1. society

Pluralism Revisited
Robert A. Dahl
The struggle of individuals and groups to gain autonomy in relation to the
controlof othersis, like the effortsto acquirecontrolover others, a fundamental
tendency of political life. Struggles for autonomy result from conflicts and
cleavages; when these strugglesare successful, as they often are, they resultin
turn in tendencies toward pluralism. Because conflicts and cleavages are
ubiquitous, so too are tendencies towardpluralism.
By suppressingautonomy and preventingthe public manifestationof conflicts and cleavages, a hegemonic regime can prevent the developmentof a
pluralisticsocial andpoliticalorder.Wheneverthe barriersto organizedoppositions arelowered, however, the thrusttowardautonomyandpluralismbecomes
evident in political and social life. In polyarchies, where by definition these
barriersare lowest, comparativelyspeaking, subsystemautonomyand organizationalpluralismare always markedfeaturesof the social and political order.
However, even if a considerabledegree of pluralismis a necessarycondition,
an essential characteristic, and a consequence of a democratic regime,
pluralismalso creates problemsfor which no altogethersatisfactorysolution
seems yet to have been found.
You will have noticedthatI have alreadyused the word "pluralism" several
times, andin severalsenses. "Pluralism''is often used simplyas a synonymfor
"diversity." Thus "culturalpluralism" seems to mean essentially the same
thing as "cultural" or "social diversity," except that "pluralism" may be
intendedto carrya more favorable, "diversity" a more neutral,connotation.I
see no profit in this usage, and prefer to use the term "diversity" instead.
I want to use the term here in two different but related ways. First, by
'conflictive pluralism''I intendto referto the numberandpatternof relatively
191
Comparative Politics
January 1978
enduringcleavages that must be taken into account in order to characterize
conflicts among a given collection of persons. I wish to distinguishconflictive
pluralismfrom strict bipolaritywhich, I shall argue, is a comparativelyrare
cleavage pattern,if we considerany numberof public, politicalconflicts within
the variouscountriesof the world, or at any rateamongthose wherethe barriers
to the public expression of conflict are relatively low. Second, by "organizationalpluralism"I intendto referto the numberandautonomyof organizations
thatmustbe takeninto accountin orderto characterizeconflicts among a given
collection of persons. Organizationalpluralismis greater, other things being
equal, the greaterthe numberof organizationsand the greatertheir autonomy.
Systems that permit a significant measureof autonomyto importantunits or
subsystemsare, in fact, frequentlycalled pluralistic,or at leastpluralisticin that
respect. ThuscontemporaryYugoslav writerssometimesdescribetheirsystem
of decentralizedsocialism as pluralistic,in contrastto strictlyhegemonic rule
with a commandeconomy, as in the Soviet Union, or Yugoslavia itself before
1951. In Italy, and until the military coup also in Chile, Communist party
leaders spoke of their commitmentto pluralismin orderto signify their intention of maintaininga regime that would permit opposition parties. Thus, the
termis no longer limitedto Westernbourgeoisthought,nor is it necessarilyan
epithet among Marxistsor other socialists.
I will now very briefly, and necessarily thereforesomewhat superficially,
examine the causes of organizationalpluralism, some of the problems or
pathologies it gives rise to, and some of the remediesthat are often proposed.
AlthoughI can hardlydo more thanhint at the main lines of the argument,my
considerations will lead me to the conclusion that none of the commonly
proposed solutions would accomplish much. What is more, with some the
remedy seems to me ratherworse than the disease.
Causes of Organizational Pluralism
The extent of organizationalpluralism within a political system, and more
concretely within a country, is mainly to be explained, I think, by: (1) the
amount of latent conflictive pluralism;(2) the nature of the socioeconomic
order;(3) the natureof the political regime; (4) the concrete structureof the
political institutions. These four factors are not wholly independentof one
another;their relationshipsare complex and by no means well understood.
Conflictive pluralism As Joseph LaPalombara'srecent analysis of the evidence indicates, and as othershave also found,1 in most countriesthereexist a
numberof different lines of cleavage, and the intersectionof these cleavage
lines has produceda patternof conflictive pluralism,not bipolarity.Bipolarity
along a cleavage line formed by social class tends to exist only in highly
homogeneouscountrieslike New Zealandor Finlandwhere otherdifferences,
192
Robert A. Dahl
as of language, religion, race, or ethnic group, are not sufficiently salient to
disturbthe effects of differences in social class. Ironicallyfor orthodoxclass
theory, however, it is precisely because of their high homogeneity that such
countries manageto deal rathereasily with conflicts arisingfrom class cleavages. In these countriesthe patternthat tends to emerge, then, is not extreme
polarizationaccompaniedby acute antagonismsbut rathera moderatebipolarity within a ratherconsensual political environment.
It is possible to interpretcleavages otherthanthose formedby social class as
crumblingobstaclesto class consciousness, left behindby the slow recessionof
traditionalsociety and early capitalism, destined neverthelessto rapiderosion
by the push of new economic relationships. Historically, however, this interpretationhas led to a persistentunderestimationof the continuingstrengthof
identificationsformedby subculturescenteredaroundreligion, region, ethnic
group, race, and language, and failure to foresee the emergence of new
identificationscenteredarounda varietyof economicdifferencesthatdo not fall
nicely along a single prominentcleavage line but rathergenerate several or
many cleavages, as among skilled and unskilledworkers,organizedand unorganized, blue collar and white collar, service workers, professionals, executives, and so on. Finally, in reducingideology to an epiphenomenonof class (a
position, it should be added, that many neo-Marxistshave long since abandoned), orthodoxclass interpretationshave tendedvastly to underestimatethe
extent to which ideological diversityamongelites leads to fragmentationrather
than solidarity. Nowhere is this last tendency more visible, incidentally, than
among the intellectualpartisansof the working classes.
To accountsatisfactorilyfor the powerfulthrusttowardconflictive pluralism
exhibited in practically all countries in the present world, and certainly in
countriesin the latersequencesof economic change, wouldcall for a deeperand
more extensive explanationthanthe accountI have given here. It would start,I
think, with the conjecturethat concrete humanexperiences provide a narrow
base for creating strong identifications and attachmentsthat extend much
outside the small, specific, and idiosyncraticcluster of human beings with
whom each of us is most intimatelyassociatedduringthe importantoccasions
of our lives.
Whateverthe explanationmay be, the importantpoint is not that "class" is
unimportantbut ratherthat, however it may be measured, in practically all
countrieswhere oppositions are relatively free to organize and express themselves, "class" in its variousmanifestationsis only an element, albeit nearly
always a significantone, in a fragmentedpatternof cleavages andconflicts that
is persistentlypluralisticand not bipolar, much less consensual.
So far I have emphasized the prevalence of conflictive pluralism. But it
would be wrong to suppose that the amount of latent conflict available for
expression when the barriersto oppositions are lowered is approximatelythe
same in every country. Whatis known from studiesof particularcountriesand
193
Comparative Politics
January 1978
from such cross-nationaldataas now exist stronglysupportsthe hypothesisthat
thereexist significantvariationsin the amountof conflictive pluralismamong
countrieswith similarregimes, particularlyamong polyarchies,and withinthe
same countryover long periods.
The socioeconomic order It is reasonableto ask whether a high degree of
organizationalpluralismis, at least in moderntimes, so exclusively a productof
capitalismthatit would necessarilydisappearin an economic orderthatboth in
theoryandpracticetook for grantedthe fact thatgiantfirmsarenecessarilyboth
public enterprisesand political systems. In particular,would a high degree of
organizationalpluralism necessarily vanish in an order where the principal
meansof productionwere socially ratherthanprivatelyowned-in, then, what
would ordinarilybe called a socialist economic order?
Although such a view is widely held, it seems to me unambiguouslyfalse,
and rests upon a theoretical confusion that makes ownership equivalent to
control. Advocatesof capitalism,like theirsocialistcritics, have often assumed
thatprivateownershipis both a necessaryand a sufficientconditionfor control
of the enterpriseby the owners; conversely, ownershipby the governmentof
the state is thoughtto be botha necessaryand a sufficientconditionfor control
by the government of the state. If an enterprise is privately owned, it is
assumed, then of course the owners make the key decisions, either directlyor
throughmanagerswho are no more than agents. If enterprisesare owned by
society, the government,the people, or the workers,then it mustfollow thatthe
decisions of enterpriseswill be made by society, the government,the people,
the workers.
This is an egregious error, simple-mindedin its concepts and tragic in its
consequences. For experience in this century has conclusively demonstrated
that ownershipis definitely not a sufficient condition for control. One cannot
even be certainthat a particularform of control requiresa particularform of
ownership.Publiclyor socially ownedenterprisesrangeall the way fromhighly
hierarchicalsystems of managerialdominance,whereeven tradeunions areof
negligible importance-as in the Soviet Union-to the system of selfmanagementor workers' control practicedin Yugoslavia.
If we accept the axiom thatin general a specific form of ownershipis not a
sufficient condition for a specific control relationship (and may not be a
necessarycondition), then the question of control is theoreticallypriorto the
question of ownership. (Unfortunately,this has rarely been appreciatedin
controversiesover capitalism vs. socialism.) Seen in this perspective, what
capitalismdid in theory, and in substantialmeasurein practice,was to inauguratea system of decentralizedcontrolover economic organizationsthatwere to
a relativelyhigh degree autonomousvis-a-vis the centralgovernmentand one
another.If socialism by definitionentails social ownershipof economic enterprises, andunless by definitionit mustbe centralized,then a socialisteconomy
194
Robert A. Dahl
can be highly decentralizedand thereforepluralistic. A socialist government
might well elect to grantextensive autonomyto enterprisesin orderto permit
internal controls far more democratic than have ever existed either under
capitalismor in centralizedsocialist systems like the U.S.S.R. Obviously no
socialist government-probably no government-would eliminate all external
controls, whetherby markets,the governmentof the state, or both. A decentralizedsocialist ordermight nonethelessgeneratejust as much organizational
pluralismas exists in any nonsocialistorder, and perhapsa good deal more.
Nor is organizationalpluralismin a socialist ordernecessarily at odds with
Marxism.On this as on so manyotherquestionsthe corpusof Marx's work is,
taken as a whole, ambiguous. For half a century, Marxistswho looked to the
Soviet Union as the very embodiment of Marxist verities assumed that a
socialist order must necessarily operate as a centralizedcommandeconomy.
Yet there are passages in Marx that lend eloquent support to the idea that
socialism would be highly decentralized.
Again, the experienceof Yugoslaviaafterturningaway from Stalinismin the
1950s lends supportto a perspectiveon Marxismthatcannotbe lightly dismissed as simply hereticalor cryptocapitalistapologetics. Unless one is preparedto
argue that by definition the Yugoslav economy is capitalist (or anyway nonsocialist) andthatby definitionYugoslav Marxistsare not real Marxists,one is
compelledto concludethatMarxismas interpretedby some Marxistsis compatible with a high degree of organizationalpluralism, and socialism with a
comparativelyhigh degree of organizationalautonomy.
To understandthe really crucial alternatives,for both the political and the
economic order, one must focus on control, not ownership. In sorting out
economic alternatives,the key question is not whetheran orderis socialist or
nonsocialist (though that may be an importantsecondaryquestion) but how
much autonomyis permittedto economic enterprisesand the natureof internal
and externalcontrols. A nonsocialist, privately-ownedeconomy can be dominated by a hegemonic political order that closely regulates the activities of
economic enterprises, as in Nazi Germanyduring war time. Conversely the
experience of Yugoslavia, even if it is unique up to now, demonstratesthat a
socialist economic order can be highly decentralized and organizationally
pluralistic.
A shift from capitalismto socialism, then, need not necessarily reduce the
amountof organizationalpluralismin a country.It is altogetherpossible thatin
some countries(the United States mightbe one) where importantdecisions are
strongly influenced by giant corporationshierarchicallydominated by their
managers,the inaugurationof a decentralizedsocialisteconomy-in which, for
example, the decisions of an economic enterprisewere made accordingto the
principles of full proceduraldemocracy, where all persons employed by the
firm (and only those) enjoyed rights of citizenship in the governmentof the
195
ComparativePolitics January 1978
firm-would result in an increase, not a decrease, in the number and autonomy
of economic organizations.
To sum up:
The amountof organizationalpluralismin a countrydoes not dependon whether
the economic order is capitalist, in the sense that the enterprisesare privately
ratherthan publicly or socially owned. It does depend on the extent to which
decisions are decentralized, that is, on the amount of autonomy permittedto
enterprises.And the amountof autonomypermittedto enterprisesappearsto be
theoreticallyindependentof forms of ownership, hence of capitalism and socialism as such. A capitalistordermay be, but need not be, highly decentralized.
A socialist order may be, but need not be, highly centralized.
Regime Suppose there were a country with a remarkable degree of diversity
among its people on a very considerable number of characteristics: language,
religion, ideology, region, ethnic group, national identification, race, etc. To
prevent the manifestation of these cleavages in political life would doubtless
require a regime in which a small set of unified rulers is in a position to mobilize
an overwhelming preponderance of political resources for their own use and for
the maintenance of a severely hierarchical bureaucracy, and to deny to
everyone else practically all access to political resources. Given a highly
hegemonic regime of this kind, no public conflict would appear and the
underlying predisposition toward conflictive pluralism would remain unmanifest.
Suppose now that the barriers to oppositions were gradually reduced. It
would be reasonable to expect that as the barriers went down, relatively
autonomous organizations would spring up, and that some of these organizations would seek to advance the claims of the hitherto politically latent groups
and subcultures. Up to some point, the more the barriers to organization and
participation were reduced, the greater would be the proliferation of relatively
autonomous organization. In time, no doubt, a limit would be reached and more
stable patterns would emerge.
Fortunately, historical experience provides us with a nice laboratory for
exploring these conjectures. For the fact is that something very much like this
did occur in Italy, Austria, Germany, and Japan after the displacement of
hegemonic regimes in those countries as World War II came to a close.
De-Stalinization in Yugoslavia after 1950 also led to a great multiplicity of
interest groups. To take another example, a rich organizational life began to
proliferate in Czechoslovakia during the famous Prague Spring of 1968. Finally, Portugal offers a current, if still unfinished, experiment. And the reverse
development often occurs when leaders establish hegemonic control over the
government, destroy all autonomous organizations, prevent the manifestation
196
RobertA. Dahl
of publicconflict, and build a hierarchicalstructureof orderand cohesion over
the remains of the silenced opposition.
It is importantto keep in mind that whether the movement is toward
organizationalpluralismor away from it, the process need not come to rest
either in a highly hegemonic regime at the one extreme or in a comparatively
open polyarchy at the other. It may simply stop somewhere along the way.
Consequently a regime may be rather "pluralistic" without being highly
"democratic.'"Not only do polyarchiesvary in the amountof conflictive and
organizationalpluralism they exhibit but, as Juan Linz has shown, so do
hegemonies .2
Yet it would be an egregious mistaketo supposethatthe extent of organizational pluralism has nothing to do with the nature of the regime. As our
illustrationsare intendedto show, in the same countryin a period of time too
shortfor significantchanges to occur in the latentpatternof social cleavages,
changesin regimehave resultedin enormouschangesin the amountof manifest
organizationalpluralism.Indeed, one of the most strikingdifferences among
regimesin the modernworldis to be foundpreciselyin theextentto which those
who oppose the conduct of the government of the state are permitted to
organize, expressthemselves, andparticipatein political life. Let me continue
to use the term"polyarchy"to referto a regimein which the rightto participate
is broadly extended and the institutionalguaranteesto oppositions are, by
historicalstandards,comparativelystrongand the-barriersto oppositionscomparatively low. Let me continue to use the term "hegemonic" for regimes
where the institutional guarantees are weak or absent and the barriersto
oppositions high.3
The propositionI now wish to advance is that:
Organizationalpluralismis ordinarilya concomitant,bothas cause andeffect, of
the liberalizationand democratizationof hegemonic regimes.
In particular,the comparativelyhigh level of institutionalguaranteescombined with the broad inclusiveness that (by definition) together characterize
polyarchy are inevitably associated with extensive organizationalpluralism.
The guaranteesof the rightto form andjoin organizations,freedomof expression, the right to vote, the right of political leaders to compete publicly for
support, particularly in elections; the existence of alternative sources of
information-all these are importantconditions for the growth of organizations, particularlypoliticalorganizations,for they both heightenthe incentives
for forming political organizationsand reduce the costs of doing so. Hence:
Given essentiallythe same latentpatternof cleavages in a country,if its regime is
polyarchalit will exhibitmoreconflictive andorganizationalpluralismthanif it is
hegemonic.
197
Comparative Politics
January 1978
Concrete political institutions Althoughthe concretepolitical institutionsof
a countryare of course partlydeterminedby the natureof the regime and the
extent of conflictive pluralism,they can also makean independentcontribution
to the numberandautonomyof organizationsin a country.These consequences
are perhapsmost visible in polyarchies, among which there are markedvariations in political institutions.Threekinds of variationsseem to be particularly
significantfor their effects on organizationalpluralism.First, multipartysystems of course increase the number,if not the autonomy, of political parties.
Second, in some polyarchies, such as Switzerland and the United States,
constitutionalnormsand political practicesprovide for an extensive partitioning of governmentalauthorityby means of both federalismand separationof
powers;the resultis to increasethe numberand autonomyof governmentaland
other political organizations.In other polyarchies, such as New Zealand and
Britain, there is much less constitutionalpartitioning;a unitaryratherthan a
federal system, combined with parliamentarygovernmentratherthan a strict
separationof powers betweenexecutive and legislative, makefor considerably
greaterconcentrationof governmentalauthorityand correspondinglyless organizationalpluralismamonggovernmentaland otherpolitical organizations.
Finally, the numberor the relative autonomy of organizationsis also increased by the institutionsand practices of "consociational democracy," as
practicedin the Netherlands(even if the degree of conflictive pluralismthat
originally stimulated the growth of these institutions has waned), and by
systems of "corporate pluralism" or "democratic corporatism" that have
developed, and may be increasing in strength, in Norway, Sweden, and
elsewhere.
Because each of these threesourcesof variationin concretepolitical institutions seems to be capable of varying independentlyof the others over a wide
range, and becausethe concreteinstitutionsof a particularcountryalso change
fromtime to time, even amongcountrieswith essentially similarregimes, such
as polyarchies, differences in concrete political institutionshelp to create a
staggeringvarietyin the specific patternsof organizationalpluralismexisting in
differentcountries.
Problems
Any specific constellationof organizationalpluralismis, then, a highly complex phenomenon.The particularconstellationsvary in importantways from
one country to another. A good deal of evidence appears to support the
hypothesis that:
In different countries, the constellation of organizationalpluralism varies in
amount, in inclusiveness, and in the patternof cleavages and conflicts in which
organizationsparticipate.Countrieswith similar regimes, e.g., polyarchies or
hegemonies, may have significantly different constellations of organizational
198
Robert A. Dahl
pluralism. Moreover, although some degree of organizationalpluralism is a
necessary condition for polyarchy and a fairly high degree of organizational
pluralismappearsto be a consequenceof the institutionalguaranteesand exclusiveness of polyarchy, the particularconstellation of organizationalpluralism
presentin a given countrywith a polyarchalregime is not necessarilyessentialto
maintainpolyarchy, and may be undesirableon other grounds.
Inequality In particular,it is prettywidely held that in its effects on decisionmakinginstitutions,organizationalpluralismin polyarchiesoften fails to meet
reasonablecriteriafor equality and, partly but not wholly as a result, for a
broader"public" or general interest. It is hardlyopen to doubtthat organizational pluralismand the institutionalguaranteesof polyarchyare not sufficient
conditions for a high degree of equality in the distributionof control over the
governmentof the state or other organizations,or the distributionof political
resources, or, more broadly, status, income, wealth, and other key values.
Stasis This conclusion would be less consequential, no doubt, if it could be
shown that organizationalpluralism is a dynamic force with a more or less
steady thrust toward the reduction of inequalities. I am not aware of any
theoreticalreasoningthathas been advancedon behalfof the existence of such a
dynamicforce. Over againstthis possibility, therearegroundsfor thinkingthat
despite markedinequalities in control over the governmentof the state in a
polyarchy, organizationalpluralismmay sometimes develop a self-sustaining
patternover fairly lengthy periods. A stable system can develop in which the
most disadvantagedare unorganizedor poorly organized and thereforecomparatively powerless to remedy their condition; in which, more broadly, a
stable patternof inequalityin the distributionof majorvalues is preserved,at
least in part, by the resultsof organizationalpluralism;in which majorpublic
problems go unsolved because every solution that does not have substantial
agreementamong all the organizedforces is, in effect, vetoed; in which public
policies in every sector are prettyclearly not determinedby considerationof
what mightserve the best interestsof the greaternumberbut resultinsteadfrom
the play of organizedgroups,each concernedexclusively with its own interests;
in which, finally, some such perceptionas this of the way the system operates
becomes widely diffusedandleads to disillusionanddiscontentnot merelywith
the particularsystem of organizedpluralismthat creates stasis and its consequences but with the constitutionalstructureof the polyarchy, and more, to
polyarchy in general or even to the desirabilityand feasibility of democracy
itself. So:
A particularconstellationof organizationalpluralismcan producea stablesystem
in which mutualvetoes preventthe reductionof inequalitiesand, moregenerally,
structuralchanges in the status quo.
199
ComparativePolitics January 1978
For most polyarchies, no doubt, my portraitis overdrawn. But there is
enough resemblanceto conditionsin many, includingmy own, to cause one to
wonder how a stable equilibriumof inequalitiesmight be overcome.
Solutions and Their Difficulties
Quite possibly thereis no generalsolution. That is, desirablesolutions may be
specific to a particularsituation or country. However, some of the most
commonly offered solutions run into serious difficulties of their own.
Substance: Common interests versus particular interests One common
genre of solutions looks to the discovery of a broad, overarching public,
national, collective or general interest (or good) that, if it were properly
understood,would clearly specify the properdecision to be made in cases of
conflicting andparticularisticclaims. The properdecision mightspecify which
claim is to be given priority, or an outcome transcendingall the particular
claims. If everyonewere to know the generalgood, andgive priorityto it, then
presumablythe autonomousorganizationswould no longer oppose changes
aimed at the common good. Inequalitiesthatremainedwould be the resultless
of the relativestrengthsof the variousorganizationsinvolved in a conflict than
of some broaderprincipleof justice or generalgood deliberatelyadoptedby all
the membersof the polity, and concurredon by the variousorganizedforces.
In this view, a greatdeal of conflict comes aboutbecause of disagreements
thatare, at base, irrational.These conflicts arise because the social consciousness of some or all of the political actors is, so to speak, deformed. Irrationalities occur because some people simply misunderstandthe nature,
causes, and consequencesof mattersin dispute;or they neglect their long run
interestsin favor of their immediateinterests;or they unwittinglysacrifice the
interestsof theirbroader,more social selves to the claims of a narrowerself
identifiedwith the primaryindividualego, or the family, the kin group,or some
other narrowsocial fragment.
If all thepartiesto a conflict alwaysunderstoodtheirtrueor realinterests,it is
argued, many conflicts would vanish. At the very least, betterunderstanding
would facilitatea searchfor and the discovery of mutuallybeneficial solutions
to conflicts. Fromthis perspective,then, properlyconductedinquirywill lead
to betterknowledge of the true natureof self and society, and this knowledge
will in turn help to foster harmonyratherthan conflict.
Unhappily,this proposalis less a solution than it is a prescriptionto search
for a solution. All assertionsas to the specific natureof a general, collective,
public, or national interest are, unless they are merely vacuous, themselves
likely to become mattersof publiccontroversy.I don't meanto say thatthe idea
of the generalgood is meaninglessbutonly thatit inherentlyinvolves debatable
philosophicalassumptions,togetherwith judgmentsof fact andvalue that will
200
Robert A. Dahl
ordinarilytransforminto a controversialpolitical question any claim publicly
advancedon behalfof a specific policy said to be in the public interestor for the
generalgood. I have yet to runacross a propositionaboutthe generalgood that
is concrete enough to specify outcomes and not highly controversial.
More specifically, argumentsattemptingto demonstrateto membersof an
advantagedmajoritythattheirbest interestsrequirethem to make sacrifices in
favor of a relativelydisadvantagedminority,convincing as they seem to their
advocates, are not necessarily convincing to the majority. Nor do advocates
fare betterwho seek to demonstratethat a relativelyprivileged minorityought
in its own interests to yield some of its benefits in favor of a relatively
disadvantagedgroup, whether another minority or a majority. Anyone who
puts forthargumentsof this kindis eitherignoredor soon embroiledin political
controversy.What seem to be crystalline, self-evident truthsto advocates are
falsehoods to their opponents. In a concrete sense, then, advocates and opponents of particularchanges are partisans. If they organize to further their
respective demands, they simply foster organizationalpluralism.
Thus what sets out to be a way of ending conflicts among organizedgroups
may only deepen an existing cleavage or create a new one. In the absence of
general agreementon the substantivecontent of a transcendentpublic good,
each group strives to impose its policies on the rest.
This is why I statedearlierthat solutionsof the kindI have been considering
seem reduciblemerelyto recommendationsthatin all cases of conflict a search
ought to be made for mutuallybeneficial and mutuallyacceptableoutcomes.
These proposalsdon't and evidently can't guaranteethat mutuallybeneficial
and acceptable outcomes will actually be discovered. To search for such
outcomes seems to me desirable,butunless we aretold whatpoliticalprocesses
and structureswill increase the likelihood of discovering and bringing about
such outcomes, if andwhenthey exist, an exhortationto searchfor the common
good does not take one very far.
Structures: Centralization versus decentralization Anotherkind of solution fills the void of the firstby specifyingchangesin politicalstructuresthat, it
is thought, would lead to the reductionof inequalities. Typically, structural
solutions require either greater centralizationand hierarchic controls over
subsystems or greaterdecentralizationand democratizationof subsystems.
Let me briefly clarify what I mean by centralizationand decentralization.If
organizationsare subsystemsin some largersystem of controls, then thatmore
inclusive system is decentralizedto the extent that its subsystemsare autonomous. By definition,A is autonomousin relationto B, with respectto some set
of actions x, to the extent that B does not control A with respect to x. (To
simplify, I am going to omit the phrase "with respect to x." However,
referenceto a scope,x, is always implied.) Also by definition, if A's autonomy
increases in relation to B, then B 's control over A decreases.
Sometimes one subsystem controls all the rest, but not necessarily com201
Comparative Politics
January 1978
pletely. Let me call a special subsystem of this kind the "Center." Then
startingfrom any given situation, to decentralize means to increase the autonomyof othersubsystemsin relationto the Center.By definition, decentralization also means a decrease in the Center's control of the subsystems. To
centralize means exactly the reverse.
Now it is sometimes arguedthat because organizationalpluralismleads to
unacceptableinequalities and prevents the adoption of policies designed to
reduceinequalities,the solutionis to reducethe autonomyof organizationsand
to increase the control of a Center. In relation to private organizationsthe
Center is ordinarilythe governmentof the state. In relation to governmental
organizations,the Centeris usuallythe leadershipof the governmentalbureaucratichierarchies.Advocatesof this solutiontypicallyproposegreatercentralization of control in the government and a correspondingreduction in the
autonomyof business firms and certainother organizations.Within the government,they advocategreatercentralizationof control in the chief executive
and a correspondingreductionin the autonomyof governmentbureaucracies.
Some advocates of the centralizingstrategyrecommendthe abandonmentof
polyarchy itself, arguing that polyarchy is inherently too pluralistic; they
recommendthe adoptioninsteadof a highly centralizedhegemonic regimeand
a commandeconomy. Only so, they argue, is it possible to reduceunacceptable
inequalities in both the opportunitiesand rewardsthat are generatedby any
modern industrialor "post industrial"society.
Whether the solution of centralizationis intended to preserve and even
strengthenpolyarchy, or to transformpolyarchy into a hegemonic regime, it
faces one grave difficulty. As we have seen, centralizationmeans that the
autonomy of certain subsystems is reduced and the control of the Center is
increased. To increase the control of the Center requires that the political
resourcesof the Centerbe increasedrelativeto the resourcesof the subsystems.
If the Center is the central leadershipin the governmentof the state, as this
solution usually prescribes,these political resourceswill ordinarilyinclude an
increasein the access to the means of coercion availableto the centralleaders.
The more the subsystemsmustbe deprivedof theirautonomyin orderto bring
about the reallocations necessary if greater equality is to be achieved, the
greatermust be the resourcesof control, including coercion, available to the
leaders at the Center. At the limit, subsystemswill have no autonomyand no
resources for resisting the control of the central leaders. To say thatothersubsystemshave little autonomyand little controlover the
Centeris to say that the political system is based on a high degree of political
inequality. Since reciprocalcontrols by others are weak or absent, the incentives of the centralleadersto reduceinequalitiesaregeneratedonly by theirown
consciences and ideology. Historically, over the long run these have been
weak, particularlyas new generationsof leaders succeed to top positions in a
centralized system. The leaders at the Center are free to redirect privileges
202
Robert A. Dahl
toward themselves, or more generally to their own political stratum,ruling
class, or elite group. Given the weakeningclaims of conscience and ideology,
they generally do. They may further intensify the system of privilege by
recruitingtheirsuccessorsmainlyfromtheirown stratum,including,of course,
their own children.
Thus a strat egy of centralizationthatis initiallyjustifiedas a way of reducing
inequalitiesrunsa seriousriskof establishinga systemthatnot only is basedon
a high degreeof political inequalitybut in due time facilitatesthe development
of socially and economically privileged ruling elite as well.
Faced by this prospect,or perhapseven the reality, one may recommendan
alternativestrategyof decentralization,together with the democratizationof
relatively autonomoussubsystems, whethergovernmental,partisan-political,
economic, or other. Muchas I preferthis approachto the other, I am compelled
to confess that it also has difficulties. Except under highly unusual circumstances,the resourcesof the varioussubsystemsare likely to vary. Differences in resourcesarise from variationsin skills, energies, culturalqualities,
previousaccumulationsand investments,scarcities, markets,readilyavailable
naturalresources, and many other factors. Thus each subsystem might be
internallydemocratic,and the principlesof distributionand allocation within
each subsystemmightbe acceptedas perfectlyjust by all the membersof each
organization, yet if there were no central controls over distributionsand
allocations, social and economic inequalities would surely arise among the
membersof differentorganizations.
To reduceinequalitieseven amonghighly democraticsubsystemsandtherefore among the membersof differentsubsystemsrequirescentralizedcontrols
over allocations, limits of some sort on organizationalautonomy. How much
and what kinds? Are there trade-offs among different forms of inequalitypolitical, economic, social, and other? If so, what principles of distributive
justice should guide decisions and what do these principlesrequire?In short,
which trade-offsare acceptable, which unacceptable?
I doubtwhetheranswerscan be foundthatareappropriatefor manydifferent
situations and many different countries. Wherever people are free to raise
questionssuchas these, the answersarelikely to become, implicitlyat least, the
subject of political controversy.
NOTES
1. JosephLaPalombara,Politics WithinNations (EnglewoodCliffs, 1974), pp. 440 ff. Robert
A. Dahl, ed. Political Oppositionsin WesternDemocracies (New Haven, 1966), pp. 380 ff. Dahl,
ed. Regimes and Oppositions (New Haven, 1973), pp. 14 ff. MorrisJanowitz and John Segal,
"Social CleavageandPartyAffiliation:Germany,GreatBritain,andthe UnitedStates," American
Journal of Sociology, LXXII (May 1967), 601-18. S M. Lipset and Stein Rokkan, eds. Party
Systemsand VoterAlignments(New York, 1967).
2. JuanJ. Linz, "TotalitarianandAuthoritarianRegimes," in FredI. GreensteinandNelson W.
Polsby, eds. Handbookof Political Science, vol. 3, MacropoliticalTheory (Reading, [Mass.],
1975), pp. 175-411, particularlypp. 277 ff.
3. Dahl, Polyarchy, Participationand Opposition (New Haven, 1971), pp. 1-13.
203