Why a Fair Compromise Requires Deliberation

I argue in this paper that the process of compromising needs to be deliberative if a fair compromise is the goal. More specifically, I argue that deliberation is structurally necessary in order to achieve a fair compromise. In developing this argument, this paper seeks to overcome a problematic dichotomy that is prevalent in the literature on deliberative democracy, which is the dichotomy between compromise and deliberation. This dichotomy entails the view that the process preceding the achievement of a compromise is essentially a process of negotiating or bargaining, which, I claim, should not be the case if a fair compromise is the goal. The reason for this claim is, in a nutshell, that negotiation or bargaining processes do not provide for an in-depth understanding of the reasons that each party has for holding their respective position. However, an in-depth understanding of each other’s reasons, is, as I will show, a necessary condition for achieving a fair compromise. In contrast to negotiation or bargaining, the deliberative process, by its very structure, provides for mutual understanding and is therefore a necessary condition for achieving a fair compromise.


Introduction
This paper argues that the process of compromising needs to be deliberative if a fair compromise is the goal. 1 More precisely, I argue that deliberation and compromise are inextricably connected in that deliberation is a structural necessity for achieving a fair compromise. This is not to say that the importance of deliberation for compromise, generally speaking, has not been recognized before. The idea that compromise is connected to deliberation has been pointed out by several deliberative democrats. Amy Gutmann and Dennis Thompson, for example, suggest that without deliberation, 'we forsake the possibility of arriving at a genuine moral compromise' (Gutmann & Thompson, 2004: 12). Similarly, Daniel Weinstock points out that 'the attempt to arrive at a compromise is an exercise in moral deliberation rather than a simple exercise of "horse trading"' (Weinstock, 2013: 540). More recently, Weinstock emphasizes that in the context of pluralist societies, ' compromise should be the goal that political deliberation sets for itself' (Weinstock, 2017: 636). And Peter Jones and Ian O'Flynn state that ' a substantively fair compromise is more likely to arise if the compromising process takes the form of deliberation rather than bargaining' (Jones & O'Flynn, 2013: 127).
However, while these authors acknowledge the necessity of deliberation for compromise in general, their accounts do not demonstrate why exactly deliberation is required for achieving a fair compromise. In clarifying the procedural nature of compromising, this paper aims to fill this gap. More precisely, this paper contributes to ongoing debates about the relationship between compromise and deliberation in the following way: I argue that a fair compromise needs to be based on a process that is purely deliberative, from the beginning to the end. None of the existing discussions regarding the connection between deliberation and compromise have yet made this specific claim, which is, however, crucial to our understanding of achieving a fair compromise in practice. If my argument is correct and we can indeed achieve a fair compromise only through a process of proper deliberation, it would be detrimental to seek a fair compromise through (deliberative) negotiation, as existing accounts tend to suggest.
My argument is structured in the following way. In section I, I explain why it is necessary to clarify the procedural nature of compromising. For this purpose, I illustrate a problematic dichotomy that is prevalent in the literature on deliberative democracy, which is the dichotomy between compromise and deliberation. This dichotomy entails the view that the process preceding the achievement of a compromise is essentially a process of negotiating or bargaining, which I claim should not be the case if a fair compromise is the goal. The reason for this claim is, in a nutshell, that negotiation or bargaining processes do not provide for an in-depth understanding of the reasons that each party has for holding their respective position. However, an in-depth understanding of each other's reasons, is, as I will show, a necessary condition for achieving a fair compromise. In section II, I propose a conception of fair compromise according to which a fair compromise requires that each party's concessions are proportional to what is at stake for the parties respectively. In section III, I use this conception of a fair compromise to develop the argument that deliberation is structurally necessary for achieving a fair compromise. From this argument follows the claim that the process of compromising needs to be deliberative if a fair compromise is the goal. Section IV concludes the paper with a reflection on the question how important it is that a compromise is fair.

I. Compromise vs. Deliberation: A Problematic Dichotomy
One of the primary concerns for deliberative democrats is the question of how we can legitimately accommodate moral disagreement in society and politics. As Gutmann and Thompson state, '[t]he general aim of deliberative democracy is to provide the most justifiable conception for dealing with moral disagreement in politics' (Gutmann & Thompson, 2004: 10). Traditionally, it has been assumed that disagreement is best addressed through a deliberative process that focuses on the common good and that eventually leads to consensus. According to Jane Mansbridge et al. 2010: [i]n the classic ideal [of deliberation], individuals enter a deliberation with conflicting opinions about what is good for the polity, but after voicing and hearing the reasons for different options, converge on one option as the best, for the same reasons. Ideally, the deliberation is based on reason. It aims at consensus and the common good (66).
This classic ideal is represented in the works of 'firstgeneration' deliberative democrats, such as Jürgen Habermas and John Rawls (Elstub, 2010: 291). According to these early deliberative democrats, public deliberation should aim at consensus, because consensus is the ideal outcome of a rational exchange of reasons. That is, ideally, the participants to a deliberative process will find a consensus in light of new arguments and evidence. Consensus therefore reflects the reasoned points of view of the (previously) disagreeing parties. Thus understood, achieving a consensus meant that the deliberation process was successful (Elstub, 2010). In contrast, the classic ideal of deliberation excludes compromise as an acceptable outcome of a deliberative process. 2 As Jon Elster puts this point, '[n]ot optimal compromise, but unanimous agreement is the goal of politics on this view' (Elster, 1997: 12).
However, deliberative democrats increasingly acknowledge compromise as a legitimate solution to disagreement. Indeed, as Gutmann and Thompson point out, 'many deliberative theorists now not only recognize but also insist on the need for, and value of, political compromise' (Gutmann & Thompson, 2012: 84). For Gutmann and Thompson, compromise is a particularly desirable way to respond to fundamental disagreement that would otherwise persist at the expense of political progress and peace.
Other deliberative democratic theorists argue that the deliberative process can and should aim at compromise as a proper goal rather than merely as an alternative in case a consensus turns out to be unrealistic. For example, Richard Bellamy states that ' differences have to be continually and democratically negotiated with compromise not consensus as the goal' (Bellamy, 2000: 216). Similarly, Weinstock argues that ' compromise rather than consensus is the goal toward which political debates aim' (Weinstock, 2006: 244). Here, the idea is that compromise can be a more adequate solution to disagreement than consensus, especially in cases of reasonable disagreement where all parties to the disagreement have valid claims. More specifically, compromise is endorsed as an adequate response to what John Rawls has called 'the fact of reasonable pluralism' (Rawls, 2001: 4). 3 The argument here is that compromise, but not consensus, can accommodate the reasonable disagreements that are part of pluralistic societies. That is, in the case of reasonable disagreement, compromise, but not consensus, allows for equal concern and respect for all the reasonable views that are in conflict (Bellamy, Kornprobst & Reh, 2012). 4 More recently, Federico Zuolo and Giulia Bistagnino (2018) address the desirability of compromise from an epistemic perspective. Seeking an answer to the question of how we should deal with deep, normative disagreements in pluralist, democratic societies, the authors argue that the recognition of epistemic parity may provide principled reasons for seeking a compromise. Christian Rostbøll sheds light on the desirability of compromise from yet another angle. Questioning the pervasive conception of compromise as ' a regrettable necessity' (Rostbøll, 2017: 620), he proposes that compromise has a proper democratic significance that reveals itself if we adopt a conception of respect according to which we conceive of fellow citizens as co-rulers.
The increasing acknowledgement of the importance of compromise for deliberative democracy involves, however, a problematic dichotomy. A default assumption in the literature on deliberative democracy is that deliberation is required only if we aim for consensus. If, on the other hand, we aim for compromise, non-deliberative processes such as negotiation or bargaining are considered to be more appropriate. 5 This view is represented, for example, in Philippe Van Parijs's statement that 'negotiation can lead to a compromise that avoids the costs and risks of conflict, exit or arbitration, whereas deliberation can lead to a consensus about what is required for a fair solution' (Van Parijs, 2012: 469;emphasis in original). Or, as Simon May puts it, ' compromise is the paradigmatic feature of negotiation' (May, 2018: 150). In brief, the idea is that we achieve consensus through deliberation and compromise through negotiation.
This view is problematic because it indicates that deliberation and compromise are mutually exclusive phenomena, which, as I argue in this paper, is not the case. In this context, one might ask whether the idea of a ' deliberative negotiation', which has recently been introduced by Mark Warren and Jane Mansbridge (2016), does not already constitute the required alternative to the problematic dichotomous view. Deliberative negotiation denotes a negotiation process that entails deliberative features, such as mutual justification, respect and a concern with fairness ( Warren & Mansbridge, 2016: 151). By introducing this concept, the authors intend to do justice to the fact that negotiation processes can entail deliberative elements, which, in their view, has been largely neglected not only in theory but also in practice.
However, deliberative negotiation constitutes a conceptual alternative to a different dichotomy than the one that I criticize in this paper. More precisely, deliberative negotiation addresses the dichotomy between deliberation and negotiation, not between deliberation and (fair) compromise. That is, Warren and Mansbridge problematize the distinction between deliberation and negotiation by pointing out that negotiation processes can be characterized by distinctively deliberative features. But they still conceive of compromise, including fair compromise, as a possible outcome of (deliberative) negotiation and not of deliberation proper (or, as they call it, pure deliberation). Put differently, even though Warren and Mansbridge's conception of deliberative negotiation entails the possibility that a fair compromise is based on pure deliberation, it is not clear from their account that a fair compromise requires pure deliberation necessarily. According to Warren and Mansbridge, a fair compromise could also result from a negotiation process that merely entails deliberative moments. This paper, in contrast, argues that a fair compromise necessarily needs to be based on deliberation properly speaking.
To be clear, I do not claim that the assumptions inherent in the dichotomous view are mistaken as such. Of course, deliberation can and should lead to consensus; and clearly, negotiation and bargaining can and should lead to compromise. My point, rather, is that we are mistaken to assume that negotiation or bargaining can reliably lead to a fair compromise. By providing an in-depth understanding of the reasons why a fair compromise requires deliberation, the arguments put forward in this paper not merely question those accounts that perpetuate the dichotomous view, but also support existing accounts that generally acknowledge a connection between deliberation and fair compromise.

II. A Conception of Fair Compromise
Since the argument put forward in this paper concerns fair compromise specifically, it is crucial to consider the sense in which a compromise may be said to be fair. According to Jones and O'Flynn, a compromise can be fair with regard to its content or with regard to the procedure on which it is based. As a matter of procedural fairness, a compromise is fair if it is based on a fair process. As a matter of substantive (or end-state) fairness, a compromise is fair if it reflects a fair outcome (Jones & O'Flynn, 2013). This paper is concerned with fair compromise in the substantive rather than the procedural sense. The question then is: What kind of outcome qualifies as a fair compromise?
Before addressing this question, a brief conceptual clarification is in order. Van Parijs argues that once the disagreeing parties consider an outcome to be fair, we no longer speak of a compromise, but of a consensus. He provides the following example: 'Suppose a cake needs to be split between you and me in circumstances in which we would each like to eat the whole thing. There is no compromise, but rather a consensus if we both believe that cutting the cake in half is fair' (Van Parijs, 2012: 470). I disagree. A fair compromise is still a compromise and not a consensus, because neither party gets what they initially wanted -which, in Van Parijs's example, is the whole cake. Put otherwise, the parties have not changed their minds with regard to their initial desire to eat the whole cake. It is, however, precisely this change of mind that is required for speaking of a consensus. In Van Parijs's example, even though the disagreeing parties might consider it to be a fair solution that each party gets half of the cake, they do not consider this agreement to be superior (or equally good) to getting the whole cake. Therefore, a fair compromise is still a compromise, not a consensus. 6 To resume the characterization of a fair (substantive) compromise, it is useful to acknowledge that every compromise is characterized by mutual concessions (Bellamy, Kornprobst & Reh, 2012;Gutmann & Thompson, 2012;Jones & O'Flynn, 2013;Lepora, 2012;Margalit, 2010;Warren & Mansbridge, 2016). As Van Parijs points out, '[a] compromise is an agreement, but not just any agreement. Its distinctiveness resides in the mutual concessions involved' (Van Parijs, 2012: 469). I therefore suggest that the fairness of a compromise is to be evaluated with regard to the concessions that each party makes. More precisely, I suggest that a fair compromise is characterized by a proportionality of concessions; and I further suggest that the proportionality of concessions is to be determined in relation to the stakes that are involved for each party to a disagreement.
This conception of a fair compromise is based on the 'proportionality principle' that has been introduced by Harry Brighouse and Marc Fleurbaey. The proportionality principle holds that '[p]ower in any decision-making process should be proportional to individual stakes' (Brighouse & Fleurbaey, 2010: 138; emphasis in original). To be sure, Brighouse and Fleurbaey's primary concern is with the fair distribution of power in democratic decisionmaking processes, not with determining the fairness of compromise. But I suggest that the proportionality principle is well suited to determine the fairness of compromise as well, for the following reason. Brighouse and Fleurbaey justify the proportionality principle by pointing out that a proportional distribution of power is better suited to implement equal respect than an egalitarian distribution of power. 7 The reason for this claim is that in those cases where the stakes are not equal for all citizens and where those citizens with greater stakes are in the minority, an equal share of power often means that minority interests get buried under majority interests -even though it is the minority that has higher stakes in an issue. I suggest that the same reasoning applies to the context of compromise as well. We can hardly profess to show equal respect for two disagreeing parties if we require that both parties split their concessions equally, even though one party has much higher stakes involved in a disagreement. Thus understood, only a distribution of concessions that is proportional to individual stakes can manifest equal respect.
A fair compromise, as it is understood in this paper, therefore requires that each party's concessions are proportional to what is at stake for them in a specific disagreement. More concretely, this means that the concessions should be lower if the stakes are high and vice versa the concessions should be higher if the stakes are low. Stakes are here understood to pertain to a person's interests and moral values that are affected by a disagreement. For example, a person can be said to have high stakes in a disagreement if the values in question pertain to her identity, her conception of justice, or to general moral principles that play an important role in leading her life. Conversely, a person can be said to have low stakes in a disagreement if, for instance, she is merely interested in the topic of disagreement without having a deeper connection to the issue, in the sense that the values in question do not affect her sense of self or do not significantly violate moral principles that are important to her. I discuss the difference between high and low stakes more concretely in the next section.

III. Why Deliberation is Structurally Necessary for Achieving a Fair Compromise
In this section, I explain why deliberation is structurally necessary for achieving a fair compromise. 8 More specifically, I argue that the process towards achieving a fair compromise needs to be constituted by a deliberative structure from the beginning to the end. 9 As such, the process of compromising is characterized by a mutual exchange of reasons with the goal of mutual understanding and a fair accommodation of the claims in play -which means that the process of compromising is deliberative in nature. Or, to put this point differently, I claim that in the process of achieving a fair compromise, there is no room for bargaining or negotiation at all. This also means that the activity of compromising (with the goal of achieving a fair compromise) is not a process that comes after deliberating; rather, for the purpose of achieving a fair compromise, the activity of compromising is deliberating. 10 The claim that deliberation is structurally necessary for achieving a fair compromise is based on the conception of a fair compromise that I have developed in the previous section (i.e. the idea that a fair compromise is characterized by concessions that are proportional to the stakes that are involved for each party). I argue that if we accept this conception of a fair compromise, we also have to accept the claim that the process of compromising needs to be deliberative, from the beginning to the point where a compromise is agreed upon.
To develop the argument that compromising needs to be deliberative if a fair compromise is the goal, it is helpful to break the process of compromising down into three successive stages. All three stages are interrelated in the sense that each stage requires the previous stage. The first stage is to understand each other's reasons for holding a particular view. The second stage is to evaluate the stakes that are involved for each party. This second stage requires the previous reason exchange because what is at stake for each party is determined by the reasons that each party has for holding their view. The third stage is to determine the concessions that each party should make. This stage requires the previous evaluation of stakes because the concessions -if they are to be fair -need to be proportional to the stakes involved.
In what follows, I explain why deliberation is required for each of these three stages. I claim that for the purpose of achieving a fair compromise, all three stages need to be deliberative and only deliberative and that there is, therefore, no room for negotiation or bargaining in the process of compromising. 11

Understanding reasons
One of the central characteristics -if not the most distinctive feature -of the deliberative process is the mutual exchange of reasons between the participants (Gutmann & Thompson, 2004;Mansbridge et al., 2010;Thompson, 2008;Weinstock, 2013). By exchanging the reasons for holding their respective positions, the disagreeing parties come to understand each other's position (and possibly their own position as well) at a deeper level than they do before participating in the deliberative process. Such an in-depth understanding of each other's reasons is provided by the structure of deliberation, whereas that is not the case for the structure of bargaining or negotiation processes.
Quite to the contrary, negotiation and bargaining structurally support the ideal of 'winning' rather than the ideal of reaching mutual understanding. The ideal of winning that characterizes negotiation processes also involves the use of deception, lies and threats. In a survey article on negotiation, Leigh Thompson, Juinwen Wang and Brian Gunia point out that '[g]iven the mixed-motive nature of negotiation, it is tempting for negotiators to use deception to maximize their personal gain' (Thompson, Wang & Gunia, 2010: 501). 12 Similarly, Van Parijs characterizes bargaining as a process 'whereby each party uses threats and bluff in order to extract as many concessions from the other as it can get away with' (Van Parijs, 2012: 472).
To be sure, negotiation can also be viewed in a more positive light. A good example of a more benevolent perspective on negotiation is what David Luban calls the 'PPP Paradigm'. 13 According to the PPP Paradigm, 'negotiation is appeal to standards rather than psychological manipulation, it seeks joint rather than individual gains, and the opposite party is treated as a collaborator rather than an adversary' (Luban, 1985: 399). Similarly, May points out that negotiation typically involves both adversarial and cooperative aspects (May, 2018). But even if negotiation is not only adversarial, but also collaborative in the sense that it appeals to standards and that it seeks joint gains, the point remains that negotiation is structurally inapt to enable the kind of in-depth understanding that is required for achieving a fair compromise. As Gutmann and Thompson formulate this contrast between deliberation and bargaining: 'When citizens bargain and negotiate, they may learn how better to get what they want. But when they deliberate, they can expand their knowledge' (Gutmann & Thompson, 2004: 12). More specifically, negotiation only provides for what I call 'first-level understanding' but not for 'second-level understanding'. In a nutshell, first-level understanding pertains to understanding that someone holds a particular position, while second-level understanding pertains to understanding why someone holds a particular position. In what follows, I illustrate the difference between firstand second-level understanding with reference to an example of a disagreement on abortion (an example to which I will refer throughout this section).
Let us assume that a person, Ann, believes that abortion is morally right (R) and should therefore be legal (L), whereas another person, Bert, believes that abortion is morally wrong (W) and should therefore not be legal (NL). Ann and Bert therefore disagree on (R) vs. (W) and correspondingly on (L) vs. (NL). Let us further assume that Ann and Bert are unable to find a consensus on their disagreement and therefore aim to find a fair compromise. 14 Ann and Bert, as citizens who engage in a public agreement-seeking process, now have two options. They can try to find a fair compromise by negotiation or by deliberation.
If they negotiate, they can achieve first-level understanding. At this level of understanding, Bert understands that Ann wants (L) (and Ann understands that B wants (NL)). 15 First-level understanding is thus a minimal form of understanding, which is however sufficient for the purpose of negotiation because it enables the participants to comply with the structural requirements of negotiation. That is, first-level understanding enables Ann and Bert to each propose offers and counteroffers of what they are willing to concede.
The question is whether such a negotiation process can lead to a fair outcome in the sense that Ann and Bert's concessions are proportional to their stakes in this disagreement. This, I claim, is highly unlikely, because negotiation processes do not provide for second-level understanding, which, as I will show, is necessary for evaluating stakes in the first place. Instead, in a negotiation process, Ann and Bert make offers and counteroffers that are independent of the reasons that the other party has for holding their respective position. If Ann and Bert negotiate, their interaction is characterized by a willingness to mislead and deceive rather than to understand each other. In contrast to first-level understanding, second-level understanding means that Bert understands the reasons why Ann wants (L). That is to say, Bert understands that Ann wants (L) because she believes (R) and he also understands why Ann believes (R). It might, for example, turn out that Ann is a feminist who considers abortion to be right primarily from a feminist perspective -as a feminist, Ann believes that no one should have a say in what happens to a woman's body but the woman herself. To gain this second-level understanding of Ann's reasons, Bert needs to be structurally required to listen to Ann, while Ann needs to be structurally enabled to explain her reasons (e.g. by attributing specific time slots for reason exchanges of this kind). And it is deliberation, not negotiation, that can enable the parties to a disagreement to exchange their reasons in this way and thus to develop a second-level understanding of the other person's view. This is significant because second-level understanding is necessary for evaluating stakes and thus, ultimately, for achieving a fair compromise. The reason why this is so will become clear in the subsequent argument. But in a nutshell, the idea is that second-level understanding is a necessary precondition for a meaningful evaluation of stakes because it is the reasons that each party has for holding their position that determines what is at stake for them. And evaluating mutual stakes is, in turn, the precondition for determining fair concessions because a fair compromise requires that each party's concessions are proportional to the stakes that each party has in a disagreement.
In sum, the second-level understanding that is the foundation for achieving a fair compromise can only be achieved through deliberation, but not through negotiation. This puts the deliberative process in a unique position to enable a fair compromise, while negotiation structurally fails in this regard.

Evaluating stakes
Second-level understanding, through deliberation, is the first of three stages towards achieving a fair compromise. As indicated above, second-level understanding is the precondition for evaluating what is at stake for each party. This is so because what is at stake for each party is determined by each party's deeper reasons for holding their respective views.
To continue with the above example, we have established that Ann supports (L) for feminist reasons. Indeed, let us assume that Ann has been identifying with the feminist movement for several years and that she has invested a significant amount of her time in advancing the cause. Feminism has therefore become an important part of Ann's identity. Thus, given that the disagreement on abortion directly pertains to feminist values, Ann can be said to have high stakes involved in this disagreement.
As for Bert, let us assume that Bert is currently enjoying the relaxed days of retirement and he realizes that he finally has enough time to engage with topics of political interest. Even though Bert has never given much thought to the topic of abortion, when he comes across the argument that abortion is murder, this image immediately sticks with him and he finds himself appalled by the idea of abortion. Bert therefore wants (NL). Now let us assume, for the sake of the argument, that in this case of disagreement between Ann and Bert, the stakes are much higher for Ann than they are for Bert. 16 In this scenario, a fair compromise requires that Bert recognizes that Ann has higher stakes involved in the disagreement than he does. But, or so I propose, if Ann and Bert negotiate, it is highly unlikely that Bert can adequately assess Ann's stakes. This is so because, as elaborated above, negotiation processes do not intend for the participants to evaluate each other's stakes. By its very structure, negotiation has no room for stake evaluationwhich makes sense. After all, why would I be interested in knowing what is at stake for you, if what matters, in the end, is that I win against you?
In contrast to negotiation, deliberation allows for an evaluation of stakes because it structurally provides for a meaningful exchange of arguments. This is necessary not merely at the first stage of understanding each other's reasons, but also at the stage of evaluating stakes, for three reasons.
First, to evaluate what is at stake for Ann, Bert will need to learn more about how the abortion question pertains to Ann's identity as a feminist and perhaps also how abortion relates to other values that are important to Ann. To broaden his understanding in this way, Bert will need to deliberate, not negotiate, with Ann. 17 Secondly, the process of evaluating stakes is not only deliberative in the sense of clarifying emerging questions, but also in the sense of discussing the other's evaluation of one's own stakes. For example, Bert might initially suggest that Ann has only slightly higher stakes, an assessment that Ann would disagree with. A deliberative process allows Ann to express her disagreement with Bert's assessment and to explain why, in her view, she not only has slightly higher, but much higher stakes. This process of explaining and justifying one's views is fundamentally deliberative.
Thirdly, the deliberative process of exchanging and justifying reasons might not only enable Bert to evaluate what is at stake for Ann, but it might also enable him to develop a clearer picture of what is at stake for himself. 18 He might, for example, realize that he has overestimated the importance that the issue of abortion plays in his own life. Alternatively, deliberating with Ann and learning about her feminist perspective, Bert might realize that a pro-life view does not fit with other values that he endorses, such as the empowerment of women. So, even though Bert continues to believe in the foetus's right to life, he might hold that belief less strongly, given that it contradicts other values that are important to him.

Determining concessions
Let us assume that Bert and Ann have agreed, through deliberation, that Ann has higher stakes in their disagreement on abortion than Bert. Let us further assume that they have also agreed that Ann has much higher stakes involved, not only slightly higher stakes. But this is still a vague agreement and it is not at all clear how this agreement will translate into fair concessions for both. A distinct stage in the process of compromising therefore consists in determining fair concessions that are proportional to the stakes that are involved for each party. Importantly, the process of determining fair concessions also needs to be deliberative, because Ann and Bert have to exchange their reasons why their respective stakes should translate into the concessions that they propose.
Let us further assume that Bert agrees to concede to Ann's primary claim that abortion should be legal (L).
He does so because he acknowledges that Ann has much higher stakes in the disagreement on abortion than he does. 19 Despite Bert's acknowledgement that Ann's stakes are much higher than his own, this is still a major concession on Bert's part -after all, he continues to believe in the foetus's right to life. Bert might therefore argue that since he concedes to Ann's primary claim (L), Ann should concede more in terms of her minor claims, such as, for example, the specific conditions under which abortion should be legal. Bert might propose, for example, that abortion should not be legal after the first trimester, or that there should be a certain amount of mandatory consulting sessions, etc. -all of which, let us assume, Ann opposes.
Ann, on her part, believes that because she has much higher stakes in the issue than Bert, her view on the conditions for abortion should also have more weight than Bert's. In this hypothetical scenario, the need for further deliberation arises, because it needs to be clarified whether Ann's claims regarding the specifics of abortion regulation should indeed also weigh more heavily than Bert's claims in this regard. After all, even though both have agreed that Ann has much higher stakes in the disagreement than Bert, both have also agreed that Bert has at least some stakes involved that need to be accommodated if the compromise is to be fair. Determining fair concessions therefore continues to be a deliberative process.
The question then is: Can negotiation structurally enable the determination of fair concessions? The answer, I submit, is once again 'no'. First of all, it seems rather unlikely, psychologically speaking, that Ann and Bert would even want to negotiate at this stage. That is, after having achieved the mutual agreement that Ann has much higher stakes involved in their disagreement, it seems very unlikely that Bert would suddenly switch into 'winning mode' and try to get away with as few concessions as he can. But, for the sake of argument, let us assume that at the stage of determining mutual concessions, Ann and Bert indeed start negotiating, with the goal of enforcing their claims as best as they can, regardless of fairness concerns.
That is, Bert will negotiate for (NL) and Ann will negotiate for (L). This not only means that the preceding deliberative process becomes meaningless (since, as I have argued, second-level understanding and stake evaluation do not play a role in negotiation processes); but it also means that the outcome of Ann and Bert's negotiation will reflect factors that are independent of fairness concerns, especially factors that are related to power.
A primary source of power in negotiations that is widely recognized is BATNA, i.e. a negotiator's 'best alternative to a negotiated agreement' (Thompson, Wang & Gunia, 2010: 494). Research has shown that negotiators with attractive alternatives are significantly more assertive in negotiations than negotiators with less attractive BATNA. The influence of power on negotiation outcomes is particularly detrimental for the purpose of achieving a fair compromise (in the sense discussed in this paper), because a fair compromise requires a distribution of concessions according to stakes, not according to power.
Indeed, the fact that negotiation outcomes tend to be influenced by BATNA-based power differences puts negotiation in an inverse relation to fairness, because negotiators with attractive BATNA have, by definition, also less at stake in a disagreement. Put differently, if you have attractive alternative options to a specific agreement, you do not have high stakes involved, because high stakes imply that you rely on that agreement to work. The influence of BATNA-based power differences on negotiation outcomes therefore means that precisely those negotiators with the lowest stakes will have the most weight in determining the outcome, which is diametrically opposed to what is required for a fair compromise -namely that those with the lowest stakes should have the least weight in determining the outcome.
Gender has been found to be another crucial factor in determining power differences in negotiation, with a negative effect for female negotiators. Female negotiators are at a disadvantage especially if gender stereotypes operate at an implicit level. As Thompson, Wang and Gunia state in this context, '[g]ender salience might thus operate like a low-power state, preventing women from acting assertively. In [a] study, women did, in fact, get worse outcomes than did men in mixed-gender negotiations, when an implicit gender stereotype was subtly activated' (Thompson, Wang & Gunia, 2010: 496). 20 In addition to BATNA-and gender-based power inequalities, we can easily think of other fairness-averse factors or skills that can determine negotiation outcomes, such as rhetorical skills, experience in negotiating or the ability to keep one's nerve in stressful situations.
Therefore, unlike deliberation, negotiation processes advantage whoever has more power in terms of BATNA or gender, or whoever has more of the skills that make you a successful negotiator generally speaking. To be sure, by chance, negotiation might result in a fair outcome -but do we want to rely on chance when our moral values are at stake? 21

Possibilities of failure
Finding a fair compromise on issues of moral or political disagreement is a delicate process, even more so since each stage of the process is vulnerable to failure. At the first stage, the participants can fail to adequately convey their reasons to each other -mutual understanding might simply not emerge. If that is so, finding a fair compromise is doomed to failure because the following two stages require the first stage of mutual understanding to be successful. But even if mutual understanding is successfully established, compromise can still fail at the next stage, in that the participants might not be able to agree who has higher stakes involved. They might understand each other's reasons, but they disagree on what these reasons mean in terms of what is at stake for each other. Failure at the second stage would, again, ruin the chances of achieving a fair compromise, because the next stage of determining fair concessions requires an agreement on stakes. And even if the first two stages are successful, achieving a fair compromise can still fail at the third stage, because the disagreeing parties might have incompatible views on how their respective stakes should translate into concessions.
In this sense, deliberation is a necessary, but not a sufficient condition for achieving a fair compromise. Much depends on contextual circumstances, such as the participants' mindsets, including their willingness to understand each other and to find a fair agreement. But, and this is the crucial claim of this paper, without deliberation, finding a fair compromise is doomed to failure, because only deliberation provides the structure that enables the participants to (at least potentially) understand each other's reasons, evaluate each other's stakes and determine concessions that are proportional to the stakes involved.

IV. How Important is it that a Compromise is Fair?
Given the multiple possibilities of failing to achieve a fair compromise, the question arises: How important is it that a compromise is fair? If the participants to a process of compromising fail at any of the three stages, what is to be done? After all, a compromise, even though in that situation an unfair one, is still on the table. For example, in case the parties to a disagreement fail to agree on a fair compromise deliberatively, they might alternatively start to bargain or they might simply decide to split the difference. However, as I have argued in this paper, neither option is likely to result in a fair outcome. The question then is: Should the disagreeing parties still compromise, even if the outcome is not fair? Put differently, how important is it that a compromise is fair? While a satisfactory answer to this question is beyond the scope of this paper, it is worthwhile to outline a few thoughts on this matter.
To start with, in some cases of disagreement, fairness might not be the most important aspect of achieving a compromise. Whether and to what degree fairness matters depends to a significant extent on the context of a disagreement. For example, if a buyer and a seller on a farmers' market disagree on the price of a pound of apples, a fair outcome of their negotiation, even though desirable, is likely not of paramount importance. 22 Indeed, it seems somewhat inadequate that fairness should be a major concern for both parties, given that they would have to engage in a deliberative process, aiming at second-level understanding of each party's reasons of why a pound of apples should cost a little more or less, and, based on that second-level understanding, they would have to evaluate what is at stake for each other etc. -which seems excessive for the purpose of finding a compromise on the price of a pound of apples.
But I submit that fairness should be of paramount importance for compromises that affect sociopolitical cooperation, such as compromises on moral disagreements that occur in civil society or between legislators. 23 For one thing, a fair compromise in these situations can be a manifestation of mutual respect, which makes (fair) compromise an indispensable feature of democracy (Bellamy, 2012;Weinstock, 2013). And for another thing, compromises at the socio-political level are often a means to avoid undesirable consequences, such as social strife or stagnation in passing legislation. In both cases, the affected parties will want to make sure that the compromise lasts. This, in turn, is much more likely if both sides consider their compromise to be fair, whereas an unfair compromise is not likely to be stable over time.
Indeed, seen from a long-term perspective, an unfair compromise can be worse than no compromise at all, in case the parties to that compromise become aware of the unfairness of their agreement after the compromise has been made. In that case, the feeling of deception can exacerbate the initial conflict -for example, if the deceived party feels justified in their initial judgement that the other party is morally unreliable. The other party is then likely considered to be unreliable not merely with regard to their judgement, but also as a matter of character. The negative personal perception that is in this case added to the initial disagreement will likely make it harder to resolve related disagreements in the future. 24 In conclusion, then, we might say that if a compromise is desired in terms of socio-political cooperation or stability, a fair compromise should be the goal. And, as I have argued in this paper, to achieve a fair compromise, we need to deliberate rather than negotiate.

Notes
1 The term ' compromise' can be understood to designate both a process and an outcome. In this regard, Daniel Weinstock distinguishes between ' compromise', which refers to compromise as a process, and ' a compromise', which designates compromise as an outcome (Weinstock, 2013: 554-555). Furthermore, Chiara Lepora introduces a third possible meaning of compromise -in addition to compromise understood as 'the act of agreeing' and 'the content of the agreement', Lepora points out that compromise can also designate 'the actions pursuant to the agreement' (Lepora, 2012: 1). For the sake of conceptual clarity, I use the term ' compromising' to designate the agreementseeking process and I use the term ' compromise' or ' a compromise' to designate the outcome of that process. 2 See Mansbridge et al., 2010. 3 The 'fact of reasonable pluralism' refers to the idea that modern democratic societies are inherently characterized by a variety of religious, philosophical and moral doctrines that can be equally reasonable and yet irreconcilable (Rawls, 2001). 4 For more detailed discussions of the reasons for including compromise in deliberative democracy, see Bellamy, 1999, chapters 4 and 5;Bellamy 2012;Gutmann & Thompson, 2004, chapter 1;Gutmann & Thompson, 2012, chapter 1;Mansbridge et al., 2010;Warren & Mansbridge, 2016;Weinstock, 2006. 5 Both terms, negotiation and bargaining, are used synonymously in this paper. 6 Similarly, May points out that 'moral correction involves the recognition that one's earlier commitments were mistaken. But if the agent comes to accept an alternative as a moral compromise, she still views her initial position as morally superior' (May, 2005: 318). 7 Brighouse and Fleurbaey also propose two other justifications for the proportionality principle. They argue that the proportionality principle increases individual autonomy and that it leads to the best social outcomes from a consequentialist perspective (see Brighouse & Fleurbaey, 2010: 142). Here, I focus on their argument from respect. 8 Note that I merely claim that deliberation is necessary, not that it is sufficient for achieving a fair compromise.
There are other factors, such as the mindsets of the persons involved, that can still prevent a fair compromise, even though the required deliberative structures are in place. 9 The word 'structure' refers here to the way in which an agreement-seeking process is set up in terms of how the participants are supposed to interact with each other. It is, therefore, the procedural structure of an agreement-seeking process that enables or inhibits the participants to interact in a certain way. The structure of deliberation, for example, provides for a mutual exchange of reasons while the structure of negotiation provides for a mutual exchange of offers and counteroffers. 10 For the opposite claim that compromising and deliberating are analytically distinct activities, see Jones & O'Flynn, 2016. The authors argue that deliberation can 'facilitate' a fair compromise in the sense that compromising can set in after deliberation, while deliberation is 'not part of the process of making a compromise' (Jones & O'Flynn, 2016: 17). 11 Note that I use the term ' compromising' or 'process of compromising' as an umbrella term for the activity that spans all three stages. 12 The phrase 'mixed-motive nature of negotiation' refers to the idea that the participants to a negotiation process are usually characterized by two conflicting motives, cooperation and competition. See Thompson, Wang & Gunia, 2010: 499. 13 'PPP' is short for Positive-sum games, Pareto-optimality and Principled bargaining. See Luban, 1985: 399. 14 Ideally, if a fair compromise is agreed upon by citizens in a public setting, their compromise will affect macropolitics, e.g. by influencing political agenda setting or even legislation. An increasingly discussed venue for deliberative processes with macropolitical 'uptake' is called minipublics. Minipublics can have macropolitical uptake in different ways, even though macropolitical influence is not guaranteed (see Goodin & Dryzek, 2006). 15 For the sake of space, in the following I only refer to the case of Bert understanding Ann, but my argument applies in the same way to Ann understanding Bert. 16 I should emphasize that the disagreement between Ann and Bert is meant to be purely hypothetical. For the purpose of my argument, it does not actually matter whether Ann's stakes are indeed higher than Bert's. The purpose here is only to show that deliberation is necessary for evaluating mutual stakes; the purpose is not to develop an actual assessment of the stakes that can be involved in a disagreement on abortion. This means that 'real-life Ann and Bert' could well agree on an evaluation of stakes that is contrary to the one that I propose in this paper. Whatever the outcome of deliberation on abortion can be, the crucial point is that deliberation, not negotiation, is needed for evaluating stakes. 17 It is worth pointing out that even though Bert will have to engage deliberatively with Ann in order to evaluate what is at stake for her, a large proportion of the evaluative process occurs in Bert's own head. This does not make the process of evaluating stakes any less deliberative. Indeed, as Robert Goodin argues, interpersonal deliberation has an indispensable internal dimension. Goodin points out that 'it remains significant how very much of the work of deliberation, even in external-collective settings, must inevitably be done within each individual's head' (Goodin, 2000: 81). For an in-depth discussion of the intrapersonal aspect of compromising specifically, see Lepora, 2012. 18 Similarly, Weinstock points out that ' deliberation geared toward compromise engages participants in a process through which they gain greater selfclarity as to their own conceptions of the good' (Weinstock, 2017: 651). 19 To emphasize this again, the point here is not to argue that Bert should (or would) necessarily react to Ann's arguments in this way. Rather, the point here is to show that deliberation is necessary for enabling Bert to react to Ann's arguments in a meaningful way, i.e. in a way that is based on second-level understanding. Whether the deliberative structures in place will actually lead Ann and Bert to deliberate in this way depends on other factors as well. Indeed, as I elaborate in the next section, in some situations of disagreement, it might well be possible that 'real-life Bert and Ann' are not able to agree on a fair compromise at all. 20 The authors suggest that gender-based power inequality can be reduced by explicitly counteracting gender stereotypes, in the sense of saying: 'Well, unassertive behavior and accommodation may be the cultural stereotype of women, but it is surely not me!' (Thompson, Wang & Gunia, 2010: 496). However, it seems to me that this strategy of explicitly counteracting gender stereotypes cannot reliably avoid the problem of gender-based power inequality in negotiations, because female negotiators actually have to explicitly activate and oppose gender stereotypes, which is certainly not something all women can do or want to do. 21 In this context, Jones and O'Flynn (2013) describe the example of two parties with equal bargaining strength that -due to their equality in bargaining strength -agree on a compromise that consists in equal concessions. If splitting the difference equally is actually what a fair outcome consists in, these parties have indeed arrived at a fair compromise; but they did so, and this is the crucial point, without having aimed at a fair outcome. The fairness of a compromise that is based on bargaining or negotiation is therefore only a matter of chance (see Jones & O'Flynn, 2013: 121). 22 In those cases where fairness is not of importance, negotiation is an appropriate way to find a compromise. As I have emphasized previously, negotiation is only inappropriate if we aim for a fair compromise. 23 For the opposite claim that fair compromises are never a moral desideratum (and that we sometimes even have the moral obligation to seek an unfair compromise), see Wendt, 2018. 24 See Kennedy and Pronin (2008) for the argument that the perception of one's opponent as biased is a crucial reason why disagreements escalate into full blown conflicts rather than be resolved.