Although the last half century has seen many experiments randomly selecting participants for deliberative bodies that advise duly elected or appointed officials, we have little experience in giving bodies of this sort direct power to legislate rather than just advise. Prudence therefore suggests that any steps in this direction be small, tentative, and easily revocable. Prudence also suggests that before beginning our first experiments we try to think through as best we can what problems might arise so that as we take our first tentative steps forward, we can be particularly alert to the potential problems we identified in advance. One of these is accountability.
Many commentators in this volume have pointed out that representatives chosen by lot (or “sortition”) do not have the kinds of accountability to their constituents that representatives chosen by election have, at least in theory and very often in practice. This problem of accountability is an important feature of the larger constituent-representative relationship. I will discuss it primarily within the context of that relationship.
Stepping back, the larger context of this question is our growing need, in a world of increasing interdependence, for state coercion to handle the free-rider problems that almost inevitably accompany such interdependence. (The reader should feel free to substitute “regulation” for my preferred term, “state coercion.”) As we collectively try to produce “free-use” goods, such as lower crime rates, efficient financial transactions, and a stable climate, we face the problem that the coercion (regulation) needed to produce these goods must be as legitimate as possible. This is true both because that legitimacy is good in itself (that is, “normative legitimacy,” or legitimacy that can be supported with good reasons) and because coercion perceived to be legitimate is more effective than coercion perceived as illegitimate. Yet as our needs for legitimate coercion have increased, our capacity to legitimate that coercion has decreased. Demand for legitimacy is up; supply is down.1
As human beings, we have evolved the capacity to govern ourselves in ever-larger entities, currently up to the level of the nation-state but now tentatively extending beyond it. In the last few hundred years, we have come to rely heavily on the mechanism of representative elections to produce state coercion that, when the system works well (as it does, for example, in Denmark), is both perceived as legitimate and has a good claim to legitimacy on normative grounds. But partly because of the nature of incentives and barriers to becoming an electoral candidate and partly because of other details of specific competitive electoral processes, these systems do not always produce an assembly that the citizens of a country feel speaks for them. In these circumstances the regulations the assembly promulgates have correspondingly less legitimacy.
Over the last half century, the degree to which the citizenry thinks that the elected legislature speaks for them has declined in many countries. Over that same half century, the absence of major wars in Europe and the Americas also may have eroded the sense of urgent common purpose that enhances the legitimacy of the citizen-legislator bond. For these and other reasons, the gap between the amount of legitimacy we need to regulate our growing interdependence and the amount of legitimacy we can generate with our current institutions has grown. The attempted solutions to this problem include a growing number of advisory bodies based on lot and even serious consideration of legislative chambers based on lot.
Combining two moves, from an advisory body to an empowered lawmaking one and from a temporary body to a more long-standing one, turns the citizen participants in any deliberative body drawn by lot into formal representatives of the full citizenry. Ideally, what should this constituent-representative relationship be? Equally important, what is it likely to be? In comparison to current electoral representation, what is lost in the constituent-representative relationship, and what is gained? The answers to these questions are not currently clear. Both the normative theory and the practical implications are likely to evolve with experimentation in practice.
Here, I will compare conventional electoral representation to representation by lot only in regard to constituent-representative relationships, ignoring other processes and effects. In these relationships, it seems that the most significant losses might come in the constituents’ agency and their capacity to inflict formal sanctions on their representatives, whereas the most significant gains might come in the constituents’ greater identification (and possible communication) with their representatives. We should also expect changes in the character of accountability—both in the forms of sanction and in the particular ways representatives give their constituents an account of what they did, what they plan to do, and why.
The Loss of Agency and of Sanction
In a functioning system of electoral representation, citizens exercise power when they cast their votes. The effect on the outcome may be very small and is hardly ever decisive, but in many elections the margin of victory also influences the future. Citizens are often aware of the power in their vote. Activists chant, “We’ll remember in November!” Ordinary citizens may feel some pride not only in fulfilling their civic duty or expressing their views at the ballot box, but also in contributing their small bit toward the outcome.
I would argue against interpreting the act of casting a vote in an election as “consent.”2 I do, however, see voting as a regular, repeatable, self-constitutive act of agency—one that helps make an individual into a citizen. The equal weight of each vote also makes a powerful normative statement about the status of the citizen, even when undercut by surrounding inequalities in political power. If a polity’s legislature were composed entirely of representatives selected by lot, this small but meaningful power in the vote and the agency experienced in exercising it would be lost.
The constituent’s power exists not only in the moment of the vote and immediately afterward, when the constituent’s vote helps to change or maintain the composition of the legislature. In addition, the constituent’s potential vote in the next election changes the incentives of the representatives in the legislature, because elected officials anticipate the next vote and try to please future voters. If a polity’s legislature were composed entirely of representatives selected by lot, that power through anticipation would be lost.
In a legislative system based on electoral representation, the constituent-representative relationship is informed to some degree by this constituent power and by both representatives’ and constituents’ awareness of that power. Citizens and their representatives have a deeply unequal relationship, because the representative’s vote can make or unmake the law, and their expertise in the matters under consideration quickly surpasses that of most constituents. Nonetheless, the constituent’s vote and the representative’s recognition of the power inherent in that vote give some dignity to the constituent and some hesitation to the representative. Although the effect is far less strong—and perhaps nonexistent—for permanent minorities in a district, for those who are often in the majority, a potential majority, or a potential winning coalition, having a vote emboldens the constituent and commands the representative’s attention. For the constituent, the anticipatory power of voting in the next election plays a crucial role in the relationship. As constituents see events unfolding in the legislature with which they disagree, they can nurse their own anticipation of the next election. They do not have to stand by helpless, without recourse. They can organize with allies to depose, or at least fight against, their elected representative in the next election. They are not powerless. With a legislature composed entirely of representatives selected by lot, this equalizing, agentic feature of the constituent-representative relationship would be lost.
For reasons anchored in this loss of constituent power and the healthy relationship between constituents and representatives that it fosters, I would oppose any legislative system based entirely on lot. Too much of value to democracy would be lost.
Accountability
A focus on sanction and on power, however, misses a significant part of the constituent-representative relationship. That focus, which I have just described, problematically lies at the core of most discussions of “accountability” today. Just as the use of the word accountability in English has increased dramatically since the late 1960s (and the use of transparency has increased dramatically since 1980),3 the meaning of the word accountability has also slowly changed. Today it primarily means the capacity to sanction (and to monitor with potential to sanction). Accordingly, when modern writers decry the loss of “accountability” in a legislature based on lot, they usually mean the loss of the constituents’ capacity to threaten the representative in the next election with the sanction of withdrawing their votes. A representative chosen by lot for a fixed term (and replaceable only by lot) is not subject to this threat.
Accountability did not always have this restrictive and punitive meaning. It meant giving an account—in French rendre compte and in German Rechenschaft abgeben. Being accountable meant having to describe, explain, and justify one’s actions to those to whom one is responsible. This more deliberative meaning of the word signals a constituent-representative relationship based not on threats of sanction but on mutual attempts at describing, explaining, justifying, and persuading.4 In this understanding, the representative’s responsibility to the other representatives, the public, and constituents may be fulfilled through explaining, perhaps ideally in a process of recursive two-way interaction.
How, then, could the members of a representative body chosen by lot be accountable to one another, to the larger public, and to their “constituents” in this more deliberative sense? This question has never been studied. It may never have been asked. Despite many past and current experiments with deliberative bodies chosen by lot, no one as far as I know has investigated the deliberators’ relations of accountability to one another or to members of the public who might be considered “constituents.”
Such an investigation would look at three levels: the accountability of the participants in the deliberative body to one another, their accountability to the public at large, and their accountability, if any, to specific “constituents” whose opinions, perspectives, and interests they represented. On all of these levels, it would help to distinguish a) sanction-based accountability versus deliberative accountability and b) formal versus informal accountability.
Table 9 shows the four forms of accountability created by combining these characteristics. In electoral representation, the accountability of representatives to their constituents falls in all four quadrants of the table. It is both sanction-based and deliberative, both formal and informal. In representation by lot, the formal sanction-based quadrant is relatively empty. In the other quadrants the appropriate and available forms of accountability vary according to who may rightly hold the representative accountable: the other representatives, the public at large, or specific constituents.
Table 9. Types of accountability in legislatures chosen by lot.
Level of formality |
|||
Accountability type |
Formal |
Informal |
|
Sanction-based |
Laws against bribery and other forms of wrongdoing. |
Representatives monitoring and sanctioning the norms of discourse within their own deliberations. Members of the public exerting informal pressures on the representatives. |
|
Deliberative |
Collective written accounts to fellow representatives or the public |
Representatives mutually listening, explaining, and justifying perspectives, opinions, and interests to one another. Representatives doing the same with the larger public or specific constituents, face-to-face or through the media. |
|
Accountability to One Another
In representation by lot, the accountability of the representatives to one another will be largely informal and deliberative. Of course they will have some formal and informal sanctions against one another, and presumably their committees will produce some formal explanatory and justificatory documents in the course of accounting for their actions to the body of the whole. But we can expect the representatives in a legislature selected by lot to be accountable to one another primarily through their informal mutual explanations and justifications.
In the many modern experiments in which lay participants drawn by lot deliberate only to register their opinions at the end of the experiment, draw up an advisory statement, or at the most legislate on small and relatively “cool” issues, the participants almost always act toward one another with generosity, goodwill, and respect. As far as I know, their relationships of mutual accountability have never been studied. In every instance that I have attended, viewed on tape, or read about, the participants seemed to act as if they felt accountable to the others in the group; they tried to listen to what the others had to say, treat the others with respect, and back up their own views with as good—and as good-faith—justifications as they could muster. Occasionally, some of those general norms were spelled out formally, but most were informally generated and maintained. These norms seemed to arise primarily through the participants’ own sense of the appropriate relations of mutual accountability in the informal private sphere. A smaller part derived and was maintained within the specific deliberative setting through informal sanctions of social approval and disapproval wielded by the participants themselves, or occasionally by the facilitator backed by the participants.
It is impossible to predict what might happen in a developed legislature by lot in the heat of political passion on issues that present irreconcilable material and ideological conflicts. As several commentators in this volume have suggested, the underlying conflicts of interest and opinion, perhaps sharpened or even exaggerated by political parties, might well create far greater conditions of mutual hostility than in usual social interactions. Experimentation might suggest that such high-conflict issues be, through some mechanism such as a vote, tabled and sent to the electoral house. In those cases, informal mechanisms of mutual accountability might not be enough to sustain constructive deliberation. On the other hand, the extreme polarization present in the United States today is in part driven by the relatively equal competitive electoral prospects of the two political parties since 1980, which has produced a dynamic in which each party benefits from preventing the other from acting constructively, thereby undermining the other’s reputation.5 This dynamic would be weakened greatly in a legislature chosen by lot.
In discussions of accountability, both social scientists and the public often ignore the accountability of representatives to one another. Theorists of organizations, however, recognize the importance of “peer accountability”—the constellation of formal and informal communicative practices and behavioral rules generated by common confrontation with a problem, common experiences of past productive human interaction, and the combination of formal and informal sanctions against significant deviations from those rules. The accountability of members of the professions to one another, for example, is maintained primarily by ongoing mutual justificatory communication, resting on a core of informal commitments to duty and feelings of solidarity and sustained by a small periphery of formal and informal sanctions.6 University faculties also work largely through peer accountability, as do EU committees and the best instances of the civil service. Even the ways that in most democracies elected representatives are accountable to one another for decent behavior are maintained largely by informal mechanisms of peer accountability. In assemblies of representatives drawn by lot, such mechanisms of peer accountability must play a major role. Without a robust system of primarily informal and deliberative peer accountability, where the character of interaction is drawn largely from larger norms of good informal deliberation, legislation by lot will fail to deliver much of democratic value.
Accountability to the Public at Large
In the accountability relations of a representative body drawn by lot to the public at large, the most important form of accountability will almost certainly be formal and deliberative. We can expect formal statements from the assembly, giving an account of the reasoning behind their decisions, to be the rule and perhaps required by the legislation setting up the assembly. The statements of the Oregon Citizens’ Initiative Review panel, placed on the legal referendum ballot by law along with statements by advocates on both sides, exemplify the kind of formal deliberative accountability that is likely to legitimate (or if done badly, delegitimate) representation by lot. The British Columbia Citizens’ Assembly concluded with a long formal report explaining and justifying their decision to advocate the Single Transferable Vote electoral system. The US Supreme Court’s explanations and justifications of their decisions also exemplify this formal deliberative accountability.
Formal sanctions will not be entirely absent. Although the electoral sanction will be missing, it is unlikely that the public would ever authorize a legislature by lot to operate without formal legal sanctions for various forms of wrongdoing, including bribery. An appropriate oversight body, with judicial due process, would have to monitor the representatives’ actions sufficiently to make such sanctions viable. Transcripts of the representatives’ formal interactions, even if kept from the public to allow more frank deliberation, would have to be available not only to the participants themselves, to refresh their memories, but also to any oversight body. Various mechanisms would generate sanction-based accountability for wrongdoing. The most significant formal sanction would be the underlying capacity of the public to dismantle the legislature by lot and return to a purely electoral representative system.
Informal sanctions by the public will also have their place. Indeed, fear of such sanctions could provide a significant disincentive for participation in an assembly drawn by lot. In a small New England town meeting, some citizens do not attend the meeting for fear of being laughed at or criticized. This dynamic would be accentuated by participation in a legislature by lot on a state or national scale. Even with high pay and the strong incentives of duty, many potential participants are likely to decline to serve. The most marginal potential participants would probably be most vulnerable to informal sanctions from the public, thus exacerbating the other inequalities in participation that are likely to emerge.7
Informal deliberative accountability can be expected to play a relatively large role in the pattern of accountability relations between representatives selected by lot and the public at large. Spokespeople will emerge; the media will fasten on certain explanations and justifications, and some representatives will take upon themselves the task of describing, explaining, and justifying the decisions of the assembly and the thinking behind those decisions. In the British Columbia Citizens’ Assembly process, for example, after the conclusion of the assembly’s deliberations, some participants informally tried to explain and justify to the public at large the reasoning behind their assembly’s decision to advocate a Single Transferable Vote electoral system to replace the existing Single Member Plurality system.8 One problem for the representatives in a legislature chosen by lot will be the openness of this informal deliberation to the public’s informal sanctions. In the primary form of accountability in this process, the formal statement of explanation, the collective production of the statement to some degree protects each individual in the assembly. But in the informal processes of deliberative accountability, the participants are more individually vulnerable.
Accountability to Constituents
Accountability to “constituents”—if there are any constituents beyond the general public—is the most problematic and least well-considered aspect of the patterns of accountability we might expect or want to institute in a legislature chosen by lot. Unlike the jury, the legislature chosen by lot will not be able to promise its members confidentiality.
In the British Columbia Citizens’ Assembly (BCCA), the French High Council of Military Function (Conseil Supérieur de la Fonction Militaire, or CSFM), and the Oregon citizens’ Initiative Review, the members of the deliberative decision body drawn by lot were and are all known.9 The members of the BCCA and CSFM also had specific constituents, the BCCA from the various “ridings” or districts in British Columbia and the CSFM from their branches of the armed forces.
Beyond geographic or other formal constituencies, we would also very probably find “constituencies” for each member developing around common interests and descriptive characteristics. Thus, even if the representatives did not have specific formal constituents, or take any actions to create informal constituencies around specific issues, certain groups of citizens would probably constitute themselves informally as the constituents of specific representatives, coalescing around those representatives’ expressed positions, their descriptive features, or even their perceived vulnerabilities to pressure or persuasion. Opponents of the actions of specific representatives would also be likely to coalesce, although we would not call such opponents “constituents.”
If the representatives were chosen by lot from geographic districts, as was the case in the BCCA, it is unclear normatively what forms of formal and informal deliberative accountability the representatives ought to have with those constituents. In the CSFM, each member chosen by lot reports in meetings open to all members of the branch of the military from which he or she was chosen. The lively discussions at these meetings serve as a significant mode of formal deliberative accountability to these constituents. In the BCCA, no such interaction with constituents was formally mandated, but some members took it upon themselves to make speeches and otherwise contact constituents in their ridings to explain the assembly’s decisions and try to persuade their listeners to vote for the assembly’s recommendation in the upcoming referendum. This mode of deliberative accountability was, however, informal and voluntary; many BCCA members did not engage in this kind of outreach.
Mandating formal deliberative accountability of the sort practiced in the CSFM would probably, in the United States, put considerable strain on the 40 percent of the public who say they fear speaking in public in front of an audience. Those with only high school degrees are almost twice as likely to have those fears as those with college educations.10 In electoral representation, by contrast, those who run for office are highly skewed toward the rich and well educated. They also self-select—and are selected for—interest in politics and the ability to communicate with constituents, in addition, one would expect, to relatively low fears of speaking in public.
Informal deliberative accountability might, however, be enhanced in representation by lot. In the United States, African American constituents are more likely to contact their elected representatives when those representatives are African American themselves.11 Constituents across a range of ethnicities and classes may be more likely to contact, communicate with, and identify with representatives who are descriptively closer to them than the current predominantly white, male, rich elected representatives. Representation by lot from geographic districts would be likely to produce greater feelings of identification and therefore contact than representation by lot from the state or nation at large, and the range of descriptive characteristics among representatives drawn by lot from a geographic district would presumably be greater than the range among current elected representatives.
If the assembly had no geographic districts and the participants were simply drawn from the state or nation as a whole, the kinds of “constituencies” described earlier might nevertheless form. Even in elected legislatures, advocacy and descriptive representation play a role in creating surrogate constituencies outside a representative’s district. For instance, as an advocate of consumer rights, Massachusetts senator Elizabeth Warren has a constituency in this substantive field outside Massachusetts. The openly gay Massachusetts congressional representative Barney Frank had a national constituency concerned with securing LGBTQ rights. The African American congressional representative Mickey Leland was a surrogate representative for African Americans in “the entire Southwest.”12 The same dynamic would appear in a legislature selected by lot. Interest groups might also constitute themselves the “constituencies” of sympathetic representatives, providing them information and expecting support.
Conclusion
Even in the field of electoral representation, few empirical political scientists or normative theorists have interested themselves in the quality of the constituent-representative relationship.13 Perhaps for the same reasons, among those studying representation by lot, few have engaged the question of constituent-representative relations in general. Although many have worried about the loss of sanction-based accountability, no one, to my knowledge, has addressed the potential losses or gains in deliberative accountability. I have attempted to begin such an analysis, including both formal and informal sanction-based accountability and formal and informal deliberative accountability. Among the representatives themselves, one would expect primarily informal deliberative accountability to predominate. In the relations of the representatives with the public at large, one would expect primarily formal deliberative accountability, through written statements of explanation and justification. The relations of the representatives with their constituents, whether geographic or self-selected, are currently an open question, both normatively and in practice. With the power of the vote removed, in all cases the level of formal sanction would be low. The level of informal sanction seems, at the moment, unpredictable.
As institutions embodying representation by lot evolve, processes of trial and error are likely to illuminate both problems and potential solutions in the realm of constituent-representative relations as in all other realms. A problem arises in finding testing grounds where the decisions are important enough to generate the dynamics one might expect in a legislature by lot but unimportant enough not to be catastrophic if unexpected dynamics arose to torpedo the experiment. One testing ground of the potential for legislation by lot might be the self-governing boards of tenants’ associations.14 Another could be high school student governments. In both cases, an association or school could institute an elected government one year and a government chosen by lot another year, allowing comparison of internal dynamics, outcomes, and constituent-representative relations. In a high school, however, the lack of issue seriousness might combine with the playful tendencies of that age group to produce in an assembly drawn by lot results that were entertaining for the participants but not useful as stand-ins for what might take place in a state or national legislature. In conducting such experiments one would ideally want to measure and compare not only the quality of the outcomes but also the degree to which constituents felt adequately represented by the two groups, the quality of constituent-representative communication, and the forms taken by both sanction-based and deliberative accountability.
As thinking about representation by lot advances in tandem with the practice, I believe it would be a mistake to assume that empowered assemblies are automatically either better for the participants or more democratic than advisory ones. We, the people, should choose for ourselves those institutions that, after consideration, we conclude can represent us best in various ways. Historically, most peoples have chosen—either explicitly at a founding moment or more subtly by accretion—systems that mixed different forms of representation. The time is probably ripe for adding more institutions chosen by lot into the mix. As we add those institutions, however, we should pay close attention to their many characteristics, including the currently neglected characteristics of deliberative accountability and constituent-representative relations.