Compromises and Fairness

 

Juha Räikkä

 

University of Turku, Department of Philosophy, jraikka@utu.fi

Many philosophers have pointed out that a compromise that is fair in one sense can be unfair in another. In this paper, I will briefly introduce different ways in which compromises can be “fair” and then analyze them. In particular, I compare the importance of what I call (a) split-the-difference fairness and (b) end-state fairness. I will defend split-the-difference fairness against an important objection—that a person’s false belief about her fair share does not change what her fair share actually is. To allow such a notion would be rather like allowing her false belief to alter what is actually true. Nevertheless, I aim to provide support for the idea that sometimes it may make sense to consider people’s actual beliefs, even if they are false. However, I do not take a strong position on the question of what the most important sense of fairness is with regard to compromises.

 

Keywords: Compromise, fairness, distribution, agreement, false beliefs

 

Introduction

Compromises tend to be unfair. All compromises involve concessions, and usually, some party or parties have to make immoderate concessions to achieve at least something. Powerful parties, in turn, use their power, but compromises are voluntary. They create obligations for all parties. Compromises have a normative force that does not depend on fairness (Jones & O’Flynn 2013: 125). Their fairness or unfairness is a side issue when their moral relevance is considered. Like promises, compromises are morally binding agreements, assuming that what is agreed upon is not based on overt coercion or blatant deception. Compromises can be (and often are) disturbingly unfair, but they are still clearly legitimate and binding.

 

That said, however, it is clear that fairness is a morally desirable feature of compromises. A person who can choose whether to make a fair or an unfair compromise has a prima facie obligation to make a fair compromise. If this is so, why are compromises so often unfair, and yet many of us find this fact unsurprising? There are many explanations. First, people do not always do what they should do. Possibly, they even have the right to do wrong, as their personal autonomy confers the freedom to choose between right and wrong (Herstein 2012). Second, people have the right to make contracts, and possibly, this right in many cases overrides the obligation to aim for only fair compromises. Third, fairness is just one feature of many that make compromises good. It is conceivable, for example, that a compromise that secures social stability can be better, morally speaking, than a fair compromise. These and similar issues explain why unfair compromises are common, despite fairness clearly being a morally important factor. Other things being equal, we should further the fairness of compromises.

 

One problem that arises is that “fair compromise” can mean many different things, and a compromise that is fair in one respect may not be fair in another. As many philosophers have pointed out, a compromise that is fair in one sense can be unfair in another (Luban 1979: 403; Wendt 2018: 2859). Below, I briefly introduce various ways in which compromises can be “fair” and analyze them. In particular, I will compare the importance of so-called (a) split-the-difference fairness and (b) end-state fairness. I will defend split-the-difference fairness against an important objection, but my aim is not to take a strong position in determining the most important sense of fairness to consider with regard to compromises. Surely, all meanings of fairness have their own roles to play.

 

The paper proceeds as follows. I will first define how the notion of “compromise” is used in this paper and introduce four ways in which compromises can be fair. Next, I present an example that illustrates the difficulty in deciding the sense in which a particular compromise should be fair. Then I evaluate the example and defend split-the-difference fairness against an interesting and tempting objection. Finally, I draw conclusions and assess the meaning of the debate.

 

The Nature of Compromise

Our everyday understanding of compromise is relatively clear, but philosophers disagree about how to understand and define the notion of compromise. This chapter describes how the term “compromise” is used in this paper.

 

A compromise involves two or more parties. When the parties settle disputes through mutual concessions, they compromise. If only one party makes concessions, it is no longer a compromise (Day 1989; Archard 2012). Often, compromising parties are in asymmetrical positions, which means that one side is powerful, while the other is weak (Golding 1979: 15). However, this does not prevent compromises.¹ The motive behind a compromise is that by agreeing to compromise, both sides can gain something. A compromise ensures that each party gets at least something they value, and making a compromise is often preferable to continuing a disagreement. Having at least one solution is better than having no solution at all. The pragmatic justification for compromises is that it is often important to get something done (Menkel-Meadow 2016: 4). We can say that by compromising, the parties achieve their second-best options when the disagreeing party makes the best option unfeasible (Wendt 2019: 2856; O’Flynn & Setälä 2020: 9).

 

Compromises are proposed when the original situation involves (1) disagreement and (2) a motive to come to an agreement. To see this, let us consider an example developed by Chiara Lepora (2012: 8). The example concerns patients’ right to know:

 

Imagine two doctors caring for a terminal patient who is unaware of her diagnosis.  One doctor, worried about the patient’s frail psychological state, resolves to avoid broaching the subject of end-of-life decisions with her. The other doctor, convinced of the patient’s right to know, wants to start discussing end-of-life decisions with her as soon as possible. These are matters of principle from each doctor’s perspective: the first takes it as a matter of principle that physicians should not needlessly jeopardise the well-being of their patients; the second takes it as a matter of principle that physicians should be completely open and honest with their patients. Suppose finally that the two physicians share responsibility for the patient’s care, and must come to a joint decision whether to tell the patient or not. It is clearly impossible for both doctors to stand by their respective principles in this case, but conflict forces them to search for a compromise. So, for example, each might backpedal on his position in deference to the other, and they finally agree to talk to the patient’s family only. That is not what either doctor’s principle requires – the one doctor is not being fully honest with his patient in that way, while the other doctor might still be worried that the patient will be stressed if family members raise awkward questions with her. The two doctors have simply agreed to do something different than what either of their principles requires. (Lepora 2012, 8)

 

The doctors disagree on whether they should reveal the patient’s condition to her, but they agree that they must come to a joint decision; otherwise, they are unable to provide care. This agreement motivates them to seek a solution to their disagreement and find a compromise. When they have found it, their disagreement has not disappeared, but they have found a pragmatic way to proceed.²

 

Many authors have emphasized that compromises must be voluntary – otherwise, they are not compromises (Jones & O’Flynn 2020: 119). The voluntariness of a compromise requires that the overt act of compromising, such as handshaking, verbal statements, or a signature, is unforced. However, it also requires that the parties understand what they are sacrificing and what they stand to gain from their concessions. Notice that the decision to agree to compromise in a certain way can happen much earlier than the actual act of crafting a compromise. This personal deliberation should also be voluntary and not based on manipulation or disrespectful deception, for instance. Of course, compromises can and often do involve elements of dishonesty. This seems acceptable, as parties need not reveal their cards during the negotiations (Kuflik 1979: 47).

 

The importance of compromises being voluntary is simply because only then do they carry obligatory force. To be a party in a compromise is to have freely accepted a moral commitment (Jones & O’Flynn 2020: 119). Compromises can be compared to promises. A person may promise something because it sounds good, but if it ceases to sound good, this does not imply that the person should not keep her promise. The reason to keep a promise is not reducible to the reason for making it (Raz 1975). If A wants to have a cup of coffee with B, A has a reason to meet B and may promise to meet B. Suppose that A promises to meet B. In this case, even if A would change her mind and not want to have a cup of coffee with B, she would still have a reason to meet B because of her promise. The same holds for compromises. They have obligatory force, even when a person is no longer willing to agree to compromise. An involuntary promise does not commit the promisor to keep the promise, and the same holds for compromises (cf. Owens 2006: 118).

 

It is sometimes claimed that a special feature of compromises is that, unlike in other negotiations, the parties do not try to get as much as they can (Benditt 1979: 27). However, perhaps it is better to say that it is possible that the parties in a compromise are not trying to get as much as they can. They may agree only because they think that although the opponent’s view is mistaken, the view makes some sense, and therefore compromising in a suboptimal way has merits. Thus, the parties may make “principled” compromises instead of purely strategic (or “goal-oriented”) compromises (Jones & O’Flynn 2013: 120–121). Note also that one may accept a suboptimal compromise because there are advantages to be gained from conceding, such as the satisfaction of coming to an agreement or gaining respect from others. It is not always a case of people trying to get all they can, even when thinking strategically.

 

In ordinary language, the results of all kinds of bargaining processes can be called “compromises,” given that the processes lead to solutions in which all parties gain something, even if less than they hoped for (Braybrooke 1982, 141; Jones & O’Flynn 2013: 120). Plausibly, it makes sense to distinguish between “compromises” and “deals” when the discussion concerns a compromise about “serious” matters that are very important to the parties (Lepora, 2012; Lepora & Goodin 2013). However, in this paper, I call “bargains” and “deals” compromises if they involve mutual concessions, are voluntary, and do not resolve the parties’ disagreement completely (cf. Wendt 2019: 2858).

 

Four Kinds of Fairness

Let us now turn to the concept of fairness. As many authors have pointed out, compromises can be fair and unfair in many ways (Luban 1979: 403; Kappel 2018: 79; Wendt 2019). Let us distinguish four senses of fairness: (1) bargaining-position fairness, (2) procedural fairness, (3) end-state fairness, and (4) split-the-difference fairness.

(1)  Bargaining-position fairness. The initial bargaining position can be fair or unfair (Wendt 2019: 2859). The position seems unfair when parties are clearly asymmetrical, that is, when one party is powerful and has greater threat potential, while the other party is weak and its threat potential is small. The powerful side can make offers that look like demands, and the weak party cannot refuse an offer when the value of having an agreement is great enough (Golding 1979: 15). The relative strengths (such as financial resources) of the parties obviously influence the outcome of the negotiations (Carens 1979: 131; Luban 1985: 403). Usually, the more powerful party achieves a better deal, although it is possible that the stronger party does not want a result that oppresses the weaker party. In principle, the powerful side may prefer to advance equality and justice. However, this does not change the fact that the initial bargaining situation is unfair when one side has too prominent a say regarding the content of the compromise. Usually, not much can be done to balance bargaining-position unfairness.

(2)  Procedural fairness. The negotiation process that ends in a compromise can be fair or unfair (Ceva 2016). A number of requirements should govern negotiations. Often, it is difficult to say whether the rules derive from procedural fairness or from more general ethical considerations in making a compromise.³ Usual norms include truthfulness, honesty, rich information sharing, openness, transparency, and willingness to understand the other side (Hall 2022: 226). The parties are supposed to respect each other and to be reflective in their discussions (Huxtable 2014: 476). The norms may be somewhat demanding, and one wonders whether compromise negotiations can ever completely meet them. More realistic norms are those that set limits to threats and determine the types of information that the parties can legitimately keep to themselves and those they should reveal (Jones & O’Flynn 2013: 123). Possibly, fair procedures also include requirements having to do with equal awareness of the issues involved and equal access to creative problem-solving techniques. When people say that a compromise is fair, they sometimes mean simply that it is the result of sincere and open discussions.

(3)  End-State Fairness. When people make compromises, they distribute things (however abstract the things may be). It is natural to ask whether the distributive arrangements made are fair (Jones & O’Flynn 2013: 115). If they are, we can say that the compromise is fair in the sense of end-state fairness. If the initial bargaining position happens to be fair and the procedures are fair, this raises the probability that the final distribution will be fair, too. However, bargaining-position fairness and procedural fairness do not guarantee end-state fairness and, on the other hand, bargaining-position unfairness and procedural unfairness do not necessarily prevent end-state fairness (Wendt 2019: 2859). The idea that compromises include distribution fits better with compromises on conflicts of interests (e.g., what is the value of this used car?) than to compromises of conflicts of principles (e.g., should we legalize euthanasia?). However, the line between conflicts of interest and conflicts of principle is often unclear (Carens 1979: 136). If parties disagree on the appropriate criterion for distribution of certain goods, they seem to have a conflict of principle, but surely they both have an interest in the goods. They disagree on how the goods should be divided between them.

(4)  Split-the-difference fairness. The idea of split-the-difference fairness is that a compromise between two sides is fair when the object of dispute is split down the middle. At the heart of “fair compromise” is giving an equal share to all parties (Braybrooke 1982: 149; Day 1989, 472; Huxtable 2014; Menkel-Meadow 2016: 5). Nevertheless, it may be difficult to determine when parties have received an “equal” share (Luban 1985: 403). Perhaps we can say that all parties deserve equal gains and should make equal sacrifices, but telling whether gains and sacrifices are “equal” may be complicated. Saying that a compromise is fair when all parties make equal concessions relative to how much they would lose without an agreement (Wendt 2019) requires determining what kinds of concessions are implied. An interesting detail is that when split-the-difference fairness is defined as a fifty–fifty position between the parties’ demands, the point of fairness becomes relative to the parties’ claims. When you split the difference with someone, you agree on a deal that is halfway between two proposals. However, overall, splitting the difference provides an independent and intuitively plausible way to see the fairness of compromises.

All four ways of understanding the fairness of compromises are important. However, evaluating their merits is a difficult task, and the task is remarkably challenging when we compare end-state fairness and split-the-difference fairness. To understand this better, let us consider a concrete example.

 

The Dilemma of Champagne Bottles

Suppose that two people are walking together on a nature trail. Suddenly, they notice a wooden box on the nearby shore. It appears to have been afloat for a relatively long time. They open the box and, to their surprise, find eight champagne bottles. Although the bottles are not new, they are not terribly old, either. There is no identification or any reason to turn them over to the authorities. They decide to keep the bottles.

 

Problems start when the older person, who is the team leader, says that she thinks she should keep all the bottles. The younger team member does not believe her ears and replies immediately to her boss that a fair distribution would be that they each get four bottles. They start to argue. The team member says that they found the bottles together, and therefore should get four bottles each. The team leader claims that a fair distribution would be that she alone gets the bottles because she is an expert on champagnes and wines. Her fondest hobby is champagne and wine tasting. In her view, giving four bottles to the team member would be almost a crime, as she cannot appreciate the nature and spirit of champagnes. The team leader promises to buy a beer or two for the team member, but the team member does not accept such a disparaging solution.

 

The team leader, because she’s the boss, could just keep the bottles. However, she wants the team member’s acceptance, as not having it could mean that the member would leave her job. This gives the team leader an incentive to prefer making a compromise. The team member is not ready to accept the distribution, which is in her view clearly unfair. On the other hand, she understands that her cards are not particularly strong. She does not want to risk her job, and this gives her an incentive to make a compromise.

 

A third person who happens to be there hears the discussion and approaches. She understands that the demands presented are sincere. The boss does not just say that she deserves all eight bottles. She really thinks that such a distribution would be best, as it would cause the most pleasure overall. Similarly, the team member is not just pretending to support the four-four distribution. That is, she truly does not agree with her boss. She believes that the four-four split would be fair.

 

The team leader and the team member ask the third party for help. They ask her to suggest a fair compromise. They do not ask that she be neutral or impartial, or any of the usual qualities required of mediators. They ask only that the third party suggest a fair compromise.

 

Luckily, the third party accepts the invitation. She first observes that an eight–zero distribution is out of the question, as it requires no concession from the boss and is therefore not a compromise at all. The same holds for the four-four distribution. It would not be a compromise either, as the team member would not concede anything. She would get the number of bottles she wanted. The third party realizes (correctly, let us assume) that the objectively right distribution would be to give four bottles to both. However, since this option is not available, from the standpoint of end-state fairness, the third party thinks it might make sense to give five bottles to the boss and three bottles to the team member. That solution would be closest to the objectively fair four-four solution. However, she then notes that from the standpoint of split-the-difference fairness, a six-two distribution would give each party two bottles less than they demanded. This solution would be considered wrong by both the team leader and the team member, but it would require equal sacrifices from both, given their opinions.⁴

 

The third party suggests that the boss will get six bottles, and the team member will get two bottles. In her view, that would be a fair compromise given the situation.⁵ The boss and the team member shake their heads but agree to the suggested compromise.

A Moderate Defense of the Solution

Did the third party make the right suggestion? We may think that she should not have suggested a compromise at all since it was clear that the four-four distribution suggested by the team member was correct. They found the bottles together; therefore, they should divide them equally. But suggesting a four-four solution to the parties would have implied that the boss would use her stronger status and just take all the bottles. So, it was good that the third party suggested a compromise. However, it is not clear whether the suggestion was correct. In this chapter, I propose a defense of the suggestion.

 

In the champagne bottle dilemma, the team leader’s erroneous understanding of the distributive criterion influenced the result of the compromise, as the third party relied on split-the-difference fairness. In their interesting paper, “Can a Compromise Be Fair?” Peter Jones and Ian O’Flynn (2013) argue against the idea that a person’s wrong beliefs can be relevant to compromise arrangements. They write as follows:

 

Can the fact that A and B are in disagreement alter what is substantively fair and make their compromise determinative of what is substantively fair? Suppose, for example, that A and B are in conflict over how a good should be divided between them. Each believes he should receive 75 percent and the other 25 percent. Looked at objectively, the fair outcome is that A should receive 75 percent and B should receive 25 percent. That objective assessment might be based, for example, on our assessment of what each has contributed to producing the good and our discovery that A has contributed three times as much as B. Can the fact that B believes he should receive 75 percent and enters the compromising process demanding 75 percent alter the standing of A’s and B’s claims, so that dividing the good equally between them becomes the substantively fair outcome? Perhaps, alternatively, rather than altering the fair outcome, the way in which A and B disagree creates a different sort of fair outcome that is somehow additional to, rather than a substitute for, the objectively fair outcome. It is very hard to see why we should accept either of these possibilities. Allowing that B’s false belief about his fair share might alter what his fair share actually is would be rather like allowing that his false belief might alter what is actually true. (Jones & O’Flynn 2013: 125-126)

 

This is a good argument. There is something deeply disturbing in letting a party’s false views influence compromise arrangements. However, something can be said in defense of the split-the-difference fairness. The following four points are important:

1.   The team leader’s claim that the ability to enjoy is relevant to the distribution of bottles is not an insane opinion. A hardcore utilitarian could easily support such an opinion. It is important to note that although there are cases in which it would be awkward to allow false beliefs to influence distribution, the dilemma of champagne bottles is not necessarily one of them. The team leader’s opinion is surprising, perhaps, but it is not one that we should set aside as completely foolish. Although the right criterion for distribution is not an “ability to enjoy,” such ability may still be somehow relevant. The team leader certainly has a point. “Fittingness” is important. Possibly, this is something we should keep in mind when we decide whether we can rely on end-state fairness (that suggests a five-three distribution) or split-the-difference fairness (which implies a six–two distribution). The claim that we should not allow a person’s false belief about her fair share change what her fair share actually is blurs the fact that not all false beliefs are totally unreasonable.

2.   The team leader thinks that it would be morally wrong and plainly unfair to distribute the bottles equally, that is, that both finders take four bottles. To ask her to agree to a four-four distribution is to ask her to act against her conscience. If we want to respect people who act in accordance with their conscience, then we should respect the boss who proposes a distribution that is fair in her view (the eight-zero distribution). One way to show respect is to rely on split-the-difference fairness. To rely on end-state fairness (with a five-three distribution) is to tell her, at least tacitly, that she should forget her conscience and do what is in fact morally best. Of course, in some cases, it does not make sense to respect persons who act as their conscience dictates. After all, a person’s conscience can direct her to do morally terrible things (Zimmerman 2008: 14). However, this is not the case with the team leader. She just does not want to waste valuable champagne. She wants to produce the most pleasure overall. Possibly, we should respect her.

3.   According to the formal principle of justice, similar cases should be treated similarly and different cases differently. If the third party considers the team leader and the team member to be similar cases, she is on the right track. The two parties are similar in many relevant respects. They are both interested in sharing the bottles fairly. They both hate the idea of the distribution not being correct. They each think that their own opinion on appropriate distribution is the correct one. In this situation, it is natural for a third party to decide to treat the cases similarly, as this is what justice requires. Treating the parties similarly means that they both must make equal sacrifices, which leads the third party to suggest that they each receive two bottles less than they think they deserve. Of course, someone may say that the cases are clearly different, as the team leader has the wrong opinion and the team member has the right opinion about the distribution criterion. However, the third party does not need to determine whose opinion is correct—although she agrees with the team member. If the third party is clever, she understands that she may be mistaken.

4.   When compromises are made, both parties think that their opponent is wrong, is making undue demands, has false beliefs, and so on. Compromises are not needed when parties agree on what is true, reasonable, and fair. When someone seeks a fair compromise, her starting point should not be that the compromise arrangements should somehow reflect who is actually right. Both parties think that they are actually right. Split-the-difference fairness applies justice based on the idea that the parties’ demands should be treated equally. Of course, this solution easily leads to a situation that is unfair in the sense of end-state fairness, but if we are interested in the fairness of compromises rather than the fairness of the end-state arrangements, we must consider the parties’ claims, however unrealistic they may seem. Often, both sides have mistaken views and undue requirements. Compromises must then be made between these claims if the compromises are to be fair. In the champagne bottle dilemma, the third party makes a proposal that aims to be fair, given the demands presented by the parties. She does not take God’s point of view but listens to what is going on in the debate between the parties. She is interested in the fairness of compromise.

These four points do not show that Jones and O’Flynn’s (2013) objection is mistaken. However, the points above provide support for the idea that sometimes it may make sense to consider people’s actual beliefs, even if they are false.⁶ In some cases, this may be how we promote the fairness of compromises.
 

Concluding Remarks

As have many other philosophers, I have argued that there are multiple ways to understand the fairness of compromises. I distinguished between (1) bargaining-position fairness, (2) procedural fairness, (3) end-state fairness, and (4) split-the-difference fairness. Obviously, all definitions of fairness have their own roles to play. However, I compared end-state fairness and split-the-difference fairness and aimed to show that, at least in some cases, it is difficult to choose between them when they point in different directions. I defended split-the-difference fairness against an objection and attempted to provide support for the idea that sometimes it may make sense to consider people’s actual beliefs, even if they are false. The fairness of a compromise may require this consideration.

 

The argument I presented has some important and concrete implications. Here is an example: Suppose that country A claims that country B poses a threat but the claim is mistaken. In reality, B does not threaten A. They disagree on what kinds of weapons and allies B can have. Suppose also that both countries have a strong incentive to make a compromise and arrangements that secure peace. Should we ignore the fact that A considers B a threat because we know that the judgment is (understandable and reasonable but) false? This is unlikely. Surely, we should pay proper attention to A’s viewpoint, whether precisely true or not. This is the way to make a fair compromise, which will very likely increase stability as well. Setting aside A’s opinion could lead to war.

 

It would be easy to give many similar examples. The assumption that we are right should not blind us when we aim to make fair compromises..

 

Acknowledgements

I would like to thank Olli Koistinen, Hemmo Laiho, Sanni Ojala, and Helena Siipi for helpful discussions. I presented this paper at the University of Oxford (St. Anne’s College) in April 2024. I would like to thank the audience, in particular professor Roger Crisp, for excellent points. Finally, I’m grateful to the editors and reviewers of Etikk i praksis. Their comments helped me to improve the manuscript considerably.

Notes

1 When a person makes a compromise with someone else, she may be criticized from a variety of perspectives. One such perspective is moral. Sometimes, moral criticism comes from one’s opponent, who can claim that ethical rules that should govern the negotiations are being violated, or that the compromiser does not have a morally appropriate problem-solving attitude. On other occasions, the critics are those represented by the compromising person. They may argue that their representative has made morally unacceptable sacrifices, or that the results of the compromise are unpredictable in a morally problematic way. Rather often, outsiders blame a party or parties for the compromise. They may say that, because of the compromise, a party lost her moral integrity or that the parties agreed to a clearly unjust compromise. Avishai Margalit (2010) famously argued that some political compromises are “rotten,” namely ones that establish or maintain an inhumane regime. In such cases, morally responsible parties include both the side who allows and enables establishing such a regime and the side who actually establishes it.

2 Lepora (2012: 7–9) distinguishes among three types of compromises. (1) In a substitution compromise (the Two Doctors example), parties agree to do something different from what either of them primarily wanted. They replace their goals and do not preserve even parts of the original goals. (2) In an intersection compromise, compromising parties preserve some of their principles or goals. Only a subset of the principles or goals that each held at the outset is sacrificed. In this sense, intersection compromises require fewer concessions than substitution compromises. (3) In a conjunction compromise, parties originally have completely opposing principles (or goals), but they decide to agree to a compromise so that both can preserve some of their original principles (or goals). Conjunction compromises tend to be particularly unpleasant, as the parties must agree to pursue things they originally opposed.

3 Fabian Wendt (2019: 2858) makes an important point when he argues that “[i]t should be noted, though, that not all procedural constraints are requirements of fairness. Some procedural requirements are moral requirements whose violation undermines a party’s consent (see Wendt 2016, Chapter 4). When A threatens to kill B or her family, then this undermines the validity of B’s consent. Invalid ‘consent’ does not generate moral obligations for the party. When a party’s consent is not genuine, then the compromise is in the end not appropriately called a ‘compromise’ at all. Procedures that constrain people in the use of violence, threats of violence, and deception thus do not constitute ‘fair procedures’. Their role is more basic.”

4 Note that the 6-2 division of the champagne bottles fits split-the-difference view only in so far as the two parties are seen to lose equal amounts of the good that they distribute (based on their views of fair distribution, i.e., each loses two bottles). However, this might not translate to them making equal sacrifices as the 4-4 team member is in fact sacrificing 50% of her initially claimed share whereas the 8-0 team boss only sacrifices 25% of her claim. It is not always clear what split-the-difference fairness actually implies.

5 I have interpreted the idea of split-the-difference mathematically, but another interpretation might be based upon how hard it would be for the parties to live with the results. Here the notion of fairness would track the actual psychological price each party would suffer. The moral idea here would be that a fair compromise is one where both parties pay the same psychological price.

6 Suppose that two medical doctors disagree about the appropriate treatment of their patient. Also suppose that, if the doctors relied on split-the-difference fairness in their compromise, the patient would die. In this case, it would be morally right to avoid split-the-difference fairness