Extrinsic Democratic Proceduralism: A Modest Defence

Disagreement among philosophers over the proper justification for political institutions is far from a new phenomenon. Thus, it should not come as a surprise that there is substantial room for dissent on this matter within democratic theory. As is well known, instrumentalism and proceduralism represent the two primary viewpoints that democrats can adopt to vindicate democratic legitimacy. While the former notoriously derives the value of democracy from its outcomes, the latter claims that a democratic decision-making process is inherently valuable. This article has two aims. First, it introduces three variables with which we can thoroughly categorise the aforementioned approaches. Second, it argues that the more promising version of proceduralism is extrinsic, rather than intrinsic, and that extrinsically procedural accounts can appeal to other values in the justification of democracy without translating into instrumentalism. This article is organised as follows. I present what I consider to be the ‘implicit view’ in the justification of democracy. Then, I analyse each of the three variables in a different section. Finally, I raise an objection against procedural views grounded in relational equality, which cannot account for the idea that democracy is a necessary condition for political legitimacy.


Introduction
There is a long history of controversy concerning the best way to vindicate democracy as a desirable and legitimate and/or authoritative political regime, 1 as well as concerning the features that characterise a regime as democratic. While not all political theorists have been explicit about it, it is reasonable to divide this justificatory enterprise into three fundamental components: definitional, evaluative and normative one. Accounts of democratic legitimacy generally include all of these, though they are not always addressed according to a specific order of importance. This justificatory enterprise can be referred to as the implicit view.
The first, definitional component is tasked with the conceptual analysis of democracy; it aims to provide a proper definition of it as a set of institutions and practices or as a distinctive decision-making procedure. The main challenge is to determine how general vs. specific and how normatively loaded vs. purely descriptive a definition of democracy ought to be in order to be both descriptively adequate and critical vis-à-vis existing institutions. As Robert Dahl points out, self-described democratic countries have adopted a wide variety of institutions-parliamentary vs. presidential systems and proportional vs. first-past-the-post electoral laws-and it is important to understand what specifically is democratic about these institutions (1998).
Hence, a useful definition should account for essential features of modern democratic regimes, such as universal suffrage and equal rights, while also maintaining some degree of generality. 2 Scholars interested in providing an account of democratic legitimacy are generally concerned with democracy as an egalitarian procedure for collective decisionmaking (Arneson 2003, pp. 130-131;Wall 2006, p. 91, Martí 2006Christiano 2008, pp. 2-9;Peter 2009, p. 2;Viehoff 2014, p. 342;Kolodny 2014a, p. 197;Saffon and Urbinati 2013, p. 445;Rostbøll 2015a, p. 268). 3 The reason for this focus 1 Political legitimacy and authority have a complex relationship. In democratic theory, most scholars take the two to be related: John Rawls refers to the 'legitimacy of the general structure of authority ' (1993, p. 136); Tom Christiano, following Joseph Raz, explicitly states that 'the idea of legitimate authority as a right to rule to which citizens owe obedience gives each citizen a moral duty to obey, which it owes to the authority' (2008, p. 242); Fabienne Peter uses legitimacy to qualify the notion of political authority, which under democratic institutions belongs to the people-'democratic legitimacy thus qualifies the right of the democratic constituency to impose laws and regulations on itself ' (2009, p. 56); according to Philip Pettit, authority and legitimacy go together (2012, p. 149). However, some are more resistant to tying up the two. Allen Buchanan famously distinguishes political legitimacy from authority, claiming that the latter concept is dispensable (2002, p. 703) and he takes democratic decisionmaking to be a condition for both legitimacy and obligation (2002, p. 714). Daniel Viehoff refers only to 'genuine authority' and avoids talk of legitimacy altogether (2014, p. 340). David Estlund (2008) and Niko Kolodny (2014a, b) distinguish between legitimacy, which identifies moral permissibility of coercion, and authority, which constitutes the moral power to issue authoritative commands. For the purposes of this article, however, no distinction is required.
is that, if the final aim of the justificatory enterprise is to vindicate democracy as a legitimate authority, the process through which binding laws are issued takes priority over other aspects of a democratic system, such as its political or social culture. Importantly, this does not mean that these latter aspects are inconsequential. For instance, Elizabeth Anderson distinguishes between three levels of democracy 'as a membership organization, a mode of government, and a culture' but also stresses how the three elements are inextricably linked to the extent that 'democracy as a mode of government cannot be fully achieved apart from a democratic culture' (2009, p. 214). 4 While such a procedural outlook may be insufficient to fully grasp the essence of democracy, it is nonetheless the case that when one argues in favour of the legitimacy of democratic decisions, the process through which these decisions are brought about is of primary concern. Therefore, the definitional component is by no means purely stipulative but rather strictly connected to the evaluative and normative ones. 5 The second, evaluative component investigates the value of democracy; it asks what is distinctive about democracy as a decision-making process thus defined and why it is desirable. There are two well-known approaches to this task: instrumentalism, which justifies democratic institutions by virtue of their outcomes or by-products, and proceduralism, which views democracy as inherently legitimate based on distinctive features of the process (Christiano 2004, p. 266).
Finally, the third component serves the normative purpose of explaining why democracy is a legitimate authority. As the argument typically goes, once we have defined democracy and justified why it is valuable, democratic decisions are legitimate (i.e. collectively binding for citizens or permissibly enforced on them). Marking democratic outcomes as legitimate means that they are binding by virtue of the fact that they are produced in line with a procedure that is viewed as democratic. Naturally, conditions and limitations may apply depending on how democracy is justified. 6 While these three components are connected, the distinction among them is crucial to assess various models of justification, particularly when it comes to the divide between proceduralism and instrumentalism. the actual collective authorization of laws and policies by the people subject to them ' (2008, p. 38). Since nothing in this article touches upon whether the principle of 'one person one vote' is a necessary component of democracy, I will leave this issue aside.

Footnote 3 (continued)
As is generally asserted in the literature, proceduralism and instrumentalism hold opposite claims. While the former takes democracy to be intrinsically valuable or just (Griffin 2003, p. 118;Christiano 2008, p. 102) and necessary for political legitimacy (Christiano 2008, p. 100;Viehoff 2014, pp. 368-369), the latter derives democracy's value from its valuable outcomes, and considers democracy legitimate only if it brings about these outcomes (Arneson 2010, p. 33;Wall 2007, p. 437). Both of these approaches are after something important. On one hand, a strong account of democratic legitimacy should be able to vindicate the value of democracy in a way that is not merely contingent. Indeed, if democracy only has a contingent value, we should abandon it as soon as another decision-making procedure is shown to be better equipped at promoting the values we seek. Naturally, this is quite problematic if we acknowledge the existence of reasonable pluralism, which sows disagreement over the values relevant for political legitimacy (Waldron 1999;Weale 1999;Peter 2009). On the other hand, a convincing account of democratic legitimacy should be able to speak to those who are not already convinced democrats. While people clearly disagree also over which values justify democracy, the lack of reference to any other value beyond democracy itself will weaken the justification for democracy in the eyes of those asking why they should support it.
Proceduralism has been famously criticised for being tautological, as it seems to defend democracy by appealing to principles that merely express what democracy is and how it functions (Estlund 2008, p. 93;Rostbøll 2015a, p. 273); instrumentalism is criticised for its vindication of democracy in a purely contingent manner, as it reduces democracy to a mere means to an end. This article has two aims: (1) to scrutinise the existing distinction between instrumentalism and proceduralism and replace it with a tripartite model including both intrinsic and extrinsic proceduralism; and (2) to defend extrinsic proceduralism as a promising account of democratic legitimacy, with one caveat. Contrary to what some scholars think (e.g. Estlund 2008), proceduralism can refer to further values without moving towards instrumentalism; however, not all extrinsic proceduralist accounts are able to vindicate the necessity of democracy for political legitimacy, as the last part of this article will argue.
The paper is organised as follows. In the next section, I introduce the three variables and analyse each of them individually. While addressing the last variable, I argue that at least some versions of extrinsic proceduralism (notably those grounded in relational equality) seem bound to abandon a core proceduralist idea, namely that democracy is necessary for political legitimacy. I conclude by outlining one important takeaway for any conception of democratic legitimacy.

Three Variables of Distinction
The opposition between instrumental and procedural accounts seems intuitively appealing. On the one hand, instrumentalism acknowledges that democracy, like any other form of government, is valuable only because of the goals it achieves. These may range from the protection of negative freedoms to the enhancement of social equality (Hamilton et al. 1787(Hamilton et al. -1788Arneson 1993Arneson , 2003Arneson , 2004Van Parijs 1996;Goodin and Spiekermann 2018). Strategies used to justify democracy in this sense vary but it is noteworthy that they all involve the following interpretation of the implicit view: first, principles and values generally viewed as independent of and prior to politics are considered relevant; second, forms of government identified as democratic are justified as the best available means for the protection, implementation or maximisation of such principles and values; third, on the grounds that democracy is justified according to certain values, its outcomes are taken as legitimate.
On the other hand, so-called procedural justifications of democracy reject this view. They purport to highlight the intrinsic value of democracy compared to other forms of government. Accordingly, democratic decision-making embodies fundamental moral principles ranging from equal respect for persons (Rawls 1993;Cohen 1997a, b;Griffin 2003) to autonomy, non-domination (Forst 2011;Pettit 2012;Bellamy 2007;Rostbøll 2015a) and moral, epistemic and relational equality (Christiano 2008;Peter 2009;Anderson 2009;Viehoff 2014;Kolodny 2014a, b). Most of these accounts reject the idea that there are moral values that pre-date the democratic process (Waldron 1999;Bellamy 2007;Peter 2009) or the political and legal order (Pettit 2012;Rostbøll 2015a, b); they instead hold that democracy's alleged capacity to bring about positive results is irrelevant in determining its value and legitimacy. Essentially, instrumentalism views democracy as a reliable, and often the most effective, form of governance, while proceduralism views democracy as inherently valuable (Christiano 2008, p. 102;Griffin 2003, p. 118).
However, viewing something as a means is not the opposite of viewing it as intrinsically valuable, as scholars in value theory observe. It is beyond the scope of this paper to summarise this debate, which is complex and ongoing, but insights from value theory will help us navigate the distinction between instrumentalism and proceduralism. Drawing from value theory, let us call 'final value' the value that an object has for its own sake, while 'non-final value' refers to the value an object has as a means or for the sake of something else. In the first case what Christine Korsgaard calls the purpose of value is the object itself, while in the second case the purpose lies in another object (1983, p. 170). 7 The fact that something has value as an end, rather than a means, does not necessarily mean that it has value in itself-i.e. in a non-derivative way. For instance, the pen that Abraham Lincoln used to sign the Emancipation Proclamation is thought to have value for its own sake on account of its history. However, if it has this value, it is not in virtue of the fact of being a pen, but in virtue of what it represents with respect to the Emancipation Proclamation. 8 Thus, the value of an object can be noninstrumental and yet derivative. Intrinsic value in this second sense refers to the kind of value an object has non-derivatively, while extrinsic value is derivative. 9 Intuitively, all means derive the entirety of their value from the causal contribution they make to something else. 10 Hence, all non-final values are also extrinsic. Objects that are valued for their own sake, however, can derive their value from other values, qualifying them as final extrinsic values, or be non-derivatively valuable, qualifying them as final intrinsic values. 11 The Emancipation pen is an example of the former type, while the environment can be considered an example of the latter type. 12 The first variable, which I name reference, is modelled on the distinction between final and non-final values in axiology and it indicates whether democracy is valued for its own sake or for the sake of something else. The second variable, which I name contribution, corresponds to the derivation sense in value theory and discriminates between intrinsic and extrinsic justifications of democracy (i.e. justifications that see the value of democracy as derived from other values or as non-derivative). The third variable, which concerns the modality of justification, is not drawn from value theory but serves to assess whether democratic procedures are valuable necessarily or contingently. Throughout the remainder of this paper, I refer to principles and values advanced to defend democracy collectively as 'the justifying value' (without inquiring into what this value or set of values is) and examine its relation to democratic procedures and outcomes. I address each of the three variables in the following sections. 9 Rønnow-Rasmussen (2015, p. 32) distinguishes between the supervenience sense and the derivation sense; while I refer to the latter here, the former maps the intrinsic vs. extrinsic distinction in terms of the internal vs. relational properties of an object. The ways in which these two senses of intrinsic vs. extrinsic relate to one another is not of concern here. Daniel Viehoff recently made a similar observation, claiming that instrumentalism is concerned 'with the kind of value that political power and its distribution have: instrumental vs. non-instrumental value', while proceduralism turns to 'the locus of value that justifies political power: it is the value of procedures rather than outcomes' (2017, p. 276). He then goes on to describe the non-derivative or contributory value that procedures may have, following Sandbu (2007). While Viehoff's distinction is insightful, it is not the central topic of his article; this article provides a more thorough account of the variables of distinction between instrumentalism and proceduralism. 10 For an objection, see Kagan (1998). 11 Christine Korsgaard makes a similar observation when she distinguishes between the purpose and source of value. Her aim is to defend a Kantian stance according to which a thing can be valued as an end without being, for this reason, intrinsically valuable. I do not employ Korsgaard's distinction here because her notion of source has been criticised as ambiguous (e.g. Rabinowicz and Rønnow-Rasmussen 2000), and I thought that the notion of derivation would be clearer in this context. 12 Kantians and Utilitarian may disagree on this point. See, for instance, Harold (2005, pp. 95-99).

Reference
All accounts of democratic legitimacy have one thing in common-they aim to provide reasons for viewing democratic decisions as legitimate in virtue of the way the decisions were made. If we follow Estlund's example and distinguish between the content of a decision and its pedigree, we see that the former qualifies a decision as correct or justified while the latter qualifies it as legitimate insofar as it is democratic (2008, p. 106ff). This first variable, reference, mirrors the traditional way of framing the opposition between proceduralism and instrumentalism, as it indicates whether democratic procedures are valued directly as ends or indirectly as means to other ends.
Instrumentalists evaluate democratic performance on the basis of the conformity of its outcomes or by-products to the justifying value (Arneson 2004, p. 42). 13 It is worth noting that instrumental justifications for democracy do not purport to be forms of 'correctness theory' (Estlund 2008, p. 99), as instrumentalism does not view only correct or just decisions as legitimate. If this were the case, instrumentalism could not afford a theory of democratic legitimacy, since democratic decisions would be binding only in virtue of their substantive correctness and not their democratic pedigree. 14 Rather, when instrumentalist approaches attempt to justify democracy as legitimate or authoritative, they aim to provide reasons to view all outcomes as prima facie legitimate. Instrumentalism does this by valuing democratic procedures indirectly, as they are thought to produce good results on average-that is why all of them can count as legitimate (unless gravely unjust). 15 In contrast, proceduralist approaches disregard outcomes and focus directly on the realisation of the justifying value in the making of decisions. Regardless of the quality of outcomes, the fact that the decisions were made in line with a process that regards all participants as free and equal directly justifies the process and legitimises its outcomes. As a result, democracy has final value rather than non-final value because it is not valued for the quality of its decisions but for the way its decisions are made-i.e. with all participants treated appropriately as free and equal, public equal or social equal (Rawls 1993;Christiano 2008;Viehoff 2014;Kolodny 2014a, b). 13 These outcomes need not be only distributive, however. See Dworkin (1987, pp. 19-23). 14 This is made clear by looking at various sources on the topic (e.g. Arneson 2003, p. 122;Dworkin 1987, p. 2;Wall 2006, p. 100;Martí 2006, p. 33). Whether an instrumentalist justification is convincing in deriving the legitimacy of all outcomes from their correct content on average is an entirely different question. See Kolodny (2014a, p. 202) for an objection. 15 For the same reason, the present characterisation also differs from the taxonomy proposed by Peter (2009, p. 67). Peter specifies various approaches within proceduralism arguing that some of them have a double requirement for the legitimacy of democratic decisions: democratic pedigree and substantive quality, i.e. rationality or epistemic correctness. On the contrary, I argue that instrumentalism as well is concerned with procedures and hence dispenses with the quality of each and every outcome, as long as the procedure is one that generally produces good outcomes.

Contribution
The fact that democracy is valued for its own sake does not necessarily mean that such a value is non-derivative. The second variable, contribution, aims to assess whether democracy is valuable in a non-derivative way or if it is valuable because of the contribution that it makes to other values. In the latter case, the value of democracy is extrinsic even though democratic procedures may still be valued for their own sake; in the former case, democracy's value is final and intrinsic.
One could argue that all instrumental justifications are extrinsic because they value democracy as a means to something external to democracy itself. This idea is fairly intuitive-if the value of democracy is merely instrumental, then it is derived from the contribution that democracy makes to the quality of outcomes and it is wholly reducible to this contribution. For instance, if I value my car as a means to go to work, the car's value is clearly instrumental, as its value is derived entirely from the outcomes it makes possible. Similarly, if the epistemic quality of outcomes is what an instrumental account of democracy is after, then democratic procedures are indirectly valuable as a means to epistemic quality and extrinsically valuable given their contribution to epistemic quality.
One could think that proceduralism, on the other hand, should value democracy intrinsically and for its own sake. Considering the previous example, not only would I value my car for its own sake instead of its capacity to get me to work, but also there would be no other value from which my car draws its value. Admittedly, this seems peculiar when using a vehicular analogy; however, the concept is worthwhile because procedural accounts can be seen in this way. According to intrinsic proceduralism, the source of democratic value lies completely within democracy, to the extent that it makes little sense to even consider a 'justifying value' external to the democratic procedure itself. This is what David Estlund and Fabienne Peter seem to be after when they dub these considerations the 'procedure-independent standard of justice' or 'procedure-independent criterion for ideal or correct outcomes' (Estlund 2008, p. 29;Peter 2009, p. 66). 16 In the literature, so-called pure or deep proceduralism is characterised by a denial that procedure-independent standards are necessary to vindicate democratic legitimacy.
Let us now look at an example. According to Joshua Cohen's deliberative conception, democracy is defined as reasoned, inclusive and public deliberation among free and equal citizens and is valued not only for its own sake but also non-derivatively (1997a, pp. 69, 74-75). It would be wrong to cast deliberation as a completely different value with respect to democracy, since the idea of free and public reasoning among equals seems to be an essential feature of Cohen's definition. Fundamentally, deliberation is what makes a decision-making procedure truly democratic as well as what justifies democracy as a valuable and legitimate procedure.
There are two major problems with Cohen's intrinsic view. The first concerns the relationship between the evaluative and definitional components of the account, Extrinsic Democratic Proceduralism: A Modest Defence which is quite blurry. On the one hand, Cohen's justification refers to the ideal of deliberative democracy, which views open and rational deliberation among free and equals as the legitimate decision-making procedure (Cohen 1997a, p. 72). And yet, on the other hand, he seems to take this justification to account for the legitimacy of actual democratic outcomes (Cohen 1997a, pp. 73-79). For this reason, Cohen's conception of deliberative democracy has been accused of neglecting the distinction between actual and ideal procedures (Estlund 2008, p. 92;Ottonelli 2012, p. 133ff).
This article allows one to see more clearly why this is the case. The ideal conception of deliberative democracy that Cohen endorses serves two conflicting purposes: it represents both the definition of democracy, which is supposed to identify democratic institutions, and the value that justifies those institutions. As a result, either a regime is both deliberative-democratic and legitimate or it is neither. This way of thinking obscures some important distinctions, such as, for instance, the one between actual democracies (however defective and not fully democratic they are according to the deliberative conception) and authoritarian regimes, or the one between procedural and substantive requirements for democratic decision-making. 17 A potential answer from a proponent of intrinsic proceduralism is that value and definition are not perfectly coextensive-actual democracies are such because they sufficiently realise the justifying value. To give an example, using Cohen's definition of democracy as free, public and rational deliberation among equals (Cohen 1997b, p. 412), actual institutions are sufficiently democratic if they guarantee the existence of a free public sphere and give each citizen an equal opportunity to influence collective decisions. These are still partial embodiments of the intrinsic value of deliberative democracy because they fail to ensure that public and rational deliberation takes place for all political issues and because the kind of political equality enjoyed by citizens is merely formal. 18 Indeed, this appears to be what Cohen himself has in mind when he states that actual institutions should mirror, 'so far as possible', the ideal deliberative procedure (Cohen 1997a, p. 79). This mirroring idea means that their democratic pedigree serves as a valid reason to consider outcomes as legitimate only if the process behind those outcomes sufficiently resembles the ideal deliberative procedure (Cohen 1997a, pp. 79-80). 19 This response, however, raises three related issues. First, it is difficult to precisely identify the minimal threshold-how many elements of the deliberative ideal need to be realised for a regime to be deemed legitimate? The importance of such a question cannot be overstated, because citizens need to know if the regimes they live under are democratic and have a right to rule over them. Second, it is far from clear which institutional, social or cultural features are necessary for a regime to pass the 17 A similar problem is forcefully illustrated by Peter's case of the political egalitarian's dilemma. According to Peter, Cohen's account of deliberative democracy is inconsistent because it allows to interpret political equality, one of its core features, in a conflicting way (2007, p. 384). 18 I credit an anonymous reviewer for this observation. 19 This seems to be what Cohen has in mind in another article in which he draws a clear distinction between the definition of democracy as collective authorisation of the exercise of state power (Cohen 1997b, p. 407) and its value, which is vindicated when such an authorisation proceeds 'on the basis of a free public reasoning among equals' (Cohen 1997b, p. 412). threshold and, thus, which should be given priority. For instance, is a decision-making procedure that gives equal voice to all while leaving decisions to only a few more or less democratic than a procedure wherein all have an equal right to vote, but unequal opportunities to express their opinion? Insofar as actual institutions are required to conform to the ideal procedure, questions like these will inevitably arise and we currently lack a principled way of answering them. 20 Third, as Estlund has pointed out, it is far from obvious that the so-called mirroring view (i.e. the direct embodiment of the ideal within actual institutions), represents the best strategy to realise such an ideal, especially given that actual political circumstances differ radically from the ones assumed in the deliberative ideal (2008, pp. 172-174, 190-191).
Nonetheless, the second and more serious problem with intrinsic proceduralism concerns the way the evaluative component is carried out. To demonstrate intrinsic value, one can deploy various strategies that scholars in axiology and metaethics have offered. Any expansion of this issue is beyond the scope of this paper, but it suffices to say that intrinsic value should be justified in a way that is independent of other values. Indeed, when we value friendship for its own sake and non-derivatively, we mean that there is no other value that friendship contributes to. Viewing democracy in a similar way, no other values, such as freedom or equality, should figure into the justification. In other words, if democracy's intrinsic value lies in free, public and rational deliberation among equals, this value cannot be reduced to a combination of freedom, equality, inclusion and public deliberation; if it were, the value of democracy would simply be derived from these values taken together. 21 Hence, the arguments for democracy as intrinsically valuable cannot appeal to any other values beyond the internal features of democracy itself.
This requirement, however, puts the bar gratuitously high for democratic legitimacy, as it presents us with a misleading dilemma-either we are able to show that democracy has value independent of anything else in the world or it is a mere means to some other end. To a certain extent, this seems to fit with proceduralists' famous charge against instrumental accounts, namely that the latter assumes the existence of substantive and procedure-independent values with respect to which democracy is reduced to a merely contingent means (Peter 2009;Rostbøll 2015a).
However, that there are no independent standards being applied to outcomes and that there are no standards independent of actual democratic procedures are two 20 Or, if we did answer them, it would be by prioritising certain aspects in the ideal of deliberative democracy over others and this, in turn, would not only break the unity of the deliberative democratic ideal, but also call for a justification of the priority of certain values composing that ideal over others. Such a conclusion would move towards an extrinsic form of proceduralism. A similar criticism is raised by John Dryzek against the 'deliberative systems approach', the criteria of which apply separately to various components of deliberative democracy but fail to apply to the overall system (2017, p. 621). 21 Additionally, among the various values characterising the deliberative ideal, we seem to lack a principled way to put them into order, meaning it is not obvious which values (i.e. freedom, equality, inclusion or public deliberation) a decision-making procedure should prioritise to be considered democratic and what kind of trade-offs among them are called for if they are not simultaneously realisable. To give an example, Christian List has shown that a democratic procedure cannot simultaneously satisfy the three desiderata commonly thought to collectively constitute democracy: robust pluralism, basic majoritarianism and collective rationality (2011, p. 289).

3
Extrinsic Democratic Proceduralism: A Modest Defence different claims. Certainly, all procedural accounts subscribe to the former claim, as they value democratic procedures for themselves and not only as a means to certain outcomes, but they do not need to endorse the latter. Proceduralists can admit the existence of standards independent of actual democratic procedures. What has not been sufficiently stressed is that these procedure-independent standards need not be outcome-related; there can be standards that are independent of the actual procedure and, at the same time, refer directly to it and not through its outcomes. This weaker view holds that normative standards justifying democracy do not amount to the features characterising democracy as a distinctive decision-making procedure but cannot be fully realised in practice outside of democratic institutions (Rostbøll 2015a, pp. 271-273;Peter 2009, p. 6). These accounts recognise the existence of values from which democracy derives its value-such as public or relational equality, political or epistemic fairness, and republican non-domination-and distinguish them from actual democratic procedures. Yet they reject the possibility that the justifying value is realised, institutionalised or embodied in ways that are undemocratic. In such a way, the value of democracy is derived from the contribution that it makes to the realisation of the justifying value; however, such a contribution is not a causal driver of democratic outcomes. For this reason, we can call these approaches extrinsically procedural.
Take the case of a wedding ring for a loving couple. It seems unreasonable to value it intrinsically. It is a golden ring, as such expensive, but the couple is likely to derive more value from the meaning that the ring holds for them rather from the ring itself. The value of a wedding ring is derived from its relational properties and from the contribution it makes to the love spouses have for each other. It is not a mere means yet it is valuable in a derivative way. In a similar manner, democracy can be justified by virtue of the direct and non-causal contribution that it makes to relational equality or autonomy by making them visible and politically meaningful. Contrary to a belief that some scholars hold, 22 a justification of this kind is still procedural and not instrumental.
An extrinsically procedural defence of democracy has three merits: (1) it clearly acknowledges the distinction between the definitional and evaluative components because it can draw a clear line between what democracy is and why it has value; (2) it does so by referring not to outcome-based values but rather to values that apply directly to democratic procedures and view them as derivatively and non-instrumentally valuable; and (3) it is able to convincingly argue for the value of democracy in terms of other values that are conceptually distinct but practically indistinguishable from actual democratic procedures.
The following question arises: are proponents of extrinsic proceduralism compelled to view democracy as contingently valuable, as instrumentalists do? The final variable of distinction focuses on the modality of the justification and serves to further analyse the connection between the justifying value and democratic procedures.

Modality
The last variable allows us to look more closely at the connection between the evaluative and normative components of the implicit view. Once what counts as a democracy is identified and the reasons for its value are detailed, modality serves to explain under which conditions and limitations the fact that a decision is democratic makes the decision legitimate. To shed more light on this component, we must focus on the relationship between democracy and the justifying value. We do this by turning to a set-theoretic conceptualisation of necessary and sufficient conditions; in this manner, we see four possibilities concerning said relationship. Figure 1a shows a case in which democratic procedures sometimes overlap and sometimes do not overlap with the realisation of the relevant justifying value. Here, democracy can be defended as valuable only contingently. Figure 1b illustrates how democracy is both necessary and sufficient to realise the justifying value so that all and only those procedures that we can identify as democratic will be justified. Finally, Fig. 1c shows a case in which democracy is a necessary but insufficient condition for realising the justifying value; Fig. 1d shows a case in which democracy is a sufficient but unnecessary condition for realising the justifying value. Which case do instrumentalism and extrinsic proceduralism conform to?
Clearly, instrumentalist accounts connect the justifying value to democracy in a contingent way (Fig. 1a). If democracy is only a means to something else, its value will hinge on its effectiveness as a means and hence is a matter of contingency (Arneson 1993, pp. 123-124;Van Parijs 1996, p. 110;Wall 2007, p. 437). Proponents of instrumentalism claim that all democratic decisions are legitimate on account of the democratic factor regardless of whether the decisions themselves are substantively just (Arneson 2003, p. 122;Dworkin 1987, p. 2;Wall 2006, p. 100). If most of these decisions are unjust, however, they jeopardise the justification of democracy as a procedure that produces just decisions more often than not. 23 By the same token, we should be able to consider a substantively just decision as legitimate even if it is not democratically sanctioned (Stemplowska and Swift 2018). In cases of conflict between unjust democratic decisions and just undemocratic decisions, we are left without a principled reason to know which of them should count as legitimate. It is clear that instrumentalism allows only for a weak account of democratic legitimacy, especially given reasonable pluralism on what the relevant justifying value should be.
In contrast to instrumentalism, procedural accounts generally hold the stronger claim that democratic procedures are necessary to realise the justifying value because said value directly refers to democratic procedures. Figure 1b represents intrinsic proceduralism, according to which democracy is defined as open and rational deliberation among free and equal citizens and, thus, democracy is necessary to realise the justifying value, which in this case is the deliberative ideal. Furthermore, democracy so defined is also sufficient for the realisation of the deliberative ideal.
As we have seen, however, intrinsic proceduralism does not hold the hyperbolic claim that all actual democratic procedures fully realise the justifying value; rather, according to the mirroring view, actual procedures are only a partial embodiment of the intrinsic value of democracy. If this is the case, these actual procedures cannot be coextensive with the procedures realising the justifying value (i.e. the ideal conception of deliberative democracy). Instead, one of the two other cases applies (Fig. 1c or d) because actual procedures are either necessary or sufficient for realising the value of deliberative democracy. Interestingly, in this respect, intrinsic proceduralism does not differ from extrinsic proceduralism, to which I now turn.
First, democracy may be thought of as necessary, albeit insufficient (Fig. 1c). According to Christiano, for instance, democracy is necessary to realise the principle of public equality-there is simply no other way to obtain public equality in the decision-making process (Christiano 2008, pp. 9, 78ff). Similarly, Daniel Viehoff states that a procedure must be egalitarian to guarantee equal control over mutual relations (2014, p. 369); Niko Kolodny argues that democracy is necessary to realise a 'full or ideal social equality' (Kolodny 2014b, p. 308). Accordingly, all procedures fulfilling the requirement provided by the justifying value must be democratic because these procedures are included in a subset of all possible democratic procedures. Of course, some democratic procedures will fail to realise the justifying value, despite the fact that they are themselves democratic. As we can see in Fig. 1c, if democracy is necessary for the realisation of the justifying value, then all justified procedures should be democratic (i.e. no decision can be simultaneously undemocratic and legitimate).
However, the tenet that democracy is necessary for the realisation of the justifying value cannot stand if there are procedures that realise such a value without being democratic. This is the case, for instance, with some recent contributions to democratic theory involving the concept of relational equality, for which Kolodny and Viehoff are the most renowned advocates. 24 Although their accounts are slightly different, they both justify democracy on the basis that it realises the principle of either 'rule over none' or the absence of unequal power relations. There are many other procedures that can satisfy this principle, however, without being democratic in any recognisable sense.
For instance, consider a decision-making procedure wherein individuals write legislative proposals on pieces of paper and put them in a jar from which they are randomly selected and implemented. No one would institutionally rule over anyone else, as the laws would be a matter of chance rather than a result of anyone's will. Even the person who cast the piece of paper eventually selected does not enjoy any kind of superiority with respect to her fellow citizens; the reason why her piece of paper becomes law is not that she decided for it, but that the piece of paper selected by chance was hers. Alternatively, consider a procedure wherein citizens cast an equal vote without having any idea what their options are. 25 Procedures like these, which do not technically violate the principle of relational equality, are clearly undemocratic in the conventional sense.
Kolodny, perhaps bearing in mind a similar concern, defines a democratic decision as 'a process that gives everyone subject to it equal or both equal and positive [emphasis added], formal or both formal and informal opportunity for informed influence either over it or over decisions that delegate the making of it' (Kolodny 2014a, p. 197). 26 According to this definition, however, as long as the influence that a procedure grants is equal, such a procedure would qualify as democratic even though the equal influence amounts to no influence at all. 27 Kolodny admits that a procedure characterised by positive equal influence over collective outcomes is not the only potential justified decision-making procedure according to relational equality. 28 Nonetheless, it seems peculiar to define a decision that bestows no power on citizens and does not include any kind of popular control as democratic. A procedure like this would certainly be egalitarian but it could not align with the various institutions that we generally consider to be essential components of a democratic regime, such as universal suffrage and equal political rights. Therefore, it seems plausible to Fig. 1 Set-theoretic conceptualisation of the relationship between democratic procedures (DP) and the justifying value (JV) 25 Kolodny would exclude this possibility, as it would not qualify as informed influence; however, it is difficult to see how this epistemic requirement can be grounded in the principle of relational equality. As long as all citizens are equally ignorant about the vote they are casting, the principle of relational equality is respected (2014b, p. 310). 26 Kolodny holds a dualistic justification for democracy similar to that of Estlund, according to which the first step identifies the procedures that are justified from the standpoint of relational equality and the second step singles out democracy as the only procedure that is epistemically reliable. Here, I refer only to the procedural element of his justification. 27 Similarly, Viehoff speaks of 'egalitarian procedures', broadly intended (2014, p. 369). 28 Kolodny rightly observes that choosing public officials by lottery would breach the principle of social equality, as it would instantiate unequal relations while these people hold office (2014a, p. 228). However, his account does not rule out cases of lottery voting, such as the one proposed by Saunders (2010) or others that were mentioned in the text. think of recognisably democratic procedures as representing a subset of procedures that satisfy the principle of relational equality. If this is the case, however, the bigger set that encompasses democratic procedures is that of all egalitarian procedures, as in Fig. 1d; in this case, democracy may be sufficient without being necessary to the realisation of the justifying value.
If this objection is sound, equality-based defenders of extrinsic proceduralism must dispense with the claim that democracy is necessary for the realisation of relational equality. 29 The set-theoretic explanation shows how to prove whether the same applies to other extrinsically procedural accounts-insofar as it is possible to come up with clearly undemocratic procedures that realise the justifying value, democracy does not qualify as necessary for political legitimacy. 30 Proponents of extrinsic proceduralism can claim that democracy is sufficient to the realisation of the justifying value (Fig. 1d). If all democratic procedures constitute a subset of the procedures that realise the justifying value, no democratic decision can fail to be an instantiation of the at-stake justifying value (Waldron 1999;Bellamy 2007). 31 However, these accounts must contend with the idea that proceduralism cannot vindicate the necessity of democracy for political legitimacy.

Conclusion
This article aimed to shed more light on the traditional distinction between instrumentalism and proceduralism by introducing three variables: reference, contribution and modality. Instrumentalist justifications of democracy are characterised by: (1) an indirect reference to democracy via its outcomes; (2) the idea that democracy is only extrinsically valuable by virtue of its causal contribution to the justifying value; and (3) the idea that democracy realises its value in a primarily contingent way. In contrast, procedural accounts: (1) always make a direct reference to democracy; (2) are more convincing in the extrinsic version; and (3) take democracy as either necessary or sufficient for political legitimacy. As I argued in the article, extrinsic proceduralism avoids two objections levelled against intrinsic proceduralism-it clearly maintains the distinction between the definitional and evaluative components of the justification and is 29 Or they could claim that we should extend our definition of democracy to encompass all egalitarian procedures. However, such a response first transforms extrinsic proceduralism in another form of intrinsic proceduralism with respect to all egalitarian procedures; in turn, it gives rise to the same objection concerning the relation between ideal and actual procedures that has already been mentioned in this paper. Second, this extended conception of democracy, which includes all egalitarian procedures, seems to impoverish the commonly held idea of democracy in modern regimes. 30 Consider, for instance, the value of democracy as collective self-determination, which we find in Anne Stilz (2016, pp. 109-110). There are other undemocratic institutional arrangements that can realise the principle so long as people reasonably affirm them. 31 Importantly, some decisions may still fail to be legitimate all things considered because there may be other more fundamental reasons that countervail democracy and the justifying value upon which it rests and, therefore, invalidate its legitimacy. able to account for the value of democracy in terms of other values without wading into instrumentalism. While equality-based versions of extrinsic proceduralism cannot vindicate the claim that the latter is necessary for political legitimacy, they can still defend it as sufficient. What is more interesting, however, is that intrinsic proceduralism as well cannot account for both the necessity and sufficiency of democracy for political legitimacy insofar as it needs to acknowledge the distinction between the actual procedures defined as democratic and the intrinsic value of deliberative democracy appealed to in order to justify them.
Democratic theory has many aims. One of them consists of providing the standards with which we can identify regimes as democratic, the reasons to account for their value as democratic regimes, and a theory that explains why, to what extent and under what limitations the fact that democratic regimes are justified according to some value makes their outcomes legitimate. Naturally then, the fact that a decision-making procedure is democratic does not necessarily mean that all of its outcomes are legitimate (i.e. it does not mean that democracy is a sufficient condition for realising the justifying value). Nor does it mean that the justifying value is the only ground for political legitimacy, though it can be.
The takeaway for democratic theory is as follows: the weaker the connection is among democracy, the justifying value and legitimacy, the more dubious we will be concerning the legitimacy of each democratic decision and the value of preserving our states as democratic; if this connection is too stringent, however, we will lack the normative resources to criticise democratic outcomes' legitimacy even if they are democratic. Extrinsic proceduralism constitutes the best way to achieve the proper balance, though I am inclined to think that it can do so fruitfully only if it also manages to vindicate the necessity of democracy for political legitimacy. Relational equality-based accounts fail in this crucial respect. Future research should look into whether there are extrinsically procedural accounts that succeed.