4

Random Assemblies
for Lawmaking:
Prospects and Limits¹

James S. Fishkin

After more than two decades of successful experimentation with deliberating microcosms or minipublics, it is natural to raise the question, how might they be institutionalized in the lawmaking process?² One key focus is the possibility of some sort of second chamber or legislative body. From the standpoint of deliberative democracy, a second chamber could provide an additional venue for deliberation about legislative proposals. This function might occur before or after the operation of the first chamber. The second chamber might have limited authority to delay a proposal for reconsideration, as in the British House of Lords, or it might be a coequal part of the legislative branch, such as the US Senate, with slightly different responsibilities in certain issue domains. Would it be useful to select the members of a second chamber by lot (random sampling) of the citizenry?

I will organize these reflections as follows:

1. I will argue for a deliberative body of randomly selected citizens as a supplement to our lawmaking process. This deliberative body might connect with the legislative process among elected representatives, or it might connect with an alternative lawmaking process such as direct democracy. In either case, it could have an institutional role in the lawmaking process.

2. I will offer skeptical considerations. There are issues to be confronted about the challenges facing citizen deliberation, particularly for a second (or third) chamber that would meet on a continuing basis and look much like a legislature.

3. I will argue that instead of a full-fledged legislature selected by lot, there are advantages to deliberative convenings for a short period on the model of a deliberative poll. These convenings might better serve the desired functions for a second chamber, particularly if their role were institutionalized. I believe the short convenings would best respond to the objections in number 2 above while also delivering a deliberative input into lawmaking.

4. I will draw inspiration from institutions in ancient Athens, particularly in the fourth century BC, to suggest appropriate institutional designs. Some of these designs will precede the key lawmaking process, and some of them will follow it. I will refer to the former as pre-filter designs and the latter as post-filter designs. In either case, the deliberative body does not do the actual lawmaking, but rather it filters the agenda (in the pre-filter cases), or it provides final approval (in the post-filter cases).

5. I will sketch three scenarios, two of them pre-filter and one post-filter. These three scenarios do not begin to exhaust the range of institutional designs, but they illustrate what might be done.

Four Forms of Democracy: Finding a Place for Deliberative Lawmaking

There are many notions of democracy, but in my view they boil down to a few competing democratic principles and how they combine to form what we will call four forms of democracy, four conceptions of democratic practice. In actual institutional designs, none of these forms of democracy are self-sufficient. They coexist and connect with other institutions, usually exemplifying other forms of democratic practice (as well as other modes of decision-making, such as judicial decisions or administrative processes). But we will focus here on explicitly democratic elements and on how these four forms of democracy can work alone or in combination to provide a satisfactory picture of rule by the people. A focus on competitive democracy and its limits clarifies the question of how it might be supplemented—and what forms of democracy can be engaged to respond to its limitations.

There are so many kinds of democracy; how can we get a handle on their variety?3 It is useful to think of some core component principles—political equality, (mass) participation, deliberation, and avoiding tyranny of the majority (which I call non-tyranny). Three of these principles are internal to the design of democratic institutions, and one (non-tyranny) is about the effects of democratic decision, effects that have long worried critics of democracy. If we consider these four principles essential components of a democratic theory, then the variations in commitment to them provide a kind of rudimentary grammar that allows us to specify the range of alternative normative theories of democratic practice. In other words, we can get a handle on different forms of democracy according to whether or not they accept or reject these component principles. Table 4 shows that four prominent forms of democracy (described below) each commit to two of the principles just mentioned (indicated by a “+”), yet each remains agnostic to the other principles (as shown with a “?”).

Table 4. A comparison of the democratic principles foregrounded by four forms of democracy.

	Form of democracy
	Competitive	Elite	Participatory	Deliberative
Political equality	+	?	?	+
Participation	?	?	+	?
Deliberation	?	+	?	+
Non-tyranny	+	+	?	?

Note: “+” indicates commitment to a democratic principle, whereas “?” indicates a more agnostic stance.

By political equality I mean, roughly, the equal consideration of one’s views as these would be counted in an index of voting power. Does the design of a decision process give each person a theoretically equal chance of being the decisive voter? Or, to take an obvious example, do voters in Rhode Island have far more voting power than voters in New York in selecting members of the Senate? By participation I mean actions by voters or ordinary citizens intended to influence politics or policy or to influence the dialogue about them. By deliberation I mean, roughly, the weighing of reasons under good conditions in shared discussion about what should be done. The good conditions specify access to reasonably good information and to balanced discussion with others who are willing to participate conscientiously. This summary is a simplification but should do for now. By non-tyranny I mean the avoidance of a policy that would impose severe deprivations when an alternative policy could have been chosen that would not have imposed severe deprivations on anyone.4 Obviously there are many interesting complexities about the definition of severe deprivations, but the basic idea is that a democratic decision should not impose very severe losses on some when an alternative policy would not have imposed such losses on anyone. The idea is to rule out only some of the most egregious policy choices and leave the rest for democratic decision.

There are sixteen possible positions defined by acceptance or rejection of these four principles, but the useful models of democracy reduce to just four.⁵ Variations that aspire to more than the four are either unworkable or merely utopian. Those that aspire to less include elements of one of these but are less ambitious than necessary. The rejection of all four would be an empty form of nondemocracy.

The four positions have all been influential. In some cases, I modify a familiar position to make it more defensible, in order to get the strongest version of each position. By competitive democracy I mean the notion of democracy embodying electoral competition in a context of constitutional guarantees for individual rights. Most influentially, this approach was championed by Joseph Schumpeter and more recently by Richard Posner and others.6 This approach to democracy is in fact the one that is most widely accepted around the world.

In this view, democracy is not about collective will formation but just a “competitive struggle for the people’s vote,” to use Schumpeter’s famous phrase. Legal guarantees, particularly constitutional ones, are designed to protect against tyranny of the majority. Within that constraint, the key desideratum is competitive elections. In Schumpeter’s view, it is a mythology left over from ill-defined “classical theories” of democracy to expect the will of the people to be meaningful. Electoral competition—without any constraints on whether candidates or parties can mislead or bamboozle the voters to win—is what matters, in this view.

Of course, any developed view of competitive democracy would require a specification of the people whose views are to be considered equally. Robert Dahl rightly criticized Schumpeter for leaving this unspecified.⁷ Modern democratic norms embrace the franchise for the full adult citizen population with only minor exceptions.⁸

Schumpeter argues that we should not expect a “genuine” public will, but rather “a manufactured will”: “The will of the people is the product and not the motive power of the political process.” Further, “the ways in which issues and the popular will are being manufactured is exactly analogous to the ways of commercial advertising.” In fact, he believes that competing parties and interest groups have “infinitely more scope” on public issues than in commercial competition to manufacture the opinions they hope to satisfy.⁹ Competitive democracy, at least on Schumpeterian terms, sees little likelihood and little need for deliberation by the people.

However, the model of competitive democracy, particularly when the role of constitutional protections for individual rights is foregrounded, provides some guarantees against tyranny of the majority. That is why our chart carries a + not only for political equality in elections but also for non-tyranny, at least as an aspiration.¹⁰

Turning to the second column in the chart, by elite deliberation I mean the notion of indirect filtration championed by Madison in his design for the US Constitution. The constitutional convention, the ratifying conventions, and the US Senate were all supposed to be small elite bodies that would consider the competing arguments. They would “refine and enlarge the public views by passing them through the medium of a chosen body of citizens,” as Madison said in Federalist 10 when discussing the role of representatives. Madison held that the public views of such a deliberative body “might better serve justice and the public good than would the views of the people themselves if convened for the purpose.” This position, like the last one, avoids embracing mass participation as a value. The passions or interests that might motivate factions are best left unaroused. The founders, after all, had lived through Shays’ Rebellion and had an image of unfiltered mass opinion as dangerous. If only the Athenians had had a Senate, they might not have killed Socrates.11

If modern legislatures functioned like Madison’s vision of the Senate, there would be far less of a case for new institutions to bring citizen deliberation into lawmaking. The representatives would deliberate on behalf of the people. There would not be a deliberative deficit at the legislative level to respond to. But the emergence of political parties, direct election of the Senate, and party discipline in legislatures, not only in the US but around the world, has greatly limited the opportunities for deliberation by representatives. They are constrained to follow the party whips, and only in acts of political courage or when there are explicitly open or free votes of conscience do they get to follow their deliberative preferences rather than the party line. Hence I mention elite deliberation as a theoretical alternative but one unlikely to be realized much in practice by elected representatives. Who would currently call the US Senate “the world’s greatest deliberative body”? It is an idea much invoked in theory but rarely followed in practice.

By participatory democracy I mean an emphasis on mass participation combined with equal counting. While many proponents of participatory democracy would also like deliberation, the essential components of the position require participation, perhaps prized partly for its educative function (as Carole Patemen argued12) and equality in considering the views offered or expressed in that participation (even if that expression is by secret ballot). Advocates of participatory democracy might also advocate voter handbooks, as did the Progressives, or perhaps new technology for voter information,¹³ but the foremost priority is that people should participate, whether or not they become informed or discuss the issues.¹⁴

A fourth position, which I call deliberative democracy, attempts to combine deliberation by the people themselves with an equal consideration of the views that result. One method for implementing this twofold aspiration is the deliberative microcosm chosen by lot, a model whose essential idea goes back to ancient Athens for institutions such as the Council of Five Hundred, the nomothetai (legislative commissions), the graphe paranomon, and the citizens’ jury. Modern instances of something like this idea include the citizens’ assemblies in British Columbia and Ontario and the deliberative poll. A second possible method for implementing deliberative democracy by the people themselves would involve some scaled-up institution of mass deliberation. Bruce Ackerman and I have discussed such designs in Deliberation Day.¹⁵

These four forms of democracy highlight the limited possibilities currently available for deliberative lawmaking. Competitive democracy does not incentivize deliberation. Candidates do not wish to win the argument on the merits as much as they wish to win the election. If they can do so by distorting or manipulating the argument successfully, many of them are likely to do so. Representatives elected through such processes are looking ahead to the next election while in office. They have only occasional opportunities to deliberate on the merits because of party discipline. Participatory democracy, at least at the scale of ballot propositions, is no more deliberative than party competition-based mass politics. And the fourth model, deliberative democracy by the people themselves, lacks an institutional home for any connection to lawmaking. The lack of deliberation in our current institutions of competitive, representative, and participatory democracy provides an opening for arguments that might institutionalize deliberation by the people themselves in a lawmaking process.

Legislature By Lot: Some Skeptical Considerations

Yet there are reasons for skepticism that the lack of deliberation in our current forms of democracy can be remedied by a full-scale legislature chosen by lot. I present three challenges.

Technical Expertise

The modern legislative process involves numerous technical questions. It is highly complex. Legislators who are unprepared are left in the hands of staff and lobbyists. The whole history of term limits has demonstrated that instead of creating effective waves of citizen legislators, it has created waves of legislators who are more dependent on staff and more vulnerable to lobbying.16 Random selection would greatly exacerbate these problems if it were to constitute a full-function, full-time legislature.

On the other hand, our experience with deliberative polling amply demonstrates the competence of random samples in an organized setting to weigh trade-offs and make reasoned choices to set priorities among competing, value-laden options. Other minipublics also report considerable success at stimulating citizen deliberation for this kind of task. This function should be incorporated into new lawmaking designs. But the full operation of a legislature requires engagement with many technical questions, far more complex than weighing such trade-offs and setting priorities. I will refer to the function amply demonstrated in deliberative polls as priority setting. Priority setting is at the heart of determining support or opposition to a piece of legislation. But it is only one component in the functions required to create new legislation from scratch.

It might be argued that the Citizens’ Assembly in British Columbia produced a ballot proposition in appropriate, modern legal language as a result of extensive deliberation. But this process of producing a single piece of legislation from scratch took a whole year of deliberation.17 A fully functional modern legislature would be expected to produce and deliberate about many pieces of legislation in a comparable period.

Corruption

With a full-function, full-time legislature, the period of vulnerability to corruption is greatly extended as compared to our experience with minipublics convened for a weekend. In a deliberative poll, there is a highly structured set of activities for the weekend, and the names and identities of the deliberators are generally not made available. But over a multiyear period of legislation, all the deliberators would become public figures. If the decisions were consequential ones, with great interests at stake, there is a serious risk that efforts to bribe or promise later employment to members of the sample would distort the deliberations. We have never actually encountered this problem in any of the deliberative polls, now held in twenty-seven countries. But only a subset of the deliberative polls have been part of government decision processes. And the identities of most or all the participants have not been publicly available beforehand. As the decisions of the random sample become more consequential and the public identities of the deliberators become more available, and as the duration of their time in office increases to a year or two, the challenge of protecting the process from corruption increases dramatically.

Maintaining the Conditions for Deliberation

Deliberating microcosms such as the deliberative poll, which convene a sample for the weekend, are carefully organized to enable citizen deliberators to weigh competing arguments, have access to competing experts, engage in mutually respectful and moderated small-group discussions, and carefully work through an agenda of choices, ensuring that the pros and cons of each choice have gotten a hearing. The process produces relatively equal discussion, promotes reasoned choices, and avoids distortions, such as domination by the more advantaged or polarization in Cass Sunstein’s sense of increasing extremity. There is considerable evidence that the participants arrive at their conclusions based on the merits of the argument rather than on some distorting pattern of small-group psychology.18

If the roles and behaviors of randomly selected full-time legislators were anything like the ones we are familiar with for legislators in modern society, then there would be a host of behaviors outside any such deliberative structure. There would be many individual meetings, caucuses, and meetings with lobbyists, staff, constituents, and so forth. Individual representatives might deliberate or they might not. But we have no institutional design to reliably predict that they would. The deliberation-enabling structure that we know works for a weekend’s task of priority setting with ordinary citizens would not be maintained for the much broader range of activities we can expect in a legislature. Without a structure, we have no way of knowing how much they would deliberate on the merits rather than simply bargain or build coalitions. Of course, there is a deliberative element in such activities. But it is not deliberation that answers to the key hypothetical, which would be the point of inserting deliberation into the legislative process.

The key hypothetical is, what would the people think under good conditions for contemplating the issue in question? That is a disarmingly simple idea, but it requires both external validity in the representativeness of the sample in attitudes and demographics and internal validity for the causal inference that some version of “good conditions” is producing the resulting opinions or considered judgments about what is to be done. That representation of what deliberative opinion on a topic is, or would be, can provide an input to legislative priority setting. The aggregation of all the individual deliberative opinions on a given issue can be taken as a recommendation for the appropriate direction for collective self-rule. Many individual side bargains and coalition activities, all part of normal legislation, would not amount to any kind of coherent answer to this simple but basic hypothetical question.

I would not claim that the precise deliberative poll design is the only way of providing an answer to this key hypothetical question. I am sure that as technology and innovation advance, there will be many credible ways of doing so. But my point is that the key hypothetical is what needs to be inserted into the policy process and indeed the lawmaking process for the deliberations of the people to contribute to collective self-rule. Sometimes, as in an Ackermanian constitutional moment, the hypothetical might be the same as what actually emerges from national public engagement. But such situations are admittedly rare. For lawmaking amid normal politics, it would be immensely useful if there were a deliberative input showing the considered judgments of the public about what needs to be done. That input can have an effect on an advisory basis or just as a media project sponsored by civil society. But here we are considering a further step, one that has been taken on occasion for deliberative polls (for example, in Texas, Japan, Macau, Greece, and Mongolia) of incorporating the poll as an official input into government decision.19 All of these cases involve short-term convenings of a sample from the relevant population. None of them involve a full-function, full-time legislature.

These skeptical arguments are not necessarily dispositive. One might argue, for example, that the Council of Five Hundred, an Athenian institution I invoke in the next section, was not a brief convening and functioned well with great responsibilities.²⁰ It met for a full year. However, it was nothing like a full-function, full-time legislature. Its only role in the legislative process was agenda setting. And it had many other administrative functions and alternated responsibilities among groups of fifty on a rotating basis. It also drew on a citizenry that was constantly rotating in and out of public responsibilities, providing considerable preparation. So it seems a slim basis for precedent that a modern random sample of the citizenry as we know it could function equally well as a full-function, full-time legislature. But here we are pushing historical analogies too far. Perhaps an appropriate modern analogue could be developed and tested at some point. In the meantime, the Athenian case offers ample inspiration for reflecting on democratic possibilities. Let us turn to its design now in more detail.

Athenian Reflections

Hence our focus is on how the more limited task of deliberative priority setting might be institutionalized in a legislative process. Consider the Athenian system as it developed in fourth-century Athens. It suggests various roles that randomly selected deliberators could serve in an actual lawmaking process. While the scale and context are utterly different from modern nationstates, some of the insights suggested by their rich institutional innovations may well be applicable.

The “first democracy” in ancient Athens is often pictured primarily in terms of the assembly where the people made authoritative decisions.21 In sight of the Acropolis, about six thousand citizens could fit in an area called the Pnyx, discuss proposed laws, and vote on them by show of hands in the assembly. However, the citizenry of Athens ranged between thirty thousand and sixty thousand males during the periods of democracy. Women, slaves, and metics (legal resident aliens such as Aristotle) could not vote. Hence most of the population—indeed, most of the eligible citizenry—could not vote at any given meeting of the assembly. Nevertheless, this first democracy set an example for direct rule that has reverberated through the ages.

However, after the disasters of the Peloponnesian War with Sparta, the Athenians briefly lost their democracy. When they managed to reinstate it (in 401–2 BC), they devised a number of reforms that emphasized what we call deliberative democracy. Some of these institutions claimed earlier vintage, but they were put together in a systematic way with the reforms. It is the redesigned Athenian democracy of the fourth century that we want to examine to get a glimpse of deliberative democracy institutionalized.

Mogens Herman Hansen sees a clear motive for the redesign: “The tendency of the reforms is clear: the Athenians wanted to obviate a return to the political crises and military catastrophes of the Peloponnesian War.”22 The orators could goad the assembly into hasty or unwise actions, including disastrous wars. The Athenians had learned that “a skillful demagogue could win the citizens to his project irrespective of whether it was really in their interest.”²³

In the new system, a decree passed by the assembly could not become a law unless it was approved by the nomothetai, a randomly selected sample of five hundred or more citizens who would deliberate for a day, hearing the arguments for and against the proposal. Only if the proposal got majority support by this body could it become a law. A. R. W. Harrison suggests that they had “deliberately invented a perfectly democratic brake to slow down the machine.” It was designed to maintain “the restored order against the possible ill effects of snap votes in the ekklesia” (the assembly).²⁴ “A perfectly democratic brake” suggests that instead of restricting their democracy, they had introduced another kind of democratic institution, one that was also democratic but in a different way.

The system now had deliberating microcosms chosen by lot before the assembly, during the assembly, and after the assembly. Before the assembly, only proposals approved by the randomly selected Council of Five Hundred could be considered in the assembly. During the assembly, orators had to be mindful that they were subject to a special court, the graphe paranomon, which could prosecute an illegal or unwise proposal made in the assembly. The purview of this special court, which also had five hundred or more randomly selected members, was broad (and sometimes misused).²⁵ But the intention was clearly to provide incentives against irresponsible demagogues turning the assembly to their will. After the assembly, there was now a clear distinction between mere decrees, which the assembly could pass, and laws, which had to be approved by the nomothetai.²⁶ This provided a multistage process hemmed in before, during, and after the meetings of the assembly, so that the direct democracy was fused with deliberative institutions representing all the people through random sampling.

As George Grote, the noted historian of ancient Greece from Victorian times, concluded, “There can be no doubt that the Nomothetae afforded much greater security than the public assembly for a proper decision.” The revised system “hedged about the making, annulling, or amending of nomoi (laws) with an elaborate process in which the nomothetai played an important role.”27

The reforms were designed to “hedge about” the assembly with deliberative groups chosen randomly who could ensure more responsible decisions. The samples were not precisely what modern experts would call random samples, but they seem to have been regarded as such.²⁸ People had to put themselves on the list from which the random sample would be drawn. But the sense of public duty was widespread among those privileged enough to be male citizens, presumably motivating participation. Participation in all aspects of Athenian self-governance was extraordinary.²⁹ And the sampling process was taken seriously. In earlier times, the method was to draw beans from a container.³⁰ The Athenians perfected the process with an allotment machine, the kleroterion, which yielded random samples of those who put themselves on the list. The sampling was conducted in public ceremonies. Some argue that random sampling was an embodiment of equality. Some argue that it was a guarantee against corruption and a method of dispute resolution. Both rationales are relevant for our purposes.

There were no property qualifications for serving in the assembly or in the courts and the various randomly selected institutions. There was, however, an age requirement of at least thirty years for the Council of Five Hundred and the nomothetai. Some have thought this was an effort to get more sober judgment in these institutions. Still, on balance the whole process was remarkably democratic—within the limitations of who was considered a citizen.

In viewing the system as a whole, there was also another key point: rotation. There were so many opportunities to be selected randomly and so many meetings of the assembly that people could take turns “to rule and be ruled,” as Aristotle noted in the Politics.³¹ Hansen calculates that “something like every third citizen served at least once as a member of the Council” and three quarters of all members had to serve as the rotating head of government for a day. “Simple calculation leads to this astounding result: Every fourth adult male Athenian citizen could say, ‘I have been twenty-four hours President of Athens.’”32

Fourth-century Athens did not rely entirely on deliberative democracy any more than fifth-century Athens before it had relied entirely on direct democracy. The reformed design was clearly a mixed system still with a very prominent element of direct democracy. But this system gives the first sustained picture of deliberation playing a key role in popular control of the laws. The people deliberated, they had impact, they made choices. The Athenian system has often been dismissed, like the democracy of the modern town meeting, as something only suitable for the small scale. But that limit is most clearly posed by the assembly. There are only so many thousand people who can gather together in a face-to-face meeting. But the deliberative elements of Athenian democracy do not face the same limitation. The random samples that deliberated could, in theory, scale to much larger populations. We now know from modern statistics that one does not need a larger sample to accurately represent a larger population. The statistical precision with which a random sample can represent a population varies primarily with the size of the sample, not the size of the population. Hence these deliberating microcosms can be applied with credibility to much larger populations than the Athenian demos. The rotation aspect is also in principle replicable, but it would take a design offering numerous opportunities at various levels of government. One might imagine local, state, and national deliberations occurring frequently as inputs to government for various kinds of issues. But that takes us beyond our topic here.

Modern Adaptations: Three Scenarios

Consider three patterns: the deliberating microcosm preceding legislative decision, the microcosm following legislative decision, or the microcosm preceding participatory democracy (via ballot propositions) to set an agenda. We can picture these three as:

Deliberating microcosm		Representative democracy
Representative democracy		Deliberating microcosm
Deliberating microcosm		Participatory democracy (ballot propositions)

Pre-Filter: Mongolia

It is worth examining a modern instance of the pre-filter to fix ideas. It is institutionalized now in Mongolia, a competitive democracy with a private property market system. It is a democratic system with both the advantages and the ills now all too common in other competitive democracies (hyperpolarization between the parties, issues of corruption, a dialogue often distorted by social media, and so forth).

In February 2017, Mongolia passed legislation requiring “deliberative polling” before a constitutional amendment could be considered by the legislature. To prepare for its first implementation, a cross-party committee in the legislature advised by two other committees that had conducted public consultations and done research on the need for amendments came up with a bipartisan agenda of possible proposals affecting the legislature, the role of the president, the powers of the prime minister, and the issue of protecting the civil service and the judiciary from politics. These topics yielded eighteen specific proposals, some supported by members from one of the two major parties, some supported by members from the other. On April 28–30, 2017, a national random sample was convened in the government palace in the capital city of Ulaanbaatar. The National Statistical Office, the same government agency that conducts the census, recruited an excellent stratified random sample: 1,568 households were randomly selected with participants randomly selected within the households; 96 percent of the people selected completed the initial interview; half of those interviewed were randomly selected to be invited; and out of the 785 who were invited, 669 came and completed the two full days of deliberation. All expenses were paid by the government, but no honorariums were offered. Our analyses show that the sample was highly representative of the citizenry, both in attitudes and demographics.

The Deliberative Poll included eighteen questions about specific aspects of the proposals. Responses to ten of those eighteen questions changed significantly between the predeliberation survey and the final survey taken afterward. It is worth discussing both the significant changes and the proposals that rated most highly at the end, regardless of change. The highest rated proposals withstood all the criticisms and still came out at the top of the list.

Support for two of the most ambitious proposals dropped dramatically with deliberation. The proposal for a second chamber based on the model of the 1990 constitution, a proposal with strong support in some opinion polls, dropped dramatically. “Creating a Parliament with two chambers: a people’s representative body (People’s Great Khural) and legislative body (State Baga Khural)” went from 63 percent to 32 percent, a drop of 31 points. With deliberation, the public became more skeptical that “a second chamber would provide effective oversight of the lower house of Parliament.” Agreement with this idea dropped from 72 percent to 41 percent. More specifically, there was increased agreement with the criticism that “both chambers would be controlled by the same political parties, thereby not providing proper oversight.” Those agreeing with this proposition rose from 43 percent to 59 percent. Lastly, there was a significant increase from 53 percent to 58 percent in those who agreed that “adding a second chamber would create too many politicians.”

A second major drop in support occurred with the proposal for an indirectly (rather than directly) elected president for only a single six-year term. There were two components to this proposal, the change in the term and the indirect election. Support for “Electing the President for a single six-year term, without reelection” dropped from 61.5 percent to 41 percent with deliberation. Support for “Electing the president for a single six-year term by an expanded plenary session of the Parliament that includes Parliament members and the Citizen’s Representative Councils of aimags and the capital city” started at 40 percent and ended at 34 percent (not a significant drop, but showing a low level of support after deliberation). There was a significant drop in support for one of the arguments in favor of indirect election: “If the President is indirectly elected by the Parliament and the Citizens Representative Councils, then he/she will be someone acceptable to all sides and above political fray.” Agreement with this conclusion dropped from 61.5 percent to 41 percent. On the other hand, there was strong agreement before and after with one of the key arguments in favor of direct rather than indirect election: “If the President is directly elected s/he can better speak for the interests of all people” (85 percent before, 80 percent afterward, no significant change).

By contrast, the deliberators supported an amendment that would increase the power of the prime minister: “Granting the Prime Minister the authority to appoint and dismiss the members of his/her Cabinet.” This proposal increased significantly from 62 percent to 73 percent. Deliberators agreed that “If the Prime Minister cannot even appoint the members of his/her own Cabinet then s/he lacks the authority to get anything done.” Seventy-two percent agreed before and 67 percent agreed afterward (not a significant drop).

After all the arguments for and against, it is worth noting that nine of the top ten proposals were directed at protecting the civil service and judiciary from political interference or corruption. These were clearly the public’s highest priorities after deliberation. The concerns from the ruling party that the president needed to be indirectly elected and from many members of the main opposition party that there was a need for a second chamber were rated much lower.

The results of the deliberative poll served as an effective filter on the proposals that the parliament can take seriously in formulating its amendment. The deliberative poll did not generate proposals from scratch. Rather, it filtered the proposals put forward by the competing parties and allowed the public’s deliberations to judge which ones deserved to go forward. Before the law on deliberative polling, the parliament would have had the power to pass an amendment by two-thirds or the power to put an amendment proposal to referendum. At the moment, the majority party has more than a two-thirds majority. Now the deliberations of the people in microcosm serve as an effective filter on what they will consider.33 This experience so far argues for the viability of the pre-filter legislative model for a limited purview of issues, such as constitutional change.

Post-Filter: Return of the Nomothetai?

Let’s return to the Athenian examples. We can imagine modern applications of the nomothetai idea in the context of the current debate about second chambers selected by lot.³⁴ Just as the ancient nomothetai, selected by lot, had the role of providing a “perfectly democratic brake” on the decisions in the assembly, a modern minipublic selected by lot could provide a brake on decisions in the parliament or lower house of a modern nationstate. I am not thinking of a fully developed second chamber by lot. Rather, imagine a role parallel to that of the nomothetai, convened briefly to pass judgment on a specific proposed law with the case for and against the proposal deliberated for a weekend on something like the deliberative poll model. But which laws? It would not seem practical to convene a new minipublic for every law.

Imagine the challenge if such an institution were inserted into the last stage of decision for a new law in a modern constitutional democracy with competing parties. Suppose there was a requirement that the final decision go to the minipublic for any law that does not pass by two-thirds in the parliament. If there is merely a majority but not a supermajority, then the people, convened in microcosm, have the final say. But if there is a supermajority in the parliament, then there is no triggering of the requirement to convene the people. This design would arguably create real incentives for even highly polarized parties to work together. It would also provide a barrier to turbocharged bare majorities that wish to ram though a proposal without any regard to the other side or sides. Placing the deliberations of the people in this position in a deliberative system might provide strong incentives for overcoming polarization among the elites, and it would motivate elite deliberation across the aisle in the parliament. We can picture it thus:

Parliament (if 2/3 support)		passage of legislation
Parliament (if majority support but less than 2/3)		minipublic for final decision

Perhaps for efficiency, the minipublic could be convened when there are multiple bills to be considered. The identities of the random sample should not be announced beforehand to protect against bribery or lobbying of ordinary citizens.

Such a design might incentivize the parties to work together, limiting the current extreme polarization characteristic among representatives in so many other countries. It would also perform as a “brake” on unwise proposals, allowing the people, as in ancient Athens, to stop proposals in the final stage. A positive vote is required, allowing the people in effect a veto on objectionable new laws.

Once again, the idea is brief convenings under specified conditions, not a standing body. A standing body would trigger the concerns mentioned earlier about corruption, lack of structure, and the need for broad-based expertise. Note that by imagining the nomothetai role served by a deliberative poll design, we add the element of small-group discussion. The ancient legislative commissions of five hundred or more heard the arguments in an amphitheater. Modern investigations demonstrate that much of the deliberative opinion change comes from the discussion.35

Pre-Filter: Deliberative Agenda Setting for Ballot Propositions

In 2011, a statewide deliberative poll in California was convened to consider governmental reforms in the initiative system, the structure of the legislature, taxation and budgeting, and other issues. A highly representative statewide sample of 412 deliberators gathered for a long weekend in Torrance, California, and produced results on proposals in all these areas. Some of the initiative reform proposals fed into the deliberations of an initiative reform effort and have become law after having been adopted by the legislature.36 But the underlying motivation for the project was to pilot a bigger idea. That pilot was only partly successful. The idea was that a deliberating microcosm of the public could be periodically convened to propose public-interest initiatives that could go on the ballot. In fact, a proposition loosely connected with the deliberations, Proposition 31, did go on the ballot. But it included additional provisions—supported by well-meaning public-interest organizations—that attracted opposition, and Proposition 31 was defeated. We showed that there would have been stronger and perhaps successful support for the elements in the proposition that corresponded to the results from the deliberative poll. This experience suggests a possible institution that could be adapted for lawmaking by the people.

Consider the California deliberative poll (What’s Next California, or WNC) as a pilot for an institution that could empower the public to help set the agenda for the initiative. How would it work? Several problems have to be solved. It has to be nonpartisan and scientifically credible yet also connected enough to the political fray that the selected proposals have actual proponents to advocate them to the electorate at election time. It needs to live up to the same sorts of criteria that we have used to evaluate WNC and other deliberative polls. It needs to be representative of voters in both attitudes and demographics. It needs to be large enough in scale that its claims to representativeness and its results are meaningful statistically. If there are briefing materials or sources of information for the deliberations, they need to be balanced, and deliberators must be given an appropriate opportunity to interact and seek further information. And lastly, the proposal or proposals selected in the deliberative process have to be followed up with something on the ballot appropriately connected to the deliberations and advocated at election time. Voters around the state should have an opportunity to consider the same reasoning that led the microcosm to support the proposal.

To fix ideas, imagine this scenario. A random sample of voters, about the size of WNC, is convened every two years to consider possible proposals to go on the ballot. Where do these proposals come from? Groups that wish to be proponents of initiatives develop proposals and satisfy a low threshold of signatures, low enough that civic groups could satisfy them with reasonable effort but without necessitating paid signature gathering. The reason for the (low, rather than onerous) signature threshold is that there has to be some way of distinguishing serious from frivolous proposals. If a proposal is selected by the microcosm, it then qualifies without the burden of the full signature collection. Saving the expense of most of the signature collection is an incentive for groups to submit their proposals to this process. A second incentive is that they can identify their proposal as endorsed by a representative and informed microcosm of the people. Once voters become familiar with the process, such an endorsement could be very valuable. We know that a prime question voters ask about any initiative is, who supports it and why?37 Endorsements are a key heuristic or informational shortcut influencing support or opposition to ballot propositions.³⁸ In this case, the answer is that a proposal got on the ballot partly because a random sample of the people thought it was a good idea after they thought about it in depth. As that idea catches on, so that less of the history needs explaining, it is likely to become more effective and valuable as a property of referendum campaigns, increasing the incentives for groups that could act as proponents to seek the thoughtful and representative endorsement of the people.

We can imagine that the entire process would be supervised by a nonpartisan commission or advisory group. Several key functions would need to be fulfilled. The briefing materials for and against each proposal would need to be scrutinized for balance and accuracy. Perhaps proponents could provide the case in favor and potential opponents the case against, but all have to pass scrutiny from a balanced advisory committee appointed by the commission to have final say on the briefing materials. When there are contested facts, the competing versions are included in a balanced format. Perhaps proponents and opponents would each have a right to reply. If there are no opponents, then some could be appointed to serve that role for this preparatory stage. At the deliberations, there would be a list of experts who could respond to questions in the plenary sessions, and these again would be scrutinized by the advisory committee. These are all functions that have been accomplished in past deliberative polls, even on highly controversial issues.

How could the design ensure a connection between the people’s deliberations and what goes on the ballot? The people would deliberate in choosing between developed proposals. We might imagine a small window for the proponents to improve their proposals in light of the deliberations with the opportunity for a follow-up confirming vote from the microcosm if the proposal changed. There are two aims at this stage—identify the preferred proposals and get the best version of them in light of the deliberations. Obviously, there are many variations and details, but this scenario sketches an approach that builds directly on the What’s Next California pilot.

We can imagine such a process as an alternative route to the ballot, not the sole route. The idea would be to provide a supply of at least some public-interest propositions that the people would find meaningful. If such a design proved successful, it could be expanded. Perhaps it might begin with the selection of one proposal each cycle. Then the number could be increased, and perhaps more than one microcosm could be convened if the number of proposals became large. Given the extraordinary expenditures on campaigning for proposals once on the ballot, it seems appropriate to imagine relatively modest expenditures for the crucial agenda-setting process. Like Oregon’s Citizens’ Initiative Review process, it might even begin with foundation funding and only then move to government funding after a track record of successful implementation.

What’s Next California showed that it is practical to convene a microcosm of the state’s voters to consider propositions in a balanced and thoughtful way. The microcosm satisfied our expectations: it was representative in attitudes and demographics, it gained knowledge, it evaluated a number of proposals, and it produced many significant changes on the basis of identifiable reasons. Some of its conclusions even fed into the legislative and the initiative process. It showed that a deliberating microcosm could provide a possible institutional design for setting the agenda for initiatives.

The initiative process is supposed to be the people’s process. But the agenda-setting function has been captured by those who can afford it. Why not recapture it for the people, using this ancient device? The key would be institutionalization and follow-up, not only to put the people’s choice before the entire electorate but also to make the reasoning available as a basis for choice. That would add a truly deliberative element to mass direct democracy and fulfill many of the initial aspirations of the initiative to empower the people to engage in thoughtful self–government.

Conclusion

In all three scenarios a deliberative-poll-like process offers a brief convening for the limited role of priority setting—for possible constitutional amendments, for final approval of legislation, or for agenda setting for ballot measures. These roles and entry points for deliberation hardly exhaust the possibilities. But already, they illustrate the many possibilities for officially incorporating the conclusions of a deliberating microcosm in the lawmaking process.

I am not advocating the position that every deliberative input should be conducted via deliberative polling. As noted earlier, I am sure that with the development of technology and innovative experimentation, we will find even better models. However, the deliberative poll satisfies some basic design features that I believe should be satisfied if we are to have confidence in the key hypothetical inference: This is what the people would think about the issue under good conditions for thinking about it. Below is a list of considerations. Other decision processes can also be constructed to satisfy them.

1. There should be credible random sampling with attention to both attitudinal and demographic representativeness. If the microcosm is unrepresentative in either respect, then that could easily distort the deliberations, as some views or interests will not be represented. Hence it is all the more important that initial data be collected to ensure that it is possible to evaluate the representativeness of the sample.39

2. The size of the sample of deliberators needs to be large enough so that statistically meaningful analyses can be conducted not only of their representativeness but also of any opinion changes. Hence the need for data collection at the individual level both before and after deliberation.

3. The agenda and basis for deliberation need to be balanced and transparent. An advisory group is very useful as a source of public assurance. The deliberative poll usually has initial arguments in favor and against each and every proposal listed in tabular form for easy reference. But there are many techniques to ensure both the balance and accuracy of the arguments and information in the deliberations as well as many different techniques for how the information can be presented.

4. Any informational background materials need to be accessible to ordinary citizens, whether the materials are in written or video form.

5. Access to competing experts to answer questions can be immensely useful. This aspect was employed in the British Columbia Citizens’ Assembly as well as in the deliberative poll. In any ongoing deliberation, new questions will arise. They cannot all be anticipated. Having balanced experts facilitates engagement with the competing sides of an argument.

6. Moderated small-group discussions facilitate relatively equal participation. In our view, moderators should be trained never to offer opinions or provide information but rather only to facilitate civil discussion.

7. Collection of qualitative as well as quantitative data to provide information on the reasons that weigh with the sample in coming to its conclusions is critical. Transcripts of small-group discussion can be analyzed systematically. Other techniques could be used, such as formulation of agreed statements after the individual-level data is collected.

8. Also necessary are analyses of the small groups to ensure that they are not subject to systematic patterns of distortion, such as small-group polarization or domination by the more advantaged. Such distortions would cast doubt on whether it is “the force of the better argument” that is determining the conclusions, rather than social coercion within the groups.

These are only some of the key components that have served deliberative polling in its claim to provide good conditions for a random sample to deliberate. Perhaps one or another can be dropped or modified after study in further controlled experiments. Or perhaps simulations in virtual reality (or some other new technology) can provide an entirely different method for a population to deliberate—to weigh the competing arguments and come to considered judgments. The future is more inventive than we can imagine. But we know already that the aim of deliberating with random samples is practical for convenings long enough to perform certain essential political functions in a lawmaking process. I urge the continuation of such experimentation and dialogue.

In my view, some of the key functions of a second chamber could be successfully performed by randomly selected microcosms in brief convenings. Perhaps my list of challenges to fulltime, full-function legislatures by random selection can eventually be overcome with the right design and experimentation. But that is an unsettled question. In the meantime, we have ample basis for credibly advocating the insertion of random assemblies in brief convenings to accomplish priority setting in a lawmaking process.