2.1 Responsibility

The word responsibility has three common meanings, all of which are important for understanding how policy work ought to be done. Having responsibility for something implies a duty to do it. Taking responsibility for something implies having the discretion to do it with more or less independence from other agents. Being responsible for something implies being held accountable to someone or to some collective for doing it.

Reference BertelliBertelli (2021, 197) argues that responsible policy work in a representative democracy must meet the following three criteria:

1. 1. “A policy worker, when performing policy work, has an individual duty to reinforce political values through governance values.”

2. 2. “Policy workers cannot perform ultra vires policy work (i.e., that which is beyond their legal powers), even if it reinforces values.”

3. 3. “Policy workers must act to reinforce values if they have the capability to use means that are in fact likely to succeed in reinforcement. They are positively obligated to reinforce political values whenever possible and not ultra vires.”

It is important to recognize that the responsibility of policy workers has an important implication for those who design governance structures, be they academics, advocates, or policymakers. Their designs ought to adhere to the complementarity principle. This is because well-designed governance structures are those which respect and facilitate the obligation of policy workers to behave responsibly, and that is both to refrain from acting outside of their purview and to reinforce, whenever possible, the values of their political systems through their individual positive actions. This is not yet the argument in favor of the indispensability of the complementarity principle, but rather the substantive view of responsible policy work that supports that positive argument, which we shall defend in due course. For our current purposes of formulating the argument, it is enough to state the criteria plainly.

2.2 The Formal Argument for Value-Congruity

Given the central idea underlying the complementarity principle, its basic formulations, and the substantive view of responsible policy work articulated above, we can formalize an argument in support of the complementarity principle as follows:

1. 1. Responsible policy work requires policy workers to reinforce the political values of the system in which they operate.

2. 2. Since responsible policy work requires policy workers to reinforce the political values of the system in which they operate, governance structures ought to be designed with the relevant political values and their reinforcement in mind.

3. 3. The complementarity principle directs institutional designers to create governance structures with the relevant political values and their reinforcement in mind.

4. 4. Thus, governance structure designers ought to adhere to the complementarity principle.

Value-congruity is essential to democracy, we contend, because the responsible conduct of policy work is essential to democratic governance. Governance structures that work against responsible policy work do not facilitate democratic self-rule. Notice that, at its core, responsible policy work as we have described it concerns normative judgments – normative discretion – and provides a substantive view about the legitimate sources of the values that guide discretionary judgments.

2.3 The Problem of Inductive Normative Reasoning

A common practical problem for which we believe the complementarity principle offers a much-needed remedy arises because of everyday contingencies in policy work. Policy workers regularly meet with circumstances in which they must use their discretion – sometimes their technical discretion, but often their normative discretion as well – in order to execute the functions of their positions.Footnote ⁶ In theorizing about “good” public administration, it is often suggested that we should look to the kinds of institutional designs and discretionary actions that have yielded favorable outcomes and extrapolate from those examples general guidance about the appropriate way to design administrative systems, including the scope of and constraints on administrative discretion. This is an ordinary process of inductive reasoning, but it presents us with a perennial problem: It is impossible to determine (with certainty) a general principle on the sole basis of observed patterns. Doing so is especially problematic when the “sample size” for those patterns is very small.

In the domain of public administration, we observe a tendency to reason from what has been demonstrably effective for one or a few agencies to some general principle that purportedly describes what it means to do “good public administration.” With respect to administrative discretion, this kind of reasoning has led to a false choice between two positions. On one hand is the position that policy workers should be free to exercise their own normative discretion at will; on the other is the view that normative judgments are off limits to rank-and-file policy workers, and their discretion is limited only to technical aspects of policy implementation. But we know that it is a mistake to infer general principles from limited data, and governance structures designed to reflect either of these positions on authorized professional discretion are not likely to produce responsible policy work. A handful of instances, often taking place under idiosyncratic circumstances, where normative decisions made at the sole discretion of policy workers produced “good” results does not imply that a general policy or practice of freely wielded normative judgments will also “work” or “add value” across any given context. Nor do instances of institutional failure precipitated by policy workers’ normative judgments imply that all normative discretion should be out of bounds.

The issue is one of filling in the general premise needed to validly reach a general conclusion on the basis of restricted or localized evidence. The evidence in question may seem reliable, but nevertheless its general applicability is likely inconclusive. To see this, consider the following number sequence puzzle:

How should we fill in the unknown quantity? What number should come next in this sequence?

A reader may be thinking that 31 looks like a good candidate, because it seems to fit the probable operating principle generating the pattern observed thus far: 1+10 is 11; 11+10 is 21; 21+10 is 31; and so on. This infers the rule from the pattern by inductive reasoning rather than deducing the next number in the sequence by applying a known or specified rule. In fact, the next number in the sequence is 1211. The number after that is 111221. The reader has no way of knowing this and little chance of guessing it. The piece of information missing, which is crucial to accurately predicting the next number in the sequence, is the rule governing the generation of the sequence. This sequence (known as the Conway sequence) is self-describing; it is a “see-and-say” sequence.Footnote ⁷

Once you know the rule, you can deduce from that rule and the pattern you see what the next number in the sequence will be. The problem in real life is that there are few instances in which we actually know the rule. Since in many domains of human life, all we have are observations, we are often forced to make our best approximations of underlying organizing principles as we go along. Our best option is to infer a general principle from what we in fact observe. That is how scientific reasoning proceeds and how we make scientific progress. There is no reason, though, that determinations of what constitutes good public administration need to be based on this kind of reasoning. That is not a scientific question, but a normative one, and in the normative domain, one need not reason on an inductive basis alone.

General normative principles can be constructed by top-down processes beginning in conceptual analysis rather than by a bottom-up process that begins in observation. The complementarity principle is of the former variety, and it provides an alternative to inductive reasoning on normative matters in day-to-day policy work. We think that theorists are mistaken to default to inductive reasoning on normative questions about good policy work, and that the complementarity principle corrects that mistake by supplying a general rule which, if followed in governance structure design, should reliably facilitate good (i.e., responsible) public administration. It does this by building value-reinforcement mechanisms into the structures themselves and by providing general guidance for policy workers’ everyday use of normative discretion. The hope is that rather than reasoning like Commissioner Hector did from his circumstances to the general conclusion that independent commissions should not be charged with making policy, people in Hector’s position can look to the guiding principle and decide whether it would be responsible in these circumstances for an independent agency to be involved in policymaking.

2.4 Inductive Normative Reasoning in Public Administration

The complementarity principle is meant to supply an alternative to inductive normative reasoning in public administration, that is, the tendency to conclude that whatever works is what constitutes good policy work – that whatever gets the desired outcome is what makes a policy or its implementation good. We think that employing this kind of reasoning in the design and execution of governance arrangements, especially with respect to the permissibility of normative discretionary judgments, is a mistake that often leads to undemocratic practices and to irresponsible policy work.Footnote ⁸

A wide variety of arguments about “good” policy work succumb to the problem of inductive normative reasoning. To take one example, Reference GoodsellGoodsell (2011) contends that policy work done within organizations with “mission mystique” can improve the work that is done and enhance the organization’s link to democracy. He describes the “mission mystique agency” as one that “is endowed with an aura of positive institutional charisma that is derived from the nature of its mission and how well it is carried out” (477). Its benefit for “career employees” of the organization is that “the mystique fosters a personal commitment to advancement of the mission” and for “attentive outsiders, it generates admiration and respect” (477–478). He argues, inter alia, that mission mystique makes it possible for policy workers to maintain and build capacity even as representatives change the aims of policy work. His premises are observations like these: (a) The National Park Service has its “source of mystique” in “the magnificent beauty and deep cultural meaning of the national parks under their care” and (b) “[a]lthough its work has experienced varied treatment by presidential administrations since its creation under President Kennedy … its popularity on college campuses and fame around the world are as strong as ever” (482). His conclusion is “to urge agencies to look to a normative ideal called mission mystique” so that they can shore up their “institutional center,” which means a “clearly articulated and necessary mission; continuity of a competent workforce and institutional memory; and a generous and confident capacity for change” (490, emphasis added).

Another example is the value of a reputation for doing “good” policy work (Reference CarpenterCarpenter 2001; Reference Carpenter2010). When an organization develops a reputation doing “good” policy work, say, protecting whistleblowers, that organization is seen by representatives as the right source of policy workers for protecting whistleblowers, and this gives the organization lasting authority with both governments and key stakeholders. Reference Bertelli and BusuiocBertelli and Busuioc (2021, 40) note that observations about reputation-enhancing authority become premises in an “epistemic view” of reputation: (a) “the preferences of elected politicians, whether or not they are induced by the wants of the electorate, will not (or have not) achieved a ‘correct’ policy,” (b) reputation “over time and through experience with changing states of the world … bureaucrats to become better at making ‘correct’ policies, and they can use their reputation with audiences to maintain them,” so (c) the policy workers should negotiate with, rather than be controlled by, elected representatives. The basis for the second point is the observation of evidence of it in agencies such as the Food and Drug Administration (Reference CarpenterCarpenter 2010) or the US Post Office (Reference CarpenterCarpenter 2001). Reference Bertelli and BusuiocBertelli and Busuioc (2021) warn that this epistemic claim sets the stage for irresponsible policy work because a “good” reputation diminishes representatives’ incentive to monitor policy work and allows policy workers to use claims to stakeholders as a way to pressure representatives to agree with their version of “good” policy work.

2.5 Defending the Complementarity Principle

We think that the complementarity principle supplies the general normative guidance necessary for agency designers to facilitate, and indeed for policy workers to identify and to do, “good” (i.e., responsible) policy work in any given context. The task of the rest of this essay is to put the meat on the bones of the argument, explicated in Section 2.2, for the indispensability of adherence to the complementarity principle in governance structure design. We proceed as follows.

Section 3 supports the first premise with reasons why meeting the criteria enumerated above constitutes responsible policy work. This discussion pieces together what exactly we mean by democracy and democratic values and ultimately roots out the values of representative democracy. It also specifies what we mean by “good” or “bad” governance. One notable outcome of the discussion in this section is the conclusion that adherence to the complementarity principle is outcome independent. Outcomes are not and cannot be the only things that matter. Adhering to complementarity when designing the means of public administration is positively prescribed even if it sometimes results in suboptimal end states.

Section 4 supports the second premise by addressing the problem of irresponsible policy work. That is, we discuss the consequences of governance structures that divert the policy workers laboring within them from responsibility. One such structure is the representative bureaucracy (Reference KingsleyKingsley 1944; Reference MosherMosher 1968), which is designed to improve outcomes according to an ambiguous criterion of social equity (see Reference FredericksonFrederickson 1990). We show that while the policy-worker-as-citizen can adhere to a vision of social equity, doing so on the job in a way that elides responsibility as defined above is not consistent with democracy.

In Section 5, we further support the second premise of our argument for the indispensability of the complementarity principle through an illustrative case that contrasts two different instantiations of a governance structure that Reference BertelliBertelli (2021, 110) calls representative agency, which allows groups of citizens “to facilitate collective decisions that legitimate subsequent policy work.” The case illuminates the difference between a representative agency structure that reinforces the political values of representative democracy with one that (perhaps inadvertently) undermines them.

Section 6 offers support for the third premise in our argument by explaining the features of the complementarity principle in greater detail. We show that the principle is genuinely action-guiding and explicit in its purpose, that the kind of guidance it offers has up to now been lacking, and that without such guidance, institutional designers run the risk of wasting their time and energy designing what might turn out to be bad governance structures. That is, they will either fail to facilitate responsible policy work or, worse, positively encourage policy workers to act irresponsibly.

In Section 6, we summarize the argument and briefly recap the highlights in support of its premises. More importantly, we offer a few thoughts about how the complementarity principle can help to improve the democratic content of the governance structures we critique.

3.1 Democracy and Representative Government

Let us begin, then, by articulating a broad conception of democracy and the values it embodies. To fully situate the argument, we reiterate that Reference BertelliBertelli’s (2021) project is concerned with what the complementarity principle can do for governance structure design in representative democracies. With this scope restriction, his intention is to address the majority of operating democracies in the current global landscape; it is not to imply that complementarity is only applicable within such states. On the contrary, and as we shall argue toward the end of this section, the principle applies regardless of the form a government takes. Representative government is both widespread and largely taken to be among the best forms of government available from the perspective of human freedom, and it has flourished. It is also the form of government under which most of our likely readership lives. Its supremacy is not, however, a foregone conclusion for the purposes of this argument. Under any form of government, responsible policy work must meet the criteria specified above, but our primary defense of those criteria will be articulated from within the framework of representative democratic government.

Democratic values are found in the very notion of democracy. Democracy’s bumper-sticker slogan reads: government of the people, by the people, and for the people. As with any system of government, it organizes a society, and oversees a populace. Its governing members count themselves as numbers among that populace, serving and governing themselves in the same manner that they serve and govern the rest of the population (Reference Hobbes and CurleyHobbes 1668 ; Reference Locke and LaslettLocke 1689; Reference RousseauRousseau 2019). They serve and govern in accordance with a legal code (Reference DworkinDworkin 1990; Reference HartHart 1961; Reference RazRaz 2009). Those living in a democratic society are free to live their lives as they see fit within the bounds of the law, and to pursue their own ends in order to build fulfilling lives (Reference RawlsRawls 1999; Reference RazRaz 1986). Essentially, members of a democratic society must be afforded equal voice as well as equal rights and liberties, and they must be able to lead autonomous lives. These are the fundamentals of democratic society, at least from an ideal perspective, and they yield a rich characterization of democracy:

From this description of democracy, we can begin to extract a preliminary list of democratic values. These include self-governance, self-representation on matters of public concern, free and equal participation in public decision-making processes, equal voice (i.e., each person’s say matters equally), freedom of thought and action, and reasonable value pluralism. Left open are choices about how to organize a system of self-governance, how to conduct public decision-making processes, how to count or group constituent inputs, and what kinds of matters count as matters of public concern (among other things). For instance, a democratic system might rely on different kinds of input mechanisms and different methods of input aggregation, and it might rely on direct or indirect representation (see, e.g., Reference BertelliBertelli 2021, chs. 2–3).

3.2 Fidelitous Representation

Representative democracy has all the values of democracy in general – broad public participation, equal voice, collective will, and reasonable pluralism – plus the value of fidelitous representation, that is, the accurate representation of the collective will, which requires preserving the salience of any number of competing considerations. A representative represents with fidelity by taking into consideration the interests of not just those who voted for her, but all of those who reside within her district. In practice, constituencies are delineated along geographical, not ideological lines. Thus, to represent one’s entire constituency with fidelity, a representative must take into consideration not only the victorious bid on any given matter of public concern, but also the relative weights of support between it and any competing bids, and the relative weights of support among the bids that did not prevail in addition to any socially relevant features that set backers of different bids apart from each other, and the relative importance of those bids to those constituent subgroups. Why? Because this is what is required to represent everyone in a pluralist society, valuing their inputs equally, enacting the public will while also respecting minority opinions. This kind of weighing and measuring, used in concert with an appropriate principle of distributive justice, gives us a formula with which to prioritize the various public matters at issue in any given place at any given time.

When we say that fidelitous representation is that which accurately represents the collective will, our notion of accuracy is pluralistic and democratic. Everyone’s voice counts equally, everyone’s interests count equally, and nobody’s voice or interests are swept aside by virtue of being in the minority. Notice that this is not the same thing as impartial representation. To be truly impartial is to do more than leave out one’s party affiliation in representative decision-making; it is, by definition, to leave out the kinds of considerations mentioned above. Moreover, fidelitous representation does not require one to leave one’s political party affiliation at the door; it only requires that one’s partisan zeal be appropriately tempered by the presence and interests of nonparty members in one’s constituency. Fidelitous representation, then, is not impartial, but, rather, embraces a kind of “weighted partiality.” For instance, because certain career civil servants are likely to find themselves in positions that charge them with making policy decisions, the concept of fidelitous representation extends also to them in something like the Weberian civil service neutrality sense. But, here again, “neutrality” is not quite the right word, since using professional discretion in order to enact policies with fidelity may involve personal neutrality, but it does not necessitate full (or “true”) neutrality among all competing interests, including partisan ones, the consideration of which may be relevant.

One way to achieve fidelitous representation is by bringing the demos into the process of policy work via representative agency structures, which is the focus of Section 4.4. Before we get into examples, though, it is worth considering why we ought to prefer a representative democratic system (if, in fact, we should) to something like direct democracy, which is imbued with fidelitous (because it is one-to-one) representation from the start.

Opting for a representative system of democratic government is not, as some have argued (e.g., Reference CordelliCordelli 2020), a non-defeatable, logical consequence of basic human dignity and reason.Footnote ⁹ It is a matter of practicality and pragmatism. Ideally, a fully democratic system would consider every individual person’s will on every separate matter of public concern. But one-to-one representation (that is, self-representation) tends only to work on a small scale with a population made up entirely of able-minded adults. There is strength in numbers. There are many other good reasons to band together as a unified polity under largely shared political values. Individual voices get lost and muddled in vast seas of other voices, so as a matter of practicality, in large and largely pluralistic polities, representative government is the next best option. It pares down the number of voices at play, but it is intentionally designed such that representatives act as conduits for their constituencies, bundling the interests and individual wills of the people they represent, and expressing and advocating for these to the best of their abilities. It also provides a mechanism for distilling myriad positions on individual issues into broad policy platforms which reduces the number of options from which voters must choose, and which thereby combats indecision. A voter need not assent to every policy position in a political platform to prefer it over its rivals. And besides, it is easier to exercise influence over individual policy positions one does not expressly endorse from a vantage point within a party or a friendly constituency than it is from without.

As we shall argue shortly, responsible policy work flows from fidelitous representation, which is itself an instance, and an expression, of value-congruity. If a type of policy work is duly authorized through fidelitous representation, it would be utterly revolutionary for policy workers to act beyond their legal discretion when doing it. This would allow a policy worker to thwart the efforts of representatives to make democracy work. Of course, policy workers may well believe that any of the democratic values we have discussed above are not being upheld in policy work, and if they have the legal authority to maintain those values, responsibility dictates that policy workers are individually obligated to do so. We offer much more detail on these points in Section 3.3, but flagging them here helps to clarify the trajectory of the argument.

Before moving on, it is worth taking stock of the argument up to this point. In this section, we have so far offered an idealized characterization of representative democracy and a list of values that responsible policy workers in representative democratic systems are duty-bound to uphold. As indicated above, we will flesh out and defend the criteria of responsible policy work articulated in Section 2.1, but first, we should address a probable objection to our defense so far.

3.3 The Problem of Roles

One might object at this point that while complementarity is beginning to seem like a good principle for governance structure design in a representative democracy, adherence to the complementarity principle within any system of government means reinforcing the values of bad governments (the kind that systematically violate human rights, for instance) as well as good ones. Creating this kind of reinforcing feedback decreases the odds of dismantling and replacing bad governmental systems with good ones (those more conducive to human flourishing).

We consider this objection constructive insofar as it helps to clarify one of the primary features of the principle. The complementarity principle applies only to those who are concerned with designing governance structures that are supposed to be effective within a particular political framework. Just as the concept of democracy is silent on the desirability of particular political institutions, the complementarity principle is silent on the desirability of any given political system. Its purpose is to facilitate – or at least not to preclude – responsible policy work on the part of everyday policy workers. This is, in part, a matter of basic social stability. Policy workers carry out the everyday governmental operations that enable society to function. Doing responsible policy work enables it to function under conditions of stability. As we mentioned briefly above, stability is one of the primary reasons for forming a government. Social stability is essential to basic human flourishing. It is the foundation on which one can build a life, cultivate an identity, start and raise a family, embark on and build a career, and so on. And this is true under any kind of regime. Even if one’s options are severely restricted under an oppressive government, the expectation that things will remain as they are and that the basic structure of society is not likely to radically change overnight is essential to building a life for oneself in one’s present circumstances. Those circumstances cannot be vulnerable to serious upheaval, which adherence to complementarity guards against.

This does mean that governance structures designed to operate in an authoritarian state should be designed such that day-to-day policy work reinforces the values of that authoritarian system. Authoritarianism may be a suboptimal state of affairs, but rogue policy workers carrying out their work within such a state, enacting policies in ways that complement, say, socialist values instead of the ones embedded in their own governmental system, will not magically create more optimal conditions. Deviating from authoritarian norms in such a way is unlikely to make anyone’s life easier or to instill socialist values in the government. In fact, it is far more likely only to make the administration of public services less efficient, less stable, and less effective. Heeding the complementarity principle ensures that the operating government, however “bad” it may be, will at least keep serving the public in whatever ways it does so. Might worsening socio-economic conditions accelerate public agitation for change? Maybe, but to do so would nevertheless be a revolutionary, unilateral, unauthorized move (by either authoritarian or socialist standards), and an abuse of the policy worker’s position within the government. In short, it would be irresponsible. It is irresponsible to knowingly and intentionally mobilize what government machinery is under one’s own control against the values of the system, and in so doing, to jeopardize the stability of people’s social environments in the service of one’s own values.

Moreover, to behave in such a revolutionary way is to make the perhaps inaccurate assumption that all people in fact would prefer not to live under an authoritarian regime. It assumes what constitutes “good” government in substantive terms. The complementarity principle assumes nothing about “good” government in that sense. Its operating notion of goodness is purely functional. This government is “good” insofar as it is functioning optimally according to its own principles, whatever those may be. We remain agnostic as to the substance of good government because we respect the fact that reasonable people disagree on how best to fill in the details, and our aim is to supply a general principle that anyone designing governance mechanisms in any system can use to positive effect. Hobbes argued that imbuing a single ruler with the power to make governmental decisions for the people was the only way to secure “good” social order. Rousseau believed social order could best be achieved through governance systems that reflect the unified, collective will of the people, as established by majority rule. Whatever the details, though, the complementarity principle applies.

Substantive applications of the complementarity principle assess the goodness of government according to whether or not the government is functioning in accordance with a particular set of political values. Reference HeathHeath (2020), for instance, argues that public administration ought to be guided by the values of liberalism. His argument is essentially a special case of the more general argument we make here. Heath argues for value complementarity between the substantive values of the political system and the norms that ought to guide public policy work within a specifically liberal democratic framework. He argues that in a liberal democracy, the values of liberalism (and not those of democracy more generally) are those that ought to guide the administration of public goods and services because liberal values more comprehensively capture the values of the system. That is, specifically liberal values, because they are more comprehensive, are more practically applicable than general, perhaps somewhat abstract or obscure democratic values. Therefore, hewing closely to them as practical guides will yield better outcomes for people living in liberal democratic regimes. The same line of argument is available to people who believe that authoritarian rule is desirable.

While adherence to the complementarity principle means designing governance structures that are meant to function in authoritarian or despotic regimes to reinforce the values of authoritarianism or despotism, it does not foreclose the possibility of sociopolitical change. The crucial point is that it is not the place of policy workers to undermine the sovereignty or jeopardize the stability of the governments they serve. To do that is to do irresponsible policy work. Things may not be going well for the demos under authoritarian rule. Nevertheless, consider how much worse things could be for the people if policy workers were to follow institutional designs that (intentionally or otherwise) gummed up the works of the public administration they do have. Radical political change ought not to be led or carried out by the bureaucracy.

This is also not to say that policy workers must acquiesce or submit to an unjust or otherwise bad governmental structure outside of their formal roles as policy workers. A person is a person, and a policy worker qua person has the freedom to express discontent, to agitate for change, to vote as she sees fit, and to pursue her own conception of the good just as any other person does. This distinction between a policy worker qua policy worker and a policy worker qua person (or civilian, or citizen, or community member) draws out the first of two problems we take to be at issue with respect to responsible policy work. We call this the problem of roles:

We shall have more to say about this problem (and its resolution under the complementarity principle) shortly, but for now our point is that it is preferable to have policy workers behaving responsibly even if that means reinforcing bad political values, if only for the sake of having things go as well as can be expected given the circumstances. To see what makes this the clearly preferable state of affairs, consider what conditions might be like in any given society if all it took to fatally undermine a government’s individual initiatives, its overarching policy agenda, or indeed its sovereignty was going straight to its rank-and-file policy workers and marshaling their actions against the very system of which they are a part in a campaign to completely change the political landscape. Such a system would be vulnerable not only to the whims of policy workers themselves, but also to outside influence – even coercion – by more powerful forces than any directly targeted policy worker could reasonably be expected to resist. Designing governance structures in ways that take seriously and, in a sense, “bake in” responsible policy work (i.e., policy work that reinforces the political values of the system) guards against such corrupting influences, which itself remains a good feature even if those corrupting influences might keep the society in question from coming closer to realizing democratic ideals.

While the problem of roles is not unknown among public administration scholars, its democratic implications are less carefully charted. Consider cases of “guerrilla government” as explored by Reference O’LearyO’Leary (2010, 8), which is constituted of “actions of career public servants who work against the wishes – either implicitly or explicitly communicated – of their superiors.” She sees this as “a form of dissent that is usually carried out by those who are dissatisfied” with the policy work in which they are involved, though “for strategic reasons” these guerrillas “choose not to go public with their concerns” with only a few “outing themselves as whistle-blowers” (8). This is a clear instance of the problem of roles, and the guerrillas seem well aware of this: Guerrillas “work on the assumption that their work outside their agencies provides them a latitude that is not available in formal settings” (9). Their discretionary action is unguided. To wit, “[m]ost have a wider conceptualization of their work than that articulated by their agency’s formal and informal statements of mission” and some guerrillas are less guided still, “more freewheeling, doing what feels right to them” (9). Guerrillas irresponsibly resolve the problem of roles precisely because they “bring the credibility of the formal, bureaucratic, political system with them, as well as the credibility of their individual professions” to lend legitimacy to their positions.

Reference O’LearyO’Leary (2010, 9) recognizes the essence of the problem of roles, namely that “guerrillas run the risk of being unregulated themselves” and that they may be “promoting policies that may not be compatible with the system as a whole.” Still, the democratic importance of responsible policy work is not fully recognized in her conclusions. In making them, Reference O’LearyO’Leary (2010, 16) defines “ethical decision making” by quoting Reference CooperCooper (1998, 256–257), who claimed, “[a] truly responsible administrator will bear an obligation to propose changes when they become problematic for the wishes of the public, inconsistent with professional judgment, or in conflict with personal conscience.” We believe that Cooper’s statement resolves the problem of roles and the problem of levels (to be articulated in Section 4.2) in an irresponsible way because it identifies the wrong object of responsibility. Policy workers are responsible to the demos and the boundaries of responsible action are demarcated by the values of the system they serve and the law. Representatives are obligated to represent the same people with fidelity, raising what we will soon call the problem of levels, and policy workers cannot undermine that representation because their moral compass points them elsewhere without irresponsibly addressing the problem of roles. Crucially, we contend that policy workers in democratic regimes are obligated to act to reinforce democratic values if it is possible to do so and if the action likely to succeed in reinforcing those values. They are not obligated to act because of their own “conscience” or because their profession would encourage such action. Conscience and profession are valid reasons to “propose changes” just as they are valid reasons to resign one’s position as a policy worker. They are not criteria for responsible action. We will take up this matter in greater detail momentarily.

To see the problem of roles in the context of a representative democracy and how this problem might be alleviated by adhering to the complementarity principle, consider the contrasts between the cases of Edward Snowden and Cassidy Hutchinson. Snowden, acting in his capacity as a concerned citizen, believed himself to be upholding democratic values in exposing potentially explosive classified information to multiple entities, including journalists and foreign government agents.Footnote ¹⁰ Hutchinson also acted in her capacity as a concerned citizen in testifying before Congress about sensitive and potentially explosive information regarding the prior knowledge and states of mind of top government officials whose actions on and leading up to the events at the Capitol on January 6, 2021, may have jeopardized the stability of the United States government, and did in fact compromise the immediate safety of myriad representatives, officials, and civilians.

Many see Snowden as a hero: He exposed the government surveillance of ordinary citizens on a massive scale, at great risk to himself, for the benefit of us all. Others see him as a criminal: He stole government secrets and disclosed them without regard to the ways in which doing so might jeopardize national security. In his capacity as a policy worker, a contractor with the National Security Agency (NSA), he acted irresponsibly. In his capacity as a citizen and a member of the affected community, his behavior seems warranted. His position was not enviable. A serious tension developed for him between his obligations as dictated by his role as a policy worker and his obligations as befitting his values as a citizen. Note, though, that he only had the capacity to act as he did as a citizen because of his role as a policy worker. He only had access to the information he did because of the security clearance he was granted by the NSA, and he was, in his official capacity, sworn to respect the secrecy of sensitive and classified government information. Importantly, in his role as a policy worker, he was not forced to behave irresponsibly. He was not constrained in such a way that he had no choice but to behave irresponsibly to achieve his purported goals as a concerned citizen.

Snowden considered himself to be a textbook whistleblower. He exposed waste and inefficiency (extensive governmental resources being used to secure miniscule intelligence advantages), serious abuses of power (the unwarranted invasion of citizens’ privacy by their government on a massive, national scale), and gross mismanagement with respect to data handling by the NSA. Questions about how he went about exposing these problems, why he felt justified, and why he went about it the way he did bring his status as a whistleblower into question. Whistleblowing is protected because it offers policy workers a way to reinforce democratic values responsibly by disclosing waste, fraud, abuse and so forth in an appropriately controlled environment.

According to Snowden’s own version of events, he first raised his concerns somewhat informally with colleagues, asking how they thought the public would react to these programs if they were exposed and explained on the front page of the newspaper. He said that colleagues agreed that the public’s reaction would be overwhelmingly negative. He said that he brought these concerns to his direct supervisors (a claim which those supervisors dispute), and that he formally raised the issue several times through the appropriate channels that were available to him.Footnote ¹¹ Feeling as though his concerns had been dismissed, he proceeded to steal classified information and subsequently to disclose the classified materials he had taken to journalists and foreign government agents. Included in those materials were details about how the agency went about gathering the information it did, the disclosure of which he argued was necessary to prove his claims. In disclosing these details, he exposed state secrets, and probably cutting-edge intelligence-gathering methods and technologies to potentially hostile foreign entities.

According to our criteria, Snowden acted irresponsibly. Confronting the problem of roles, he was torn between upholding his values as an everyday citizen of the United States and his duties as an NSA contractor. Had the NSA structure been designed in accordance with the complementarity principle, Snowden’s dilemma might have been easier to resolve without resorting to taking professionally irresponsible action. If the agency was designed with democratic value reinforcement in mind, he might have had better guidance as to how he could uphold his obligation to act responsibly in his capacity as an NSA contractor. Snowden’s is a particularly illuminating case because the various factors at play within it (a) explain why his professional obligations deserved to be given priority over his personal values, (b) illustrate the importance of value complementarity in policy work more generally, and (c) demonstrate the salience of fidelitous representation. Moreover, his case shows how all of this holds true even when the government in question is behaving badly.

Snowden expressly believed that he was acting in accordance with democratic values. He rightly thought that, upon learning what he knew about the surveillance programs being used to collect their data, their personal correspondence, and their personal information, the people of the United States would consider the NSA program to be a gross invasion of privacy in violation of their constitutional rights against unlawful searches and seizures. The first problem for responsibility was that Snowden acted ultra vires. He acted in defiance of both the letter and the spirit of the law. He has correctly pointed out that he was not protected under federal whistleblower protection laws at the time of his disclosures. Had he disclosed the NSA’s programs and the corroborating documentation he procured to, say, a member of Congress with the proper security clearance to review that classified information, he would have been subject to arrest and prosecution. Though he would not have been formally protected as a contractor, it would have been reasonable to disclose his information according to procedures open to regular NSA employees and to demand the same kind of protection that any such employee would be afforded. Moreover, the responsible thing for his colleagues employed directly at NSA to do would have been to take his concerns to the appropriate entities for him, since doing so would reinforce the political values of the system of which they are a part, since they could do so without acting ultra vires, and since they (unlike Snowden) had the capability to do so using means within their legal constraints that would likely succeed in reinforcement.

It would have been equally reasonable for those to whom Snowden might have disclosed to act in their official capacity to uphold democratic values by doing whatever it took to provide him with the same protection his colleagues would have had. Members of Congress could have introduced a whistleblower protection bill to cover his actions, NSA officials could have defended his actions in court because that would square with the spirit of the law. However, such disclosure, to be responsible, also would have required that Snowden could count on those to whom he disclosed the information to respond appropriately, that is, with fidelity to the public interest, to his interests as a whistleblower, and to the democratic values that Snowden was trying to maintain. We cannot emphasize this enough. Snowden did not feel as though he could count on that kind of treatment, so instead, in the second problem for responsibility, he absconded with the information and disclosed it in spectacularly irresponsible fashion to reporters and foreign government agents. This reportedly damaged American intelligence gathering capabilities worldwide and potentially jeopardized the stability of the United States by unilaterally undermining the people’s trust in their government and in their elected representatives.

Having bypassed the channels deemed appropriate by, at least, the spirit of whistleblower protection laws, Snowden bypassed the people’s elected representatives, effectively denying them the opportunity to represent with fidelity – that is, to act in their constituents’ interests to stop the program of mass surveillance being conducted against them by their own government agencies. Because existing whistleblower protection laws and the democratic values underwriting them would have protected any regular NSA employee, there were obvious channels open to Snowden to be able to responsibly disclose the waste, abuse, and mismanagement he had identified. The problem lies in the structure of his policy work. Without the appropriate value reinforcement mechanisms built into the NSA organization and its contracts, particularly whistleblower protection for those with high-level security clearance, and without the assurance that his disclosure would be handled with fidelity by those to whom he ought to have disclosed, Snowden felt obligated to both expose the agency’s abuses and to protect himself through irresponsible actions, specifically those that violate the second criterion for responsible policy work.

Hutchinson’s case is slightly different, but it illustrates how it is possible to navigate the problem of roles in the wake of a government behaving badly without violating either personal values or professional obligations. Hutchinson recently defied her former boss and the former President of the United States by testifying before the Select Committee to Investigate the January 6 Attack on the United States Capitol, disclosing personal conversations between top advisors to President Trump along with details about the states of mind of the President and other actors as well as their behavior leading up to and on January 6, 2021. In her capacity as assistant to the White House Chief of Staff, Hutchinson had a duty to keep the confidence of West Wing officials while reinforcing the values of representative democracy to the best of her ability and where possible. As a citizen, she wanted the American people to know what she knew about the thoughts and actions of key officials in the days and hours before an attempt to stop the transfer of presidential power.Footnote ¹²

As policy workers, both Snowden and Hutchinson had an individual duty to reinforce the political values of the system in which they operated, to do so within the law, and to act on that duty if likely to succeed in reinforcing the relevant values. Both acted at great personal risk. One of them behaved responsibly and the other did not. In the absence of governance structures designed with value complementarity in mind, getting it right can come down to the toss of a coin. Though Hutchinson maintained confidence until she could offer testimony to an appropriate authority, she might just as well have reached out to any number of reputable journalists covering the story. Her actions have given elected representatives the opportunity to represent their constituents with fidelity, which reinforce the democratic values at the heart of the electoral process and the democratic value of fidelitous representation. That outcome was not preordained. The governance structure within which she was operating did not facilitate, much less ensure, responsible action. But the congressional committee did. In speaking to it, and through it, to the nation, she upheld her individual duty to take actions that were likely to reinforce the political values of representative democracy, where she could do so within the bounds of the spirit and the letter of the laws than constrain her in her official capacity. Snowden did not.

Designing institutions in accordance with the complementarity principle positions policy workers for success in their efforts to look out for the good of the people. The structures of public administration should be designed to facilitate responsible policy work – work that reinforces the political values of representative democracy. But why, one might ask, should I believe that these specific criteria are and ought to be the criteria for responsible policy work?

3.4 Components of Responsible Policy Work

At long last, the ground has been sufficiently cleared for us to flesh out and to defend the three criteria of responsible policy work articulated in Section 2.1. The first criterion Reference BertelliBertelli (2021, 197) gives for responsible policy work is that “a policy worker, when performing policy work, has an individual duty to reinforce political values through governance values.” Call this the individual duty component of responsibility. In a representative democracy, the individual duty component of responsible policy work holds because one’s role as a policy worker in the larger political system is that of a policy worker who is meant to carry out or to enact the policies that have been chosen by the public through an official electoral process. In other words, it is one’s job as a policy worker to function not as an individual with her own will, but as a component part of the machinery of the governmental system of which she is a part – this, simply put, is the functional explanation of what a policy worker does in a representative democracy. Because a policy worker’s role is to function as a part of the system rather than as an individual, value reinforcement is an intrinsic feature of “good” policy work in the same way that teeth meeting their counterparts in order to turn gears and belts using friction to spin plates makes them good parts of a machine, as opposed to defective or broken ones.Footnote ¹³

Importantly, this explanation of what it is to be a policy worker does not foreclose the use of authorized normative discretion in policy enactment. Just as an algorithm uses hypotheticals in logic gates to choose a proper course of action, policy workers have different avenues available to them, any of which, if followed with fidelity, might constitute a responsible execution of the function in question. The key to using one’s authorized normative discretion responsibly is reasoning according to the right set of values – the values of the system one serves.

All of these criteria, including the individual duty criterion, follow from the concept of government, or what it means to govern. Having opted for a particular system of government is to have opted for a particular set of political values – the set of political values according to which the system shall operate. As a component of that system, a policy worker is bound by the same set of political values – the values by which the system operates. Her personal values do not and ought not to enter into contention when she is deciding how she should execute her function in her official capacity. For all intents and purposes, while she is serving in her official capacity, her values are the values of the system. The individual duty component of responsibility, then, delimits which values may and may not enter into the deliberative processes of persons engaged in responsible policy work.

The second criterion says that “policy workers cannot perform ultra vires policy work, even if it reinforces values” (Reference BertelliBertelli 2021), because to do so is to act beyond the constraints on what a public servant is validly authorized to do. This criterion highlights and follows from another important feature of the very idea of government. Call this the rule of law component of responsibility. The legal code of a representative democracy specifies what a person – whether a citizen, a noncitizen community member, a public servant, or otherwise – may or may not do, full stop. If a policy worker is not, by some reasonable interpretation of the codes set forth, legally authorized to act in her official capacity in a way that would reinforce the values of representative democracy, it is incumbent on her to refrain from taking that action, if only because to do so at one’s discretion is to violate the terms of one’s public office. It is to take matters into one’s own hands, which is to step outside of the purview of one’s official role.

Even if taking matters into one’s own hands would in fact yield better consequences or better uphold the values of the system than would taking any course of action authorized to one’s position, to do so would constitute a dereliction of duty. The ends do not justify the means. If they did, any policy worker would be authorized to act on the basis of his or her own judgments to secure whatever he or she alone determined to be the result that best squares with the values of the system, which even if he or she turned out to be right would be to unacceptably circumvent the public process by which public policy is chosen and by which constraints on public actions are determined, and to thereby undercut the duly determined will of the people. The ends must be achieved by authorized means – and only by authorized means – if social stability is to hold.

The rule of law component imposes the second limit on policy workers’ authorized normative discretion; the values of the system impose the first. Together the individual duty and the rule of law components set the bounds within which one may perform one’s role as a public servant responsibly. While the first two components of responsible policy work circumscribe what one may do in one’s official capacity, the third deals entirely with what one must do.

The third and final criterion of responsible policy work says that “policy workers must act to reinforce values if they have the capability to use means that are, in fact, likely to succeed in reinforcement” (Reference BertelliBertelli 2021, 197). It is not a limit, but, rather, something of a propulsion mechanism. That is, policy workers are positively obligated to reinforce political values whenever it is possible, likely to succeed, and they are validly authorized to do so. Call this the capability component of responsibility. This criterion holds because the failure to reinforce political values when doing so is both possible and within one’s purview as a public servant is an abdication of duty, which is no more permissible than executing functions beyond those legally authorized and delegated to one’s office. Abdication of this kind threatens the sovereignty of the system from within. It would be as if a sentient bomb, having recently become a pacifist, declined to detonate and instead bonked its target with an ineffectual thud (Reference HofstadterHofstadter 1981). Even if the outcome would have caused a great deal of destruction, if doing so would have reinforced the values of the political system that sent it, it ought to have detonated in accordance with its job description. To willfully abdicate (perhaps on the grounds of one’s own deeply held values) is to defy one’s duty to execute the functions duly and legally delegated to one’s office. If you cannot or will not execute the functions delegated to you, you ought not hold the office at all. Resolving the problem of roles in this event means quitting the office.

When a policy worker does not have the means or authority to reinforce values, the criteria of responsibility respect the Kantian principle that “ought implies can,” though they do so with some nuance that is worth exploring. Many strands of the normative literature in public administration incorporate more-or-less deontological intuitions about what policy workers ought to do for society, as our discussion of social equity in Section 4.3 exemplifies. Consider a claim that all of us, policy workers included, “ought to feed the hungry.” There are certainly moments at which it would be impossible for a person to comply with this prescription, but the mandate to feed the hungry is not rendered void in these moments. Its normative force comes from the moral judgment that no one should have to suffer the psychological and physical consequences of hunger or to die for lack of basic necessities and so forth. Therefore, we should not feed the hungry only when we can, but always, and sometimes we will not be able to live up to what is morally required of us.

This speaks directly to claims that policy workers should “stand on principle.” Consider an argument that a policy worker should hand over classified documents to foreign entities because the people ought to know the information within them, and they should get it by any means necessary. Suppose that handing over the documents goes beyond the physical capabilities of that policy worker, or that it would require that individual to act illegally, or that delivering the documents to a foreign agent would undermine a core value of the system in which the policy worker operates. The three components of responsibility incorporate an understanding of all these scenarios, showing that a policy worker can responsibly fail to live up to some deeper sense of duty. It provides a rationale for a policy worker defying broader ethical principles even when, all else equal, they could be upheld. While duty does imply power in responsible policy work, the power is not merely physical and is not absolute. It is both conferred and circumscribed by these considerations: You are obligated to act if you in fact can act (capability), if it would reinforce the values of the system (individual duty), and if it would not require you to exceed your legally circumscribed authority (rule of law). It is crucial to recognize that the individual duty and rule of law components of responsibility contribute directly to what it means for a policy worker to have the power to act.

Not one word of our discussion of responsibility should be read to mean that responsible policy work means upholding or complementing the values of any particular office holder, but rather the values of the system of which one’s official function is a part. As our imagined President Eisenhower said, representative democracy does not necessitate that he, the President, should rule as a king, and that all policy workers are duty-bound to execute his vision. Instead, the system itself is imbued with certain values from the start, and those are the values with which policy workers’ actions – and the actions of the President – are duty-bound to comport.

The infamous “Saturday Night Massacre” on October 20, 1973, which occurred during the Watergate scandal, illustrates what we have been arguing in the preceding paragraphs. Special Prosecutor Archibald Cox refused to abide by an agreement between the White House and the Senate Watergate Committee in regard to taped conversations after being ordered by President Nixon to do so. Cox explained himself in detail in a televised press conference wherein he was asked “is not your intention in direct conflict with the President’s orders to you, and if it is and you’re fired by the end of this news conference, what happens then?” He replied, “I was appointed by the Attorney General. Under the statutes, the Attorney General and those to whom he delegates authority are in charge of all litigation, including the obtaining of evidence. I think there is a question whether anyone other than the Attorney General can give me any instructions that I have any legal obligation to obey” (New York Times 1973a, 60). As a consequence, President Nixon ordered Attorney General Eliot Richardson to fire Cox.

Richardson, who had appointed Cox, refused to carry out that order and resigned, writing in his letter to Nixon:

Nixon then turned to Deputy Attorney General William Ruckelshaus with the same order and the same result. Ruckelshaus, in his resignation letter, said only that “my conscience will not permit me to carry out your instruction to discharge Archibald Cox” (New York Times 1973b, 61). Years later, he explained “I was convinced that Cox had only been doing what he had the authority to do; what was really of concern to the president and the White House was that he was too close. He hadn’t engaged in any extraordinary improprieties, quite the contrary” (Reference McFaddenMcFadden 2019).

Ruckelshaus’s resignation made Solicitor General Robert Bork the Acting Attorney General, and Nixon wrote to him straightaway, ordering him to fire Cox who had “made it apparent that he will not comply with the instructions I issued to him … Clearly the government of the United States cannot function if employees of the executive branch are free to ignore in this fashion the instructions of the President” (Reference KilpatrickKilpatrick 1973, A1). Bork complied, Cox was fired and the FBI sealed his office as well as those of Richardson and Ruckelshaus.

While that night he said very little, years later, Bork explained his decision in an interview:

The US District Court for the District of Columbia disagreed, holding that

In their statements above, Richardson, Ruckelshaus, and Bork base their claims to responsibility on an individual duty to reinforce the values of the United States government – namely, maintaining the lawful place of the Department of Justice and the President within it – by ensuring the stability of their system of government in a constitutional crisis. It is the rule of law component of responsibility that focuses our attention on the bounds of their authority. However, they apparently differed in their views about capability: Richardson and Ruckelshaus saw no “extraordinary impropriety” and thus could not act. Bork thought he was authorized to remove Cox, and he could successfully maintain constitutional valuesFootnote ¹⁴ by making the dismissal explicitly on the President’s direction and without reference to the “extraordinary impropriety” standard.Footnote ¹⁵ While a court later held that he did not have that authority, Bork seems to have acted on a belief that he did, and that belief completes his claim to responsible action. Believing that he acted responsibly would not have been enough to end the matter, of course. Had Archibald Cox sought reinstatement as Special Prosecutor after the decision in Nader v. Bork (which he did not), Bork could not have ignored the court’s order and blocked him on the instructions of the President or his own view of responsible action. The rule of law and capability components would have come together in this instance to make such an act irresponsible as well as illegal.

Together, the individual duty, rule of law, and capability criteria offer a full description of responsible policy work. The first two describe the limitations on policy workers’ discretion which follow from the political values of the system of government in which they operate; the third imposes a positive obligation to act when doing so reinforces those values. So long as a policy worker complies with these criteria, she is executing her official function responsibly. To assent to this description is to accept the first premise of the argument for complementarity. The truth (or at least the plausibility) of that premise established, we are free to move on to our second premise.

4.1 Identity and Representation in Policy Work

Theories of representative bureaucracy posit a link between a particular group identification on the part of policy workers and a bias in the use of those workers’ discretion that favors the groups with which they identify. Consider a social identity group A, for example, women, Latin@ people, immigrants, veterans. Three claims prevail in the literature.

First, a claim of passive representation is that the proportion of policy workers identifying with A should reflect the proportion of A-identifiers in the polity. Writing in the United Kingdom, Reference KingsleyKingsley (1944) considered the issue to be class, and passive representation beyond the Oxbridge elite in the civil service was the object of his argument. In the United States, Reference LongLong (1952, 811) begins from a premise that the value of pluralism, of a political order in which everyone in the polity can participate, is crucial for the state to embrace, but “[i]mportant and vital interests in the United States are unrepresented, underrepresented, or malrepresented in Congress.” His view is that “[t]hese interests receive more effective and more responsible representation through administrative channels than through the legislature” (811). What legitimates representation at the administrative level of the state, for Long, is the value of responsiveness. Reference SaltzsteinSaltzstein (1979, 469) clarifies, “the position taken by such proponents is that bureaucracy will be responsive to the interests and desires of important social groups in society if its personnel are drawn proportionately from these social groups and share the same values and attitudes as the groups they represent.” This happens, writes Reference MeierMeier (1975, 528), because the mirroring of social identity groups among the cast of policy workers means that “the decisions made by the bureaucracy will be similar to the decisions made if the entire American public passed on the issue.” Social equity, on this view, demands that the community of policy workers becomes a microcosm of the pluralistic interests – proxied by the social identity groups – in the polity with the discretion to do policy work that is responsive to all those interests.

Second, a claim of active representation contends that A-identifiers hold “particular sets of values that, when better represented in the bureaucracy, translate into better policy and administrative outcomes” for A-identifiers in the polity (Reference Riccucci and Van RyzinRiccucci and Van Ryzin 2016, 23). As voiced in the representative bureaucracy literature, this claim is a positive, rather than a normative one. Relevant bureaucratic discretion and within-group similarities in policy objectives provide the opportunity for transforming passive to active representation in policy work (24). When active representation occurs, scholars in this tradition contend that the public administration “becomes a representative organization that may supplement the representation provided through other political institutions including elected legislatures” (Reference Bradbury and KelloughBradbury and Kellough 2011, 158, emphasis added). Active representation is not a full-on revolution from within the bureaucracy from this perspective, but both a complement and supplement to the lawmaking of the elected representatives of the people with the goal of embedding pluralism into administration, that is, “Democracy denotes all of the people, not just some … those who may not otherwise receive effective service provisions through traditional political mechanisms can rely on career bureaucrats to the extent that passive and active representation are linked” (Reference Riccucci and Van RyzinRiccucci and Van Ryzin 2016, 24).

A third claim is that of symbolic representation, which holds that “the mere existence of a passively represented bureaucracy … can produce a sense of trust and legitimacy among citizens who share those social origins resulting in cooperation from the citizens and ultimately the production of more effective policy outcomes” (Reference Riccucci and Van RyzinRiccucci and Van Ryzin 2016, 25). By manifesting trust in administrative processes, the claim is that the symbolism of passive representation legitimates policy workers’ efforts to act consistent with the wants of A-identifiers, bringing pluralism into the government as A-identifiers interact with other groups in government, shaping all policy workers’ perspectives and their policy work in the process.

These theoretical claims suppose that isomorphic group identification between members of the public and the pool of policy workers in public administration should translate to better representation of relevant group-related interests in public policy one way or another, for one reason or another. Though there is some empirical evidence to support this supposition, that evidence is (perhaps inevitably) piecemeal and inconsistent. At this point, the problem with relying on inductive normative reasoning becomes acute. The evidence that, in certain circumstances, the inclusion of in-group identifiers among decision-makers can secure desirable results for in-group members does not imply that, for every group identity issue, the inclusion of in-group members in the decision-making body will do the same. While noting and cataloging successful implementations of the in-group policy-making strategy is good as a matter of positive social science, the normative conclusion, explicit or implicit, that more isomorphic group identification improves the representation of all the people in the process of policymaking is not warranted on the basis of the available evidence. The problem lies in the initial supposition which oversimplifies the necessarily diverse and complex constellation of factors that go into determinations of interests and the reasons for pursuing or promoting some interests over others.

The group-identity-to-group-representation supposition essentially sets up at least one of the following unsupported inferences:

1. 1. Identity group membership implies representation;

2. 2. Identity group membership implies trust;

3. 3. Trust implies representation.

Both passive and active representation claims make the first kind of inference. Passive representation does this axiomatically: A-identifiers represent A-identifiers. Active representation does so with an inductive generalization: A-identifying policy workers have done their work so as to provide greater (or more frequent) benefits to A-identifiers when compared with non-A-identifying policy workers, which implies that A-identifiers represent A-identifiers. But mere membership in the same identity group does not guarantee identical interests, similar sets of interests, or even overlapping interests. Reference ShelbyShelby (2007) argues, for instance, that there is little reason to suspect that there exists a sufficient overlap in the substantive values and interests of Black people (whether in a single country or worldwide) to establish a robust conception of a Black political identity (1–23). Still, the fact that there is no monolithic Black experience – that “individual or group differentiation within the black population” exists and “that there are many, perhaps incommensurable, ways to be black” (3) – does not preclude a kind of Black solidarity. It just means that Black representation in public offices does not guarantee that the interests of individual Black people will be better represented, either actively or passively. There are simply too many differences among intra-group interests to expect that kind of outcome. The same kind of problem exists for both the establishment of trust (inference 2) and the achievement of better representation through symbolic representation (inference 3). It is not enough to establish trust between A-identifiers and institutions of government that some number of people holding public offices or working in public administration are themselves A-identifiers.

A related problem is associated with the logical importance of trust for inferences 2 and 3. Even if shared identity is sufficient to establish for some a kind of (perhaps misguided) trust by in-group members, observing that one does in fact trust those who share one’s identity does not imply that one’s interests will be better represented in the public sphere by those who share their group identity. One can be a woman and a misogynist, and acting on the latter trait, one can institute policies that might work against the interests of some large contingent of people who identify as women. Clearly, being of the same social identity group is not only insufficient to establish trust, it does not, by itself, warrant trust. In the absence of a much stronger connection between identity and shared interests, and without being able to ground trust or representation in identity, these kinds of representation ought not to be expected to generate the outcomes proponents claim – at least not at the scale at which they envision.

One other important unifying thread running through these theories of representative bureaucracy – one we address more substantially in the pages to come – is the somewhat cynical presumption of failure at the level of representative government. The assumption is that representative government is incapable of responding to particularized, group-differentiated interests. Basically, it assumes institutional failures of sensitivity to certain public interests as a standard hazard of broader democratic procedures. Rather than offering a way to shore up these real or perceived shortcomings (or democratic deficits) at the governmental level, these theories offer a sort of resigned remedy ex post – a shortcut that evades, or an end run around, presupposed inevitable failures. In a sense, this avenue of remedy is democracy-defeating, because it accepts certain failures of representation (or representational cracks) as intractable consequences of large-scale democratic processes. Proponents of these views have more or less conceded that democracy is inadequate to its greater purpose and in need of ad hoc repairs, stitches, or bandages to hold together its broken pieces. This observation brings us to a deeper discussion of what we call the problem of levels.

4.2 The Problem of Levels

The pursuit of social equity in public administration is a frequent justification for violating the criteria for responsible policy work that we described in the preceding section. A prominent school of thought holds that it is part of a policy worker’s job to use her discretion to strive for social equity in the everyday administration of public goods and services. Ultimately, we take the imperative to achieve social equity through day-to-day policy work to be obviated by the novel concept that we have introduced, and continue to defend, in the pages of this essay: fidelitous representation. We understand fidelitous representation to be a bedrock value and an essential component of any well-functioning representative democracy. It is, in a sense, an obvious precondition of such systems, but its obviousness has somehow gone unnoticed. As we said at the outset, there is real value in bringing what ought to be relatively obvious into sharper relief.

Fidelitous representation seems an obvious precondition of a well-functioning system of representative government because without it, the system is only nominally, as opposed to actually, representative. In its absence, the system cannot be legitimately representative. Surely, the very idea of representative government implies fidelitous representation. Nevertheless, the essential role of fidelitous representation in the concept of representative government is all too often overlooked. This essay demonstrates the benefits of bringing this value into the light. For one thing, it explains why building (unguided) normative discretion on issues of social equity into rank-and-file policy work risks directing public servants to behave irresponsibly. Doing so shifts the locus of legitimate social justice concerns from what we take to be the proper level of representatives, institutions, and institutional designers to the ground floor, so to speak, where such concerns can and, on the competing view, should be addressed by policy workers, according to their own views. Notice how this directly engages the problem of roles that we have spelled out above.

The disconnect lies in the pursuit of social equity at one’s professional, but unguided, discretion and the pursuit of social equity at one’s discretion as specified by institutional design. If value-congruity is built into the institution by design and the system itself values equity and diversity (as representative democracy does as a consequence of its commitments to equal voice and reasonable pluralism), then there will be fewer opportunities for policy workers to violate the conditions of responsible policy work in pursuit of social equity because those kinds of concerns are already built into what it means for the institution to function properly. We call this the problem of levels:

The problem of roles articulated in Section 3.3 draws out the problem of levels. As we have argued, it is not part of the role of a policy worker, when acting in her official capacity as a policy worker, to use her own judgment subject only to her own, unguided discretion to pursue whatever she sees as legitimate aims in her day-to-day work. Doing so means running the risk of engaging in either ultra vires activities (violating the rule of law component) or efforts that are out of step with the actual values of the system (violating the individual duty component of responsible policy work). But the contemporary focus in public administration on issues of social equity advises policy workers to do just that: to use their own judgment and their own (unguided) professional discretion to pursue what they see as legitimate social equity aims in their day-to-day work. Rank-and-file policy workers need guidance, grounded in a stable and uniform set of values, as to what social equity means and what measures to achieve it should look like if they are going to pursue such ends as a matter of routine policy work. That guidance can only come from the higher-order values embedded in the system of government within which one’s department or agency is situated. This goal is achievable by clearly connecting democratic values to what is meant by social equity.

4.3 Social Equity in Public Administration

The argument for active bureaucratic representation relies on a concept of social equity that is unclearly defined. Attempts by public administration scholars to define social equity have produced a variety of conceptually vague, often inconsistent, and sometimes contradictory descriptions of the term. Writes Reference Frederickson and MariniFrederickson (1971, 311), “the procedures of representative democracy presently operate in a way that either fails or only very gradually attempts to reverse systematic discrimination against disadvantaged minorities. Social equity, then, includes activities designed to enhance the political power and economic well-being of these minorities.” In 2000, a panel of the National Academy of Public Administration tautologically called it “The fair, just and equitable management of all institutions serving the public directly or by contract, and the fair, just and equitable distribution of public services, and implementation of public policy, and the commitment to promote fairness, justice, and equity in the formation of public policy” (Reference Svara and BrunetSvara and Brunet 2004, 101, emphasis added). This confusing state of affairs was apparent to Reference RosenbloomRosenbloom (2005, 248), who found it “worrisome for a term that gained prominence in the field of public administration as far back as 1968.” He was concerned, as we are, with the use of the elusive idea to influence policy work, writing that in the United States, “[c]learly, public administrators need a mandate from the constitutional branches of government to legitimate their pursuit of redistribution and social change, just as they would to justify repression in the name of administrative efficiency or national security” (251).

The conceptual confusion arises from several sources (for details on each of the following descriptions, see Levinson et al. 2022, 2–11). First, social equity is sometimes described in terms of conflicting notions of equality: in terms of equal access, equal opportunity, or equal outcomes. Sometimes, it is couched in terms of equal respect, experience, or treatment. Sometimes, it is described in terms of just distribution, demanding in some cases a “fair” distribution of resources, or in others the equal distribution of outcomes across populations. And still other times, it is described in terms of the priority of the less advantaged, sometimes according to relative disadvantage, sometimes according to absolute disadvantage. More often than not, a single source’s definition of the term will contain two or more of these characterizations, and more often than one might expect, two or more of them that stand in tension with one another, are incompatible, or are mutually inconsistent (Levinson et al. 2022, 10–11; see also Reference RosenbloomRosenbloom 2005).

The common thread present throughout these various descriptions of the concept consists of (a) the recognition of some kind of connection (either merely correlational or causal, or both) between certain morally irrelevant social differences (race, socioeconomic status, disability, etc.) and disparate social outcomes, along with (b) a goal of severing that connection in order to eliminate the disparate outcomes. What is unclear even to social equity advocates and scholars, it appears, is whether the connection to be severed is one of mere correlation or one of causation and whether severing the connection can actually be expected to successfully eliminate the disparities in social outcomes.

Scholars and advocates often come to endorse conflicting aims under the same umbrella term because they are trying to accommodate multiple possibilities. This is especially well illustrated by mission statements that advocate for both equal treatment and equal outcomes.Footnote ¹⁶ Consider the goals of equal treatment, as expressed through equality of opportunity, and equal outcomes. If being treated to the same opportunities for success fails to produce equally successful outcomes, presumably, we ought to engineer the right outcomes, but to do so, we will be forced to engage in disparate treatment. Why not just aim to engineer the right outcomes from the start? If the connection is causal, treating people to equal opportunities for success should help to eliminate the disparities in outcomes. If the connection is merely correlational, engineering equal outcomes may be the only way to eliminate the social disparities at issue. A presumption in favor of earned positive outcomes might explain the double-dipping. Still, when people advocate for “social equity” without specifying more clearly what they mean by it, what they mean to achieve, or the means by which they mean to achieve it, “[they] risk misleading [themselves] and others about the values for which [they] stand … [and] also risk sidestepping the hard judgments and potential trade-offs” (Levinson et al. 2022, 22–23) that are inextricable to the design of good governance structures – that which enables and encourages policy workers to act responsibly. Hence, what appears to be lacking in the public ethos is an authoritative specification (or series of specifications) as to what we as a polity mean by “social equity,” what we mean to achieve in the pursuit of it, and the means by which it is legitimate to pursue it. Without these details, policy workers lack the guidance necessary to obviate reliance on their personal value sets in determining how best to pursue equitable outcomes. That is, they are left – or worse, led – to engage in potentially irresponsible policy work.

According to Reference RazRaz (2006), one feature of legitimate authority is that it preempts other reasons for action. That is, the fact that some action would comply with an authoritative directive takes the place of other reasons for or against taking that action; the fact that it would comply with an authoritative directive is not merely added to the balance of other reasons for or against doing it (Reference RazRaz 2006, 1018–1019). Determinations of the legitimacy of any purported source of authority are and must be made by individuals, but these determinations are not made in a vacuum. Relevant structural norms help a person to decide whether an authority is legitimate or not. In the case of public administration, a person taking on the role of a policy worker must accept as authoritative the laws and the values of the system. When assuming the role of a policy worker and accepting these sources of authority, one also implicitly accepts the preemptive nature of any and all authoritative directives that issue from these sources. It is those sources – the system’s values and laws – that must determine the character, the relative importance, and the legitimate avenues of pursuit of matters of social equity in public administration if uniformity and consistency in such pursuits are to prevail.

One might understandably worry at this point that the very problem policy workers are confronting and trying to solve in using their (unguided) discretion to advance social equity is the inaction, or worse, the counteraction, on such issues at the higher levels of government. Laws are made by elected representatives, many of whom have no (politically motivated or constituency-induced) interest in advancing the goals of social equity, and institutional designers have other concerns when deciding how any given agency or institution will function. The very reasons public administration scholarship tends to foreground the pursuit of social equity at the unguided discretion of policy workers are (a) the failure at higher levels to attend to such issues and (b) the positions of policy workers, who must facilitate such pursuits, relative to the public.

The first reason is a failure in practice; it does not indicate a deeper shortcoming in our argument so far. If anything, it illustrates the heretofore (incredibly) nonobvious nature of the value of fidelitous representation. Should that value be acted on and reinforced as it should at every level of a representative democratic system, the impetus behind leaving important and difficult determinations on issues of social equity to the (again, professional but unguided) discretion of street-level workers would disappear. The second reason counts in favor of our proposed resolution to this practical failure. That fidelitous representation is hard and sometimes not achieved cannot justify the substitution of unguided policy work. Perhaps this statement has seemed obvious enough to political philosophers to make them less interested in discretionary policy work than they have been until recently.Footnote ¹⁷ Treating the arguments of public administration scholars carefully, which we are doing presently, makes it clear that fidelitous representation demands more attention than it has received by either scholarly community.

Policy workers are in fact those who must execute social equity pursuits in public administration; they are not, however, those who must or ought to be deciding on the substantive details of what those pursuits should be or aim to achieve. Our proposed resolution has already been set out in these pages: It is adherence to complementarity by embedding in governance structures expectations regarding the representative democratic value of fidelitous representation. Failure to attend to issues of social equity is a failure to represent with fidelity. Building expectations of fidelitous representation into what it means to do one’s job as a representative correctly – in line with the specified functions of the office – orients substantive social equity concerns to the level of governance at which they ought to operate. Once the relevant determinations are made at the appropriate level, rank-and-file policy workers have the guidance necessary to use their (guided) discretion to responsibly pursue issues of social equity. This does not make for a representative bureaucracy in the usual sense, but for a responsible one.

Recall that fidelitous representation involves actively preserving the salience of and responding with positive action to competing public interests within one’s constituency. This is a core democratic value implied by the characterization of democracy as government of the people, by the people, and for the people, where “the people” encompasses the whole population of the polity. Recall as well that we are trying to show that adherence to complementarity is indispensable to good governance structure design, where such “good” design means designing public agencies and their procedures to facilitate responsible policy work, and thus “administer democracy” (Reference BertelliBertelli 2021). In this sense, our project here uses certain ideals (e.g., fidelitous representation) as ultimate aims which in turn prescribe appropriate actions in real-world (and thus, nonideal) circumstances. Our point is that the solution to the failure at the higher levels of government to attend to issues of social equity is not to accept that failure and resort to delegating responsibility for such concerns down the chain to policy workers (see, e.g., Reference ZackaZacka 2017, 62–65). If the designers of governance structures adhered to the complementarity principle, and thus agencies were imbued with an expectation of fidelitous representation from the start, pursuits of social equity by policy workers would be less likely to result in ultra vires actions or in actions that defy or fail to uphold the individual duty and capability components of responsible policy work. This is because they would have stable guidance as to what it means to pursue issues of social equity in line with their roles as policy workers.

Notice that policy work that reinforces the democratic value of fidelitous representation resolves the problem of levels by strengthening the notion that questions of social equity need to be set at the level of representatives, institutions, and institutional designers. In so doing, it also helps to ease the tensions and internal struggles inherent to the problem of roles. If policy workers have express guidance on these issues, which themselves issue from the values of representative government, their own personal values need not enter their deliberations about what to do in any given situation. Because the values of the system, including the values of social equity and its legitimate avenues of pursuit, have been built into the governance structure within which policy work occurs, a policy worker is better positioned to act responsibly – that is, to choose a course of action that reinforces the values of the system and that falls within her legally authorized purview – to achieve what have been deemed (through the appropriate democratic channels) to be publicly desirable, socially equitable ends. If a circumstance arises that presents an as-yet unsettled question, she is better positioned to responsibly refrain from immediate action, and to decisively kick the question back up the chain of decision-making. An expectation of fidelitous representation built into the governance structure gives her solid footing from which to make demands of her superiors. A well-designed institution is one that facilitates responsible policy work – one in which governance values complement or reinforce the values of the larger political system. Thus, a well-designed institution is one that builds in expectations about and mechanisms which reinforce the value of fidelitous representation. Recall that Cassidy Hutchinson had a well-designed institution, namely, the January 6th Committee, but Edward Snowden did not. Notice that this also preserves the accountability of representatives to the public. Accountability for any street-level policy worker’s responsible course of action does not (as well it should not) fall on the policy worker. It is her job to carry out the orders given by her superiors (i.e., by elected representatives), and thus it is her superiors who are accountable for the outcomes.

Advocates of social equity as a guiding principle of public administration might argue that even in a pluralistic democracy characterized by fidelitous representation, the practice of fidelitous representation must be calculated on majoritarian grounds which will crowd out the genuine availability of certain valuable life options for minority groups. The only way to counter this is to design governance structures that actively attend to membership in the relevant groups in ways that defy the majoritarian principle at the heart of democracy.

What this line of argument gets wrong is the assumption that the value of majoritarianism enjoys an insuperable centrality in the concept of democracy. The value of pluralism is not superficial – it has teeth, and these two legitimately democratic values can – and often do – run up against each other because an essential feature of democratic pluralism is its grounding in autonomy. Trade-offs between these values must be made if autonomy is to be respected.

Reference SchoutenSchouten (2019) has recently argued not just for the permissibility, but, rather, for a positive obligation on the behalf of modern democratic governments to promote certain substantive and indeed group-differentiated interests on liberal democratic, substantive-value-neutral grounds. Her argument essentially constitutes a rebuttal to the view that democracy is incapable of responding to concerns about social equity. We have reasons that are neutral with respect to substantive values, she argues, to promote autonomy because autonomy is necessary for and essential to democratic governance (Reference SchoutenSchouten 2019, 170–197). Basically, in the absence of an autonomous populace, one in which people freely and independently decide on and pursue their own conceptions of the good life, democratic inputs are of little value. They simply do not and cannot express the interests of the people because the interests of the people are not being decided upon by the people. Consequently, the preservation and indeed the promotion of autonomy is itself a substantively neutral, purely political, democratic value. Democracy as we understand it is simply impossible without it.

Genuine autonomy, though, requires having a sufficient number of good options from which to choose (Reference RazRaz 1986). A person without stable housing, for instance, cannot make a genuinely autonomous decision between sleeping out in the cold or taking refuge in a crowded shelter where she will have a roof over her head, but is likely to be assaulted or robbed in the night. As the saying goes, she is stuck between a rock and a hard place. The adage itself acknowledges that this kind of decision is, in a very real sense, no decision at all. To be forced into choosing any single option from a pool of only bad options is to have one’s autonomy effectively denied. Notice that not much about the situation changes when one is free to choose the only good option from a pool of otherwise bad options. That option essentially becomes the only option. To choose it is not exactly to make an autonomous decision. Just how many good options must exist for the whole pool of options to contain a “sufficient number” of them is unclear, but certain kinds of practical considerations can help to bring an answer into focus.

On Schouten’s view, one such consideration is the fact that there are certain valuable and in fact widely valued ways of life that people cannot enact because they are not genuinely available options (Reference SchoutenSchouten 2019). She argues that if a way of life is valued widely enough or, perhaps, is simply a genuinely valuable one, and is also widely foreclosed as an option, a person’s autonomy is thereby thwarted or denied. That alone is bad from a democratic perspective given that autonomy is essential to democracy, but when one considers the potential social consequences of a widespread denial of autonomy, the threat to social stability becomes clear (198–227). The upshot of Schouten’s argument is that government interventions to promote the enactment of widely valued (and perhaps genuinely valuable) ways of life are warranted on purely democratic, autonomy promoting grounds, and that subsidization is one kind of intervention that might be warranted (Reference SchoutenSchouten 2019, 170–227).

Notice that, according to this argument, substantive interests based on social difference may be the proper locus of concern for policy workers. Reference SchoutenSchouten’s (2019) argument is, after all, about the gendered division of labor. She defends the claim that government action to promote gender egalitarianism in the home is permissible on grounds that are neutral with respect to substantive values. Certain ways of life are systematically prohibitively costly for members of different minority groups. And though not all valuable or widely valued ways of life must be available at equal cost, they do have to be within reach to be genuinely available options (Reference SchoutenSchouten 2019, 170–197). Seeing a way of life as a live option depends on seeing it enacted by people like you, but when it becomes apparent that a particularly valuable and widely valued way of life is out of reach for a large contingent of the population, it threatens social harmony which, in turn, constitutes a threat to social stability. Since all of these (social stability, the ability to choose and to pursue one’s own conception of the good, governmental support for and reinforcement of value pluralism in society, and so forth) are primary democratic values, this kind of scenario ultimately licenses certain social-equity-based pursuits – things like affirmative action policies, which bring certain social forms into reach for members of groups for whom they would otherwise be out of reach on autonomy promoting grounds, and without the need for governance structures that objectionably undercut democratic values by directing policy workers to make decisions about what constitutes an equitable solution at their own, unguided discretion. The difference is in the level of governance at which it is appropriate to attend to social equity concerns like access, opportunity, and equitable social outcomes.

Schouten offers an independent line of reasoning that supports our conclusion that the appropriate level for deciding on legitimate aims and pursuits of social equity in a democracy is the level of institutional design, not the level of policy work, by grounding the warrant for governmental interventions on issues of equity in political values (e.g., democratic values) as opposed to in the substantive value sets of individual policy workers. The public supplies the input as to what kinds of options ought to be available, and institutional designers determine how to make that happen. Public-facing policy workers ought to carry out their discretionary duties within the bounds that have been set and according to the determinations that have been made in the proper, substantive-value-neutral way.

The lynchpin for the democratic legitimacy of pursuits of social equity is the democratic value of pluralism. Democracies value pluralism because they value and respect a genuine diversity of values, opinions, and inputs, all of which are necessary for genuine democratic rule. They could not genuinely value and respect these things if they did not also genuinely value and respect the diversity of persons, personal histories, and other background conditions (socio-economic, cultural, racial, and so forth) that give rise to those values, opinions, and inputs. People must be able to determine their own interests, and those interests need to be taken into consideration in order for a government to be genuinely democratic. Thus, it is in the government’s interest to preserve, perhaps even to promote, a diverse, autonomous demos. Insofar as pursuits of social equity in public administration contribute to that aim, they are democratically licensed, but since they are a part of public policy, they must also be specified, circumscribed, and initiated by democratic mechanisms.

4.4 Representative Bureaucracy and the Problem of Levels

Recall that advocates of both passive and passive-to-active representation as tools for enhancing social equity tend to assume that people of a particular social identity group have higher stakes in the members of that identity group’s outcomes than do nonmembers. It is therefore imperative to design agencies such that they are largely composed of in-group identifiers and often, according to some theorists, such that the selected representatives that make up the organization are tasked with policymaking that affects in-group identifiers (Reference FungFung 2003). This, they argue, frequently makes for better policy outcomes for in-group identifiers than would any other governance structure design. But even if it does make for better policy outcomes, it defies the democratic value of deference to the collective will, which constitutes an irresponsible execution of public policy work. As Reference LafontLafont (2020) would observe, it “shortcuts” democracy. The representatives making policy in such agencies are not elected representatives, but appointed ones. They are neither bound by nor accountable to the will of the electorate (of which they, presumably, are just one part), and their independent choices are not necessarily bound to improve outcomes.

Aiming for fidelitous representation through targeted policies like affirmative action, by contrast, builds the relevant stakes into what it means for the agency to function properly, all the while maintaining both accountability and deference to the collective will. Regardless of the demographic makeup of the representative agency, the desired outcomes determined according to the collective will constitute the policy goal, and if the desired outcomes are meant to counteract systemic disadvantages associated with a particular social identity group, the result should be comparably advantageous for that group. Does this way of grounding group-differentiated policy goals proliferate the possibilities for legitimate government interventions that promote substantively valuable and contested ways of life? Yes, but so what? If such interventions are licensed by the democratic values of the state and the polity the agency serves – as we have argued they should be – it is possible to attend to the kinds of concerns championed by advocates of social equity without engaging in irresponsible policy work or violating the complementarity principle.

Value-congruity distinguishes the actions of a fidelitous representative enacting, or working together with other representatives to enact, a policy that addresses such concerns at the policy level, within the parameters permitted by law and democratic principles, from the actions of an administrator in the Department of Motor Vehicles deciding to adjudicate between equally qualified applicants for a job on the basis of membership in a given identity group. The law and the principles of a representative democracy collectively circumscribe which possible means to the desired ends count as legitimate and which do not. When those constraints are ignored in the interest of achieving “good” outcomes by any means necessary, the result is a kind of policy work that is potentially corrosive to the system it aims to serve, rather than one that is complementary to its values.

Social equity has been seen as one of the “pillars” of public administration (Reference FredericksonFrederickson 1980). We contend that it can only be so inasmuch as it is built into the governance structures in which policy work is done, and this means that it must confront the problem of levels. When managers have the authority to hire a diverse workforce, reasonable pluralism suggests they ought to do so. The manner and extent to which that diverse workforce should improve outcomes for previously underrepresented groups depends on the authority they have qua policy workers to make such improvements, not on their own decisions about the substance of those improvements. Active representation through policy work must resolve the problem of levels by gaining the endorsement of representatives. Bureaucrats seeking to achieve active representation must offer proposals to both representatives and the people for governance structures and administrative procedures that embody their understanding of social equity. If they are adopted, these ideas can be legitimately revolutionary as they can bring valuable ways of life into the grasp of more people, enhancing pluralism through the fidelitous representation of elected representatives, who can judge their connection to the wants of the people. However, it is crucial to recognize that in a democracy, the people make the choices about options for the good life, not the policy workers. Representatives must continue to represent with fidelity when designing how to make all options possible, and they must face accountability for their choices about the means, not just the ends of policy work. An illegitimate revolution occurs when policy workers take autonomy away from the people, making and implementing the options themselves without the constraints imposed by a requirement of fidelitous representation and violating the rule of law requirement of responsible policy work. Should a representative bureaucracy encourage this, it would seek to do something undemocratic, while adhering to all the conditions of responsible action we have discussed will keep that from happening.

5.1 Cool v. Hot Deliberation in Representative Agency

Reference AnsellAnsell (2011, 168) describes a system of “collaborative governance” in terms of “fruitful conflict,” that is, “conflict that in some manner enhances or advances knowledge, understanding, meaning, or capacity between different or opposing perspectives and interests.” The examples that he marshals involve the creation of temporary “public agencies” (representative agencies), which draw participants from local publics, and which facilitate a deliberative process between parties with conflicting interests to arrive at a better understanding of the stakes at issue in the circumstances. The goal is not necessarily consensus, nor is it to hash out a winning and losing policy position. Rather, the goal is for the party entrusted by the public with the power to devise and to execute public policy on a particular issue, be it a local representative or the manager of an independent agency, to acquire more fine-grained information in order to better address the issue.

Within his representative agencies, Reference AnsellAnsell (2011) advocates for cool deliberation, in which deliberators have strong opposing opinions, responsive debate (designed to come to an agreement or an understanding rather than to win, all-or-nothing), and low stakes: The outcomes of these representative (“public”) agencies are evaluable decisions to be used as input for policy-making decisions at the appropriate level. Reference Klijn and SkelcherKlijn and Skelcher (2007, 588) see this as the sort of model that emphasizes pluralism, consisting, as it does, of “horizontal interdependencies through which actors steer the development of policy and its implementation.” They contrast it with networks that are “centres of power and privilege that give structural advantage to particular private interests” because of their “strong managerial character …, their incorporation of strategically powerful actors, and the opacity of their formal rules and constitutional position.” Readers may be tempted to think of these latter networks only as “iron triangles” or agencies captured by interests (cf. Reference JordanJordan 1981; Makkai and Braithwaite 1991), but we think that they (ironically) resemble a form of representative agency as well.

The power center conjured by Klijn and Skelcher might be created by a form of representative agency that Reference FungFung (2003, 345–346) advocates in certain circumstances. He argues for a mini-public characterized by hot deliberation, in which deliberators have strong opposing opinions, adversarial debate (designed to convince rather than to compromise), and real stakes – the outcomes of these representative agencies are to be enacted. They essentially produce policy. Reference FungFung (2003; Reference Fung2006) does not take this always to be the best form of deliberation, but he does suggest that, often, it enhances democracy by offering a more representative, deliberative policy-making process, and this by virtue of the fact that it actively involves people from the community in the business of setting local policy.Footnote ²¹

It is important to recognize that Fung himself remains fairly neutral among public participation schemes. His position with respect to the circumstances in which various levels of public involvement in public policymaking might be appropriate is nuanced (Reference FungFung 2003). It seems that his intention lies more in providing a useful taxonomy of participatory structures, which are more and less commonly used in attempts at more and less collaborative governance, than in making substantive claims about the general desirability of one structure or another (Reference FungFung 2006). If we had to describe his substantive position on the matter, we think it is that different circumstances warrant different levels of public involvement. Fung does, however, tend toward championing greater levels of “empowered” public involvement, or cogovernance, especially in circumstances in which “a democratic deficit manifests itself as a lack of state accountability or when the minipublic is a component of a governance or problem-solving scheme” (Reference FungFung 2003, 346). Careful as he is to attenuate his enthusiasm for such schemes in circumstances where public involvement might produce less than desirable results or constitute a detriment rather than an enhancement to democratic processes, the practitioners following Fung’s lead might not be so careful. And what he describes as cogovernance enjoys broad support in public administration (e.g., Reference Bryson, Quick, Slotterback and CrosbyBryson et al. 2013; Reference NabatchiNabatchi 2012).

Our contention is not that these kinds of governance structures are always a bad idea, but rather that the complementarity principle can be useful to those designing them. Specifically, it is instructive for figuring out when a participatory structure is appropriate and how to design one that promotes responsible policy work. Adherence to the complementarity principle guards against representative agencies that are incompatible with representative democracy, a result which occurs when each (the agency and the democracy in which that agency operates) finds itself operating under a different set of institutional rules (Reference Klijn and SkelcherKlijn and Skelcher 2007, 590). Ansell’s coolly deliberating advisors, with which we contrast Fung’s hotly deliberating policymakers, seems to implicitly follow the complementarity principle, giving it a democratically constrained quality that certain implementations of Fung’s model might lack.

Ansell’s collaborative governance model adheres to the complementarity principle because it builds the core values of representative democracy into the function of the representative agency. The board, in its advisory role, gives the representative who formed it the information she needs to represent the entire constituency with fidelity by initiating a process designed to draw out the competing concerns and interests on a given public matter as expressed by actual members of the relevant constituency. But, unlike Fung’s model of cogovernance directed by engaged and empowered mini-publics, it does this without granting actual policymaking power to the mini-public, thereby (perhaps unwittingly) skirting responsibility or accountability for the policies it makes. Ansell’s collaborative governance process does not bypass the collective will; instead, it brings the collective will into greater relief, leaving the representative entrusted with its execution in full control of the relevant policymaking. Note that this process gives the representative the information she needs to determine what a socially equitable outcome looks like in a particular domain, so that she can set up or direct the policy workers on the ground to execute their functions accordingly.

The People’s Maps CommissionFootnote ²² organized by Governor Tony Evers of Wisconsin is an example of cool deliberation in action. The commission met with constituents all over the state, gathered information on demographics, political affiliations, and all manner of other relevant considerations in an attempt to draw up the most equitable, nonpartisan redistricting proposal possible. Upon completion of fact-finding and deliberations, they submitted to the Governor’s office what they took to be their best proposals. The office then examined these proposals and chose which one it would submit for approval to the state legislature. This process incorporates the general public in a way that satisfies Reference FungFung’s (2003; Reference Fung2006) intuitions about more and less democratic or representative processes in matters of local concern. The Commission also appears to better democratize a process that can and often does result in a serious lack of fairness when executed by people with partisan interests, but it does this in a way that does not delegate powers away from elected officials. The Governor takes the nonpartisan recommendations, decides which among them to submit, and the legislature votes. Absent any evidence of partisan motivations or manipulations on the part of the commission’s members or those choosing among their proposals in the Governor’s office, the legislature ought to approve the selected map (or at least one among them), since to do so would be to represent the people of Wisconsin with fidelity.

Were it to follow a cogovernance model, the process would be quite different. The People’s Maps Commission would simply deliberate and select the map, present that decision to the Governor’s office, and in so doing, trigger the relevant redistricting processes to bring the state’s districts into conformity with the chosen map. Why might this seem preferable? Just look at what, in fact, happened with this process in Wisconsin. The Governor’s office selected the map it deemed appropriate from those submitted by the commission and the map was not only rejected by the state legislature, but the Commission’s process was challenged in the Wisconsin Supreme Court. The Governor was obliged to submit instead the map with the least changes as the battle continued to work its way through the courts, and it seems unlikely at this point that the People’s Maps Commission will have been able to produce a fair, nonpartisan districting map for the state. The better outcome, one might reasonably insist, is the one in which the map’s fairness and accuracy is determined by the nonpartisan commission selected to do the relevant research and map drawing, and that map is imposed on the lawmakers in accordance with the democratic value that “the people ought to choose their representatives, not the other way around” (The People’s Maps Commission 2021).

Reference FungFung (2003, 345) tends toward hot deliberations with high stakes outcomes like the political geography with which the People’s Maps Commission is concerned. He argues that hot deliberation in these cases often results in better outcomes for the people. We argue that those outcomes, even if they are “better” than or preferable to others, are not necessarily more democratic, bypassing as they do the collective will as expressed by the voters in electing the representatives they did. The People’s Maps Commission case is especially interesting because it represents something of a Catch 22. The representatives in question were chosen based on the very maps that the People’s Maps Commission suggests are unfairly stacked in favor of one party over others, which is unfair because it is not representative – according to their data – of the partisan breakdown of the state. This state of affairs makes Fung’s model look very attractive indeed, but there must be a better way of alleviating the trouble with these maps than simply putting it to the Commission to replace the existing ones. Why? Because when processes make an end-run around the agreed-upon channels for democratic decision-making, they defy or undermine certain democratic values that the whole redistricting process is supposed to reinforce.

Both Ansell and Fung situate their projects within a system of representative democracy. Because one policy proposal must defeat all others, because people have registered differing opinions about which proposal should be chosen, and because representatives are obligated to take pluralism seriously, the responsive rather than the adversarial model of representation is that which reinforces the political values of the system. In this case, those values are both pluralism and public accountability.

Ansell’s responsive, collaborative governance model is designed to reinforce the values of pluralism and the accountability of representatives to the people for two reasons. It takes the voters’ will as a starting point for directing public policy, and it uses the recommendations of a group of citizens as more fine-grained data about how best to implement the will of the voters. It facilitates a recursive process of policy goal refinement and explores different implementation options, and all the while, the representative remains accountable to the voters. The model accords with the conjecture that representative agencies can “engage a wider range of actors in the policy process, connecting them in new ways, and thus ‘oil the wheels’ of representative democracy as it struggles to govern in a complex environment” (Reference Klijn and SkelcherKlijn and Skelcher 2007, 594). This is essentially a version of what Reference FungFung (2003, 341) calls a participatory advisory panel.

Fung’s “empowered” version of representative agency threatens to undercut rather than enhance the value of pluralism by circumscribing a new, narrower “people” – some subset of the voters whose popular will elected the representative – to whom the representative then becomes accountable, thus obviating the value of public accountability. It takes the voters’ will as a starting point for directing public policy, then sets them aside, allowing the policy implementation decisions to be made by a smaller group of individuals who are selected from the population at large. The representative, on this model, is bound to execute what a mini-public decides, and since the members of the representative agency are not elected, no one is accountable for the outcomes of the policy to the voters. Essentially, the will of the representative agency supplants the will of the broader constituency even if it takes the will of the broader constituency under advisement. Importantly, though, it need not.

According to the criteria specified in Section 2.1, responsible policy work means that wherever possible, policy workers must reinforce the values of the political system of which they are a part. Representative agency structures like those championed by Fung and others in this camp do not necessitate this, and this is because these structures fail to embed complementarity.Footnote ²³ They do not take seriously one of the major positive obligations of responsible policy work (the capability component), instead directing policy workers to engage in actions that might well run counter to that obligation.

This further illuminates the values that undergird systems of representative government. Responsiveness is necessary for the genuine preservation of pluralism under a representative system. The adversarial model tends to obviate pluralism as a value altogether. Consider the aggressive agendas of straight opposition to the majority party over the past twenty-five years or so in the US Congress. The adversarial stance gives lawmakers strong reasons to decline to represent their constituencies with fidelity, rather, it gives them positive reasons to represent only their own voters – that is, only those who actually voted for them. More insidious still is the fact that once one side has chosen the adversarial model, pluralism is lost. This is so because such an adversarial model is an exercise in anti-democratic, factional people-making. It undermines the democratic value of equal voice – having an equal opportunity to influence public policy.

To insist that policies made by mini-publics (constituted of members of the public and which are, in fact, a subset of the public) rather than by representatives is to confuse subordinate values – things like transparency, efficiency, and equity that are valuable insofar as they help to achieve higher, overarching goals – for bedrock democratic values like equal voice, maximal public participation, collective decision-making, respect for reasonable disagreement, fidelitous representation, and so forth. Proponents of this kind of model make an inductive inference from the actual outcomes of specific initiatives to the general desirability of the mechanisms that yielded those results. They use inductive normative reasoning to infer a general claim from an existential proposition: Socrates is a man; Socrates is bald; therefore, all men are bald. What “works” or what produces desirable results in a given context does not necessarily improve or enhance democracy – for that matter, what “works” may not turn out to be very democratic at all. It is one thing to claim that deliberative methods (in Reference FungFung’s [2003] case, using selectively chosen mini-publics in hot deliberations to produce actual policy outcomes) yield better results; it is quite another to claim that without adhering to complementarity they constitute an improvement to or enhancement of the democratic process of policymaking.

5.2 What Makes for a Better Policy?

One final question raised by the foregoing discussion is this: Who decides whether one policy is “better” than another? Pluralism – maintaining the ability of the people to choose and to pursue different reasonable conceptions of the good – is a core value of representative democracy. Because people living in representative democratic political systems have, and are permitted to have, different views regarding what is good or bad, right or wrong both generally and for themselves, differing views about the desirability of public policies (either in general or for someone or some group) are an inevitable consequence of the sociopolitical landscape. Taking questions of public policy to the demos as a whole respects pluralism by giving everyone the chance to weigh in on them. Empowering selectively chosen mini-publics to set public policy does the opposite. It makes it possible, for instance, for a select group with certain strong interests to enact policies that run counter to other people’s strong interests, thus eroding the value of pluralism. Perhaps it all seems well and good that something like that could happen so long as the decision goes in the way one would like it to go, but that will not always be the case.Footnote ²⁴

Making an end-run around established law-enforcement procedures and standard public accountability measures in order to achieve what many consider a “better” outcome is exactly what the recent Texas anti-abortion law aimed to do (before Dobbs v. Jackson Women’s Health Organization overturned the precedent it was designed to evade).Footnote ²⁵ It aligned with certain deeply held beliefs of a large number of Texans, and so constituted, in at least one widespread opinion, a better outcome than having enacted a law that comports with democratic values (like that of pluralism and basic autonomy) and procedures (like the state’s enforcement of its own laws). Moreover, this law purportedly democratized the enforcement of law by putting it in the hands of the people. Is that a “better” outcome? And is it – by virtue of putting public decision-making and public law enforcement power in the hands of the people – in fact more democratic or democracy-enhancing? It certainly is not if you consider pluralism to be a core democratic value. It is an instance of minority rule. Designing governance structures that delegate policymaking authority to selected subsets of the public undercuts or erodes pluralism. That is, in failing to adhere to the complementarity principle by designing governance structures with the express goal of reinforcing core democratic values, the Texas anti-abortion law fails to promote or to facilitate responsible policy work. Adopting such governance structures when they are likely to promote “good” outcomes according to your own sensibilities opens the door to legitimizing all such governance structures, whether they get “good” outcomes or not. Hence, our argument focuses on “good” (that is, responsible) policy work rather than on outcomes simpliciter. Whatever else one can say about the Texas anti-abortion law, it certainly did not seem to encourage or facilitate responsible policy work.

The solution to the problem of roles and the problem of levels is, at least in part, one and the same: adherence to the complementarity principle in governance structure design. This is why Reference BertelliBertelli (2021, 197) and we connect the criteria of responsibility to the complementarity principle. Part of what determines a “good” or “better” outcome is the aggregate will of the public, or how well the outcomes and the policies and procedures that produce them adhere to the values of the representative democracy of which they are a part.