voting models 2009-11-16 - hathitrustideological proximity) between voters.” (felsenthal and...

25
1 Weighted Voting Models for the HathiTrust Constitutional Convention by Julia Lovett I. Introduction and Scope In March 2011, the HathiTrust Digital Library partner institutions will meet for a Constitutional Convention (CC). At this meeting, the partners will either develop a new governance model for HathiTrust or articulate a set of questions to frame a post‐Convention discussion of a new governance model. Questions might include: What should the governing boards be, and how should they be constituted? Which institutions should have representative power, and how many representatives should there be per institution? Should HathiTrust have a new host institution and, if so, what criteria should be used to select a new host institution? These decisions will have a large impact on the future of HathiTrust, and should be arrived at through a method that accurately reflects varying levels of institutional investment in the project. The best way to ensure fair decision‐making in a situation where voters are inherently unequal is to establish a weighted voting system (see Barrett and Newcombe, 1968). To this purpose, this paper explores the following questions: How is voting power calculated in a weighted voting system? What factors and principles should inform allocation of voting power? How do these models apply to HathiTrust? The first question is the most straightforward one to answer, drawing from scholarly literature about a priori voting power, a subset of general voting theory and social choice theory. Voting power theory can be applied to “any collective body that makes yes‐or‐no decisions by vote.” (Felsenthal and Machover, 1998) The classification a priori refers to voting power that is “determined without taking into consideration voters’ prior bias regarding the bill voted upon, or the degree of affinity (for example, ideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting power depend on the assumption that every possible combination of votes in a committee is equally likely to occur. In reality, of course, not all combinations are equally likely. Certain scholars strongly object to the fact that a priori methods exclude external factors such as voters’ past biases and affiliations (see Gelman et al., 2004; Margolis, 2004; Albert, 2003). This objection does not seem to present a problem for HathiTrust. Such biases and affiliations, if they exist, would be impossible to quantify without the existence of any voting prior to the CC. Voting power theory’s restriction to binary (two option) decisions is necessary for a number of reasons. First and foremost, a method of calculating weighted voting power for a system involving many options or complicated voting procedures has not been widely studied. Many voting theorists have detailed procedures with more than two choices and perhaps more than one result (Black, 1998; Dummett, 1984; Tideman, 2006; Brams, 1983), but these studies usually assume that constituents have equal votes. In addition, such studies often focus on elections and large‐scale voting processes, as opposed to

Upload: others

Post on 20-Feb-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

1

WeightedVotingModelsfortheHathiTrustConstitutionalConvention

byJuliaLovett

I. IntroductionandScopeInMarch2011,theHathiTrustDigitalLibrarypartnerinstitutionswillmeetforaConstitutionalConvention(CC).Atthismeeting,thepartnerswilleitherdevelopanewgovernancemodelforHathiTrustorarticulateasetofquestionstoframeapost‐Conventiondiscussionofanewgovernancemodel.Questionsmightinclude:Whatshouldthegoverningboardsbe,andhowshouldtheybeconstituted?Whichinstitutionsshouldhaverepresentativepower,andhowmanyrepresentativesshouldtherebeperinstitution?ShouldHathiTrusthaveanewhostinstitutionand,ifso,whatcriteriashouldbeusedtoselectanewhostinstitution?ThesedecisionswillhavealargeimpactonthefutureofHathiTrust,andshouldbearrivedatthroughamethodthataccuratelyreflectsvaryinglevelsofinstitutionalinvestmentintheproject.Thebestwaytoensurefairdecision‐makinginasituationwherevotersareinherentlyunequalistoestablishaweightedvotingsystem(seeBarrettandNewcombe,1968).Tothispurpose,thispaperexploresthefollowingquestions:

• Howisvotingpowercalculatedinaweightedvotingsystem?• Whatfactorsandprinciplesshouldinformallocationofvotingpower?• HowdothesemodelsapplytoHathiTrust?

Thefirstquestionisthemoststraightforwardonetoanswer,drawingfromscholarlyliteratureaboutapriorivotingpower,asubsetofgeneralvotingtheoryandsocialchoicetheory.Votingpowertheorycanbeappliedto“anycollectivebodythatmakesyes‐or‐nodecisionsbyvote.”(FelsenthalandMachover,1998)Theclassificationapriorireferstovotingpowerthatis“determinedwithouttakingintoconsiderationvoters’priorbiasregardingthebillvotedupon,orthedegreeofaffinity(forexample,ideologicalproximity)betweenvoters.”(FelsenthalandMachover1998)Inotherwords,thesystemsthattheoristshaveusedtodeterminevotingpowerdependontheassumptionthateverypossiblecombinationofvotesinacommitteeisequallylikelytooccur.Inreality,ofcourse,notallcombinationsareequallylikely.Certainscholarsstronglyobjecttothefactthatapriorimethodsexcludeexternalfactorssuchasvoters’pastbiasesandaffiliations(seeGelmanetal.,2004;Margolis,2004;Albert,2003).ThisobjectiondoesnotseemtopresentaproblemforHathiTrust.Suchbiasesandaffiliations,iftheyexist,wouldbeimpossibletoquantifywithouttheexistenceofanyvotingpriortotheCC.

Votingpowertheory’srestrictiontobinary(twooption)decisionsisnecessaryforanumberofreasons.Firstandforemost,amethodofcalculatingweightedvotingpowerforasysteminvolvingmanyoptionsorcomplicatedvotingprocedureshasnotbeenwidelystudied.Manyvotingtheoristshavedetailedprocedureswithmorethantwochoicesandperhapsmorethanoneresult(Black,1998;Dummett,1984;Tideman,2006;Brams,1983),butthesestudiesusuallyassumethatconstituentshaveequalvotes.Inaddition,suchstudiesoftenfocusonelectionsandlarge‐scalevotingprocesses,asopposedto

Page 2: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

2

votingincommittees.FelsenthalandMachover(1998)explicitlynotethesetrendsintheliterature,describingthemeasurementofvotingpoweras“orthogonaltotheconcernsofthegeneraltheoryofvoting.”(5)ThemostimportantaspectofavotingmodelfortheHathiTrustCCisthateachpartnerinstitutionhasalevelofinfluencecommensuratewithitsdegreeofinvestmentinHathiTrust.Givenexistingmethodsofcalculatingvotingpower,thesurestresultscanbedrawnfrombinaryvoting.Inaddition,votingtheoryoftenassumesthatasimplemajority(morethanhalf)willberequiredforadecisiontopass.Itispossible,ofcourse,forthequota—thenumberofvotesrequiredtopassadecision—toberaisedhigherthanasimplemajority.Severalstudieshavebeenmadeoftheeffectofvariousquotasonrealvotingbodies(Leech,2003;DreyerandSchotter,1980).Forthesakeofsimplicity,thispaperwilldiscussvotingtheorybasedonasimplemajority.

Themoredifficultquestionstoanswerarethesecondandthirdintheabovelist:Whatfactorsandprinciplesshouldinformallocationofvotingpower?HowcanthesemodelsbeappliedtoHathiTrust?Whereasmethodsofcalculatingvotingpowerhavebeenwidelycitedandacceptedbymanytheorists,thereisnostraightforwardmethoddetailinghowtofairlyallocatesuchpower.Theweightingfactorsseemtodependontheorganization,andcouldincludeprinciplesoffairness,monetarycontributions,andadesiretobalancelargerandsmallerpowers.CasestudiesofvotinginrealorganizationssuchastheUnitedNations,InternationalMonetaryFund,andEuropeanUnionCouncilofMinisters,couldprovidesuggestionsforHathiTrust.

II. VotingPowerTheory

Therearetwowidelyacceptedandcitedmethodsforcalculatingapriorivotingpower.OnewasoutlinedbyPenrose(1946)andlaterarrivedatindependentlybyBanzhaf(1965).TheothermethodwasfirstproposedbyShapleyandShubik(1954).Furthermore,Leech(2003)hasdemonstratedhowtousethesemethodstoworkbackwardsandcalculatevotingweightbasedonpredeterminedvotingpower,accordingtotheBanzhafmethod.Therearealsosomeminormethodsthathavenotbeenwidelyadoptedandwillnotbediscussedinthispaper(seeFelsenthalandMachover,2004).

Thehistoryofvotingpowertheoryhasbeencharacterizedbyfitsandstarts,asmanyscholarshaveundertakentheproblemofmeasuringapriorivotingpowerwhileapparentlyunawareofpriorworkinthesubject.BanzhafindependentlyarrivedatthesamesystemasPenrose,inignoranceofthelatter’s1946paper.Coleman(1971)approachedtheproblemunawareofbothShapley‐ShubikandBanzhaf.(FelsenthalandMachover,1998)Onlyrelativelyrecentlyhasthefieldcometogetherinacoherentway.

InboththeUnitedStatesandEurope,interestinvotingpowertheoryhassurgedwithchangesindecisionrulesoflargevotingbodies.Forexample,intheEuropeanUnion,manytheoristsarguedfororagainsttheuseofaprioripowermeasuresasaguideforthere‐weightingofvotesintheCouncilofMinistersin2001(PajalaandWidgren,2004).IntheUnitedStates,thetopicofvotingpowerenjoyedthegreatestpopularityinthe1960’swhenmanyscholarsadvocatedapplyingweightedvotingtostatelegislatures.(FelsenthalandMachover,2004;Leech,2003)ItwasduringthiscontroversythatJohnBanzhaffirstarguedagainstequatingvotingweightwithvotingpower.

Page 3: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

3

A. PenroseandBanzhaf

Inhisseminalarticle“WeightedVotingDoesn’tWork:AMathematicalAnalysis,”JohnBanzhaf(1965)arguesthat“votingpowerisnotproportionaltothenumberofvotesalegislatormaycast.”(318)Inotherwords,agreatervotingweightdoesnotautomaticallytranslateintogreaterpower.Banzhafcraftedthisargumentinresponsetowidespreadsuggestionsatthetimethatweightedvoting,inwhichlegislativerepresentativeswouldholdvotingweightsproportionaltothepopulationoftheirrespectivedistricts,couldbeimplementedasanalternativetoreapportioningthelegislativeseats.Theresultingdistributionofpower,proponentsargued,wouldbethesameineithercase.

Banzhafdemonstratesthefallacyofequatingweightwithpowerthroughaseriesofexamplesandmathematicalreasoning.Heasserts,“itwouldbemoreeffectivetothinkofvotingpowerastheabilityofalegislator,byhisvote,toaffectthepassageordefeatofameasure.”(318)Alegislator’spowerliesinthechancethathisindividualvotewillswingthecollectivevote.Thosechancesareindeedhighlydependentontherelativevotingweightsinthecommittee,butnotinapredictableorintuitiveway.Banzhafpresentsthefollowingtable(p.339)asonerealexamplewhereseeminglyappropriatevotingweightsresultinradicallyinappropriatepowerdistribution:

Inordertotallythenumbersinthelastcolumn,Banzhafhastakenallthepossiblecombinationofvotesandlookedatthecombinationswherethelegislatorcouldswingthevote.Heassumesthatasimplemajority(morethanhalf)willcausethebilltopass.ThemostsurprisingresultisthatNorthHempstead,withapopulationof213,225and21weightedvotes,canneverbethepivotalvoteandthereforeeffectivelyhasnovotingpower.

Banzhaf’sargumentechoestheworkofL.S.Penrose(1946),whosearticle“TheElementaryStatisticsofMajorityVoting”firstundertookaseriousstudyofapriorivotingpower.Penrosewrites,“thepoweroftheindividualvotecanbemeasuredbytheamountbywhichhischanceofbeingonthewinningsideexceedsonehalf.”(53)ParaphrasedbyFelsenthalandMachover(2004),Penrose’sbasicargument(thesameasBanzhaf’s)isthat“themorepowerfulavoteris,themoreoftenwilltheoutcomegotheways/hevotes.”(3)Banzhaf,whowasapparentlyunawareofPenrose’swork,explainsthissameconceptwithasimplemathematicalargument:

Page 4: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

4

Moreexplicitly,inacasewherethereareNlegislators,eachactingindependentlyandeachcapableofinfluencingtheoutcomeonlybymeansofhisvotes,theratioofthepoweroflegislatorXtothepoweroflegislatorYisthesameastheratioofthenumberofpossiblevotingcombinationsoftheentirelegislatureinwhichXcanaltertheoutcomebychanginghisvotetothenumberofcombinationsinwhichYcanaltertheoutcomebychanginghisvote.(331)

WhilePenroseframedhisargumentaroundtheabsoluteprobabilityofavoter’ssuccess,Banzhafwasmoreinterestedinrepresentingrelativepower.Thefollowingtable(p.342)showshismethodoflistingallpossiblevotingcombinationsandtallying“combinationsinwhicheachlegislatorcastsadecisivevote:”

Page 5: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

5

Again,therearesomesurprisingresults:VoterL,withaweightofone,hasjustasmuchrealvotingpowerasJandK,whoeachhaveaweightofthree.InordertodeterminetherelativevotingpowerofH,I,J,K,andL,onewoulddivideanindividual’svotingpowerbythetotalamountofvotingpowerintheassembly.Essentially,thiscalculationresultsintheindividualvoter’sshareofthetotalavailablevoting

Page 6: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

6

power.Convertingthenumberstopercentages,sothattheentireweightaddsupto1,resultsinwhatistermedtheBanzhafIndex(FelsenthalandMachover,2004,p.5).FelsenthalandMachover(2004,p.6)demonstrateBanzhafIndexvaluesfortheEuropeanUnionCouncilofMinistersinthefollowingtable:

IftheHathiTrustpartnersaccepttheBanzhafIndexasameasureofrelativevotingpower,suchanindexwouldbesimpletocomputeforanygivenvotingweights.Themoredifficultpartwouldbeworkingbackwardstofindvotingweight,givenapredeterminedpowerdistribution.Leech(2003)pointsout,“Therehavebeenmanystudiesofthedistributionofaprioripowerinactualvotingbodieswherethedecisionruleandallocationofvotestovotingmembersisgivenbutrelativelyfewwheretheapproachhasbeenusedasatoolfordesigningweightedvotingsystems.”Inthissamepaper,LeechdemonstrateshowtousetheBanzhafindextodesignavotingsystem.Thesecalculationsarepossiblewithsomemathematicalwork,discussedinmoredetailinthesectionbelow,“DeterminingWeightforaGivenBanzhafPowerAllocation.”

B. ShapleyandShubikTheShapley‐Shubik(1954)methodistheotherwidelyacceptedsystemofdeterminingvotingpower.Atfirstglance,thismethodseemstoreplicateBanzhaf’sapproach.ShapleyandShubikstate,“Ourdefinitionofthepowerofanindividualmemberdependsonthechancehehasofbeingcriticaltothe

Page 7: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

7

successofawinningcoalition.”(787)This“chanceofbeingcritical”isimportantinBanzhaf’smethodaswell.ButtheirconceptofvotingpowerbeginstosoundabitdifferentfromBanzhaf’s:“Itispossibletobuyvotesinmostcorporationsbypurchasingcommonstock.Iftheirpoliciesareentirelycontrolledbysimplemajorityvotes,thenthereisnomorepowertobegainedafteronesharemorethan50%hasbeenacquired.”(788)WhereasPenroseandBanzhafbothdescribedvoterpowerasthepowertoinfluenceadecision,Shapley‐Shubikdescribevoterpowerasashareinanexpectedpayoff.Thereisafundamentalconceptualdifferencebetweenthetwomethods;whereastheBanzhafmethodisbasedonprobability,theShapley‐Shubikmethodisbasedongametheory.

FelsenthalandMachover(1998,2004)stronglyarguethispointabouttheconceptualdifferencesbetweenthetwomethods.ShapleyandShubik’sviewofpower,theyargue,“isthatpassageordefeatofthebillismerelytheostensibleandproximateoutcomeofadivision.Therealandultimateoutcomeisthedistributionofafixedpurse(theprizeofpower)amongthevictorsinthecasewhereabillispassed.”(10)VotersaccordingtoPenroseandBanzhafarepolicy‐seeking;votersaccordingtoShapleyandShubikareoffice‐seeking.(FelsenthalandMachover2004,p.11)Theideaofvotermotivationissomethingthatcomesupofteninthegeneraltheoryofvoting(seeColeman,1971;Dummett,1984).Thispaperwillnotexplicitlydetailthosearguments,butithelpstobeawarethatcertainideasaboutvotermotivationunderlieboththePenrose‐BanzhafandShapley‐Shubikmethods.IndecidingonamethodofcalculatingvotingpowerintheHathiTrustCC,thepartnersshouldconsiderwhichnotionofvotermotivationfitswithHathiTrust.Mostlikely,thepartnerswillallbeinterestedinwhatisbestforHathiTrust,andthereforecouldberegardedaspolicy‐seekingratherthanoffice‐seeking.

Beyondhavingconceptualdifferences,theBanzhafandShapley‐Shubikmethodssometimesproducedifferentnumericalresultswhenappliedtothesamesetofvoterweights.ThedifferencescanbeseeninthistablefromLaneandBerg(1999),detailingvotingpowersinGermany(p.314)accordingtothePenrosemeasure(BI),BanzhafIndex(Banzhafnorm.),andShapley‐ShubikIndex(SSI):

Page 8: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

8

OnepotentialdrawbackofusingtheShapley‐ShubikIndexisthatitismuchmoredifficulttocalculatethantheBanzhafIndex.InShapleyandShubik’soriginalarticle(1954),aftergivinganexampleofanindividualvoter’spowerinacertainsituation,theystate:“Thecalculationofthisvalueandthefollowing[values]isquitecomplicated,andweshallnotgiveithere.”(789)PerhapsthesimplestexplanationisgivenbyDixon(1983,p.298),whowrites:

Thismeasurereflectsthepriorprobabilitythatanindividualwillcastthedecidingvoteonanyissuebybeingthelastmembertojoinaminimalwinningcoalition...Todeterminetheexpectationthatanindividualwillpivot,onemustconsiderallofthepossiblesequences(n!inannmembervotingbody)inwhichaminimalwinningcoalitionmightform.Foranymemberi,theShapley‐Shubikpowerindex,φ,maybedefinedas

φ=(totalpivotsfori)/n!

TheShapley‐ShubikIndextakesintoaccountthe“possiblesequences,”theorderinwhichvotesoccur.Again,thispointstoaconceptualdifferenceaswellasamethodologicaldifferencebetweentheShapley‐ShubikmethodandtheBanzhafmethod:WhereasBanzhaffocusesoncombinationsofvotes,

Page 9: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

9

theShapley‐Shubikmethodfocusesonpossiblesequencesofvotes.Thepivotalmember,accordingtoShapley‐Shubik,isthelastmemberwhosevoteformsa“minimalwinningcoalition,”renderingthesubsequentvotesmeaningless.Again,forHathiTrust,thepartnersshoulddecidewhichmethodofpowerdeterminationfitsHathiTrust’svotingscheme.Ifvoteswillbecastsimultaneouslyratherthansequentially,thentheBanzhafIndexmakesmoresensethantheShapley‐ShubikIndex.

C. DeterminingWeightforaGivenBanzhafPowerAllocation

Calculatingtheweightsforagivenpowerallocationrequiresacomputerandaprogramtorunthenecessaryalgorithm.InLeech’s2003paper,“PowerIndicesasanAidtoInstitutionalDesign:TheGeneralisedApportionmentProblem,”heexplainsastep‐by‐stepapproachtocalculatingvotingweightsbasedonagivenBanzhafIndexofpowervalues.Essentially,hismethodreliesontrialanderror,startingwithanarbitraryguessofweightsandthenrepeatingtheguessesatcertainintervalsuntilhittingontheclosestpossiblematch.Keepinmindthat,asLeechexplains,thereisnooneuniqueweightdistributionforeveryuniquepowerindex.Usinghisalgorithm,itispossibletofindanextremelycloseapproximation,withafewimportantlimitations:Ingeneral,themorevotersthereare,themorepossiblepowerdistributionsthereare,andtheeasieritistofindweightstomatchagivenpowerallocation.Forexample,Leechdemonstratesthatwith3voters,foranycombinationofweights,thereareonlyfourpossiblepowerallocations.Upto5voters,therearestillrelativelyfewresultingpowerallocations.

LuckilyforHathiTrust,bythetimeoftheCCtherewilllikelybeatleastsevenpartners,andtheweightscanbecalculatedtocloselyapproximatemostpowerallocations.Inthespreadsheetattachedtothisdocument,eachexampleHathiTrustpowerallocationcorrespondstoacertainweightdistribution,andtheseweightsarebasedonLeech’smethod.CreditforthisworkisduetoDanielKneezel,aUniversityofMichiganPhDstudentinmathematics,whodesignedaprogramtorunLeech’salgorithm.(ForamoredetailedexplanationofthisspreadsheetandspecificrecommendationsforHathiTrust,seethelastsectionofthisdocument.)

III. PrinciplestoGuideVotingPowerAllocation

Apriorivotingpowertheoryservesanextremelyusefulpurpose:itenablesthecalculationofvotingpowerbasedonunequalvotingweights.Itdoesnot,however,makethejobofdecidinghowtoallocatevotingpoweranyeasier.Manyscholars,whethertheyincorporateformalvotingpowertheoryornot,havetackledtheproblemofhowtoallocatevotingpowerinorganizationswithdiverseconstituents.Severalgeneralprinciplescanbegleanedfromthesestudies,asbasiccriteriatoguidethedistributionofvotingpower.

BarrettandNewcombe(1968)completedanextensivestudyofvariousweightedvotingformulasandhowtheymightapplytotheUnitedNations.Theauthorsneatlylayoutfourprinciplesthatavotingmodelshouldsatisfy:“Decision‐makingbyvotingisasocialinventiondesignedtosatisfyseveraltypesofdemands:Thoseofjustice(orequity),thoseofwisdomoreffectiveness,thoseofreflectingandformalizingactualpowerrelationships,andthoseofacceptabilitytotheparticipants.”(2)Corresponding

Page 10: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

10

tothese“typesofdemands”areseveralreasonsthatweightedvoting,ratherthanequalvoting,isnecessaryincertainsituations.Thefollowingtableofprinciples(“typesofdemands”),withcorrespondingreasonsforweightedvoting,isadaptedfromBarrettandNewcombe’sexplanationonp.1‐5.

TypesofDemands ReasonsforWeightedVotingEquity • Somemayhaveagreaterfinancialstake

• Somemaybemoreaffectedbythedecisions

• Somemayhavegreaterseniorityrights• Somemayrepresentlargerorganizations

Effectiveness • Somemaybeinabetterpositiontocarryoutthedecisions

• Somemaybebetterinformedabouttheissues

Reflectingandformalizingpowerrelationships • Somemayhavemorepersonalpowerthanothers,andwewishtoformalizethispowerratherthanhavingitexercisedinformallybyinfluencingthevotesofothers

Acceptabilitytoparticipants • Noexplicitreasons;weightedvotingis“acceptable”whenitreachesacompromisebetweenequityandpower

Lookingatthesedemandsandreasons,itiseasytoseethatmanyofthemapplytotheHathiTrustpartnership.ThespecificpermutationsofthefactorsastheyrelatetoHathiTrustwillbediscussedintheRecommendationssectionofthisdocument.Itisimportanttofirstunderstandtheprinciplesthemselves,thereasonsforthoseprinciples,andhowrealorganizationsattempttosatisfythoseprinciplesthroughweightedvotingschemes.

A. Equity

Theprincipleofequityisintuitive.Inasituationwhereeveryvoterhasanequalsay,themeansofsatisfyingtheprincipleofequityissimple:eachpersonshouldhaveonevote.Manytheoristsarguethatinrealinternationalorganizations,whosemembersrepresentthecitizensoftheirrespectivecountries,theidealvotingpowerallocationwillequalizethepowerofthosecitizensthroughweightedvotingintheorganization.Ifeachmemberoftherepresentativeboardgetsonevote,thenthecitizensoflargercountriesactuallyhavelesspowerthanthecitizensofsmallercountries.Likewise,BanzhafandShapley‐Shubikhaveproventhatgivinglargercountriesagreatervotingweightdoesnotresultintheappropriateallocationseither.Itwasthisprincipleoffairnessofrepresentation,andtheconcernthatentiregroupsofcitizensmightnothaveanyactualvotingpower,thatdroveBanzhaftocreatetheBanzhafIndex.Acountry’spopulation,therefore,isoftensuggestedasamainfactorindeterminingvotingpower.

HathiTrustisnotarepresentativebody,andcannotrelyonthedemocraticprincipleofone‐person‐one‐vote.Thereareotherfactors,though,thatarebeing“represented”bythevotingmembersof

Page 11: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

11

HathiTrust.Thesecouldincludethenumberofvolumescontributedtotherepository.Thegreaternumberofvolumesaninstitutionhascontributed,thegreatertheinstitution’sstakeinHathiTrustandthegreatertheamountofmaterialthatisbeingrepresentedfromthatinstitution’scollection.

B. Effectiveness

Theprincipleofvotereffectivenessisperhapsthemostdifficulttoquantify,andyetitseemsrelevanttoanyorganizationinwhichmembershavevaryingdegreesofexperience.InHathiTrust,certainmemberinstitutionshavemoreintimatefamiliaritywiththegovernanceofHathiTrustsimplybyhavingworkedontheprojectatagranularlevelonadailybasis.Ifanynew,non‐foundingpartnersjoinHathiTrustbeforetheCC,certainlythosenewpartnershavelessexpertisethanthecurrentpartners,aswellaslesscapabilitytoeffectivelyactondecisionmaking.BarrettandNewcombeclassifytheabilitytoactondecisionsaspartoftheeffectivenessprinciple.Voterswhoarewiseandexperiencedareprobablymoreinvolvedintheissuesathand,andthereforebetterpoisedtocarryoutdecisions.ThiscorrelationiscertainlytrueinHathiTrust,wherethememberswhohavemoreexperiencealsohavemoreexistingtiestotheprojectandgreaterabilitytocarryoutdecisions.

C. Power

Inanyorganization,somemembershavemorerealpowerthanothers.BarrettandNewcombeexplainthatthereareseveraltypesofinformalpoweramembermightpossess.Forexample,certainmembersmaybemorecharismaticthanothers.Certainmembersmayhavemorepoliticalpower.ThefactisthatcertaininstitutionsinHThavemorerealpowerthanothers.Alongwithacommitmenttohostthephysicalinfrastructure,MichiganandIndianahavemadedeepercommitments(e.g.theMichiganlegalorientation,alargerfinancialcommitmentbyeachuniversity,andIU’stechnologicalinvestments)thatpartnerswouldbechallengedtofindatanotherinstitution.

Theneedtopreventstrongermembersfromdominatingdecision‐making,whilestillensuringthatsmallerpowershaveavoice,isacentralneedthatweightedvotingattemptstosolve.Intheir2006article“ReformingtheIMF’sWeightedVotingSystem,”RapkinandStrand(2006)arguethatreformstotheIMF’svotingregimemustanswera“fundamentalproblemthatisboththeoreticalandpractical:howbesttoreconciletheprincipleofsovereignequalitywiththefactofwidepowerasymmetriesamongmembers.”(p.2)IntheIMF,variousmembersdohavemorerealpowerthanothers,andtherearecertainvotingmeasuresthatattempttoformalizethatpower.TheUnitedStates,forcertaintypesofdecisions,“retainstheonlysingle‐countryvetoovermajorIMFdecisions,includinganydecisionthatwouldreduceitsvotingpowerandincreaseordecreasethatofothercountries.”(p.12)Vetopowerisoneoptionincaseswhereonemembersimplyhasmorerealpowerthanalloftheothermembersputtogether.Itcanbeharmful,however;RapkinandStrandalsopointoutthattheU.S.’svetopowerfosters“perceptionsofsystemicunfairness”(p.2)andinfluencesdecisionsevenwhenitisnotexplicitlyexercised.

Whetheranorganizationgrantscertainmembersvetopowerornot,thefactremainsthatmorepowerfulmembersofanorganizationwillnotacceptavotingmodelinwhichtheyarenotgrantedenoughpower.Neitherwillthelesspowerfulmembershaveconfidenceinamodelthatgrantsthem

Page 12: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

12

verylittleornopower.Theprincipleofpower,therefore,ishighlyinterrelatedwiththeprincipleofacceptability.

D. Acceptability

Acceptability,accordingtoBarrettandNewcombe(amongothers),representsacompromisebetweentheprinciplesofpowerandequity.Anacceptableschemewillrepresenttheinterestsofboththestrongerandweakerparties.Acceptabilityisarguablythemostimportantprinciplethataweightedvotingmodelshouldsatisfy,simplybecauseavotingmodelthatisnotacceptabletovoterswillnotwork.Newcombe,Wert,andNewcombe(1971)alsoechothisfocusonacceptabilityintheirlaterstudyofpossibleUNformulas:

Onlyaworldbodythathastheconfidenceofasmanymember‐nationsaspossible,EastandWest,richandpoor,largeandsmall,willbeabletoobtainthepowersitneeds.The‘balancing’neededtogaintheconfidenceofmemberstatesisadifficultmatter.Preciselythisbalancingisthepurposeofanyweighted‐votingscheme....Themaincriterioninevaluatingaweighted‐votingformulaisacceptability,noteitherabstractjusticeortheoreticalreasoning.”(452)

Theirastuteobservationisagoodonetokeepinmind,andwillmostlikelybethedrivingprinciplebehindselectingavotingprocedureforanyorganization.Inthelongterm,novotingprocedurewillworkifitdoesnotinspiretheconfidenceoftheconstituents.

ManyofthevoteallocationformulasproposedbyBarrettandNewcombe,andlaterdiscussedbyDixon(1983),inrelationtotheUnitedNations,aimatsomesortofcompromisebetweenthefactorsofequityandpower.Thefactorofacountry’spopulationcorrelatestothedemandsofequity,whilethefactorofGDPcorrespondsto“aroughindicationofastate’spower.”(Dixon,1983)Thesefunctionsincludetakingdirectnumbers,logarithms,squareroots,andothervariationsoneachcountry’srespectiveGDPandpopulation.Interestingly,Dixonalsodebatestheacceptabilityofapriorivotingtheoryitself,asthepointofhisstudyistoapplytheBanzhafIndextoeachformula.Hepointsoutthatifaweightedvotingsystemisdesignedspecificallytomeettheneedsofagivenpowerallocation,thevotingweightsnecessarytogenerateagivenpowerdistributionwillseemsoarbitrarytovotersastorenderthemunacceptable.

Onepossiblecompromisebetweenweightedvotingandequalvoting,whichisRapkinandStrand’s(2006)mostimportantrecommendationtotheIMF,isthesystemofvotingbydoublemajority.The“CountandAccount”(doublemajority)votingscheme,asdescribedbelowbyO’NeillandPeleg(2000),representsamorestableandconsistentcompromisebetweenweightedvotingandequalrepresentation:

Votingbycountandaccounttakesintoconsiderationbothsizeandequalityanddoessoinasimpleway.Votesarecountedtwice,firstwitheachpartyweightedequally,andthenwitheachweightedbyitsfinancialcontributionorsomeotherobjectivemeasure.Aproposalpassesifitgetsamajorityinbothways.Theruleformalizestheideathatanorganizationshouldactonlywhenithasthesupportofboththegeneralmembershipandtheimportantmembers.(3)

Page 13: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

13

Onethingtobeawareofisthatthecountandaccountschemeis“necessarilymoreconservativethanusingastraightmajorityoftheaccountweightsalone,sincetwocriteriahavetobesatisfiedratherthanoneandfewerresolutionsgetpassed.Thiscanbeseeneitherasadisadvantageofthemethod.”(O’NeillandPeleg8)Itisalsopossibletoapplyvotingpowerindexestothedoublemajoritymethod;Turnovec(1997)hasdonesoinhisusefulstudyofvotingintheEuropeanUnion.Thedoublemajoritymethodisonepossiblewaytoincreaseacceptability.ItcouldbeparticularlyusefulasatechniqueforHathiTrust.

IV. RecommendationsforaHathiTrustVotingModel

Drawingfromtheprevioussections’discussionsofapriorivotingpowertheoryaswellascasestudiesofvotingpowerallocation,thissectionwillproposerecommendedvotingmodelsforHathiTrust.Recommendationsfallintotwodifferentcategories:basicrecommendations,andspecificrecommendations.Whereasbasicrecommendationshavetodowiththeassumptionsandscaffoldingofthevotingmodel,specificrecommendationswillentailanumberofdifferentformulasthatcouldworkforHathiTrust.

TodesignavotingmodelthatworksforHathiTrust,wemustask:

• Whichofthefourprinciples—equity,effectiveness,power,andacceptability—arerelevanttoHathiTrustattheConstitutionalConvention?

• Howcantheprinciplesbesatisfiedthroughaweightedvotingmodel,andwhatfactorsareavailabletoinformthemodel?

Toanswerthefirstquestion,itishelpfultoturntoexistingHathiTrustdocumentation,particularlytheMissionandGoals,FAQ,andFunctionalObjectivesontheHathiTrustwebsite(http://www.hathitrust.org/).Acrossallofthedocumentation,thereisastrongfocusoncollaboration,co‐ownership,andopenness.TheMissionandGoalsexplainHathiTrust’scorevalues,statingthattheorganizationisa“collaborationofthethirteenuniversitiesoftheCommitteeonInstitutionalCooperationandtheUniversityofCaliforniasystemtoestablisharepositoryfortheseuniversitiestoarchiveandsharetheirdigitizedcollections.Partnershipisopentoallwhosharethisgrandvision.”Atthesametime,theHathiTrustdocumentationmakesreferencestothefactthatdespitethisunityofvision,partnerswillretaintheirindividuality.Forexample,HathiTrustwillbe“co‐ownedandmanagedbyanumberofacademicinstitutions”andwillstriveto“dramaticallyimproveaccesstothesematerialsinwaysthat,firstandforemost,meettheneedsoftheco‐owninginstitutions.”Thesestatementssuggestatoncecollectiveownership,collectivemanagement,andcollectiveneeds,butalsoindividualownership,individualmanagement,andindividualneeds.ItissafetosaythatHathiTrustisacollaborativeorganizationwithacohesivevisionthatisnonethelesscomprisedofindividualanddistinctconstituents.OneofthechallengesindesigningavotingmodelwillberecognizingthefundamentalvalueofeachHathiTrustinstitution,whileallowingfordifferencesinpowerandinfluence.

ItisalsoclearfromtheHathiTrustdocumentationthatthefoundingpartnersofHathiTrustdeserveagreatdealofcreditfortheirhardworkontheprojectthusfar.UndertheFAQ“Who’stakingthelead?”

Page 14: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

14

thefollowingstatementemphasizesthecommitmentandexpertiseofthecurrentpartners:“TheUniversityofMichigan,IndianaUniversity,theUniversityofVirginia,andtheUniversityofCaliforniasystem,allhighlyregardedfortheirexpertiseintheareasofinformationtechnology,digitallibraries,andprojectmanagement,areleadingthepartnershipeffortthroughtheirexpertiseandfinancialcommitment.AllmembersoftheCICarefoundingpartners.”Recognizingpowerdifferencesamongthepartnerssuggeststhattheprinciplesofbothequityandpowerneedtobesatisfiedbyaweightedvotingsystem.Currently,therearefivepartners:

• UniversityofMichigan• IndianaUniversity• CaliforniaDigitalLibrary(includesalltheUniversityofCaliforniainstitutions)• CommitteeonInstitutionalCooperation(excludingMichiganandIndiana)

o MichiganStateUniversityo NorthwesternUniversityo TheOhioStateUniversityo PennStateUniversityo PurdueUniversityo TheUniversityofChicagoo UniversityofIllinoiso UniversityofIllinoisatChicagoo TheUniversityofIowao UniversityofMinnesotao UniversityofWisconsin‐Madison

• UniversityofVirginia

Inaddition,ColumbiaUniversityandYaleUniversitymayalsobejoiningtheranksofHathiTrustrelativelysoon.Amongthepartners,thefollowingfactorscontributetotheirdifferencesincommitmenttoHathiTrust:

• Financialcontributions• Projectednumberofvolumesinrepository• Lengthofcommitment(sinceJanuary2008)• Repositoryadministratorstatus

o UniversityofMichigan,IndianaUniversity• Founderstatus

o UniversityofMichigan,IndianaUniversity,CommitteeonInstitutionalCooperation,UniversityofCalifornia

Toillustratetherelationshipofthesefactorstoprinciples,or“typesofdemands,”thefollowingtablemimicstheprevioustablewhileaddingineachfactorandwhereitwouldcorrelatetoaprinciple:

TypesofDemands HathiTrustReasonsforWeightedVoting

HathiTrustWeightingFactors

Equity • Somehaveagreaterfinancialstake

• Somearemoreaffectedbythedecisionsduetogreaterinvestmentofresourcesand

• Financialcontribution• Numberofvolumes

contributed(projected)• Lengthofcommitment,

startinginJanuary2008• Founderstatus

Page 15: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

15

Inthistable,severalofthefactorsoverlap;thatis,theyfulfillseveraldifferentprinciples.Theprincipleofeffectivenessistheonlyonethatdoesnothaveanyuniquefactors.Itissafetosaythatequity,power,andacceptabilityarethedemandsthatwilldeterminetheHathiTrustCCvotingmodel.Effectivenessiscertainlyimportant,butanyvotingmodelthatsatisfiestheotherthreeprincipleswillautomaticallysatisfytheprincipleofeffectivenessaswell.

A. GeneralRecommendations

TherecommendationsforHathiTrust’svotingmodelcanbedividedintogeneralrecommendations,pertainingtotheoverarchingstructureofthemodel,andspecificrecommendationspertainingtothemorecomplicatedsubsetofquestionsaboutpowerallocationandvotingweight.ThosemorecomplicatedquestionswillbediscussedinSectionB.ThefollowingsetofrecommendationsappliestothebasicarchitectureofaHathiTrustvotingpowerscheme.Itgoeswithoutsayingthattheyaremeanttosatisfytheprinciplesofequity,power,andacceptability,aswellasappropriatelyreflectthefactors.

• Implementadoublemajorityvotingsystemofbothweightedandequalvotes,requiringamajorityofbothtypesofvotesinorderforadecisiontopass

• Fortheweightedvotes,setasimplemajority(50%quota)• Fortheequalvotes,themajoritycanbesetatasimplemajorityorhigherdependingonthe

importanceofthedecision• Forespeciallyimportantdecisions,itwouldalsobewisetorequireamajorityorunanimityof

theHathiTrustfoundinginstitutions:UM,UC,CIC,andIU

Equity • Somehaveagreaterfinancialstake

• Somearemoreaffectedbythedecisionsduetogreaterinvestmentofresourcesandcontributionofvolumes

• Somehavemadealongercommitmentand/ororarefoundingmembers

• Financialcontribution• Numberofvolumes

contributed(projected)• Lengthofcommitment,

startinginJanuary2008• Founderstatus

EffectivenessofVoters • SomeareHathiTrustadministrators

• SomehavebeenmoreinvolvedthanothersinshapingHathiTrust,andaremoreinformedabouttheissues

• Administratorstatus• Lengthofcommitment,

startinginJanuary2008

Power • SomeareHathiTrustadministratorsand/orfoundingmembers,andthereforehavemorerealinfluenceoverdecisions

• Administratorstatus• Founderstatus

Acceptability • GiventhepowerasymmetriesinHathiTrust,aweightedvotingschememakessense

• Compromisebetweenfactorsofpowerandequity

Page 16: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

16

• HathiTrustadministrators,UMandIU,shouldeachhaveaninstitutionalveto• Sticktobinary(twooptions,oryes/no)voting,whichwillenabletheapplicationofapriori

votingpowertheorytotheweightedvotes• Determineavotingpowerallocationbasedonanappropriateformulaandfactors,andthen

calculatethevoterweightsaccordingtotheBanzhafIndexusingLeech’s(2003)method(seeSectionBandattachedspreadsheet)

o Recommendedpowerallocationformula:

VotingPower=(squarerootofvolumes)+(squarerootoffinancialcontribution)

Mostoftheserecommendationsareself‐explanatory,orhavebeenjustifiedelsewhereinthedocumentalready.ThedoublemajorityvotingmodeleffectivelyreflectsthetensionevidentintheHathiTrustmissionstatementbetweeninstitutionalequalityandpowerdifferences.Itfulfillstheprincipleofacceptabilitybecauseitachievesinexecutionwhatitpromisesintheory.Itisalsoanextremelyusefulmodel,asitissimpleandyetadaptabletodifferenttypesofdecision‐making.Thisadaptabilitymakesitlikelytobeanacceptablescheme.Asnotedabove,theequalvotesportionoftheschemecanbeeasilymanipulatedintermsofitsquotaandalsothesubsetsoftypesofvoters.Forimportantdecisions—movingtheHathiTrusthostinstitution,forexample—itmakessensetorequirenotonlyamajorityofvoters,butahighermajoritythanusualandperhapsamajorityoffoundingmemberinstitutions.

ImplementinganinstitutionalvetofortheHathiTrustadministratorsislikelytobethemostcontroversialoftheserecommendations,butitmakessensebothpracticallyandtheoretically.TheUniversityofMichiganandIndianaUniversityhavebothmadetremendouscommitmentsofadministrativeresourcesandlegalresponsibilitytoHathiTrust.

Asfortheweightedvotingportionofthescheme,thequotaislessflexible.Thequotashouldbesetat50%because,withahigherquota,itissimplymoredifficulttofindweightsthataccuratelyreflectagivenpowerallocation(seeLeech2003).Sincethegoalofweightedvotingistoachieveanaccurateallocationofpower,itmakessensetouseaquotathatachievesthisaccuracy.Theadherencetobinaryvotesisalsomeanttosatisfytheneedforaccuratevotingweights,asonlybinaryvotingcanbeanalyzedusingapriorimethods.

TherecommendationofusingtheBanzhafIndex,asopposedtotheShapley‐ShubikIndexofcalculatingvotingpower,isbasedonthetwomethods’verydifferentconceptionsofvotermotivationandprocedure.TheBanzhafIndexassumesthatvotersaredisinterestedparties(policy‐seeking)ratherthanseekersoftheirownadvancement.TheHathiTrustpartnerscertainlyhavetheproject’sbestinterestsatheart,andthereforetheBanzhafconceptionofmotivationismoreappropriate.TheShapley‐ShubikIndexisalsobasedonsequenceofvotes,whichseemsirrelevanttoHathiTrust.VotesattheHathiTrustCCshouldbetakensimultaneously.Asforthevotingweightsthemselves,thereareanumberofdifferentformulasthatcanbeusedtoallocatepower,butsomeworkbettertoaccuratelyreflectinstitutionalcontributionsandachieveabalanceofpowerandequity.

B. SpecificRecommendations:PowerAllocationandVotingWeights

Page 17: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

17

Thefollowingrecommendationsrefertothefirstpointofthebasicrecommendations:“Determineavotingpowerallocationbasedonanappropriateformulaandfactors,andthencalculatethevoterweightsaccordingtotheBanzhafIndexusingLeech’s(2003)method.”Theattachedspreadsheetillustratesseveraldifferentpowerallocationsaccordingtoformulas,aswellastheweightsdeterminedbyLeech’salgorithmdiscussedabove.Asfounderstatusandadministratorstatushavealreadybeenaccountedforthroughotherrecommendations,theseformulasarederivedfromthequantifiablefactorsoffinancialcontributionsandnumberofvolumescontributedtoHathiTrust.

Thereareseveralpossiblepowerallocationsthatcouldwork,andthissectionwillproposefourformulas.Asmentionedintheprevioussection,Dixon(1983),drawingfromNewcombe,Wert,andNewcombe(1971)aswellasBarrettandNewcombe(1968),proposedseveralformulasthatpurporttosatisfytheprinciplesofequityandpower(andthusachieveacceptability)bytakingeitherdirectproportionsorsimplemathematicalderivationsofbothpopulationandGDP.Recallthatpopulationsatisfiestheprincipleofequity,whileGDPequatestopower.

Theformulasareasfollows:

1. Population+GDP2. PopulationxsquarerootofGDPpercapita3. Logpopulation+logGDP

Thefirstformulaobviouslydirectlycorrelatestothetwofactorsinvolved,simplytotalingthetwo.Thethirdformulauses“theapplicationoflogarithmsasawaytodiscountextremevalues,”(p.300)whilethesecond“compensatesforpopulationdifferencesbydistributingvotesaccordingtotheproductofpopulationandthesquarerootofGDPpercapita.”(p.300)OutofDixon’sfifteenformulas,thesearethethreethatusemultiplefactorsandalsousemathematicalfunctionstospecificallymitigatetheinequalitiesoftherawnumbers.TheyseemappropriateandtranslatabletoHathiTrust’sneeds.(Notethattheseformulaswereoriginallyproposedasformulasforweight,butwhattheyaretrulymeanttosignifyispower.Itissafetousethemtoallocatepowerratherthanweight.)

WhileHathiTrustmembersdonothaveeitherpopulationorGDP,therearecertainfactorsinHathiTrustthatsatisfytheprinciplesofbothequityandpower,respectively,andcouldbesubstitutedintheformulas.Theprincipleofequity,forHathiTrust,couldberoughlyfulfilledbyafactorofthenumberofvolumescontributedtotherepository.Theprincipleofpower,meanwhile,couldbesignifiedbyfinancialcontribution.Substitutingthesefactorsfortheoriginalones,wehave:

1. Numberofvolumes+financialcontribution2. Numberofvolumesxsquarerootoffinancialcontribution3. Log(numberofvolumes)+log(financialcontribution)

Totheseformulascanbeaddedafourthformulathatdoesnotappearintheliteraturebutreflectssomeofthereasoningbehindthe2ndand3rdformulaabove,andturnsouttobebetterforHathiTrust’sneeds:

Page 18: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

18

4. Squarerootofvolumes+squarerootoffinancialcontribution

ThisformulawillbetherecommendedformulaforHathiTrust,butletusexplorehowthefourformulascompare.

First,herearetheprojectedestimatedvaluesforeachHathiTrustinstitution’sfinancialcontributionandvolumescontributed,asofMarch2011.Thesenumbersaremerelyguessesatthispoint,andcanbeadjustedinthefuture:

Institution Financialcontribution Volumes

UniversityofMichigan 2,155,404 5,000,000IndianaUniversity1 900,000 600,000CommitteeonInstitutionalCooperation 1,650,000 1,000,000UniversityofCalifornia 1,244,142 3,000,000UniversityofVirginia 72,000 500,000ColumbiaUniversity 72,000 500,000YaleUniversity 40,000 30,000TOTALS 6133546 10,630,000

Itismoreequitabletousetheseprojectedamounts,ratherthancurrentamounts.Someinstitutionshavemorevolumescurrentlyintherepositorysimplybecausetheyweretherefirst.Inaddition,asexplainedpreviously,Leech’salgorithmofcalculatingvotingweightforagivenpowerdistributionworksfarbetterongroupsofmorethanfivevoters.

Theattachedspreadsheetshowsdifferentpowerandweightvaluesofthesefourformulas,accordingtotheinstitution.Thepowervalueswerecalculatedaccordingtotheformulas,whiletheweightswerecalculatedseparatelyusingLeech’salgorithm.Notethatchangingtheinputvaluesofvolumesorfinancialcontributionswillreadjustthepowervalues,butwillnotre‐calculatetheweights.Lookingatthevaluesthatresultfromeachoftheseformulas,itisimmediatelyapparentthatcertainformulascompensatefornumericalextremesbetterthanothers,andthatsuchcompensationisanextremelyimportantcriteriaforselectingapowerallocationformula.

Informula1,theUniversityofMichiganhas43%ofthetotalpower,whileYaleUniversityreceivesonly0.42%ofthepower.Suchanextremedifferencewouldmostlikelybeconsideredunacceptableandunfairbyallofthepartners,andcertainlybyYale.Thefollowingtablegivestheformula1powerandweightvalues:

Formula1:(financialcontribution)+(volumes)

1TogenerateIU’svotingpower,IU’svolumeestimateshavebeenseparatedfromthetotalCICvolumes.

Page 19: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

19

Institution Power RelativePower2 RelativeWeights3UniversityofMichigan 7,155,404 42.68% 41.73%IndianaUniversity 1,500,000 8.95% 7.81%CommitteeonInstitutionalCooperation 2,650,000 15.81% 15.60%UniversityofCalifornia 4,244,142 25.32% 26.73%UniversityofVirginia 572,000 3.41% 3.87%ColumbiaUniversity 572,000 3.41% 3.87%YaleUniversity 70,000 0.42% 0.37%TOTALS 16,763,546 100% 100%

Onepositiveaspectofformula1isthatthepowerratiosareveryclosetotheweightratios.Thisstrongcorrelationmakestheweightsmoreintuitiveandthereforemoreacceptable.Still,thewildlyunevenpowerallocationmakesthisformulalessthanideal.

Formula2resultsinevengreaterhighsandlowsofpowervalues,andsomeoftheweightsgeneratedasaresultoftheseextremesarezeroornegativevalues.Obviously,negativeweightvaluesareuselessfromapracticalandtheoreticalstandpoint.Herearethevaluesresultingfromformula2:

Formula2:volumesxsquarerootof(financialcontribution)Institution Power RelativePower RelativeWeightsUniversityofMichigan 7,340,647,110 57.28% 50.08%IndianaUniversity 569,209,979 4.44% 0.00%CommitteeonInstitutionalCooperation 1,284,523,258 10.02% 0.06%UniversityofCalifornia 3,346,233,405 26.11% 50.01%UniversityofVirginia 134,164,079 1.05% ‐0.07%ColumbiaUniversity 134,164,079 1.05% ‐0.07%YaleUniversity 6,000,000 0.05% ‐0.01%TOTALS 12,814,941,909 100% 100.00%

Note,also,thatthepowervaluesandweightvaluesaccordingtothisformulaarequitedifferent.Arelativepowerof10.02%yieldsaweightof0.06%,whichseemscounterintuitiveevenifitismathematicallycorrect.

UnlikeFormulas1and2,whichresultinextremehighsandlowofpowerallocations,Formula3attemptstoalleviatetheseextremesbytakingthelogarithmofeachfactor.Theresultingpowerallocationisperhapsnotdiverseenough,withallofthepercentageslyingveryclosetogether:

2RelativePowerrepresentstheinstitution’spowerdividedbythetotalpowerofallinstitutions.3RelativeWeightsrepresenttheclosestpossiblematchofweightratios,accordingtoLeech’salgorithm,thatyieldthedesiredpowerallocation.Anyweightsthatmaintainthegivenratioswillresultinthesamepowerallocation.Thesimplestwaytomaintaintheratioswouldbetoallocatevotesasintegerstotaling100.

Page 20: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

20

Formula3:log(financialcontribution)+log(volumes)Institution Power RelativePower RelativeWeightsUniversityofMichigan 13.03 16.34% 17.51%IndianaUniversity 11.73 14.71% 14.96%CommitteeonInstitutionalCooperation 12.22 15.32% 15.91%UniversityofCalifornia 12.57 15.77% 16.61%UniversityofVirginia 10.56 13.24% 12.64%ColumbiaUniversity 10.56 13.24% 12.64%YaleUniversity 9.08 11.39% 9.73%TOTALS 79.75 100.00% 100%

Sofar,twoofthethreeformulashaveresultedintoomanyextremevalues,andthethirdyieldsapowerallocationthatistooevengiventhevariationinpartnercommitmentstoHathiTrust.Evidently,aswiththeoverallvotingmodeldesign,itisdifficultwiththeseformulastoachieveamiddlegroundbetweenextremeinequalityandcompleteequality.The4thformula,derivedinresponsetothisneedforamiddleground,seemstoachievethenecessarybalance.Thosevaluesareasfollows:

Formula4:squarerootof(financialcontribution)+squarerootof(volumes)Institution Power RelativePower RelativeWeightsUniversityofMichigan 3,704 28.75% 28.30%IndianaUniversity 1,723 13.38% 13.98%CommitteeonInstitutionalCooperation 2,285 17.73% 17.84%UniversityofCalifornia 2,847 22.10% 21.69%UniversityofVirginia 975 7.57% 7.70%ColumbiaUniversity 975 7.57% 7.70%YaleUniversity 373 2.90% 2.78%TOTALS 12,884 100% 100%

Takingthesquarerootofeachfactor,andthentotalingthem,mitigatestheproblemofextremes,butnotasthoroughlyastakingthelogarithm.Itseemstorepresentanacceptabledistributionofpower,andinaddition,thecorrespondingweightsareextremelyclosetotheactualpowerdistribution.Theclosenessofweightandpowerservestomakethisformulamoreacceptableandintuitive.ThefinalrecommendationforHathiTrust,therefore,istouseformula#4toallocatepower,equatingvotingpowerwiththesumofsquarerootsofaninstitution’sfinancialcontributionandnumberofvolumescontributed.Thisweightedvotingformulawillserveasacorecomponentofalargerdoublemajorityscheme,satisfyingtheprinciplesofbothequalityandpowerdisparitiesinHathiTrust.Toseehowthisproposedvotingmodelwouldplayoutinrealdecision‐making,letusexamineafewpotentialscenarios.

Page 21: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

21

C. PotentialVotingScenarios

Althoughwecanonlyguessatconfigurationsofvoterpreferencesforanygivenquestion,itisworthwhiletolookathowthevoteswouldplayoutforafewhypotheticalsituations.AttheCC,thepartnerswillrespondtorecommendationsfromtheHathiTrustthree‐yearreview.ImaginethatoneoftheserecommendationspertainstowhetherornottheHathiTrustmodelshouldcontinuetocombinepreservationandaccess.The3‐yearreviewrecommendsthatHathiTrustcontinuetoembracetheconceptofa“lightarchive,”enablingbothpreservationandaccess.NowimaginethatofthepartnersexceptfortheUniversityofMichigansupporttherecommendation.TheUniversityofMichiganopposesthismeasureandsupportsconvertingHathiTrustintoa“darkarchive”onthegroundsthatitismoreeconomicalandlesscomplex.Accordingtoformula4,theweightedvoteswouldplayoutasfollows:

Institution Weights(Integers) Maintain“lightarchive”?UniversityofMichigan 28.3 NOIndianaUniversity 14.0 YESCommitteeonInstitutionalCooperation 17.8 YESUniversityofCalifornia 21.7 YESUniversityofVirginia 7.7 YESColumbiaUniversity 7.7 YESYaleUniversity 2.8 YESTOTALS/OUTCOME 100.0 YESNUMBEROFYES n/a 71.7NUMBEROFNO n/a 28.3

Tomakethetablemorereadable,theweightshavebeenturnedintopositivenumberstotaling100,roundedofftothenearesttenth.WithallofthepartnersexceptforUniversityofMichigancominginat71.7%ofthevote,theirpreferencetokeepthelightarchivemodelwillpassintheweightedvotingphase.

Accordingtotheothercomponentsoftherecommendedvotingmodel,thereisalsothedoublemajorityrequirementtobeconsidered,aswellasthefactthatMichiganhasavetopower.Thisscenarioalsoeasilypassesaccordingtotheunweightedvotescomponentofthedoublemajority,withsixoutofsevenmembersinagreement.Italsohappens,though,thattheonememberindisagreementhasvetopower.ThefactthattheUniversityofMichiganpossesstheabilitytovetodoesnotautomaticallymeanthatUMwouldexercisethatpower.Inthishypotheticalsituation,Michiganwouldhavetocarefullyconsidertheopinionsoftheotherpartnersaswellasthepoliticalramificationsofopposingthecollectivemembership.IfMichigandidchoosetovetothedecision,itwouldnotautomaticallymeanthatthevotewouldgoMichigan’sway;rather,thequestionwouldrequirefurtherdiscussion,possiblealterationstotheproposal,andanothervote.

Page 22: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

22

TakingthissamehypotheticalpropositionofmaintainingHathiTrustasa“lightarchive,”nowsupposethatIndianaandCaliforniajoinMichiganinopposingthemeasureandvotingtoconvertHathiTrustintoadarkarchive.Theweightedvoteswouldlooklikethis:

Institution Weights(Integers) Maintain“lightarchive”?UniversityofMichigan 28.3 NOIndianaUniversity 14.0 NOCommitteeonInstitutionalCooperation 17.8 YESUniversityofCalifornia 21.7 NOUniversityofVirginia 7.7 YESColumbiaUniversity 7.7 YESYaleUniversity 2.8 YESTOTALS/OUTCOME 100.0 NONUMBEROFYES n/a 36.0NUMBEROFNO n/a 64.0

Inthisscenario,the“no’s”gettheweightedvotewitha64%majority.Buttakingintoconsiderationthedoublemajorityrule,the“yes”voteshavefouroutthesevenunweightedvotes.Themeasureneitherpassesnorfails,andthepartnersgobacktodiscussion.Inasituationlikethisone,wherethereareconflictingoutcomesbetweentheweightedandunweightedvotes,itisclearthathavingadoublemajorityrulemakespassingadecisionmoredifficult.Thiscouldbeseeninapositivelight—anopportunityforthepartnerstohavefurtherdiscussionandperhapsmodifytheproposal—orinanegativelight,asanobstructiontodecision‐making.

Thereareaninfinitenumberofscenariostobeconsidered,andtothispurpose,theattachedspreadsheetincludesa“votecalculator”basedonformula4.

V. Conclusion

Torevisittheinitialgoalsputforthinthisdocument,thequestionsthatthisdocumentsetouttoanswerwere:

• Howisvotingpowercalculatedinaweightedvotingsystem?• Whatfactorsandprinciplesshouldinformallocationofvotingpower?• HowdothesemodelsapplytoHathiTrust?

Thesethreequestionswereansweredthroughanexplorationofapriorivotingpowertheory,anexaminationofthebasicprinciplesthatvarioustheoristshavesuggested,andfinally,adetailedsetofrecommendationsforHathiTrustalongwithseveralpotentialscenariosandtheiroutcomesaccordingtotherecommendedvotingmodel.Thatrecommendedmodel,whichinvolvesthreecomponents—weightedvotes,unweightedvotes,andinstitutionalvetopower—attemptstosatisfytheprinciplesofequity,power,andacceptabilitydetailedbytheoristsandalsoevidentinHathiTrust’sonlinedocumentationoftheproject’smission.

Page 23: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

23

ThegoalofthisdocumenthasbeennotonlytorecommendaspecificvotingmodelforHathiTrustasitiscomprisedintheforeseeablefuture,butalsotooutlineflexibleapproachesandformulasthatcanbeadaptedasthemembershipandcontributingfactors(financialcontributionandnumberofvolumes)change.Forexample,ifthehostinstitutionchanges,itmaynolongerbepracticaltogivetheUniversityofMichiganvetopower.Intheweightedvotingcomponent,itmightmakemoresensetouseadifferentformulaforpowerallocationifthefactorschange;formula3,forinstance,equalizespowertoomuchinthecurrentconfiguration.Withmoreextremedifferencesincontributingfactors(financialcontributionsandnumberofvolumes),formula3mightworktooffsetthosedifferences.Inaddition,theexecutivecommitteemaywishtoconsiderdifferentbreakdownsofthemembership,suchasbreakingtheCICandUCintotheirrespectivememberinstitutionstobeuniqueHathiTrustpartners.AstheHathiTrustDigitalLibrarywillcertainlyevolveinunexpectedways,itsdecision‐makingprocessesshouldalsoevolvetosupportthosechanges.

Page 24: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

24

BibliographyAlbert,M.2003.“Thevotingpowerapproach:measurementwithouttheory.”EuropeanUnionPolitics4(3):351‐366.Banzhaf,J.1965.“Weightedvotingdoesn’twork:amathematicalanalysis.”RutgersLawReview19(2):317‐343.Barrett,C.andH.Newcombe.1968.PeaceResearchReviews2(2).Barthélémy,F.andM.Martin.2007.“ConfigurationstudyfortheBanzhafandtheShapley‐Shubikindicesofpower.”THEMAWorkingPaper,accessibleathttp://www.u‐cergy.fr/thema/repec/2007‐07.pdfBlack,D.1998.Thetheoryofcommitteesandelections.2nded.Norwell,MA:Kluwer.Brams,S.andP.Fishburn.1982.Approvalvoting.Boston:Birkhäuser.Brauninger,Thomas.“WhenSimpleVotingDoesn’tWork:MulticameralSystemsfortheRepresentationandAggregationofInterestsinInternationalOrganizations.”BritishJournalofPoliticalScience(33):681‐703.Coleman,J.1986.Individualinterestsandcollectiveaction.Cambridge:CambridgeUniversityPress.Dixon,W.1983.“TheevaluationofweightedvotingschemesfortheUnitedNationsGeneralAssembly.”InternationalStudiesQuarterly27(3):295‐314.Dreyer,J.andSchotter,A.1980.“PowerrelationshipsintheInternationalMonetaryFund:theconsequenceofquotachanges.”TheReviewofEconomicsandStatistics62(1).Dummett,M.1984.Votingprocedures.Oxford:ClarendonPress.Farquharson,R.1969.Theoryofvoting.NewHaven:YaleUniversityPress.Felsenthal,D.andM.Machover.2004.“Apriorivotingpower:whatisitallabout?”Politicalstudiesreview2004(2):1‐23.Felsenthal,D.andM.Machover.1998.Themeasurementofvotingpower.Cheltenham,UK:EdwardElgar.Gelman,A.,J.KatzandJ.Bafumi.2004.“Standardvotingpowerindexesdonotwork:anempiricalanalysis.”BritishJournalofPoliticalScience34:657‐674.Gold,J.“DevelopmentsintheLawandInstitutionsofInternationalEconomicRelations.”TheAmericanJournalofInternationalLaw68(1974):687‐708.Lane,J.andS.Berg.1999.“Relevanceofvotingpower.”JournalofTheoreticalPolitics11(3):309‐320.

Page 25: Voting Models 2009-11-16 - HathiTrustideological proximity) between voters.” (Felsenthal and Machover 1998) In other words, the systems that theorists have used to determine voting

25

Leech,D.2002.“VotingpowerinthegovernanceoftheInternationalMonetaryFund.”AnnalsofOperationsResearch109:375‐397.Leech,D.2003.“Powerindicesasanaidtoinstitutionaldesign:thegeneralizedapportionmentproblem.”InEuropeanGovernance,editedbyM.Holleretal.,107‐121.Tübingen:MohrSiebeck.Leech,D.andM.Machover.2003.Qualifiedmajorityvoting:theeffectofthequota[online].London:LSEResearchOnline.Availableat:http://eprints.lse.ac.uk/archive/00000435Margolis,H.1983.“TheBanzhaffallacy.”AmericanJournalofPoliticalScience27(2):321‐326.Newcombe,H.,J.Wert,andA.Newcombe.“ComparisonofWeightedVotingFormulasfortheUnitedNations.”WorldPolitics23(1971):452‐92.O’Neill,B.andB.Peleg.“VotingbyCountandAccount:ReconcilingPowerandEqualityinInternationalOrganizations,”(2000)http://www.sscnet.ucla.edu/polisci/faculty/boneill/c&a.html(accessedJuly9,2009).Pajala,A.andM.Widgrèn.2004.“AprioriversusempiricalvotingpowerintheEUCouncilofMinisters.”EuropeanUnionPolitics5(1):74‐97.Penrose,L.1946.“Theelementarystatisticsofmajorityvoting.”JournaloftheRoyalStatisticalSociety109(1):53‐57.Rapkin,DavidandJonathanR.Strand.“ReformingtheIMF’sWeightedVotingSystem.”TheWorldEconomy29(2006):305‐24,accessibleathttp://papers.ssrn.com/sol3/papers.cfm?abstract_id=924287.Shapley,L.andM.Shubik.1954.“Amethodforevaluatingthedistributionofpowerinacommitteesystem.”TheAmericanPoliticalScienceReview48(3):787‐792.Tideman,N.2006.Collectivedecisionsandvoting:thepotentialforpublicchoice.Aldershot,England:Ashgate.Turnovec,F.1997.ThedoublemajorityprincipleanddecisionmakinggamesinextendingEuropeanUnion.EastEuropeanSeries,No,48.Vienna:InstituteforAdvancedStudies.