Liquid Democracy: Potentials, Problems, and Perspectives

Christian Blum, Rutgers University (chrstnblm@googlemail.com)

Christina Isabel Zuber, Universitt Konstanz (mail@christinazuber.com)

Accepted for publication in The Journal of Political Philosophy on 1 May 2015

Abstract
Pirate Parties are suggesting a solution to citizens dissatisfaction with representative
democracy that seeks to remedy the system from within. By combining direct democratic
participation with a highly flexible model of representation, liquid democracy promises the
best of both worlds: Citizens can freely choose to either vote directly on individual policy-
issues, or to delegate their votes to competent representatives who vote on their behalf. This
delegation is policy-area specific and can be retracted instantly. So far, democratic theory has
neglected this model of democratic decision-making. This article fills this gap. We first define
the basic model of liquid democracy. We then defend two theses: First, liquid democracy
mobilizes more political expertise than purely representative democracy; second, liquid
democracy is more egalitarian than the latter. We draw on social epistemology, arguments
from collective intelligence, and empirical research on voter competence to argue that we can
expect ordinary citizens to master their tasks in a liquid democracy. Finally, we discuss two
problems affecting the basic model of liquid democracy: a problem of unequal voting power,
and a problem of policy-inconsistency. They can be addressed by combining liquid democracy
with a trustee model of representation, requiring decision-makers to adhere to deliberative
norms, and balancing liquid decision-making in legislatures with an executive that reviews the
formal feasibility of policies and moderates package deals between proposals from different
policy areas.

Acknowledgments
The original ideas for this paper were developed during a workshop on liquid democracy that
was organized by the authors at the University of Cologne in spring 2012. We are indebted to
all participants of the workshop, but in particular to Saskia Ruth, Holger Reinermann, and
Gregor Zons for their contributions during the inspiring discussions we had. We would further
like to thank Ren Rderstein for giving us the opportunity to present insights from the
workshop to members of the Cologne Pirate Party and receive their feedback. Previous
versions of this paper have been presented at the PDD Specialist Group Conference, 9-11 July
2014, Newcastle, England and at the 4th Global International Studies Conference, 6-9 August
2014, Frankfurt am Main, Germany. We wish to thank Sergiu Gherghina, Bob Goodin and
three anonymous referees of the Journal of Political Philosophy for their excellent comments
that have helped us improve this article. Both authors wish to acknowledge financial support
received from the Fritz Thyssen Foundation while drafting the manuscript.

1
1. Introduction

The first pirate party was founded in Sweden in 2006. Since then, many have followed the

example, and the non-governmental organization Pirate Parties International (PPI) now

counts members from 43 different countries. Pirate parties have gained seats in local, regional

(e.g. Berlin, North Rhine-Westphalia, Saarland, Schleswig-Holstein), and national (Iceland,

Czech Republic) parliaments, as well as the European parliament. These parties are mostly

known for opposing Internet censorship and defending transparency, civil rights, and open

content. However, they are also suggesting a model of collective decision-making that seeks to

remedy democratic systems from within, namely liquid democracy, a model they are already

applying for intra-party decision-making.1 By combining direct democratic participation with

a highly flexible model of representation, liquid democracy promises the best of both worlds:

Citizens can freely choose to either vote directly on individual policy-issues, or to delegate

their votes to issue-competent representatives who vote on their behalf. This delegation is

policy-area specific and can be retracted instantly. 2

So far, normative democratic theory appears to have neglected this model of

democratic decision-making. The substantive idea was introduced into the academic debate by

James C. Miller in 1969 under the somewhat unwieldy term of a program for direct and

proxy voting.3 Millers work received very limited attention from contemporary authors with

1
See Pirate Parties International, About the PPI, http://www.pp-international.net/about (2009), retrieved
4/29/2014; Anna Litvinenko, Social Media and Perspectives of Liquid Democracy: The Example of Political
Communication in the Pirate Party in Germany, Proceedings of the 12th European Conference on
eGovernment, ed. M. Gasc (Reading: Academic Publishing International Limited, 2012), pp. 403-407, and
Simon Franzmann, Aufgespiet. Wie der Erfolg der Piratenpartei Gesellschaft, Politik und
Politikwissenschaft herausfordert, Mitteilungen des Instituts fr Deutsches und Internationales Parteienrecht
und Parteienforschung 18 (2012), pp. 123-126.
2
See Bryan Ford, Delegative Democracy, Manuscript (2002), http://www.brynosaurus.com/log/2002/0515-
DelegativeDemocracy.pdf, retrieved 5/13/2014.
3
James C. Miller, A Program for Direct and Proxy Voting in the Legislative Process, Public Choice 7 (1960),
pp. 107-113, at p. 107.

2
just one rather skeptical review that Martin Shubik dedicated to his proposal. 4 Since the

early 2000s, the model has been rediscovered and discussed alternatively under the terms

liquid democracy, delegative democracy, or proxy voting.5 More recent contributions have

added further detail to Miller's proposal and have aimed to defend liquid democracy with

indirect and direct arguments. The indirect argument holds that liquid democracy comes

closest to the ideal of complete direct democracy by making compatible the better features of

direct voting with the practical necessity of some representation. 6 Direct arguments hold,

first, that liquid democracy improve[s] the legislatures performance 7 by allowing members

to select issue-competent delegates, rather than representatives who decide on all issues

regardless of expertise. Second, it is more democratic 8 than its representative counterpart

since the entitlement to either vote directly or to delegate votes enables members to

participate fully9 in political decisions. However, the arguments have so far remained highly

general in nature, and the authors have not pursued the aim of justifying liquid democracy

explicitly based on criteria derived from normative democratic theory. 10

We seek to fill this gap. We first define the basic model of liquid democracy through

four properties: direct democracy, flexible delegation, meta-delegation, and instant recall. We

then offer a democratic-theoretical justification of liquid democracy, understood as a

procedure for democratic decision-making at the systemic level.11 Normative democratic

4
Martin Shubik, On Homo Politicus and the Instant Referendum, Public Choice 9 (1970), pp. 79-84.
5
See Ford, Delegative Democracy; Dan Alger, Voting by Proxy, Public Choice 126 (2006), pp. 1-26; and
James Green-Armytage, Direct Voting and Proxy Voting, Manuscript (2014),
http://inside.bard.edu/~armytage/proxy.pdf, retrieved 5/13/2014.
6
Miller, A Program for Direct and Proxy Voting, p. 107; see also Ford, Delegative Democracy, p. 1.
7
Alger, Voting by Proxy, p. 9.
8
Green-Armytage, Direct Voting and Proxy Voting, p. 6
9
Ibid.
10
In addition, Fords Delegative Democracy and Green-Armytages Direct Voting and Proxy Voting are draft
manuscripts reflecting work in progress.
11
The interaction between liquid democracy as a procedure for intra-party decision-making and an overall
representative system is thus not the subject of this paper. We also set aside all normative issues implied by
the practical application of liquid democracy through software tools such as LiquidFeedback, including inter
alia the problem of secret voting. These constitute important topics for future considerations.

3
theory claims that we can identify criteria in virtue of which some variants of collective

decision-making are better than others. These criteria specify which primary function

collective decision-making must fulfill, and they allow measuring the goodness of different

variants of decision-making based on how well these variants fulfill that function.12 Two major

competing accounts provide such criteria: the epistemic account and the equality-based

account. They paint different and, arguably, incompatible pictures of what is at stake in

decision-making and presuppose distinct notions of the common good. 13 However, proponents

of liquid democracy need not opt for one account. We argue that their basic model is

preferable to representative democracy in terms of the criteria offered by both approaches: (1)

liquid democracy mobilizes more political expertise than purely representative democracy; (2)

liquid democracy is more egalitarian than representative democracy. We choose representative

democracy as the normative benchmark because we follow Steffen Ganghof's argument that

the justification of democratic institutions should be comparative, judging whether they can

fulfill their purpose more effectively and efficiently than the best alternative. 14

However, unlike representative democracy, liquid democracy puts high demands on

ordinary members of the political community who are expected to select competent experts on

the epistemic account and to choose representatives that best further their subjective interests

on the equality-based account. We argue that social epistemology, collective intelligence, and

the literature on voter competence provide grounds for optimism that they are likely to

perform these tasks in a satisfying way.

12
Unlike many authors writing about democratic theory, we prefer to use the basic normative notions
goodness and betterness in this context, rather than legitimacy, because we consider legitimacy to be a
non-gradual property of political systems. On our reading, political systems are either legitimate or
illegitimate, but never more or less legitimate than others. They can, however, be better or worse than others
by degrees.
13
See Christian Blum, Why the Epistemic Justification of Deliberative Democracy Fails, Challenges to
Democratic Participation. Antipolitics, Deliberative Democracy, and Pluralism, ed. Andre S. Campos and Jos
Gomes Andr (Plymouth: Lexington Books, 2014) pp. 47-65; at p. 56.
14
Steffen Ganghof, Equality-Based Comparison: How to Justify Democratic Institutions in the Real World,
Politics 33 (2013), pp. 101-111; at p. 106

4
 In addition, liquid democracy encounters the challenges of (1) avoiding to recreate

representative democracy's problem of two classes of citizens, though in a different way (the

problem of unequal voting power) and (2) achieving consistency across the whole set of

decisions being made across different policy areas (the problem of policy-inconsistency). We

therefore suggest a modified account that combines liquid democracy with a trustee model of

representation, requires decision-makers to adhere to deliberative norms, and balances liquid

decision-making in legislatures with an executive that reviews the formal feasibility of policies

and moderates package deals between proposals from different policy areas.

The next section introduces the basic model of liquid democracy. Section three

provides an epistemic justification of liquid democracy while section four defends liquid

democracy applying the criteria of the equality-based account. Section five argues that

ordinary members of the political community can be expected to competently perform their

liquid democratic tasks. Section six suggests modifications of the basic model in light of the

problems of unequal voting power and policy-inconsistency. Section seven concludes the

article and points to topics that should be addressed in the future.

2. The Basic Model of Liquid Democracy

Liquid democracy is a procedure for collective decision-making that combines direct

democratic participation with a flexible account of representation. Its basic model consists of

four components that can be stated as follows: All members of a political community that

satisfy a set of reasonable participatory criteria (adulthood, baseline rationality) are entitled to

(I) directly vote on all policy issues (direct democratic component)

5
 (II) delegate their votes to a representative to vote on their behalf on (1) a

singular policy issue, or (2) all policy issues in one or more policy

areas, or (3) all policy issues in all policy areas (flexible delegation

component)

(III) delegate those votes they have received via delegation to another

representative (meta-delegation component)15

(IV) terminate the delegation of their votes at any time (instant recall

component)

These specifications are best illustrated by an example. Consider a member A who ponders on

three upcoming policy decisions. The first decision concerns a bill that imposes stricter

sanctions on tax dodgers in order to drastically reduce fiscal evasion in the future. According

to (I), A is entitled to directly vote on this issue; as luck would have it, she is an expert in this

field (maybe she majored in tax law), and thus it seems natural that she would do so. The

second decision is about the authorization to cultivate genetically enhanced corn variety. A has

a keen environmental conscience, but possesses little knowledge about this policy area.

However, she is closely acquainted with B who is a renowned biologist with a background in

bioethics. Since A trusts Bs judgment and (II) entitles her to delegate her vote to B, it seems

sensible that she should authorize him to vote on her behalf on issues that pertain to

environmental policy. As a consequence, B now possesses two votes that he can bring to bear

on the issue at hand: As vote and his own. The third decision concerns a trading agreement

with another country. A knows little about said country or about economic policy; and neither

is she acquainted with any expert in this field. However, she knows another member C who is

15
The meta-delegative component is not featured in Millers original account. Ford added it to the basic model
(see Ford, Delegative Democracy, p. 10).

6
well connected with a number of specialists on international relations and economics. Since

(II) entitles her to delegate her vote on this issue to C, and since (III) entitles C to delegate As

vote to a further member of the community, A would do well, it appears, to transfer her vote to

C with the express instruction that he transfer it to a specialist of his choosing. Once all

decisions have been made, A surveys the results. Doing so, she might unearth new information

that cast doubts on whether B has used her vote with her best interests in mind (maybe he was

swayed in his decision by a lobby group). Realizing this, A makes use of (IV) and terminates

her representative relationship with B. Although this does not affect the previous decision, it

deprives B of As voting power regarding future decisions.

The basic model of liquid democracy stands in stark contrast to the classical model of

representative democracy. According to the latter, political decision-making proceeds in two

stages: In the first stage, all members elect representatives who compete for votes with

different political programs for a fixed term. In the second stage, the elected representatives

directly vote on all policy issues at hand; and they do so with equal voting power. 16 Thus,

representative democracy divides the members of a political community into two political

classes: On the one hand, there is the class of members who authorize other members to vote

on policy issues on their behalf, but do not vote on policy issues themselves; this class is

commonly labeled the class of principals. On the other hand, there is a group of members who

are authorized by others to vote on their behalf on all issues, namely, the class of

representatives.

We can begin delineating the differences by noting that, first, liquid democracy does

not require members to be authorized via election in order to vote on policies. All members
16
For normative arguments in favor of combining proportional representation for selecting representatives with
majority vote to decide on policy issues in legislative assemblies, see Steffen Ganghof, Politische Gleichheit
und echte Mehrheitsdemokratie. ber die normativen Grundlagen institutioneller Arrangements, Zeitschrift
fr Politikwissenschaft 15 (2005), pp. 741-763, drawing on Thomas Christiano, The Rule of the Many.
Fundamental Issues in Democratic Theory (Boulder, Co: Westview Press, 1996). We take this combination
that Ganghof terms real majoritarian democracy as the model when referring to representative democracy.

7
are entitled to vote on all policies irrespective of whether they represent other members.

Second, it allows for area-specific representation rather than representation with respect to

all policy issues such that members directly authorize experts to vote on their behalf with

respect to a limited policy area; or indirectly authorize experts by conferring their vote to

members who possess meta-expertise and can select experts to vote on their behalf. Third, it

makes short shrift of turn-based representation by allowing principals to immediately retract

their votes when their representatives have not voted to their satisfaction. Fourth, it allows that

representatives can have unequal voting power, since their voting power is a function of the

votes conferred upon them plus their own vote.

3. The Epistemic Justification of Liquid Democracy

The epistemic account of democracy is advocated, among others, by Carlos S. Nino, David M.

Estlund, Fabienne Peter, Christian List & Philip Pettit, and Hlne Landemore. 17 It is based on

two premises that may be called the error premise and the epistemic premise:

1. error premise: the members of a political community can err about matters

of the common good.

Members of a political community can err about matters of the common good and thus

authorize political actions that are detrimental to their collective welfare. Causes for such

errors are: lack of information, false information, inconsistent inferences drawn from correct
17
Carlos S. Nino, The Constitution of Deliberative Democracy (New Haven/London: Yale, 1996); David M.
Estlund, Democratic Authority. A Philosophical Framework (Princeton: Princeton University Press, 2008);
Fabienne Peter, Democratic Legitimacy (New York: Routledge, 2009); Christian List and Philip Pettit, Group
Agency. The Possibility, Design, and Status of Corporate Agents (Oxford/New York: Oxford University
Press, 2011); and Hlne Landemore, Democratic Reason. Politics, Collective Intelligence, and the Rule of
the Many (Princeton, NJ: Princeton University Press, 2012).

8
information, or manipulation by agenda-setters.18 The possibility of erring about the common

good presupposes objective normative standards that transcend the currently held interests of

the members and that can be determined by truth-apt beliefs. 19 If there were no such standards,

it would be conceptually impossible that the realization of the members currently held

interests was ever detrimental to their collective welfare. These standards must be agent-

independent and observer-independent, i.e. their validity neither depends on the judgments or

desires of those to whom they apply, nor on the judgments and desires of anyone in their

environment.20 As expressed in the second premise, these standards allow assessing the

goodness of political decision-making procedures:

2. epistemic premise: political decision-making procedures derive their goodness

from reliably generating true beliefs about the common good.

This view is encapsulated in Estlunds claim that it must count in favor of a social decision

procedure that it tends to produce the better decision. 21 In other words: It is irrational not to

choose a system whose procedures are more reliable in generating knowledge about the

common good than alternative systems. The central function of political procedures is then the

18
See Joseph A. Schumpeter, Capitalism, Socialism and Democracy (New York: Harper & Row, 1976) and
John S. Dryzek, Deliberative Democracy and Beyond. Liberals, Critics, Contestations (Oxford, New York:
Oxford University Press, 2002).
19
See Estlund, Democratic Authority, p. 98.
20
For a more detailed discussion see David Enoch, Taking Morality Seriously. A Defense of Robust Realism
(Oxford: Oxford University Press, 2011), p. 10. We remain neutral on the material question of whether the
common good is best construed via a so-called objective list account i.e. as consisting of a set of
intrinsically valuable goods , a constructivist account i.e. as constituted by the hypothetical agreement of a
group of fully informed and fully rational individuals or some other account. On objective list accounts in
ethics see Thomas Hurka, Perfectionism (Oxford: Clarendon Press, 1993), on constructivist accounts see
Ronald Milo, Contractarian Constructivism, Journal of Philosophy 94 (1995), pp. 181-204. The important
point is that the common good, as presupposed by the epistemic account, is an objective normative entity that
members of the political community can either ascertain or fail to ascertain.
21
Estlund, Democratic Authority, 98. In more technical terms: there ought to be a considerable above 0.5
probability that the collective decision-making procedure indicates p if p is true, and there ought to be a
considerable above 0.5 probability that it does not indicate p if p is false (for a detailed discussion see List
and Pettit, Group Agency, pp. 81-103).

9
identification of normative truths about what is good for a given society. Hence, the epistemic

account provides the following criterion for assessing the goodness of different forms of

decision-making procedures, namely: the greater the epistemic reliability of the procedure, the

better the procedure.

How does liquid democracy fare on this account? We argue that it is more reliable in

generating true beliefs about the common good and choosing corresponding policies than

representative democracy because it has a greater capacity for mobilizing policy area

expertise. A few words on the notion of policy area expertise and its relation to the reliability

of decision-procedures: We hold that a member of a community has expertise with respect to a

policy area if and only if (1) she has considerably more true beliefs and less false beliefs than

the average member about which policies pertaining to that area are conducive to the common

good and which are not; and (2) there is a considerable above average probability that she will

form true beliefs and will not form false beliefs of the above-mentioned types in the future.

Thus, the reliability of a decision-making procedure co-varies positively with its capacity for

mobilizing policy area expertise. The more expertise can be brought to bear on policy

decisions, the more likely is it that these decisions will advance the common good.

The first and main reason why liquid democracy has a greater capacity for mobilizing

policy area expertise than its counterpart is that it allows for policy area specific

representation. In a purely representative democracy, members must authorize representatives

to vote on their behalf on all policy issues regardless of their domains of expertise. Thus, it

is highly likely that, on the one hand, votes will be negatively influenced by decision-makers

lack of expertise; and that, on the other hand, many members who possess more expertise on

the issues at hand get no say. In short: Representative democracy with its one-size-fits-all

model of representation has a tendency to include the wrong kind of decision-makers and

10
exclude the right kind of decision-makers. 22 By contrast, members in a liquid democracy can

handpick representatives to vote exclusively on issues that fall into their domain of

competence. Thus, it is far less likely that votes are cast by representatives who have no

expertise that they can bring to bear on policy decisions. The second reason is that the instant

recall provision allows for the replacement of experts, whose track record proved suboptimal,

by choosing more promising representatives. Consequently, it is possible to effectively and

efficiently filter out the best experts without the time lags of electoral cycles.

We can summarize the argument as a logical inference: if (1) the goodness of political

decision-making procedures is dependent on their reliability in generating true beliefs about

the common good and choosing corresponding policies; and if (2) their reliability is dependent

on their capacity for mobilizing policy area expertise; and if (3) liquid democracy has, due to

its flexible delegation component and instant recall component, a greater capacity for

mobilizing policy area expertise than representative democracy, then (4) liquid democracy is

better than representative democracy. Since we believe that premises (1), (2), and (3) are

plausible, we hold that on the epistemic account, liquid democracy is indeed superior to its

counterpart.

22
It should be acknowledged that representative democracies also have mechanisms to mobilize policy area
specific expertise, e.g. through specialized parliamentary committees and expert hearings. However, two
important differences should be noted: First, comparative studies of parliamentary committees in Western
Europe show that the power of parties and the prestige of the committee (i.e. factors unrelated to expertise)
play the dominant role in appointment; see Erik Damgaard, How Parties Control Committee Members,
Parliaments and Majority Rule in Western Europe, ed. H. Dring (New York: St. Martin's Press, 1995), pp.
308-325. The committees are therefore affected by partisan biases. Second, in the case of expert hearings,
experts are only advisors, not decision-makers, and their influence is conditioned by the willingness of MPs
to take up their suggestions.

11
4. The Egalitarian Justification of Liquid Democracy

The equality-based account of democracy is advocated, among others, by Peter Singer, Jeremy

Waldron, and Thomas Christiano.23 It is based on what may be called the sovereignty premise

and the equality premise:

1. sovereignty premise: the members of a political community shape the welfare of

their community on the basis of their own subjective interests.

Following this account, a political communitys common good is not a normative entity that is

out there waiting to be discovered, as proponents of the epistemic account would have it.

Rather, its members are sovereign authors of their own welfare who have the last word on

deciding what is good and bad for their community.24 Hence, their subjective interests are

constitutive of and prior to the common good itself; and the positive value of a political action

for the common good is a function of the valuing of that action by the members themselves. 25

In light of the heterogeneity of interests in modern political communities, it is highly

unlikely, though, that all members always agree on the content of the common good. This is

where the second premise comes into play:

23
Peter Singer, Democracy and Disobedience (Oxford: Oxford University Press, 1973); Jeremy Waldron, Law
and Disagreement (Oxford: Oxford University Press, 1999); Thomas Christiano, The Rule of the Many;
Knowledge and Power in the Justification of Democracy, Australasian Journal of Philosophy 79 (2001), pp.
197-215; The Authority of Democracy, Journal of Political Philosophy 12 (2004), pp. 266-290; The
Constitution of Equality. Democratic Authority and Its Limits (Oxford/New York: Oxford University Press,
2008).
24
See Christian Blum, Determining the Common Good: A (Re-)Constructive Critique of the Proceduralist
Paradigm, Phenomenology and Mind 3 (2012), pp. 176-188; at p. 180.
25
Following David Sobels standard account (On the Subjectivity of Welfare, Ethics 107 (1997), pp. 501-508),
we define subjective interests as intentional and prospective pro-attitudes that are directed towards states of
affairs that are considered desirable by the bearers of those attitudes.

12
 2. equality premise: political decision-making procedures derive their goodness

from allowing each member to assert her interests in an equal fashion.

Given that the members subjective interests are constitutive of their collective welfare, it

seems reasonable to assume that each members interests ought to be given equal weight in

determining the content of the common good. Christiano argues with great force for this

assumption by stating that [t]his equality proceeds from the importance of interests as well as

the separateness of persons. [...] no ones interests matter more than anyone elses. 26 Hence,

the equality-based account provides the following criterion for assessing the goodness of

different forms of decision-making procedures: the greater the participatory equality of the

procedure, the better the procedure.

How does liquid democracy fare on the equality-based account? There are two respects

in which liquid democracy facilitates a greater equality of participatory resources and is hence

preferable to representative democracy. First, consider again the decision-making model of

representative democracy. By design, the latter facilitates an unequal distribution of

participatory resources insofar as its two stage decision-making process divides the

community in two political classes (see section two): the class of principals, who only

indirectly assert their interests by electing a representative or group of representatives whose

policy proposals best match their own preferences; and the class of representatives who

directly assert their interests by voting on policies. Liquid democracy remedies this inequality

by introducing a direct democratic component, thus allowing all members regardless of

whether they act as representatives to directly assert their interest by voting on all policy

issues.

However, liquid democracy is not only more egalitarian in virtue of its direct
26
Christiano, The Authority of Democracy, p. 269.

13
democratic component. We argue that, second, the flexible delegation component also allows

for a more egalitarian political representation than purely representative democracy.

Consequently, even when members make use of representation (which they most likely will do

on many occasions), they are better off in terms of participatory equality. To see why this is so,

consider the fact that in a representative democracy principals must choose between

preconceived bundles of policy goals proposed by candidates. It is highly unlikely that all

principals can wholly identify with specific bundles proposed by candidates, i.e. acknowledge

them as completely representing their own interests. Rather, they will have to compromise by

electing candidates whose bundles cover most of their political concerns and thus forsake the

assertion of some of their interest. Furthermore, it is likely that some principals will have to

compromise more than others: Some may find most of their interests represented by a specific

bundle; others may only be able to identify with a singular aspect of a bundle or, indeed, by no

aspect of any bundle whatsoever. Simply put, the capacities of principals to indirectly assert

their interests by electing representatives varies with respect to how well their interests match

with the bundles of policy proposals offered by candidates; and this variance makes for an

unequal distribution of participatory power.

The flexible delegation component of liquid democracy, on the other hand, allows all

members regardless of the combination of their interests to select specific representatives

for each of their political concerns. Since all members are entitled to authorize representatives

to vote on their behalf with regard to a limited policy area, they can customize a set of

representatives each of which represents a specific interest held by them. The more fine-

grained account of political representation offered by liquid democracy makes redundant the

need to compromise in the indirect assertion of interests. Consequently, it remedies the

problem of unequal participatory power that is generated in representative democracy by

14
differences between how well the different members interests match with available policy

bundles.27

We can, again, summarize the argument as a logical inference: if (1) the goodness of a

political decision-making procedure is dependent on its participatory equality; and if (2)

participatory equality is dependent on the equality of resources for directly asserting interests

(via voting on policy issues) and for indirectly asserting interests (via authorizing

representatives whose agendas match the interests of their principals); and if (3) liquid

democracy facilitates a greater equality of resources with respect to both forms of asserting

interests than representative democracy, then (4) liquid democracy is better than representative

democracy. Since we believe that premises (1), (2), and (3) are plausible, we hold that on the

equality-based account liquid democracy is superior to its counterpart.

5. Are Ordinary Members of the Political Community Overburdened by Liquid

Democracy?

We have argued that liquid democracy is preferable to representative democracy on both

major accounts of normative democratic theory. On the one hand, it allows members to

delegate their votes to issue-competent experts (epistemic superiority). On the other hand, it

allows them to vote directly on policy issues and to customize a set of representatives each of

which represents a specific subjective interest held by them (egalitarian superiority). These

advantages depend on whether members can competently perform a number of tasks: they

must select the right expert for the right job; they must decide when their interests are better

served by voting directly on a policy issue and when by delegation; and they must decide

27
This shows that allowing for liquid democracy within political parties is already an important step towards
greater equality since it allows individuals to influence the party's electoral program, thus helping to design a
policy-bundle that comes closer to their individual interests.

15
when to make use of the instant recall option.

However, one might call into question whether ordinary members are capable of

performing these tasks. The informational demands placed on the individual in a liquid

democracy appear to be much greater than those placed on a voter in a representative

democracy, where no policy-area-specific knowledge is required and where representatives

are elected for a fixed term. It seems that if a layperson with respect to a domain of expertise E

must decide whether to give credence to a putative expert E 1 or a rivaling putative expert E2,

she cannot acquire good epistemic reasons for favoring one over the other. 28 For, to be in that

position she would require expert knowledge about E herself, i.e. knowledge that is

unavailable to laypersons. This skeptical worry is particularly pressing with respect to the

political sphere where citizens must navigate a vast spectrum of distinct policy areas that are

characterized by steep expert-layperson gaps.

In the following, we argue that it is nonetheless plausible to expect members of a

political community to perform their liquid democratic tasks, both individually, and, perhaps

more importantly, as a collective. First, social epistemology provides good reasons for holding

on to the general claim that laypersons can make reasoned choices between conflicting

putative experts. Second, arguments from collective intelligence show (1) that collective

epistemic systems (such as the World Wide Web) can support reasoned expert selection

through practices of ranking and rating, and (2) that if sufficiently diverse, the collective of

people in a liquid democracy can produce better decisions even if reasoned expert selection

fails. Third, empirical studies of voter competence in direct democratic decision-making show

that even the poorly informed make reasoned choices.

Theorists of social epistemology argue that there are several sources of evidence that

28
See e.g. John Hardwig, Epistemic Dependence, Journal of Philosophy 88 (1985), pp. 693-708, Tyler Burge,
Content Preservation, Philosophical Review 102 (1993), pp. 457-488, Scott Brewer, Scientific Expert
Testimony and Intellectual Due Process, The Yale Law Journal 107 (1998), pp. 1535-1681.

16
laypersons can bring to bear on the evaluation of putative experts without having to become

experts themselves.29 The first source of evidence concerns the performance of putative

experts in discussions. When following a discussion between E 1 and E2 a layperson may not

be able to assess the content of the arguments advanced by the speakers. She may, however,

assess the form of presentation of those arguments. On the one hand, she can take into account

the dialectical performance of the speakers, i.e. the quantity, quickness, and smoothness of

rebuttals to ostensive counter-evidence presented by rival speakers.30 If, e.g., E1 presents an

ostensive defeater whenever E2 offers evidence for her conclusion, while E2 rarely manages to

offer a rebuttal to E1s evidence, the layperson has good reasons for placing greater trust in E1.

On the other hand, she can take into account the speakers receptivity to new information, i.e.

the disposition to consider seriously new ostensive evidence and to reconsider ostensive

evidence in light of new findings.31 If, e.g., E1 provides more charitable responses to E2s

opinions, affords fairer opportunity for E2 to express her views, and expresses greater interest

in E2s ostensive evidence, there are good reasons for attributing to E 1 a greater open-

mindedness and, thus, a greater epistemic credibility. 32

The second source of evidence pertains to biases and distorting interests that might lie

behind the claims of a putative expert. If a layperson possesses good evidence for such a bias

in E1 (stemming e.g. from economic interests, extreme loyalties towards a specific group,

sexist or racist views) and no evidence for such a bias in E 2, she has better reasons for giving

credence to E2. This idea is, as Alvin Goldman points out, supported by common sense and

29
See Alvin Goldman, Experts: Which Ones Should You Trust?, Philosophy and Phenomenological Research
63 (2001), pp. 85-111, David Matheson, Conflicting Experts and Dialectical Performance: Adjudication
Heuristics for the Layperson, Argumentation 19 (2005), pp. 145-158, David Coady, When Experts
Disagree, Episteme 3 (2006), pp. 68-79, and Thomas Christiano, Rational Deliberation Among Experts and
Citizens, Deliberative Systems, ed. J. Parkinson and J. Mansbridge (Cambridge: Cambridge University
Press, 2012), pp. 27-51.
30
Goldman, Experts: Which Ones Should You Trust, p. 95.
31
See David Matheson, Conflicting Experts and Dialectical Performance, p. 151-153.
32
On the merit of open-mindedness for the pursuit of knowledge see also James A. Montmarquet, Epistemic
Virtue, Mind 96 (1987), pp. 482-497.

17
experience: If two people give contradictory reports, and exactly one of them has a good

reason to lie, the relative credibility of the former is seriously compromised. 33

The third source of evidence consists in track records of putative experts cognitive

successes: The better a persons track record, the greater the likelihood of her having correct

answers to current questions.34 At first glance, this suggestion is puzzling, for it seems that in

order to use past track records to assess the credibility of a putative expert with respect to a

domain of expertise E, the layperson must possess expert knowledge about E herself.

However, this is not necessarily true: Many statements pertaining to E are epistemically

inaccessible for laypersons at a certain point in time, but become accessible later on. Consider

the following claim: Introducing a green zone in the city center will reduce particulate

matter pollution (PM). Laypersons at time t 0 may not be in a position to determine the

credibility of this claim. However, at time t 1, one year after a green zone has been put into

practice, they may ascertain its credibility by checking whether the average monthly PM rate

has dropped.35

The fourth source of evidence stems from the cognitive capacities of other members of

the political community. First, laypersons may deliberate about the arguments presented by

putative experts, reciprocally fill gaps in their knowledge about the relevant area, and

collectively uncover fraudulent claims.36 Second, laypersons can appeal to meta-expert

agreement, i.e. the fact that the vast majority of putative experts regarding E side with E 1,

rather than her competitor E2.37 Third, they can make use of what Thomas Christiano calls

overlapping understanding and explains as follows: [S]uppose P knows about intellectual

33
Goldman, Experts: Which Ones Should You Trust, p. 104.
34
See ibid. p. 106.
35
We owe this example to an anonymous referee.
36
See John Dryzek and Christian List, Social Choice Theory and Deliberative Democracy: A Reconciliation,
British Journal of Political Science 33 (2003), pp. 1-28; p. 10.
37
Ben Almassi, Climate Change, Epistemic Trust, and Expert Trustworthiness, Ethics and the Environment 17
(2012), pp. 29-49; at p. 37 and Goldman, Experts: Which Ones Should You Trust?, p. 97.

18
disciplines a, b, and c, and Q knows about disciplines b, c, and d. Their knowledge overlaps at

b and c. This overlap allows Q to understand some of a because P can translate the ideas of a

into b and c. Now suppose that R has expertise c, d, and e and S has d, e, and f. [] S and P

do not overlap at all []. Despite this, Ss knowledge may constrain P [] through the

intermediaries Q and R.38 Thus, even if a member of the political community is unable to

directly assess the truth-value of claims that lie outside her domain of expertise, she may do so

indirectly by relying on intermediaries whom she trusts and who translate the relevant content

and make it intelligible to her.

The arguments from social epistemology show that liquid democracy does not need to

rely solely on each and every individual's solitary capacity to select competent experts.

Instead, it can make use of the fact that individuals are connected in a whole system of

delegative relationships where the evaluation of experts can turn into a collective enterprise.

Goldman's sources of evidence thereby mirror the features of a reputation system based on

ranking and rating which is seen as an important condition for a functioning collective

epistemic system as discussed in the literature on collective intelligence.39 As Gloria Origgi

argues, given an overwhelming informational environment with a multitude of competing

sources, such as the World Wide Web (and, we could add, such as a liquid democracy),

successful epistemic practice becomes possible if we gain access to information about how

others value and rate things. Not all crowds are wise, but we can argue with Origgi that they

become wiser where online reputation systems collect, distribute, and aggregate information

about participants past behavior.40 The claim that we can find the right experts for the right

job thus gains further plausibility, since the epistemic practice of processing information

through online rankings and ratings is already a part of many peoples daily routine, for
38
Christiano, Rational Deliberation Among Experts and Citizens, p. 38.
39
See Gloria Origgi, Designing Wisdom through the Web, Collective Wisdom. Principles and Mechanisms,
ed. H. Landemore and J. Elster (New York: Cambridge University Press, 2012), pp. 38-55.
40
Origgi, Designing Wisdom through the Web, p. 49.

19
example when selecting from a range of book-sellers on Amazon. 41

Another argument from collective intelligence goes one step further: Even if reasoned

expert selection fails, liquid democracy could still plausibly be expected to be epistemically

superior to representative democracy, since it does not only mobilize the expertise of the few,

but also brings the wisdom of the many to bear on a decision. The theoretical foundation for

this claim can be found in Scott Page's work on cognitive diversity. 42 The more diverse

individuals ways of representing situations and problems, of categorizing or partitioning

perspectives, of generating solutions to problems and of inferring cause and effect, the better

the community at developing predictive models and solving problems. He provides

experimental evidence and mathematical proof for a Diversity Trumps Ability Theorem and

a The Crowd beats the Average Law. The former states that under certain conditions, a

randomly selected collection of problem solvers outperforms a collection of the best

individual problem solvers.43 The latter holds that given any collection of diverse predictive

models, the collective prediction is more accurate than the average individual prediction. 44

Relating his own findings to democratic decision-making, Page explicitly acknowledges that

individuals may indeed have poor incentives to invest in improving their own predictive

models in order to competently decide on policies. He then argues that this is not necessary for

good decisions, since the lack of individual ability can be compensated by cognitive diversity:

An effective democracy [] may depend as much on its citizens having diverse predictive

41
Interestingly, recent survey-based evidence suggests that people using the Internet as a source of information
are more likely to develop issue-specific knowledge than those who rely on TV, newspapers, or the radio; see
S. Mo Jang and Yong Jin Park, The Internet, Selective Learning, and the Rise of Issue Specialists, First
Monday 17 (2012), http://dx.doi.org/10.5210/fm.v17i5.3888.
42
Scott Page, The Difference. How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies
(Princeton/Oxford: Princeton University Press, 2007).
43
Scott Page, The Difference, p. 162. The four conditions are that the problem is difficult (which is typical for
most policy decisions), that all problem solvers have at least some ability to solve the problem (in technical
terms, they need to be able to recognize a local optimum), the group is diverse to the extent that some
problem solver can find improvement (i.e. local optima are different but they intersect in only one global
optimum), and the set of problem solvers must be sufficiently large.
44
Page, The Difference, p. 209.

20
models as on their having accurate predictive models. 45 Thus, liquid democracy is likely to

produce better decisions even if expert selection fails. With its greater openness to

participation, liquid democracy makes it more likely that a large, cognitively diverse group of

people provide input to each decision.

Finally, since the skeptical worries concern the capacities of real members of a

political community, empirical research on voter competence in direct democratic decision-

making can be instructive. Direct democratic institutions are often confronted with skepticism,

since a range of widely cited studies show that the average voter is poorly informed and lacks

coherent and stable political preferences and beliefs. 46 However, these findings have been

countered with several more optimistic responses. One response has been to bite the bullet and

accept the verdict of the poorly informed voter, only to then show that the uninformed

nonetheless successfully use heuristics to make competent choices, i.e. they are capable of

predicting the outcomes of their choices and of evaluating whether they are in line with their

preferences. Using heuristics allows individuals to keep the challenges of information

processing in a complex environment within manageable boundaries. The punch line in this

literature is: Yes, information is costly, yes, people are poorly informed, but observational and

experimental evidence tell us that this does not matter for the quality of their decisions. 47 A

prominent heuristic is to rely on cues from the more informed. Experiments show that in the

absence of shared interests (e.g. when individuals are from different social groups), it becomes
45
Page, The Difference, p. 347.
46
See most prominently: Philip E. Converse, The Nature of Belief Systems in Mass Publics, Critical Review:
A Journal of Politics and Society 18 ([1964] 2006), pp. 1-74, and John R. Zaller, The Nature and Origins of
Mass Opinion (New York: Cambridge University Press, 1992).
47
For experimental and observational evidence, see e.g. Arthur Lupia, Shortcuts versus Encyclopedias:
Information and Voting Behavior in California Insurance Reform Elections, American Political Science
Review 88 (1994), pp. 63-76, Arthur Lupia and Mathew D. McCubbins, The Democratic Dilemma: Can
Citizens Learn What They Need to Know? (New York: Cambridge University Press, 1998), and Paul M.
Sniderman, Richard A. Brody and Philip E. Tetlock, Reasoning and Choice: Explorations in Political
Psychology (Cambridge: Cambridge University Press, 1991). Experiments by Lau and Redlawsk provide
nuance to these findings, cautioning that the politically sophisticated make better use of heuristics than the
less sophisticated, see Richard R. Lau and David P. Redlawsk, Advantages and Disadvantages of Cognitive
Heuristics in Political Decision Making, American Journal of Political Science 45 (2001), pp. 951-971.

21
crucial to introduce institutions such as penalties for lying and threats of verification to make

it unlikely that cue-givers will use their position to deceive and manipulate cue-takers. 48 This

reiterates the importance of a functioning reputation system discussed above, though from a

different angle.

A second response has been to dispute whether the lack of information people display

when answering factual questions about politics in a survey has any connection to the

information they need to perform their tasks in direct democratic polls, 49 and to call for paying

closer attention to the importance of how we measure ignorance. For example, Jason

Barabas et al. provide a typology of knowledge and show that the answers to substantive

research questions (e.g. whether education affects political knowledge) differ depending on

whether we measure knowledge as knowledge about the rules of the game (affected by

education) or as policy-specific knowledge (unaffected by education). 50 Finally, whereas

findings about ignorance are mostly based on US voters, research that compares cantons in

Switzerland has shown that increased possibilities for participation in fact lead to citizens

being better informed.51 In addition to theoretical arguments from social epistemology and

collective intelligence, empirical research on direct democratic decision-making thus provides

further ground for optimism with regard to the practicality of liquid democracy. It shows that

reasoned choice is not dependent on the level of information as long as more motivated

individuals act as cue-givers and are prevented from lying through a functioning reputation

system. Furthermore, in the long run increased possibilities for participation may actually set

incentives for more members to become better informed.

48
Cheryl Boudreau, Making Citizens Smart: When Do Institutions Improve Unsophisticated Citizens
Decisions?, Political Behaviour 31 (2009), pp. 287-306.
49
Boudreau, Making Citizens Smart, p. 291.
50
Jason Barabas, Jennifer Jerit, William Pollock and Carlisle Rainey, The Question(s) of Political Knowledge,
American Political Science Review, 108 (2014), pp. 840-855.
51
Matthias Benz and Alois Stutzer, Are Voters Better Informed When They Have a Larger Say in Politics?
Evidence for the European Union and Switzerland, Public Choice, 119 (2004), pp. 31-59.

22
6. Problems and Modifications of the Basic Model of Liquid Democracy

However, even if members are well-informed, or make at least good use of heuristics, liquid

democracy still encounters the challenges of (1) avoiding to recreate representative

democracy's problem of two classes of citizens, though in a different way (the problem of

unequal voting power) and (2) achieving consistency across the whole set of decisions being

made across different policy areas (the problem of policy-inconsistency). Neither problem

arises to the same extent in representative democracy. In order to maintain our argument that

liquid democracy is superior to representative democracy, modifications to the basic model are

therefore in order.

6.1. The Problem of Unequal Voting Power

The problem of unequal voting power is a direct consequence of allowing members of a

political community to choose freely whether to cast their vote on an issue themselves or

whether to delegate it to a representative. Let us reconsider the example given in section two.

A decided to delegate her vote on whether to authorize planting genetically enhanced corn to

B, a renowned expert on this matter. It is plausible that not only A, but several other members

of the community also delegate their votes to B. After all delegations have been made and the

time for casting the vote has come, B can cast 450 votes. Consider now that unlike A,

individual C has chosen to vote himself. B and C now cast their vote in the final decision on

genetically enhanced corn, the slight difference being that B casts 450 votes, whereas C casts

one vote.

23
 The example shows that liquid democracy runs the risk of reproducing a problem of

representative democracy, though in a different way. Rather than dividing the political

community into a class of ordinary citizens without and a class of representative elites with

decision-making power for the duration of a whole legislative period, liquid democracy

divides the decision-making body in a given policy area into ordinary members with one vote

and members with 2+N votes.52 This problem does not occur to the same extent in

representative democracy where each citizen has one vote when it comes to selecting

representatives and where each representative has one vote when deciding in parliament. This

inequality of voting power may provide counter-incentives for ordinary members to

participate directly in collective decision-making. Since their political resources to influence

decisions are marginal compared to those of super-representatives with thousands of votes,

ordinary members have little reason to directly vote on policy issues. Thus, a conflict comes to

light between the direct democratic component and the flexible delegation component of the

basic model. However, if one holds on to the idea that all members should be entitled to

directly vote on policy issues, it seems necessary to qualify the flexible delegation component.

6.2. The Problem of Policy-Inconsistency

The beauty of liquid democracy is that it gives individuals the free choice on whether or not to

delegate their vote, and it does so for policy areas individually. However, on the downside,

this flexible and issue- or area-specific delegation implies that the composition of the

decision-making body in each policy area varies. Whereas non-delegates with just one vote

may have incentives to form alliances within a policy-area to rival the voting power of popular

52
See also Ford, Delegative Democracy, p. 3.

24
delegates with many votes, there are no comparable incentives for forging alliances across

policy-areas. This poses severe challenges to the quality of the overall policy output of a given

democracy since all policy decisions are inherently connected through a common and public

budget in virtue of which measures in one policy field impact on the resources available in

another. Furthermore, even if sufficient resources for all measures were available, there might

be substantive contradictions between decisions taken in different areas. In the absence of a

coordinating body of decision-makers with oversight, those contradictions are doomed to

remain unnoticed until the later stage of policy implementation.

By contrast, representative party democracy is strong at bundling together solutions to

societal problems from different policy areas and also trading them against each other, if

necessary. First, at the beginning of the electoral phase, parties negotiate internally over the

programmatic policy bundle they want to offer to voters. Each electoral program is thus the

result of negotiating goals and measures to reach these goals across different policy areas.

Second, during the inter-electoral phase (i.e. the phase between representative elections)

parties put forward programs for governing that have to get the support of the parliament

where the same people are deciding on all policy areas, thereby becoming necessarily aware

of the interconnectedness and potential trade-offs between individual decisions. In liquid

democracy, there is no distinction between electoral and inter-electoral phases, since new

issues can continuously be brought onto the agenda. Parties as coordinating mechanisms are

superfluous since members of the community either participate directly or delegate their votes

on the basis of individual relationships with their delegate. Finally, the principle of instant

recall ensures that the group of people dealing with issues in a given area may change

composition between decisions. To be sure, any method of aggregating preferences where

more than two alternatives are involved is potentially bedeviled by a problem of cycling

25
collective preferences.53 In that regard, liquid democracy is no better or worse than any other

collective decision-making procedure. However, whereas in a representative democracy, at

least the preferences to be aggregated are consistent across policy areas since the same

parliamentarians vote on all issues (even if the social preference may cycle), in liquid

democracy not even the preference input is consistent across policy area because different

people decide. As a result, liquid democracy can be expected to display inferior performance

in its public-policy output.

6.3. Deliberating Trustees and an Executive with Oversight Towards a Solution

We suggest that these problems can be ameliorated by combining the basic model of liquid

democracy with three components: (1) a trustee, rather than a mere delegate model of

representation, (2) elements of deliberative democracy, and (3) an executive with oversight.

We have seen that participatory incentives for ordinary members with just one vote

will be minor if there are super-representatives with hundreds of votes in the same policy

forum. However, what if, once delegation has happened, there was the additional option of

deliberating on possible answers to the societal problem at stake before casting votes? 54 In that

case, even an ordinary member with a single vote could bring forward the decisive argument

that may convince a representative with 450 votes. Rather than merely aggregating votes once

delegation has happened, there should be a phase were decision-makers engage in giving
53
The mathematical foundation for the problem of cyclical social preferences is Kenneth Arrow's impossibility
theorem, famously used to argue that democratic rule is irrational by William Riker (1982. Liberalism Against
Populism: A Confrontation Between the Theory of Democracy and the Theory of Social Choice. Prospect
Heights: Waveland Press). The empirical relevance of the theorem, however, has been disputed (see Gerry
Mackie. Democracy Defended. Cambridge: Cambridge University Press, 2003).
54
In keeping with the standard definition, we understand political deliberation as a rational and egalitarian type
of communicative contestation about policies where all participants are required to justify their interests to
others, take into account their respective views, and revise their positions in light of better arguments (see
Dryzek, Deliberative Democracy and Beyond, pp. 47-50.

26
reasons for why they intend to vote in favor or against a proposal. Including a deliberative

phase between delegation and voting thus gives members with just one vote the chance to

make a difference.

However, including a deliberative phase implies that voting intentions have to be

revisable in principle. Representatives therefore need to have some degree of discretion and

independence. Including the possibility for deliberation thus has important implications for

how we conceive the representational role of the delegates. In her seminal analysis of the

concept of representation,55 Hanna Pitkin argues that political representation lies on a

continuum between the extreme poles of mechanic execution of the principals instructions

(delegate) and complete independence (trustee). It is defined by Pitkin as follows: The

representative must act in such a way that, although he is independent, and his constituents are

capable of action and judgment, no conflict arises between them. He must act in their interest,

and this means that he must not normally come into conflict with their wishes. 56 In light of the

problem of unequal voting power, liquid democracy needs representatives that lean enough

towards the independence side of the continuum and have discretion to change their voting

intention on the basis of justified reasons presented to them within a deliberative stage. They

may, however, do so only if they consider such a change to be in the best interest of the

individuals who delegated their votes to them in the first place. Delegates thus need not

merely cast the vote their principals would and could have cast themselves, but have

discretion to take into account new arguments presented during deliberation and update their

voting intentions accordingly. However, at the same time, the principal's wishes must be

potentially there and potentially relevant, 57 and liquid democracy ensures this through the

55
Hanna F. Pitkin, The Concept of Representation (Los Angeles: University of California Press, 1967).
56
Ibid, p. 166.
57
Ibid, p. 155.

27
principle of instant recall. If the principals become aware that there is a conflict between their

interests and the actions taken by the representative, they can withdraw their delegated vote.

However, including a deliberative phase does not only ameliorate the problem of

unequal voting power. Following David Miller, John S. Dryzek and Christian List, it also

helps to reduce the structure of political conflict to one, rather than several issue dimensions,

thereby preventing preference cycles and ensuring the transitivity of collective preferences. 58

In addition, it opens up a possibility of linking individual decisions to others within the same

policy area, thus ameliorating the problem of policy-inconsistency. This shows why it might

be more advantageous to structure relationships of delegation by policy area, rather than by

individual topic. Only the former offers the possibility to connect individual decisions within

the same policy area. For example, when discussing a reform that shortens the years of high

school education, it seems advisable to consider the wider implications for the quality of

higher education, since universities now have to deal with a greater number of potentially less

well-prepared entrants. The decision-makers should therefore simultaneously address

measures to support universities in dealing with this challenge. At the same time, only

delegation by policy area lowers information costs compared to full-scale direct democracy. If

I still need to engage with every single decision being taken in order to decide on a case-by-

case basis whether and to whom to delegate my vote, this will not be much easier than directly

voting on all issues myself. However, once I have established that E is an expert on education

policy and have delegated my vote to her for all issues on the agenda that are tagged as

belonging to this policy area, I can retreat from following this policy-area more closely (apart

from occasionally checking whether her voting behavior corresponds to my interests).

58
David Miller, Deliberative Democracy and Social Choice, Political Studies 40 (1992), pp. 54-67, John S.
Dryzek, Deliberative Democracy and Beyond. Liberals, Critics, Contestations (New York: Oxford University
Press, 2000), pp. 38-42, and John S. Dryzek and Christian List, Social Choice Theory and Deliberative
Democracy, pp. 7-27.

28
 However, as argued above, coherent public policy output of the democratic political

system requires not only coordination of decisions taken within one policy-area but also

across policy-areas. In order to address this challenge, we suggest to strictly limit the role of

liquid democracy to legislative decision-making. Liquid legislatures are then to be

accompanied by a democratically elected executive serving a fixed term that does not blindly

implement, but also reviews laws in terms of financial viability and overall consistency. When

financial trade-offs and/or inconsistencies between laws from different areas are detected, the

executive should return the affected legislative proposals to the area-specific forums. It should

finally have the competence to initiate and moderate negotiation processes between different

areas for reaching package deals. Importantly, however, the executive should not be entitled to

veto laws from a substantive perspective (i.e. on the basis of disagreeing for some reason or

another with the content of the law), but only from the perspective of financial resources and

inconsistencies between individual laws from different policy areas. In liquid democracy,

connections between policies are thus the end product of a bottom-up process (where peoples

preferences are aggregated at the level of individual polices) rather than being proposed in a

top-down manner by political elites, as is the case in representative democracies.

7. Conclusion

This article has justified liquid democracy on the basis of arguments from normative

democratic theory. We have argued that collective decision-making according to the basic

model of liquid democracy is more epistemically reliable and more egalitarian than

representative democracy. Drawing on arguments from social epistemology, and research on

collective intelligence and voter competence in direct democratic decisions, we have shown

29
that there are good reasons to be optimistic that ordinary members of a political community

can indeed perform their liquid democratic tasks.

We have further addressed two problems of the basic model of liquid democracy: the

problem of unequal voting power and the problem of policy-inconsistency. We have therefore

suggested three modifications: (1) combining the basic model with a trustee, rather than a

mere delegate model of representation; (2) requiring decision-makers to adhere to deliberative

norms; and (3) balancing liquid decision-making in legislatures with an executive that reviews

the fiscal feasibility of policies and moderates package deals between proposals from different

policy areas.

Naturally, this discussion of problems and potential solutions does not exhaustively

address the many issues that should be taken into consideration when putting liquid

democracy into practice. However, fleshing out all details involved in putting liquid

democracy into practice was beyond the scope of this article. We have aimed to provide a

normative justification of liquid democracy to inspire further discussion on this previously

neglected type of collective decision-making. We hope that future theoretical work will now

address questions such as: Should the executive consist of a president with an appointed

cabinet? How exactly would a budgeting system work under liquid democracy? How should

the agenda-setting process be structured? How should the reputation and ranking system be

designed?

Finally, our article has justified liquid democracy at the systemic level, comparing it to

representative democracy as the best systemic alternative. Currently, liquid democratic

decision-making is emerging on smaller scales as a decision-making procedure applied within

political parties, or institutions of local government. 59 Future empirical research should

30
therefore address the interactions between liquid decision-making and an overall

representative democracy. We hope that these works will benefit from the clarification and

justification of the basic model of liquid democracy we have sought to provide in this article.

Electronic Participation: 5th I.F.I.P. W.G. 8.5 International Conference Proceedings, ed. M. Wimmer, E.
Tambouris, and A. Macintosh (Heidelberg/New York/Dordrecht/London: Springer 2013), pp. 147-158.

31