Resumé

In a recent paper John D. May has nominated Responsive Rule for a definition of
democracy. Responsive Rule prescribes necessary correspondence between
governmental acts and the wishes with respect to those acts of the persons who
are affected. In this paper the definition is criticized for being unclear on a vital
conceptual issue and for not fully expressing the essence of democracy. On the one
hånd it is maintained that democracy's subject population remains undefined as
the meaning of the word 'affected' is unclear, and three distinctions are brought
forward to illustrate the problem of separating affected persons from persons not
affected. On the other hånd May's identification of democracy with equality in the
weight of wishes is called in question, and it is maintained that wishes should meet
aqualifying criterion. Three procedures for identifying qualified wishes are
discussed, and it is suggested that information must be included in a conception
of democracy in order to separate rational wishes from irrational wishes.

Nominating Responsive Rule

Some years ago the Swedish political scientist Carl-Johan Westholm in his dissertation on John Stuart Mill pointed out the ambiguity of the term 'democracy. Westholm argued that the term was used in so many differing meanings that a statement that the good society is good is no more a tautology than a statement that the good society is democratic (Westholm 1976, 184), and he illustrated his point by listing four different definitions of democracy. The first definition identifies democracy with majority rule. Only form counts in this definition, and a majority decision depriving the minority of its political rights is thus democratic. The second definition identifies democracy with majority rule as well as with guarantees for civil rights, and in this definition both form and content count. The third definition gives still more weight to content, in so far as it adds, for instance, demands for a certain economic policy to demands for civil rights. Finally the fourth definition observes content only and disregards form. Decisions are 'democratic' if they have a certain content; arrangements for decision-making are in themselves irrelevant (Westholm 1976, 180-181).

There are of course varying ways of making this flora of definitions more lucid and manageable. One method is to classify definitions in two main categories, those focusing on how political decisions are made (process definitions), and those focusing on the content of decisions (content definitions). For this second type of definitions political processes and methods then appear as arrangements that might or might not further democracy; the adequacy of these processes and methods in this respect is the object of empirical rather than normative assessments. Olof Ruin has proposed a similar classification, as he suggests that two different aspects of democratic theory are emphasized in the literature on the problems of democracy (Ruin 1974, 171). One concerns form: policies should be made through the active participation of those affected by the policies. The second concerns content: the content of policies should reflect, the demands and the wishes of the majority of the people to whom the policies apply. In this paper we shall discuss a definition of democracy belonging in this second category.

There are two reasons for this endeavour of ours. The first is that we believe that the definition merits attention in so far as it distinguishes the nature of democracy from the nature of democracy's prerequisites. In short, the definition lies to our mind at the hearth of what democracy is about. Secondly, the definition however has some shortcomings, which should be subjected to discussion and criticism in order to improve the clarity and the exploitation of the definition.

The definition of democracy which is here subjected to discussion and criticism was nominated by John D. May in a paper published in Political Studies some years ago (May 1978). In this paper it is the ambition of May to facilitate the definition of democracy 'in terms that are intelligible and palatable alike to behaviouralists and anti-behaviouralists and nonbehaviouralists, platonists and positivists and phenomenologists, avowed democrats and avowed royalists, leftists and centrists and rightists, Westerners and Easterners and Third Worlders and Other Worlders'(l).¹ To this end he proposes for consideration a definition of democracy alluded to as Responsive Rule (hereafter RR):

necessary correspondence between acts of governance and the wishes with
respect to those acts of the persons who are affected (1).

May compares RR with more conventional alternatives and he finds
that RR scores highly on a number of points. These are:

1. Terminological clarity. May argues that the RR definition uses
terms whose sense is immediately clear and that the RR version of
democracy therefore is 'remarkably clear', as 'its terms do not themselves

cry out for a definition' (2). He alleges that equal measures of clarity
cannot be ascribed to definitions alluding without elaboration to 'popular
sovereignty', 'majority rule', 'political equalityV minority rights', etc. (2).

2. Essentiality. RR distinguishes the nature of democracy from the nature of democracy's prerequisites, indicators, by-products and merits. Statements ascribing additional properties to democracy, May argues, 'probably express beliefs about phenomena which logically or empirically, causally or consequentially, invariably or occasionally, coincide with democracy'(3).2

3. Generality. May finds that the RR concept is congenial with the
notion that democracy is an arrangement which in principle can exist in
all sorts and sizes of groups (3).

4. Nominal quantifiability. The RR definition is compatible with the notion that democracy is the name for both an ideal-type or model state and a condition which exists in greater or lesser degree. RR can thus be used for comparing regimes in terms of democracy and for the making of nominal classifications (3 - 4).

5. Phenomenological clarity. May argues that RR is clear in that it expresses definite positions on major conceptual issues which traditionally accompany talk about democracy. According to May this surely is a vital test of intelligibility for a definition of democracy. As to the basic relation between Governance and the Governed, RR makes 'a clear, univocal choice in favour of the "government by" tradition' (4), and May finds this choice commendable, as an univocal concept is easier than an equivocal concept to use in ordinary discourse, as the RR concept is more consistent with conventional usage, and as it is functional for normative disputation (4 - 5).

Recapitulating his discussion, May thus declares that the nominated definition seems noteworthy 'for terminological clarity; for essentiality, or concentrating on the nature of democracy as distinct from its concomitants and merits; for generality, or congeniality with the notion that democracy can be sought in all sorts and sizes of human groups; for nominal quantifiability, or compatibility with the notion that democracy is both an ideal-type state and a matter of degree, and for dealing clearly with conceptual issues that commonly accompany talk about democracy' (13- 14). In the following we shall not try to dispute the merits of RR as far as terminological clarity, generality, and nominal quantifiability are concerned. May's argumentation, which cannot be repeated here for lack of space, is to our mind convincing and coherent. Admittedly, we have some difficulties to perceive the benefit of the property of generality. May takes it to be a non-partisan notion that democracy is an arrangement

which in principle can exist in all sorts and sizes of groups (3). We a.re for our part inclined to reserve the term 'democracy' for talk about politics, and we fail to see the point of applying the concept to 'all sorts and sizes of groups', such as the family or a chess-club or other parapolitical structures (for terminology, see Easton 1965, 50-52). In fact, we believe it confusing rather than clarifying to talk about democracy in all sorts of contexts, which may differ in many vital respects. However, we shall not elaborate this point here. Our interest is with the categories of essentiality and phenomenological clarity.

This means that we shall deal with internal as well as external aspects of the definition. The internal aspects concern logic and clarity, and we shall maintain that the RR concept is fairly unclear concerning one: vital issue, that of democracy's subject population. We shall discuss this issue in the next section of this paper, which accordingly focuses on the property of phenomenological clarity. The external aspects concern premises and basic assumptions, and we shall treat such aspects in the third and concluding section of the paper. The key word for our discussion is responsiveness.

However, we shall not dispute the very foundation of the RR conception, namely the government by the people tradition. This conception catches to our mind the very essence of democracy, and we concur in the statement of Sidney Verba and H. N. Nie that 'responsiveness is what democracy is supposed to be about' (1972, 300). The nature of democracy is of course a highly controversial issue, and one that cannot be scientifically proved to be right or false. This issue is basically ideological. It can be approached from non-ideological angles, and this is how May proceeds. He recognizes the government for the people tradition as Beneficent Rule, implying 'that a group is democratic to the extent that its governmental acts serve the needs - the wants or interests or objective requirements, as distinct from the felt preferences - of its members' (4). And he makes a choice in favour of RR on grounds of methodology and compatibility with conventional thought (usefulness in ordinary discourse, consistency with conventional usage, functionality for normative disputation); he considers the RR concept's 'ecumenical appeal' (14). We are here making the same choice, but for different reasons. Our reasons are predominantly normative and ideological. In fact, we do not really see how May's criteria could decide the matter in favour of RR to the disadvantage of Beneficent Rule. To our mind Beneficent Rule lends itself equally well for ordinary discourse, it is widely accepted and is therefore consistent with conventional usage and has the same ecumenical appeal; it is certainly no less functional for

normative disputation than RR. But Beneficent Rule is not in agreement with the central view of anti-paternalist liberalism that a person should himself be considered the final arbiter of what he wants, and as we are adherents of this view we therefore reject Beneficent Rule to the advantage of RR.

Our normative acceptance of the government by the people tradition does however not imply that we see the RR concept as unproblematic in terms of essentiality. As defined by May, this tradition identifies democracy 'with a constitution ensuring that the desires of the governed, however remote they may be from the needs of the governed, determine the contents of governmental acts' (4). Basically we agree with this definition. But we are inclined to add one condition, which concerns the origin of the desires of the governed. We are therefore obliged to penetrate further in the government by the people tradition and to take a closer look at the concept of responsiveness.

On Democracy's Subject Population

Democracy's subject population has been variously named as 'the people', 'the community', 'the whole population', 'the general will', 'popular consent', 'the many', 'the majority', 'the multitude', 'the hitherto oppressed', 'the poor', and 'the proletariat', May states. He adds that each phrase is 'conspicuously incomplete' (6). And he finds that RR makes a clear decision about the identity of the subjects in a democratic regime (5). RR defines the subject population as all persons who are affected by acts of government; in so doing it employs a functional approach to identifying the population and it does so without qualification. 'It is impartially inhospitable to preferential weighting of preferences according to the age, legal status, place of habitation, rate of political activity, intensity of concern, class status or morals of persons affected by governmental acts' (6 - 7). May argues that this way of defining democracy's subject population meets the need for clear choice on a vital conceptual issue, and is consistent with important elements of conventional thought (7). However, two objections can be raised against this functional conception of democracy's subject population, and they concern the logical and practical consequences of RR. The one objection is recognized and refuted by May; the other is not.

(1) As RR defines the subject population as all persons affected by acts of governance, it obviously assumes that the population can vary in composition with the issue at stake. A foreign policy issue affects not only the people in the actual country which makes a decision concerning

this issue but obviously also people in other countries. On the other hand a law regulating, say, accommodation allowance in a country affects only a part of the inhabitants of that country. The subject populations are clearly different. And on some issues the population must obviously include in its ranks 'infants, lunatics, felons and foreigners' (8). If we now assume, as we commonly do, that in defining democracy's subject population one is simultaneously identifying the persons who must exercise positive controlling rights of power (freedom to express opinions, voting power, etc.), then RR appears troublesome to say the least. In some cases controlling powers must be given to 'infants, lunatics, felons and foreigners', and in most cases we will have varying sets of persons entitled to the use of controlling powers. In fact, RR could for obvious practical reasons not possibly be used as a device for the making of binding political decisions.

May answers to this objection by emphasizing the need to separate the task of identifying democracy's subject population from the task of determining what rights and powers need to be assigned to what persons. 'We are not immediately obliged', May tells us, 'to identify democracy with a system in which all of the affected persons, and only those, are entitled to vote or are endowed with equal shares of political control' (9). The subject population is one thing, the controlling population is another. Identifying the first group challenges us to say what persons' wishes need to be taken into account; identifying the second group challenges us to discern what constitution, what power structure, what distribution of rights and powers among what persons expedites compliance (9). The two tasks are not identical, neither are the populations necessarily identical. It is not so that affected infants, lunatics, felons and foreigners must exercise controlling rights of power; it is rather so that controlling rights of power must be exercised in a way which ensures that the wishesishes of affected infants, lunatics, felons and foreigners are taken into account. Democracy, to repeat, is not the same as democracy's prerequisites. RR does not deal with the rights and powers of persons; it deals with the power or weights of policy preferences (10). To our mind, May by providing this answer shows that the objection is false. It is not an implication of RR that the controlling population must vary in composition with issues. The implication is that one is resorted to find a power structure which promotes the prescribed correspondence between governmental acts and popular desire. And this task, although difficult, is not impossible in terms of logic or practice.

(2) Another ambiguity however remains, and on this difficulty May
offers no comments. We refer to the phrase 'who are affected', and we

ask for a clarification of the word 'affected'. May provides no answer and
there is little in his paper indicating that he is aware of this problem.

A first difficulty is brought to the fore by the familiar distinction between intentions and outcomes. A certain governmental act may be intended by the authorities to affect a certain segment X of a population, or the authorities may at least be aware of the possibility that the act affects X. The outcomes of this act may however a) affect this segment X and leave other segments unaffected; b) affect X as well as segments Z and Y; c) leave segment X unaffected but affect segments Z and Y instead, etc., etc. These alternatives obviously offer differing solutions to the problem of identifying a proper subject population and a choice has to be made between the solutions. In order to specify the problem we may refer to the systematization given in Table 1 below, showing four different subject populations. The first case (A) and the last case (D) are ideal cases and remain unproblematic as they express correspondence between intentions and outcomes. The cases in between (B and C) are however problematic. The former covers a population which is not affected although it was the intention of the authorities to affect this very population, whereas the latter covers a population which is affected although it never was the intention of the authorities to affect this very population. Are these populations part of democracy's subject population or are they not?

Obviously there are two answers and the logic of RR offers no definite solution. RR assumes an obligation for the authorities to pay attention to popular desire, but is unclear as to the scope of this obligation. On the one hand it could be maintained that the obligation cannot possibly be extended to desires held by quarters who are outside the calculations and the awareness of the authorities; if this view is accepted there follows that a regime is democratic in so far as its arrangements yield correspondence

between its governmental acts and the preferences of the persons who form the intended subject population. It thus follows that B is a subject population whereas C is not. On the other hand it could be maintained that the obligation must cover desires held by all persons affected; if this view is accepted there follows that a regime is democratic in so far as its arrangements yield correspondence between its governmental acts and the preferences of the persons really affected by those acts. It thus follows that C is a subject population whereas B is not. This latter view then brings the claim that a regime must have perfect knowledge of the consequences of governmental acts in order to be truly democratic.³

Another choice relates to a distinction between directly affected and indirectly affected persons. Clearly, if we take only directly affected persons into account we are delimiting a subject population that differs from a subject population which comprises indirectly affected persons also. Let us pick one example. If a great power (A) with a population of say 200 million people invades a smaller state (B) with a population of say 5 million people, then this act is democratic if a) it is supported by the majority of the population of A and opposed by the population of B; and b) we regard only these populations as affected. These conditions are certainly compatible with conventional ways of thinking, which takes us to the rather startling conclusion that this military intervention is a democratic act. On the other hand, if we take into account the long-range effects of this intervention on the international system and the level of international tension, and if we thus are obliged to regard mankind as 'affected', then we might classify this act of brutality in another manner.

Furthermore a choice must be made as to the degree of effect. Let us pick another, less dramatic example. A law in the field of housing policy may prescribe conditions for receiving accommodation allowance. Who are affected by this law? Certainly those who are receiving the benefits. But what about those who have to pay the costs (through taxation) or those who are affected if only in the indirect sense that a law is binding for all citizens? In other words: how much 'affected' should one be to be classified as 'affected'?

The RR definition does not answer this question. The definition only distinguishes between persons 'affected' and persons 'not affected', but this is a distinction that is unclear indeed. When illustrating his discussion of democracy's subject population by reference to university governance, May concludes that 'staff and students are only a portion of the governed, i.e. the persons who are affected by university policies' (10). Other persons and groups are affected as well. 'Governing boards of universitites often adopt policies which affect not only local staff members and

incumbent students, but also foreign staff members, townspeople, alumni, incipient students, purveyors, potential employers of graduates, parents, taxpayers, and other interest groups', May writes (10). But surely students and potential employers are not affected to the same degree by university policies, regulating, for instance, the amount of courses or the length of terms. What May therefore really tells us is that there are degrees of effect; that there are differences between populations as to the degree to which they are affected by acts of governance. This however implies that 'being affected' is something that expresses a continuum rather than a dichotomy, and this means that we in fact may have differences in terms of intensity of effect that exceed the difference between being affected or non-affected. Let us as one example consider a hypothetical situation (reproduced in Table 2) where 5 persons are affected by an act so that one person A is affected strongly, his intensity measure being 4. Corresponding measures are 3, 2 and 1 for persons B, C and D respectively, whereas person E is not affected at all. The difference between D (who is affected) and E (who is not affected) is therefore 1.

However, within the segment of'affected' persons (A - D), picturing a total of six distance relations, we find three relations exceeding the distance between D and E. It does not seem at all clear that the difference between D and E should be given more weight than the difference between, say, A and C or B and D. So there is every reason to repeat the question: How much 'affected' should one be to be classified as 'affected'?

The point we have made in this section is that the RR concept leaves some loose ends in identifying democracy's subject population. Only the three distinctions introduced here result in eight overlapping but certainly not identical sets of'affected' persons, as illustrated in Table 3. The matter of identifying affected persons can be approached in different ways which provide different conclusions, and the RR concept does not make a clear choice between these ways and conclusions. It therefore remains unclear on a vital conceptual point.

On the other hand, definite solutions do not suggest themselves. We shall add only a few general comments. Firstly, it is of course vital that empirical applications of the definition are consistent. If, for instance, we want to compare the democratic performance of a given system at two points of time (A and B), and we hold the socio-economic structure of that system as a valid indicator of 'preferences' (Godwin & Shepard 1976), then we should take care not to use as indicators of governmental acts, say, legislative documents at time A and policy outcomes at time B. These indicators do not necessarily catch same sets of affected persons, as the first is based on intentions whereas the second is based on outcomes. Secondly, when applications are considered, it is to our mind important that cutting points are established in a direction favouring 'directly affected' to 'indirectly affected' and 'more affected' to 'less affected'. If the very point of introducing the component of 'affected' and of thus discriminating between subject populations is to be secured, then the component must be credited with a certain amount of discriminating power. If slightly and indirectly affected persons are to be considered on an equal basis with much and directly affected persons, then the functional approach to identifying democracy's subject population surely is watered down. A final and third remark is however that the benefits of the functional approach are just as easily lost if the search for directly and much affected persons is performed only mechanically and without paying attention to the possibilities opened by the RR concept. For instance, one should not hesitate to count foreigners in democracy's subject population for some types of issues at least (cf Anckar 1981, 17); likewise one should not overlook the possibilities of utilizing RR as an instrument for analyzing regional policy.4

On Responsiveness 5

The RR concept makes a clear choice in favour of the government by the people tradition, and this choice is one without exceptions. The concept demands that the wishes of a majority of affected people are fulfilled, regardless of the content of these wishes. It does not recognize as democratic a situation where the wishes are being overridden, be that for any reason pertaining to the interest or welfare of the people.

For those extending sympathy to some other tradition, as for instance Beneficent Rule, RR therefore appears unacceptable. They certainly are inclined to voice a criticism, the content of which is that RR demands a complete adaptation to preferences which renders it difficult for the elite to react creatively on behalf of the people. Rationality, creativity and responsible government, the argument runs, are therefore alien to RR. According to Amitai Etzioni, a complete adaptation to incoming demands would negate the system's capacity to follow a collective policy and would 'render the system passive and drifting, merely reflecting the fluctuating changes in the preferences of the members and fluctuations in the environment' (Etzioni 1968, 504). In like manner Heinz Eulau and Paul D. Karps expect a person elected from the multitude to be a representative, not merely to be reactive but to take the initiative. Eulau and Karps see the leader in the participatory theory of democracy as largely a reactive agent guided by the collective wisdom of the group. 'He is at best the executor of the group's will, indeed a human facsimile of Pavlov's dog', they write. 'One is in fact back to the instructed-delegate model in which there is no room for discretion in the conduct of the representative' (Eulau & Karps 1978, 69).

However this criticism is to our mind only partly justified. It is based
on a fairly one-sided view of responsiveness.

In fact we think it highly unfair to regard the preference-obeying representative as a reactive agent only, which is exposed to and reacts to stimuli. We believe that power of initiative and innovation, creative ability, and other similar good activities and qualities are compatible with the notion of a responsive representative. This is because we presuppose sets of interactions between the representative (A) and those represented by him (B) and because we believe that mutual influences are expressed in these interactions. Let us imagine a situation where A has received an instruction from B telling him that B prefers a certain policy X. However, A finds that another policy Y would correspond better to the interest of B (as conceived of by A) and he therefore acts to convince B that B should change his preference. If B does not agree to Y, then responsiveness

entails that A yields to the wishes of B. If A acts in favour of Y his conduct is unresponsive and violates RR. But if A is successful in his argumentation B will change his instruction and voice a demand for Y, meaning also that A now becomes responsive to B. A has been active, he has been committed to the cause of the represented, he has argued in favour of his conviction and he has tried to mobilize support for his view. In short, he has been an initiator and he has been responsive. His conduct has been active; still it has been in full agreement with RR. This is of course also valid for a situation when B maintains his preference for X and A submits to this view.

However, the fact that RR is compatible with creative acts from the part of the representative and with an exchange of views between those who govern and those who are governed does not imply that RR demands or prescribes creativity or interaction. From the point of view of RR wishes may be rational and informed or they may be irrational and uninformed, but they must be met. Rational and informed wishes are, however, for evident reasons to be preferred to irrational and uninformed wishes, and are sometimes regarded as a condition for the origin of democracy. In a recent paper Walter Carlsnaes, while sympathetic to the notion of responsiveness, maintains that no democratic process worthy of the name is well served by either constituency or governmental ignorance and irrationality. Carlsnaes finds that three different aspects are involved here, namely 'an enlightened and critically reflective public, a corps of politicians sufficiently well-informed not to be the pawns of experts and professional bureaucrats, and a dynamic area of public debate not beholden to any particular - private or public - interest' (Carlsnaes 1981,88).

Then, can RR be rewritten in a form including such aspects while retaining the notion of responsiveness? Can the choice in favour of the government by the people tradition be replaced by a choice in favour of a government by the enlightened people tradition?

Obviously the answer is yes and no. The answer is yes in so far as; RR can be rearranged in a form prescribing correspondence between governmental acts and qualified wishes of those affected by these acts. The answer is no in so far as such a rearrangement certainly entails difficulties for terminological and phenomenological clarity. These: are drawbacks, to be sure. But it could be argued that such drawbacks must be accepted. In the form suggested by May RR is no doubt a short and definite definition of democracy (1). But even those who agree that responsiveness is what democracy is about may hesitate to concur in May's identification of democracy with equality in the weight of

preferences (14). Carlsnaes does not accept this identification; neither do we. To our mind essentiality demands that wishes meet a qualifying criterion. We accept the view that popular desire should determine the outcome of politics, but we take a negative stand against the view that ignorant wishes should determine the outcome of politics.

One objection would be that if all preferences are equal in weight, then one might envisage forms of responsiveness where the crucial distinction between Responsive Rule and Beneficent Rule is actually rubbed out. Let us imagine two political systems both of which perform in a responsive manner. The one is characterized by member consciousness, by a high degree of interaction between elite and masses, and by occasional differences of opinion between these quarters. Responsiveness here denotes that the elite subordinates to the masses and obeys the instruction from the people. The other is characterized by a lack of real interaction. The government advocates viewpoints that are then echoed by ignorant and nonreflective system members who do not bother much about politics and do not maintain well-considered wishes. Responsiveness here denotes that the elite is tied to its own instructions, echoed by the people. It requires at least some argument to put these systems at equal footing in terms of democracy and responsiveness. RR draws this parallel but provides no argument. For our part we believe that the second system in our example in fact acts in a manner very similar to Beneficent Rule and alien to Responsive Rule. The government governs for the people, not by the people. RR sees the two systems as similar. We do not.

As we have suggested that RR should be rearranged to prescribe correspondence between governmental acts and qualified wishes of those affected by the acts, we now face the task of discussing the meaning of 'qualified. Qualified in what respect? And what criteria could be applied for separating qualified wishes from unqualified wishes?

We shall here consider three possibilities (Table 4). The first relates to the content of wishes and makes a distinction between wise and unwise wishes, meaning that some wishes are to be regarded as wise (qualified) in terms of some standard, whereas others are to be regarded as unwise (unqualified) in terms of the same standard. The two other possibilities relate to the origin of wishes. The first emanates from the concept of manipulation and implies that wishes that are not manipulated are qualified, whereas wishes that are manipulated are unqualified. The second departs from the notion of rationality, depicting rational wishes as qualified and irrational wishes as unqualified. We shall briefly comment on these three alternatives.

Discriminating between wishes on the basis of content appears troublesome to say the least. Obviously a classification can be made in objectively and subjectively wise wishes. In the first case the analyst decides what is wise and what is not, in the second case the decision is made by a political actor. For the sake of simplicity only two categories of actors are considered here, namely those who govern on the one; hand and those who are governed on the other hand. Objective classifications presuppose that the analyst makes use of some standard as for instance maximizing the happiness, the equality, the moral and intellectual calibre, or the self-development opportunities of a designed population (4). This however puts us at distance from RR and in fact equals some variant of Beneficent Rule, where the meaning of 'beneficent' is decided by the analyst. This crossing of the borderline between the government by and the government for traditions, implying also that the analyst is given the right to resort to politically and ideologically coloured choices of standards, is unacceptable.

Subjective classifications do not make the task easier. If we decide on wisdom in terms of elite choice, then we in fact apply Beneficent Rule and the government for tradition. What government sees as appropiate for the needs or interests of the governed is wise, however remote this may be from the wishes of the governed. This is of course not in accordance with a notion of responsiveness. On the other hand, if we define wise wishes in terms of popular desire, then we in fact apply Responsive Rule and the government by tradition. What the people wish is wise and must therefore be met; no qualifications come in question. Efforts to determine wisdom in terms of content are therefore doomed to failure. They result in either Beneficent Rule or Responsive Rule, they do not result in a revision of RR admitting the incorporation of wise wishes in the definition.

Other possibilities to depict qualified wishes concern the origin of
wishes. One could suggest that wishes that have come about in a certain
way are qualified and can therefore not be overridden, whereas wishes

that have not come about in this way are unqualified and can therefore be overridden. The concept of manipulation is of interest here. It is a central thought in the liberal tradition that governments are justified in overriding wishes (wants, preferences) only when these wishes have been manipulated. It could now be argued that this distinction between wishes that are manipulated and wishes that are not manipulated is helpful when it comes to separate qualified wishes from unqualified wishes. One argument would be that manipulation occurs where ignorance is to be found (Goldmann 1981, 127).

But what then is manipulation?

Alan Ware has suggested that manipulation only occurs when some specified person is both morally and causally responsible for a change in some other person's want. Wants determined by institutional bias or the exercise of power cannot be set aside, as this would open the way for an authoritarian approach to government (Ware 1981). In a comment to Ware's suggestion Peter Morriss argues that Ware has not provided an argument to justify this view. Morriss points out that some wants determined by institutional bias may be as harmful to those who have them as those wants created by manipulation as defined by Ware. He therefore finds no reason to accept Ware's distinction: 'Suppose I develop a want for something through manipulation, and you develop a want for the same something through the operation of institutional bias. It requires at least some sort of an argument to justify overriding my want whilst refusing to override yours' (Morriss 1982, 117 - 118). In his reply to Morriss, Ware admits that victims of institutional bias and power are unfree, and may have been as unfree in acquiring a want as the victims of manipulation. However, he argues that the rejection of wants that are harmful to the interests of those who were unfree at their acquisition is potentially dangerous for the liberal state, as the criteria for identifying wants that are the product of power or institutional bias are controversial (Ware 1982, 121).

This dispute clearly illuminates the dilemma which we confront. The RR definition states that democracy requires necessary correspondence between governmental acts and wishes with respect to those acts of the persons who are affected; no conditions relating to the acquisition of wishes are stated. As we find it necessary to demand that the necessary correspondence is between governmental acts and qualified wishes, we have here asked if the concept of manipulation is helpful in providing a criterion for the identification of qualified wishes. We have thus established a condition relating to the acquisition of wishes. This effort is however of limited value. On the one hand we can regard as manipulated

all wishes for which causal responsibility can be located outside the actor. Morriss claims that this means that there are very few wishes which might not justifiably be overridden (Morriss 1982, 119), and he is certainly right. If manipulation is conceived of in this manner, then wishes are, on the whole, manipulated; hence they are unqualified and candidates for being overridden. Once again, Responsive Rule in fact becomes Beneficent Rule. We therefore believe that Ware is right when he rejects this view of manipulation and regards it as dangerous to the liberal state; we therefore also believe that he is right in searching for some more restricted criterion for identifying manipulated wishes. However, we must agree with Morriss that Ware does not provide a convincing criterion, and we are for our part unable to suggest a better solution. To our mind the concept of manipulation offers little help for achieving the task of identifying qualified wishes while retaining responsiveness.

A third possibility to identify qualified wishes is by utilizing the concept of rationality. The distinction would be between rational and irrational wishes, and the rearrangement of RR would be that governmental acts should correspond to rational wishes with respect to those acts of affected persons. Rationality is of course a much debated concept, but we shall not tackle a conceptual analysis here. Suffice it to say that we take rationality to denote two aspects of decision-making: deliberation and information. In this context the former aspect means that wishes should originate from the relating of means to goals, whereas the second aspect means that such deliberations should be based on a sufficient amount of open information. Expressions like 'deliberations' and sufficient information' of course imply difficulties for terminological clarity at least. Nevertheless, this way of approaching the acquisition of wishes offers to our mind the best possibility to lay down a government by the enlightened people tradition.

This is because the criterion separating qualified wishes from unqualified wishes (a) is not fortuitous; (b) does not a priori classify most wishes as unqualified, and (c) corresponds to our disclaiming of ignorant popular desire as the object of responsiveness. This does not mean that the criterion is unproblematic. For it to be utilized in concrete research, unclearnesses need to be solved. One concerns classification.

There is little doubt that wishes that are deliberated and based on information should be classified as rational; likewise there is little doubt that wishes that are not deliberated and not based on information should be classified as irrational. In so far as the category of informed but undeliberated wishes is at all conceivable, it obviously ranks as unqualified. But what about wishes that are deliberated although based on an

insufficient amount of information? To our mind, they cannot qualify as rational. This choice is once again based on our aspiration to avoid Beneficent Rule. Deliberated wishes may stand out abundantly in a society amongst those who are governed. But if these wishes are preferably based on information provided by or supervised by those who govern in the interest of those who govern, then we are in fact again near Beneficent Rule. Responsiveness here means that the authorities submit to wishes the content of which is largely determined by the authorities themselves; this is not the kind of responsiveness we have had in mind.

Our conception of democracy thus prescribes the fulfillment of rational wishes. There are two conditions for democracy: there must be rational wishes and they must be met. When these conditions are crossed in a four-fold table (Table 5), some governmental traditions emerge, which attain different scores on Rationality and Responsiveness. These differences can also be made the basis for comparisons in terms of democracy, and as they are based on a more nuanced conception of democracy they also offer a wider range of alternatives. Paraphrasing one example given by May (4), we can compare three regimes (A, B, C) in rates of achievement (See Table 6). When observing Responsiveness and Beneficence

only6 and giving paramount importance to Responsiveness, regimes A
and B are ascribed the greater measure of democracy. They both score 70

on Responsiveness and 30 on Beneficence, whereas regime C's rates of achievement on each scale are 50. However, if we make use of our distinction between rational and irrational wishes and thus give paramount importance to responsiveness to rational wishes, we come to an opposite conclusion. Regime C now appears more democratic than regimes A and B, and B appears slightly more democratic than A.

This much simplified example is functional also in so far as it demonstrates that various traditions of government may exist side by side within the frame of the same regime. Kjell Goldmann has in a recent paper (1981) called our attention to the fact that there are differences not only between systems but also within systems as to the amount of information. Some political sectors are characterized by a great deal of information, others by a lack of information and public interest. There are in other words differences between sectors as far as prerequisites for democracy are concerned; Goldmann suggests that foreign policy and defence policy are typical areas where information is meagre and democracy undeveloped. Goldmann also emphasizes that one must consider the problem of scanty resources for democracy (1981, 133 - 134). There are, for instance, limits to the amount of information that can be placed at disposal. This means that attempts at increasing democracy in one political sector entail either a decrease in democracy in some other sector or sectors or an overall increase in the amount of information.

---

NOTES

Parts of this paper were presented at guest lectures at the University of Luleå 12.5.1982 and the University of Umeå 13.5.1982. Thanks are due to Sten Berglund, Tom Bryder and Kjell Lundmark for useful suggestions.

1. Figures in brackets refer to pages in May's paper.

2. 'Such beliefs may be exemplified by declarations identifying democracy as or with inorganic fraternity (Proudhon), despotic rule (Bonald), the idea of community life itself (Dewey), a petit-bourgeois counter-revolutionary ideology (Marx), mediocrity (J.S. Mill, Sorel), equality of fortunes and intellects (Tocqueville, Stephen), shared power (Lasswell), the consummation of No-government and Laissez-faire (Carlyle), the absence of a State apparatus (Marx, Bakunin, Lenin), the political system in which society achieves consciousness of itself (Durkheim), the most political and complicated of systems (H. Mayo), institutionalized opposition (Lipset), the good society itself in operation (Lipset), maximal opportunities for self-development (Macpherson), and the worship of jackals by jackasses (Mencken)'. May 1978, 3.

3. The distinction between being subjectively or objectively affected (affected in terms of own perceptions or in terms of real outcomes) of course creates a similar difficulty. My colleague at Åbo Academy, Krister Ståhlberg, has called my attention to this point.

---

4. The possibility of including preferences of still unborned generations in the concept of democracy should also attrået attention. Concerning some issues at least (nuclear wastes) these generations are clearly to be regarded as 'affected'. Wishes of unborned generations do not of course exist in an empirical sense and can therefore not be observed. But estimations can be included in calculations for decision-making.

5. This section includes some materials reported also in Anckar 1980 and Anckar 1981.

6. Regarding 'government against the people' and 'government for the people' as Beneficence, is here of course an oversimplification, used for illustrative purposes only.

REFERENCES

Anckar, D. 1980. 'Responsivitet - om begreppet, företeelsen och forskningsområdet'.
Politiikka 1, 28 - 38.

Anckar, D. 1981. 'The New Social Movements: Promoters of Democracy?' Paper,
International Seminar on 'Societies in Transition', Cuttack, Orissa.

Carlsnaes, W. 1981. 'Foreign Policy and the Democratic Process'. Scandinavian
Political Studies 3, 81-108.

Easton, D. 1965. A Framework for Political Analysis. Englewood Cliffs: Prentice-Hall,
Inc.

Etzioni, A. 1968. The Active Society. New York: The Free Press.

Eulau, H. and Karps, P. D. 1978. 'The Puzzle of Representation: Specifying Components
of Responsiveness', pp. 55-71 in Eulau & Wahlke, J., eds., The Politics of
Representation. London: Sage Publications.

Godwin, R. K. and Shepard, W. B. 1976. 'Political Processes and Public Expenditures.
A Re-Examination Based on Theories of Representative Government'. American
Political Science Review 4, 1127 — 1135.

Goldmann, K. 1981. T utrikes-och försvarsfrågor fungerar inte demokratin'. Statsvetenskaplig
Tidskrift, 3, 123 - 134.

May, J. D. 1978. 'Defining Democracy: A Bid for Coherence and Consensus'. Political
Studies, 1, 1-14.

Morriss, P. 1982. 'Manipulating manipulation'. British Journal of Political Science, 1,
117-119.

Ruin, O. 1974. 'Participatory Democracy and Corporatism: The Case of Sweden'.
Scandinavian Political Studies, Vol. 9 (old series), 171 - 184.

Verba, S. and Nie, H. N. 1972. Participation in America: Political Democracy and Social
Equality. New York: Harper & Row.

Ware, A. 1981. 'The Concept of Manipulation: Its Relation to Democracy and Power'.
British Journal of Political Science 2, 163 - 181.

Ware, A. 1982. 'Manipulation, Liberalism and Paternalism: A Reply to Morriss'. British
Journal of Political Science 2, 119-122.

Westholm, C.-J. 1976. Ratio och universalitet. John Stuart Mill och dagens demokratidebatt.
Kungälv. Skrifter utgivna av Statsvetenskapliga föreningen i Uppsala, Nr. 72.