The Citizens’ Assembly Model

Full text

The Citizens’ Assembly Model
John Ferejohn
Stanford University
February 2006
Introduction
Since the time of Ancient Athens students of democracy have been skeptical that
elections are a way of implementing democracy, at least if democracy requires something
like government “by” the people rather than “of” or “for” them.1 There are several
grounds for this skepticism: first, elections may select unusual people – better, more able,
or merely more ambitious leaders – and thereby exclude ordinary people from a regular
role in government. Second, whether or not elected leaders are especially skillful, those
elected to office tend to become a professional class with knowledge and interests
separate from those of ordinary people. The Athenians themselves thought that lottery
was the democratic way to fill office, as it ensures that anyone who wanted to serve
would be able to do so.2
The modern reaction has generally been to tone own expectations for democracy,
settling for electoral representation as the only feasible kind of democracy in modern
1 For an account of representative or electoral democracy see Bernard Manin, Representative Government,
Cambridge University Press, 1997.
2 Athenian lotteries were moderate in several ways: an aspirant to office had to agree to serve and serving
as a magistrate was risky as well as costly. Every office hold was continuously subject to impeachment and
to a thoroughgoing audit after his year in office. So, not everyone was willing to stand for office. See
Manin, op cit. and, for the best institutional description of Athenian practices, Mogens Herman Hansen,
The Athenian Democracy in the Age of Demosthenes, Oklahoma University Press, 1999.
1

circumstances. Moreover, democratic elitists following Joseph Schumpeter have thought
that electoral elitism is not only unavoidable but is positively desirable in modern large
scale government, both because of the technical nature of a modern government, but
especially because citizens cannot be expected form real opinions about distant and
unfamiliar political matters.3 Without specialized leaders, chosen by competitive
elections, we would be completely at the mercy of unelected technocratic elites, or
powerful private interests. And that would be worse than democratic elitism which, at
least, aims at a government for the people if not by them. Of course, this is a negative
argument and not an assertion that competition will actually impose real discipline on
public officials. But there is reason to think that even if elections are very competitive –
an implausibly strong assumption – elected officials will still enjoy a great deal of
uncontrolled discretion.4
Whatever the merits of democratic elitism, there have been many attempts to
permit direct popular input into lawmaking by such devices as the referendum, the
initiative, the recall election, and provisions for electing more and more officials
including judges. There has been a century long surge in the use of mechanisms of direct
democracy, not only in the United States, where the populist tradition is strong but also in
Europe where a number of countries attempted to ratify the proposed European
Constitution by referendum rather than through parliamentary approval. And California,
ground zero of the initiative process, where an unpopular governor was recently recalled,
has seen a spate of proposals proposed by the new governor intended to enact legislation
3 Schumpeter’s point anticipated the findings of modern students of public opinion on this point. Joseph
Schumpeter, Capitalism, Socialism, and Democracy, Harper, 1942. John Zaller argues that the mass
public has almost no attitudes about the public issues and policies; see his Nature and Origins of Mass
Opinion (Cambridge, 1992)
4 John Ferejohn, 1986, “Incumbent Performance and Electoral Control,” Public Choice, Vol 30 (1986).b
2

through plebiscite.5 In the broader context, the long history of tax revolts in the United
States must be seen as partly attributable to the availability of the initiative and
referendum. And, no doubt, the timidity of state judges in policing the death penalty
must be partly attributed to the looming presence of judicial election and recall. If
nothing else, the institutions of direct democracy make political officials a bit nervous.
Moreover, a number of recent studies indicate that these institutions probably do
have the effect of moving legislation, whether enacted through popular processes or by
the legislature, in the direction of the opinion of the median voter.6 And some scholars
think that that is a good thing.7 Whether it is actually good or not, however, depends on
whether the median voter’s wants really reflect some attractive conception of the
common interest or, at least, or her own genuine interests, or whether, instead, they are
ephemeral reactions to emotionally charged events or are manipulated by special interests
for their own purposes.8 There are good reasons, drawn from decades of research on
public opinion, to suspect that one or both of these worries may be warranted.9 It seems
5 These attempts have been only partially successful; the recent initiatives proposed by Governor
Schwartzenegger in 2005 all failed by substantial margins.
6 Elizabeth Gerber, 1999 The Populist Paradox: Interest Group Influence and the Promise of Direct
Legislation, Princeton: Princeton University Press.
7 Matsusaka, John G., 2004, For The Many or The Few. The Initiative Process, Public Policy, and
American Democracy, Chicago, Il: University
of Chicago Press.
8 This is not the place to go into alternative justifications of the mechanisms of direct democracy. It may be
argued that the “agency” problems of elected representatives are intractable without ways of going around
the legislature. But, this line of thought is convincing only if could be shown that the effect of direct
democratic mechanisms is welfare improving, in light of the informational and cognitive limitations of
ordinary voters.
9 John Zaller, op cit. has provided a pretty convincing case for a claim of this sort. James Fishkin’s
advocacy of deliberative polls is based on the plausible assumption that the median voter is likely an
untutored person, even as to her own “real” values and preferences, and that there is no reason to defer to
her unreflective preferences as a guide to policy. James Fishkin, 1991, Democracy and Deliberation: New
Directions for Democratic Reform, New Haven: Yale University Press.
3

significant, for example, that one noted consequence the popular initiative seems to be to
establish and retain the death penalty in places in those states that have the initiative.10
Thus traditional direct democratic institutions have various limitations as modes
of implementing direct democracy but the most troubling one is this: each is easily
captured by interest groups and can be, and will very likely be, used to introduce
distortions into the political process at least on issues of little salience to the average
voter and are, in this respect, undemocratic, even anti-democratic, in practice.11 The case
of California’s popular initiative is typical. It is relatively easy for an interest group to
get the required number of signatures to put a proposal on the ballot – a proposal that
faces the voters with a take-it-or-leave-it offer which is in the interest of the proposer –
which only slightly improves the (perceived) well being of the median voter. The other
devices of direct democracy are nearly as easily captured and employed either for private
purposes or public projects of little interest to the general public.12
The flaw in the traditional mechanisms is that they permit the proposal power to
be captured: to be removed from the voters and be placed either with the governmental
institutions, in the case of the referendum, or to be put out for bid among private parties.
And, these institutions provide no public defense of capture.13 There is no way for voters
to amend a proposed initiative or referendum or to deliberate about what proposals ought
10 Gerber, op cit. There are of course other objections from the standpoint of minority rights or of the
stability or rule of law. But those objections are irrelevant to the present discussion.
11 There are of course other objections to these institutions raised by those who doubt the value of direct
democracy no matter how well it may be implemented. Many political thinkers think that direct rule is
inevitably fickle and despotic and that representative forms of democracy are not only inevitable but more
attractive than direct democracy.
12 A good treatment of referendum and initiative politics is given in Gerber, Elisabeth R., 1999, The
Populist Paradox: Interest Group Influence and the Promise of Direct Legislation, Princeton, NJ: Princeton
University Press.
13 Courts may, of course, review either proposed referenda or initiatives but the basis of such review has
little to do with legally permissible ways by which private interests may capture the proposal power.
4

to be subjected to public consideration. It has generally been thought to be unwieldy or
even impossible to do any better than this. How would one actually permit open public
debate and amendment on a proposal to be voted on? Simply to ask that question is to
see the unwieldy nature of possible solutions. But, perhaps open debate by the general
public is not needed to cure the chief mischief of traditional modes of direct democracy.
Perhaps what is needed is a genuinely public forum for deliberation over the agenda; one
that cannot be easily captured by special interests.
In fact, this is precisely the idea expressed in classical Greek institutions.14 Most
of the city states had “democratic” chambers in which every citizen could vote.15 But
debate in these large citizen assemblies was often chaotic and unmanageable and so most
cities had a council, the Athenians called theirs the Boule, that had the power to establish
an regulate the assembly’s agenda. The Athenian Boule was chosen randomly from
geographically defined districts from among all citizens over thirty years of age, and
these “representatives” rotated control over its meetings over the course of the year. The
Boule was a small deliberative body, in which every Athenian could expect to serve
during his lifetime, that could consider and revise proposals that had to be decided in the
Assembly.16
14 See Hansen, passim. Indeed the practice extended throughout the classical world. The Roman republic,
while scarcely democratic as a government, confined the making of laws to the comitia, which were
nondeliberative assemblies in which all citizens had a vote (that would be counted in according to complex
procedures). Public deliberation about laws or other public actions took place in contios, which were
informal gatherings assembled by magistrates for the purpose of discussing proposed policies. See Robert
Morstein-Marx, 2004, Mass Oratory and Political Power in the Late Roman Republic, Cambridge:
Cambridge University Press. A more democratic view is given by Fergus Millar, 1998, The Crowd in
Rome in the Late Republic, Ann Arbor: University of Michigan Press.
15 There were of course enormous differences among the city states in the extent of citizenship in the
population.
16 It is probable that the assembly could amend proposals too so the Boule probably did not have a genuine
monopoly on proposal power.
5

British Columbia’s experiment in direct democracy is, in many ways, very similar
to Athenian idea, especially in its separation of democratic proposal power, which the
focus of deliberation, from decision making power where deliberation is less expected.
By requiring deliberation in a relatively small and manageable assembly and not in the
larger public sphere, it offers the prospect that an institution of genuine direct democracy
can be actually be made workable in modern conditions. It embodies the two central
requirements for direct democracy: it permits real public deliberation about future
legislation, albeit among a small a representative body of citizens; and it permits
ratification or endorsement of legislation by the whole electorate. The Citizens’
Assembly model therefore appears to offer a way to permit real popular participation in
“legislation” in a way that is separated from the normal elite-advantaging processes that
form the core of modern governmental practice.
To be sure, the aspirations of those who established the Citizens’ Assembly were
more modest than this. They did not aim to establish a new and generally applicable
legislative process. Their aim was only to permit the people to decide what to do on a
specific issue where politicians had self serving interests and could not be trusted to
decide dispassionately: the choice of the electoral rules by which they themselves would
compete for office. There are other, similar, issues which the motivations of politicians
might be distrusted as well such as the choice of rules governing campaign finance, or
drawing electoral boundaries, or deciding whether or not legislative terms ought to be
limited. But, I think that the promise of the Citizens’ Assembly model is broader than
this. To the extent that the CA experiment showed that ordinary people can conduct a
serious and deep deliberative process in a way that marshals relevant evidence and
6

expertise in making decisions on important or, perhaps, foundational political matters, the
CA model offers a way to rethink the venerable problem of real democracy.
I. The Constitution of the Citizens’ Assembly
British Columbia Citizens’ Assembly did not originate in abstract political theory;
nor did it arise from a conscious attempt to emulate the democratic practices of the
Athenians. Its creation was a local political act: a maneuver in partisan politics aimed at
least in part at securing political advantage for one party over another.17 Of course the
Athenian experiment was itself the result of practical politics. Notwithstanding the
venerable Solonic myth, Athenian democracy was not a creation of a farsighted lawgiver,
but was the result (possibly unintended) of a struggle among bitterly opposed aristocrats.
But whatever its genesis, the CA may represent an important institutional innovation: one
that can be applied, in whole or part, to other political contexts.
The situation that was faced in British Columbia is quite familiar in other political
circumstances as well: the desire among some political leaders to alter the electoral
system conflicts with the resistance of incumbent politicians and parties to changing the
system that permits them to win office. Even when there are good reason to change
electoral rules, politicians may be afraid to do so out of fear of having to compete under
new rules. For this reason electoral systems tend to be very stable except perhaps at
certain special moments when forces line up to permit a rapid change in the electoral
17 The Liberal party, which has had a majority in the Province for years, was still smarting over the fact that
in 1996 it received fewer seats than the NDP despite winning a majority of the votes and was shut out of
the government for five years as a result. It was believed that under a more proportional electoral system
such a thing could never again happen. There are of course other reasons for wanting to adopt some form
of PR: in 2001 for example, the Liberals were returned to power with 58% of the vote and won all but two
seats. Among political observers (not to mention politicians) Canadian seat results are notoriously volatile
even among first past the vote systems.
7

rules. A moment of this kind may have occurred in British Columbia in xxxx. That
moment was created by the combination of particular political circumstance and the
appearance of a new idea: a new proposal that offered the possibility for breaking up a
long established and robust political equilibrium.
The solution proposed was to establish a congress of citizens, drawn randomly
from the voter rolls that would meet and listen to experts and form their own opinion,
free from considerations of self interest, as to what kind of electoral system would be best
for British Columbia, with their recommendation to be submitted to the voters for final
approval. The members were paid for going to weekly meetings and were provided with
staff and budget sufficient to allow them to deliberate seriously and deeply about the
issue. Incumbent politicians were excluded from the process and only one career public
servant had a formal role within the assembly and that was to preside over its
deliberations and to keep parliamentary order. Otherwise, as is evident from the papers
in this volume, the members were free to deliberate together, to draw on information and
analysis from other electoral systems, to hear from interests who thought they could be
affected by any proposed reforms.
While it was obviously important that the members of the CA were chosen in way
that ensured that they would be impartial, in the sense of having no personal stake in the
choice of an electoral system, it was equally important that they reflected, or represented
in some sense, the legitimate interests of British Columbia residents. Its members could
have been expected to share the public interest in having a fair and responsive electoral
system, but they also should reflect the particular interests of significant groups and
geographic locations. The CA was therefore chosen on a geographic basis with two
8

citizens, a man and a woman, selected from each election district and there was concern
that significant groups in the population be represented. And, while some groups were
under-represented in the originally constituted CA, it appears that the overall makeup of
the Assembly represented most major population groups in the Province (need chapter
reference). The inevitable problems of self selection and drop out seemed, in the end, to
have been quite minor in their significance (need reference). In any case, whatever
representational defects there may have been could be remedied in the process of taking
testimony, ensuring that under-represented groups had ample opportunity to present their
views. And the public nature of this process provided substantial notice to voters and
interests about what reforms were being considered and how they were to be evaluated.
II. Deliberation in The Citizens’ Assembly
The assembly engaged in an extensive deliberative process. It took testimony
from whichever groups and individuals wanted to appear before it. It actively sought
information about a wide variety of electoral systems and experiences from other
countries. Its members asked probing and difficult questions of these witnesses. In these
efforts it was guided by a number of expert political scientists who attempted to help CA
members to frame issues and organize discussions. These proceedings were open and
many of them were actually broadcast throughout the province. While it is doubtful that
many voters followed these meetings carefully, it seems that there was a fairly
widespread familiarity among the voters that the process was ongoing.
Initially, the CA began its deliberations with the aim of narrowing the focus to a
small set of alternative electoral systems. The idea was that this narrowed set of options
9

could be effectively communicated to the wider public, provoking a more focused set of
responses in subsequent hearings that would, then, allow the CA to produce a final
recommendation to be put before the electorate. In the end, the CA was able to restrict its
intensive deliberations to two alternative systems – a single transferable vote system, and
a modified proportional representation system, each of which had support within the
Assembly.
There are of course a lot of caveats that need to be made about the Citizens’
Assembly process, both as a learning process and as a decision making process. Leaders
emerged within the CA and there were various staff members who had and exercised
power at various moments not all of which were very visible to the members. The chair
was, in addition, a professional public servant and he made a number of rulings that had
to be accepted as final that may have shaped deliberations decisively. For example, he
ruled out the possibility of expanding the size of the legislature, a ruling that might very
well have altered subsequent deliberations. Those rulings could not be appealed to the
floor as they could have been in an ordinary parliament, and this fact represents an
important reminder of the political constraints on the process.
Moreover the staff organized the meetings and the agenda and decided when to
meet in plenary sessions and when to break out into smaller groups. And one imagines
that they must have helped shape, perhaps decisively, the list of experts that were called
at least in the early stages. Moreover, among the members of the CA, some were more
willing to commit time and effort and pay the price to become real technical experts and,
predictably, these people played a larger role in deliberation. So there were
opportunities for manipulation. Still, the members generally did not seem to worry that
10

this possibility distorted their proceedings or recommendations in a significant way.
After all, whatever their opportunities, those in a position to manipulate the results didn’t
have a tangible interest in doing so.
The view of most of the authors represented in this volume is that the process
worked well in important ways. A representative group of ordinary citizens came
together and deliberated about the BC electoral system and alternatives to it in an
intelligent and informed way. Their deliberations were intense and serious: members
came with open minds and did in fact change their views and arguments over the course
of the proceedings. Moreover, in the course of the hearings and deliberations several of
the members became quite expert in the technical details of electoral systems. There
were hard issues faced along the way and deep conflicts as to what was defective in the
current system and which alternative would be best and, importantly, which system they
could hope to convince their fellow voters to accept.18 Finally, they were able to agree
on a technically sound recommendation, one that reflected values that they had become
convinced had to be respected by any electoral system in British Columbia.19 And they
then offered this proposal with near-unanimous endorsement, to their fellow citizens.
And many of them spoke publicly for this proposal in the ensuing campaign.
18 One reason for the adoption of STV rather than the modified PR system, perhaps the most important one
at the final stage, was that members of the CA were convinced that STV could be described fully and
explained to the voters, while being much less sure that the alternative could be communicated effectively.
19 It is a nice question which values those were. It was clear from the proceedings that people from remote
ridings were very insistent that their constituencies not be submerged in large, geographically
heterogeneous districts. It was also clear that representatives from urban areas were disposed to respect this
strongly expressed idea. This turned out to make the STV system with relatively small districts an
especially attractive alternative. The design of the CA itself expressed two special concerns by giving
“quota” status to riding (by chosing two members from each) and gender (one man and one woman). There
was no special preference expressed in the CA’s design for ethnicity or First Nation status. Although,
when there were no First Nation Representatives in the original Assembly two were added to it. Within the
CA deliberations, the body rejected every “affirmative action” concern except for geography: which it
accepted by chosing a districting system.
11

III. Deliberation in the Public Sphere
In turn, the voters considered this proposal in the context of a campaign in which
voters (in repeated polls analyzed by Richard Johnston and Fred Cutler) manifested
general familiarity with the CA. And significant elements of the public appeared to
arrive at two distinct beliefs: that the CA was composed of “ordinary” Canadians (like
themselves), and that its members had become experts on the arcane subject of electoral
systems. As in the end, the voters approved the recommendation with a very substantial
majority. That this majority was insufficient under the politically set rules of the game
ought not to distract us from the impressiveness of the achievement itself from a
majoritarian perspective.
The question that remains – the focus of the remainder of this paper – is this: does
the CA process offer a viable model of public deliberation that offers the prospect of
making the political process more democratic? (cite to chapters by Chambers and
Rattner) Or, is the CA model simply another kind of blue ribbon commission, put
together for the purposes of advising public officials or, perhaps to legitimate a proposal
favored by officials that those officials think they are in no position to make on their
own? Does the CA model permit genuine public involvement and input into the policy
making process or is it more likely that CA was coopted by public officials for their own
purposes. Is there, in short, an intractable gap between deliberation in the CA and
genuine deliberation in the public sphere?
The CA model could be defended as a model of public deliberation in two ways:
it could be considered a model of hypothetical deliberation. Since its members are a
relatively representative sample of the voters, the deliberations of this body might be
12

considered to model or predict what would happen if ordinary citizens were able to sit
through the same (or a similar) process. James Fishkin has offered such a justification for
his method of deliberative polling, in which a sample of citizens spends a weekend of
intense (and well managed deliberation) in order to form a policy judgment. I suppose it
might be the case that from a statistical viewpoint there would be something to be said for
this view, notwithstanding the literal impossibility of the whole voting body realizing the
counterfactual condition. It seems plausible, for example, that repeated samples put
through deliberative processes might tend to converge to the same proposal and that this
limit could play the role of the population proposal. Something like that.
Or, one could take an Aristotelian line and define democracy as a system of
government where all the citizens take turns ruling and being ruled. Of course, not every
citizen can plausibly hope to be chosen to sit in a Citizens’ Assembly, even if these were
to become a regular part of the normal practice of government. But each citizen has, in
principle, an equal chance of service and so, statistically, something like the Aristotelian
principle is satisfied. On this account, the reason that voters ought to respect such
deliberations is that, ex ante, they had the same chance of serving as anyone who ended
up on the CA, and in this sense the judgment of the CA is, statistically, approximately
their judgment. Suppose for example that the CA is thought in this way: every person
places an “intelligent ballot” – a ballot that acts as her deliberative agent within the CA if
she is drawn – into a container. Then the CA is drawn from that container and produces
its deliberative judgment from the deliberative interaction of those ballots. It seems clear
on this description that ex ante, each person’s “considered” judgments have an equal
13

chance to influence the outcome and that, in this way, the recommendation of the CA
stands as an approximation for the judgment of electorate generally.
But I am interested in pursuing less abstract interpretations and want to argue that
the CA model instantiates a real public deliberative process of some kind. (cite to
chapters of Thompson, Bohman, Chambers and Rattner) This deliberative process
instantiates a kind of division of deliberative labor and accord a real role for the voting
public at large as well. This deliberative process, like those in Athens and the classical
world, divides the public into two parts: the CA which deliberates and proposes, and the
electorate which does not directly deliberate but makes the final decision.
There are of course real reasons to doubt that the CA could be a genuine model
for public deliberation. The CA was only a small body of people who rearranged their
lives in ways that permitted them to get deeply involved in the recommendation they
were to make to the electorate. Indeed, from what we know of public opinion during the
STV campaign, the members of the electorate – the wider public whose deliberative input
democrats seek – did not seem to form their own judgments about the comparative
advantages of the various electoral systems under discussion. The public generally did
not deliberate about the actual choice to be made in the way the way that philosophers
have wanted. Perhaps, as public opinion research suggests, it is unrealistic to think that
the public is at all capable of directly evaluating a technical recommendation of this kind.
Rather, as far as can be seen (in Johnston’s and Cutler’s chapter, need cite), the
public’s deliberation was indirect or, perhaps, “virtual.” Different segments of the public
formed judgments about the CA itself – some people saw that it was composed of
ordinary Canadians like themselves (i.e. not politicians with self interested motivation),
14

while others were convinced that its members had become experts on the subject – and
decided to trust its recommendation on the basis of these judgments. In other words,
rather than direct deliberation on the merits of the alternatives, the CA process itself
acquired a trustworthy reputation and this gave reason for voters to support its
recommendation. It is striking, however, that the combination of these two judgments
about the CA was not held by many members of the electorate but was segmented: some
people thought that the CA was trustworthy because it was composed of ordinary
citizens. Others thought it trustworthy because of its acquired expertise. Despite this
segmentation however, there was something of a consensus that the recommendations of
the CA deserved a trust (if not deference), even if there was division on why that trust
was warranted. (Numbers are needed from the Johnston-Cutler chapter).
Political scientists may recognize this phenomenon as an instance of the typical
relationship of a member of a legislature to her constituents.20 It is the view that Edmund
Burke expressed to his constituents at Bristol -- that they ought to trust him to make the
right decisions and not expect him to parrot their own, less informed views. But Burke’s
view is usually understood as elitist and anti-democratic. Perhaps, then, the CA process is
no more democratic in the end than the ordinary political processes that it is meant to
supplement or reform.
There seem to be two lines of response to this worry. The first is to admit that,
yes, the CA or at least its staff and some of its members rapidly became a new and
influential elite, capable of exerting power on a significant issue. But, because of the
composition of the CA and the way it actually worked, this new elite was sharply
20 The classical reference for the US is Richard Fenno, Home Style, Boston: Little Brown, 1978. See also
Bruce Cain, John Ferejohn, Morris Fiorina, The Personal Vote, Cambridge: Harvard University Press,
1987.
15

separated from the existing political establishment and in structural conflict with it on an
issue about which politicians normally care deeply. And, this conflict offered ordinary
voters a choice they do not often get: to choose between the positions of genuinely
competing elites. This is a kind of neo-Schumpeterian argument that has, at best, a
shallow democratic pedigree but it does see the CA process in ways that I think the
participants might have seen it: as a genuinely new mode of taking an important and
technical political decision by opening the deliberative process to people outside of the
political establishment.
The second response is to try to articulate a new model of public deliberation in
the “public sphere,” one that recognizes the inevitability of a political division of labor
and of relations of power and deference within deliberative processes. This is what I aim
to do in the following section.
IV. Deliberation in Democracies
Lots of deliberation actually takes place in democracies. It takes place in widely
varying institutional settings and, of course, many of those settings are “political” in the
sense that some participants can take actions based on power rather than reason. And
others of these settings are highly technical as well and only a few specialists – many of
them neither elected nor politically appointed – are able, or are thought to be able, to
participate meaningfully in the discussions. Between the temptations of power and the
need for expertise there seems little room for truly public consideration of important
decisions even within a representative legislature let alone among the public at large.
Indeed, it seems a rare thing for the legislature to meet and deliberate about and decide an
16

issue simply on the basis of the better reasons. At the end of the day the legislative
majority, if it is sufficiently disciplined, has the power to impose its will whether or not it
has the better reasons.
Democratic deliberation is imperfect in this sense. But political institutions are
not equally situated in this respect (cite to Warren’s chapter). There is a reason why the
legislative majority could feel entitled to impose its will on an issue: it must, later on,
stand before the electorate if it wants to be returned to power as a majority and be held
accountable for its actions in office. It has reason, therefore, to take account of this fact
in deciding what to do on particular issues. So, to point to legislative deliberations as
defective because marred by power relations is to miss the wider deliberative context in
which the majority is accountable to the electorate. That is not to say that this wider
electoral context is fully deliberative either – it may not be deliberative at all -- it is only
to point to the need to evaluate the whole “deliberative system” in the light of the
political circumstances within which it operates ( cite to Bohman’s chapter).
Aristotle thought that a legislative assembly was the place for deliberative speech
– for deciding together which future options should be chosen. Courts were, by contrast,
the place for forensic speech, for deciding what has already happened and for making
judgments of praise and blame about something that has occurred in the past. Legislative
assemblies are, in this way, forwarding looking, and courts, backward looking. Rawls
famously turned this notion on its head when he said that the (US) Supreme Court was an
exemplary deliberative institution. He could have meant two separate things: first, he
could have thought that Aristotle’s view of courts was simply too narrow: perhaps it fits
ordinary trial courts, with which Aristotle was familiar, well enough, but higher level
17

courts do more than decide cases. They make law and, in this sense, legislate for the
future when settling past disputes. If that is right it is no surprise that such courts, which
did not exist in Athens, deliberate in much the same way that a legislature does. But we
can take a contemporary Aristotelian, who is familiar with modern governmental
institutions, to mean something very different: that courts, within a democratic political
system, are forced to operate on the basis of reasons to a greater extent than are
legislatures. This is the idea I want to explore.
Actual democracies divide political labor in various ways: they separate powers in
one way or another between legislatures, courts, public agencies, the executive, and of
course ordinary voters. The relations among all these entities are complex of course and
they are not my subject here. Insofar as we are speaking of democracies there must be
some sense in which, ultimately, all of the other offices are responsible directly or
indirectly to the citizens.21 But the chain of responsibility itself is highly variable in
other respects. More important in this context is that these different institutions are
characteristically confronted with what I will call distinct deliberative expectations:
public expectations as to how much justification or reason is supposed to accompany an
action.
Let’s start with courts: courts are expected to deliberate and are given elaborate
deliberation-forcing procedures to ensure that they do. Secondly, when a court makes a
decision it is expected to offer its reasons for the decision – at least this is true of
appellate courts that are made up of unelected judges. The appellate system within which
courts are embedded plays a part in enforcing this expectation that good reasons will be
21 The way in which these responsibilities are to be discharged is also institutionally specific and I will not
elaborate this here. What it is that a judge is responsible for doing differs from that for a legislature. And
the modes by which each may be chastised or removed differ as well.
18

given. But beyond the appellate system court orders generally need to be complied with
and respected if they are to have effect. Giving good reasons grounded in law is the best
tool that judges have to ensure that their orders will be respected by others and, therefore,
have real effects. Put another way, because courts have no direct access to legitimate
political power, they are forced to rely on reasons.
One step below courts are public agencies. Agencies are delegated powers by
legislatures and their leaders politically appointed. They are generally expected to make
decisions based on reasons – legal and factual – of roughly the kind that could be offered
by a court. But, agencies are usually not generally constrained by evidentiary procedures
in the way that courts are and usually their decisions are less constrained by the need that
the reasons offered for a decision actually be good reasons than courts are. In the United
States for example, except when a constitutional issue is at stake, agency decisions need
only satisfy what is called the rational basis test: in effect there must only be some
plausible rationale for the decision even if that rationale is not very convincing and no-
one actually believes it. Otherwise, whether the reasons are good or bad, courts will
generally let the agency action stand.
Legislatures face even fewer deliberative expectations. No doubt the members try
to give persuasive reasons in parliamentary debate – perhaps not so much aimed at
persuading the minority as to persuade constituents – but at the end of the day, the
majority can impose its will. The members of the majority must eventually stand to
account of course before the electorate but few political observers think that this fact
seriously disciplines the relation between particular legislative actions and the reasons
19

given for them. Voters must decide on keeping or changing a majority that has a long
record of failures and successes, all things considered.
Finally we come to the voter. Voters are not expected to give reasons for their
votes. Indeed, in modern democracies they are effectively prevented from doing so by
institutions such as the secret ballot and other protections of the “sanctity” of the polling
place. So voting is a pure exercise of power or will and cannot be seen as a form of
persuasion or deliberation at all. Rousseau expressed this notion very clearly in the
Social Contract where he prohibited citizens from speaking or even asking questions in
the assembly: they were there simply to listen to argument and to vote.22
I argue that deliberative expectations are inversely correlated with democratic
pedigree.23 The people, as voters, are not required to present reasons for their actions and
in effect are actually prohibited from doing so by such institutions as the secret ballot and
prohibitions against campaign activity near polling places. The legislature, which is
separated from the people by “one degree,” is informally expected to give reasons of
some kind but there is no requirement that their actions are closely connected to these
reasons or even that there are reasons given at all. Moreover a failure to give reasons does
not invalidate legislation enacted according to constitution. Reasons are, in this sense,
optional in the legislature: if you have the votes you don’t need reasons too.
22 It seems likely that Rousseau’s endorsement of a kind of nondeliberative legislative process was a
concession to defects in human nature or likely practice. He thought that an uncorrupted polity would
permit open and free debate and public ballots as he thought Rome did in its early days. But as corruption
sets in (by which he meant the appearance of factions or partial interests), it is simply too dangerous to
permit the legislature to conduct itself in this open manner and it is better to place strict limits on its
procedures: the members are to listen to proposals prepared by magistrates and then vote in secret without
any prior discussion. It is of some interest to note that the Athenians used precisely these procedures for
court proceedings while permitting open debate and open voting in the Ekklesia.
23 This argument is exposited in AConstitutional Adjudication: Lessons From Europe,@ (with Pasquale
Pasquino) University of Texas Law Review, 2004, vol 82 (June 2004), 1671-1704.
20

Agencies, which are separated from the people by “two degrees,” typically face
more exacting deliberative expectations. They need to give reasons and those reasons
ought to be somehow justify their actions, in the sense of showing how the agency has
the authority, under the legislative statutes, to take the action it did. And finally courts,
which are often separated completely from the people (if they are filled by examinations)
must give reasons that are even more closely connected to the actions themselves;
reasons that can persuade others to follow. Failure to do so may lead to an invalidation of
their holding by a higher court, or to its being ignored by other political actors or by
members of the public. Legal authority is, in this respect, tied closely to judicial holdings
being shown to follow validly from statutory and constitutional law (and perhaps
morality as well).
Having laid out this simple model we can begin to ask about the place of the
citizens’ assembly and, in the next section, about the ways in which the deliberative
expectations that it faced were met or not met (cite to chapters of Bohman, Thompson,
Chambers and Rattner). From a narrowly legal standpoint, the CA was established by
executive action and legislation and in this way is placed in the same way that an agency
or a court might be, at least two degrees from the people. But the CA is different from
agencies and courts in another crucial respect: it was composed in a way as to be
representative of the people in much the way legislatures are composed (cite to chapters
of Thompson and Warren). But the mode of choice was lottery rather than election and,
for this reason, it was much more representative of ordinary Canadians than an elected
legislature could ever be. Secondly, it was charged with deciding an issue in which its
21

members had no venal or partisan interest whatsoever, and was in this respect similar to
courts.
It seems a tricky question where to locate the CA in the terms I have laid out.
Insofar as the CA is itself an institution of direct democracy – and its actions direct
popular expressions – it would have little reason to offer justifications: their
recommendations were simply those that ordinary voters would have made had their
names been drawn and had they taken the time to participate in the process. Of course,
ordinary voters – who support is needed if the CA proposal was to become law – did not
participate in the deliberations, and a reason-free recommendation might not be
persuasive to them. That is, unless those ordinary voters found reasons to trust the
participants in the CA. (acknowledge disagreement among authors on this point:
Chambers has the view that voters must assess the actual alternatives whereas Warren
thinks that it is sufficient that voters trust the CA). Moreover, it is unlikely that many
ordinary voters would have the time, inclination or competence to evaluate reasons for or
against the specific proposals even if they were fully presented (but see some evidence in
Johnston and Cutler). As I suggested earlier, in many respects, this relationship of trust
without reason-giving is similar to the situation of the ordinary legislature which has only
a weak reason-giving requirement (cite to Mansbridge).
In this case, however, voters probably had more reason to trust a recommendation
from the CA than they would have had to trust a recommendation from the legislature. At
least this seems true in the special case of a recommendation about electoral reform,
where members of the legislature and political parties have special interests to protect. I
mentioned already two grounds from this trust: the first “structural” ground is that the CA
22

was, in fact, made up of ordinary Canadians and not professional politicians and this fact
was widely believed. The second is the public evidence presented by the CA during its
proceedings and during the ensuing campaign that its members had taken their task
seriously, had become technical experts in electoral reform, and had deliberated hard to
produce a sensible recommendation (see chapter of Blais, et al). There was, in effect,
good reason for ordinary voters to believe that the recommendations of the CA were
sound and deserving of support even if most could not evaluate them directly (cite to
Chambers who is optimistic that voters can actually deliberate about the alternatives). On
this account the CA faces substantially weaker deliberative expectations than the
legislature would have, at least if it can establish itself as trustworthy in the two ways we
have described. If this is right, we ought to take the CA model as a serious attempt at
direct democracy and not merely as the creation of an ad hoc legislature or as a special
kind of blue ribbon commission.
V. Internal and External Deliberation
I have argued that the voters had good structural reasons to trust the CA’s
recommendations: the composition of the CA itself as visibly made up of ordinary
citizens seems especially significant. Johnston’s and Cutler’s research done during
campaign process suggests that voters were generally convinced that the CA was indeed
made up ordinary citizens. And, elected politicians were excluded from any role in the
CA process. These were not, of course, choices made by the CA itself but were made in
kind of self denying ordinance by the politicians who set it up. This is not to say that
there was nothing the CA, or those who established it, could do to give the voters
23

additional (nonstructural) reasons to trust its recommendations. I think there were
several important steps, all of which were taken as far as I can see, that led to an
establishment of a high level of trust. The way the CA conducted itself – holding lots of
public hearings, members speaking to community meetings to explain its process and its
reasoning – seem also to have been significant in reassuring voters that the process was
substantially controlled and directed by the members of the CA and not by the staff,
political party leaders or interest groups. Moreover, the openness of the process gave the
public and the press plenty of chance to see the developing expertise of CA members.
And at least to a significant fraction of the public, the reassurance of technical mastery
was important. At least that is the suggestion of the early research findings at this point in
time.
Pasquale Pasquino and I, in earlier work, have called this form of deliberation
external, in that it exposed the deliberative process to public examination by media,
politicians and ordinary voters if they chose to watch.24 We contrasted it with internal
deliberation in which members attempt to reason among themselves, in camera, to
produce a joint recommendation on behalf of the whole body. The internal form of
deliberation is typical of the European Constitutional courts, the subject that Pasquino
and I studied. It was also the choice of the US Constitutional Convention in Philadelphia
and indeed was the mode of deliberation in the early American Senate. External
deliberation was employed by the French Constituent Assembly, the proceedings of
which took place openly amid the Parisian mob.25 And, perhaps ironically, it is a form of
24 Ferejohn and Pasquino, op cit.
25 For a perceptive consideration of alternative deliberative models see Simone Chambers, “Behind Close
Doors: Publicity, Secrecy, and the Quality of Deliberation”, Journal of Political Philosophy, Vol 12 (2004)
389-410. Her focus is largely on what we call internal deliberative models.
24

deliberation employed, to an extent, by the American Supreme Court with its practice of
making visible multiple voices on most important issues.
There is much to be said in favor of internal deliberation: it permits participants to
speak to each other without an external and poorly informed audience to impress.
Members of internally deliberative body are free to change their minds or view without
worrying about their public reputations for inconsistency. Moreover, internal
deliberation works to shield members from pressures of bribery or intimidation from
external groups. Perhaps, for this reason, we see that courts sometimes move toward the
internalist model when their members feel threatened by external actors. The Supreme
Court of John Marshall faced with the hostility of the Jeffersonians adopted this tactic.
And the post World War II Constitutional Courts in Germany and Italy did the same (the
German Court opened up its process a little in 1969 when it decided to permit dissenting
opinions. The Italian Court recently refused to go this far, essentially for the reasons
presented in this paragraph. In any case, internally deliberative processes may provide an
environment in which issues can decided free from coercion and temptation and in that
way approach what Habermas has called an ideal speech situation. At least from the
viewpoint of those in the room.
External deliberation aims to convince those who are not in the room. And,
insofar as the CA was only part of the public process by which the electoral reform was
to be decided, it was forced to adopt an externalist stance. And, by adopting a largely
externalist deliberative practice, the CA was able to convince ordinary voters that its
recommendation could be trusted.26 The risk of such a strategy was of course that
26 The CA did hold some discussions in groups that met in non-public sessions. I believe, however, that
most of their important deliberations fit the externalist model better than the internalist one.
25

internal divisions in the CA would be exposed and exploited and, to an extent, some of
this occurred. A small number of members of the CA opposed its principal
recommendation preferring the MMP system that was rejected at the end. And some of
them campaigned prominently against the STV proposal. That is always a risk of
externalist deliberation.
VI. The CA Model
The BCCA was assembled to work on electoral rules, an issue that is particularly
difficult for representative democracies to handle credibly. Could it be a useful model for
democratic participation more generally? I believe, as I have said, that in some respects,
the structure of the CA model exemplifies a deeply attractive form of popular decision
making. It separates the proposing of legislation from its acceptance, permitting a great
deal more specialization and expertise in the proposal stage than in the ratification stage.
This separation is familiar from many settings. The American constitution itself was
proposed by a body of delegates who met for months to develop and refine the proposed
constitution an then ratified in popular conventions elected on a very wide franchise. It is
noteworthy that the CA model bears a striking resemblance to the structure of Athenian
democratic institutions. The Athenian Ekklesia – the assembly – could only take up
proposals that had been considered first in the council or Boule. The Boule was chosen
by lot (from voting districts) and great care was exercised to ensure that its internal
processes could not be captured by a small group. For one thing, no-one could serve on
the Boule more than twice in his lifetime. For another, the presidency of the Boule was
rotated among the ten “tribes” over the course of the year.
26

Before rushing to an affirmative answer, however, it is well to keep in mind the CA
model has some special advantages in dealing with electoral rules that would not
necessarily be available to it in other cases. Most importantly, when it comes to
establishing or changing electoral rules, incumbent politicians are automatically
suspected of being “judges in their own cases.” They are, in effect, morally if not
factually disqualified from managing this issue. This leaves the door open to some other
body to do the job – perhaps judges or a panel of academic or technical experts or, as in
BC, a panel of ordinary citizens. Thus there is something special to say, in the case of
electoral rules management, for each of these alternatives. Judges are specialists in
managing procedures aimed at getting right answers (even if they have no expertise about
elections). Experts are presumably especially able to make choices among rules based on
the best available knowledge. The fact is, however, that the choice of electoral rules
involves trading off some values against others and expertise, whether procedural or
substantive, is not is not adequate to settle these issues. It seems that a citizens’ body has,
on this account, a unique competence that the other models cannot claim (cite to
discussions of citizen’s competence by Baum, Thompson, and Blais, et al.): it can claim a
right to base its recommendations on its own substantive value judgments as well as on
relevant information from the community of experts.
But, in ordinary cases, where the legislature is not thought to have a special interest
separate from the public interest, it is not so clear that the CA model is so attractive.
After all, something is lost by moving an issue away from a multipurpose legislature to a
randomly chosen body of citizens assembled to deal with that single issue. We lose at
least the experience and political ability that elected leader can be expected to have
27

developed for example. We lose the institutional capacity that a legislature has to make
complex multi-issue choices that embody tradeoffs in law. We also lose some of the
agility and flexibility that professionally staffed governmental processes might be
expected to display. And, if we make many decisions by a CA process, we probably lose
the ability to hold officials responsible for how government policy has been working.
After all, when the CA is finished, the members go back the farms as Cinncinatus did.
Why pay these prices if there are no strong reasons to distrust politicians to decide the
issue?
There seem to be two reasons on offer. One is that representative democracy as
presently constituted is simply not adequately democratic. The principle of election and
electoral competition reliably leads to a kind of elective aristocracy and an institution that
corrects that tendency is attractive for that reason alone. Indeed there are attempts in
other Canadian provinces that offer seek to expand the CA model in a way that would
expand that locus of institutional competition.
The second reason is that we already have numerous institutions of direct
democracy and each of those are quite deeply flawed in ways already discussed. So we
have already, in important ways paid the price I alluded to and the question now is, what
have we got in return for that payment. Perhaps the CA model can provide a way to
correct or supplement those institutions in a way that redeems their original promise.
Here is what I have in mind. In California for example, the popular initiative has
essentially no subject matter limitations. And if, as I suggested above, that institution is
deeply flawed as a democratic institution, the CA process may offer a path of reform that
could actually help it to realize its original promise.
28

As pointed out earlier, the initiative process is flawed largely because it is so easily
captured by interest groups and public officials. It is worth it for an interest group to pay
the price to capture the proposal power precisely because the right to make a take it or
leave it offer to the electorate is extremely valuable – especially when the status quo
policy is unpopular. But suppose that whenever an initiative is proposed, a CA would be
automatically convened to deliberate about and possibly amend the proposal. Like the
BCCA, this CA would be entitled to a competent professional staff, have the authority to
call witnesses, and its members would receive a reasonable level of compensation,
adequate to make it an attractive task for most voters. It would be expected to take the
time necessary to establish a high enough level of competence to allow it to draft an
unbiased recommendation to the electorate.
Insofar as the CA is statistically representative and conducts fair and thorough
hearings and deliberation, and develops a reasonable level of expertise, one would expect
that the CA process would have the effect of making the amended initiative more
attractive to the median voter than the original proposal would have been.27 This ought
to have two effects. First, it would probably make the proposal more likely to pass, both
because it is more attractive to the median and because voters have no reason to suspect
that it is the proposal of a special interest group. Second, the CA process itself would
diminish the value to special interests of proposing an initiative to the ballot as they
would lose the capacity to control the language and legal effect of that initiative (unless
27 I don’t think this claim can be more than a statistical expectation. After all, the members of the CA have
participated directly in an intensive deliberative process that may well have transformed their preferences
and beliefs on the issue and the electorate has not. Insofar as the CA is externally deliberative some
members of the general public may change their views in the same way. But the evidence in this volume
suggests that an indirect process is more plausible whereby voters come to trust the CA to be both
“representative” and competent. That is only conjecture at this point.
29

they could persuade the CA that their language is in the public interest). If this is right
there should be much less abuse of the initiative process by special interests. Of course,
there is no reason to expect that a CA process would make it hard to enact proposals that
target unpopular minorities such as those accused or convicted of crimes or of terrorism.
The Citizens’ Assembly model offers two broad pathways of reform. First, as it
was initially implemented, it provides a way to make certain specific political decisions
that are poorly made in ordinary political processes. That is, it provides a model for
taking some issues out of ordinary politics. Electoral reform, redistricting, campaign
finance law, and the regulation of political speech are not well managed by self interested
politicians. There are good reasons to remove them from ordinary politics and place
them in the hands of some independent and institution: a panel of judges perhaps or else a
blue ribbon commission of distinguished citizens. But panels and commissions of this
kind have their own weaknesses precisely because they are elite dominated. They may
lack imagination or boldness; and they may lack democratic legitimacy and so their
proposals may be pedestrian or politically irrelevant. The CA offers a way to cure these
problems. Its great contribution was to show that it was possible to construct a
democratic institution that not only exhibited independence and impartiality (this was no
surprise given how its members were selected) but also developed a high level of
technical competence with respect to a difficult subject matter.
The second contribution of the CA model builds directly on its democratic features.
The CA provides a way to redeem, to some extent, the ancient promise of democracy as a
form of popular government -- in Lincoln’s words, as government “by” the people. The
CA provides a way for ordinary people to play a regular role in public life outside of the
30

channels of elite or Schumpeterian democracy. It demonstrated that this role can be
played responsibly and with intelligence. By demonstrating that this power to propose
could be rationally and deliberately exercised by a genuinely democratic institution, the
CA offers a new pathway of democratic reform. This path nee not be restricted to the
narrow range of issues on which people do not think politicians can be trusted to act in a
publicly spirited way. Rather, it offers a way for a democracy to consider and revise its
fundamental laws: framework laws that a people need to see as “their own.” The CA
offers, in this way, a way of vindicating some notion of public authorship without, at the
same time, falling prey to plebiscite or mystery.
31