Trustees vs. Delegates in Historical Context

The treatment of the “trustee/delegate” problem as distinctive to democratic representative government (rather than a more general problem of decision making) may be due to the simultaneous emergence of the practice of political representation and notions of popular sovereignty in which political representatives were increasingly used to express the decisional authority of the people. The distinction between “trustees” and “delegates” traces back at least to Magna Carta in the early 13th century. At that time, representatives had little authority to act, let alone judge or enact law, but were merely to give their constituency's (town or borough) assent to the king's demand for money (Fasolt Reference Fasolt, Bowman and Cody1991). Over the next 500 years, that relationship would be complicated as representatives were given more decisional authority over law, and, simultaneously, popular sovereignty emerged as a governing ideal. By the time of the American Revolution, the question about whether representatives should be “trustees” or “delegates” of their constituents overlaid with the separate view that law should in some sense reflect the preferences of the people (Rehfeld Reference Rehfeld2005). The ideal that law should be both just and reflective of popular preferences—an ideal to which any kind of lawmaker might in practice aspire regardless of whether institutionally or legally bound to the people's will—became hard to distinguish from the practice of using political representatives as lawmakers themselves.

The underlying distinctions that were later subsumed by the “trustee/delegate” labels arose in debates about political representation in the 18th century. Scholars point to Edmund Burke as the exemplar of the “trustee” position and to a more defuse set of anti-Federalists to characterize the “delegate” position. In truth, their positions were more complicated and held by a wide variety of actors, each of whom had nuanced positions on the aims, sources of judgment, and responsiveness that should animate lawmakers more generally (Burke Reference Burke1774; Parsons [1778] 1983; see also Kenyon Reference Kenyon1966; Pitkin Reference Pitkin1967; Rakove Reference Rakove1996; Rehfeld Reference Rehfeld2005).

Consider, for example, the proposal to constitutionally protect the right of U.S. citizens to instruct their representatives. The proposal, made by Representative Thomas Tudor Tucker in 1789, would have become part of the First Amendment to the U.S. Constitution (concerning basic freedoms of speech, assembly, and religion). The proposal was soundly defeated in part by arguments, such as this one by Thomas Hartley:

Hartley's response illustrates two critical points. First, Hartley identifies all three moving parts that I claim comprise the “trustee/delegate” distinction. The aims of legislation should be the good of all (modified when appropriate by the good of a part), as judged by the lawmaker because of his or her proper deliberative position in the legislature. Furthermore, the lawmaker should be less responsive to external sanctions that would reflect “the prejudices and acrimony of the party,” relying instead on his or her own “honest reason” and the desire for “sound policy.” Second, although other speakers (and Hartley himself) would refer to “representatives” as well, his argument here is framed in terms of what a lawmaker ought to consider. The conflation of “representative” with the role of “lawmaker” (or, more generally, “decision maker”) that they take on once they are representatives is thus evidenced in the language used.

Stepping back from this particular example, we can characterize the trustee and delegate positions by systematically fixing their underlying three moving parts. The trustee position, often attributed to Edmund Burke, specifies that (1) national legislation ought to aim at the national good; (2) the representative, in deliberation with other legislators, should be the ultimate judge of what constitutes that national good; and (3) the representative should be less responsive to electoral sanctions, motivated instead by some form of civic virtue. Burke, for example, clearly identifies the first two features as separate features of the whole. Defending the national good as the proper aim of legislation, he writes, “If the local constituent should have an interest, or should form an hasty opinion, evidently opposite to the real good of the rest of the community, the member for that place [i.e., the representative] ought to be as far, as any other, from any endeavour to give it effect” (Burke Reference Burke1774, 448). Separately, Burke defended the view that representatives rather than their constituents or other third party ought to be the source of judgment about what constitutes the national good. “Your representative owes you, not his industry only, but his judgment; and he betrays, instead of serving you, if he sacrifices it to your opinion” (Burke Reference Burke1774, 446).

Given the attribution of “trustee” to Burke's conceptions, it may come as some surprise that he never uses the term in his Speech to the Electors at Bristol, keeping the parts more clearly separate. But two centuries later, those views have been conflated into the designation “trustee.” Although Pitkin's description of Burke was rightly complicated and nuanced, she attributes to Burke the “trustee” distinction, and rightly noted that in Burke's view a representative ought to act from a sense of right and wrong and be less responsive to sanctions. “The first concept of representation encountered in Burke's thought is thus an aristocracy of virtue and wisdom governing for the good of the entire nation” (Pitkin Reference Pitkin1967, 218, emphasis added). Adrian Vermeule's recent formulation of “trusteeship” is characteristic of its academic and popularFootnote ¹¹ treatment that also collapses its underlying variables. “Burke's trustee model of representation [is] a normative stance that sees a good representative as one who exercises independent judgment for the common weal, rather than simply acting so as to satisfy the preferences of a . . . constituency” (Vermeule Reference Vermeule2007, 218, emphasis added).

Delegate views of representation, like those Burke was countering,Footnote ¹² press instead for a close correspondence between the views of an electoral constituency and the votes of representatives. The view was perhaps best captured by the opponents of the American Constitution in the 1780s, the so-called anti-Federalists. For the delegate (1) the aims of legislation are the good of a particular electoral constituency, (2) citizens are the source of judgment about what constitutes that good, and (3) representatives are supposed to be highly responsive to the threat of sanction. Thus, so-called “delegates” seek to secure their constituents' interests as their constituents so define them.

These are characterizations, and a lot of variation can be found, a point I develop later. But in broad summary, the representative as delegate is historically one who (1) aims at the good of his or her constituents, (2) as judged by his or her constituents, (3) more responsive to external sanction (election or the avoidance of legal penalties), whereas the representative as trustee is one who (1) aims at the good of the whole, (2) as judged by the representative, (3) less responsive to sanction, but acting on some form of civic virtue instead.

The Dependency and Presumptions of Empirical and Normative Research

Within the current academic literature, there appear to be three separate positions on which scholars focus, covering a range of possibilities for each variable. Empirical scholars generally presume that democratic representatives should aim at constituent interests, as judged by the constituents, and be highly responsive to electoral sanction. Normative scholars appear on the other side of the spectrum, endorsing either mixed strategies or something rather close to Burke's trusteeship.

The reigning view within empirical political science presumes the delegate model of representation as the ideal. I say “presumes” because the norms that underlie most empirical research on the subject are not defended, even as they are asserted. Most empirical studies, for example, equate “better” or “good” representation with “electoral responsiveness,” measured by how closely a representative's voting record corresponds to his or her constituents' preferences (Gelman and King Reference Gelman and King1994). Others measure “good representation” by how many voters vote for winning candidates (Brunell Reference Brunell2006). Indeed, Zoltan Hajnal (Reference Hajnal2009) took this approach in this journal, critiquing current practice only to the extent that it is gets at imperfect measures of the correspondence between votes and election outcomes, not that that correspondence itself was to be questioned as the proper measure of “good” representation.

Empirical scholars may favor delegate views of representation because they are easier to measure: one need only compare roll call votes of representatives with public opinion surveys, or election outcomes with votes cast, to evaluate whether “good” representation in this sense is achieved.Footnote ¹³ But such a view ignores or presumes away the other more complicated question of what the ends of lawmaking should be, how to measure good deliberation, and other questions that make the trusteeship view of political representation a more plausible alternative, or at least not one to be summarily dismissed.Footnote ¹⁴

If there is any consensus in the normative literature on democratic theory, it is reflected by Hanna Pitkin's view that no single hard and fast rule should govern a representative in all cases. A democratic representative need not always be in agreement with his or her constituents, but “he must not normally come into conflict with their will when they have an express will. . .” (Pitkin Reference Pitkin1967, 163). The conflict, in Pitkin's mind, is hard to resolve, and instead of resolution we should look for accountability. “[W]hen a representative finds himself in conflict with his constituents' wishes, this fact must give him pause. It calls for a consideration of the reasons for the discrepancy; it may call for a reconsideration of his own views. It is not sufficient for him to choose; it is necessary that the choice be justifiable” (Pitkin Reference Pitkin1967, 163–4).

In real world politics, our evaluation of the appropriateness of a lawmaker so deviating from popular sentiment does not appear to be based on a principled view about whether representatives should be trustees or delegates. Rather, it appears to be based on whether we agree with the substantive issue at hand. Thus, some praise as a “profile in courage” Marjorie Magolis-Mezvinsky's 1993 decision to cast the decisive vote in favor of President Bill Clinton's first budget, despite its predictable electoral consequences for her as a representative of a conservative constituency, and ridicule President George W. Bush a decade later when he dismissed protesters as irrelevant to his own thinking about “just” policy in Iraq. Conceptually, each acted in pursuit of what he or she believed justice demanded, overriding the contrary judgment on the issue of those whom he or she represented.Footnote ¹⁵

Deliberative democrats have been more receptive to representative independence for precisely the reason just hinted at. Proper deliberative conditions require that representatives be able to revise their views and change their mind when confronted by better arguments.Footnote ¹⁶ This deliberative commitment requires greater independence of judgment than empirical political scientists presume when they measure “representation” as congruence between voter preferences and a representative's votes. Even given its variation, deliberative democrats are still democrats and thus add all sorts of democratic checks on the required deliberative independence of the representative. Thus are conditions of publicity, accountability, and transparency such critical checks on representative independence. In these views, democratic control unfolds in a process and relationship between the representative and voters not merely at the time of voting and reelection, but during the term of office.

Mansbridge's “Rethinking Representation”

If empirical and normative scholars effectively endorse different views of the proper role of representatives, Jane Mansbridge (Reference Mansbridge2003) proposes altogether eliminating the language of trustees and delegates, recasting the set of issues in terms of four new ways to “rethink representation”: “promissory,” “anticipatory,” “gyroscopic,” and “surrogate.” Mansbridge describes these as “legitimate forms of representation” even, she argues, if none of the last three “meets the criteria for democratic accountability. . .”. These ideal types are meant to be just that—descriptions of conceptual points that are not necessarily realized in any pure form. In practice, representatives act in a way that mixes these forms.

Rethinking the terms of representation in four categories is supposed to provide a better conceptual map than “trustee/delegate” by making possible the identification of “the underlying power relation in each form, the role of deliberation in each, and the normative criteria appropriate to each” (Mansbridge Reference Mansbridge2003, 515). But the conceptual claims that Mansbridge makes are less helpful to empirical and normative analysis than they ought to be because they simply trade one kind of complexity (the distinctions collapsed in the “trustee/delegate” distinction) for another. As the passage just quoted indicated, each of the four kinds of representation collapses at least three different things—power relations, modes of deliberation and normative criteria. Because power relations need not be dichotomous, and because there are many forms of deliberation and probably even more normative criteria by which to judge actions, it would appear that relying on these three conceptual descriptors we would arrive at an infinite number of ways to “rethink representation.” In short, it would be far more useful to isolate each subpart independently of, or in conjunction with, the other subparts to see how they do covary and to judge how they ought to covary.

Consider, for example, how Mansbridge's treatment of “promissory representation” adds to, rather than unpacks, the complexity of the “delegate” model. “Promissory representation” is a model in which representatives “promise” to do what they are authorized to do, and their failure to redeem the promissory note results in their facing electoral sanction by being turned out of office. “In promissory representation, the power relation from voter to representative, principal to agent, runs forward in linear fashion. By exacting a promise, the voter at Time 1 (the election) exercises power, or tries to exercise power, over the representative at Time 2” (Mansbridge Reference Mansbridge2003, 516). In sum, “Promissory representation thus focuses on the normative duty to keep promises made in the authorizing election (Time 1), uses a conception of the voter's power over the representative that assumes forward looking intentionality, embodies a relatively unmediated version of the constituent's will, and results in accountability through sanction” (Mansbridge Reference Mansbridge2003, 516).

As we saw previously, so-called delegate representation involved two of these parts—delegates were to use their constituents' judgment in determining how to vote and be responsive to external sanction for failing to do this. But here the promissory representative is to be responsive to the voter's power and nonresponsive to that power in so far as he or she simply has a “normative duty to keep promises” quite apart from fearing the electoral sanction he or she is sure to face. On top of this, “promissory representation” adds three more distinctions. First, the will of the voters should be “relatively unmediated” (so we could presumably have a “mediated” version). Second, the location of the commands is shifted: unlike delegated representation in which instructions are issued from the constituency to the representative, in promissory representation, the terms of the agreement are set by the candidate or the party in the form of campaign promises.

Finally, by referring to “forward looking intentionality,” promissory representation makes explicit a confusing distinction expressed in the widely used terms “prospective” and “retrospective” voting. Reflecting those terms, Mansbridge means that promissory representation uses so-called prospective voting in which voters consider future performance when casting a vote rather than using a vote as sanction for past actions. But the terms and description “forward looking” applied to intentionality and “prospective/retrospective” applied to voting are confusing because intentionality and the power that we express when we cast a vote are always, by necessity, forward looking. Unless we were able to move back in time, we cannot express “backward intentionality,” nor can we vote retrospectively, that is, in a manner that affects the past. It is an important philosophical point: votes for political representatives always amount to this intention, “I intend that going forward you should be my representative.” Even in so-called retrospective voting we use the forward-looking, prospective intention of our vote to reward or punish past performance; in “prospective voting,” we use our expectations about future performance to cast our vote. In both cases, it is the evidence we consider, not our votes, that is prospective or retrospective, forward or backward looking. Votes may be given as sanction and reward for past action, but they are always grants of power for the future.

Despite its increased complexity, the core of promissory representation is remarkably similar to the “delegate” model: representatives are to closely track the judgment of their home constituency, as set out in a specific set of known terms that serve as the condition of service, and be responsive to the sanctions they face for failing to comply. In both cases, representatives follow a precommitted strategy, and their failure to follow that precommitment is seen as good reason to terminate their future service. Of course, there may be situations in which these variables take on the particular characteristics that Mansbridge attributes to them, and in those cases, there is no harm in calling it “promissory representation.” But for the purpose of analyzing political phenomena, such complexity is more likely to introduce confusion than clarification. Mansbridge's other categories face similar problems.Footnote ¹⁷

That said, Mansbridge's terms “gyroscopic” and “induced” in her last set of distinctions is more conceptually helpful to describe the extent to which a representative is responsive to sanctioning. Here, we can use “gyroscopic” to refer to a representative who is relatively unresponsive to sanction—he or she acts based on principles and policies to which he or she is committed, quite apart from the personal consequences of pursuing them. In contrast, an “induced” representative is more responsive to sanctioning, acting with an eye toward the personal consequences of his or her actions. This, however, is a much pared-down version of Mansbridge's conception. Mansbridge herself uses a more complex formulation: in “gyroscopic representation, the representative looks within, as a basis for action, to conceptions of interest, ‘common sense,’ and principles derived in part from the representative's own background” (Mansbridge Reference Mansbridge2003, 515). By her treatment, the gyroscopic representative is not merely less responsive to sanctioning, but also looks to herself as her own source of judgment and basis for determining the proper ends of legislation.

Restricting the terms “gyroscopic” and “induced” to describe the relative responsiveness of a representative to sanction, but jettisoning the other features of Mansbridge's description, isolates the critical ideas at its core and helps demonstrate why “gyroscopic” is an apt term. As Mansbridge argued, voters often want to support representatives who are less likely to change their mind based on the prospects of being voted out of office. What makes a gyroscopic (i.e., less responsive) representative different from a responsive one does not thus rely on the substantive position either takes ex ante. Rather, as Mansbridge argues, gyroscopic representatives may be desirable because they need less auditing to make sure they pursue their objectives and will stick to their guns even if it will cost them their jobs (Mansbridge N.d.; see also Fearon Reference Fearon, Przeworski, Stokes and Manin1999).

Conceptually, gyroscopic representation is of a piece with traditional concerns about “virtue.” Like gyroscopic representation, “virtuous” representatives were supposed to act by a “proper” motivation stemming not from external rewards or punishments, but from an internal, properly attuned sense of right and wrong. But using “virtue” in this way is itself a gross concept spanning two dimensions (Shapiro Reference Shapiro2005). It signified that a person was internally motivated (rather than externally induced), and a representative who had “political virtue” was one who had the purported right substantive view on issues and personal conduct. Thus, the truly virtuous person not only did the right thing, but also did it for the right reasons (i.e., motivated by the right sorts of considerations). This is why James Madison could argue that institutions might be designed to encourage people to act as if they were virtuous, that is, so that their external behavior would be similar enough to those who were in fact virtuous regardless of whether their internal motivations were good, right, or just. That two representatives voted the same (right) way was not enough to know that they were virtuous, but it was enough in his mind for a reasonably just system. Using Mansbridge's term “gyroscopic” to describe those who are less responsive to sanction is so helpful in this regard: it allows us to focus on a representative's responsiveness to sanction without specifying the content of the representative's views.

The Three Distinctions and the Independence/Mandate Controversy

One reason the previous argument might sound a little fishy is that the three distinctions of aims, judgment, and responsiveness do not encompass the authority and will of the voters. This is problematic if only because the traditional “trustee/delegate” debate has closely tracked the “independence/mandate” controversy in which we ask whether a representative acts independent of or mandated by the will or authority of voters (Pitkin Reference Pitkin1967, 144–67). Traditionally, so-called trustees are said to have greater authorial independence from their constituents' will, and so-called delegates appear constrained by the will of their constituents. The failure to include the location of authority as part of the underlying tripartite structure of the “trustee/delegate” debate certainly cuts against our general intuitions and long-standing historical use of these terms and is in need of explanation.

At first, it might appear that “responsiveness” captures this dimension because responsive representatives will tend to do what their constituents want (as if following their instructions), and less responsive representatives will tend to do less of what their constituents want (as if acting independently of them). But such a view misconstrues the nature of “responsiveness” and “authority.” Responsiveness, as I have used it here, simply describes a representative decision maker's sensitivity to sanction; by “authority,” we mean to describe the conditions under which an actor has the right to act in whatever way he or she is acting. Foreshadowing the following discussion, we can have representatives who are instructed and authorized to be unresponsive to sanctions, just as we can have fully independent representatives who, on their own authority decide to be extremely responsive to sanctions. Voters, for example, might want to authorize nonresponsive representatives in order to constrain themselves because they know that otherwise they would improperly meddle in his or her affairs. So, whether representatives have the authority to be more or less responsive is an open question that requires a separate determination. Authority regards where the right to decide lies; degrees of “responsiveness” will be specified as part of the content of that right.

To be sure, the question of where the authority lies to make a binding decision—even one that is otherwise just—is a critical question of sovereignty and may well determine whether we characterize a regime as reasonably democratic at all (Dworkin Reference Dworkin1977; Estlund Reference Estlund2008; Locke [1690] Reference Locke and Laslett1988; Rawls Reference Rawls1993; Raz Reference Raz1979; Simmons Reference Simmons1979). On one side, where citizens retain a constant authority to decide each piece of legislation, we have something approaching direct democracy. On the other side, we are unlikely to call “democratic” a case where citizens at one moment authorize someone to be the lawmaker and govern them, in perpetuity, never again to hold that person accountable.Footnote ²⁴ In between, we have a situation where citizens have at least the indirect authority over the law, through the representatives they authorize to vote for a time and later hold to account. The independence/mandate distinction thus also describes how much ongoing control constituents have over their representatives (Pitkin Reference Pitkin1967).

The role of authority in the traditional “independence/mandate” controversy, however, is somewhat murky because it collapses two separate questions: “who has the right to cast a vote on legislation?” and “who has the right to decide how a vote should be cast?” By “cast a vote,” I mean having one's voice (or will) count in some way toward determining whether a law is enacted. Members of Congress and other law makers, for example, presumably have this first right, the rest of us do not. By “decide how a vote should be cast,” I mean determining how a vote ought to be cast. This second concern is separate from the first—it explains what kinds of considerations those who have the right to vote ought to take up (or be bound by) when determining what vote to cast. It is this second presumption that explains why a lawmaker might be making a normative error in deciding his or her vote by relying on a horoscope rather than, say, the deliberation of his or her peers, without necessarily denying that he or she—rather than some other person—has the first right to have his or her vote count in determining the law.

I presume that a complete account of authority will specify the precise relationship between the right to vote on legislation (or binding decisions, more generally) and the right to decide how that vote should be cast. It might turn out that the first right (to cast a vote) depends on properly exercising the second right (to determine how the vote ought to be cast), or vice versa. Or, it might turn out that the first right (to cast a vote) is determined solely by a selection procedure without any restrictions on the proper exercise of that right once obtained. The point here is only conceptual: the right to cast a vote is conceptually separate from the right to decide how that vote should be cast, and we may remain formally agnostic about the relationship between these two rights.

With these two distinctions in mind, the debate about “instructions” or “independence” and “mandates” should be understood to concern this second right over who has the authority to decide how a vote should be cast (or other decision made), and not over the right to vote on legislation itself. This should not be controversial: even if a representative is mandated and obliged to follow someone else's will (about how to determine how to vote), it is still the representative, and only the representative, who we presume has the right to actually cast the vote in the legislature. In the extreme case, an instructed or mandated representative is someone who lacks the second right to decide how to cast his or her vote; an independent representative is someone who has the right to decide how to cast his or her vote. But, in both cases, the representative-decision maker, and only that person, retains the first right, the right to cast the vote itself. Indeed, if someone else were to attempt to cast the vote in the legislature, it would not count, even if she were casting the vote in exactly the same way as the person who had that right would have cast it.

So far this clarifies but is not at odds with the conventional story. The complexity arises when we refer to Table 2, and consider that each of the eight cells are actually descriptions of different ways that a person might determine how to cast his or her vote. In other words, deciding how to cast a vote will entail a specification of one of the eight cells. The right to decide how a vote should be cast thus means, in part, that one has the right to determine toward which of the eight ideal types the holder of the first right (e.g., the member of the legislature) should strive. No matter the degrees of freedom a representative has relative to his or her home constituency, he or she (or his or her constituents) will have to fix the 3 central distinctions of aims, sources of judgment, and responsiveness that were part of the traditional trustee/delegate model. Put differently, each of the eight ideal types is completely, and unproblematically, consistent with either the independence or mandated view of decisional authority. An independent representative is one who has full authority to decide how to cast his or her vote; a mandated representative is one who must follow his or her constituents' instructions in deciding how to cast his or her vote. As I show here, this explains why the seemingly paradoxical “mandated” trustee and “independent” delegate can arise. In short, the location of authority concerns who decides how votes should be cast, and that involves specifying in which of the eight cells decision makers ought to operate.

To illustrate this, I turn to the margins where unusual and seemingly paradoxical cases arise. First, the case of “mandated” Burkean trustees (cell A) and, second, the case of “independent” delegates (cell H). Again, I emphasize that I am not offering a substantive account of where the right to determine how to vote ought to reside. Rather, I am explaining why such an account is separate from the three subsidiary distinctions outlined previously.

The Mandated Trustee

We start by considering how the apparently paradoxical “instructed” or “mandated” Burkean trustee might arise.

In the traditional mandated view, representatives must follow the instructions given to them by their constituents concerning how to vote on any legislation. By this view, the only way the representative, by right, may exercise his or her first right to cast a vote on legislation is by following the instructions given to him or her by his or her constituents. Thus, to conform, constituents must issue instructions of this sort:

where “X” is a directive that determines how the representative must vote. Often, “X” takes on a very narrow value (e.g., “X = Our will about how to vote on the proposed law”). But decisional authority is agnostic with regard to “X”: as long as the determination of “X” is retained by the constituency, and as long as the representative is required to follow “X” when casting his or her vote, the representative is an instructed representative.

So far, this is consistent with the usual story that links “instructions” with “delegacy.” For it is often the case that the content of the instructions a constituency uses to instruct their representative is something like, “Pursue our good, as we see it, and you will be rewarded with reelection,” thus appealing to the distinctions described by the label “delegate.” Let us consider, though, what happens when “X” takes on more complicated values. Instead of the narrow case where X is set to the will of the constituency and that will is decisive (e.g., “for,” “against”), we can imagine cases where X takes on conditional values like this:

where “P” is some policy (e.g., “raise taxes,” “go to war”). In theory, we can imagine all kinds of complicated instructions that a constituency might command, even while retaining their authority over how their representative ought to vote. Complicated, contingent instructions are undoubtedly cumbersome and demanding on a constituency, but that's a separate problem: as long as the representatives must follow their constituent's instructions as to how to cast their vote and so long as their constituents provide instructions, we have mandated representation and instructed representatives.

So now, let's imagine that the constituency's instruction, “X,” takes on this value:

If we add to this a self-binding provision: “do not act with regard to future sanctions,” we now have the case of a fully instructed Burkean trustee (Table 2, cell A). Indeed, we could imagine an account of authority in which a representative should be so instructed before every vote by their home constituency to act in this way. This would be a case where the representative's authority to determine how to vote rested on the instructions from the home constituency, yet the representative would descriptively be indistinguishable from the Burkean trustee.

It might even turn out that this is exactly how democratic authority ought to operate: constituents instructing their representative on every vote, to use their own judgment about which of the eight cells in Table 2they should follow, in determining how to vote on proposed legislation. We could even add to this normative possibility a sociological story that makes its exercise more probable. Over time, as the public came to see the value of, and rely on, a Burkean trustee, they might begin to alter their judgments about how authority should be exercised, and come to the view that constituents need not instruct their representative before each vote to act as a Burkean trustee. In such a case, they might arrive at the quite different view that home constituencies could by right instruct the representative to act for a time (i.e., the electoral period) as a Burkean trustee. Indeed, given Burke's interest in clearly articulating to his or her constituents how he or she would act as their representative (what I called a “republican, self-reliant, gyroscopic” decision maker) and given his or her acceptance of voters as the proper source of his or her authority to be such a decision-maker, I believe Burke was historically an instructed trustee of precisely this sort. But strictly speaking, that Burkean trustee would have the authority to rely on his or her own judgment about the public good, being less responsive to sanctions, only to the extent that this is what his or her constituents instructed him or her to do. As the example demonstrates, these concerns, which touch directly upon issues of sovereignty and authority, are conceptually independent of the three distinctions underlying the trustee/delegate debate.

The Independent Delegate

Now let's consider the other extreme, the apparently paradoxical “independent delegate.” In this case, we would have an account of representative authority in which a representative has both the first right to cast a vote and the second right to determine how that vote should be cast. Doing so, she nevertheless acts as his or her constituents' delegate, doing exactly what they want on every issue when voting in the legislature.

In the traditional independence view, constituents decide who has the first right to cast a vote in the legislature by selecting a lawmaker, and in doing so, create political actors who have the second right to decide how to vote. By this view, the representative may, by right, exercise his or her first right to vote on law in the legislature by following his or her own decision rules on how to cast his or her vote. Thus, to be independent of their constituents, representatives must be free to follow their own rules in the following form:

where “X” is a directive that determines how the representative must vote. Often “X” takes on a very narrow value (e.g., “X = My own will about how to vote on the proposed law”). At the time, the representative goes to cast his or her vote, this results in a clear judgment (e.g., “For”), but it is based on a more substantive view about how the representative ought to cast his or her vote. But the location of decisional authority is agnostic with regard to the substance of “X”: as long as the representative has the authority to determine what “X” should be, he or she is formally independent of his or her constituency's authority to decide how to make his or her decision.

So far, this is consistent with the usual story that links “independence” with “trusteeship.” For it is often the case that when representatives retain decisional authority, the rule they follow is something like “Pursue the nation's good, as I see it, and ignore public opinion polls,” thus expressing the content of cell A, the Burkean trustee. Paralleling the previous case, let us consider what happens when “X” takes on more complicated values. Instead of the narrow case (e.g., “For,” “Against”), we can imagine cases where “X” takes on conditional values like this:

where “P” is some policy (e.g., “raise taxes,” “go to war;” etc). In theory, we can imagine all kinds of complicated rules that a representative might follow decision rules fully consistent with retaining his or her authority over determining how to vote. Complicated, contingent decision rules are undoubtedly cumbersome and demanding on a representative—perhaps he or she should empower a think-tank to come up with recommendations—but that's a separate problem: as long as representatives follow their own views about how to decide how to vote, and not their constituents', we have an independent representative.

So now, let's imagine that the representative's decision rule determining how to vote takes on this value:

If we add to this, “and I will act in a manner extremely responsive to my future electoral success,” we now have the case in which a completely independent representative exercises his or her independence by becoming a delegate. Indeed, it might turn out that the proper account is one in which a constituency alienates its authority to a representative who is then obliged to act as their delegate. This would be a case where the decisional authority on every vote rested with the representative, yet the representative would descriptively be indistinguishable from what we usually call a delegate.

It might even turn out that this is exactly how democratic authority ought to operate: independent representatives turning to their constituents' judgment about which of the eight cells in Table 2they should follow, in determining how to vote on proposed legislation. We could even add to this normative possibility a sociological story that makes its exercise more probable. Perhaps we believe that citizen preferences ought to determine legislation as much as possible, but because it is often hard to determine when collective citizen preferences exist at all, we believe it would be imprudent if not catastrophic to enact direct democracy. Instead, we believe that representatives have the authority to decide how to decide: they use that authority by first consulting their citizen's judgments, but then choose another way to decide when that one proves indeterminate.Footnote ²⁵ Again, a normative account of authority would explain which position we ought to endorse, but it is at least not implausible to see how an independent delegate might arise.