Democracy's Wisdom: An Aristotelian Middle Way for Collective

20 Pages·2013·0.26 MB·English

psr1200062 xxx(xxx) January4,2013 22:3
AmericanPoliticalScienceReview Vol.00,No.0 January2013
doi:10.1017/S0003055412000627
Democracy’s Wisdom: An Aristotelian Middle Way for
Collective Judgment
Q1
JOSIAH OBER StanfordUniversity
Asatisfactorymodelofdecision-makinginanepistemicdemocracymustrespectdemocraticval-
ues, while advancing citizens’ interests, by taking account of relevant knowledge about the
world. Analysis of passages in Aristotle and legislative process in classical Athens points to
a “middle way” between independent-guess aggregation and deliberation: an epistemic approach to
decision-making that offers a satisfactory model of collective judgment that is both time-sensitive and
capable of setting agendas endogenously. By aggregating expertise across multiple domains, Relevant
ExpertiseAggregation(REA)enablesabodyofminimallycompetentvoterstomakesuperiorchoices
among multiple options, on matters of common interest. REA differs from a standard Condorcet jury
in combining deliberation with voting based on judgments about the reputations and arguments of
domain-experts.
Ifademocracyistoberobustlysustainedovertime, Plato (in the Republic), along with other ancient
public decisions must respect democratic values, and modern critics, argued that democracy’s commit-
while advancing citizens’ interests. Democracy is menttolibertyandpoliticalequalitynecessarilyleads
a sham if meaningful decisions, leading to significant citizens to pursue arbitrary desires rather than real
publicoutcomes,arenotmadebyfreecitizens,secure interests, and to make choices based on false opin-
intheirdignity,actingaspoliticalequals(Ober2012). ion rather than knowledge. The critics conclude that
Butlegislativeprocessmustdomorethanexpressval- democracy is inherently anti-epistemic and that only
ues. It must seek to achieve outcomes aligned with anon-democraticregimecouldmakepolicyfavorable
citizens’ interests in, for example, security, fair rules, to people’s real interests (Roberts 1994; Ober 1998).
andsocialwelfare.Allotherthingsbeingequal,antic- If they are right – if, in contrast to a well-ordered
ipatedoutcomesaremorelikelytobeachievedwhen epistemic autocracy, democracy’s core values render
legislationispredicatedonknowledgeaboutrelevant a democratic collectivity inherently incapable of em-
featuresoftheworld.Sinceantiquity,politicaltheorists ploying knowledge to make policies leading to favor-
have asked whether a political regime can be at once ableoutcomes–wemustaskwhethersustainingvalues
democraticandepistemic.Canpolicy-makingprocesses justifiesthecostof,forexample,lesssecurity,lessfair
expressdemocracy’scorevaluesandservecitizen’sin- rules, and less social welfare. It obviously would be
terests when decisions are based on well-justified be- betterforthosewhocherishdemocraticvaluesifthat
liefs, rather than ill-founded popular opinions? How questionweremoot.
a democratic community might employ knowledge in The promise of epistemic democracy is that, under
choosing among alternatives is a question of institu- therightconditions,aprocessofdecision-makingthat
tional design that concerned classical Greek political expressesandsustainsdemocraticvaluescandobetter
theorists and that remains central for contemporary than making random choices among policy options –
political scientists (Callander 2011). It is a pressing andthuscanpromotetheinterestsofcitizensbyachiev-
question,notleastbecauseitexceedstheboundsofthe ingrelativelyfavorableoutcomes.Ifthatisthecase,and
state.Universities,businessfirms,NGO’s,federations, if no non-democratic epistemic process can be shown
andtransnationalagenciesallconfrontthequestionof to do better, the presumptive normative benefits of
how many individuals, who share certain interests in liberty,politicalequality,andcivicdignityneednotbe
common,canchoosewiselyamongavailableoptions.1 traded off against the expected costs of inferior pol-
icy.Insum,ifepistemicdemocracy’spromisewereful-
JosiahOberisTsakopoulos-KounalakisProfessorinHonorofCon- filled, we could add interest- and-outcome-based rea-
stantineMitsotakis,DepartmentsofPoliticalScienceandClassics, sonstovalue-based reasons forpreferringdemocracy
StanfordUniversity,100WestEncinaHall,616SerraStreet,Stanford
to“epistocracy.”2
CA94301([email protected]).
Thispaperonexpertisecouldnothavebeenwrittenwithoutthe
helpofexpertsinseveraldomains.MythankstotheUCLAeditorial
board(especiallyKirstieMcClure),totheUNTboard(especially omy,reciprocity,fallibility,experimentalism,transparency,andprac-
StevenForde),andtothejournal’sanonymousreadersforthoughtful ticalreason:Brettschneider2007;Christiano2011;andworkscited
comments and expert guidance. For helpful discussion, comments inwhatfollows.Expressionofdemocraticvaluesinmajoritariande-
on drafts, bibliography, and access to unpublished work to I owe cisionprocesses:Waldron1999;Schwartzberg2007.Importantwork
specialthankstoChrisBobonich,BruceCain,DanielaCammack, onepistemicdemocracyincludesCohen1986,1996;ListandGoodin
Alexandre Debs, Brian Garsten, Paula Gottlieb, Justin Grimmer, 2001;List2005;BovensandRabinowicz2006;Anderson2006;Page
Monte Johnson, Lily Kelting, He´le`ne Landemore, Melissa Lane, 2007;Estlund2008;Furstein2008;Fischer2009;Schwartzberg2010;
ArthurLupia,HelenMilner,RobertMorestein-Marx,PhilipPettit, ListandPettit2011(chapter4);ElsterandLandemore2012.Areal-
andGlenWeyl.VersionsofthispaperwerepresentedattheAPSA istictheoryofepistemicdemocracyoughttobeincentivecompatible
(2009);attheUniversitiesofBergen,Chicago,Colorado,andLille; (Ober2008:5–22)butnoformalmodelisofferedhere.
andatHarvard,Stanford,UCSB,UCSD,andYale. 2Estlund(2003,2008),whocoined“epistocracy”asatermforruleby
1Inadditiontothecorevaluesofliberty,equality,anddignity,demo- experts,developsamoralargumentagainstit.Myargumentsuggests
cratictheoristssometimesincludevaluesof(amongothers)auton- thattheargumentforepistocracygoeswrongattheoutsetbecause
1
psr1200062 xxx(xxx) January4,2013 22:3
Democracy’sWisdom January2013
One enabling condition for epistemic democracy is ofREAundercompleterules.Inthisfirst,epistocratic,
participationindecision-makingbycitizenswhoattend model,thebestchoiceamongthreeoptionsisafunc-
torelevantsourcesofknowledge,totrueexpertiseand tionoftheaggregatedrelevance-weightedrankingsof
genuineexperts.Yethowcanattentiontoexpertsavoid individual experts who are assumed to be infallible.
devolution into rule by experts, thus leading to the Thefifthsectionexplainshowexpandingthemodelto
eclipseof politicalequality, and at least potentially to includemassvotingsolvestheproblem,raisedbyAris-
theeclipseoflibertyanddignityaswell?Thequestion totle,offallibleexperts–andtherebybringsREAout
of whether a democracy can make appropriate use of of epistocracy into the realm of epistemic democracy.
diverseformsofexpertise,whilepreservingitscoreval- The sixth section introduces a less stylized model of
ues,hasconcernedpoliticaltheoristsandpractitioners REAunderincompletelyspecifiedrules.Theproblems
since antiquity. It remains an issue for contemporary andpotentialofthisversionareillustratedbyreference
theorists.PhilipPettit,forexample,arguesforadelib- todecision-makingindemocraticAthensin481BCE.
erativeconstitutionalorderinwhich“authorial”power The final section (7) concludes by suggesting that a
to legislate lies with depoliticized deliberative bodies groupemployingREAmightimproveitsperformance
possessingappropriateprofessionalexpertise,whileor- overtimethroughlearning,andsketchesthekindsof
dinarycitizensarereducedtoan“editorial”role(Pettit empirical studies that will be be required if the theo-
2004:57–62;contra:Urbinati2012).Democracy’srela- reticalaccountofdemocracy’swisdomofferedherein
tionship to expertise is a live issue in current policy istobetested.
debates,withcriticscontendingthatdemocracy’santi-
epistemic character renders it unequal to, for exam-
ple, the challenge posed by long-term climate change
INTERESTS, KNOWLEDGE, EXPERTS
(ShearmanandSmith2007).
This article seeks to define an appropriate role for How ought a democratic process to make decisions,
expertknowledgeinarobustdemocracyinwhichleg- ifitistosustaindemocracy’scorevaluesandpromote
islative authority is retained by citizens. It does so citizeninterests?Oneinfluentialansweristoaggregate
by drawing upon two promising trends in democratic the preferences (over representatives or policies) of
theory: adapting Greek political theory (e.g. Wilson freecitizensbycountingtheirequallyweightedvotes.
2011) and political practice (e.g. Schwartzberg 2010) Robert Dahl (1989, 1998), among others, has argued,
to contemporary purposes, and seeking a productive againstPlatoandotherepistemicregimetheorists,that
middle ground between deliberative and aggregative democraticvaluesarepreserved,andcitizens’interests
approaches to democratic process (e.g. Dryzek and advanced, when policy is set by a majority of voters
List 2003). Borrowing from Aristotle’s discussion of whose preferences express their own opinions about
the“wisdomofthemany”andfromancientAthenian theirownbestinterests.Dahl’sapproachpreserveslib-
politicalpractices,Isuggestonewaythatthepromise erty, political equality, and dignity by asserting that
ofepistemicdemocracymightbefulfilledthroughap- each individual voter is the best (even if necessarily
propriateinstitutionaldesign. imperfect) judge of his or her own interests, and that
The following section (1) establishes the precon- a majority of such individually-chosen interests, ex-
ditions for epistemic democracy, notably the identifi- pressed as equal votes, deserves to be established as
cation of common interests and relevant domains of statepolicy.Yetthemajority’spreferences,evenifthey
expertise. The next section (2) compares salient fea- dotracktherealinterestsofthemajority,willoftenfail
tures of deliberation and independent guess aggrega- to reflect the interests of all citizens. It is because the
tion with Relevant Expertise Aggregation (REA) – preferences of a majority might ignore or harm the
a “middle-way” system for making good decisions mostbasicinterestsofindividualsorofaminoritythat,
among two or more options on issues with multiple in liberal democracies, certain fundamental interests
relevant criteria. In REA the best overall choice is arelegallyprotectedasrights.
a function of how the options score in terms of the Classicaltheoristsapproachedthematterofpreserv-
criteria.Eachcriterionisdefinedasarelevantdomain ingvaluesandfosteringinterestssomewhatdifferently.
ofexpertise.Optionsarerankedbyexpertsineachdo- Aristotleregardedadvancingthespecialinterestsofa
main,orbymassvotingbasedonrecommendationsof majorityinademocracy(orofanempoweredminority
multipleexperts.Thethirdsectionre-analyzesthewell- inanoligarchy)asunjust,ifitcameattheexpenseof
known “wisdom of the many” passage in Aristotle’s promotingcommoninterests.Hesupposedthatajust
Politics(3.11),arguingthatitisacompressedaccount community would identify common (rather than par-
of REA. The fourth section employs a passage from tial/factional)interestsand,byappropriateuseofprac-
Aristotle’s Poetics to fill out a non-democratic model tical wisdom (phroneˆsis), would select policies most
likely to advance those interests. Political decision-
making, for Aristotle, was an epistemic endeavor in
itwronglysupposesthat,becausethereareexpertsindomainsrel- that it was meant to discover the best answers to
evant to politics there are also general experts in politics (as op- questions of appropriately-shared concern. If we are
posedtorelativelycompetentpoliticalleaders).Forameta-analysis willingtoacceptthecommon(althoughnotuniversal)
ofthevast“democracyandeconomicperformance”literature,see
thoughtsthat(1)peopledohaverealinterests(andnot
DoucouliagosandUlubas¸ogˇlu2008,whoconcludethatreal-world
merelypreferences)and(2)thatsomeinterestsarein
modern democracies do no worse on this outcome-measure than
non-democracies. factsharedbysomemembersofacommunity,thereis,
2
psr1200062 xxx(xxx) January4,2013 22:3
AmericanPoliticalScienceReview Vol.00,No.0
byextension,noreasontorejectapriorithepossibility outcomes, must be transitively ordered, such that if
thatcertaininterests(e.g.insecurity,fairrules,welfare) A > B and B > C then A > C.4 Collective decisions
couldbesowidelysharedastobereasonablydescribed made through the participation of many individuals,
as commonly held. Moreover, one need not embrace includingdecisionsaboutcommoninterests,canbere-
Aristotelian eudaemonism to regard pursuing certain garded as the rational judgment of a collective agent
commoninterestsasanormativelychoice-worthypo- without reference to metaphysically mysterious con-
liticalgoal,ortosupposethatpracticalwisdomisequal ceptions of agency only when decisions are made un-
tothetaskofidentifyingcertaininterestsasshared.For dertherightconditions(ListandPettit2011).Finally,
thepurposesofthisarticle,asuperiorcommon-interest by definition, epistemic democracy must decide mat-
choiceisonethatselectstheavailableoptionthat,all ters democratically, avoiding capture by elites. I will
thingsconsidered,bestadvancesaninterestsharedby hope to show that Relevant Expertise Aggregation,
themembersofacommunity.3 the “middle-way” decision-making process described
I do not propose that we follow Aristotle in as- here, addresses the challenges of transitivity, agency,
suming that politics can or ought to be reduced to a andelitecaptureaswellorbetterthandeliberationand
search for the best answers about common interests independent guess aggregation in their conventional
andtheiradvancement:Manypoliticalmattersinvolve forms.
hard choices among conflicting social values; many Insofarasdemocraticpoliticsisameansforchoosing
othermattersconcerntheinterestsonlyofcertainper- whichavailableoptionbestservesasharedinterest,it
sons.Yetitseemsimplausibletosaythatpoliticsnever involves the use of knowledge (accurate information,
is,andneveroughttobe,concernedwithintereststhat truebeliefs)fordiscoveringthebestoption.Givenin-
are reasonably held to be shared. Democratic politi- determinacyandcontingency,thiscannotmean“iden-
cians, in antiquity (Ober 1989) and modernity alike, tifyingtheapproachthatinfalliblyachievesacommon
frame proposals as promoting the common interest; interest.” It must mean instead, “choosing the option
theirrhetoricisnotemptyinsofarasitacknowledgesa amongavailablealternativesthathasthebestchance,
sharedconvictionthataddressingcommoninterestsis allthingsconsidered,ofadvancingacommoninterest.”
at least part of what politics ought to be about. Com- Twofundamentalpremisesofepistemicapproachesto
mon interests might, hypothetically, be identified by democracyarethat(1)therearebetteroptions,inthat
consensus, but, for my purposes, consensus is unnec- thechancesofagoodoutcomearebetterifthatoption
essary.Majoritarianmechanismsmaybeemployedto ischosen,and(2)decision-makerscan,undertheright
identify and advance a common interest, so long as conditions,identifybetteroptions.5
the majority decision actually represents a superior Options are likely to be better insofar as they
common-interest choice, as defined in the previous take fuller account of relevant facts about the world.
paragraph. It is implausible to suppose that majority Reality-trackingisvalued(inthiscontext)becauseof
votesalways(oroften)accuratelyidentifyoradvance itsconsequences:Allotherthingsbeingequal,options
commoninterests.Itissufficientformyargumentthat thattakeaccountoftherelevantfactsaremorelikely
certain interests are sometimes shared and, in such toleadtoabetteroutcome,allthingsconsidered,than
cases, there is such a thing as the better policy. Al- those that do not. In this article, I am not concerned
though common and partisan interests can never be withassessingthemoralstatusofinterestspursuedby
fully disentangled, focusing on shared interests limits a democratic community (e.g. by weighing the shared
thesubjectivityofdecision-makingandtherebyallows interests of a community against global interests or
theperformanceofdecision-makinggroupstobeeval- universal human rights). Nor am I concerned with
uated(YatesandTschirhart2006). the normative value of epistemic democracy relative
To be plausible on the face of it, any approach to to, e.g., agonistic pluralism (Honig 1993; Lacau and
epistemic democracy must address the challenges of Mouffe 2001). Nor, finally, am I concerned to show
transitivity (and thus of cycling), collective rational- thatepistemicdemocracieswillinvariablyout-perform
ity, and elite control. Options, in order to have stable autocracies. I am concerned with designing institu-
tional mechanisms that enable citizens in a democ-
racy to make relatively better decisions, and thereby
3Aristotleonjusticeasacommoninterestinthepublicgood:Politics advance common interests, while sustaining core
3.1278b20–25withGottlieb2009:201–202n.17.BecauseAristotle values.
supposed that there was a single, specifiable human good (Kraut
2002),“thecommoninterest”mightbetakenasthetruthaboutthe
humangood.Ipropose,instead,thatacommoninterestisidentified
whenthepreferencesforageneraloutcomeamongthemembers 4Oncyclingasanissueinepistemicdemocracy,seeListandGoodin
ofareasonabledecision-makinggroup(i.e.agroupinwhichpref- 2001;DryzekandList2003;List2011;Elster2011,withworkcited.
erencestrackinterests)seekingapublic(non-exclusive)goodare 5AsCallander(2011)notes,decision-makerscanneverbesure,ex
well-alignedandwhenthedignityofeachandallisrespected.Here, ante,thattheyhavechosenwell,norcantheybesure,expost,that
reasonably-heldinterestsaretakenasthefunctionalequivalentof anoutcomecameaboutbecauseoftheirchoiceBut,ashedemon-
realinterestsand“reasonable”referstothelevelofpracticalwis- strates,thisneednotobviatethehopeofidentifyingbetteroptions.
dom(reflectiononwhatoughttobedone,givingdueattentionto Hawthorne (und: 5) argues that “for a wide range of philosophi-
relevantfacts,causes,andeffects)assumedtopertaininthedecision- callyrespectableviews[onthepublicgood,citingAristotle,Locke,
makingbodyofordinarypersons,discussedinthethirdsection.Cf. Rousseau,Mill,Rawls]thereissuchathingasthebetterpolicyin
Pettit2004:59,oncommoninterestsas“publicvaluation.”Onthe atleastsomecasesand...thatsuchviewsmayfindaidandcomfort
challengeposedbysocialchoicetheorytodemocraticrationality,see fromwhatJuryTheoremsimplyabouttheabilityofmajoritiesto
note4,withdiscussioninsection2,below. findthebetterpolicy.”
3
psr1200062 xxx(xxx) January4,2013 22:3
Democracy’sWisdom January2013
Epistemicdecision-makingisnecessarilyconcerned elaborates upon passages in Aristotle’s texts. It is not
with expertise. Experts in a given domain (say, chess fullyspecifiedinthetextsbutitis,Ibelieve,aplausible
masters) are more capable than others at producing extension of Aristotle’s thinking on political science.
a desired outcome (winning) and the probability of Anapproachtocollectivejudgmentcloselyresembling
achieving the outcome is increased by better choices REAIIandIIIwas,Iwillargue,thebasisforAristotle’s
(goodmoves).TheCallipolisofPlato’sRepublicisan celebrated claim that, under the right conditions, the
example of an ideal epistemic regime in which rulers manymaybewise:abetterjudgeofsomemattersthan
are experts. Their choices accurately track the Form anexcellentindividualorsmallgroup.8
of the Good and thereby produce a just society. Cal- The conditions necessary for REA are demanding,
lipolis is, however, neither realistic nor democratic. It but realistic: (1) Decision-makers addressing an issue
is undemocratic because a few experts rule without seek the best available option. (2) The issue is divisi-
consultingtheothercitizens.Itisunrealisticbecauseit bleintoparts,eachofwhichhasaspecifiable(relative
assumestheexistenceofgeneral,architectonicpolitical to other parts) relevance to the issue at question. (3)
experts. The philosopher-kings are “architect-like” in Eachofthoserelevantpartsisexplicableasadomain
their master knowledge of a blueprint that perfectly of knowledge that can be enlightened by identifiable
directs the actions of all those whose work (and sub- domain-experts willing and able to disclose private
sidiary expertise) is necessary to create and sustain a information.9 Certain forms of REA require a fur-
justsociety.AbsentaFormoftheGoodtowhichsome ther assumption that (4) conditions 1–3 are common
individualshaveprivilegedaccess,thereisnoreasonto knowledgeinagroupthatexistsovertimesuchthatits
believethatsuchgeneralpoliticalexpertsexist.6 decisions take the form of a repeated game, and that
Politics is unlike domains of endeavor in which in- (5)itsmembersupdatebeliefsaboutexpertsinoverall
dividuals achieve true expertise (say, chess or violin reality-trackingways.10
playing:Ericsson2006)initslevelofcomplexity.There
are,exhypothesi,nogeneralexpertsinpoliticsbecause,
DELIBERATION, INDEPENDENT GUESS
lackingaccesstotheFormoftheGood(orsomesimi-
AGGREGATION, AND MIDDLE-WAY
larmetaphysicalresource),suchexpertswouldneedto
DECISION THEORY
masterarangeofhard-to-acquirespecializedexpertise
that exceeds the bounds of human cognitive capacity. The primary mechanisms discussed in the epistemic
It is, however, highly plausible to suppose that there democracy literature are deliberation (DEL), under-
are true experts in many domains relevant to polit- stoodasaformofreciprocalreason-givingamongciti-
ical decision-making (Ericsson et al. 2006). Domain- zens(Cohen1996;GutmannandThompson2004),and
experts may come to believe, wrongly, that they are independent-guess aggregation (IGA), understood as
general experts. The catastrophic results that can fol- a method for mathematically aggregating votes with
lowwhenpoliticalauthorityiscededtodomain-experts a specified probability of being correct (Condorcet
are well documented (Scott 1998). But this ought not 1785;ListandGoodin2001).DELandIGAhavereal-
be a reason for excluding expertise in relevant do- worldapplications.DELhasbeenputintopracticeas
mains from democratic decision-making. The goal of
Relevant Expertise Aggregation is to bring relevant
domain-specificexpertiseintotheprocessofdecision- 8Thefollowingpassages,inadditiontothepassagesfromthePol-
makingwithoutcedingpoliticalauthoritytoexperts.7 itics and Poetics (sections 4, 5) and from the Rhetoric (notes 17,
REA captures some of the ways by which better 28),supporttheargumentthatArisotledoeshaveinmindanap-
proach similar to REA: Metaphysics 1.2.993a30-b7: no individual
options were chosen in antiquity and are chosen in
knowsthewholeofthetruth,buteveryonehaspartofit.Partsof
modernity. While there are epistocratic versions of Animals1.1.638a1–12:therearetwolevelsofproficiencyinevery
REA (REA I: fourth section, below), other versions human endeavor: true expertise and education adequate to form
(REA II and III: fifth and sixth sections, below) are judgmentsonexpertarguments;someindividualsarecompetentto
judgeargumentsinmostdomains,othersonlyinspecificdomains.
democratic in retaining the principle of equal votes
Rhetoric1.2.1357a2–4,1359b19–21:thegoalofdeliberativerhetoric
and(REAIII)inresistingelitecapturethroughagenda
isexpositionofcomplexmattersbyexpertstominimallycompetent
control.REAisAristotelianinthatitisinspiredbyand decision-makers(“personswhocannottakeinataglanceacom-
plicatedargument,orfollowalongchainofreasoning”)especially
inthegeneraldomainsofwaysandmeans,warandpeace,national
6Inhisearlywork,Protrepticus(fragmentcitedinIamblichus,Pro- defense,importsandexports,andlegislation.
trepticus10.54.12–56.2),Aristotleusesthearchitectasamodelforthe 9BovensandRabinowicz(2006)compareCondorcetjurymodels
ideallegislator.Aristotlelikewiseappliesan“architectonic”frame predicated on voting on parts with voting on complex issues as
topoliticalscience(NicomacheanEthics1.2.1094a26-b7),aswellas wholes.Their“premise-basedprocedure”differsfromREAinad-
tophilosophicalfoundations(Metaphysics1.2.981a30–982b7),butin dressingonlybinary(yes-or-no)questions,inrequiringamajority
theselaterworkshedoesnotsuggestthatthereisamaster“archi- “yes”voteoneachpremise(domain/part),andinassumingvoter
tect.”MythankstoMonteJohnsonforcallingtomyattentionthese independence.
andotherkeypassagesinAristotle,notablythosecitedinnotes8, 10Thefactthatindividualsholdprivateinformationrelevanttodeci-
17,and28. sionsaffectsbest-choicevotingmodelsinvariousways,dependingon
7Dividinganissueintodomainsinwhichexpertiseisidentifiable theassumptionsofthemodel:Austen-SmithandBanks1996;Fed-
is described in literature on decision expertise as “the process- dersenandPesendorf1996.TheREAmodelswouldbecomplicated
decompositionperspective”:YatesandTschirhart2006:426–2.See if we assume (as I have not) that domain-experts with especially
ibid:435onhowthemultiplefactorsinvolvedincomplexdecisions valuableprivateinformationareexecutives,officialsseekingreelec-
willpresumablyrender“trueacross-the-boarddecision-makingex- tion,oraresubjecttojudicialreview:seeCanes-Wrone,Herron,and
pertise”inanyindividual“exceedinglyrare.” Shotts2001;FoxandStephenson2011.
4
psr1200062 xxx(xxx) January4,2013 22:3
AmericanPoliticalScienceReview Vol.00,No.0
amethodofpollingandformakingpoliticaldecisions sibility,instability,andarbitrariness”(DryzekandList
(Fishkin2009;Baiocchi2005;WarrenandPearse2008). 2003, quote: 2). Deliberation has subsequently been
IGAincludespredictionmarkets(Sunstein2007),but incorporated into work on Condorcet juries (Gerardi
the primary mechanism considered here is the Con- andYariv2007;Elster2011).Thisarticlecontributesto
dorcet jury, in which (in its canonical form) jurors the“middleway”literatureofreconciliationbetween
make independent judgments and the likelihood of DELandIGAbydrawingfromancientGreekpolitical
each voter choosing correctly on a binary decision is thoughtandpractice.
assumedtobebetterthanhalf.Asthenumberofvotes Reconciliation is especially desirable because nei-
increases,themajoritychoiceisevermorelikelytobe therDELnorIGA,intheircanonicalforms,fullysuits
correct. theneedsofademocraticcommunitythatmustmake
In a spectrum of approaches for organizing diverse complex, highly consequential, time-sensitivechoices.
informationanddispersedknowledgeforthepurposes IGAsuffersfromademocraticdeficitinthatitispred-
ofdemocraticdecision-making,DELandIGA,intheir icated on agendas (issues and options) that are set
strong conventional forms, stand at opposite poles: exogenously. Lacking the opportunity to deliberate,
DEL assumes non-strategic exchange of information the voters constituting the group cannot themselves
and reasons, and regards that interaction as neces- determine the issues on which choices must be made,
saryforachievingbetteranswersandoutcomes.DEL ortheoptionsfromamongwhichtheywillchoose.Nor
values updating by decision makers, on the basis of can they, as a group, establish rules governing issue
new information and reasons offered by others, both selection or the option menu. The issues and avail-
as a means to achieving better outcomes and as an able options must ordinarily be presented to voters
expression of the values of equality and reciprocity. bysomeexternalauthority.11 Insofarasitsticksbyits
CanonicalformsofIGAassumevoterindependence– own premise of voter independence, IGA therefore
thereisnopre-decisioninformation-sharing.Indepen- comeswithexternal(presumptivelyelite)agendacon-
dence is valued as preserving freedom of individual trolbuiltin.
choice, but also because it prevents the informational Under DEL, issues and options may be set by an
cascades (group-think) and polarization (extremism) external authority. An example is a deliberative poll
that have been associated by Cass Sunstein, among in which the issue and options are set by the authori-
others,asinherentanti-epistemicfeaturesofdelibera- tiescommissioningand/orconductingthepoll(Fishkin
tion (Sunstein 2000, 2002; cf. Mendelberg 2002; Mutz 2009). But a group employing DEL also has the ca-
andMartin2001).Pre-decisioncommunicationamong pacitytosetitsownagenda:Itcandecidewhatissues
decision-makers,inwaysthatviolatetheindependence to take up, and what options to choose among. It can
of their individual choices may be taken as a source establishrulesgoverning whatissueswillbetakenup
of corruption (List and Pettit 2004). Yet, in the real when, and what options will be available on a given
world, choices are rarely, if ever, truly independent: issue. Strong forms of DEL require equality of de-
Thepresenceofopinionleaderscancompromisefree liberative opportunity and are committed to rules of
speech,therebymakingvotesdependent onalimited neutrality.Assuch,evenwhenitdoesnotrequirecom-
numberofschoolsofthought(Ladha1992).Moreover, plete consensus, the conventional forms of DEL lack
jurorsinacourtroomvoteforguiltorinnocenceonthe a mechanism for closure (when have enough reasons
basisoftheevidencetheyhaveallheard,ratherthanon been offered to holdouts?), and thus cannot offer a
theirindependentknowledgeofthestateoftheworld. practical way forward in time-sensitive decision con-
Iftheevidenceismisleading,orifjurorsareincapable textswithoutviolatingitsownpremises.
of assessing relevant evidence, the classic Condorcet In real-world democracies, citizens must be able to
result,inwhichcertaintyofcorrectnessisapproached settheirownagendas(ordemocraticallyestablishrules
asthesizeofthejuryincreases,willbeweakened(Di- fordoingso)andmustmakedecisionsundertimecon-
etrich and List 2004). These considerations point to straints. Middle-way theory potentially answers those
thelimitationsofIGA,butdonotobviateitsvaluein demands, by relaxing certain assumptions on which
enabling groups to make better-than-random choices DEL and IGA are predicated. REA may be thought
among options. Real-world IGA does not offer infal- ofasavariantofDELsincesomeversionsofDELdo
libility,but democratic theoristsoftenvalue fallibility, incorporate voting (Fishkin 2009; Warren and Pearse
especiallywhentherecognitionofhumanfallibilityis 2008).OritmayberegardedasavariantofIGA,since
conjoined with experimentalism (Schwartzberg 2008; someCondorcetjurymodelsdoallowcommunication
KnightandJohnston2011). amongvoters(GerardiandYariv2007),andallowfor
In some recent empirical work on decision-making division of issues into parts (Bovens and Rabinowicz
(Sunstein2007),DELandIGAaresetupasincompat- 2006). The innovative feature of REA is establishing
iblealternatives,tothedetrimentofDEL.Bycontrast, the functional relationship between multiple relevant
JohnDryzekandChristianListarguepersuasivelythat criteria and the best choice among options through
deliberation, rather than being incompatible with ag-
gregative approaches to democratic decision making,
actually offers a solution to the challenge posed by 11Alternatively,inextremecases,anexistentialthreat(seesection6:
social choice theorists (e.g. Riker 1982), who claim REAIII,foranexample)mayserveasanexogenously-setagenda,
insofarasthegroupwillceasetoexistifitdoesnotaddresstheissue.
thatthedemocraticaggregationofviews,interests,or
Thissecondsortofexogenousagenda-settingiscommontoeachof
preference across individuals is “bedeviled by impos- thedecision-makingapproachesconsideredhere.
5
psr1200062 xxx(xxx) January4,2013 22:3
Democracy’sWisdom January2013
TABLE1. Threeapproachestoepistemic-democraticdecision-making
Deliberation IndependentGuess RelevantExpertise
(DEL) Aggregation(IGA) Aggregation(REA)
Issuechoice Exogenousor Exogenous Exogenousorendogenous
endogenous
Options Exogenousor Exogenous Exogenousorendogenous
endogenous
Independence No Yes No
Updating Yes No Yes
Choiceamong Byreasonsoffered Byindividualsassessing Byexpertise,orreputations&argumentsof
options bydeliberators issues domain-experts
Decisionby Consensusorvote Independentvotes Relevance-weightedvotes
Notes: Endogenous = issues or options are determined by the decision-making group. Exogenous = issues and options are
determinedbyexternalagency.
equal relevant-domain-weighted votes based on the and an infallible expert in each domain ranks the op-
reputationsandtestimonyofdomain-experts.Thisar- tionsaccordingtothecriteriaofjudgmentappropriate
ticle argues that, by shifting the focus from judging to that domain. REA II also assumes pre-established
complex issues “in the round” to judging experts in rules, but introduces mass voting by domain, on op-
circumscribeddomains,REAenablesacollectivityto tions,advisedbymultipledomain-experts,asasolution
make superior common-interest choices, under time- to the problem of the fallibility of individual experts.
constraint and without elite capture. Table 1 summa- REAIIIassumesthatissues,options,andrelevantdo-
rizes the relevant features of REA compared to con- mains are set in a deliberative/voting stage that pre-
ventionalIGAandDEL. cedesandisadvisorytothefinalvote.Table2setsout
Undertherightconditions,amiddle-wayepistemic theprimaryfeaturesofeachversion.
approach to democracy can answer the challenges
of sustaining transitivity and collective agency, while
avoidingelitecontrol.12Yetitstillfacesproceduralhur- ARISTOTLEPOLITICS3.11: THE WISDOM
OF THE MANY
dles: Determining what issues ought to be addressed
andwhen;definingtheoptionsamongwhichachoice InacelebratedpassageofthePolitics(3.11=1281a42-
must be made; choosing well among those options
b10), recently and skillfullyanalyzed by James Linley
under time constraint. The distinguishing features of
Wilson(2011)asanexampleof“deliberativeintegra-
REA,whichenableittoclearthosehurdles,arethat(1)
tion,”Aristotleclaimsthat,undertherightconditions,
issuesareparsedintoamanageablenumberofdomain-
“the many” judge certain matters better than any ex-
parts;(2)domainsofexpertiseareweightedaccording cellentindividualorsmallgroup.13Thissectionfillsout
totheirrelevancetotheissue;(3)expertsineachdo-
Aristotle’s bare-bones account of the wisdom of the
main rank options according to the criteria relevant
manyinordertoshowthatthepreconditionsrequired
tothatdomain.Relevantdomainsmaybeestablished
for Aristotle’s approach to epistemic democracy are
by rules or may be determined by a combination of
not so demanding as to relegate the wisdom of the
deliberationandvoting.Ineithercase,thenumberof
many to the realm of ideal theory alone. Aristotle’s
partsintowhichtheissueisparsedis(ideally)anopti-
account of collective judgment suggests how diverse
mumthatbalancestwoaspectsoftransparency:Includ-
expertisemightbeaggregatedbyagroupofdemocratic
ingmoredomainsallowsdomainstobedefinedmore
decision-makersconfrontedwithavarietyofpossible
narrowly,producinggreaterclarityabouttheexpertise
answers.
relevanttoeachdomain.Yetrestrictingthenumberof
Therelevantpassageislaidoutschematically,below,
domainsclarifiestherelativerelevanceofeachdomain
withclarifyingnotesinbracketsandkeyGreektermsin
totheissuebeingdecided.
parentheses.Thesubdivisionofthepassageintoeight
Thediscussionthatfollowsthenextsectionconsid-
sectionsismyownandisthebasisofsubsequentcita-
ers three versions of REA. In REA I issues, options, tionsofthispassage(flagged§):
anddomainsarepre-establishedbyinstitutionalrules
12List2011showsthatdemocracyconfrontsatrilemmainthatthere
is no decision procedure simultaneously satisfying “robustness to 13Waldron1995isaninfluentialdiscussionofthispassageinterms
pluralism,”“basicmajoritarianism,”and“collectiverationality.”In ofknowledgeaggregation.SeealsoNewman1887IIIadloc.(with
REA,thetrilemmaisavoidedbyrelaxing“robustnesstopluralism” citation of relevant comparanda from Aristotle and other ancient
byassumingthatthereisadegreeofpre-existingattitudecohesion writers);Keyt1991;Aristotle,Robinson,andKeyt1995;Ober1998:
(pernote3)andthatdeliberationinthecourseofthedecisionpro- 319–26;Kraut2002:402–409;Gottlieb2009:200–207.Lanen.d.and
cessproducesfurtheralignmentofattitudes(movingtowardssingle- Cammack2013arguethatthepassageconcernsaggregatedvirtue
peakedness).REAIandIIaresubjecttoeliteagendacontrol,but alone,andhasnothingtodowithdeliberationordiversity;butsee
REAIIIisnot(section6). discusionthatfollows,withnote16.
6
psr1200062 xxx(xxx) January4,2013 22:3
AmericanPoliticalScienceReview Vol.00,No.0
TABLE2. ThreeversionsofRelevantExpertiseAggregation
REAI.Completerules REAII.Completerules REAIII.Incompleterules
infallibleexperts massvoting massvotinginstages
Issuechoice Setbyrules Setbyrules Setexogenously,orby
deliberation/vote
Options Setbyrules Setbyrules Setbydeliberation/vote
Domains&relevance Setbyrules Setbyrules Setbydeliberation/vote
Expertsperdomain One Multiple Multiple
Choiceamong Byexperts,basedon Bycitizens,basedon Bycitizens,basedon
options individualexpertiseina reputationofexperts, reputationofexperts,
givendomain reasonsgivenbyexperts reasonsgivenbyexperts
Decisionisa Domain-experts’ranking Massvotes,aggregatedby Massvoteonoptions,
functionof ofeachoption,intheir relevantdomains,on advisedbypriorvotes,by
owndomainsof options,advisedby domainsonoptions,ofa
expertise domain-experts counciladvisedinturnby
domain-experts
1. The many (hoi polloi), of whom none is individu- Aristotle’saccountishighlycompressed.Ifweareto
allyanexcellent(spoudaios)man,neverthelesscan, understandhow,intheconcludingphrase(§8),“some
whenjoinedtogether,bebetterthanthose[theex- judge a particular part of the issue, while all of them
cellentfew], together judge all of the parts taken as a whole” we
2. Not [better] as individuals but all together (hoˆs mustdosomeunpacking,inlightofinformationfamil-
sumpantas), iar to Aristotle’s original readers (Wilson 2011: 263–
3. just as potluck meals (sumphoreˆta deipna) can be 67). Aristotle’s point (§1) is that under the right con-
betterthanthoseprovidedatoneman’sexpense.14 ditions, a group of ordinary people judge better (i.e.
4. For,therebeingmany,eachpersonpossessesacon- choose more reality-tracking options with better ex-
stituentpart(morion)ofvirtue(areteˆ)andpractical pected outcomes) than a few excellent persons. The
wisdom(phroneˆsis), groupachievesitscorrectnessofjudgmentasagroup
5. and when they have come together, the multitude (§2). Aristotle offers two homey examples (potluck
(pleˆthos) is like a single person (hoˆsper hena an- meal,judgingmusic/poetry)ofacollectivityachieving
throˆpon), yet many-footed and many-handed and a superior outcome. I assume that Aristotle expected
possessingmanysense-capacities(aistheˆseis), eachexampletobefamiliartohisoriginalreaders,and
6. soit[themultitude]islikewise[likeasingleperson thatthispresumedfamiliarityaccountsforsomeofthe
with multiple capacities] as regards to its facets of compressionofthepassage.
character(taeˆtheˆ)anditsintellect(dianoia). The first example (§3) is an analogy: the “potluck
7. Thisiswhythemany(hoipolloi)judgebetter(kri- meal,”towhichprospectivedinersbringdifferentcon-
nousin ameinon) in regard to musical works and tributions.Thepotluckmealis,potentially,anexcellent
thoseofthepoets, whole.Wemustpresume,basedonAristotle’scoreas-
8. for some [judge] a particular part [of the issue] (ti sumptionsaboutjusticeasajoint-and-severalcommon
morion), while all of them [together judge] all [of good, that the result of a successful potluck meal is a
thepartstakenasawhole](pantadepantes). better experience for each contributor than would be
the case if a “simple” meal (3.1286a29–30) were pro-
Politics3.1281a42-b10.Trans.C.Lord,adapted.15 videdatoneman’sexpense.Theseveralcontributions
are the parts that constitute this potentially excellent
whole.Assuch,inordertoachievethemeal’spotential,
14Thispassageisreprisedinsimilarlanguageat3.1286a24–31:“The eachofthepartsmustbeoftherightsort.Thepotluck,
polis is made up of many persons, just as a feast to which many asawhole,willgowrongiftheparts,thevariouscontri-
contribute(hestiastissumphoreˆtos)isfinerthanasingleandsimple
butions,arenotatoncediverseandgood.Yethowwill
one,andonthisaccountamob(ochlos)judgesmanymattersbetter
diversityandgoodnessofcontributionsbeassured?If
thananysingleperson.”
15ThepassageshouldbereadintheframeworkestablishedbyAris- we assume independence of contribution choice (i.e.
totle,NicomacheanEthicsbook6:adetaileddiscussionoftherela- nodinerknowsexantewhatanotherwillbring),then
tionshipamongsensecapacity,virtue,practicalwisdom,character, each diner may just happen to bring the same thing -
intellect, and experience – and their bearing on deliberation and
if we have six diners, we may end up with six courses
decision.Waldron(1995),Kraut(2002:402–409),andWilson(2011)
ofpastasalad.Regardlessofthegoodnessofthepasta
emphasizethedeliberativecharacterofPolitics3.11.Gottlieb(2009:
200–207)arguespersuasivelythatAristotleisneitherbeingironicnor salad, this will not be an excellent meal and certainly
presentingsomeoneelse’sargument.Sheshowsthattheoptimistic no better than one provided by an individual. More-
accountofdemocraticdecision-makinginthisandrelatedpassages over,potluckdinnersaresusceptibletofree-riding:A
iscompatiblewiththediscussionoftheunityoftheseveralvirtuesin
free-riding diner might choose strategically to bring
NicomacheanEthicsandEudemianEthics,becauseunlikedisunited
vices,thevirtues(likecollectivejudgments)cohere. somethingcheapandpoor,anticipatingthatotherswill
7
psr1200062 xxx(xxx) January4,2013 22:3
Democracy’sWisdom January2013
bringbetterfare.Ifeachdinerfearsbeingstuckwitha introducestheissueofparts(contributions)andwholes
sucker’spayoffbyfree-riders,therewillbearacetothe (the meal). The analogy suggests, moreover, that the
bottomandthecommonrepastwillbecorrespondingly optimalnumberofpartsfallswithinarange:withtoo
poor. few dishes the meal will be excessively simple. Yet at
If the meal is to be excellent, as Aristotle specifies somepointitwillnotbeimproved,indeedmaybemade
itcanbe,therightconditions–theoperatingassump- worse, by the addition of further dishes. The analogy
tions of the contributors to the dinner – must include alsoshowswhythemembersofthegroupmustshare
somecommonknowledgeandsomerulesintheform somerelevantformsofknowledge,andwhytheymust
of social norms. First:even assuming that they do not sharesocialnormsaboutquality.
deliberateaboutspecificcontributionsinadvance,the Aristotle’s second example (§7) concerns judgment
multiplecontributorsmusthavegoodreasontoexpect of a musical/poetic production: He states as a matter
particular people to bring a particular sort of food or of fact that hoi polloi do judge better (than the few)
drink.Thatistosay,theymustsharecommonknowl- in regard to “musical works and those of the poets.”
edge of one another so as to be able to predict what Aristotlemayhavevariousmusical/poeticproductions
eachislikelytobringtothetable.Next,theremustbe inmindhere.Butitseemscertainthatamongthepro-
a norm that ensures that each brings something good ductionsAristotleandhisoriginalreadershadinmind
to the dinner. These are not excessively demanding wastheperformanceofdrama,andespeciallytragedy.
conditions. Anecdotally: I have gone on picnics with Tragictheaterwas,inAristotle’sviewandthatofmod-
thesamegroupoffriendsformanyyears.Weseldom ern scholars, the definitive “musical/poetic” venue in
prearrangewhowastobringwhat,butwehaveagood classicalAthens,whereAristotlelivedformostofhis
ideaofwhatsortofthingeachislikelytobring.More- workinglife(Wilson2000).ItisasafeguessthatAris-
over, there is a norm of contributing good things. Per totle’s original readers were reasonably familiar with
Aristotle’s specification, it is arguably a better expe- the rules governing the judgment of dramatic perfor-
rience – gastronomically as well as socially – than a mances.
dinnerprovidedatoneindividual’sexpense.Iassume In classical Athens, prior to a Dionysian festival,
thatthisoutcomeisnotuncommonamonglong-lived threetragicpoets(andthreecomicpoets)werechosen
associations with shared norms and stable member- by a state official to present works for performance.
ships.Aristotle’soriginalreaderswerecertainlyfamil- Each tragic poet presented a group of three plays.
iar with the practices of voluntary associations, which After the performances, the three play-groups were
wereprevalentinclassicalGreece(Ismard2010);some judged; on the basis of that judgment the poets were
will have had experience with potluck meals; others awarded first, second, and third prizes. The judging
wouldknowofthemthroughreading.16 was by a panel chosen by lottery from among the
Aristotle’s potluck analogy gets us some way to- citizens of Athens. In practice, the mass audience of
wards understanding what Aristotle was after in the severalthousandspectators,byitsresponsetotheper-
“wisdom of the many” passage. The potluck analogy formances,gavethejudgestheircue-asappealsforthe
audience’sgoodwillinAristophanes’extantcomedies
clearly demonstrate. It is, in sum, a likely hypothesis
16AlongwiththestandardGreeklexicon(Liddell,Scott,Jones1968:
thatAristotlehadthemassjudgmentofatheaterau-
s.v.sumphoreˆtos:“broughttogether,“collected”)andmostotherin-
terpretersofthetwo(Politics3.1281b2–3,1286a29–30)relevantpas- dience in mind when he stated that hoi polloi judge
sages(recently:Wilson2011:263–64),Itakethemeal(sumphoreˆton betterthananyoneindividualinregardtomusicaland
deipnon/sumphoreˆtos hestiasis) to be a potluck. In the alternative poeticworks.17
proposed by Cammack 2013 and Lane n.d., Aristotle refers to a Aristotle specifies (§4) that each member of the
meal provided at the financial expense of many, through a public
decision-makinggrouppossessesaconstituentpartof
process,andthesuperiorityofthemealisafeatureoftheaggre-
gatedvirtueofthefinancialcontributors.Thisseemsunlikely.The virtueandpracticalwisdom.Thesepartscanbeeffec-
rare term sumphoreˆtos is unattested in Greek public documents. tively aggregated: Aristotle states (§5–6) that a mass
PaceCammack2013theconcept“mealsharedbythemembersof can, under the right conditions, be like a single per-
anassociation,towhichseveralmemberscontributefoodstuffs,”was
son – yet unlike a single person it possesses a multi-
familiartoAristole’sreaders.ItcomesupinHesiod,WorksandDays
(722:daitosekkoinou),beforethemonetizationofGreekeconomies tude of sense-capacities, and likewise the facets of its
madecommonfinancialcontributionsfeasible.Xenophon(Consti- character and its intellect (two components of practi-
tutionoftheLacedaemonians5.3)referstoSpartancommonmeals cal wisdom): are multiplied. Aristotle seems to mean
(sussitia)featuring“manyadditionalcontributions”(pollaparaloga)
thatthegrouppossessesamongitsmembershipmulti-
providedbyhuntingamongothersources.Xenophon(Cyropaedia
pleanddiverseconjunctionsofsensibility,virtue,and
7.1.12)alsoanalogizesabattletoacommonmeal(eranos)towhich
“wemayeachcontributemanygoodthingsforourmutualbenefit” practicalwisdom,andthattheseconjunctions arerel-
(polla kagatha alleˆlois eisenegkein). In Aristophanes’ Acharnians evant to collective judgment. Individuals manifesting
(1085–1142,withscholiast)theprotagonistpacksabasketofspe-
cialfoodstobringtoadinner(deipnon)atthehouseofapriest.
Thealternativeinterpretationleavesunexplainedwhytheinferior,
provided-by-one-manmealisdescribed(1286a29–30,seenote14)as 17The audience’s “vote” was in the form of what Schwartzberg
hapleˆ:“simpleasopposedtocompoundormixed”(Liddell,Scott, (2010)callsacclamation.Atthemajorfestivals,thetragicprizewas
Jones1968.s.v.haploos/haplousIIIa;citingAristotleMetaph.989b17, for a set of three tragedies; the comic prize for a single comedy.
Sens.447a18).ThecentralityforAristotleofthecomplexwhole(e.g. Athenian audiences as judges of drama: Csapo and Slater 1994;
thefamilyorpolis)madeupofdiverseparts,thepriorityofthewhole Wallace 1997; Marshall and van Willigenburg 2005. Aristotle
toitsparts,andthepoliticalimplicationsofthathierarchy:Politics (Rhetoric 3.2.1403b31–35) also suggests ways the “vote” could go
1.1253a20,1260b14–15withOber1998:295–96. wrongifthemanyhavemisplacedpriorities.
8
psr1200062 xxx(xxx) January4,2013 22:3
AmericanPoliticalScienceReview Vol.00,No.0
these multiple conjunctions are, therefore, explicable with how subsidiary parts (mereˆ, moria: terms used
as parts (§4) that, when properly aggregated (come interchangeablybyAristotle)constituteawhole.With
together),producearightsingle-person-likedecision. referencetothepotluckanalogy,wecanthinkofeach
Since the example (§7) is judgment of musical/poetic of the six parts of tragedy as analogous to dishes that
productions,andsince(soIhaveargued),Aristotlehad makeupafinemeal,althoughtheanalogyisinexactin
tragedyinmind,someofthemultipleconjunctionsmay thatitisjudgmentofquality,notqualityinitself,thatis
be understood as forms of expertise that bear on the Aristotle’sconcernin§7and§8ofthePoliticspassage.
judgment of the multiple parts that make up a good In order for a tragedy group to be judged “best,”
tragedy(§8). each of the six constituent parts must be taken into
Aristotle’spointhereseemstobethatanindividual accountandproperlyweightedforrelevance;ignoring
may indeed have the sensibilities, virtue, and practi- orimproperlyweightinganyrelevantpartwillresultin
cal wisdom to be a very good at judgment of a given aflawedoveralljudgmentandthusinabadoutcome:
partoftragedy.Buthewilllacksomeconjunctionsof theprizeswillbeawardedtothewrongpoets.Aristotle
sensibility, virtue, and practical wisdom – some forms specifiesthatthesixpartsarerankedinthefollowing
of expertise – relevant to judging the whole. A group orderofimportance:plot,character,intellect,diction,
withtherightsortofdiversitywillpossessesallforms song, spectacle (Poetics 1450a-b). Thus, plot is most
ofexpertisenecessarytomakingagooddecision.The relevanttoqualityandmustbemostheavilyweighted;
groupwilljudgewellifitisabletobringtherelevant spectacleisleastrelevantandisweightedleastheavily,
forms of expertise to bear on the constituent parts of with the others in between. In the model that follows
the tragedies being judged, while giving each form of (Table3)therelevance-weightingofthesixparts(Pw)
expertisetherightweight.18 Iftheaccountistoavoid is8,6,5,4,3,2;thusthemostrelevantpart(plot=8)
metaphysical mysteriousness, the aggregation process is assumed to be four times as important as the least
must be accomplished without losing sight of the fact relevant(spectacle=2).Wedonotknowhowclosethis
that, while it is in some ways like a single person, the istoAristotle’sownweighting,andofcoursechanging
groupactuallyconsistsofmultipleindividuals.Theag- the weights produces different outcomes. The follow-
gregationisexplainedunmysteriouslyifwesuppose(as ingsimple(nointeractionterms)andstylizedmodelis
thecontextofthepassageclearlyimplies)thatAristo- meantonlytoshowhowthetragedyrankingproblem
tle was referring to a decision-making institution. If is solved, when we assign specific weights to the six
we assume that Aristotle was in fact referring to the parts, and (extrapolating from Aristotle) make some
judgment of tragedy in §7, we can fill in his account assumptionsaboutrulesandexperts.
of the “wisdom of the many” by reference to actual Inthisstylizedepistocraticsystem,theassumedfea-
institutional rules, domains of expertise, and a hierar- turesofthetragedychoiceareasfollows:(1)thenum-
chy of relevance. Analyzing how a group of the right ber of options (three tragedy-groups) and the actual
sort,actingasaquasi-personthroughproceduralrules, options (T1, T2, T3) are set by rules; (2) the six rele-
could correctly rank tragedies, puts some flesh on the vantparts(perAristotle’sPoetics)aresetbyrules;(3)
bonesofAristotle’saccountofthewisdomofthemany. each part is a domain for which there is an infallible
expert, who judges correctly by ranking the three op-
tionsaccurately,intermsofthecriteriarelevanttohis
REAI:COMPLETERULESWITHINFALLIBLE
domain;20(4)weightingoftherelevantparts(i.e.Aris-
EXPERTS
totle’shierarchy,nowassignedparticularweights)isset
Athenian institutions stipulated the rules for ranking by rules. Having watched the performances, each ex-
(first, second, third prize) three tragic poets based on pertrankstheoptionsbasedonhisdomain-expertise.
theperformanceofthreetragediesbyeachpoet.Inthe For example, the Plot expert, impressed by T1’s nar-
Poetics (1450a6–14), Aristotle provides an account of rative development, ranks T1 highest, yielding an un-
thesixpartsoftragedythat(forourpurposes)canbe weightedrankof3,andaweightedscoreof24.After
regarded as additional rules: “Necessarily then every eachexperthasrankedtheoptions,theaggregatescore
tragedyhassixconstituentparts(mereˆ),andonthese for each option is tallied. The final ranking (and thus
its quality depends. These are plot, character (eˆtheˆ), the outcome: distribution of prizes) is determined by
diction,intellect(dianoia),spectacle,andsong...This the aggregate scores. The procedure ranks the three
list is exhaustive, and practically all the poets employ
these elements.” In listing the parts that make up the
whole that is a proper tragedy, Aristotle specifies the (dianoia)—foritisthesewhichdeterminethequalityofanaction;
relevantdomainsofexpertise.ThePoeticspassagere- indeedintellectandcharacterarethenaturalcausesofanyactionand
calls the Politics passage. Both concern character and itisinvirtueofthesethatallmensucceedorfail—itfollowsthenthat
itistheplotwhichrepresentstheaction.By‘plot’Imeanherethe
intellect as natural causes and co-determinants of the
arrangementoftheincidents:‘character’isthatwhichdetermines
quality of an action.19 Moreover, both are concerned
the quality of the agents, and ‘intellect’ appears wherever in the
dialoguetheyputforwardanargumentordeliveranopinion.”
20Inavariantepistocraticsystem,thesingleinfallibleexpertineach
18ThepassagesfromAristotle’sMetaphysics,PartsofAnimals,and domainisreplacedbyaclassicCondorcetjuryofindependentand
Rhetoric,citedinnote8,areespeciallyrelevantinthiscontext. competent(above0.5likelytobecorrect)experts.Iftheirlevelof
19Cf.theprecedingpassage(Poetics1449b36–1450a6):“Andsince competencyishighenough,andthereareenoughvotes,thechoice
tragedyrepresentsactionandisactedbylivingpersons,whomust of option will approach certainty. This parallels the “distributed
ofnecessityhavecertainqualitiesofcharacter(eˆtheˆ)andintellect premise-basedprocedure”ofListandPettit2011,ch.4.
9
psr1200062 xxx(xxx) January4,2013 22:3
Democracy’sWisdom January2013
TABLE3. HypotheticalAristotelianrankingofthreeoptions
REAI REAII
Pw T1u T1w T2u T2w T3u T3w T1u T1w T2u T2w T3u T3w
Plot 8 3 24 1 8 2 16 425 3400 50 400 25 200
Character 6 1 6 2 12 3 18 50 300 200 1200 250 1500
Intellect 5 2 10 3 15 1 5 150 750 300 1500 50 250
Diction 4 2 8 3 12 1 4 180 720 300 1200 20 80
Song 3 3 9 1 3 2 6 375 1125 25 75 100 300
Spectacle 2 1 2 3 6 2 4 10 20 450 900 40 80
Aggregate 12 59 13 56 11 53 1190 6315 1325 5275 485 2410
Notes:Optionsaregroupsoftragedies(T1,T2,T3).Pw=weightedrelevanceofagivenpart(domainofexpertise).Pwismultiplied
bytheexpertranking(REAI:3=highest),ormassvote(REAII),toachievetheweightedscore.T(n)u=unweightedrankorvote
foragivenpartforeachoption.T(n)w=weightedscoreofagivenpart.AggregateT(n)w(thesumofweightedscoresofallparts)
determinesthecorrect(Aristotelian)rankingamongoptions:T1>T2>T3.
optionstransitively,inthecorrect(Aristotelian)order: expertopinions,andtherebydecidesonabestcourse
T1(scoreof59)isawardedfirstprize;T2(56)issecond; oftreatment.22
T3(53)third.21 Neithertheoperacriticnortheattendingphysician
REAImodelsthedecision-makingprocessesofan islikelytoassignnumericalvaluestotheexpertopin-
individual faced with making choices on an issue that ionsthatcontributetothefinaldecision,butthebasic
isdivisibleintodomainsofexpertise.Suppose,forex- featuresofREAremain:acomplexmattersisbroken
ample, that an experienced opera critic must write a down into parts, known experts judge each part, op-
review that includes a rating (say, one to four stars) tions are delimited by rules, weighting is by relative
and a ranking (better or worse than other operas this relevanceamongparts.Ineachcasethedecisionmust
season). Between acts of the performance she seeks betimely(thecritichasadeadline,thetreatmentmust
the opinion of audience-members expert in the vari- begin before the patient dies). The opera critic and
ousaspectsoftheperformanceshewillhighlightinher attendingphysicianexamplesshowhowtheinstitution-
review(singing,staging,actingandsoon).Herexperi- alizedrulesofREAIenableseveralexpertstobe“like
enceenableshertoidentifyexpertsandtoweighttothe asingle(rationalandexperienced)person”incoming
partsasshedecidesonratingandranking.Orsuppose toadecision.Itis,howeveranon-democraticmethodin
thatanattendingphysicianmustdecideamongseveral thattheagendaisnotsetbythegroup,rankingineach
treatmentoptionsforapatient.Thephysician’sexperi- domain is by a single dictator-like expert, and the ex-
encewiththissortofcaseleadshertorecognizethatthe perts’votesareunequal(duetoweighting).Moreover,
case has several distinct aspects, and that each aspect this epistocratic “cloistered expert” approach (Ober
demands a certain expertise. The physician consults 2008:1–3)isill-suitedtoaddressingcomplexsituations
medicalexperts,eachofwhommakesarecommenda- thatlieoutsidethegroup’spriorexperience.
tion based on his or her specific domain of expertise
(e.g.surgery,diet,psychology).Undertherightcondi-
REA II: COMPLETE RULES WITH MASS
tions,thephysicianaggregatesappropriatelyweighted
VOTING
InREAItheprovisionofexpertsinmultipledomains
21TheapproachdescribedhereisaBordacount,inthateachvoter isessentialtoachievingthesuperioroutcome,butthere
rankseachoption.Thereareotherpossiblesystemsforemploying isnomassvote.Therulesdeterminetherangeof op-
expertiseinrelevantdomainstojudgequality.Forexample,inan tions; after the domain-experts have ranked the op-
unweightedsystem,eachexpertmightmakeabinarychoice(0,1),in- tions,allthatisneededisforacompetentauthorityto
dicatingthatthetragedy-groupdidordidnotmeetsomeestablished
applytheweightingformulaandaddupthescores.Yet
standard. All tragedy-groups receiving six votes (i.e. meeting the
Aristotle’s“wisdomofthemany”passageassumesthat
standardineachdomain)might,forexample,receiveaprize.Thus,
therecouldbeone,two,orthreeprizes–ornoneatall.Alternatively, it is a multitude, acting like a single person (§5), that
inarelevance-weightedsystemeachexpertmighthaveaveto(hecan judgesthewhole;thesuperiorjudgmentonmusicaland
vote0,or1–3).Ifanyexpertdeterminesthatatragedy-groupfailsto
meetagivenstandardinhisdomain(voting0),itisexcludedfrom
further consideration; unvetoed groups are ranked. Both systems 22Notethatineachcasetheindividualisanexperiencedexpert,but
alloweachexperttoexertconsiderablepowerovertheoutcome,and isnotageneralexpert:i.e.shehasenoughexpertisetodetermine
couldresultinnoprizebeingoffered(eachgroupbeingvetoedby relevantdomainsandtojudgeamongexperts,butherexpertiseis
oneormorejudges).These(amongother)relevance-andexpertise- insufficienttoallowherconfidentlytojudgetheissueonherown.
sensitiveapproachestojudgmentaggregationmaybeappropriate ThisisjustthesituationIattributetoAristotle’swisemany,acting
forotherdecisioncontexts,butnoneseemstofittheAristotelian- asarationalquasi-individual.Iowetheoperacriticandattending
AthenianapproachtojudgmentoftragedysowellasthesystemI physician examples to discussions with David Large (who reports
havedescribedhere.Mythankstoananonymousreaderforthese that he has witnessed opera critics doing just this) and Adrienne
importantpoints. Mayor,respectively.
10