From Representation and Popular Rule, ed. Ian Shapiro, Susan Stokes and Elizabeth Wood and A.S.Kirshner, C.U.P. forthcoming
Varieties of Public Representation
1. Background and basics
Systems of representative government, I shall assume, are designed to give control over government to the people. Far from being an alternative to democracy, as some have taken them (Manin 1997), they embody an institutional framework — or rather a family of frameworks — for realizing the democratic ideal of giving kratos to the demos, power to the people. The distinction between a participatory and a representative system is not one between democracy proper and some faint approximation but a distinction between rival proposals for the implementation of democracy.
My focus in this paper is on representation in this democratic, popularly enabling sense. Thus the target of the paper is narrower than it might have been. As Hobbes in particular argues, the idea of representation may be used, not just of representatives who are subject to the continuing or periodic control of the people, but also of an hereditary, absolute monarch. The defenders of parliament in the 1640’s tried to give its members a monopoly right on the use of the word (Skinner 2005) but Hobbes argued against them that it was absurd that a monarch who ‘had the sovereignty’ over his subjects ‘from a descent of 600 years’ should not be ‘considered as their representative’ (Hobbes 1994, 19.3). His own view, to the contrary, was that ‘the King himself did…ever represent the person of the people of England’ (Hobbes 1990, 120).
But though my focus is narrower than Hobbes’s, it is broader than the target to which many contemporary theorists give their attention. As will appear, I use the notion of representation in such a way that any public authorities, and any citizens who assume a legitimate role in public discourse, may make a legitimate claim to represent the people (Richardson 2002). Others apply the notion, however, only to elected representatives — elected members of the legislature and, where relevant, of the executive and the judiciary. The concept of electoral representation is more tightly circumscribed than my looser concept of democratic representation but I think that it does not offer the same generality of perspective.1 I hope that this paper may help to vindicate that claim.
There are three factors in any relationship or system of representation, whether in my sense of representation or in any other. First there are the representatives or, in a seventeenth century word, the representers; I prefer this term since the other has an exclusively electoral connotation. Second there are the represented or, as I shall say, the representees. And third, there is the relationship that exists between those two parties: the representation that is exercised by representers on behalf of representees.2
Representers may be individual agents or groups of individuals agents. And in the case where a group serves in this role, the members may each act for their own ends, according to their own judgments, or they may act on a shared intention to further this or that end. In this latter case the individuals will each intend that together they promote the agreed end and they will each do their bit for the promotion of the end, expecting that others will play their parts too (Bratman 1999); they will be a cooperative grouping, not a mere collection of individual agents.
The cooperative grouping that combines around a joint intention comes itself in two forms. The members may act for shared goals, now on this occasion, now on that, without ever forming a joint intention governing their continuation over time. Or they may form the special, shared intention that over time they together should constitute a corporate agent or agency: a body that simulates the performance of a single agent with a single mind. The intention shared in this case will be that they together cooperate in the organized pursuit of agreed ends according to agreed judgments. The ends will usually be an evolving set of ends selected under agreed procedures, and the judgments an evolving body of judgments selected under agreed procedures (Pettit and Schweikard 2006). The distinctions are mapped in the following diagram.
These distinctions are readily illustrated. The member of the legislature who represents a certain constituency or the president who represents the people as a whole exemplifies the individual representer. And the legislature or the executive as a whole exemplifies the group representer. The members of such a group may behave as a collection of independent agents, each with their own brief, or as a cooperative grouping. And if they behave as a cooperative grouping, then they may or may not incorporate in the manner of a body with a coherent, evolving set of goals and judgments.
The US Congress might be seen as an unincorporated, cooperative grouping that does battle, now on this issue, now on that, looking in each case to see if a more or less ad hoc majority can be assembled to support a certain line. There is some concern with securing coherence between the lines supported over time, of course, as well as coherence with the Constitution. But this concern can take second place to the other concerns of members — say, their concern to display their colors back home. This will especially be so if in any case a presidential veto is likely, or if the Supreme Court is expected to reject or reinterpret the legislation.
The Westminster Parliament might be seen, by contrast, as an incorporated, cooperative agency whose members assume a higher degree of responsibility for legislating on coherent lines, at least within the time-frame of a given parliament. The members acquiesce in a procedure whereby a fixed majority will be established and that majority — say, a single, incorporated party — will bring forward a coherent program of legislation to be enacted after parliamentary discussion in the name of the parliament as a whole.
So much for the possibility of variation on the side of the representers in a democratic system. There is a corresponding degree of variation on the side of the representees. The representee may be a single individual, as when the member of a legislature takes up some cause on behalf of a constituent. Or the representee may be a group. The group represented may be cast as a mere collection, as perhaps with the electors in a given district, or they may have a more cooperative aspect. And the cooperative representee may be incorporated or unincorporated. In other words the distinctions on the side of the represented may correspond to those that we map on the side of the representing.
The loose pressure group or the ethnic minority that succeeds in finding a spokesperson in the legislature or elsewhere will often be an unincorporated entity whose members are united around just a single issue. But equally an entity with a corporate or quasi-corporate form of organization may figure as representee. The states that are represented by the Senate of the US or Australia are surely entities of this kind.
The ultimate representee in any democratic system will be the people as a whole: a body that might be taken as a mere collection but is usually depicted as an incorporated entity. It is of the essence of a democratic system that it is supposed to create a state that represents the people as a whole, acting in their name on the international stage and in dealings with individual citizens and groups of citizens (McLean 2004). Thus John Rawls (1999) says that the government in a well-ordered society — ideally, a liberal democratic society — will be ‘the representative and effective agent of a people’ (38) — ‘the political organization of the people’ (26). The disordered society where a small group usurps power, like the unordered society where political organization fails utterly, will be marked by the absence of precisely this system-wide level of representation.
Not only are there variations of these kinds in who are represented and who do the representing: that is, in the relata at either end of the representative relationship. There are also variations in the nature of the relationship. I argue that there are two fundamentally contrasting varieties of representation, indicative and responsive. Indicative representers stand for the representees in the sense of typifying or epitomizing them; how they act is indicative of how the representees would act.3 Responsive representers act for or speak for the representees, playing the part of an agent in relation to a principal; how they act is responsive to how the representees would want them to act. Both sorts of representation, so I shall assume, have to be authorized by the representees.
Authorized indicative representers I describe as proxies, authorized responsive representers as deputies. Deputies divide, in a traditional distinction, into delegates who are more or less explicitly directed by representees and trustees who have interpretive discretion in determining how to construe their representees. This is how the distinctions look.
Indicative (proxies) Responsive (deputies)
Directed (delegates) Interpretive (trustees).
Three metaphors have dominated the tradition of thinking about the meaning of representation (Skinner 2005). One of these metaphors is drawn from representation in the pictorial arts, and it maps onto indicative representation. As the painting is indicative of how the subject of the painting looks so on this image should representers be indicative of representees; they should be proxies. The other two metaphors are drawn respectively from the courts and from the theater: from the way in which an attorney represents a client, and an actor represents a character. These reflect the two different modes of responsive or deputy-style representation, one directed, the other interpretive. As the attorney acts under the explicit or implicit direction of a client, so should representers act under the explicit or implicit direction of their representees; they should be delegates. As the actor constructively interprets the mind of a character, so should representers interpret the mind of representees; they should be trustees.
The following two sections review these two conceptions of representation, looking at how they might apply with different sorts of representers and representees but focusing in particular on the representation of a people. The fourth section then asks after how these forms of representation might be organized in a democracy, focusing on a particular problem that is raised by responsive representation of the interpretive kind. In order to illustrate the usefulness of the distinctions, the paper concludes with an appendix that considers issues of representation in the presidential and parliamentary systems that are exemplified respectively by Washington and Westminster.
2. Indicative representation
The standard case
In the standard form of this first conception, the assumption is that there are a number of representers, that the representees are the people as a whole, and the indicative requirement is that the representers should be a reliable or representative sample of the representees. They should faithfully reproduce significant differences among the population, and reproduce them in proportion to their realization within the community. We are familiar nowadays with the idea of indicative representation in the context of statistics. The random sample that is polled in the statistical exercise is a representative sample in the sense of being a good indicator of the population as a whole.
This conception of representation goes naturally with the pictorial or figurative metaphor. As Quentin Skinner (2005, 163) argues, it appears in those parliamentarian writers in England of the mid-seventeenth century who look for a ‘speaking likeness’ of the people in those who rule them, ‘describing Parliament as a “representation” — a picture or portrait — of the body of the people’.
The relationship that is envisaged between the body of representatives and the body of the people is certainly figurative, as in the likeness between a portrait and the subject of the portrait. But it might best be conceptualized as the relationship whereby a painting or sculpture, particularly one of an abstract kind, can exemplify a feature of the subject: the mass of a body, the elegance of a movement, and so on. As a tailor’s swatch contains pieces of cloth that exemplify the purportedly significant features of various fabrics, so figurative works exemplify purportedly significant aspects of an object or substance or situation or whatever (Goodman 1969). And as figurative representations can play this part to an aesthetic purpose, so politically representative bodies ought to serve in a parallel, democratic role. They ought each to exemplify and indicate the presumptively significant aspects of the people they represent, reproducing salient variations amongst its members and in a proportion that corresponds to their distribution in the population.
The idea of indicative representation figures early in democratic theory, since it is the sort of representation that is achieved or is likely to be achieved under the lottery system that was favored by the Athenians and that also played an important part in later regimes like those of the Italian city republics (Waley 1988; Hansen 1991). This lottery system might be taken as a version of the technique of random sampling but random sampling put to use in the service of advancing goals espoused by the people as a whole. While it may have been motivated by a desire to have a regular turnover in the representer body, the important thing from our viewpoint is that it would have ensured a degree of proportional and indicative representation.
The indicative idea also appears in the jury system, as that was developed in medieval Europe (Abramson 1994). To be subjected to the judgment of one’s peers, whether in determining that there is a legal case to answer, or that one is legally liable, is to be exposed, not to a random arbiter — a chance enemy, perhaps — but to a body that stands in for the community as a whole. The idea is that the jurors should represent a cross-section of the community or at least of the fully enfranchised members: in medieval Europe, the mainstream, propertied males.4
The ideal of indicative representation was entrenched in the thinking of many of those associated with the American war of independence and the French revolution. Melanchton Smith could write in 1788, in opposing the American constitution: ‘The idea that naturally suggests itself to our minds, when we speak of representatives is, that they resemble those they represent; they should be a true picture of the people’ (Ketcham 2003, 342). Again, it was powerfuly endorsed in a speech given by Mirabeau to the French Constituent Assembly in January 1789, though he used the image of a map rather than a picture to get it across. According to this version of the model, ‘a representative body is to the nation what a chart is for the physical configuration of its soil: in all its parts, and as a whole, the representative body should at all times present a reduced picture of the people — their opinions, aspirations, and wishes, and that presentation should bear the relative proportion to the original precisely as a map brings before us mountains and dales, rivers and lakes, forests and plains, cities and towns’ (Pitkin 1969, 77).
With the growth of electoral machinery, the indicative idea was naturally applied to elections for the legislature, providing support for making the electoral system more and more proportional (Mill 1964). Is it also behind the practice of organizing the legislature around geographically dispersed districts? It is hard to believe that it did not play some role in justifying that practice but the evidence, according to (Rehfield 2005), is against this hypothesis. Still, districting does induce a similarity in one dimension — nowadays a fairly unimportant one — between the population as a whole and the legislature that represents it.
The indicative idea survives in the continuing enthusiasm for proportional representation and has been given new life in campaigns for supplementing electoral representation with novel, statistically representative bodies. It is there in the general policy of organizing citizens’ juries that would review various policy issues (Stewart, Kendall and Coote 1994). It is present in the notion of the deliberative opinion poll that is chosen as a random sample and then canvassed for its view on one or another issue at two separate times: first, before members of the sample make contact, and, second, after they come together to receive background information, to hear different points of view and to debate the right line to take on the issue under consideration (Fishkin 1997). A particularly striking example appears in the Citizens’ Assembly that was recently established in the Canadian province of British Columbia (Fung and Fagotto 2006; Ferejohn forthcoming). A more or less representative sample of 160 citizens was assembled and given the task, over much of 2004, of reviewing the existing electoral system in the light of various hearings and discussions, and making a recommendation on whether or not it should be amended. The group recommended a change that then went to referendum and won more than 50% support — just short of the quota required to trigger a change.5
The similarity between representer and representee does not mean, in itself, that the representee has any degree of control over the representer. The connection between them is not one under which the representer has to track the dispositions of the representee, and respond by acting in conformity to those dispositions. Rather it is a connection under which the representer is a good indicator or model of how the representee might have been disposed to do act or speak, if in the position of the representer. Thus the people of British Columbia did not control the Citizens’ Assembly but, given that the Assembly was a microcosm of the people, it might be thought to model the decision that the people as a whole would have made had it been given the same information.
But while the indicator relationship does not necessarily give control over representers to representees, it is possible for representees to exercise control by selecting suitable indicators as their representers, by subjecting them to constraints designed to ensure that they remain indicators, and by deselecting them if they fail to act appropriately. Such authorized, indicative representers, as indicated earlier, I describe as proxies of the representees; they stand in for those who determine their selection.
Suppose that I see someone who thinks like me, as we say, and is a good indicator in a certain domain of how I tend to respond on various issues. While I do not directly control what he or she does, I do assume a position of indirect control when I decide to authorize that person to act in my stead, say on a committee. By appointing that person rather than someone else, I make it more likely that the committee will act in a congenial manner: act, as they would have been led to act, had I myself been a member. And this indirect control can be further strengthened by additional measures. I may put other constraints or incentives in place that make it more likely that the person will simulate what I would have done on the committee: for example, I might take steps to insulate the person against special, warping motives. And I may retain the power of de-selecting the person, should he or she, for whatever reason, not serve my interests well. The person is my proxy, someone who takes my place, with my authority.
The extended case
These observations about the control that is possible with indicative representers should prompt us to recognize that while indicative representation has traditionally been associated with the proportional representation of groups, the model can also apply in very different contexts.
Consider the professional or quasi-professional public figures who are appointed by government, often under strict regulations and often with a (limited or unlimited) tenure that is independent of those who do the appointing. Obvious examples will be judges and prosecutors, public auditors and ombudsmen, agency heads and the members of public commissions. These figures will serve as proxies for the public, being appointed under public authorization because of the likelihood that they will serve the public well.
This likelihood may not be based on a similarity of nature between representers and representees but it will be based on a similarity in the tendency of their wishes. There will be a convergence between the goals the representers are professionally constrained and motivated to implement — at least ideally —and the goals that the public would want to promote in the domain of their operation. The representers are recruited to public service, plausibly, on the grounds that their interests, if things go well, are likely to be indicative of the interests of the public. And they can be subjected to constraints that make it even more likely that this will be so. Such figures can be cast as proxy representers of the people on the same grounds that certain proportional bodies can be cast as proxies.
Think of those who serve on a central bank like the Federal Reserve in the US. Or think of those who are members of an election commission of the kind that is given responsibility in many countries, at arm’s length from parliament, to draw the boundaries of electoral districts. Absent dependency on the elected officials who appoint them — absent reliance on those officials for continuing tenure, for example — such individuals may be moved by motives of professional reputation to act reliably in a manner that answers to the interest of the people as a whole. And this will be especially likely, if their performance is subject to public check and commentary. It is surely more likely that an independent electoral commission would act in this way, for example, than that elected members of parliament or congress would do so, since elected representatives would have special party interests in the shape of district boundaries (Pettit 2004).
The idea of appointing professional figures was not unknown to traditional democratic theory. A good example is provided by the official known as a ‘podesta’, who played a very important role in the life of twelfth and thirteenth century Italian city republics (Waley 1988). The podesta, who might serve at different times in different cities, was an outsider appointed for six months to a year in order to run a city’s affairs. He was given a range of issues to control, was often forced to act without having contact with special interest groups within the citizenry, and was subjected to intense scrutiny at the end of his tenure. Since he might hope to be appointed as podesta by different cities at different times, he would also have been susceptible to reputational motives for doing his job to the satisfaction of the citizenry. Such a figure would have been a good example, in our terms, of an indicative representer: a proxy agent who takes the place of those he represents.
But it is not just public appointees who can count as indicative representers in the extended sense we are examining. For another sort of example, think of the whistle-blowers who can expose abuses in public life: the ‘private attorneys general’, as they have been called (Rabkin 1998), who serve the public well by challenging certain laws before the courts. The fact that the public or people give whistle-blowers protection, and the fact that they give private attorneys general a license to use the courts as they do, means that these figures are authorized to act in their characteristic manner. In the aggregate they serve the public well, by most accounts, acting as they do on convergent if not always identical interests. Their authorization by the public means that like formally appointed officials, they can be seen as representers who are employed in their characteristic roles, precisely because of the indicative relationship between their dispositions and the presumptive dispositions of the people as a whole.