specificȱ algorithmȱ isȱ notȱ accessibleȱ byȱ competitors,ȱitȱisȱnotȱtheȱcaseȱthatȱanotherȱviableȱalgorithmȱisȱinaccessible,ȱnorȱtheȱ dataȱ sufficientȱ toȱ manageȱ it.ȱ And,ȱ ofȱ course,ȱ inȱ additionȱ toȱ Google’sȱ majorȱ comȬ petitorsȱthereȱareȱdozensȱofȱotherȱsearchȱenginesȱcompetingȱwithȱGoogle’sȱspecificȱ algorithm.ȱ Seeȱ Listȱ ofȱ searchȱ engines,ȱ WIKIPEDIA,ȱ http://en.wikipedia.org/wiki/ȱ List_of_search_enginesȱ(lastȱvisitedȱNov.ȱ6,ȱ2010).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
43ȱ
andȱ theȱ presenceȱ ofȱ moreȱ advertisersȱ isȱ aȱ strongȱ costȱ toȱ otherȱ advertisers;ȱratherȱthanȱaȱnetworkȱeffect,ȱthereȱisȱaȱcongestionȱefȬ fect.140ȱ Thisȱ congestionȱ costȱ isȱ magnifiedȱ byȱ theȱ impositionȱ ofȱ Google’sȱqualityȱscoringȱintoȱtheȱauctionȬplacementȱprocess,ȱasȱ everyȱ advertiser’sȱ placementȱ isȱ dependentȱ onȱ theȱ qualityȱ ofȱ otherȱ advertisersȱ biddingȱ inȱ eachȱ auction.141ȱ Whileȱ advertisersȱ mayȱ orȱ mayȱ notȱ benefitȱ fromȱ indirectȱ networkȱ effectsȱ onȱ theȱ advertiser’sȱ sideȱ ofȱ theȱsearchȱ engineȱ platform,ȱ theȱ congestionȱ effectȱresultingȱfromȱtheȱsimpleȱcompetitiveȱdynamicȱofȱmanyȱ buyersȱcompetingȱforȱaȱscarceȱresourceȱisȱundeniable.ȱ Theȱ uncertaintyȱ surroundingȱ theȱ economicȱ consequencesȱ ofȱ Google’sȱbusinessȱandȱitsȱbusinessȱpracticesȱshouldȱcompelȱexȬ tremeȱ cautionȱ viewedȱ withinȱ anȱ errorȬcostȱ framework.ȱ Theȱ risksȱ arisingȱ fromȱ misapplicationȱ ofȱ economicȱ theoryȱ andȱ aȱ woefullyȱ poorȱ understandingȱ ofȱ theȱ consequencesȱ ofȱ Google’sȱ innovativeȱ productsȱ andȱ businessȱ practices,ȱ coupledȱ withȱ theȱ dramaticȱ costsȱ ofȱ suchȱ errors,ȱ shouldȱ counselȱ againstȱ antitrustȱ interventionȱ withoutȱ someȱ significantȱ directȱ economicȱ dataȱ toȱ contradictȱaȱplausible,ȱprocompetitiveȱanalysis.ȱ IV.
THEȱMONOPOLIZATIONȱCASEȱAGAINSTȱGOOGLEȱ
Thisȱ Partȱ willȱ discussȱ theȱ mostȱ fullyȱ developedȱ realȱ caseȱ againstȱ Googleȱ embodiedȱ inȱ theȱ TradeCometȱ complaint,142ȱ andȱ otherȱ aspectsȱ ofȱ aȱ hypotheticalȱ caseȱ againstȱ Google.ȱ Thisȱ Partȱ willȱ addressȱ theȱ legalȱ andȱ economicȱ theoriesȱ underlyingȱ theȱ actualȱ andȱ hypotheticalȱ cases,ȱ highlightingȱ theȱ pitfallsȱ ofȱ antiȬ trustȱenforcementȱagainstȱGoogle,ȱbutȱmoreȱgenerally,ȱtheȱanaȬ lyticalȱ weaknessesȱ ofȱ anȱ enforcementȱ approachȱ thatȱ eschewsȱ errorȬcostȱ principlesȱ inȱ itsȱ decisionȬmakingȱ processes.ȱ Asȱ aȱ prefatoryȱmatter,ȱthisȱtaskȱrequiresȱknowledgeȱofȱtheȱapplicableȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 140.ȱEvidenceȱ demonstratesȱ thatȱ thickerȱ auctionsȱ resultȱ inȱ higherȱ prices,ȱ irreȬ spectiveȱ ofȱ theȱ characteristicsȱ ofȱ additionalȱ bidders—aȱ presumablyȱ undesirableȱ consequenceȱ(toȱtheȱbidders)ȱofȱaȱbroaderȱnetwork.ȱSee,ȱe.g.,ȱJosephȱFarrellȱ&ȱPaulȱ Klemperer,ȱCoordinationȱ andȱ LockȬin:ȱCompetitionȱ withȱ Switchingȱ Costsȱ andȱ Networkȱ Effectsȱ inȱ 3ȱ THEȱ HANDBOOKȱ OFȱ INDUSTRIALȱ ORGANIZATIONȱ 2018ȱ (M.ȱ Armstrongȱ andȱR.ȱPorterȱeds.,ȱ2007)ȱ(“Second,ȱthereȱmayȱbeȱnoȱintraȬgroupȱnetworkȱeffects;ȱ thereȱmayȱevenȱbeȱintraȬgroupȱcongestion.ȱThus,ȱgivenȱtheȱnumberȱofȱphotograȬ phers,ȱaȱdeveloperȱprefersȱfewerȱotherȱdevelopersȱforȱcompetitiveȱreasons,ȱjustȱasȱ withȱmerchantsȱacceptingȱcreditȱcards.”);ȱPaulȱKlemperer,ȱWhatȱReallyȱMattersȱ inȱ AuctionȱDesign,ȱ16ȱJ.ȱECON.ȱPERSP.ȱ169,ȱ172ȱ(2002).ȱ 141.ȱȱAtheyȱ&ȱEllison,ȱsupraȱnoteȱ123.ȱȱ 142.ȱComplaint,ȱsupraȱnoteȱ134.ȱ
44ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
monopolizationȱ standards.ȱ Weȱ beginȱ withȱ someȱ preliminaryȱ discussionȱ ofȱ theseȱ standardsȱ beforeȱ turningȱ toȱ ourȱ antitrustȱ analysisȱofȱGoogle’sȱspecificȱbusinessȱconduct.ȱȱ A.
FirstȱPrinciplesȱofȱMonopolizationȱEnforcementȱȱ
Sectionȱ2ȱofȱtheȱShermanȱActȱmakesȱitȱunlawfulȱforȱanyȱpersonȱ toȱ “monopolize,ȱ orȱ attemptȱ toȱ monopolize,ȱ orȱ combineȱ orȱ conȬ spireȱwithȱanyȱotherȱpersonȱorȱpersons,ȱtoȱmonopolizeȱanyȱpartȱ ofȱtheȱtradeȱorȱcommerceȱamongȱtheȱseveralȱStates,ȱorȱwithȱforȬ eignȱ nations.”143ȱ Theȱ languageȱ ofȱ theȱ statuteȱ isȱ notoriouslyȱ illȬ equippedȱ toȱ helpȱ resolveȱ actualȱ cases.144ȱ Nonetheless,ȱ itȱ isȱ wellȱ establishedȱthatȱtheȱoffenseȱofȱmonopolizationȱrequiresȱdemonȬ strationȱ ofȱ bothȱ “(1)ȱ theȱ possessionȱ ofȱ monopolyȱ powerȱ inȱ theȱ relevantȱmarketȱandȱ(2)ȱtheȱwillfulȱacquisitionȱorȱmaintenanceȱofȱ thatȱ powerȱ asȱ distinguishedȱ fromȱ growthȱ orȱ developmentȱ asȱ aȱ consequenceȱofȱaȱsuperiorȱproduct,ȱbusinessȱacumen,ȱorȱhistoricȱ accident.”145ȱ Courtsȱ andȱ antitrustȱ scholarsȱ haveȱ struggledȱ withȱ assigningȱ administrableȱ contentȱ toȱ theȱ languageȱ ofȱ Sectionȱ 2,146ȱ spurringȱaȱscholarlyȱdebateȱoverȱwhetherȱconstructingȱaȱunifiedȱ monopolizationȱtestȱtoȱapplyȱtoȱallȱvarietiesȱofȱbusinessȱconductȱ fallingȱwithinȱtheȱscopeȱofȱtheȱstatuteȱisȱpossibleȱorȱdesirable.147ȱȱ Therefore,ȱasȱdiscussedȱinȱPartȱIIȱabove,ȱtheȱkeyȱchallengeȱfacȬ ingȱanyȱproposedȱanalyticalȱframeworkȱforȱevaluatingȱmonopoȬ lizationȱ claims,ȱ isȱ distinguishingȱ procompetitiveȱ fromȱ anticompetitiveȱconduct.ȱAntitrustȱerrorsȱareȱinevitableȱbecauseȱ muchȱ ofȱ whatȱ isȱ potentiallyȱ actionableȱ conductȱ underȱ theȱ antiȬ trustȱlawsȱfrequentlyȱactuallyȱbenefitsȱconsumers,ȱandȱgeneralistȱ judgesȱareȱcalledȱuponȱtoȱidentifyȱanticompetitiveȱconductȱwithȱ imperfectȱinformation.ȱAsȱJudgeȱEasterbrookȱhasȱnoted,ȱtheȱopȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 143.ȱ15ȱU.S.C.ȱ§ȱȱ2ȱ(2006).ȱ 144.ȱȱBORK,ȱsupraȱnoteȱ31,ȱatȱ57ȱ(“TheȱbareȱlanguageȱofȱtheȱShermanȱActȱconveysȱ littleȱ.ȱ.ȱ.ȱ.”);ȱFrankȱH.ȱEasterbrook,ȱVerticalȱArrangementsȱandȱtheȱRuleȱofȱReason,ȱ53ȱ ANTITRUSTȱL.J.ȱ135,ȱ136ȱ(1984)ȱ(“TheȱlanguageȱofȱtheȱShermanȱActȱgovernsȱnoȱrealȱ cases.”);ȱ Thomasȱ E.ȱ Kauper,ȱ Sectionȱ Twoȱ ofȱ theȱ Shermanȱ Act:ȱ Theȱ Searchȱ forȱ StanȬ dards,ȱ93ȱGEO.ȱ L.J.ȱ1623,ȱ1623ȱ(2005)ȱ(“SectionȱTwoȱofȱtheȱShermanȱActȱhasȱbeenȱaȱ sourceȱ ofȱ puzzlementȱ toȱ lawyers,ȱ judgesȱ andȱ scholars,ȱ aȱ puzzlementȱ derivedȱ inȱ largeȱpartȱfromȱtheȱstatute’sȱextraordinaryȱbrevity.”).ȱ 145.ȱUnitedȱStatesȱv.ȱGrinnellȱCorp.,ȱ384ȱU.S.ȱ563,ȱ570–571ȱ(1965).ȱ 146.ȱU.S.ȱ DEP’Tȱ OFȱ JUSTICE,ȱ COMPETITIONȱ ANDȱ MONOPOLY:ȱ SINGLEȱ FIRMȱ CONDUCTȱ UNDERȱ SECTIONȱ 2ȱ OFȱ THEȱ SHERMANȱ ACTȱ (2008),ȱ availableȱ atȱ http://www.usdoj.gov/atr/public/reports/236681.htm.ȱȱ 147.ȱSee,ȱe.g.,ȱKauper,ȱsupraȱnoteȱ144.ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
45ȱ
timalȱantitrustȱrulesȱminimizeȱtheȱcostsȱofȱtheseȱerrorsȱbyȱestabȬ lishingȱandȱallocatingȱappropriateȱburdensȱofȱproof.148ȱGivenȱtheȱ tendencyȱinȱantitrustȱtoȱcondemnȱbusinessȱpracticesȱthatȱareȱnotȱ wellȱ understood,ȱ orȱ forȱ whichȱ anȱ efficiencyȱ explanationȱ cannotȱ beȱ profferedȱ thatȱ fitsȱ intoȱ theȱ categoriesȱ establishedȱ byȱ earlierȱ cases,ȱitȱisȱkeyȱthatȱanyȱburdenȬshiftingȱapproachȱtoȱmonopolizaȬ tionȱ retainsȱ theȱrequirementȱthatȱplaintiffsȱ demonstrateȱthatȱacȬ tualȱconsumerȱharmȱhasȱoccurred.149ȱȱ DespiteȱtheȱvigorousȱdebateȱoverȱtheȱappropriateȱlegalȱstanȬ dardsȱtoȱapplyȱinȱspecificȱSectionȱ2ȱcases,ȱaȱsensibleȱandȱcomȬ monȱ startingȱ placeȱ forȱ discussionȱ ofȱ modernȱ monopolizationȱ analysisȱ isȱ theȱ D.C.ȱ Circuitȇsȱ analysisȱ inȱ Microsoft.ȱ Inȱ theȱ moȬ nopolizationȱ context,ȱ theȱ D.C.ȱ Circuit’sȱ Microsoftȱ opinionȱ setsȱ forthȱ theȱ leadingȱ burdenȬshiftingȱ approachȱ forȱ distinguishingȱ exclusionaryȱ fromȱ competitiveȱ acts.150ȱ Theȱ plaintiff’sȱ initialȱ burdenȱisȱdescribedȱasȱfollows:ȱ [T]oȱbeȱcondemnedȱasȱexclusionary,ȱaȱmonopolist’sȱactȱmustȱ haveȱ anȱ ‘anticompetitiveȱ effect.’ȱ Thatȱ is,ȱ itȱ mustȱ harmȱ theȱ competitiveȱ processȱ andȱ therebyȱ harmȱ consumersȱ.ȱ.ȱ.ȱ[And]ȱ theȱ plaintiff,ȱ onȱ whomȱ theȱ burdenȱ ofȱ proofȱ ofȱ courseȱ rests,ȱ mustȱdemonstrateȱthatȱtheȱmonopolist’sȱconductȱindeedȱhasȱ theȱrequisiteȱanticompetitiveȱeffect.”151ȱ
Next,ȱ “[I]fȱ aȱ plaintiffȱ successfullyȱ establishesȱ aȱ primaȱ facieȱ caseȱ underȱ §ȱ2ȱ byȱ demonstratingȱ anticompetitiveȱ effect,ȱ thenȱ theȱ monopolistȱ mayȱ profferȱ aȱ [nonpretextual]ȱ ‘procompetitiveȱ justification’ȱ forȱ itsȱ conduct.”152ȱ Finally,ȱ “[I]fȱ theȱ monopolist’sȱ procompetitiveȱ justificationȱ standsȱ unrebutted,ȱ thenȱ theȱ plainȬ tiffȱmustȱdemonstrateȱthatȱtheȱanticompetitiveȱharmȱofȱtheȱconȬ ductȱoutweighsȱtheȱprocompetitiveȱbenefit.153ȱ Theȱkeyȱeconomicȱfunctionȱofȱtheȱplaintiff’sȱburdenȱtoȱdemȬ onstrateȱ actualȱ competitiveȱ harmȱ atȱ theȱ onsetȱ ofȱ litigationȱ is,ȱ consistentȱ withȱ theȱ errorȬcostȱ approachȱ describedȱ above,ȱ toȱ minimizeȱtheȱsocialȱcostsȱofȱantitrustȱenforcement,ȱand,ȱinȱparȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 148.ȱSeeȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ14–15.ȱ 149.ȱSeeȱ Benjaminȱ Klein,ȱ Exclusiveȱ Dealingȱ asȱ Competitionȱ “Onȱ theȱ Merits”,ȱ 12ȱ GEO.ȱMASONȱL.ȱREV.ȱ119ȱ(2003).ȱ 150.ȱUnitedȱStatesȱv.ȱMicrosoftȱCorp.,ȱ253ȱF.3dȱ34ȱ(D.C.ȱCir.ȱ2001)ȱ(enȱbanc)ȱ(perȱ curiam).ȱ 151.ȱId.ȱatȱ58–59ȱ(citationȱomitted).ȱ 152.ȱId.ȱatȱ59ȱ(citationȱomitted).ȱ 153.ȱId.ȱ
46ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
ticular,ȱ theȱ costsȱ associatedȱ withȱ falseȱ positives.ȱ Theȱ D.C.ȱ CirȬ cuitȱnotedȱtheȱdifficultyȱofȱthisȱtask:ȱ Whetherȱ anyȱ particularȱactȱ ofȱ aȱ monopolistȱisȱ exclusionary,ȱ ratherȱ thanȱ merelyȱ aȱ formȱ ofȱ vigorousȱ competition,ȱ canȱ beȱ difficultȱ toȱ discern:ȱ theȱ meansȱ ofȱ illicitȱ exclusion,ȱ likeȱ theȱ meansȱ ofȱ legitimateȱ competition,ȱareȱ myriad.ȱ Theȱ challengeȱ forȱanȱantitrustȱcourtȱliesȱinȱstatingȱaȱgeneralȱruleȱforȱdistinȬ guishingȱ betweenȱ exclusionaryȱ acts,ȱ whichȱ reduceȱ socialȱ welfare,ȱandȱcompetitiveȱacts,ȱwhichȱincreaseȱit.154ȱ
Withȱ thisȱ challengeȱ inȱ mind,ȱ courtsȱ haveȱ longȱ struggledȱ toȱ developȱ administrableȱ testsȱ that,ȱ atȱ aȱ minimum,ȱ identifyȱ imȬ plausibleȱclaims.ȱTheseȱscreens,ȱsuchȱasȱtheȱ“monopolyȱpower”ȱ requirement,ȱ filterȱ outȱ nonmeritoriousȱ claimsȱ whereȱ theȱ comȬ plainedȬofȱ conductȱ isȱ incapableȱ ofȱ harmingȱ theȱ competitiveȱ processȱ andȱ whereȱ aȱ findingȱ ofȱ liabilityȱ wouldȱ beȱ especiallyȱ likelyȱ toȱ chillȱ procompetitiveȱ businessȱ practices.ȱ Similarly,ȱ theȱ requirementȱthatȱplaintiffsȱsatisfyȱtheirȱprimaȱfacieȱburdenȱwithȱ evidenceȱofȱanticompetitiveȱeffectȱservesȱtheȱpurposesȱofȱreducȬ ingȱtheȱadministrativeȱcostsȱofȱlitigatingȱnonmeritoriousȱclaimsȱ andȱminimizingȱtheȱsocialȱcostsȱofȱerrors.ȱ Theȱmeritsȱofȱanyȱspecificȱapplicationȱofȱtheȱapproachȱdescribedȱ above,ȱofȱcourse,ȱlieȱinȱtheȱdetailsȱofȱitsȱexecution.ȱForȱexample,ȱtoȱ theȱextentȱthatȱevidenceȱofȱmereȱharmȱtoȱindividualȱcompetitorsȱisȱ sufficientȱ toȱ satisfyȱ theȱ plaintiff’sȱ primaȱ facieȱ burdenȱ ofȱ harmȱ toȱ competition,ȱ theȱ socialȱ valueȱ ofȱ theȱ requirementȱ willȱ beȱ diminȬ ishedȱandȱconsumersȱwillȱsuffer.ȱHarmȱisȱfurtherȱexacerbatedȱbyȱ theȱtendencyȱinȱantitrustȱcasesȱtoȱcondemnȱbusinessȱactivitiesȱforȱ whichȱthereȱisȱnoȱimmediateȱandȱintuitiveȱefficiencyȱexplanation.ȱ Forȱ theseȱ cases,ȱ minimizingȱ antitrustȱ errorȱ dependsȱ criticallyȱ onȱ ensuringȱ thatȱ theȱ evidenceȱ plaintiffsȱ areȱ requiredȱ toȱ profferȱ isȱ aȱ relativelyȱstrongȱsignalȱofȱharmȱtoȱcompetition.ȱ WhileȱMicrosoftȱsetsȱforthȱtheȱmodernȱburdenȬshiftingȱframeȬ workȱ forȱ monopolizationȱ claims,ȱ thereȱ areȱ otherȱ importantȱ sourcesȱ ofȱ Sectionȱ 2ȱ jurisprudence.ȱ Despiteȱ heatedȱ rhetoricȱ aboutȱtheȱideologicalȱnatureȱofȱmodernȱantitrust,ȱperhapsȱbestȱ capturedȱinȱtheȱeventsȱsurroundingȱtheȱwithdrawalȱofȱtheȱSecȬ tionȱ2ȱReport,ȱtheȱSupremeȱCourt’sȱantitrustȱjurisprudenceȱhasȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 154.ȱId.ȱatȱ58.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
47ȱ
exhibitedȱanȱimpressiveȱdegreeȱofȱconsensus.155ȱIndeed,ȱtheȱSuȬ premeȱ Court’sȱ consensusȱ withinȱ antitrustȱ jurisprudenceȱ isȱ strongestȱwhenȱoneȱanalyzesȱSectionȱ2ȱspecificallyȱandȱinȱisolaȬ tion.ȱConsiderȱthatȱsinceȱNYNEXȱCorp.ȱv.ȱDiscon,ȱInc.,156ȱallȱfourȱ ofȱtheȱSupremeȱCourt’sȱdecisionsȱaddressingȱclaimsȱunderȱSecȬ tionȱ 2ȱ andȱ settingȱ forthȱ theȱ relevantȱ principlesȱ haveȱ beenȱ deȬ cidedȱ unanimously.157ȱ Inȱ theseȱ recentȱ decisions,ȱ theȱ Supremeȱ Courtȱhasȱarticulatedȱaȱnumberȱofȱfirstȱprinciplesȱthatȱguideȱtheȱ Court’sȱdecisions,ȱandȱinformȱourȱmonopolizationȱanalysis:ȱ x
MereȱpossessionȱofȱmonopolyȱpowerȱisȱnotȱanȱantitrustȱofȬ fense.ȱTheȱSupremeȱCourt’sȱdecisionȱinȱTrinkoȱrepreȬ sentsȱtheȱmostȱpowerfulȱarticulationȱofȱthisȱprincipleȱ ofȱmodernȱantitrust.ȱTheȱunanimousȱCourtȱnotedȱ thatȱtheȱprospectsȱofȱmonopolyȱprofitsȱareȱwhatȱ“atȬ tractsȱ‘businessȱacumen’ȱinȱtheȱfirstȱplace;ȱitȱinducesȱ riskȱtakingȱthatȱproducesȱinnovationȱandȱeconomicȱ growth.”158ȱAntitrustȱcommentatorsȱhaveȱalsoȱinȬ creasinglyȱrecognizedȱthatȱaȱsignatureȱfeatureȱofȱ U.S.ȱmonopolizationȱpolicyȱisȱitsȱunderstandingȱofȱ theȱtradeoffsȱbetweenȱinnovationȱandȱdynamicȱeffiȬ ciencyȱgainsȱandȱtheȱstaticȱwelfareȱlossesȱassociatedȱ withȱmonopolyȱpower.159ȱShermanȱActȱmonopolizaȬ tionȱjurisprudence,ȱtherefore,ȱclearlyȱendorsesȱantiȬ trustȱrulesȱthatȱprotectȱtheȱcompetitiveȱprocessȱbutȱ doȱnotȱpunishȱsuccess,ȱrequireȱfirmsȱtoȱpullȱtheirȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 155.ȱLeahȱ Brannonȱ &ȱ Douglasȱ H.ȱ Ginsburg,ȱ Antitrustȱ Decisionsȱ ofȱ theȱ U.S.ȱ SuȬ premeȱCourt,ȱ1967ȱtoȱ2007,ȱ3ȱCOMPETITIONȱ POL’Yȱ INT’Lȱ3ȱ(2007).ȱBrannonȱandȱGinsȬ burg,ȱforȱexample,ȱfindȱthatȱfromȱ1997ȱtoȱ2006,ȱeightyȬfiveȱpercentȱofȱallȱantitrustȱ decisionsȱ wereȱ decidedȱ byȱ aȱ supermajorityȱ marginȱ (andȱ eachȱ inȱ favorȱ ofȱ theȱ deȬ fendant,ȱ althoughȱ thisȱ isȱ fairlyȱ predictableȱ givenȱ theȱ proplaintiffȱ natureȱ ofȱ 1960sȱ antitrustȱ jurisprudence).ȱ Id.ȱ atȱ 20.ȱ Whenȱ oneȱ considersȱ theȱ Supremeȱ Courtȱ deciȬ sionsȱduringȱtheȱBushȱadministration,ȱforȱexample,ȱtheȱaggregateȱvoteȱcountȱisȱ86Ȭ 9ȱwithȱsevenȱofȱelevenȱopinionsȱgeneratingȱunanimousȱagreement.ȱ 156.ȱ525ȱU.S.ȱ128ȱ(1998).ȱ 157.ȱSeeȱPac.ȱBellȱTel.ȱCo.ȱv.ȱLinklineȱCmmcȇns,ȱInc.,ȱ129ȱS.ȱCt.ȱ1109ȱ(2009);ȱWeyȬ erhaeuserȱCo.ȱv.ȱRossȬSimmonsȱHardwoodȱLumberȱCo.,ȱ549ȱU.S.ȱ312ȱ(2007);ȱVeriȬ zonȱ Commc’nsȱ v.ȱ Lawȱ Officesȱ ofȱ Curtisȱ V.ȱ Trinko,ȱ 540ȱ U.S.ȱ 398ȱ (2004);ȱ NYNEX,ȱ 525ȱU.S.ȱ128.ȱForȱanȱanalysisȱofȱtheȱRobertsȱCourtȱantitrustȱdecisions,ȱseeȱJoshuaȱ D.ȱWright,ȱTheȱRobertsȱCourtȱandȱtheȱChicagoȱSchoolȱofȱAntitrust:ȱTheȱ2006ȱTermȱandȱ Beyond,ȱ 3ȱ COMPETITIONȱ POL’Yȱ INT’Lȱ 25ȱ (2007)ȱ (arguingȱ thatȱ Chicagoȱ Schoolȱ ecoȬ nomicȱprinciplesȱcharacterizeȱtheȱRobertsȱCourtȱantitrustȱjurisprudence).ȱ 158.ȱTrinko,ȱ540ȱU.S.ȱatȱ407.ȱ 159.ȱSeeȱDavidȱS.ȱEvansȱ&ȱKeithȱN.ȱHylton,ȱTheȱLawfulȱAcquisitionȱandȱExerciseȱofȱ Monopolyȱ Powerȱ andȱ itsȱ Implicationsȱ forȱ theȱ Objectivesȱ ofȱ Antitrust,ȱ 4ȱ COMPETITIONȱ POL’YȱINT’Lȱ203,ȱ203ȱ(2008).ȱ
48ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
competitiveȱpunches,ȱorȱdemandȱthatȱfirmsȱrollȱoverȱ onceȱtheyȱhaveȱlawfullyȱachievedȱmonopolyȱpower.ȱ Instead,ȱtheȱantitrustȱlawsȱcondemnȱonlyȱspecificȱ actsȱthatȱresultȱinȱtheȱimproperȱacquisitionȱofȱorȱ maintenanceȱofȱmonopolyȱpowerȱandȱthatȱharmȱtheȱ competitiveȱprocess.ȱ x
Theȱmereȱexerciseȱofȱlawfulȱmonopolyȱpowerȱinȱtheȱformȱ ofȱhigherȱpricesȱisȱnotȱanȱantitrustȱviolation.ȱAȱcorollaryȱ ofȱtheȱpreviousȱprinciple,ȱtheȱSupremeȱCourtȱhasȱreȬ peatedlyȱrecognizedȱthatȱaȱmonopolistȱisȱentitledȱ underȱtheȱShermanȱActȱtoȱreapȱtheȱrewardsȱofȱitsȱinȬ novation.ȱTheȱsuccessfulȱmonopolistȱfirmȱisȱentitledȱ toȱchargeȱwhateverȱpriceȱtheȱmarketȱwillȱbear.ȱAȱ contraryȱfinding,ȱlimitingȱtheȱreturnsȱtoȱsuccessfullyȱ competingȱinȱtheȱmarketplace,ȱisȱlogicallyȱinconsisȬ tentȱwithȱaȱcompetitionȱpolicyȱregimeȱdesignedȱtoȱ fosterȱinnovationȱandȱeconomicȱgrowth.160ȱ
x
CourtsȱmustȱbeȱconcernedȱwithȱtheȱsocialȱcostsȱofȱantiȬ trustȱerrors,ȱandȱtheȱerrorȬcostȱframeworkȱisȱaȱdesirableȱ approachȱtoȱdevelopingȱstandardsȱwhichȱincorporateȱtheseȱ concerns.ȱTheȱfundamentalȱandȱvexingȱrealitiesȱofȱ Sectionȱ2ȱenforcementȱareȱthatȱfirst,ȱitȱisȱbothȱexceedȬ inglyȱdifficultȱtoȱreliablyȱidentifyȱanticompetitiveȱ conduct,ȱandȱsecond,ȱthatȱerrorsȱareȱlikelyȱtoȱharmȱ theȱintendedȱbeneficiariesȱofȱtheȱantitrustȱlaws.ȱThisȱ taskȱisȱeasierȱsaidȱthanȱdone.161ȱTheȱSupremeȱCourtȱ hasȱconsistentlyȱandȱrepeatedlyȱexpressedȱitsȱconȬ cernȱwithȱantitrustȱerrors,ȱespeciallyȱfalseȱpositivesȱ whichȱareȱlikelyȱtoȱbeȱmoreȱfrequentȱandȱmoreȱ costlyȱthanȱfalseȱnegatives.162ȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 160.ȱSee,ȱe.g.,ȱLinkline,ȱ129ȱS.ȱCt.ȱatȱ1118;ȱTrinko,ȱ540ȱU.S.ȱatȱ407–08;ȱNYNEX,ȱ525ȱ U.S.ȱatȱ136–37.ȱ 161.ȱSeeȱ Frankȱ H.ȱ Easterbrook,ȱ Whenȱ Isȱ Itȱ Worthwhileȱ toȱ Useȱ Courtsȱ toȱ Searchȱ forȱ ExclusionaryȱConduct?,ȱ2003ȱCOLUM.ȱ BUS.ȱ L.ȱ REV.ȱ345,ȱ345ȱ(“Aggressive,ȱcompetiȬ tiveȱconductȱbyȱanyȱfirm,ȱevenȱoneȱwithȱmarketȱpower,ȱisȱbeneficialȱtoȱconsumers.ȱ Courtsȱshouldȱprizeȱandȱencourageȱit.ȱAggressive,ȱexclusionaryȱconductȱisȱdeleteȬ riousȱ toȱ consumers,ȱ andȱ courtsȱ shouldȱ condemnȱ it.ȱ Theȱ bigȱ problemȱ liesȱ inȱ this:ȱ competitiveȱandȱexclusionaryȱconductȱlookȱalike.”).ȱȱ 162.ȱSee,ȱe.g.,ȱLinkline,ȱ129ȱS.ȱCt.ȱatȱ1120;ȱTrinko,ȱ540ȱU.S.ȱatȱ414ȱ(“Theȱcostȱofȱfalseȱ positivesȱcounselsȱagainstȱanȱundueȱexpansionȱofȱ§ȱ2ȱliability.”);ȱNYNEX,ȱ525ȱU.S.ȱ atȱ 136–37;ȱ Spectrumȱ Sports,ȱ Inc.ȱ v.ȱ McQuillan,ȱ 506ȱ U.S.ȱ 447,ȱ 456,ȱ 458ȱ (1993)ȱ (“[T]hisȱ Courtȱ andȱ otherȱ courtsȱ haveȱ beenȱ carefulȱ toȱ avoidȱ constructionsȱ ofȱ §ȱȱ 2ȱ whichȱmightȱchillȱcompetition,ȱratherȱthanȱfosterȱit.”);ȱBus.ȱElecs.ȱCorp.ȱv.ȱSharpȱ Elecs.ȱCorp.,ȱ485ȱU.S.ȱ717,ȱ728ȱ(1988);ȱMatsushitaȱElec.ȱIndus.ȱCo.ȱv.ȱZenithȱRadioȱ
ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
49ȱ
Inȱ reflectionȱ ofȱ theseȱ principles,ȱ theȱ errorȬcostȱ approachȱ hasȱ becomeȱaȱmainstreamȱandȱwellȬacceptedȱapproachȱtoȱevaluateȱ antitrustȱstandardsȱandȱpolicyȱdecisions.163ȱTheȱSupremeȱCourtȱ hasȱadoptedȱanȱerrorȬcostȱapproachȱinȱatȱleastȱtheȱfollowingȱdeȬ cisions:ȱNYNEXȱCorp.ȱv.ȱDiscon,ȱInc.,164ȱStateȱOilȱv.ȱKhan,165ȱBrookeȱ Groupȱ v.ȱ Brownȱ &ȱ Williamsonȱ Tobaccoȱ Corp.,166ȱ Leegin,167ȱ WeyerȬ haeuser,168ȱTrinko,169ȱCreditȱSuisse,170ȱandȱLinkline.171ȱJusticeȱScaliaȇsȱ articulationȱofȱtheȱCourtȇsȱconcernsȱinȱTrinkoȱisȱinstructive:ȱ Againstȱtheȱslightȱbenefitsȱofȱantitrustȱinterventionȱhere,ȱweȱ mustȱ weighȱ aȱ realisticȱ assessmentȱ ofȱ itsȱ costsȱ.ȱ.ȱ.ȱ.ȱ Mistakenȱ inferencesȱandȱtheȱresultingȱfalseȱcondemnationsȱ“areȱespeȬ ciallyȱcostly,ȱbecauseȱtheyȱchillȱtheȱveryȱconductȱtheȱantitrustȱ lawsȱ areȱ designedȱ toȱ protect.”ȱ Theȱ costȱ ofȱ falseȱ positivesȱ counselsȱagainstȱanȱundueȱexpansionȱofȱ§ȱ2ȱliability.172ȱ
Withȱtheȱgeneralȱmonopolizationȱlandscapeȱandȱfirstȱprinciplesȱ inȱhandȱtoȱprovideȱtheȱlensȱforȱanyȱspecificȱapplicationȱofȱSecȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ Corp.,ȱ 475ȱ U.S.ȱ 574,ȱ 594ȱ (1986)ȱ (statingȱ thatȱ mistakenȱ inferencesȱ inȱ predatoryȬ pricingȱ casesȱ “areȱ especiallyȱ costlyȱ becauseȱ theyȱ chillȱ theȱ veryȱ conductȱ theȱ antiȬ trustȱ lawsȱ areȱ designedȱ toȱ protect”);ȱ Copperweldȱ Corp.ȱ v.ȱ Independenceȱ Tubeȱ Corp.,ȱ467ȱU.S.ȱ752,ȱ767–68ȱ(1984)ȱ(notingȱthatȱscrutinyȱofȱsingleȱfirmsȱunderȱtheȱ ShermanȱActȱisȱappropriateȱonlyȱwhenȱtheyȱposeȱaȱdangerȱofȱmonopolization,ȱanȱ approachȱthatȱ“reducesȱtheȱriskȱthatȱtheȱantitrustȱlawsȱwillȱdampenȱtheȱcompetiȬ tiveȱzealȱofȱaȱsingleȱaggressiveȱ[competitor]”).ȱ 163.ȱSee,ȱ e.g.,ȱ POSNER,ȱ supraȱ noteȱ 10,ȱ atȱ ixȱ (“Almostȱ everyoneȱ professionallyȱ inȬ volvedȱinȱantitrustȱtoday”ȱagreesȱthatȱ“theȱdesignȱofȱantitrustȱrulesȱshouldȱtakeȱintoȱ accountȱ theȱ costsȱ andȱ benefitsȱ ofȱ individualizedȱ assessmentȱ ofȱ challengedȱ pracȬ ticesȱ.ȱ.ȱ.ȱ.”).ȱ Forȱ aȱ generalȱ discussionȱ ofȱ anȱ errorȱ costȱ approachȱ toȱ antitrust,ȱ seeȱ Manneȱ&ȱWright,ȱsupraȱnoteȱ10,ȱatȱ6.ȱForȱspecificȱapplicationsȱofȱtheȱerrorȬcostȱapȬ proachȱ toȱ variousȱ antitrustȱ topics,ȱ seeȱ Becknerȱ IIIȱ &ȱ Salop,ȱ supraȱ noteȱ 10;ȱ Evansȱ &ȱ Padilla,ȱsupraȱnoteȱ10;ȱFroebȱetȱal.,ȱsupraȱnoteȱ10;ȱHyltonȱ&ȱSalinger,ȱsupraȱnoteȱ10.ȱ 164.ȱ525ȱU.S.ȱatȱ133ȱ(1998)ȱ(“[C]ertainȱkindsȱofȱagreementsȱwillȱsoȱoftenȱproveȱsoȱ harmfulȱtoȱcompetitionȱandȱsoȱrarelyȱproveȱjustifiedȱthatȱtheȱantitrustȱlawsȱdoȱnotȱ requireȱproofȱthatȱanȱagreementȱofȱthatȱkindȱis,ȱinȱfact,ȱanticompetitiveȱinȱtheȱparȬ ticularȱcircumstances.”).ȱ 165.ȱ522ȱU.S.ȱ3,ȱ10ȱ(1997)ȱ(Certainȱ“typesȱofȱrestraintsȱ.ȱ.ȱ.ȱhaveȱsuchȱpredictableȱ andȱperniciousȱanticompetitiveȱeffect,ȱandȱsuchȱlimitedȱpotentialȱforȱprocompetiȬ tiveȱbenefit,ȱthatȱtheyȱareȱdeemedȱunlawfulȱperȱse.”).ȱ 166.ȱ509ȱU.S.ȱ209,ȱ223–24ȱ(1993).ȱ 167.ȱLeeginȱCreativeȱLeatherȱProds.,ȱInc.ȱv.ȱPSKS,ȱInc.,ȱ551ȱU.S.ȱ877,ȱ886ȱ(2007).ȱ 168.ȱWeyerhaeuserȱCo.ȱv.ȱRossȬSimmonsȱHardwoodȱLumberȱCo.,ȱ549ȱU.S.ȱ312,ȱ 319–20ȱ(2007).ȱȱ 169.ȱ540ȱU.S.ȱ398,ȱ414ȱ(2004).ȱ 170.ȱCreditȱSuisseȱSec.ȱ(USA)ȱLLCȱv.ȱBilling,ȱ551ȱU.S.ȱ264,ȱ283ȱ(2007).ȱ 171.ȱ129ȱS.ȱCt.ȱ1109,ȱ1122ȱ(2009).ȱ 172.ȱTrinko,ȱ540ȱatȱ414ȱ(citationȱomitted)ȱ(quotingȱMatsushitaȱElec.ȱIndus.ȱCo.ȱv.ȱ ZenithȱRadioȱCorp.,ȱ475ȱU.S.ȱ574,ȱ594ȱ(1986)).ȱ
50ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
tionȱ 2ȱ law,ȱ weȱ turnȱ toȱ aȱ moreȱ detailedȱ discussionȱ ofȱ theȱ twoȱ elementsȱofȱaȱpotentialȱmonopolizationȱcaseȱ(monopolyȱpowerȱ andȱexclusionaryȱconduct)ȱandȱtheirȱapplicationȱtoȱGoogle.ȱ B.
MonopolyȱPowerȱ
Monopolyȱ powerȱ isȱ theȱ firstȱ elementȱ ofȱ theȱ monopolizationȱ offenseȱ andȱ refersȱ toȱ theȱ “powerȱ toȱ controlȱ pricesȱ orȱ excludeȱ competition.”173ȱAsȱanȱantitrustȱconcept,ȱmonopolyȱpowerȱmustȱ beȱdistinguishedȱfromȱtheȱtypeȱofȱeconomicȱmarketȱpowerȱthatȱ refersȱ merelyȱ toȱ theȱ abilityȱ toȱ haveȱ someȱ discretionȱ overȱ oneȇsȱ ownȱ priceȱ withoutȱ losingȱ allȱ sales.ȱ Althoughȱ marketȱ powerȱ inȱ thisȱ senseȱ isȱ ubiquitousȱ inȱ theȱ modernȱ economy,ȱ monopolyȱ powerȱ ofȱ theȱ typeȱ requiredȱ toȱ establishȱ aȱ Sectionȱ 2ȱ violationȱ impliesȱ theȱ powerȱ toȱ controlȱ eitherȱ marketȱ pricesȱ orȱ output.ȱ Further,ȱthisȱpowerȱmustȱbeȱdurableȱratherȱthanȱtransitory.ȱȱ AppliedȱtoȱaȱpotentialȱmonopolizationȱcaseȱagainstȱGoogle,ȱaȱ monopolyȱ powerȱ inquiryȱ raisesȱ severalȱ issues.ȱ Theȱ firstȱ isȱ thatȱ theȱmarketȱdefinitionȱinquiryȱplaysȱaȱcentralȱroleȱinȱdiscipliningȱ anyȱ monopolyȱ powerȱ analysis.ȱ Thus,ȱ inȱ assessingȱ aȱ claimȱ ofȱ aȱ Sectionȱ 2ȱ violation,ȱ carefulȱ considerationȱ ofȱ theȱ potentiallyȱ relevantȱ marketsȱ inȱ whichȱ anticompetitiveȱ conductȱ mightȱ haveȱ occurredȱ isȱ aȱ must.ȱ Theȱ secondȱ isȱ that,ȱ asȱ withȱ anyȱ modernȱ marketȱ definitionȱ analysisȱ involvingȱ webȬbasedȱ productsȱ andȱ services,ȱoneȱmustȱconsiderȱwhetherȱnetworkȱeffectsȱareȱrelevantȱ toȱtheȱmonopolyȱpowerȱanalysis,ȱand,ȱifȱso,ȱtoȱwhatȱextent.ȱ C.
MarketȱDefinitionȱandȱMonopolyȱPowerȱ
Seriousȱ marketȱ definitionȱ problemsȱ ariseȱ inȱ Google’sȱ case— notȱ surprisingly,ȱ differentȱ marketȱ definitionsȱ translateȱ toȱ veryȱ differentȱconceptionsȱaboutȱtheȱlevelȱofȱmonopolyȱpowerȱexistȬ ingȱinȱGoogleȇsȱmarketsȱandȱhaveȱdifferingȱimplicationsȱforȱSecȬ tionȱ2’sȱmonopolizationȱanalysis.ȱȱ AsȱdiscussedȱinȱPartȱIII.B,ȱGoogle’sȱmarketȱisȱfarȱmoreȱcomȬ plicatedȱ thanȱ isȱ commonlyȱ assumed.ȱ Googleȱ sellsȱ advertising,ȱ firstȱ andȱ foremost,ȱ andȱ itȱ givesȱ awayȱ severalȱ otherȱ products,ȱ includingȱ searchȱ results.ȱ Ifȱ theȱ relevantȱ advertisingȱ marketȱ inȬ cludesȱ allȱ advertisingȱ acrossȱ media,ȱ Googleȱ hasȱ aȱ minisculeȱ marketȱ shareȱ andȱ essentiallyȱ noȱ marketȱ power.ȱ Ifȱ theȱ relevantȱ advertisingȱ marketȱ includesȱ onlyȱ onlineȱ advertising,ȱ Googleȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 173.ȱUnitedȱStatesȱv.ȱE.I.ȱduȱPontȱdeȱNemoursȱ&ȱCo,ȱ351ȱU.S.ȱ377,ȱ391ȱ(1956).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
51ȱ
stillȱhasȱaȱrelativelyȱsmallȱshareȱofȱtheȱmarket.ȱOnlyȱwhenȱdifȬ ferentȱ typesȱ ofȱ onlineȱ advertisingȱ (suchȱ asȱ searchȱ adsȱ versusȱ contextualȱdisplayȱadsȱversusȱbehavioralȱdisplayȱads)ȱareȱsepaȬ ratedȱ intoȱ differentȱ marketsȱ doesȱ Google’sȱ marketȱ shareȱ growȱ substantiallyȱinȱpaidȬsearchȱadvertising.ȱ Someȱ careȱ isȱ requiredȱ inȱ evenȱ limitingȱ ourȱ attentionȱ toȱ theȱ paidȱ searchȱ advertisingȱ market.ȱ Mostȱ casualȱ discussionsȱ ofȱ Google’sȱ marketȱ shareȱ referenceȱ itsȱ shareȱ ofȱ theȱ searchȱ market.ȱ Althoughȱ theȱ sizeȱ ofȱ Google’sȱ searchȱ marketȱ isȱ relevantȱ toȱ asȬ sessingȱitsȱsignificanceȱinȱtheȱsearchȱadvertisingȱmarket,ȱtheȱtwoȱ areȱnotȱtheȱ same.ȱThus,ȱclaimsȱthatȱ “Googleȱhasȱ70%ȱofȱtheȱ USȱ searchȱmarket”ȱmayȱbeȱtrue,174ȱbutȱareȱnotȱclearlyȱrelevantȱtoȱtheȱ questionȱofȱwhetherȱGoogleȱhasȱmonopolyȱpowerȱinȱtheȱsearchȱ advertisingȱmarket,ȱwhereȱthisȱfigureȱisȱmerelyȱaȱmeasureȱofȱtheȱ numberȱ ofȱ searchesȱ performedȱ onȱ theȱ majorȱ generalȱ searchȱ enȬ ginesȱ byȱ endȱ usersȱ inȱ theȱ Unitedȱ States.ȱ Otherȱ measuresȱ assessȱ theȱshareȱofȱimpressionsȱandȱclicksȱonȱsearchȱadsȱthatȱareȱservedȱ byȱ Google.ȱ Thisȱ isȱ alsoȱ relevantȱ butȱ incomplete,ȱ asȱ itȱ doesȱ notȱ addressȱ theȱ shareȱ ofȱ advertisersȱ orȱ advertiserȱ dollarsȱ repreȬ sented.ȱNoneȱofȱthisȱisȱtoȱsayȱthatȱGoogleȱdoesȱnotȱhaveȱaȱlargeȱ shareȱofȱtheȱsearchȱadvertisingȱmarket,ȱifȱsuchȱaȱmarketȱexistsȱatȱ all,ȱbutȱthatȱitsȱspecificȱshareȱandȱitsȱmarketȱpowerȱareȱfarȱmoreȱ difficultȱtoȱcalculateȱthanȱtypicallyȱpresented.175ȱȱ InȱitsȱreviewȱofȱtheȱGoogleȱandȱDoubleClickȱmerger,ȱtheȱFTCȱ foundȱ thatȱ searchȱ advertisingȱ andȱ displayȱ advertisingȱ wereȱ inȱ distinctȱmarkets,ȱandȱthatȱonlineȱadvertisingȱandȱadvertisingȱinȱ otherȱmediaȱwereȱinȱdistinctȱmarkets.176ȱWhileȱweȱhaveȱnotȱunȬ dertakenȱaȱsubstantialȱassessmentȱofȱthisȱparsingȱofȱtheȱproductȱ markets,ȱitȱisȱconcerningȱisȱthatȱtheȱFTC,ȱinȱmakingȱitsȱdecisionȱ itself,ȱdidȱnotȱappearȱtoȱundertakeȱaȱparticularlyȱcarefulȱanalysisȱ ofȱtheȱmarketȱdefinitionȱproblemȱpeculiarȱtoȱtwoȬsidedȱmarkets.ȱ ThisȱcasualȱapproachȱtoȱanalysisȱleadsȱtoȱaȱconflationȱofȱmarȬ ketȱpresenceȱandȱmarketȱpowerȱandȱfrequentlyȱmarsȱdiscussionsȱ ofȱGoogle’sȱpositionȱinȱanyȱmarket.ȱAȱsensibleȱstartingȱpoint,ȱalȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 174.ȱStephenȱShankland,ȱGoogle’sȱU.S.ȱSearchȱShareȱNearsȱ70ȱPercent,ȱCNETȱNEWS,ȱ Julyȱ15,ȱ2008,ȱhttp://news.cnet.com/8301Ȭ1023_3Ȭ9991866Ȭ93.html.ȱ 175.ȱForȱ aȱ discussionȱ ofȱ theȱ substantialȱ complexitiesȱ inȱ determiningȱ relevantȱ marketsȱandȱmarketȱshareȱinȱtwoȬsidedȱmarkets,ȱseeȱDavidȱS.ȱEvansȱ&ȱMichaelȱD.ȱ Noel,ȱ Definingȱ Marketsȱ thatȱ Involveȱ MultiȬSidedȱ Platformȱ Businesses:ȱ Anȱ Empiricalȱ Frameworkȱ withȱ anȱ Applicationȱ toȱ Google’sȱ Purchaseȱ ofȱ DoubleClickȱ (Nov.ȱ 6,ȱ 2007),ȱ http://ssrn.com/abstract=1027933.ȱ 176.ȱFed.ȱTradeȱComm’n,ȱsupraȱnoteȱ94,ȱatȱ3–7.ȱ
52ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
thoughȱnotȱcomplete,ȱwouldȱbeȱtoȱassessȱmarketȱdefinitionȱ(andȱ competitiveȱeffects)ȱonȱtheȱsideȱofȱaȱplatform’sȱmarketȱwhereȱitȱ chargesȱ pricesȱ aboveȱ marginalȱ costȱ (whereȱ itȱ isȱ customaryȱ inȱ twoȬsidedȱplatformsȱforȱtheȱplatformȱtoȱattractȱusersȱtoȱoneȱsideȱ withȱbelowȬcostȱorȱzeroȬcostȱpricingȱandȱtoȱearnȱrevenueȱonȱtheȱ otherȱsideȱofȱtheȱmarket),ȱparticularlyȱasȱinȱGoogle’sȱcaseȱwhereȱ itȱchargesȱaȱzeroȱpriceȱtoȱendȱusersȱofȱitsȱsearchȱplatform.177ȱButȱ anȱassessmentȱofȱaȱplatform’sȱabilityȱtoȱraiseȱpricesȱonȱoneȱsideȱ ofȱ itsȱ platformȱ withoutȱ considerationȱ ofȱ whatȱ happensȱ toȱ theȱ otherȱ sideȱ (andȱ itsȱ feedbackȱ effectȱ onȱ theȱ firstȱ side)ȱ isȱ incomȬ plete.178ȱAgain,ȱthisȱisȱnotȱtoȱsayȱthatȱaȱnarrowȱ“onlineȱsearchȱadȬ vertising”ȱ marketȱ definitionȱ isȱ inappropriateȱ toȱ assessȱ Google’sȱ marketȱpower,ȱbutȱthatȱtheȱdeterminationȱisȱcomplexȱandȱerrorȱ prone—andȱessentialȱtoȱallȱanalysisȱthatȱfollows.ȱȱ Theȱmoreȱimportantȱquestionȱisȱhowȱultimatelyȱtoȱdetermineȱ theȱmarketȱwithinȱwhichȱGoogle’sȱactivitiesȱshouldȱbeȱassessed.ȱ Theȱ riskȱ isȱ thatȱ aȱ marketȱ determinationȱ madeȱ onȱ theȱ basisȱ ofȱ commonȱ senseȱ andȱ corporateȱ documents—particularlyȱ thoseȱ thatȱ equateȱ channelsȱ ofȱ distributionȱ withȱ markets,179ȱ mayȱ draȬ maticallyȱ overstateȱ Google’sȱ powerȱ toȱ influenceȱ advertisingȱ prices.ȱ Advertisingȱ isȱ aimedȱ atȱ loweringȱ searchȱ costs180—ofȱ buyersȱforȱsellers,ȱofȱsellersȱforȱbuyers,ȱandȱofȱbuyersȱforȱprices.ȱ Whetherȱ thisȱ isȱ doneȱ byȱ bolsteringȱ brandȱ recognitionȱ forȱ purȬ posesȱ ofȱ facilitatingȱ orȱ encouragingȱ futureȱ purchases,ȱ orȱ byȱ providingȱ aȱ readyȱ outletȱ forȱ aȱ consumerȱ lookingȱ toȱ makeȱ anȱ immediateȱ purchase,ȱ itȱ isȱ hardȱ toȱ seeȱ theseȱ asȱ ultimatelyȱ disȬ tinctȱ functions.ȱ Yetȱ theȱ latterȱ optionȱ hasȱ beenȱ madeȱ widelyȱ availableȱandȱimmenselyȱeffectiveȱthroughȱonlineȱsearchȱadverȬ tising,ȱandȱitȱtakesȱonȱaȱdistinctiveȱcast.ȱȱ Inȱreality,ȱallȱformsȱofȱadvertising—andȱrelatedȱendeavorsȱlikeȱ storeȱplacementȱandȱdesign—areȱaboutȱbringingȱbuyersȱandȱsellȬ ersȱtogetherȱbyȱminimizingȱsomeȱofȱtheȱtransactionȱcostsȱthatȱothȬ erwiseȱkeepȱthemȱapart.ȱGivenȱaȱconsistentȱfunctionȱforȱdifferentȱ channelsȱ ofȱ distribution,ȱ theȱ burdenȱ isȱ onȱ thoseȱ propoundingȱ aȱ distinctȱ economicȱ relevanceȱ forȱ eachȱ channelȱ ofȱ distributionȱ toȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 177.ȱSeeȱDavidȱS.ȱEvans,ȱTwoȬSidedȱMarketȱDefinition,ȱinȱMARKETȱ DEFINITIONȱ INȱ ANTITRUST:ȱ THEORYȱ ANDȱ CASEȱ STUDIES,ȱ ABAȱ SECTIONȱ OFȱ ANTITRUSTȱ LAWȱ 7ȱ (Forthcomingȱ2010).ȱ 178.ȱId.ȱatȱ10.ȱ 179.ȱSeeȱManneȱ&ȱWilliamson,ȱsupraȱnoteȱ121,ȱatȱ612–13.ȱ 180.ȱStigler,ȱsupraȱnoteȱ69,ȱatȱ224.ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
53ȱ
demonstrateȱtheȱproposedȱdistinctionȱwithȱeconomicȱevidence.ȱInȱ theȱ absenceȱofȱ suchȱ directȱ evidence,ȱtheȱ monopolyȱ powerȱ deterȬ minationȱ oftenȱ turnsȱ onȱ inferencesȱ drawnȱ fromȱ marketȱ shares.ȱ Suchȱ inferencesȱ areȱ notȱ uncommonȱ inȱ antitrustȱ analysis,ȱ andȱ Google’sȱclaimedȱmarketȱsharesȱareȱcertainlyȱnotȱoutȱofȱlineȱwithȱ theȱsharesȱthatȱhaveȱgivenȱriseȱtoȱtheseȱpresumptions.181ȱ D.
TheȱQuestionȱofȱNetworkȱEffectsȱ
Theȱroleȱofȱnetworkȱeffectsȱinȱtheȱ“NewȱEconomy”ȱgenerallyȱ raisesȱ aȱ hostȱ ofȱ questionsȱ forȱ antitrustȱ enforcement,ȱ especiallyȱ inȱestablishingȱtheȱdurableȱmonopolyȱpowerȱrequiredȱtoȱproveȱ aȱ Sectionȱ 2ȱ violation.ȱ Weȱ nowȱ turnȱ toȱ consideringȱ theȱ roleȱ ofȱ networkȱ effectsȱ inȱ Google’sȱ productȱ markets,ȱ asȱ wellȱ asȱ whetherȱ andȱ howȱ anyȱ networkȱ externalitiesȱ implicateȱ anȱ antiȬ trustȱinterventionȱagainstȱGoogle.ȱ Considerȱ firstȱ theȱ caseȱ ofȱ aȱ networkȱ withȱ soȬcalledȱ “direct”ȱ networkȱeffects.ȱInȱsuchȱaȱcase,ȱaȱuser’sȱparticipationȱinȱtheȱsysȬ temȱ confersȱ aȱ soȬcalledȱ networkȱ benefitȱ onȱ otherȱ users,ȱ unȬ capturedȱ byȱ theȱ priceȱ theȱ userȱ paysȱ toȱ accessȱ theȱ system.ȱ Inȱ contrast,ȱ indirectȱ networkȱ effectsȱ areȱ fullyȱ internalizedȱ byȱ theȱ system,ȱasȱtheȱpriceȱtheȱuserȱpaysȱsimplyȱreflectsȱtheȱincreasedȱ valueȱofȱtheȱplatformȱfromȱhavingȱmoreȱusersȱonȱtheȱotherȱsideȱ ofȱtheȱtwoȬsidedȱplatform.ȱȱ TheȱconsequenceȱofȱparticipationȱinȱaȱnetworkȱwithȱdirectȱnetȬ workȱeffectsȱratherȱthanȱindirectȱisȱdramaticȱinȱtermsȱofȱtheȱabilityȱ ofȱfirmsȱtoȱcompeteȱwithȱGoogle.ȱInȱtheȱindirectȱcase,ȱaȱcompetingȱ adȱ platformȱ withȱ aȱ somewhatȱ smallerȱ networkȱ ofȱ endȱ usersȱ (searchers)ȱwould,ȱifȱtheȱvalueȱofȱtheȱadvertisementȱisȱdependentȱ onȱ theȱ sizeȱ ofȱ theȱ networkȱ ofȱ endȱ users,ȱ simplyȱ receiveȱ aȱ lowerȱ priceȱforȱitsȱproduct.ȱTheȱdifferenceȱinȱqualityȱattributableȱtoȱtheȱ endȬuserȱnetworkȱsizeȱwouldȱbeȱreflectedȱinȱtheȱprice,ȱandȱadverȬ tisersȱwouldȱhaveȱtheȱsameȱmarginalȱincentiveȱtoȱadvertiseȱonȱeiȬ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 181.ȱSee,ȱ e.g.,ȱ Eastmanȱ Kodakȱ Co.ȱ v.ȱ Imageȱ Technicalȱ Servs.,ȱ Inc.,ȱ 504ȱ U.S.ȱ 451,ȱ 481ȱ (1992)ȱ (findingȱ eightyȱ toȱ ninetyȬfiveȱ percentȱ predominant);ȱ Unitedȱ Statesȱ v.ȱ Grinnellȱ Corp,ȱ 384ȱ U.S.ȱ 563,ȱ 571ȱ (1966)ȱ (findingȱ eightyȬsevenȱ percentȱ predomiȬ nant);ȱUnitedȱStatesȱv.ȱE.I.ȱduȱPontȱNumoursȱ&ȱCo.,ȱ351ȱU.S.ȱ377,ȱ379ȱ(1956)ȱ(findȬ ingȱ seventyȬfiveȱ percentȱ predominant);ȱ Americanȱ Tobaccoȱ Co.ȱ v.ȱ Unitedȱ States.,ȱ 328ȱU.S.ȱ781ȱ(1946)ȱ(findingȱoverȱsixtyȬsixȱpercentȱpredominant);ȱUnitedȱStatesȱv.ȱ DentsplyȱInt’lȱInc.,ȱ399ȱF.3dȱ181,ȱ188ȱ(3dȱCir.ȱ2005)ȱ(findingȱseventyȬfiveȱtoȱeightyȱ percentȱpredominant).ȱȱ
54ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
therȱ platform.ȱ 182ȱ Theȱ possibleȱ complicationȱ isȱ dependentȱ onȱ theȱ fixedȱcostsȱofȱadvertising,ȱwhereȱtheȱinitialȱcostsȱofȱaccessingȱmulȬ tipleȱ systemsȱ couldȱ beȱ largeȱ enoughȱ toȱ precludeȱ currentȱ Googleȱ advertisersȱfromȱswitchingȱtoȱaȱcompetingȱplatform.ȱ ButȱtheȱargumentsȱinvolvingȱnetworkȱeffectsȱinȱGoogle’sȱcaseȱ doȱnotȱturnȱonȱtheseȱfixedȱstartupȱcosts;ȱrather,ȱtheȱargumentsȱ areȱ thatȱ competitorsȱ areȱ unableȱ toȱ obtainȱ necessaryȱ minimumȱ scaleȱtoȱofferȱqualityȱcompetitiveȱwithȱGoogle’s.ȱThatȱis,ȱtheȱinȬ terventionistsȱargueȱthatȱnetworkȱeffectsȱcreateȱanȱinsurmountȬ ableȱ barrierȱ toȱ entryȱ forȱ wouldȬbeȱ competitors.ȱ Thatȱ theȱ relevantȱ networkȱ effectsȱ areȱ internalizedȱ shouldȱ negateȱ thisȱ concern,ȱ however,ȱ andȱ atȱ anyȱ rateȱ allȱ ofȱ Google’sȱ mainȱ comȬ petitorsȱalreadyȱhaveȱsignificantȱscale.ȱ Theȱ Court’sȱ decisionȱ inȱ Microsoftȱ turnedȱ inȱ partȱ onȱ networkȱ effects,183ȱ andȱ theȱ court’sȱ approachȱ toȱ networkȱ effectsȱ inȱ thatȱ caseȱ isȱ ofȱ primaryȱ importance.ȱ Regardlessȱ ofȱ whetherȱ theȱ DeȬ partmentȱofȱJusticeȱorȱtheȱD.C.ȱCircuitȱwasȱultimatelyȱcorrectȱinȱ predictingȱ thatȱ Microsoftȇsȱ businessȱ practicesȱ wouldȱ resultȱ inȱ anticompetitiveȱ effects,ȱ Microsoftȱ offersȱ theȱ opportunityȱ toȱ evaluateȱtheȱapproachȱofȱtheȱDepartmentȱofȱJusticeȱinȱdevelopȬ ingȱ aȱ theoryȱ basedȱ onȱ aȱ particularȱ viewȱ ofȱ theȱ economicsȱ ofȱ networkȱ effects,ȱ andȱ theȱ approachȱ ofȱ theȱ courtsȱ inȱ assessingȱ thoseȱ theoriesȱ withȱ aȱ nascentȱ economicȱ literatureȱ inȱ aȱ highȬ stakesȱcaseȱinvolvingȱinnovation.ȱ Currently,ȱ ourȱ bestȱ understandingȱ ofȱ networkȱ effectsȱ viewsȱ them,ȱappropriately,ȱasȱbeneficial,ȱalthoughȱthereȱisȱdisputeȱinȱ theȱliteratureȱoverȱtheȱextentȱtoȱwhichȱtheirȱpresenceȱalsoȱraisesȱ exclusionaryȱconcerns.184ȱAlthoughȱthereȱisȱoftenȱaȱgreatȱdealȱofȱ carelessnessȱ inȱ definingȱ terms,ȱ particularlyȱ inȱ theȱ tenuousȱ translationȱfromȱeconomicȱtheoryȱtoȱjudicialȱopinions,ȱthereȱisȱaȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 182.ȱConsiderȱaȱcomparisonȱtoȱcarȱmanufacturers.ȱMercedesȬBenzȱoffersȱaȱbetter,ȱ butȱcorrespondinglyȱmoreȱexpensiveȱproductȱthatȱHonda.ȱYetȱbecauseȱofȱdifferencesȱ inȱdemandȱelasticityȱamongȱpurchasers,ȱbothȱHondaȱandȱMercedesȱareȱableȱtoȱcomȬ peteȱ vigorouslyȱ inȱ theȱ broaderȱ carȱ market.ȱ Inȱ theȱ searchȱ case,ȱ theȱ differencesȱ beȬ tweenȱtheȱcompaniesȱdoȱnotȱstemȱfromȱtheȱintendedȱendȱproductȱ(althoughȱproductȱ differentiationȱdoesȱexist,ȱofȱcourse,ȱandȱhasȱprovedȱaȱsourceȱofȱtheȱimpetusȱforȱenȬ try),ȱbutȱratherȱtheȱsizeȱofȱtheȱnetwork.ȱThatȱtheseȱnetworkȱexternalitiesȱareȱinternalȬ izedȱmeansȱthatȱalmostȱanyȱnetworkȱcanȱenterȱandȱexpectȱtoȱgrowȱbecauseȱfromȱtheȱ startȱitȱcanȱofferȱpricesȱcommensurateȱwithȱitsȱnetworkȱsize.ȱ 183.ȱUnitedȱ Statesȱ v.ȱ Microsoftȱ Corp.,ȱ 253ȱ F.3dȱ 34ȱ 49–50ȱ (D.C.ȱ Cir.ȱ 2001)ȱ (enȱ banc)ȱ(perȱcuriam).ȱ 184.ȱCompareȱLiebowitzȱ&ȱMargolis,ȱsupraȱnoteȱ128ȱwithȱShapiro,ȱsupraȱnoteȱ9.ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
55ȱ
crucialȱ distinctionȱ betweenȱ indirectȱ andȱ directȱ networkȱ efȬ fects.185ȱ Professorsȱ Liebowitzȱ andȱ Margolisȱ highlightȱ theȱ draȬ maticallyȱ differentȱ implicationsȱ ofȱ theȱ twoȱ effects,ȱ andȱ inȱ particularȱdemonstrateȱthatȱtransactionsȱinvolvingȱcomplemenȬ taryȱ products—indirectȱ networkȱ effects—fullyȱ internalizeȱ theȱ benefitsȱ ofȱ consumingȱ complementaryȱ goods.186ȱ Thus,ȱ despiteȱ frequentȱclaimsȱtoȱtheȱcontrary,ȱindirectȱnetworkȱeffectsȱareȱnotȱ aȱ sourceȱ ofȱ marketȱ failureȱ leadingȱ toȱ technologyȱ lockȬinȱ andȱ potentiallyȱexclusionaryȱeffects.187ȱ TheȱcaseȱagainstȱMicrosoftȱwasȱbuiltȱinȱimportantȱpartȱonȱindiȬ rectȱ networkȱ effects.ȱ Theȱ mostȱ importantȱ argumentȱ againstȱ theȱ company—thatȱtheȱsubstantialȱnumberȱofȱdevelopersȱwritingȱapȬ plicationsȱtoȱrunȱonȱWindowsȱsystemsȱwasȱanȱ“applicationsȱbarȬ rierȱ toȱ entry”—wasȱ anȱ argumentȱ thatȱ indirectȱ networkȱ effectsȱ insulatedȱ Microsoftȱfromȱ competitionȱ andȱ conferredȱtheȱ monopȬ olyȱ powerȱ requiredȱ forȱ theȱ courtȱ toȱ findȱ againstȱ it,ȱ despiteȱ theȱ claimedȱpersistentȱthreatȱofȱentry.188ȱSomeȱeconomicȱtheoryȱdoesȱ supportȱ theȱ possibilityȱ ofȱ thisȱ anticompetitiveȱ effect.ȱ Theȱ court’sȱ approachȱ toȱ addressingȱ whetherȱ thisȱ possibilityȱ warrantedȱ antiȬ trustȱliability,ȱhowever,ȱisȱproblematic.ȱTheȱcourtȱtreatedȱtheȱmereȱ allegationȱ ofȱ anȱapplicationsȱ barrierȱtoȱ entryȱasȱ sufficientȱtoȱ findȱ anȱanticompetitiveȱeffect:ȱItȱisȱenoughȱtoȱconcludeȱthatȱsuchȱaȱbarȬ rierȱexists,ȱandȱthatȱ“theȱapplicationsȱbarrierȱtoȱentryȱdiscouragesȱ manyȱ fromȱ writingȱ forȱ theseȱ lessȱ popularȱ platforms.”189ȱ Inȱ fact,ȱ thisȱconclusionȱmayȱbeȱcorrect.ȱWhatȱisȱtroubling,ȱhowever,ȱisȱthatȱ theȱ courtȇsȱ approachȱ isȱ notȱ sufficientlyȱ empirical,ȱ especiallyȱ inȱ lightȱofȱtheȱconflicting,ȱunderlyingȱtheoreticalȱliterature.ȱTheȱcourtȱ doesȱnotȱrequireȱproofȱthatȱtheȱconclusionȱisȱcorrect.ȱNorȱdoesȱtheȱ courtȱevenȱeffectivelyȱcanvassȱtheȱunderlyingȱ economicȱtheoretiȬ calȱliteratureȱtoȱsupportȱitsȱconclusion,ȱneverȱmentioning,ȱforȱexȬ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 185.ȱSee,ȱe.g.,ȱLiebowitzȱ&ȱMargolis,ȱsupraȱnoteȱ128,ȱatȱ135;ȱMichaelȱL.ȱKatzȱandȱ Carlȱ Shapiro,ȱSystemsȱ Competitionȱ andȱ NetworkȱEffects,ȱ 8ȱ J.ȱ ECON.ȱ PERSP.ȱ 93,ȱ 95–96ȱ (1994).ȱ 186.ȱSeeȱLiebowitzȱ&ȱMargolis,ȱsupraȱnoteȱ128,ȱatȱ149.ȱ 187.ȱȱSpulber,ȱsupraȱnoteȱ8.ȱ 188.ȱMicrosoft,ȱ253ȱF.3dȱatȱ55–56.ȱ 189.ȱId.ȱ Noteȱ thatȱ theȱ courtȱ doesȱ notȱ evenȱ assessȱ theȱ extentȱ thatȱ mereȱ discourȬ agementȱ operatesȱ asȱ anȱ effectiveȱ entryȱ barrier.ȱ Theȱ districtȱ court,ȱ weȱ note,ȱ howȬ ever,ȱwasȱnotȱquiteȱsoȱcircumspect.ȱSeeȱUnitedȱStatesȱv.ȱMicrosoft,ȱ87ȱF.ȱSupp.ȱ2dȱ 30,ȱ42ȱ(D.D.C.ȱ2000),ȱaff’dȱinȱpart,ȱrev’dȱinȱpart,ȱ253ȱF.3dȱ34ȱ(D.C.ȱCir.ȱ2001).ȱ
56ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
ample,ȱ Professorsȱ Liebowitzȱ andȱ Margolis’sȱ definitiveȱ workȱ onȱ theȱ(contrary)ȱimplicationsȱofȱindirectȱnetworkȱeffects.190ȱ Directȱevidenceȱwouldȱseemȱtoȱofferȱaȱcorrective,ȱandȱMicrosoftȱ arguedȱthatȱtheȱissueȱshouldȱbeȱdecidedȱonȱtheȱbasisȱofȱdirectȱeviȬ dence.ȱ Theȱ courtȱ dismissedȱ Microsoft’sȱ directȱ evidenceȱ onȱ moȬ nopolyȱ power,ȱ however,ȱ andȱ reliedȱ onȱ theȱ structuralȱ argumentȱ derivedȱfromȱitsȱcasualȱeconomicȱanalysisȱofȱtheȱapplicationsȱbarȬ rier.ȱ Unfortunately,ȱ asȱ Professorsȱ Liebowitzȱ andȱ Margolisȱ wroteȱ duringȱ theȱ heatȱ ofȱ theȱ Microsoftȱ case,ȱ ”Withȱ soȱ littleȱ empiricalȱ supportȱ forȱ theseȱ theories,ȱ itȱ appearsȱ atȱ bestȱ prematureȱ andȱ atȱ worstȱ simplyȱ wrongȱ toȱ useȱ thisȱ theoryȱ asȱ theȱ basisȱ forȱ antitrustȱ decisions.”191ȱTheȱcourtȱwasȱaskedȱtoȱactȱasȱtheȱultimateȱpeerȱreȬ viewerȱofȱanȱinternecineȱeconomicȱdebate—aȱtaskȱforȱwhichȱitȱwasȱ singularlyȱunsuited.ȱTheȱMicrosoftȱcaseȱrealizesȱtheȱriskȱthatȱagenȬ ciesȱ andȱ courtsȱ applyingȱ novelȱ economicȱ theoriesȱ inȱ novelȱ marȬ ketsȱwillȱtakeȱquestionableȱapproachesȱtoȱantitrustȱenforcement.ȱȱ Severalȱ commentatorsȱ haveȱ suggestedȱ thatȱ Google’sȱ searchȱ productȱ andȱ itsȱ searchȱ advertisingȱ productȱ exhibitȱ traditionalȱ networkȱeffectsȱandȱimplicitlyȱorȱexplicitlyȱleapȱtoȱtheȱimplicationȱ thatȱ increasedȱ antitrustȱ concernȱ isȱ warranted.192ȱ Thisȱ isȱ noȱ surȬ prise.ȱ Productȱ innovationsȱ coupledȱ withȱ allegedȱ networkȱ effectsȱ areȱlikelyȱtoȱattractȱacademicȱandȱregulatoryȱattention,ȱdrawȱantiȬ trustȱscrutiny,ȱandȱincreaseȱtheȱprobabilityȱofȱliability.ȱButȱseveralȱ importantȱ weaknessesȱ emergeȱ inȱ theseȱ argumentsȱ andȱ demonȬ strateȱ considerableȱ limitationsȱ inȱ ourȱ understandingȱ ofȱ theȱ dyȬ namicsȱofȱGoogle’sȱbusiness,ȱsuggestingȱthatȱaȱmoreȱempiricallyȬ mindedȱandȱcautiousȱapproachȱtoȱinterventionȱisȱdesirable.ȱ Remarkably,ȱ theȱ networkȬeffectȱ argumentȱ asȱ appliedȱ toȱ Googleȱ isȱ generallyȱ executedȱ byȱ nakedȱ assertion.ȱ Noneȱ ofȱ theȱ articlesȱ levelingȱ theȱ argumentȱ explainsȱ theȱ sourceȱ ofȱ theȱ netȬ workȱeffectsȱusingȱdetailsȱofȱGoogle’sȱactualȱmarket,ȱproducts,ȱ andȱ businessȱ practices—norȱ doȱ theyȱ explainȱ theȱ natureȱ ofȱ theȱ antitrustȱconcern.193ȱThisȱdoesȱnot,ȱofȱcourse,ȱmeanȱthatȱtheȱasȬ sertionȱ isȱ necessarilyȱ wrong.ȱ Erroneousȱ enforcementȱ isȱ moreȱ likely,ȱ however,ȱ ifȱ actionȱ isȱ takenȱ onȱ theȱ basisȱ ofȱ unsubstantiȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 190.ȱSeeȱLiebowitzȱ&ȱMargolis,ȱsupraȱnoteȱ128.ȱ 191.ȱStanȱ J.ȱ Liebowitzȱ &ȱ Stephenȱ E.ȱ Margolis,ȱ Networkȱ Effects,ȱ inȱ THEȱ NEWȱ PALGRAVEȱDICTIONARYȱOFȱECONOMICSȱANDȱTHEȱLAWȱ673,ȱ674ȱ(PeterȱNewmanȱed.,ȱ 1998).ȱ 192.ȱSeeȱsupraȱnoteȱ126.ȱ 193.ȱSeeȱid.ȱ(listingȱrecentȱarticlesȱidentifyingȱnetworkȱeffectsȱinȱGoogle’sȱbusiness).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
57ȱ
atedȱclaims.ȱErrorsȱcanȱbeȱcorrectedȱbyȱtheȱcourts,ȱbutȱtoȱdoȱso,ȱ courtsȱmustȱadoptȱapproachesȱtoȱantitrustȱenforcementȱthatȱareȱ conduciveȱtoȱerrorȱcorrection.ȱInȱparticular,ȱcourtsȱmustȱsteadȬ fastlyȱ protectȱ theȱ safeguardsȱ setȱ forthȱ byȱ theȱ Supremeȱ Court,ȱ whichȱ requireȱ plaintiffsȱ toȱ presentȱ rigorousȱ evidenceȱ ofȱ comȬ petitiveȱ harmȱ asȱ aȱ preconditionȱ forȱ liability.ȱ However,ȱ evenȱ whenȱ courtsȱ correctȱ erroneousȱ enforcementȱ (orȱ privateȱ litigaȬ tionȱ decisions),ȱ thisȱ merelyȱ reducesȱ theȱ probabilityȱ ofȱ erroneȬ ousȱ outcomes—anȱ effectȱ thatȱ isȱ notȱ insignificant,ȱ butȱ thatȱ neverthelessȱleavesȱsignificantȱerrorȬcostȱrisk.194ȱ Forȱtheȱpurposesȱofȱantitrustȱanalysis,ȱitȱisȱabsolutelyȱcriticalȱtoȱ distinguishȱtheȱexistenceȱofȱnetworkȱeffectsȱfromȱtheȱimplicationsȱ ofȱ theirȱ existenceȱ onȱ theȱ meritsȱ ofȱ antitrustȱ intervention.ȱ MoreȬ over,ȱtoȱtheȱextentȱthatȱtheȱexistenceȱofȱnetworkȱeffectsȱdoesȱmakeȱ aȱ differenceȱinȱ termsȱofȱanȱ antitrustȱcase,ȱ itȱmustȱ beȱ becauseȱtheȱ networkȱeffectȱoperatesȱasȱaȱbarrierȱtoȱentry,ȱtherebyȱdiminishingȱ theȱlikelihoodȱthatȱusersȱwillȱtransferȱtoȱcompetingȱplatforms.ȱBeȬ causeȱ usersȱ ofȱ searchȱ enginesȱ areȱ insensitiveȱ toȱ theȱ numberȱ ofȱ otherȱusers,ȱthisȱeffectȱshouldȱnotȱstronglyȱhold.ȱNorȱdoȱindirectȱ networkȱeffectsȱseemȱtoȱcreateȱaȱbarrierȱtoȱentryȱhere.195ȱȱ Theȱupshotȱisȱthatȱthereȱisȱconsiderableȱdifficultyȱinȱassessingȱ theȱcompetitiveȱimplicationsȱofȱinnovativeȱproducts,ȱandȱreflexȬ iveȱappealsȱtoȱtheȱexistenceȱofȱnetworkȱeffectsȱasȱjustifyingȱinterȬ ventionȱ areȱ likelyȱ toȱ leadȱ toȱ erroneousȱ decisionsȱ byȱ enforcersȱ andȱjudges.ȱMoreover,ȱtheȱapproachȱadoptedȱinȱMicrosoftȱtoȱanaȬ lyzeȱtheȱcompetitiveȱimplicationsȱofȱnetworkȱeffects,ȱifȱappliedȱinȱ Google’sȱ case,ȱ isȱ disconcertinglyȱ likelyȱ toȱ overȬemphasizeȱ theȱ theoreticalȱ argumentsȱ supportingȱ aȱ chargeȱ ofȱ problematicȱ netȬ workȱeffects,ȱevenȱinȱtheȱabsenceȱofȱclearȱempiricalȱevidenceȱtoȱ supportȱ thoseȱ charges.ȱ Inȱ Google’sȱ case,ȱ whereȱ networkȱ effectsȱ atȱ leastȱ superficiallyȱ appearȱ toȱ beȱ prevalentȱ andȱ itsȱ businessȱ modelȱ appearsȱ toȱ mimicȱ thoseȱ ofȱ otherȱ networkȱ industries,ȱ aȱ clearerȱ understandingȱ ofȱ theȱ particularsȱ ofȱ Google’sȱ businessȱ suggestsȱ thatȱ thoseȱ presumptionsȱ areȱ prematureȱ andȱ misdiȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 194.ȱSeeȱEasterbrook,ȱsupraȱnoteȱ10,ȱatȱ15.ȱȱ 195.ȱTheȱ generalȱ economicȱ consequencesȱ ofȱ indirectȱ andȱ directȱ networkȱ effectsȱ areȱsubstantiallyȱdifferent,ȱandȱindirectȱnetworkȱeffectsȱareȱnotȱaȱsourceȱofȱineffiȬ cientȱ technologyȱ lockȬin—andȱ theyȱ thusȱ doȱ notȱ poseȱ aȱ barrierȱ toȱ entry.ȱ Seeȱ 1ȱ Stanleyȱ J.ȱ Liebowitzȱ &ȱ Stephenȱ E.ȱ Margolis,ȱ Networkȱ Effects,ȱ inȱ 1ȱ HANDBOOKȱ OFȱ TELECOMMUNICATIONSȱ ECONOMICS:ȱ STRUCTURE,ȱ REGULATIONȱ ANDȱ COMPETITIONȱ 79,ȱ85–86ȱ(MartinȱE.ȱCaveȱetȱal.ȱeds.,ȱ2002).ȱSeeȱgenerallyȱSpulber,ȱsupraȱnoteȱ8ȱ(exȬ plainingȱwhyȱtechnologyȱlockȬinȱrarelyȱoccurs).ȱ
58ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
rected.ȱFurtherȱ understandingȱmightȱleadȱtoȱ conclusionsȱ differȬ entȱ thanȱ theȱ preliminaryȱ onesȱ describedȱ here.ȱ Theȱ immediateȱ pointȱ isȱ thatȱ theȱ likelihoodȱ ofȱ errorȱ inȱ theȱ faceȱ ofȱ Google’sȱ imȬ menselyȱcomplicatedȱproductȱandȱbusinessȱinnovationsȱisȱunacȬ ceptablyȱ high,ȱ particularlyȱ coupledȱ withȱ theȱ dynamicȱ consequencesȱ ofȱ deterringȱ innovationsȱ exactlyȱ likeȱ Google’s,ȱ whichȱhaveȱprovenȱtoȱbeȱenormouslyȱwelfareȬenhancing.ȱ E.
HasȱGoogleȱEngagedȱinȱExclusionaryȱConduct?ȱȱ
Thereȱ areȱ fiveȱ primaryȱ categoriesȱ ofȱ Sectionȱ 2ȱ claimsȱ raisedȱ againstȱGoogleȱinȱtheȱTradeCometȱcomplaint:ȱ ȱ First:ȱ Googleȱ enteredȱ intoȱ exclusiveȱ syndicationȱ agreeȬ mentsȱ withȱ certainȱ highȬtrafficȱ onlineȱ publishers,ȱ foreȬ closingȱaccessȱbyȱcompetitorsȱtoȱtheseȱimportantȱsourcesȱ ofȱsearchȱrevenue.196ȱ ȱ Second:ȱ Googleȱ manipulatedȱ itsȱ “Landingȱ Pageȱ Quality”ȱ scoreȱtoȱexcludeȱcompetitorsȱfromȱgainingȱtrafficȱthroughȱ Googleȱadvertising.197ȱ ȱ Third:ȱ Googleȱ restrictedȱ advertisersȱ accessȱ toȱ importantȱ dataȱ createdȱ whileȱ usingȱ AdWords.198ȱ Advertisersȱ oftenȱ embarkȱ onȱ complexȱ campaignsȱ thatȱ involveȱ biddingȱ onȱ hundredsȱ ofȱ thousandsȱ ofȱ keywords.ȱ Byȱ restrictingȱ acȬ cessȱ toȱ AdWordsȱ data,ȱ Googleȱ hasȱ madeȱ itȱ difficultȱ forȱ advertisersȱtoȱevaluateȱtheȱperformanceȱofȱtheirȱadvertisȬ ingȱcampaignsȱandȱdecideȱwhetherȱtoȱswitchȱtoȱorȱaddȱaȱ competitor’sȱsearchȱadvertisingȱservice.ȱȱ ȱ Fourth:ȱ Googleȱ deployedȱ defaultȱ mechanismsȱ thatȱ makeȱ itȱ difficultȱ forȱ usersȱ toȱ selectȱ aȱ searchȱ engineȱ otherȱ thanȱ Google.199ȱWhenȱindividualsȱuseȱGoogle’sȱtoolbarȱfeatureȱ theyȱautomaticallyȱhaveȱGoogleȱsetȱasȱtheȱdefaultȱsearchȱ tool.ȱIfȱaȱuserȱtriesȱtoȱsetȱanȱalternativeȱsearchȱengineȱtheȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 196.ȱComplaint,ȱsupraȱnoteȱ134,ȱ¶ȱ69.ȱ 197.ȱId.ȱ¶ȱ91–97.ȱ 198.ȱId.ȱ¶ȱ74.ȱȱ 199.ȱId.ȱ¶ȱ75.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
59ȱ
default,ȱ Google’sȱ toolbarȱ softwareȱ automaticallyȱ andȱ withoutȱtheȱuser’sȱpermission,ȱrevertsȱtoȱGoogle.ȱȱ ȱ Fifth:ȱ Googleȱ manipulatedȱ itsȱ organicȱ searchȱ algorithmȱ toȱ benefitȱGoogle’sȱownȱproductsȱandȱdisfavorȱcompetitors.200ȱȱ ȱ Theseȱ fiveȱ categoriesȱ ofȱ allegationsȱ largelyȱ revolveȱ aroundȱ twoȱbusinessȱdecisions.ȱFirst,ȱGoogleȱhasȱenteredȱintoȱexclusiveȱ syndicationȱ agreementsȱ withȱ highȬtrafficȱ websitesȱ toȱ impedeȱ competitors’ȱ abilityȱ toȱ gainȱ theȱ criticalȱ searchȱ exposureȱ necesȬ saryȱ toȱ operateȱ aȱ viableȱ searchȱ advertisingȱ platform.ȱ Second,ȱ Googleȱ hasȱ implementedȱ aȱ qualityȱ metricȱ asȱ partȱ ofȱ itsȱ keyȬ wordȱauctionȱthatȱeffectivelyȱterminatesȱitsȱvoluntaryȱprofitableȱ dealingsȱwithȱcompetitors.ȱWeȱconsiderȱeachȱbusinessȱdecisionȱ inȱturn,ȱhighlightingȱissuesȱcriticalȱtoȱanalyzingȱtheȱmeritsȱofȱaȱ potentialȱmonopolizationȱcaseȱagainstȱGoogle.ȱ F.
ExclusiveȱSyndicationȱAgreementsȱandȱOtherȱForeclosureȬBasedȱ Argumentsȱȱ
Theȱ TradeCometȱ complaintȱ allegesȱ thatȱ Googleȱ hasȱ enteredȱ intoȱexclusiveȱagreementsȱ“withȱmanyȱofȱtheȱmostȱhighlyȱtrafȬ fickedȱ websitesȱ onȱ theȱ Internet,ȱ guaranteeingȱ thatȱ anyȱ searchȱ generatedȱ atȱ thoseȱ nonȬsearchȱ websitesȱ.ȱ.ȱ.ȱisȱ directedȱ toȱ Google’sȱ searchȱ advertisingȱ platformȱ ratherȱ thanȱ toȱ rivalȱ platȬ forms.”201ȱForȱexample,ȱGoogleȱenteredȱintoȱanȱagreementȱwithȱ AmericaȱOnlineȱ(AOL),ȱwhichȱdedicatedȱitsȱsearchȱbusinessȱtoȱ Google’sȱ technology.202ȱ Othersȱ haveȱ similarlyȱ arguedȱ thatȱ Google’sȱ exclusiveȱ arrangementsȱ withȱ AOL,ȱ whichȱ involvedȱ paymentsȱ inȱ exchangeȱ forȱ placementȱ ofȱ aȱ smallȱ boxȱ onȱ everyȱ webpageȱthatȱsaidȱ“SearchȱPoweredȱbyȱGoogle,”ȱwereȱaȱcriticalȱ momentȱinȱGoogle’sȱhistory.203ȱ Theȱ antitrustȱ claimsȱ relatedȱ toȱ Google’sȱ exclusiveȱ syndicationȱ agreementsȱareȱrelativelyȱstraightforward.204ȱTheȱallegationȱisȱthatȱ Google’sȱ agreementsȱ contractuallyȱ forecloseȱ competingȱ searchȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 200.ȱId.ȱ¶ȱ76.ȱ 201.ȱId.ȱ¶ȱ68.ȱ 202.ȱId.ȱ¶ȱ28.ȱ 203.ȱVISEȱ&ȱMALSEED,ȱsupraȱnoteȱ126,ȱatȱ208–09.ȱ 204.ȱForȱpurposesȱofȱthisȱArticle,ȱweȱuseȱtheȱtermȱ“exclusiveȱagreements”ȱtoȱenȬ compassȱbothȱagreementsȱthatȱrequireȱfullȱasȱwellȱasȱpartialȱexclusivity.ȱ
60ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
enginesȱfromȱtheȱopportunityȱtoȱcompeteȱforȱtheȱdistributionȱnecȬ essaryȱ toȱ achieveȱ minimumȱ efficientȱ scale.ȱ Inȱ theȱ lessȱ technicalȱ languageȱ adoptedȱ inȱ theȱ TradeCometȱ complaint,ȱ itȱ isȱ allegedȱ thatȱ theȱ agreementsȱ areȱ responsibleȱ forȱ “lockingȱ up”ȱ theȱ businessȱ ofȱ significantȱInternetȱpublishersȱsuchȱthatȱtheȱresultȱisȱ“foreclo[sureȱ ofȱa]ȱsubstantialȱpercentageȱofȱtheȱsearchȱsyndicationȱmarket.”205ȱȱ Thus,ȱtheȱanticompetitiveȱtheoryȱofȱharmȱisȱthatȱGoogleȱhasȱ lockedȱupȱaȱsufficientȱshareȱofȱtheȱinternetȱsearchȱbusinessȱwithȱ exclusiveȱarrangementsȱsuchȱthatȱrivalȱsearchȱoperatorsȱcannotȱ achieveȱ minimumȱ efficientȱ scale,ȱ andȱ thatȱ Googleȱ doesȱ soȱ byȱ predatoryȱ“overbuying”ȱsuchȱthatȱtheȱpaymentsȱdoȱnotȱnecesȬ sarilyȱ coverȱ Google’sȱ advertisingȱ revenues.206ȱ Whileȱ antitrustȱ hasȱ longȱ recognizedȱ theȱ competitiveȱ necessityȱ ofȱ exclusiveȱ arȬ rangements,ȱ itȱ isȱ alsoȱ possibleȱ thatȱ suchȱ agreementsȱ canȱ raiseȱ barriersȱtoȱentryȱandȱviolateȱSectionȱ2ȱofȱtheȱShermanȱActȱunȬ derȱ certainȱ conditions.ȱ Google’sȱ exclusiveȱ syndicationȱ agreeȬ mentsȱhere,ȱhowever,ȱareȱnotȱlikelyȱcauseȱforȱantitrustȱconcern.ȱȱ Oneȱ reasonȱ exclusiveȱ dealingȱ arrangementsȱ haveȱ longȱ beenȱ understoodȱ byȱ antitrustȱ enforcersȱ andȱ courtsȱ toȱ resultȱ inȱ theȱ typeȱofȱcompetitiveȱharmȱrequiredȱforȱanȱantitrustȱviolationȱisȱ thatȱ exclusiveȱ dealingȱ contractsȱ soȱ frequentlyȱ ariseȱ fromȱ theȱ competitiveȱprocessȱforȱproductȱdistribution.ȱConsiderȱGoogle’sȱ syndicationȱarrangementȱwithȱAOL.ȱWhileȱtheȱTradeCometȱcomȬ plaintȱ presentsȱ theȱ temptationȱ toȱ viewȱ Google’sȱ successȱ inȱ obȬ tainingȱthatȱcontractȱasȱaȱsymptomȱofȱtheȱlackȱofȱcompetition,ȱitȱ isȱ actuallyȱ theȱ opposite.ȱ Asȱ Mssrs.ȱ Viseȱ andȱ Malseedȱ note,ȱ Google’sȱsuccessȱwasȱanȱexampleȱofȱbestingȱYahoo!ȱinȱaȱfierceȱ competitionȱ forȱ AOL’sȱ business.207ȱ Thisȱ “competitionȱ forȱ theȱ contract,”ȱinȱantitrustȱparlance,ȱexposesȱtheȱfallacyȱthatȱobservȬ ingȱ aȱ winnerȱ onȱ theȱ topȱ ofȱ hillȱ byȱ himselfȱ afterȱ aȱ raceȱ impliesȱ theȱlackȱofȱcompetition.ȱThisȱformȱofȱupfront,ȱexȱanteȱcompetiȬ tionȱ hasȱ longȱ beenȱ recognizedȱ asȱ aȱ dimensionȱ ofȱ competitionȱ thatȱgeneratesȱsubstantialȱbenefitsȱforȱconsumers—AOL’sȱextraȱ revenuesȱareȱpassedȱonȱinȱtheȱformȱofȱlowerȱprices,ȱinvestmentsȱ inȱtheȱqualityȱofȱitsȱproducts,ȱandȱotherȱbenefits—thatȱantitrustȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 205.ȱComplaint,ȱsupraȱnoteȱ134,ȱ¶ȱ70.ȱ 206.ȱId.ȱȱ 207.ȱVICEȱ&ȱMALSEED,ȱsupraȱnoteȱ126,ȱatȱ208.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
61ȱ
mustȱ protect.208ȱ Indeed,ȱ antitrustȱ courtsȱ haveȱ recognizedȱ thisȱ formȱ ofȱ competitionȱ asȱ “aȱ vitalȱ formȱ ofȱ rivalryȱ.ȱ.ȱ.ȱwhichȱ theȱ antitrustȱlawsȱencourageȱratherȱthanȱsuppress.”209ȱ Theȱcompetitiveȱprocessȱforȱproductȱdistribution,ȱorȱaccessȱtoȱ promotionȱandȱdistribution,ȱisȱespeciallyȱvigorousȱinȱhighȬtechȱ andȱwebȬbasedȱmarketsȱwhereȱvyingȱforȱconsumerȱattentionȱisȱ anȱ importantȱ elementȱ ofȱ theȱ competitiveȱ process.ȱ Becauseȱ ofȱ theȱ highȱ stakesȱ natureȱ ofȱ competitionȱ forȱ theȱ contractȱ andȱ theȱ businessȱthatȱtheȱcontractȱbringsȱwithȱit,ȱandȱbecauseȱofȱtheȱinȬ tuitiveȱappealȱofȱtheȱfallacyȱthatȱtheȱwinnerȱofȱanȱexclusiveȱconȬ tractȱ mustȱ faceȱ littleȱ competition,ȱ antitrustȱ analysisȱ ofȱ competitionȱ forȱ distributionȱ isȱ “anȱ unsettledȱ andȱ sometimesȱ incoherentȱ areaȱ ofȱ theȱ law.”210ȱ Google’sȱ exclusiveȱ syndicationȱ agreementsȱwithȱfirmsȱthatȱcanȱshiftȱtheirȱcustomerȱbasesȱtoȱ(orȱ from)ȱ Google’sȱ searchȱ technologyȱ areȱ properlyȱ viewedȱ asȱ aȱ formȱ ofȱ competitionȱ forȱ distribution,ȱ similarȱ toȱ theȱ slottingȱ alȬ lowancesȱthatȱmanufacturersȱpayȱtoȱretailersȱforȱpremiumȱgroȬ ceryȱstoreȱshelfȱspaceȱwhichȱincreasesȱproductȱsales.ȱInȱmarketsȱ whereȱ theȱ successȱ ofȱ theȱ firmȱ dependsȱ onȱ AOLȱ andȱ similarlyȱ situatedȱ firmsȱ facilitatingȱ accessȱ toȱ consumers,ȱ oneȱ canȱ expectȱ vigorousȱcompetitionȱforȱdistributionȱtoȱensue.ȱȱ Moreover,ȱGoogle’sȱsoȬcalledȱexclusiveȱsyndicationȱagreementsȱ areȱoftenȱforȱlimitedȱduration,ȱtypicallyȱapplyȱtoȱonlyȱoneȱformȱofȱ advertising,ȱ andȱ oftenȱ allowȱ theȱ syndicationȱ partnerȱ toȱ sellȱ adsȱ directly.ȱThus,ȱtheȱagreementsȱareȱtypicallyȱpartialȱexclusivesȱthatȱ allowȱtheȱsyndicationȱpartnerȱtoȱhaveȱgreaterȱchoiceȱandȱtoȱretainȱ greaterȱproductȱvarietyȱthanȱwouldȱobtainȱunderȱaȱfullȱexclusive.ȱ Eachȱofȱtheseȱfactorsȱreducesȱtheȱdegreeȱorȱextentȱofȱexclusivity.ȱ Suchȱ limitationsȱ areȱ notȱ surprisingȱ contractualȱ compromisesȱ givenȱvigorousȱcompetitionȱamongȱsearchȱengines.ȱȱ MerelyȱlabelingȱGoogle’sȱsyndicationȱagreementsȱasȱ“competiȬ tionȱforȱdistribution”ȱdoesȱnotȱmeanȱthatȱtheȱagreementsȱareȱimȬ muneȱ fromȱ antitrustȱ scrutiny.ȱ Whenȱ accompaniedȱ withȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 208.ȱSee,ȱe.g.,ȱPaddockȱPubl’ns,ȱInc.ȱv.ȱChi.ȱTribuneȱCo.,ȱ103ȱF.3dȱ42,ȱ45ȱ(7thȱCir.ȱ 1996)ȱ (“CompetitionȬforȬtheȬcontractȱ isȱ aȱ formȱ ofȱ competitionȱ thatȱ antitrustȱ lawsȱ protectȱratherȱthanȱproscribe,ȱandȱitȱisȱcommon.”).ȱ 209.ȱMenashaȱCorp.ȱv.ȱNewsȱAm.ȱMktg.ȱInȬStore,ȱInc.,ȱ354ȱF.3dȱ661,ȱ663ȱ(7thȱCir.ȱ 2004).ȱȱ 210.ȱJoshuaȱD.ȱWright,ȱAntitrustȱLawȱandȱCompetitionȱforȱDistribution,ȱ23ȱYALEȱ J.ȱ ONȱ REG.ȱ 169,ȱ 170,ȱ 191ȱ (2006);ȱ seeȱ alsoȱ Klein,ȱ supraȱ noteȱ 149ȱ (explainingȱ that,ȱ conȬ traryȱtoȱtheȱ“unfortunateȱtendency”ȱofȱviewingȱexclusiveȱdistributionȱcontractsȱasȱ uncompetitive,ȱsuchȱcontractsȱoftenȱinvolveȱ“competitionȱonȱtheȱmerits”).ȱ
62ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
exclusivityȱ provisions,ȱ conditionsȱ ariseȱ underȱ whichȱ agreementsȱ canȱ excludeȱ equallyȱ efficientȱ rivals,ȱ raiseȱ barriersȱ toȱ entry,ȱ andȱ generateȱconsumerȱharm.ȱItȱdoes,ȱhowever,ȱimplyȱthat,ȱlikeȱotherȱ formsȱofȱcompetition,ȱoneȱshouldȱhesitateȱbeforeȱcondemningȱasȱ anȱ antitrustȱ violationȱ anȱ outcomeȱ thatȱ isȱ generatedȱ byȱ theȱ comȬ petitiveȱprocess.ȱBeforeȱturningȱtoȱwhetherȱtheȱconditionsȱnecesȬ saryȱforȱGoogle’sȱagreementsȱtoȱharmȱconsumersȱareȱlikelyȱtoȱbeȱ metȱhere,ȱitȱisȱworthȱnotingȱthreeȱimportantȱandȱoftenȱoverlookedȱ benefitsȱofȱcompetitionȱforȱdistributionȱinvolvingȱexclusivity.ȱȱ TheȱfirstȱcompetitiveȱbenefitȱisȱthatȱcompetitionȱforȱdistribuȬ tionȱ generatesȱ promotionalȱ paymentsȱ toȱ distributors,ȱ whichȱ are,ȱinȱturn,ȱpassedȱonȱtoȱconsumersȱthroughȱqualityȱimproveȬ mentsȱ orȱ priceȱ reductionsȱ inȱ theȱ distributors’ȱ markets.ȱ AlȬ thoughȱ theȱ competitiveȱ benefitsȱ areȱ notȱ asȱ intuitivelyȱ obviousȱ asȱ paymentsȱ thatȱ takeȱ theȱ formȱ ofȱ aȱ priceȱ reduction,ȱ competiȬ tiveȱpaymentsȱfromȱfirmsȱlikeȱGoogleȱtoȱAOLȱcanȱimproveȱconȬ sumerȱwelfareȱasȱtheyȱareȱpassedȱonȱtoȱconsumers.ȱȱ Second,ȱ whenȱ exclusivityȱ provisionsȱ areȱ observedȱ inȱ conȬ tractsȱ resultingȱ fromȱ competitionȱ forȱ distribution,ȱ theyȱ oftenȱ haveȱ importantȱ efficiencyȱ effects.ȱ Forȱ example,ȱ exclusiveȱ dealȬ ingȱcanȱfacilitateȱinvestmentȱandȱtheȱsupplyȱofȱefficientȱpromoȬ tionȱ andȱ distributionȱ byȱ minimizingȱ freeȬridingȱ bothȱ inȱ theȱ presenceȱandȱabsenceȱofȱmanufacturerȬsuppliedȱinvestments.211ȱ Thisȱisȱoneȱreasonȱwhyȱweȱobserveȱexclusiveȱdealingȱcontractsȱ inȱindustriesȱwhereȱfirmsȱdoȱnotȱhaveȱmarketȱpower.ȱȱ Third,ȱwhenȱfirmsȱlikeȱAOLȱofferȱpartiallyȱorȱfullyȱexclusiveȱ contractsȱupȱforȱbiddingȱtoȱGoogleȱandȱitsȱrivals,ȱtheȱresultȱcanȱ beȱ toȱ intensifyȱ competitionȱ forȱ distribution.212ȱ Professorsȱ Kleinȱ andȱ Murphyȱ demonstrateȱ thatȱ offeringȱ upstreamȱ firmsȱ accessȱ toȱtheȱdistributor’sȱloyalȱcustomerȱbaseȱenablesȱtheȱdistributorȱ (inȱ thisȱ caseȱ AOL)ȱ toȱ commitȱ aȱ substantialȱ fractionȱ ofȱ itsȱ cusȬ tomers’ȱ purchasesȱ toȱ theȱ favoredȱ supplierȱ andȱ therebyȱ draȬ maticallyȱ increaseȱ eachȱ supplier’sȱ perceivedȱ elasticityȱ ofȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 211.ȱSeeȱ Benjaminȱ Kleinȱ &ȱ Andresȱ V.ȱ Lerner,ȱ Theȱ Expandedȱ Economicsȱ ofȱ FreeȬ Riding:ȱHowȱExclusiveȱDealingȱPreventsȱFreeȬRidingȱandȱCreatesȱUndividedȱLoyalty,ȱ74ȱ ANTITRUSTȱL.J.ȱ473,ȱ483–84,ȱ498ȱ(2007).ȱ 212.ȱBenjaminȱ Kleinȱ &ȱ Kevinȱ M.ȱ Murphy,ȱ Exclusiveȱ Dealingȱ Intensifiesȱ CompetiȬ tionȱforȱDistribution,ȱ75ȱANTITRUSTȱ L.J.ȱ 433,ȱ437ȱ(2008).ȱThisȱexplanationȱisȱrelatedȱ to,ȱandȱprovidesȱtheȱeconomicȱbasisȱfor,ȱtheȱargumentȱthatȱexclusivesȱ“instigated”ȱ byȱcustomersȱshouldȱenjoyȱaȱpresumptionȱofȱlegality.ȱSeeȱRichardȱM.ȱSteuer,ȱCusȬ tomerȬInstigatedȱExclusiveȱDealing,ȱ68ȱANTITRUSTȱL.J.ȱ239,ȱ240–42ȱ(2000).ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
63ȱ
demandȱbyȱmakingȱrivalȱproductsȱhighlyȱsubstitutable.213ȱThisȱ effectȱisȱimportantȱbecauseȱitȱprovidesȱaȱprocompetitiveȱrationȬ aleȱ forȱ whyȱ bothȱ Googleȱ andȱ firmsȱ likeȱ AOLȱ mightȱ desireȱ exȬ clusiveȱ syndicationȱ agreements:ȱ Theyȱ makeȱ bothȱ firmsȱ betterȱ offȱ andȱ consumersȱ benefitȱ asȱ greaterȱ paymentsȱ areȱ passedȱ on.ȱ TheȱkeyȱpolicyȱimplicationsȱforȱtheȱantitrustȱtreatmentȱofȱcomȬ petitionȱ forȱ distributionȱ isȱ thatȱ itȱ isȱ aȱ normalȱ partȱ ofȱ theȱ comȬ petitiveȱprocessȱandȱanyȱantitrustȱscrutinyȱshouldȱbeȱfocusedȱonȱ ensuringȱthatȱrivalsȱhaveȱopenȱaccessȱtoȱofferȱcompetingȱbids.ȱ Exclusiveȱ syndicationȱ agreements,ȱ likeȱ mostȱ exclusiveȱ dealȬ ingȱcontractsȱresultingȱfromȱtheȱcompetitiveȱprocessȱforȱdistriȬ bution,ȱ areȱ likelyȱ toȱ provideȱ atȱ leastȱ someȱ efficiencyȱ benefitsȱ andȱ toȱ harmȱ someȱ individualȱ competitors.ȱ Theȱ questionȱ reȬ mains,ȱhowever,ȱwhetherȱthoseȱagreementsȱmightȱalsoȱproduceȱ harmȱtoȱcompetitionȱinȱtheȱformȱofȱanticompetitiveȱeffectsȱandȱ therebyȱviolateȱSectionȱ2.ȱWeȱnowȱturnȱtoȱthatȱquestion.ȱȱ Theȱ modernȱ “ruleȱ ofȱ reason”ȱ analysisȱ evaluatingȱ exclusiveȱ dealingȱcontractsȱfocusesȱonȱaȱnumberȱofȱfactors,ȱincludingȱtheȱ defendant’sȱ marketȱ power,ȱ theȱ degreeȱ ofȱ marketȱ foreclosure,ȱ entryȱ conditionsȱ withinȱ theȱ market,ȱ theȱ durationȱ ofȱ theȱ conȬ tractsȱatȱissue,ȱwhetherȱexclusivityȱhasȱtheȱpotentialȱtoȱraiseȱriȬ vals’ȱ costs,ȱ theȱ presenceȱ ofȱ actualȱ orȱ likelyȱ anticompetitiveȱ effects,ȱ andȱ businessȱ justificationsȱ forȱ theȱ questionedȱ action.ȱ Professorsȱ Areedaȱ andȱ Hovenkampȱ articulateȱ theȱ primaȱ facieȱ caseȱforȱexclusiveȱdealingȱclaimsȱasȱfollows:ȱȱ Inȱ orderȱ toȱ succeedȱ inȱ itsȱ claimȱ ofȱ unlawfulȱ exclusiveȱ dealȬ ing,ȱ aȱ plaintiffȱ mustȱ showȱ theȱ requisiteȱ agreementȱ toȱ dealȱ exclusivelyȱandȱmakeȱaȱsufficientȱshowingȱofȱpowerȱtoȱwarȬ rantȱ theȱ inferenceȱ thatȱ theȱ challengedȱ agreementȱ threatensȱ reducedȱoutputȱandȱhigherȱpricesȱinȱaȱproperlyȱdefinedȱmarȬ ket.ȱThenȱitȱmustȱalsoȱshowȱaȱforeclosureȱcoverageȱsufficientȱ toȱwarrantȱanȱinferenceȱofȱinjuryȱtoȱcompetition,ȱdependingȱ onȱ theȱ existenceȱ ofȱ otherȱ factorsȱ thatȱ giveȱ significanceȱ toȱ aȱ givenȱ foreclosureȱ percentage,ȱ suchȱ asȱ contractȱ duration,ȱ presenceȱorȱabsenceȱofȱhighȱentryȱbarriers,ȱorȱtheȱexistenceȱofȱ alternativeȱsourcesȱorȱresale.214ȱ
Modernȱ antitrustȱ analysisȱ ofȱ exclusiveȱ agreementsȱ inȱ casesȱ involvingȱ competitionȱ forȱ distributionȱ thereforeȱ requires:ȱ first,ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 213.ȱKleinȱ&ȱMurphy,ȱsupraȱnoteȱ212,ȱatȱ444–45.ȱ 214.ȱ11ȱ PHILLIPȱ E.ȱ AREEDAȱ &ȱ HERBERTȱ HOVENKAMP,ȱ ANTITRUSTȱ LAWȱ ¶ȱ 1821ȱ (2ndȱed.ȱ2005)ȱ(citationȱomitted).ȱ
64ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
aȱdemonstrationȱofȱtheȱdefendant’sȱmarketȱpower,ȱsecond,ȱsubȬ stantialȱ foreclosure,ȱ thirdȱ contractsȱ ofȱ sufficientȱ durationȱ toȱ prohibitȱmeaningfulȱcompetitiveȱbiddingȱbyȱrivals,ȱandȱfourth,ȱ anȱ analysisȱ ofȱ actualȱ orȱ likelyȱ competitiveȱ effectsȱ arisingȱ fromȱ theȱ defendant’sȱ conduct.ȱ Weȱ willȱ focusȱ ourȱ discussionȱ onȱ theȱ secondȱ andȱ thirdȱ elementsȱ hereȱ becauseȱ weȱ haveȱ alreadyȱ disȬ cussedȱ monopolyȱ powerȱ andȱ theȱ requirementȱ facedȱ byȱ allȱ plaintiffsȱ bringingȱ monopolizationȱ claimsȱ toȱ demonstrateȱ thatȱ theȱ conductȱ atȱ issueȱ eitherȱ hasȱ generatedȱ orȱ willȱ veryȱ likelyȱ generateȱhigherȱprices,ȱreducedȱoutput,ȱorȱlessȱinnovation.ȱ G.
SubstantialȱForeclosureȱ
Itȱ isȱ typicallyȱ necessaryȱtoȱ showȱthatȱ aȱ monopolistȱ hasȱ foreȬ closedȱatȱleastȱfortyȱpercentȱofȱtheȱrelevantȱmarketȱbeforeȱantiȬ trustȱ liabilityȱ canȱ beȱ found.215ȱ Oneȱ commentatorȱ summarizesȱ currentȱ antitrustȱ lawȱ asȱ “routinelyȱ sustain[ing]ȱ theȱ legalityȱ ofȱ exclusiveȱdealingȱarrangementsȱwithȱforeclosureȱpercentagesȱofȱ 40ȱ percentȱ orȱ less.”216 Notwithstandingȱ thisȱ traditionalȱ threshȬ old,ȱaȱsmallerȱforeclosureȱpercentageȱcanȱsufficeȱsoȱlongȱasȱitȱisȱ shownȱ thatȱ competitorsȱ haveȱ beenȱ keptȱ fromȱ achievingȱ theȱ criticalȱmassȱnecessaryȱtoȱposeȱaȱthreatȱtoȱtheȱmonopolist.217ȱȱ Theȱeconomicȱlogicȱofȱtheȱforeclosureȱrequirementȱisȱsound.ȱ Theȱanticompetitiveȱtheoriesȱofȱexclusiveȱdealingȱarrangementsȱ inȱ theȱ economicsȱ literatureȱ requireȱ substantialȱ economiesȱ ofȱ scale.218ȱThisȱisȱbecauseȱinȱorderȱforȱaȱmonopolistȱtoȱsucceedȱinȱ increasingȱbarriersȱtoȱentry,ȱheȱmustȱcoverȱenoughȱdistributionȱ forȱaȱsufficientȱperiodȱofȱtimeȱthatȱrivalsȱdoȱnotȱhaveȱtheȱopporȬ tunityȱtoȱachieveȱminimumȱefficientȱscale.ȱIfȱrivalsȱfaceȱconstantȱ returnsȱtoȱscale,ȱaȱreductionȱinȱdistributionȱopportunitiesȱdoesȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 215.ȱSee,ȱ e.g.,ȱ Jeffersonȱ Parishȱ Hosp.ȱ Dist.ȱ No.ȱ 2ȱ v.ȱ Hyde,ȱ 466ȱ U.S.ȱ 2,ȱ 7ȱ (1984)ȱ (holdingȱthatȱaȱhospitalȱwithȱthirtyȱpercentȱofȱtheȱrelevantȱmarketȱdidȱnotȱpossessȱ significantȱmarketȱpower).ȱ 216.ȱJonathanȱM.ȱJacobson,ȱExclusiveȱDealing,ȱ“Foreclosure,”ȱandȱConsumerȱHarm,ȱ 70ȱANTITRUSTȱ L.J.ȱ 311,ȱ324ȱn.85ȱ(2002)ȱ(listingȱexemplaryȱdecisionsȱillustratingȱuseȱ ofȱtheȱfortyȱpercentȱthreshold).ȱ 217.ȱSeeȱUnitedȱStatesȱv.ȱMicrosoftȱCorp.,ȱ253ȱF.3dȱ34,ȱ70–71ȱ(D.C.ȱCir.ȱ2001)ȱ(enȱ banc)ȱ(perȱcuriam)ȱ(findingȱlessȱthanȱfortyȱpercentȱforeclosureȱsufficientȱforȱantiȬ trustȱliability);ȱseeȱalsoȱDennisȱW.ȱCarlton,ȱAȱGeneralȱAnalysisȱofȱExclusionaryȱConȬ ductȱ andȱ Refusalȱ toȱ Deal—Whyȱ Aspenȱ andȱ Kodakȱ Areȱ Misguided,ȱ 68ȱ ANTITRUSTȱ L.J.ȱ 659,ȱ676,ȱ678ȱ(2001).ȱȱ 218.ȱAldenȱF.ȱAbbottȱ&ȱJoshuaȱD.ȱWright,ȱAntitrustȱanalysisȱofȱtyingȱarrangementsȱ andȱ exclusiveȱ dealing,ȱ inȱ 4ȱ ENCYCLOPEDIAȱ OFȱ LAWȱ ANDȱ ECONOMICS,ȱ ANTITRUSTȱ LAWȱANDȱECONOMICSȱ183,ȱ196ȱ(KeithȱN.ȱHyltonȱed.,ȱ2ndȱed.ȱ2010).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
65ȱ
notȱ depriveȱ theȱ rivalȱ ofȱ theȱ opportunityȱ toȱ operateȱ efficientlyȱ andȱcompetitionȱcannotȱbeȱharmed.ȱTheȱkeyȱpolicyȱimplicationȱ ofȱtheȱrequirementȱofȱsubstantialȱforeclosureȱisȱthatȱsoȱlongȱasȱaȱ sufficientȱnumberȱofȱdistributorȱcontractsȱbecomeȱavailableȱforȱ competitiveȱbiddingȱwithinȱaȱreasonableȱtimeȱperiod,ȱexclusiveȱ contractsȱareȱunlikelyȱtoȱgenerateȱcompetitiveȱharm.219ȱConsisȬ tentȱwithȱtheȱeconomicȱrequirementȱthatȱanȱexclusiveȱarrangeȬ mentȱforecloseȱaȱsubstantialȱshareȱofȱdistribution,ȱantitrustȱlawȱ hasȱlongȱrequiredȱplaintiffsȱtoȱdemonstrateȱsubstantialȱforecloȬ sureȱwithinȱaȱrelevantȱmarket.220ȱ TheȱD.C.ȱCircuit’sȱanalysisȱofȱMicrosoft’sȱexclusiveȱdealingȱarȬ rangementsȱwithȱInternetȱAccessȱProvidersȱ(IAPs)ȱandȱpersonalȱ computerȱ manufacturersȱ providesȱ aȱ recentȱ exampleȱ ofȱ modernȱ antitrustȱ analysisȱ ofȱ somewhatȱ similarȱ arrangements.ȱ Theȱ disȬ trictȱcourtȱconcludedȱthatȱMicrosoft’sȱdeȱfactoȱexclusiveȱdistribuȬ tionȱ contractsȱ didȱ notȱ violateȱ Sectionȱ 1ȱ ofȱ theȱ Shermanȱ Actȱ becauseȱ theyȱ foreclosedȱ lessȱ thanȱ fortyȱ percentȱ ofȱ theȱ market.ȱ Somewhatȱ puzzlingly,ȱ theȱ districtȱ courtȱ thenȱ foundȱ thatȱ theȱ sameȱ arrangementsȱ violatedȱ Sectionȱ 2ȱ ofȱ theȱ Shermanȱ Act.ȱ Theȱ D.C.ȱCircuitȱdidȱnotȱreverseȱtheȱdistrictȱcourtȱonȱitsȱrulingȱwithȱ respectȱtoȱSectionȱ1,ȱwhichȱtheȱplaintiffsȱdidȱnotȱchallenge,ȱbutȱitȱ upheldȱ theȱ districtȱ court’sȱ determinationȱ thatȱ theȱ contractsȱ vioȬ latedȱSectionȱ2.221ȱAlthoughȱtheȱagreementsȱforeclosedȱlessȱthanȱ fortyȱpercentȱofȱtheȱmarket,ȱtheȱD.C.ȱCircuitȱconcludedȱthatȱMiȬ crosoftȱ hadȱ violatedȱ theȱ antitrustȱ lawsȱ becauseȱ theȱ agreementsȱ “help[ed]ȱkeepȱusageȱofȱNavigatorȱbelowȱtheȱcriticalȱlevelȱnecesȬ saryȱforȱNavigatorȱorȱanyȱotherȱrivalȱtoȱposeȱaȱrealȱthreatȱtoȱMiȬ crosoft’sȱ monopoly.”222ȱ Asȱ such,ȱ theȱ D.C.ȱ Circuitȱ endorsedȱ aȱ distributionȬchannelȬspecificȱ formȱ ofȱ foreclosureȱ analysis.ȱ AsȬ signingȱdifferentȱweightsȱtoȱmoreȱefficientȱdistributionȱchannels,ȱ theȱcourtȱconcludedȱthatȱMicrosoftȱforeclosedȱoverȱfortyȱpercentȱ ofȱ “efficient”ȱ orȱ “effective”ȱ distribution.223ȱ Thus,ȱ anyȱ antitrustȱ analysisȱ ofȱ Google’sȱ exclusiveȱ syndicationȱ arrangementsȱ willȱ almostȱcertainlyȱrequire,ȱasȱaȱnecessaryȱbutȱnotȱsufficientȱcondiȬ tionȱ forȱ liability,ȱ theȱ plaintiffȱ toȱ demonstrateȱ thatȱ theȱ contractsȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 219.ȱSeeȱKlein,ȱsupraȱnoteȱ149,ȱatȱ122.ȱ 220.ȱSee,ȱforȱexample,ȱtheȱcasesȱandȱdiscussionȱinȱAREEDAȱ&ȱHOVENKAMP,ȱsupraȱ noteȱ214,ȱ¶ȱ1821;ȱJacobson,ȱsupraȱnoteȱ216,ȱatȱ324ȱn.85.ȱ 221.ȱMicrosoft,ȱ253ȱF.3dȱatȱ70–71.ȱȱ 222.ȱId.ȱatȱ71.ȱ 223.ȱSeeȱKlein,ȱsupraȱnoteȱ149,ȱatȱ127–28.ȱ
66ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
forecloseȱ atȱ leastȱ fortyȱ percentȱ ofȱ availableȱ inputs—inȱ thisȱ caseȱ searchȱ business—fromȱ rivals.ȱ Thisȱ approachȱ isȱ aȱ minimalȱ safeȬ guardȱtoȱensureȱthatȱantitrustȱliabilityȱisȱnotȱerroneouslyȱthrustȱ uponȱagreementsȱbetweenȱfirmsȱthatȱposeȱnoȱthreatȱtoȱcompetiȬ tionȱandȱlikelyȱproduceȱsubstantialȱbenefitsȱforȱconsumers.ȱ InȱtheȱcaseȱagainstȱGoogle,ȱitȱisȱunlikelyȱthatȱthisȱburdenȱcanȱ beȱ met.ȱ Likeȱ inȱ Microsoft,ȱ theseȱ searchȱ agreementsȱ allegedlyȱ preserveȱ Google’sȱ monopolyȱ positionȱ becauseȱ theyȱ blockȱ searchȱadvertisersȱfromȱobtainingȱtheȱcriticalȱmassȱofȱsearchȱtrafȬ ficȱthatȱisȱnecessaryȱtoȱcarryȱoutȱaȱviableȱsearchȱadvertisingȱplatȬ formȱ thatȱ couldȱ provideȱ someȱ competitiveȱ discipline.ȱ Theȱ anticompetitiveȱ theoryȱ isȱ that,ȱ inȱ theȱ sameȱ wayȱ thatȱ Microsoftȱ blockedȱbrowsersȱsuchȱasȱNetscapeȱfromȱaccessingȱanȱimportantȱ distributionȱchannelȱnecessaryȱtoȱgenerateȱusageȱlevelsȱcriticalȱtoȱ competeȱwithȱMicrosoftȱinȱtheȱoperatingȱsystemȱmarket,ȱGoogleȱ blockedȱorȱwillȱblockȱcompetingȱsearchȱadvertisersȱfromȱgainingȱ theȱrequisiteȱlevelȱofȱsearchȱtrafficȱnecessaryȱtoȱmaintainȱaȱviableȱ andȱcompetitiveȱsearchȱadvertisingȱplatform.ȱȱ Withoutȱhavingȱtheȱbenefitȱofȱtheȱdataȱnecessaryȱtoȱconductȱaȱ fullȬscaleȱ foreclosureȱ analysis,ȱ weȱ noteȱ severalȱ criticalȱ points.ȱ Theȱ firstȱ isȱ thatȱ theȱ burdenȱ liesȱ withȱ theȱ potentialȱ plaintiffȱ toȱ demonstrateȱ thatȱ theȱ foreclosureȱ levels,ȱ whetherȱ aboveȱ orȱ beȬ lowȱ conventionalȱ levelsȱ sufficientȱ toȱ surviveȱ summaryȱ judgȬ mentȱ inȱ anȱ exclusiveȱ dealingȱ case,ȱ areȱ sufficientȱ toȱ depriveȱ rivalsȱofȱtheȱchanceȱtoȱcompeteȱandȱachieveȱminimumȱefficientȱ scale.224ȱItȱisȱunclearȱwhatȱpercentageȱofȱtheȱrelevantȱmarketȱforȱ distributionȱGoogle’sȱexclusiveȱorȱpartialȱexclusiveȱsyndicationȱ agreementsȱcover.ȱTheȱpercentageȱmightȱwellȱbeȱquiteȱsmallȱinȱ lightȱofȱtheȱtotalȱnumberȱofȱInternetȱpublishers.ȱHowever,ȱitȱisȱ atȱ leastȱ theoreticallyȱ plausibleȱ thatȱ theȱ percentageȱ exceedsȱ theȱ fortyȬpercentȱthreshold.ȱȱ Aȱ Sectionȱ 2ȱ violationȱ wouldȱ haveȱ toȱ relyȱ onȱ theȱ tenousȱ preȬ sumptionȱthatȱcompetitorsȱneedȱaccessȱtoȱhighȬtrafficȱwebsitesȱtoȱ buildȱ scale.ȱ Inȱ realityȱ theȱ minimumȱ viableȱ scaleȱ isȱ likelyȱ quiteȱ small.ȱ Google’sȱ ownȱ history—growingȱ fromȱ aȱ tinyȱ startȬupȱ competingȱ withȱ aȱ Yahoo!ȱ behemoth—andȱ thatȱ ofȱ Microsoft’sȱ Bingȱsearchȱengineȱdemonstrateȱthatȱanȱoperationȱofȱcompetitiveȱ qualityȱcanȱbeȱobtainedȱwithȱrelativelyȱsmallȱinitialȱscale.ȱAtȱtheȱ sameȱ time,ȱ severalȱ startȬupȱ searchȱ engines—fromȱ Cuilȱ toȱ WolfȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 224.ȱSeeȱJacobson,ȱsupraȱnoteȱ216,ȱatȱ326.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
67ȱ
ramAlphaȱ toȱ theȱ perennialȱ competitor,ȱ Ask.com—haveȱ enteredȱ theȱ market,ȱ atȱ leastȱ believingȱ thatȱ theyȱ wereȱ ableȱ toȱ obtainȱ theȱ necessaryȱ scale,ȱ evenȱ thoughȱ noneȱ haveȱ beenȱ particularlyȱ sucȬ cessful.ȱSomeȱlargeȱcompetitorsȱhaveȱfailedȱtoȱcapitalizeȱonȱtheirȱ sizeȱ toȱ achieveȱ marketplaceȱ success—Lycosȱ andȱ AltaVista,ȱ forȱ example.ȱ Thus,ȱ itȱ isȱ unlikelyȱ thatȱ syndicationȱ agreementsȱ thatȱ hypotheticallyȱforecloseȱoverȱhalfȱofȱtheȱpotentialȱmarketȱforȱdisȬ tributionȱ (aȱ figureȱ weȱ findȱ implausible)ȱ wouldȱ beȱ sufficientȱ toȱ depriveȱrivalsȱofȱtheȱopportunityȱtoȱcompeteȱforȱsufficientȱdistriȬ butionȱtoȱachieveȱtheȱminimumȱefficientȱscale.ȱ Aȱ courtȱ could,ȱ asȱ theȱ D.C.ȱ Circuitȱ didȱ inȱ Microsoft,ȱ conductȱ aȱ narrow,ȱ distributionȬchannelȬspecificȱ foreclosureȱ analysisȱ andȱ comeȱ upȱ withȱ muchȱ largerȱ foreclosureȱ percentages.225ȱ Again,ȱ however,ȱsuchȱanȱapproachȱwouldȱnotȱnecessitateȱaȱdifferentȱreȬ sult.ȱForeclosureȱisȱaȱnecessaryȱbutȱnotȱsufficientȱconditionȱforȱliȬ ability.ȱ Althoughȱ failureȱ toȱ demonstrateȱ substantialȱ foreclosureȱ impliesȱthatȱantitrustȱliabilityȱisȱinappropriate,ȱaȱsuccessfulȱshowȬ ingȱ onlyȱ impliesȱ thatȱ furtherȱ analysisȱ ofȱ competitiveȱ effectsȱ isȱ wise.ȱ Theȱ criticalȱ questionȱ isȱ whetherȱ theȱ agreementsȱ preventȱ openȱ andȱ vigorousȱ competitionȱ forȱ distribution.ȱ Theȱ beliefȱ thatȱ foreclosureȱofȱaȱsingleȱchannelȱofȱdistributionȱisȱsufficientȱtoȱmainȬ tainȱmonopolyȱpowerȱisȱatȱleastȱpartiallyȱbeliedȱbyȱtheȱabilityȱofȱ theseȱsearchȱcompetitorsȱtoȱemployȱcreativeȱtacticsȱtoȱgainȱmarketȱ share.ȱProductȱdifferentiationȱisȱanȱobviousȱstrategy,ȱandȱseveralȱ searchȱenginesȱhaveȱemployedȱnovelȱtechnologiesȱinȱanȱeffortȱtoȱ distinguishȱthemselves.ȱSimilarly,ȱverticalȱsearchȱenginesȱcanȱgainȱ marketȱ shareȱ byȱ offeringȱ searchesȱ relevantȱ toȱ specializedȱ prodȬ ucts—Amazon,ȱ forȱ example,ȱ dominatesȱ searchesȱ forȱ books.ȱ FiȬ nally,ȱcompaniesȱlikeȱMicrosoftȱhaveȱattemptedȱtoȱgainȱaccessȱtoȱ endȱ usersȱ byȱ attemptingȱ theirȱ ownȱ exclusiveȱ arrangements— arrangementsȱ that,ȱ becauseȱ ofȱ theȱ internalizationȱ ofȱ indirectȱ netȬ workȱ effects,ȱ areȱ economicallyȱ viableȱ forȱ theirȱ counterparties.ȱ Moreover,ȱ Microsoftȱ hasȱ aȱ substantialȱ channelȱ ofȱ distributionȱ throughȱ itsȱ ownȱ operatingȱ systemȱ andȱ otherȱ products.226ȱ Weȱ thereforeȱtentativelyȱconclude,ȱwithoutȱhavingȱaccessȱtoȱdataȱsufȬ ficientȱ toȱ conductȱ ourȱ ownȱ completeȱ foreclosureȱ analysis,ȱ thatȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 225.ȱMicrosoft,ȱ253ȱF.3dȱatȱ70–71.ȱ 226.ȱAlthough,ȱMicrosoftȱitselfȱprobablyȱmakesȱsomeȱofȱtheȱtypesȱofȱprocompetiȬ tiveȱarrangementsȱthatȱMicrosoftȱwouldȱenterȱintoȱillegal—aȱtestamentȱtoȱtheȱerrorȱ costsȱ inherentȱ inȱ thatȱ caseȱ inȱ lightȱ ofȱ theȱ thenȬunforeseenȱ competitionȱ betweenȱ GoogleȱandȱMicrosoft.ȱ
68ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
Google’sȱexclusiveȱsyndicationȱagreementsȱareȱ notȱ likelyȱ toȱgenȬ erateȱforeclosureȱsufficientȱtoȱdepriveȱrivalsȱofȱtheȱopportunityȱtoȱ competeȱforȱdistribution.ȱ Theȱfinalȱkeyȱcompetitiveȱquestionȱandȱaȱrelatedȱoneȱisȱwhetherȱ theȱagreementsȱatȱissueȱcollectivelyȱforecloseȱrivalsȱfromȱaccessȱtoȱ criticalȱtrafficȱforȱlongȱenoughȱtoȱproduceȱanticompetitiveȱeffects.ȱ Exclusiveȱ agreementsȱ ofȱ shortȱ durationȱ typicallyȱ doȱ notȱ raiseȱ competitiveȱconcerns,ȱbecauseȱrivalsȱareȱcontractuallyȱprohibitedȱ fromȱaccessȱtoȱtheȱcompetitiveȱprocessȱforȱdistribution—deprivedȱ theȱ opportunityȱ toȱ compete—onlyȱ duringȱ theȱ shortȱ periodȱ theȱ agreementȱisȱinȱforce.ȱWhenȱexclusiveȱcontractsȱareȱofȱshortȱduraȬ tion,ȱ Google’sȱ rivalsȱ haveȱ theȱ opportunity,ȱ likeȱ Yahoo!ȱ did,ȱ toȱ competeȱforȱcontractsȱwithȱAOLȱandȱothers.ȱIndeed,ȱconventionalȱ antitrustȱanalysisȱofȱexclusiveȱdealingȱarrangementsȱdictatesȱthatȱ agreementsȱofȱlessȱthanȱoneȱyearȱandȱterminableȱatȱwillȱareȱpreȬ sumptivelyȱ lawful,227ȱ andȱ agreementsȱ longerȱ thanȱ oneȱ yearȱ butȱ stillȱofȱshortȱdurationȱareȱlessȱlikelyȱtoȱresultȱinȱcompetitiveȱharm.ȱ Itȱ isȱ unclearȱ whetherȱ theȱ syndicationȱ agreementsȱ areȱ sufficientlyȱ shortȱ inȱ durationȱ toȱ beȱ viewedȱ asȱ presumptivelyȱ lawfulȱ underȱ currentȱexclusiveȱdealingȱlaw.ȱTheseȱagreementsȱdoȱnotȱseemȱtooȱ harmful,ȱhowever,ȱasȱtheȱvastȱmajorityȱofȱsyndicationȱagreementsȱ expireȱinȱlessȱthanȱthreeȱyearsȱandȱareȱstaggeredȱ(ratherȱthanȱallȱ comingȱupȱforȱrenewalȱatȱtheȱsameȱtime).ȱRivalsȱareȱthereforeȱconȬ tinuouslyȱaffordedȱampleȱopportunityȱtoȱofferȱcompetitiveȱterms.ȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 227.ȱAȱ numberȱ ofȱ courtsȱ haveȱ heldȱ thatȱ exclusiveȱ contractsȱ ofȱ oneȱ yearȱ orȱ lessȱ areȱ presumptivelyȱ lawful.ȱ See,ȱ e.g.,ȱ Concordȱ Boatȱ Corp.ȱ v.ȱ Brunswickȱ Corp.,ȱ 207ȱ F.3dȱ1039,ȱ1059ȱ(8thȱCir.ȱ2000);ȱCDCȱTechs.,ȱInc.ȱv.ȱIDEXXȱLabs,ȱInc.,ȱ186ȱF.3dȱ74,ȱ 81ȱ(2dȱCir.ȱ1999);ȱOmegaȱEnvtl.ȱInc.ȱv.ȱGilbarco,ȱInc.,ȱ127ȱF.3dȱ1157,ȱ1163–64ȱ(9thȱ Cir.ȱ1997);ȱPaddockȱPubl’ns,ȱInc.ȱv.ȱChicagoȱTribuneȱCo.,ȱ103ȱF.3dȱ42,ȱ47ȱ(7thȱCir.ȱ 1996)ȱ(“[T]heȱFTCȱandȱtheȱSupremeȱCourtȱconcludedȱthatȱevenȱexclusiveȱdealingȱ contractsȱareȱlawfulȱifȱlimitedȱtoȱoneȱyear’sȱduration.”);ȱThompsonȱEverett,ȱInc.ȱv.ȱ Nat’lȱCableȱAdver.ȱL.P.,ȱ57ȱF.3dȱ1317,ȱ1324–25ȱ(4thȱCir.ȱ1995);ȱU.S.ȱHealthcare,ȱInc.ȱ v.ȱHealthsource,ȱInc.,ȱ986ȱF.2dȱ589,ȱ596ȱ(1stȱCir.ȱ1993);ȱRolandȱMach.ȱCo.ȱv.ȱDresserȱ Indus.ȱInc.,ȱ749ȱF.2dȱ380,ȱ395ȱ(7thȱCir.ȱ1984).ȱSimilarly,ȱsomeȱcommentatorsȱhaveȱ arguedȱ inȱ favorȱ ofȱ perȱ seȱ legalityȱ forȱ suchȱ shortȬtermȱ contracts.ȱ See,ȱ e.g.,ȱ Wright,ȱ supraȱ noteȱ 210,ȱ atȱ 203.ȱ Aȱ handfulȱ ofȱ courtsȱ haveȱ supportedȱ thisȱ proposition.ȱ See,ȱ e.g.,ȱUnitedȱStatesȱv.ȱDentsplyȱInt’l,ȱInc.,ȱNo.ȱCIV.A.99Ȭ005ȬSLR,ȱ2001ȱWLȱ624807,ȱ atȱ*8ȱ(D.ȱDel.ȱMar.ȱ31,ȱ2001);ȱMinn.ȱMiningȱ&ȱMfg.ȱCo.ȱv.ȱAppletonȱPapersȱInc.,ȱ35ȱ F.ȱSupp.ȱ2dȱ1138,ȱ1144ȱ(D.ȱMinn.ȱ1999)ȱ(“3MȱhasȱproducedȱevidenceȱthatȱAppleȬ ton’sȱ soleȬsourcingȱ agreementsȱ oftenȱ includeȱ incentivesȱ thatȱ haveȱ theȱ practicalȱ effectȱofȱtyingȱupȱtheȱpaperȱsheetȱinventoryȱofȱaȱmerchantȱoverȱaȱperiodȱofȱseveralȱ years.”);ȱUnitedȱStatesȱv.ȱDairymen,ȱInc.,ȱNo.ȱ7634A,ȱ1983ȱWLȱ1899,ȱatȱ*1–2ȱ(W.D.ȱ Ky.ȱOct.ȱ26,ȱ1983)ȱ(enjoiningȱrequirementsȱcontractsȱcoveringȱlargeȱpercentageȱofȱ theȱmarketȱthoughȱonlyȱthirtyȱdaysȱtoȱoneȱyearȱinȱduration),ȱaff’dȱperȱcuriam,ȱ758ȱ F.2dȱ654ȱ(6thȱCir.ȱ1985).ȱȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ H.
69ȱ
QualityȱScoresȱ
Theȱ moreȱ interestingȱ andȱ novelȱ antitrustȱ questionsȱ relateȱ toȱ theȱrelativelyȱmoreȱinnovativeȱpracticeȱofȱusingȱqualityȱscoringȱ toȱ adjustȱ AdWordsȱ searchȱ auctions.ȱ Googleȱ isȱ allegedȱ toȱ emȬ ployȱ itsȱ qualityȱ score—whichȱ rivalsȱ complainȱ itȱ hasȱ keptȱ seȬ cret—toȱ precludeȱ accessȱ byȱ competitorsȱ toȱ itsȱ topȱ searchȱ results,ȱ andȱ toȱ increaseȱ theȱ paymentsȱ requiredȱ ofȱ competitorsȱ forȱtopȱplacement.228ȱInȱanȱeffortȱtoȱmatchȱtheȱfactsȱofȱAspenȱSkiȬ ing,ȱ moreover,ȱ theȱ TradeCometȱ complaintȱ allegesȱ thatȱ Googleȱ withdrewȱ fromȱ aȱ voluntary,ȱ profitableȱ ventureȱ throughȱ maȬ nipulationȱofȱitsȱqualityȱscores.ȱ229ȱ Theȱ appropriateȱ antitrustȱ questionȱ raisedȱ byȱ theseȱ comȬ plaintsȱ isȱ whetherȱ Sectionȱ 2ȱ imposesȱ uponȱ Google,ȱ assumingȱ thatȱitȱisȱaȱmonopolist,ȱaȱdutyȱtoȱdealȱwithȱitsȱrivals.ȱTheȱantiȬ trustȱ lawsȱ onlyȱ rarelyȱ imposeȱ aȱ dutyȱ toȱ dealȱ onȱ businessȱ firms.230ȱInȱTrinko,ȱtheȱSupremeȱCourtȱreaffirmedȱthatȱasȱaȱgenȬ eralȱ matter,ȱ theȱ antitrustȱ lawsȱ doȱ notȱ imposeȱ aȱ dutyȱ toȱ dealȱ withȱrivals.231ȱHowever,ȱtheȱSupremeȱCourtȱalsoȱidentifiedȱnarȬ rowȱconditionsȱ“atȱtheȱboundary”ȱofȱSectionȱ2ȱlawȱunderȱwhichȱ antitrustȱlawȱwillȱimposeȱsuchȱaȱduty.232ȱ Inȱ Aspenȱ Skiing,ȱ theȱ Supremeȱ Courtȱ heldȱ thatȱ aȱ skiȱ areaȱ opȬ eratorȱ violatedȱ theȱ antitrustȱ lawsȱ byȱ refusingȱ toȱ continueȱ aȱ jointȬticketȱ ventureȱ withȱ aȱ neighboringȱ operator.233ȱ Underȱ theȱ agreement,ȱ theȱ partiesȱ issuedȱ joint,ȱ multidayȱ liftȱ ticketsȱ thatȱ couldȱbeȱusedȱatȱeachȱofȱtheȱareasȱskiȱfacilities.ȱInȱfindingȱthatȱ thereȱ wasȱ sufficientȱ evidenceȱ toȱ supportȱ antitrustȱ liability,ȱ theȱ Courtȱ focusedȱ onȱ theȱ offendingȱ operator’sȱ willingnessȱ toȱ terȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 228.ȱSeeȱDanielȱLyons,ȱTheyȱMightȱBeȱaȱLittleȱEvil,ȱNEWSWEEK,ȱJuneȱ1,ȱ2009,ȱatȱ24;ȱ JoeȱNocera,ȱStuckȱinȱGoogle’sȱDoghouse,ȱN.Y.ȱTIMES,ȱSept.ȱ13,ȱ2008,ȱatȱC1.ȱ 229.ȱCompareȱ Complaint,ȱ supraȱ noteȱ 134ȱ ¶ȱ 8,ȱ withȱ Aspenȱ Skiingȱ Co.ȱ v.ȱ Aspenȱ HighlandsȱSkiingȱCorp.,ȱ472ȱU.S.ȱ585,ȱ601ȱ(1985).ȱȱ 230.ȱSee,ȱe.g.,ȱUnitedȱStatesȱv.ȱColgateȱ&ȱCo.,ȱ250ȱU.S.ȱ300,ȱ307ȱ(1919)ȱ(notingȱthatȱ antitrustȱlawsȱtypicallyȱdoȱnotȱ“restrictȱtheȱlongȱrecognizedȱrightȱofȱ[a]ȱtraderȱorȱ manufacturerȱ engagedȱ inȱ anȱ entirelyȱ privateȱ business,ȱ freelyȱ toȱ exerciseȱ hisȱ ownȱ independentȱdiscretionȱasȱtoȱpartiesȱwithȱwhomȱheȱwillȱdeal”).ȱTheȱrightȱtoȱrefuseȱ toȱdealȱwithȱrivalsȱisȱnotȱabsolute,ȱhowever,ȱbutȱitȱisȱclose.ȱseeȱalsoȱAspenȱSkiing,ȱ472ȱ U.S.ȱatȱ601ȱ(“[T]heȱhighȱvalueȱ.ȱ.ȱ.ȱplacedȱonȱtheȱrightȱtoȱrefuseȱtoȱdealȱwithȱotherȱ firmsȱdoesȱnotȱmeanȱthatȱtheȱrightȱisȱunqualified.”).ȱSeeȱgenerallyȱVerizonȱComm.ȱ Inc.ȱv.ȱLawȱOfficesȱofȱCurtisȱV.ȱTrinko,ȱLLP,ȱ540ȱU.S.ȱ398,ȱ407ȱ(2004).ȱ 231.ȱTrinko,ȱ540ȱU.S.ȱatȱ408.ȱ 232.ȱId.ȱatȱ409.ȱ 233.ȱAspenȱSkiing,ȱ472ȱU.S.ȱatȱ608.ȱ
70ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
minateȱaȱvoluntaryȱandȱprofitableȱbusinessȱrelationship.234ȱTheȱ CourtȱobservedȱthatȱtheȱoffendingȱoperatorȱpersistedȱinȱtermiȬ natingȱtheȱjointȬticketȱventureȱevenȱafterȱtheȱcompetitorȱofferedȱ toȱ payȱ fullȱ retailȱ priceȱ forȱ theȱ ticketsȱ inȱ orderȱ toȱ continueȱ theȱ arrangement.ȱRelyingȱonȱtheseȱfacts,ȱtheȱCourtȱconcludedȱ thatȱ suchȱ conductȱ suggestedȱ thatȱ theȱ offendingȱ skiȱ operatorȱ wasȱ willingȱtoȱforgoȱshortȬtermȱprofitsȱforȱfutureȱmonopolyȱprices.ȱ Asȱ aȱ result,ȱ theȱ courtȱ determinedȱ thatȱ theȱ refusalȱ toȱ dealȱ wasȱ anticompetitiveȱconductȱaimedȱatȱpreservingȱaȱmonopoly.ȱȱ Theȱ Supremeȱ Court’sȱ latestȱ wordȱ onȱ theȱ dutyȱ toȱ dealȱ limitsȱ theȱdutyȱtoȱanȱextremelyȱnarrowȱsetȱofȱcircumstances:ȱ Firmsȱ mayȱ acquireȱ monopolyȱ powerȱ byȱ establishingȱ anȱinfraȬ structureȱthatȱrendersȱthemȱuniquelyȱsuitedȱtoȱserveȱtheirȱcusȬ tomers.ȱ Compellingȱ suchȱ firmsȱ toȱ shareȱ theȱ sourceȱ ofȱ theirȱ advantageȱ isȱ inȱ someȱ tensionȱ withȱ theȱ underlyingȱ purposeȱ ofȱ antitrustȱlaw,ȱsinceȱitȱmayȱlessenȱtheȱincentiveȱforȱtheȱmonopoȬ list,ȱtheȱrival,ȱorȱbothȱtoȱinvestȱinȱthoseȱeconomicallyȱbeneficialȱ facilities.ȱEnforcedȱsharingȱalsoȱrequiresȱantitrustȱcourtsȱtoȱactȱ asȱcentralȱplanners,ȱidentifyingȱtheȱproperȱprice,ȱquantity,ȱandȱ otherȱ termsȱ ofȱ dealing—aȱ roleȱ forȱ whichȱ theyȱ areȱ illȱ suited.ȱ Moreover,ȱ compellingȱ negotiationȱ betweenȱ competitorsȱ mayȱ facilitateȱ theȱ supremeȱ evilȱ ofȱ antitrust:ȱ collusion.ȱ Thus,ȱ asȱ aȱ generalȱmatter,ȱtheȱShermanȱActȱ“doesȱnotȱrestrictȱtheȱlongȱrecȬ ognizedȱrightȱofȱ[a]ȱtraderȱorȱmanufacturerȱengagedȱinȱanȱenȬ tirelyȱprivateȱbusiness,ȱfreelyȱtoȱexerciseȱhisȱownȱindependentȱ discretionȱasȱtoȱpartiesȱwithȱwhomȱheȱwillȱdeal.”235ȱ
TheȱCourtȱwarnedȱthatȱtheȱimpositionȱofȱaȱdutyȱtoȱdealȱwouldȱ threatenȱtoȱ“lessenȱtheȱincentiveȱforȱtheȱmonopolist,ȱtheȱrival,ȱorȱ bothȱtoȱinvestȱinȱ.ȱ.ȱ.ȱeconomicallyȱbeneficialȱfacilities.”236ȱ“Refusalȱ toȱ deal”ȱ antitrustȱ jurisprudenceȱ hasȱ beenȱ heavilyȱ criticizedȱ byȱ commentators,237ȱandȱoffersȱbusinessȱfirmsȱlittleȱinȱtheȱwayȱofȱadȬ vanceȱ knowledgeȱ regardingȱ whetherȱ businessȱ decisionsȱ violateȱ theȱantitrustȱlaws.ȱBecauseȱimpositionȱofȱaȱdutyȱtoȱdealȱwithȱrivalsȱ threatensȱ toȱ decreaseȱ theȱ incentiveȱ toȱ innovateȱ byȱ creatingȱ newȱ waysȱofȱproducingȱgoodsȱatȱlowerȱcosts,ȱsatisfyingȱconsumerȱdeȬ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 234.ȱId.ȱatȱ610–11.ȱ 235.ȱTrinko,ȱ540ȱU.S.ȱatȱ407–08ȱ(citingȱColgate,ȱ250ȱU.S.ȱatȱ307).ȱ 236.ȱId.ȱ 237.ȱSee,ȱe.g.,ȱRonaldȱA.ȱCassȱ&ȱKeithȱN.ȱHylton,ȱPreservingȱCompetition:ȱEconomicȱ Analysis,ȱLegalȱStandardsȱandȱMicrosoft,ȱ8ȱGEO.ȱMASONȱL.ȱREV.ȱ1,ȱ27ȱ(1999);ȱFrankȱH.ȱ Easterbrook,ȱOnȱIdentifyingȱExclusionaryȱConduct,ȱ61ȱNOTREȱ DAMEȱ L.ȱ REV.ȱ972,ȱ973ȱ (1986).ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
71ȱ
mand,ȱ orȱ creatingȱ newȱ marketsȱ altogether,ȱ courtsȱ andȱ antitrustȱ agenciesȱhaveȱbeenȱreluctantȱtoȱexpandȱtheȱduty.ȱ Despiteȱthisȱreluctance,ȱtheȱTradeCometȱcomplaintȱcontendsȱthatȱ Google’sȱ decisionȱ toȱ implementȱ aȱ qualityȱ metricȱ toȱ effectivelyȱ terminateȱ earlierȱ dealingsȱ withȱ competitorsȱ moreȱ closelyȱ resemȬ blesȱ theȱ circumstancesȱ presentedȱ inȱ Aspenȱ Skiingȱ thanȱ thoseȱ inȱ Trinko,ȱandȱthusȱpresentsȱtheȱrareȱcircumstanceȱwarrantingȱimpoȬ sitionȱofȱaȱdutyȱtoȱdealȱunderȱSectionȱ2.ȱTheȱkeyȱallegationȱisȱthatȱ Googleȱ manipulatesȱ theȱ qualityȱ scoreȱ generatedȱ byȱ itsȱ qualityȱ scoreȱmethodology,ȱallowingȱGoogleȱtoȱadjustȱwhereȱamongȱtheȱ sponsoredȱlinksȱAdWordsȱwillȱplaceȱanȱadvertisementȱandȱwhatȱ amountȱ mustȱ beȱ bidȱ toȱ secureȱ aȱ topȱ placement.ȱ Accordingȱ toȱ TradeComet,ȱ thisȱ allowsȱ Googleȱ arbitrarilyȱtoȱ chargeȱadvertisersȱ higherȱ pricesȱ forȱ theȱ sameȱ placementȱ irrespectiveȱ ofȱ theȱ adverȬ tiser’sȱkeywordȱauctionȱbids.ȱTheȱcomplaintȱcontemplatesȱthatȱinȱ extremeȱ cases,ȱ Googleȱ couldȱ chargeȱ arbitrarilyȱ highȱ pricesȱ suffiȬ cientȱ toȱ resultȱ inȱ aȱ deȱ factoȱ refusalȱ toȱ dealȱ withȱ rivals.238ȱ TradeȬ CometȱallegesȱthatȱGoogleȱemployedȱthisȱtypeȱofȱstrategyȱonceȱitsȱ verticalȱsearchȱengineȱrival,ȱSourceTool,ȱstartedȱtoȱenjoyȱsuccessȱinȱ theȱsearchȱadvertisingȱmarket.239ȱȱ Google’sȱ useȱ ofȱ itsȱ ownȱ qualityȱ scoresȱ doesȱ not,ȱ however,ȱ createȱanȱantitrustȱdutyȱtoȱdeal.ȱTradeCometȱprecariouslyȱjustiȬ fiesȱitsȱclaimȱbyȱallegingȱthatȱGoogleȱandȱTradeCometȱonceȱenȬ teredȱintoȱaȱvoluntaryȱandȱprofitableȱdeal.ȱTradeCometȱallegesȱ thatȱchangesȱtoȱtheȱtermsȱofȱthatȱdeal,ȱsuchȱasȱanȱincreaseȱinȱtheȱ priceȱ charged,ȱ implyȱ theȱ typeȱ ofȱ shortȬtermȱ sacrificeȱ ofȱ profitsȱ atȱworkȱinȱAspenȱSkiing.ȱWeȱareȱnotȱpersuaded.ȱTheȱreasonsȱforȱ rejectingȱ antitrustȬbasedȱ dutiesȱ toȱ dealȱ citedȱ byȱ theȱ Courtȱ inȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 238.ȱComplaint,ȱsupraȱnoteȱ134,ȱ¶ȱ78.ȱ 239.ȱWeȱ doȱ notȱ separatelyȱ discussȱ thisȱ claimȱ asȱ anȱ essentialȱ facilityȱ claimȱ bothȱ becauseȱ theȱ Supremeȱ Courtȱ hasȱ refusedȱ toȱ endorseȱ suchȱ aȱ claim,ȱ seeȱ Trinko,ȱ 540ȱ U.S.ȱ atȱ 410,ȱ andȱ becauseȱ thereȱ isȱ nearȱ universalȱ agreementȱ fromȱ commentatorsȱ thatȱ itȱ shouldȱ beȱ abandoned.ȱ See,ȱ e.g.,ȱ 3Aȱ PHILLIPȱ E.ȱ AREEDAȱ &ȱ HERBERTȱ HOVENKAMP,ȱANTITRUSTȱLAW,ȱ¶ȱ771c,ȱatȱ196ȱ(3dȱed.ȱ2008)ȱ(notingȱthatȱ“theȱessenȬ tialȱfacilityȱdoctrineȱisȱbothȱharmfulȱandȱunnecessaryȱandȱshouldȱbeȱabandoned”);ȱ MichaelȱBoudin,ȱAntitrustȱDoctrineȱandȱtheȱSwayȱofȱMetaphor,ȱ75ȱGEO.ȱ L.J.ȱ395,ȱ402ȱ (1986)ȱ(notingȱ“embarrassingȱweakness”ȱofȱessentialȱfacilitiesȱdoctrine);ȱAbbottȱB.ȱ Lipsky,ȱJr.ȱ&ȱJ.ȱGregoryȱSidak,ȱEssentialȱFacilities,ȱ51ȱSTAN.ȱL.ȱREV.ȱ1187,ȱ1195ȱ(1999)ȱ (“[M]andatoryȱaccessȱremedies,ȱsuchȱasȱtheȱessentialȱfacilitiesȱdoctrine,ȱdoȱnotȱfitȱ comfortablyȱwithinȱantitrustȱlaw.”);ȱGregoryȱJ.ȱWerden,ȱTheȱLawȱandȱEconomicsȱofȱ theȱ Essentialȱ Facilityȱ Doctrine,ȱ 32ȱ ST.ȱ LOUISȱ U.ȱ L.J.ȱ 433,ȱ 480ȱ (1987)ȱ (“Courtsȱ shouldȱ rejectȱtheȱdoctrine”).ȱ
72ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
Trinkoȱandȱadvancedȱbyȱandȱleadingȱcommentatorsȱallȱmilitateȱ inȱfavorȱofȱrejectingȱsuchȱanȱallegation.ȱȱ First,ȱevenȱtakingȱtheȱallegedȱfactsȱasȱtrue,ȱthereȱisȱnoȱreasonȱ toȱbelieveȱthatȱaȱcourseȱofȱconductȱthatȱwasȱonceȱprofitableȱreȬ mainsȱ soȱ indefinitely.ȱ Marketȱ conditionsȱ change,ȱ andȱ suchȱ aȱ ruleȱ wouldȱ produceȱ perniciousȱ incentiveȱ effects.ȱ Aȱ ruleȱ thatȱ exposesȱinnovativeȱfirmsȱtoȱSectionȱ2ȱliabilityȱandȱtrebleȱdamȬ agesȱforȱinterruptingȱorȱterminatingȱaȱcourseȱofȱdealingȱthreatȬ ensȱ toȱ lessenȱ theȱ incentiveȱ toȱ innovateȱ andȱ enterȱ intoȱ agreementsȱtoȱcommercializeȱinnovationȱinȱtheȱfirstȱinstance— particularlyȱ becauseȱ theȱ innovator’sȱ incentivesȱ toȱ enterȱ intoȱ agreementsȱthatȱspotlightȱitsȱinnovationȱchangeȱoverȱtimeȱwithȱ increasedȱ consumerȱ awarenessȱ ofȱ theȱ innovation.ȱ Itȱ isȱ forȱ thisȱ reasonȱthatȱseveralȱcommentatorsȱatȱtheȱrecentȱSectionȱ2ȱHearȬ ingsȱconcludedȱthatȱterminationȱofȱanȱearlierȱcourseȱofȱdealingȱ shouldȱnotȱbeȱaȱsignificantȱfactorȱinȱassessingȱwhetherȱtheȱantiȬ trustȱlawsȱimposeȱaȱdutyȱtoȱdeal.240ȱ Secondȱ(againȱassumingȱGoogleȱisȱaȱmonopolistȱandȱhasȱforȬ sakenȱshortȬtermȱprofitsȱtoȱrefuseȱtoȱdealȱwithȱrivals),ȱimposingȱ aȱ dutyȱ toȱ dealȱ isȱ notȱ likelyȱ toȱ improveȱ mattersȱ becauseȱ ofȱ theȱ difficultiesȱ ofȱ craftingȱ andȱ enforcingȱ aȱ remedy.ȱ Asȱ theȱ Courtȱ notedȱ inȱ Trinko,ȱ “enforcedȱ sharingȱ.ȱ.ȱ.ȱrequiresȱ antitrustȱ courtsȱ toȱ actȱ asȱ centralȱ planners,ȱ identifyingȱ theȱ properȱ price,ȱ quanȬ tity,ȱ andȱ otherȱ termsȱ ofȱ dealing—aȱ roleȱ forȱ whichȱ theyȱ areȱ illȱ suited.”241ȱ Theȱ Antitrustȱ Modernizationȱ Commissionȱ recentlyȱ reachedȱaȱsimilarȱconclusion,242ȱjoiningȱtheȱgrowingȱconsensusȱ ofȱ commentators,ȱ suchȱ asȱ Judgeȱ Posner,ȱ whoȱ haveȱ concludedȱ thatȱ“itȱcannotȱbeȱsoundȱantitrustȱlawȱthat,ȱwhenȱCongressȱreȬ fusesȱorȱomitsȱtoȱregulateȱsomeȱaspectȱofȱaȱnaturalȱmonopolist’sȱ behavior,ȱtheȱantitrustȱcourtȱwillȱstepȱinȱand,ȱbyȱdecree,ȱsupplyȱ theȱmissingȱregulatoryȱregime.”243ȱȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 240.ȱU.S.ȱ DEP’Tȱ OFȱ JUSTICE,ȱ COMPETITIONȱ ANDȱ MONOPOLY:ȱ SINGLEȬFIRMȱ CONDUCTȱ UNDERȱ SECTIONȱ 2ȱ OFȱ THEȱ SHERMANȱ ACTȱ 42ȱ n.64ȱ (2008),ȱ availableȱ atȱ http://www.usdoj.gov/atr/public/reports/236681.pdf.ȱ 241.ȱTrinko,ȱ540ȱU.S.ȱatȱ408.ȱ 242.ȱANTITRUSTȱ MODERNIZATIONȱ COMM’N,ȱ REPORTȱANDȱ RECOMMENDATIONȱ 102ȱ (2007),ȱ availableȱ atȱ http://govinfo.library.unt.edu/amc/report_recommendation/ȱ amc_final_report.pdfȱ(“[F]orcedȱsharingȱrequiresȱcourtsȱtoȱdetermineȱtheȱpriceȱatȱ whichȱ suchȱ sharingȱ mustȱ takeȱ place,ȱ therebyȱ transformingȱ antitrustȱ courtsȱ intoȱ priceȱregulators,ȱaȱroleȱforȱwhichȱtheyȱareȱillȱsuited.”).ȱ 243.ȱPOSNER,ȱsupraȱnoteȱ10,ȱatȱ243–44.ȱ
No.ȱ1]ȱ
GoogleȱandȱtheȱLimitsȱofȱAntitrustȱ
73ȱ
Third,ȱ andȱ mostȱ importantly,ȱ evenȱ theȱ mostȱ aggressiveȱ inȬ terpretationsȱ ofȱ Aspenȱ Skiing,ȱ andȱ theȱ mostȱ enthusiasticȱ supȬ portersȱofȱaȱlimitedȱantitrustȱdutyȱtoȱdeal,ȱconcedeȱthatȱrefusalȱ toȱdealȱisȱentirelyȱappropriateȱifȱthereȱisȱaȱcompetitiveȱjustificaȬ tionȱ forȱ theȱ conductȱ atȱ issue.ȱ Inȱ thisȱ case,ȱ theȱ argumentȱ thatȱ Google’sȱ qualityȱ scoresȱ areȱ withoutȱ competitiveȱ meritȱ isȱ misȬ leadingȱandȱleadsȱtoȱperverseȱantitrustȱresults.ȱGoogle’sȱqualityȱ scoreȱ metricȱ isȱ anȱ innovativeȱ andȱ effectiveȱ algorithmȱ forȱ preȬ dictingȱ clickthroughȱ ratesȱ andȱ facilitatingȱ efficientȱ pricing.244ȱ Thatȱtheȱdeviceȱisȱusedȱbyȱeveryȱgeneralȱpurposeȱsearchȱengineȱ forȱ theȱ sameȱ purposeȱ furtherȱ suggestsȱ thatȱ itsȱ functionȱ isȱ proȬ competitive.ȱComplaintsȱaboutȱtheȱsecrecyȱofȱtheȱalgorithmȱareȱ aȱ redȱ herringȱ fromȱ anȱ antitrustȱ perspective.ȱ Noȱ businessȱ firm,ȱ evenȱaȱmonopolist,ȱhasȱanȱantitrustȱdutyȱtoȱrevealȱtoȱcompetiȬ torsȱformulasȱthatȱitȱusesȱtoȱsetȱprices.ȱFurther,ȱthereȱisȱanȱobviȬ ousȱ procompetitiveȱ justificationȱ forȱ keepingȱ theȱ qualityȱ scoreȱ metricȱ secret:ȱ Google’sȱ successȱ inȱ matchingȱ keywordsȱ toȱ adsȱ willȱbeȱcompromisedȱbyȱdisclosureȱofȱtheȱalgorithmȱbecauseȱitȱ wouldȱopenȱopportunitiesȱtoȱgameȱtheȱauctionȱprocess.245ȱȱ Butȱ thereȱ isȱ aȱ moreȱ fundamentalȱ point:ȱ Unitedȱ Statesȱ antiȬ trustȱlawȱnotȱonlyȱdoesȱnotȱcondemnȱGoogle’sȱabilityȱtoȱchargeȱ efficientȱ pricesȱ forȱ itsȱ servicesȱ throughȱ theȱ auction,ȱ itȱ encourȬ agesȱit.ȱEvenȱifȱaȱpotentialȱantitrustȱplaintiffȱcouldȱdemonstrateȱ ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 244.ȱSeeȱPartȱIII;ȱAtheyȱ&ȱEllison,ȱsupraȱnoteȱ123,ȱatȱ37.ȱ 245.ȱTheȱcomplaintȱoverȱGoogle’sȱrefusalȱtoȱcompletelyȱdiscloseȱitsȱpricingȱalgoȬ rithmȱ isȱ relatedȱ toȱ recentȱ attemptsȱ toȱ incorporateȱ privacyȱ issuesȱ intoȱ antitrustȱ analysis.ȱOnȱtheȱoneȱhand,ȱitȱisȱuncontroversialȱthatȱprivacyȱcanȱbeȱaȱformȱofȱnonȬ priceȱ competitionȱ andȱ thusȱ fallsȱ withinȱ theȱ domainȱ ofȱ theȱ antitrustȱ lawsȱ inȱ theȱ sameȱwayȱthatȱaȱcartelȱbetweenȱrivalsȱtoȱrefuseȱtoȱcompeteȱoverȱstoreȱhoursȱorȱfreeȱ parkingȱwouldȱbeȱillegal.ȱConventionalȱantitrustȱanalysisȱisȱsufficientlyȱflexibleȱtoȱ adaptȱ toȱ suchȱ concernsȱ whereȱ appropriate.ȱ Theseȱ arguments,ȱ however,ȱ seemȱ toȱ haveȱ noȱ applicationȱ here.ȱ Modernȱ mergerȱ analysisȱ requiresȱ oneȱ toȱ demonstrateȱ howȱtheȱmergerȱchangesȱtheȱincentivesȱofȱpartiesȱtoȱcompeteȱonȱprivacy.ȱWeȱhaveȱ notȱ seenȱ anyȱ proponentsȱ ofȱ increasedȱ scrutinyȱ ofȱ privacyȱ concernsȱ inȱ mergerȱ analysisȱ provideȱ anȱ explanationȱ forȱ whyȱ aȱ mergerȱ wouldȱ changeȱ incentivesȱ ofȱ firmsȱtoȱcompeteȱonȱprivacy.ȱWhateverȱtheȱevidenceȱsupportingȱtheȱrelationshipȱ betweenȱ marketȱ concentrationȱ andȱ priceȱ underlyingȱ someȱ ofȱ modernȱ mergerȱ analysis,ȱweȱareȱawareȱofȱnoȱevidenceȱthatȱsuchȱaȱrelationshipȱexistsȱbetweenȱconȬ centrationȱofȱdataȱandȱprivacyȱcompetition.ȱTheȱanalogousȱmonopolizationȱcomȬ plaintȱ wouldȱ beȱ thatȱ aȱ dominantȱ firmȱ wouldȱ engageȱ inȱ practicesȱ thatȱ harmedȱ competitionȱbyȱreducingȱtheȱprivacyȱprotectionsȱaffordedȱconsumers.ȱTheȱprivacyȱ complaintsȱ areȱ notȱ argumentsȱ thatȱ Googleȱ wouldȱ engageȱ inȱ conductȱ thatȱ wouldȱ reduceȱcompetition,ȱbutȱratherȱ“status”ȱargumentsȱthatȱaȱsingleȱfirmȱinȱcontrolȱofȱ dataȱisȱpresumptivelyȱbadȱfromȱanȱantitrustȱperspective.ȱThereȱisȱnothingȱinȱmodȬ ernȱmonopolizationȱlawȱtoȱsupportȱsuchȱaȱclaim.ȱ
74ȱ
HarvardȱJournalȱofȱLawȱ&ȱPublicȱPolicyȱ
[Vol.ȱ34ȱ
thatȱtheȱqualityȱscoringȱmetricȱ(orȱsomeȱotherȱauctionȱrule,ȱlikeȱ reducingȱtheȱnumberȱofȱslotsȱavailable)ȱresultsȱinȱhigherȱpricesȱ becauseȱ thoseȱ pricesȱ moreȱ accuratelyȱ reflectȱ demand,246ȱ imȬ provingȱone’sȱabilityȱtoȱextractȱmonopolyȱrentsȱsimplyȱdoesȱnotȱ violateȱ Sectionȱ 2ȱ ofȱ theȱ Shermanȱ Act.ȱ Thisȱ fundamentalȱ tradeȬ offȱ reflectsȱ preciselyȱ andȱ deeplyȱ theȱ concernȱ withȱ errorȬcostȱ avoidanceȱthatȱweȱhaveȱbeenȱdiscussing.ȱ V.
CONCLUSIONȱ
Althoughȱ ourȱ analysisȱ hasȱ focusedȱ onȱ theȱ typesȱ ofȱ argumentsȱ weȱbelieveȱareȱmostȱlikelyȱtoȱbeȱraisedȱinȱSectionȱ2ȱclaimsȱagainstȱ Google,ȱitȱisȱpossibleȱthatȱthereȱareȱothersȱweȱdoȱnotȱconsiderȱhere.ȱ Forȱ theȱ reasonsȱ discussedȱ above,ȱ however,ȱ ourȱ tentativeȱ conclusionȱ isȱ thatȱ plaintiffsȱ cannotȱ orȱ shouldȱ notȱ prevailȱ againstȱ Googleȱ inȱ aȱ monopolizationȱ claimȱ basedȱ onȱ theȱ twoȱ typesȱ ofȱ conductȱ consideredȱ here:ȱ exclusiveȱ syndicationȱ agreementsȱ andȱ useȱ ofȱ theȱ qualityȱ scoreȱ metricȱ toȱ extractȱ greaterȱ rents.ȱ Atȱ aȱ minimum,ȱasȱaȱsafeguardȱagainstȱtheȱtypesȱofȱantitrustȱerrorȱthisȱ Articleȱdiscussesȱanyȱsuchȱenforcementȱactionȱshouldȱnotȱproceedȱ withoutȱrigorousȱandȱconcreteȱevidenceȱofȱharmȱtoȱconsumers.ȱ Indeed,ȱ inȱ lightȱ ofȱ theȱ antitrustȱ claimsȱ arisingȱ outȱ ofȱ innovativeȱ contractualȱ andȱ pricingȱ conduct,ȱ andȱ theȱ apparentȱ lackȱofȱanyȱconcreteȱevidenceȱofȱanticompetitiveȱeffectsȱorȱharmȱ toȱcompetition,ȱanȱenforcementȱactionȱagainstȱGoogleȱonȱtheseȱ groundsȱ createsȱ substantialȱ riskȱ forȱ aȱ falseȱ positiveȱ whichȱ wouldȱ chillȱ theȱ innovationȱ andȱ competitionȱ currentlyȱ providingȱimmenseȱbenefitsȱtoȱconsumers.ȱ
ȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱȱ 246.ȱAndȱasȱweȱdiscussȱinȱPartȱIII,ȱsupra,ȱaȱqualityȱscoreȱadjustmentȱdoesȱpermitȱ theȱ searchȱ engineȱ toȱ captureȱ moreȱ revenueȱ byȱ increasingȱ theȱ likelihoodȱ ofȱ reveȬ nueȬgeneratingȱclickthroughsȱbyȱincreasingȱsearchȱresultȱrelevance.ȱȱ