Ditching Morality?

Political realism aims to give “greater autonomy to distinctively political thought” (Williams Reference Williams2005: 2–3). At least some political realists understand this as a postulate about the type of normativity that is appropriate for political thinking. Political ethics aims to articulate and defend claims about politics that are normative in a moral sense, for example, claims about what is morally wrong to do for politicians. But moral normativity is only one kind of normativity. Insofar as political realists suggest that proper political thinking involves some other, more distinctively political type of normativity, they reject political ethics as a form of moralism.

Of course, when political realists speak of moralism, they do not primarily have political ethics in mind, if only because political ethics is a relatively neglected subdiscipline in political philosophy. What they have in mind are primarily mainstream liberal theories of justice. Caricaturing a bit, such theories start with certain moral intuitions, for example about fairness, and then develop abstract ahistorical principles of justice in the light of which real-world institutions are to be evaluated. My interest here is political ethics, not theories of justice, though, and so I will try to focus on realist worries about moralism as they apply to political ethics. The most convenient way to approach the realist challenge to political ethics is to take a look at the types of normativity political realists have proposed as alternatives to moral normativity.

Political Prudence

One possibility is to restrict normative thinking about politics to considerations of political prudence. There is a philosophical tradition in which political prudence is understood as excellence in political leadership, and this is to involve moral self-mastery and the ability to properly attend to the situation and “through deliberation and careful judgment seek the ends of political excellence” (Dobel Reference Dobel1998: 76, cf. Coll Reference Coll1991). Obviously, this is not an understanding of political prudence that would help political realists to purge political thinking from morality.

Rather, for political realists, political prudence must be conceived in a non-moralized way, basically as the sum of skills that are necessary to thrive in politics, to successfully obtain and wield and enlarge political power. Niccolò Machiavelli in The Prince can be understood as a teacher of political prudence in that sense. He writes, for example, that

One can also read Hans Morgenthau’s political realist theory of international relations in this spirit. Political realism, says Morgenthau, “maintains the autonomy of the political sphere” (Reference Morgenthau1948/2005: 15). Politicians, at least when they are rational, “think and act in terms of interest defined as power” (Reference Morgenthau1948/2005: 5). From the point of view of political realism, a “rational foreign policy [is a] good foreign policy” (Reference Morgenthau1948/2005: 12), and prudence, understood as the skill to appropriately weigh the political consequences of one’s actions (i.e., their consequences for the distribution of power), is the “supreme virtue in politics” (Reference Morgenthau1948/2005: 14). Morgenthau’s approach is, to some extent, inspired by Thomas Hobbes, who puts “for a general inclination of all mankind, a perpetual and restless desire of power after power, that ceaseth only in death” (Reference 63Hobbes and Gaskin1651/1998: Ch. 11).

There is no doubt that thinking about political prudence is a meaningful enterprise. We have to understand political prudence if we want to understand why certain personalities and certain tactics are and are not successful in politics. Reflections on political prudence are common in political commentary, and they also have a place in political philosophy, not necessarily because political philosophy is to teach and advise politicians how to succeed, but because nothing political philosophy cares about can be realized through a politician’s actions if the politician does not have the skills to succeed in politics. (More on this in Section 3.)

But should normative thinking about politics be limited to judgments of political prudence? That seems quite implausible. If we embrace a non-moralized notion of political prudence, then it can be prudent for a political leader to secretly order the murder of journalists whom they perceive as a threat, but doing so is, of course, morally wrong. Pointing this out will be a statement of political ethics, and, to me, it seems like a perfectly reasonable statement.

Self-Serving Ideologies

Proponents of the radical wing of political realism suggest a different way to engage in normative political thinking without ethics (Prinz and Rossi Reference Prinz and Rossi2017, Cross Reference Cross2022, Aytaç and Rossi Reference Aytaç and Rossi2023): They propose to criticize ideologies from an epistemic rather than moral point of view. Uğur Aytaç and Enzo Rossi appeal to empirical studies that show the prevalence of politically motivated reasoning, and, on that basis, they suggest that an ideology that helps a dominant social group to defend or increase its power is probably based on motivated reasoning and thus epistemically flawed: “To the extent that hierarchical power is essential to its own justification, self-justification of power counts as a malignant form of epistemic circularity: given their incentives, we shouldn’t trust the powerful that their power is justified” (Reference Aytaç and Rossi2023: 1219 n.9). Epistemic ideology critique points out what we ought not to believe – because it is epistemically flawed – and so it does involve a type of normativity. This type of normativity – epistemic normativity – is clearly different from moral normativity.

Just like it is useful to think about political prudence in political philosophy, it is useful to engage in epistemic ideology critique. I would doubt that demonstrating that an ideology is advocated by people who benefit from it is enough to discredit that ideology. At best, it might make the ideology suspicious (see Sleat Reference Sleat2024: 146–147). Nevertheless, to raise such suspicions makes sense.

I am not sure where an exclusive focus on epistemic ideology critique would leave normative thinking about political behavior. Ideologies may sometimes have implications for the evaluation of political behavior, but epistemic ideology critique only tells us what not to believe about how politicians ought and ought not to behave; it does not vindicate any beliefs about how politicians ought and ought not to behave. Epistemic ideology critique is just critical, it is not constructive.

The crucial question is whether and why normative thinking about politics should be limited to such epistemic ideology critique. Why not allow epistemic ideology critique and moral thinking about politics? Ultimately, I think the radical realists’ answer lies in their general skepticism about the possibility of moral knowledge or justified moral belief. When Aytaç and Rossi criticize moralist ideology critique as an alternative to their own epistemic ideology critique, they question the “presupposition that human beings have direct knowledge of basic matters of justice under good enough social conditions” (Reference Aytaç and Rossi2023: 1218). In a similar spirit, Raymond Geuss points out that “ethics is usually dead politics” (Reference Geuss2010: 42): Sedimented power relations present themselves as moral truths. Famously, Karl Marx and Friedrich Engels held that ethics is a mere “superstructure” phenomenon that reflects nothing but economic relations: “Law, morality, religion, are to him [the Proletarian] so many bourgeois prejudices, behind which lurk in ambush just as many bourgeois interests” (Reference Marx and Engels1848/1969: Ch. 1).

I cannot attempt to defend the possibility of moral knowledge or justified moral belief here. A point I would like to make, though, is that it seems very difficult to avoid moral judgments when thinking normatively about politics. Even thinkers who are officially committed to refrain from making moral judgments can hardly consistently avoid them. G. A. Cohen remarks that Marx believed that capitalism was unjust, but did not believe that he believed it was unjust (Reference Cohen1983: 444). Relatedly, one will probably not be able to avoid motivated reasoning by limiting oneself to epistemic ideology critique. One might exclusively focus on epistemic ideology critique because one is predisposed to like the conclusions one can draw with this research agenda. There is no way getting around that, and, to me, it seems that all we can do is try our best to let the moral arguments speak for themselves, rather than to stop moral reasoning.

Does Politics Have a Function?

Carlo Burelli proposes that normative thinking about politics should focus on “functional normativity.” What does he mean by that? The function of a heart, for example, is to pump blood; it is what a heart should do. We know that, says Burelli, because, first, hearts tend to pump blood, and, second, because pumping blood contributed causally to the continued existence of hearts (Reference Burelli2022: 630). Accordingly, when a heart performs its function well, we can evaluate this heart as a good or healthy one, and this evaluation is not a moral evaluation, but a functional one. Note that a functional evaluation is not relative to ends that we may share or not; it is a fact of the matter that a heart should pump blood, and, accordingly, it is an objective matter whether a heart is healthy or not, whether it does well what it is supposed to do.

The evaluation of hearts has not much to do with politics, of course. But Burelli thinks that there is a “function of politics” as well, which he identifies as the selection of binding collective decisions under conditions of permanent disagreement (Reference Burelli2022: 637). Just like hearts tend to pump blood, and pumping blood contributed causally to the continued existence of hearts, politics tends to bring about binding collective decisions under conditions of permanent disagreement, and the fact that they do played a causal role in bringing political institutions into existence.

Burelli seems to mostly have political institutions in mind when he reflects on the function of politics. But one could try to argue that selecting binding collective decisions is not just the function of political institutions, but also of the conduct of individual politicians: The function of politicians is to do their part in enacting binding decisions under circumstances of permanent disagreement. When a politician acts in ways that create severe instability in the political system, then he or she will do something that should not be done in light of the function of politics.

Let us accept, for the sake of argument, that Burelli has correctly identified the function of politics. Burelli admits that this does not show that normative political theory should exclusively focus on functional normativity, so understood. But he claims that functional normativity has priority over moral normativity:

It is questionable, though, whether this strict priority can be vindicated. Compare a polity that is able to implement collective decisions but uses it to commit a genocide, and a polity that is not (sufficiently) able to implement collective decisions and yet manages to be relatively stable and peaceful: Clearly, it makes sense to say that the former is overall morally more vicious and less preferable than the latter, and this shows that the ability to implement collective decisions does not have strict normative priority (see Erman and Möller Reference Erman and Möller2023: 9–10).

Burelli might reply that it is not conceivable to have a peaceful and stable society without a minimum level of ability to make collective decisions. But, for one thing, ordered anarchy is conceivable, even if – perhaps – unlikely. Second, it is hard to deny that some murderous, totalitarian states are worse than anarchy, even though they have working mechanisms for collective decision-making; functional normativity does not have strict priority over moral normativity.

Moreover, while Burelli rightly emphasizes that functional normativity is an objective matter, not relative to aims an agent may or may not have, it is also clear that functional normativity only matters to us if and because we want the respective function to be performed. It is perfectly conceivable to acknowledge that the function of a mafia boss is to organize crime, and yet to hold that one ought not to be or become a mafia boss, to hold that the mafia should not exist, and so on. Similarly, one will only care about the function of politics if one thinks that it is a good thing to have politics. And that evaluation is not a functional evaluation. It is, as far as I can see, a moral one, based in the value of peace and order, or something like that (see also Erman and Möller Reference Erman and Möller2023: 14–17). Burelli sometimes seems to concede that, for example, when he says, in the earlier quote, that the ability to make binding decisions has priority over “other moral concerns” (emphasis added).

Don’t Ignore the Realities of Politics

Political realists have not given us a compelling reason to refrain from moral thinking about politics and thus to give up on the project of political ethics (see also Erman and Möller Reference Erman and Möller2015, Leader Maynard and Worsnip Reference Leader Maynard and Worsnip2018). Some political realists agree (and, indeed, some have made important contributions to political ethics). But these realists insist that moral thinking about politics is nevertheless different from moral thinking in other areas. Referring to Bernard Williams, Robert Jubb suggests that political normativity is merely distinct from moral normativity “in the sense that the weight, direction and relevance of different considerations would all systematically be altered by politics’ constitutive features” (Jubb Reference Jubb2019: 362, cf. Sleat Reference Sleat2016: 253–254, Hall Reference Hall2017: 290–291). On this understanding, the political realist idea that we should not invoke standards external to politics does not dismiss moral thinking about politics as such, but only moral thinking about politics that ignores the constitutive features of politics. That, it seems to me, is a reasonable position, and arguably it is a position that many who, like myself, do not think of themselves as political realists, can share. In any case, taking this position means to accept that we should not try to keep morality out of political philosophy and political theory (cf. also Philp Reference Philp2012, Miller Reference Miller2016, Sleat Reference Sleat2024).

What are politics’ constitutive features? Jeremy Waldron takes two features to constitute what he calls the “circumstances of politics”: “[T]he felt need among the members of a certain group for a common framework or decision or course of action on some matter, even in the face of disagreement about what that framework, decision or action should be, are the circumstances of politics” (Reference Waldron1999: 102). The need for politics arises because a common framework or decision is needed, and yet there is disagreement about what that framework or decision should be. The relevant disagreements and conflicts can, of course, concern many different things, from the question whether abortion should be illegal to the question of whether the coal industry should be subsidized, and the background for such disagreements about policies is often rooted in ideological disagreements. Liberals, conservatives, social democrats, and so on, have different values and philosophical ideas about what the state should and should not do. Political ethics should be applicable to politicians who operate in such circumstances of disagreement, and so it should properly take these circumstances of politics into account (more on this in Section 5).

Realists inspired by Bernard Williams (Reference Williams2005: 4) will add that politics always aims to be legitimate, that it aims to be justifiable to its subjects. I am not sure whether this should be seen as constitutive for politics as such, since politics sometimes seems to result in brute force that does not aim to be legitimate. But I do not want to quibble about what is constitutive of politics, and what is a more contingent feature. When it comes to politicians in current democracies, it is clear that the goal to justify what they do to citizens is indeed an important one. It is part of what politicians try to do when they give press conferences, when they debate in parliament, and so on.

There are many other features that are probably not constitutive for politics as such, but nevertheless shape politics as we know it today. Politicians are typically members of political parties, they aim to win democratic elections to gain political power, they run campaigns, and so on. These realities must also be acknowledged in political ethics, if it is to be relevant, and, perhaps, in political philosophy quite generally (see Horton Reference Horton2010: 433–434).

But what does it mean to acknowledge these things in political ethics? For one thing, political ethics should actually try to find answers for how politicians ought to behave in the context of the realities of politics. To give an example, a question that political ethics should address is what compromises politicians should be willing to accept in coalition governments (see Section 5). This question arises in the context of democratic party politics, and it presupposes the need for binding decisions against the background of conflict and disagreement. Political ethics does not ask very abstract and general questions of how human beings ought to behave in general, and it does not assume that ethical principles that apply to private life can simply be transferred to politics. Rather, political ethics essentially takes the form of a “role ethics” (cf. Hardimon Reference Hardimon1994): It investigates what politicians in their role as politicians should do.

Further, depending on specific historical circumstances, political ethics will differentiate what these roles require and permit. What we can expect from politicians in a modern Western democracy is different from what we could expect from politicians in ancient Persia, for example.

Finally, to acknowledge the realities of politics in political ethics means to accept them as constraints on what can be conceived as moral requirements and virtues in political ethics. Political ethics will not declare it a moral requirement or a virtue for politicians to try to transcend political disagreement by convincing everyone else of their position, or to win democratic elections without running a campaign, for example.

The realities of politics thus play a crucial role in constructing the moral principles of political ethics. They do not merely matter in the application of independently developed moral principles. Leader-Maynard writes that “the realist should appeal to an understanding of theory-building as profoundly generative of human intellectual capacities, rather than merely a clarificatory or deductive ‘working out’ of pre-given or self-evident notions” (Leader Maynard Reference Leader Maynard2022: 458, see also Sangiovanni Reference Sangiovanni2008).

Summary

Political realism is opposed to “moralism” in politics and, as such, it seems opposed to the very project of political ethics. Some political realists suggest that normative political thinking should exclusively (or at least primarily) focus on assessments of political prudence, epistemic ideology critique, or the functionality of politics. The arguments for doing so are not convincing, though. One cannot escape morality in normative political thinking. This means that we need not give up on the project of political ethics.

Other political realists rightly suggest that political ethics should properly acknowledge the realities of politics: What is constitutive of politics is that there is a perceived need for a common framework or decision, but disagreement about what that framework or decision should be. Other essential features for politics in contemporary democratic societies are that politicians aim to justify their decisions to the population and want them to be regarded as legitimate, and that their actions are embedded in an institutional framework that includes political parties, democratic elections, and so on. Political ethics should try to answer ethical questions that arise within these realities of politics and, as such, it will have to take the form of a role ethics, an ethics for the role of the politician. The realities of politics will thereby serve as constraints to what can be regarded as morally required and virtuous.

Rational Egoists Everywhere

Rational choice theory assumes that persons have well-ordered preferences and efficiently act to satisfy these preferences. Utility is the measure of preference satisfaction, and so one can say that persons are utility maximizers, according to rational choice theory. In principle, rational choice theory leaves open what people’s preferences substantively are. Often, though, it is more narrowly assumed that people’s preferences are selfish. This, then, is the picture of “economic man” or “homo economicus.” It is commonly presupposed in economics. Public choice theory applies it to the realm of politics.

Why is the project of political ethics (seemingly) at odds with public choice theory? No matter the details, political ethics holds that it is sometimes morally wrong to pursue one’s self-interest in politics, that there are some moral constraints on political conduct, and that politicians, at least to some extent, should be sincerely motivated to promote their vision of political morality. Of course, it is not inconsistent to hold that politicians do nothing but rationally pursue their self-interest and to hold that they morally ought to behave differently. What politicians in fact do and what they morally ought to do are simply different questions, and so are the questions of what politicians in fact are like and what good politicians would be like. But the project of political ethics would arguably be uninteresting and somewhat pointless if politicians must be assumed to behave like homo economicus.

The rational choice model of human behavior has been challenged by behavioral economics (see Kahneman Reference Kahneman2011, Thaler Reference Thaler2015 for overviews) and experimental economics (see Smith Reference Smith2007, Charness and Pingle Reference Charness and Pingle2022 for overviews): Real people do not always seem to behave like homo economicus, in particular outside of market settings. With regard to public choice theory more specifically, critics have argued that it fails to explain and predict the actual behavior of politicians and other political agents, for example, voters (Kelman Reference Kelman1988, Shapiro and Green Reference Shapiro and Green1994, Udehn Reference Udehn1996: Ch. 2).

Of course, how the findings of behavioral and experimental economics are to be interpreted is still up for debate: What might look like an anomaly at first, may actually be compatible with rational choice theory in general and public choice theory in particular. And even if people sometimes do not behave like homo economicus, this does not mean it cannot be illuminating to assume that they do, in particular when it comes to explaining and predicting patterns of human behavior in the aggregate. As Richard Posner puts it, the picture of human behavior painted in behavioral economics is that…

This is not the place to evaluate the merits and limits of the homo economicus model of human behavior. But once it is understood that public choice theory need not claim that each and every individual politician under all circumstances behaves like homo economicus, the tension between public choice theory and the project of political ethics is greatly reduced.

Utopophobia

Even though public choice theory need not claim that each and every politician behaves like homo economicus under all circumstances, it does hold that self-interest and, more specifically, a desire for power, influence, fame, and money, are important motivations in politics. And this, it seems to me, is a pretty reasonable assumption, when it comes to politicians, and in particular politicians in leadership positions (which is, as I said in the Introduction, my focus in this Element). In public administration, power, influence, and fame arguably play much less of a role. But self-interest, if just in the form of a desire to make a good and comfortable living, is probably an important motive among bureaucrats as well.

If self-interest is an important motivation in politics, what are the implications for political ethics? In Section 1, we said that political ethics should be sensitive to the realities of politics. If we follow public choice theory and assume that self-interest is an important motivation in politics, being sensitive to the realities of politics means that the ethical standards upheld in political ethics cannot be too high-minded. Political ethics cannot make it a moral requirement for politicians to engage in selfless social service; it must take it as morally permissible to pursue career ambitions in politics that are to some extent driven by a self-interest and a desire for power, influence, fame, and money. To some extent, a drive for power may even be necessary to be successful in politics.

Some philosophers who defend more idealistic theorizing against what they regard as “utopophobia” deny that facts about human selfishness set limits on what can be regarded as moral requirements. As David Estlund points out, while “cannot do” may be requirement-blocking, “will not do” is not (Reference Estlund2011: 211). If a person is too selfish to comply with an ethical standard, while otherwise perfectly able to comply, this does not show that the ethical standard should be reconsidered. He gives the example of someone who finds herself motivationally unable to comply with the moral requirement to refrain from dumping trash by the side of the road, which is hardly requirement-blocking (Reference Estlund2011: 219–220). This does not change, says Estlund, if we assume that it is quite common to be selfish like that person (Reference Estlund2011: 220). Likewise, then, if politicians are generally too selfish to bring themselves to comply with an ethical standard that is proposed in political ethics, this does not show that we should reconsider that standard, if Estlund is right.

Interestingly, though, Estlund does not deny that ethical standards should not be overdemanding. He merely argues that this is not a concession to human nature, in particular to the natural selfishness of humans: “The suggestion is not that the more partial motivations are widespread or natural but that they are morally sound or reasonable, and this is a substantive moral judgment about their content” (Reference Estlund2011: 223). In other words, if political ethics refrains from postulating high-minded ethical requirements, then this is not because politicians cannot bring themselves to comply with such standards, but because politicians legitimately pursue their selfish career goals (at least to some degree). When digging deeper, though, the reason why it must be considered legitimate to pursue one’s career goals (to some degree) will probably have to do with human nature, as far as I can see.

David Schmidtz offers an additional thought in support of modest ethical standards in politics. In his discussion of Estlund’s argument, he points out that the crucial question is not so much whether I can command my own will to comply with a demand, but whether I can expect other people to comply. “That I do not choose for everyone is the political fact of life. It is nothing like weakness of will” (Reference Schmidtz2016: 3). That I cannot bring myself to refrain from dumping trash on the side of the road is a matter of weakness of will, but that I cannot expect politicians to put their career ambitions to the side is what matters for political ethics.

Constitutions for the Worst Case

There is a second branch to public choice theory, and that is normative public choice theory, also known as constitutional political economy. Geoffrey Brennan and James Buchanan quite readily admit that people do not always behave like homo economicus and, in particular, that we find “nonself-seeking behaviour by human agents in all institutional settings” (Reference 60Brennan and Buchanan1981: 165). But, they say, public choice theory is not to predict individual behavior, but to derive normative propositions about institutional design: “The model of human behaviour that we might properly use in choosing among alternative institutions may be different from the model that would be more appropriate in making predictions about behaviour within existing institutional structures” (Reference 60Brennan and Buchanan1981: 159, cf. 164–165, Reference Brennan and Buchanan1988: 180).

This approach to institutional design has a long and honorable tradition: David Hume held that, in politics, “every man must be supposed a knave” (Reference Hume and Miller1741/1985: 42). If it is reasonable to hold that the motive of self-interest is prevalent in politics, even though not every individual politician behaves like homo economicus all the time, then political institutions should be designed to work well even when populated with politicians who behave like homo economicus. Political institutions that are designed that way protect citizens from putting too much trust in the virtue of politicians and thus from being exploited and mistreated. In addition, such institutions eliminate the incentive for politicians who are not knaves to behave as if they were (Kliemt Reference Kliemt, Rowley and Schneider2008: 213): When well-intended politicians know that there are safeguards against knavish behavior, they can feel assured that their own efforts will not be futile.

What does “knavish” institutional design and, more specifically, constitutional design look like? We need not get into details here, but a variety of constitutional tools suggest themselves, from bicameralism, federalism, judicial review, and supermajoritarian voting rules to constitutionally enshrined liberal rights (for classic treatments, see Buchanan and Tullock Reference Buchanan and Tullock1962, Brennan and Buchanan Reference Brennan and Buchanan1985). As James Madison famously puts it in the Federalist Papers:

Crowding Out Virtue

It has been argued, though, that knavish constitutions, as recommended by public choice theory, tend to crowd out virtue among politicians. For our purposes, let us say that a virtuous politician is one who is, at least to some degree, public spirited and sincerely aims to promote their vision of political morality. A virtuous politician is not a homo economicus. If it were true that a knavish constitution, as recommended by public choice theory, crowds out virtue, this would show that public choice theory and the project of political ethics are in conflict, after all. Insofar as political ethics holds that politicians should be virtuous, it could not endorse knavish constitutions, as they are recommended by public choice theory.

Now, the claim that knavish constitutions crowd out virtue may seem puzzling at first. After all, if anything, knavish constitutions make it easier for virtuous politicians to thrive (see Brennan and Hamlin Reference Brennan and Hamlin1995: 40, 43, 45). If it does not pay to be a rational egoist, then we can expect more virtuous behavior and more virtuous people. As James Madison puts it, “the best security for the fidelity of mankind is to make interest coincident with duty” (Hamilton, Madison, and Jay Reference Hamilton, Madison, Jay and Cooke1788/1982: No. 72).

So why could knavish constitutions crowd out virtue? There are a couple of reasons. As Bruno Frey explains, individuals tend to lose the intrinsic motivation to behave as they should when they feel like their behavior is externally monitored: Intrinsic motivation is substituted by extrinsic control, and the “locus of control shifts from the inside to the outside of the person affected. Individuals who are controlled from the outside would feel ‘overjustified’ if they maintained their intrinsic motivation” (Reference Frey1997: 1045).

Relatedly, he continues, individuals tend to lose the intrinsic motivation to behave as they should when they get the sense that being intrinsically motivated to do the right thing is not acknowledged, when they feel like they are not being trusted. Feeling unappreciated, their self-esteem is impaired, which results in reduced intrinsic motivation (Reference Frey1997: 1045).

Samuel Bowles adds that facing economic incentives can lead to “moral disengagement,” to people switching their ethicality off (Reference Bowles2016: 89–90): It is not just that the intrinsic motivation to do what one should do is lost – even if it is not lost, it gets overlayed by egoistic motivation when people face incentives that appeal to egoistic motivation.

Finally, Lisa Herzog, among others, argues that the appeal to incentives communicates that many people are in fact rational egoists, which might induce people who otherwise would be able and willing to restrain their egoistic desires to not restrain them because they are only willing to restrain their egoistic desires under the assumption that others do so as well. She writes:

Applied to knavish constitutions, then, we can expect politicians who operate within such constitutions to lose their intrinsic motivation to behave virtuously because they feel monitored and because they feel not trusted, to morally disengage because they face incentives that appeal to egoistic motivation, and to not restrain their egoistic desires because they get communicated that everyone else is an unrestrained egoist as well.

A Response

Brian Kogelmann offers a twofold response to this. First, he points out that the mechanisms just described are not only at play in political systems with knavish constitutions but also in political systems without knavish constitutions: Politics as such crowds out virtue (Reference Kogelmann2023: 148–150). Politicians always face restrictions and thus external control: They must win elections, they must comply with the expectations of their political party, and they must meet the expectations of interest groups that approach them. Politicians regularly receive feedback that communicates distrust to them, be it from other politicians who are competing with them for votes, be it from the press, or be it from public opinion. And even if they are not constrained by a knavish constitution, politicians face incentives that appeal to their self-interest and trigger moral disengagement, for example, when they are approached by lobbyists, when they engage in logrolling, and so on. Kogelmann does not directly address Herzog’s argument, but one could argue that the desires for power, influence, fame, and money are prevalent enough in politics that a knavish constitution does not communicate anything new or surprising about selfishness in politics. If anything, it communicates that those selfish inclinations must be kept in check, which seems like a good thing to communicate.

Second, Kogelmann argues that, even though politics tends to crowd out virtue, knavish constitutions at least “partially counteract this tendency by cultivating virtue” (Reference Kogelmann2023: 151). Virtue has to be developed through habituation, and knavish constraints help to habituate virtue by incentivizing politicians to do the right thing (Reference Kogelmann2023: 151–153). He cites Aristotle:

Kogelmann gives three examples (Reference Kogelmann2023: 153–154): Bicameralism and an executive veto force politicians to make compromises, and so knavish constitutions let politicians habituate the virtue of a compromising mindset (see Section 5). Limited and enumerated powers force politicians to exercise legislative restraint, and so knavish constitutions will help politicians to habituate legislative restraint. Judicial review forces politicians to respect and revere the constitution, thereby letting politicians learn and habituate such respect. If Kogelmann is right, then public choice and political ethics are not only not at odds with each other, but cohere well and, in fact, support each other: Political ethics articulates what politicians ought (not) to do and what character traits they should have, and knavish constitutions, as devised by public choice theory, help politicians to develop the desired character traits.

While knavish constitutions may help to foster some desirable character traits in politicians – counterbalancing some of politics’ tendencies to foster undesirable character traits – constitutions can only do so much, though. Politics is not administration, and the rules that constrain politics must leave room for political judgment. To a considerable extent, political virtue plays out in a sphere that cannot be incentivized by constitutions or other institutional rules. Knavish constitutions may prevent politicians from doing the wrong thing, but they will not help much to find the right or virtuous thing to do. That being said, the point remains that the worry that knavish constitutions crowd out virtue seems overblown.

Nevertheless, maybe we have to rely more on politicians’ virtue than the public choice idea of a maximally knavish constitution admits. As Sam Schmitt points out, even knavish constitutions permit constitutional change and reform, and purely self-interested politicians may change a good constitutional framework for the worse. At some point, we thus have to rely on the virtue of politicians to maintain the framework. Schmitt argues that, for that reason, Buchanan’s approach to constitutional design must, to some degree, rely on ethical politicians, and he postulates an ethical principle according to which politicians ought to uphold the “rules of the game,” that is, the constitutional framework (Reference Schmitt2023: 179, 193–194). One may add that his approach must also rely on virtue in those who set up the constitutional framework in the first place.

Moreover, less-than-maximally knavish constitutions might still be able to help politicians to habituate virtue, while on the other hand leaving them a little more discretion. As Adrian Vermeule points out, some precautions may be futile (Reference Vermeule2014: 54–57), and some risks may better be addressed ex-post rather than preemptively (Reference Vermeule2014: 72–76). Other precautions may be suboptimal: The optimal level of risk may not be zero, after all, since knavish precautions restrain potentially beneficial political action (Reference Vermeule2014: 58–63, see also Kogelmann Reference Kogelmann2015: 465–470). In a similar spirit, Geoffrey Brennan and Alan Hamlin point out that, if we want our institutions to work even when populated with full-blown rational egoists, we have to discard institutional devices that work fine as long as some agents have some virtue (Reference Brennan and Hamlin1995: 41–42). Their example is selection procedures for the appointment of judges that serve as a screening device; if we assume that all judges will be rational egoists anyway, we have to declare any such procedures as irrelevant, which they clearly are not. Perhaps one can think of political representatives more generally not as agents whose interests are maximally misaligned with the interests of the principals (the constituents), so that it might pay out to try to select relatively virtuous political representatives rather than primarily relying on sanctions and incentives (cf. Besley Reference Besley2006, Mansbridge Reference Mansbridge2009).

Perhaps, then, normative public choice theory should not presuppose that every politician is a full-blown rational egoist, but relax its assumptions about the motivational sources of politicians at least a little bit. We must, to some extent, rely on some virtue in politicians. Let me emphasize, though, that even if we give up on the idea of a maximally knavish constitution, this does not mean to give up on the idea of “economizing on virtue” (Brennan and Hamlin Reference Brennan and Hamlin1995: 40): Even if people must not be assumed to be full-blown rational egoists, virtue is scarce, and so there it still seems desirable to have constitutional rules that work even if “many do not behave as virtue requires all the time.”

Be that as it may. The main point of this (sub-)section is not to defend or criticize the idea of a knavish constitution. Rather, the question we pursued was whether the project of political ethics is at odds with normative public choice theory insofar as knavish constitutions – as advocated by public choice theory – tend to crowd out virtue. And our discussion suggests that the answer is negative.

The Seduction of Political Power

Normative public choice theory worries about self-interest in politics: We should not fool ourselves into believing that politicians are saints and, for that reason, we should carefully restrain the business of politics and limit its reach. But self-interest in politics may not be the only reason to be wary about political power. Lord Acton’s famous dictum that “power corrupts” (Reference Acton and Himmelfarb1887/1948: 364) expresses an even darker thought: That people change as the result of obtaining political power. As Lars Udehn writes, it is “dangerous to invest people with too much power, especially if unchecked.” But this “has little to do with self-interest, or with economic man for that matter.” Rather, “[p]ower, itself, is the problem, not petty self-interest” (Reference Udehn1996: 191).

Following Mark Philp, I would like to understand the thesis that power corrupts as the thesis that political power “distorts the character and moral judgments of those who rule” (Reference Philp2007: 104). It is a distortion that can be dubbed the “arrogance of power.” Philp explains:

“Corruption” is here understood in a loose sense. While some have defended a notion of corruption as a cognitive distortion (see Blau Reference Blau2018), this “corruption” is quite different from the type of corruption I will discuss in Section 4 – the abuse of power for personal gain: The sleazy politician is to be distinguished from the overreaching politician (Philp Reference Philp2007: 106). The overreaching politician abuses power due to hubris, a distorted way of thinking about how political power can be legitimately used for nonpersonal ends. Philp gives the example of Watergate (Reference Philp2007: 105). A tendency that makes our current times feel so dangerous is that populist rhetoric increasingly seems to allow political leaders to expand their power and ignore or eliminate checks and balances in the name of the “will of the people.” At the time of writing, Donald Trump posts on X: “He who saves his Country does not violate any Law” (February 15, 2025).

Philp offers some quite plausible explanations for why politicians in leadership positions develop hubris (Reference Philp2007: 107–110, see also Nagel Reference Nagel and Hampshire1978: 76): With growing political power, politicians experience an estrangement from others and a weakening sense of solidarity, they get surrounded by people who do not give them critical feedback, the public approaches them with a combination of great expectations on the one hand and cynicism and hostility on the other hand, and, after an often long and difficult path into power, the power politicians finally obtain feels disappointingly restricted and limited in too many ways.

What does this mean for political ethics? The thesis that power corrupts in the sense explained is a tendency, not a necessity (Philp Reference Philp2007: 113–114). It leaves space for virtue in politics. But it affects how we must think about integrity in politics. Integrity is a virtue that, vaguely speaking, has to do with the extent to which a person shows fidelity to principles they sincerely hold (more on this in Section 5). Because of the seductive nature of political power, integrity in politics will always remain a difficult and fragile achievement, Philp points out:

Lord Acton not only says that power corrupts but also that “absolute power corrupts absolutely”: The more political power is associated with political positions, the greater we can expect the distorting tendency to be. What is more, David Schmidtz suggests that the more political power is associated with a position, the more it will tend to attract people who are corruptible:

Like public choice theorists, political realists like Philp look at politics “without romance.” It is interesting that, unlike public choice theorists, political realists do not aim to shrink and heavily restrain the business of politics. They are not libertarian-leaning. But to explore the reasons why is beyond the scope of political ethics (and this Element).

Summary

Public choice theory models political agents as homo economicus, which is in tension with political ethics insofar as political ethics sometimes tells politicians to not behave like homo economicus. But, since positive public choice theory primarily aims to predict patterns of behavior in the aggregate, not individual behavior, this tension is greatly reduced. Insofar as public choice theory rightly emphasizes the important role self-interest plays in politics, it suggests that political ethics must be somewhat “utopophob,” if it wants to properly acknowledge the realities of politics.

Normative public choice theory recommends a constitutional design that works even when political institutions are populated with politicians who behave like homo economicus. It has been argued that such knavish constitutions crowd out virtue, which would bring normative public choice theory back in conflict with political ethics. These worries seem overblown, though, since the mechanisms that crowd out virtue are at play in politics as such. Nevertheless, we may have to rely on politicians’ virtue to a greater extent than normative public choice theory admits.

Self-interest in politics is not the only thing to worry about in politics. It has been argued that political power tends to distort the character and moral judgment of politicians. One implication for political ethics is that political integrity can be expected to always remain a fragile achievement.

Dirty Hands without Paradoxes

In his seminal paper on the topic, Michael Walzer states that a politician with dirty hands does what is morally right “in utilitarian terms,” but thereby also does something that is morally wrong (Reference Walzer1973: 161; for the prehistory of the dirty hands theme, see Parrish 2017). Walzer gives the example of a democratic candidate who must make a deal with a dishonest ward boss, and, even though the candidate may rightly think that he should accept the offer, he is very reluctant to do so and rightly feels guilty when he does (Reference Walzer1973: 165–166). Walzer also offers a more dramatic example involving a politician who authorizes the torture of a rebel leader who very likely knows the location of bombs hidden in apartment buildings. “When he ordered the prisoner tortured, he committed a moral crime and he accepted a moral burden. Now he is a guilty man” (Reference Walzer1973: 167). Walzer’s analysis of what it means to have dirty hands is somewhat paradoxical, as he admits himself: How can a course of action be both right and wrong? If what one does is morally right, why should one feel guilty?

It is tempting to offer an analysis of dirty hands that avoids this air of paradox. Consequentialists hold that the politicians in Walzer’s examples do what is morally right because their actions (presumably) produce the best-possible outcomes, given the alternatives. After all, as we saw, the right act is right “in utilitarian terms.” The politicians’ hands are dirty insofar as they violate moral rules that are justified from a consequentialist perspective, consequentialists will continue. Moral rules are justified, from a consequentialist perspective, if and because (widespread real or hypothetical) adherence to them leads to (maximally or at least sufficiently) good consequences. Of course, from a consequentialist perspective, politicians do not do anything wrong when they violate these rules under unusual circumstances where doing so produces the best-possible outcomes, and so they are not guilty of wrongdoing. Some, therefore, consider the thesis of dirty hands a conceptual confusion (cf. Nielsen Reference Nielsen, Rynard and Shugarman2000: 140). When politicians do what is overall morally right, their hands are not dirty. At the same time, consequentialists can explain and commend feeling guilty, in such situations, because developing dispositions to comply with justified moral rules seems commendable from a consequentialist perspective (cf. Hare Reference Hare1981, Railton Reference Railton1984).

This consequentialist take on the dirty hands thesis does not take the dirt on the politicians’ hands seriously enough, though. Walzer writes:

Similarly, Bernard Williams says that there is an “uncancelled moral disagreeableness” in dirty hands scenarios (Reference Williams and Hampshire1978: 63).

One way to make sense of this uncancelled disagreeableness is to claim that, in dirty hands cases, incommensurable spheres of morality clash: Role obligations that arise for politicians and more ordinary moral duties that apply to everyone. The incommensurability explains the uncancelled moral disagreeableness.

I think one should avoid postulating an incommensurability between different spheres of morality, though. In dirty hands cases, it seems like the agents involved do plausibly deliberate about the different moral demands that apply to them and try to come to a conclusion about what is, all things considered, right or justifiable for them to do. If there were incommensurable spheres of morality, one could not make such all things considered judgments.

The following view promises to make sense of the uncancelled moral disagreeableness without postulating incommensurable spheres of morality (cf. also Nagel Reference Nagel1972, Coady Reference 61Coady2004: 778–780, Reference Coady2008: 82–83, Zanetti Reference Zanetti2022: Ch. 3). There are moral duties that function like side constraints; one may not violate them even if doing so would produce better consequences. But moral duties hold pro tanto: Once the consequences of violating a moral duty would be sufficiently better – once a certain threshold is met – the moral duty is outweighed. It then becomes, all things considered, morally permissible or even morally required to not fulfill it. One could say that the duty came into conflict with a more stringent conflicting duty, a duty to bring about the desirable outcome. In such cases, there is a sense in which a moral wrong is done, because a moral duty is not fulfilled, and yet what is done is, all things considered, morally right.

Theorists who are skeptical of this appeal to pro tanto moral duties will object that it does not adequately capture the moral remainder that is essential to the phenomenon of dirty hands: One cannot at the same time claim that an action is all things considered morally right and that there really remains something morally wrong about the course of action, something that warrants the feelings of guilt or remorse that an agent should experience in dirty hands cases (cf. de Wijze Reference de Wijze2024: 204).

In response, though, I think the view that appeals to pro tanto duties does adequately capture the moral remainder. Consider a case where the moral duty that is outweighed is correlated with someone else’s moral rights (cf. also Williams Reference Williams and Hampshire1978: 63): Because someone’s rights are infringed, there is a victim to whom one owes some sort of compensation, even though what one does is overall morally permissible (or even required). If one breaks into someone’s car to be able to get to the hospital, for example, this may be morally permissible if there are no other options, and yet one owes some kind of compensation to the car’s owner. The moral remainder here is very real; it comes in the form of a moral duty of compensation. And, while the moral remainder is most easy to see when a moral right is justifiably infringed, there is a moral remainder in other cases, too. Whenever a pro tanto moral duty is justifiably not discharged, a new moral duty arises, a duty to – in one way or another – make up for this failure to discharge one’s duty.

What about feelings of guilt or remorse? Does it make sense to feel guilty or remorseful after doing something that is overall morally right? Arguably it does, at least in cases where one infringed upon a person’s rights or otherwise wronged someone. Feelings of guilt or remorse need not express the belief that what one did was, all things considered, morally wrong. Sometimes they may only express that one violated someone’s rights or otherwise wronged someone, even though what one did was, all things considered, not morally wrong. Perhaps this is not obvious in the hospital case, because the wrong is mundane and relatively easy to make up for. But consider the case, cited by C. A. J. Coady (Reference Coady2008: 83), of prisoner doctor Olga Lengyel, who poisoned newborn babies in Auschwitz to save their mothers; every woman with a child in tow was marked for death. She said: “I marvel to what depths these Germans made us descend. […] The Germans succeeded in making murderers of us.” One can feel deep guilt without thinking that what one did was, all things considered, morally wrong.

How Severe Must Moral Transgressions Be?

Not all cases in which an agent permissibly fails to discharge a moral duty are properly called dirty hands cases. The hospital example arguably is a case in point. To count as a dirty hands case, the wrong that is done must be sufficiently severe. Matthew Kramer appeals to the concept of evil to explain the dirty hands thesis. Evil conduct, he explains, is “impelled by sadistic malice or heartlessness or extreme recklessness that is connected to severe harm in the absence of any significant extenuating circumstances” (Reference Kramer2018: 189). In (justified) dirty hands cases, one does something that is evil – or rather would be evil in the absence of the extenuating circumstances – in order to fulfill the stronger of two duties that cannot be simultaneously fulfilled (Reference Kramer2018: 197).

Somewhat similarly, in his later works, Michael Walzer reserves the notion of dirty hands for cases of supreme emergency. These are cases where “our deepest values and our collective survival are in imminent danger.” He gives the example of the British fight against Nazi-Germany in the years 1940 and 1941, when a Nazi victory seemed close and the British decided to bomb residential areas and the centers of German cities (Reference Walzer1988/2004: 33–34, see 46). He writes that “dirty hands aren’t permissible (or necessary) when anything less than the ongoingness of the community is at stake, or when the danger that we face is anything less than communal death” (Reference Walzer1988/2004: 46).

Why limit the concept of dirty hands to such extremely severe cases, cases that involve evil or supreme emergency? Thomas Nagel suggests that it is only in extremely severe cases that one really cannot justify one’s course of action to the victims. He writes:

I am not sure I see the difference Nagel is pointing at here. In both cases, one can justify one’s actions and, it seems to me, if one cannot justify one’s action to the victim in the second example, one can also not justify it to the victim in the first example. Of course, the first victim may be somewhat more inclined to accept one’s justification, but why should this be relevant to the question whether one’s hands are to count as dirty hands or not?

I do agree, though, that dirty hands cases involve a sufficiently severe moral wrong; the dirt must be clearly visible, so to say. As Michael Stocker puts it, there must be a real “violation and a betrayal of a person, value, or principle” (Stocker Reference Stocker1990: 18, cf. 20, 25). I will leave open where exactly to draw the line between wrongs that do and wrongs that do not involve such a violation and betrayal. But, to me, some cases that do not involve evil or extreme emergency still seem to involve enough dirt to count as dirty hands cases, for example, cases where politicians justifiably lie to voters (cf. Brennan Reference Brennan, Crookston, Killoren and Trerise2016), despite – perhaps – having a moral duty to be sincere or honest in public discourse (cf. Schwartzman Reference Schwartzman2011, Quong Reference Quong2011: Ch. 9, Carey Reference Carey2018). I am less sure whether it should count as dirty hands that politicians often have to faithfully administer past political decisions they condemn (cf. Waldron Reference Waldron2018).

Being Ready for Dirty Hands

The phenomenon of dirty hands is not peculiar to politics, if we do not restrict the application of the concept to the most severe cases. Michael Stocker offers an example in which the marriage of relatives has been troubled, but is improving now. One of them suspects that the other had an affair and wants to have these suspicions allayed. You know that the suspicions are correct, but when asked to lie in order to help save the marriage, you decide to comply. This, Stocker suggests, can involve dirty hands (Reference Stocker1990: 18–19).

But even though dirty hands are not exclusive to the realm of politics, if we do not restrict the application of the concept to the most extreme cases, it has far greater significance in the realm of politics. Stuart Hampshire points at three important features of politics that explain why (Reference Hampshire and Hampshire1983: 49). First, politicians, in particular in leadership positions, have responsibility for “greater and more enduring consequences and consequences that change more men’s lives.” When politicians make law, this has wide-ranging effects for the whole country, and even when a law is overall beneficial, there are always some individuals who are harmed. Second, politics is involved in violence and coercion (see also Weber Reference Weber, Gerth and Mills1919/1977: 125–126). In politics, decisions about war and peace are made, politics is responsible for maintaining law and order, and even mundane laws ultimately have to be coercively enforced. Third, politicians bear responsibility to act in the interest of the people they represent. They are, in a sense, the agents of the people (who can be conceived as the principal). These features, taken together, make dirty hands cases more prevalent in politics than in other contexts. Dirty hands cases typically involve decisions that have wide-ranging consequences and involve the use of coercion and violence as well as special responsibilities toward persons one represents.

What does that mean for political ethics? Quite straightforwardly, it means that we cannot expect politicians to have clean hands, especially politicians in leadership positions (which is, to repeat, my focus in this Element). More strongly, we have to consider it a virtue in politicians to be prepared and willing to get (justifiable) dirty hands: One cannot be a good politician without being ready for dirty hands. This requires a certain character: A politician cannot be too squeamish if he or she is supposed to be prepared and willing to get (justifiable) dirty hands.

One can read Max Weber’s famous essay on “Politics as a Vocation” in this spirit. Weber distinguishes an “ethics of ultimate ends” and an “ethics of responsibility” (Reference Weber, Gerth and Mills1919/1977: 120). An ethics of ultimate ends upholds moral principles unconditionally; an ethics of responsibility more flexibly considers the indirect and more long-term consequences of one’s actions. According to Weber, an ethics of responsibility is appropriate for politics, and good politicians are aware that their actions are not in line with an ethics of ultimate ends: They are aware of their dirty hands. Weber writes:

Michael Walzer agrees: “Here is the moral politician: it is by his dirty hands that we know him. If he were a moral man and nothing else, his hands would not be dirty; if he were a politician and nothing else, he would pretend that they were clean” (Reference Walzer1973: 168). William Galston speaks of the virtue of toughness as the “disposition that enables leaders to recognize harsh necessities without blinding themselves to moral costs” (Reference Galston1991: 176).

Moreover, not only must a virtuous politician be prepared and willing to get dirty hands, he or she must also be reluctant to get dirty hands, and feel the “uncancelled disagreeableness” of dirty hands. This is so for two reasons, as Bernard Williams explains. One is, quite simply, that it is appropriate to feel guilty or remorseful after doing something that is wrong, even when it is overall justified (Reference Williams and Hampshire1978: 65). The other is that “only those who are reluctant or disinclined to do the morally disagreeable when it is not necessary […] have much chance of not doing it when it is not necessary” (Reference Williams and Hampshire1978: 64, cf. 65). One may be skeptical whether it is possible to find or screen for politicians who meet this criterion (Philp Reference Philp2007: 94–95), but that is what political ethics will say about a good politician, a politician we should desire to have.

Why could one reject this picture of the good politician? Some consider it romanticized, elitist, and dangerous (Shugarman Reference Shugarman, Rynard and Shugarman2000: 243–244). Politicians should not be lone decision-makers who operate in lofty heights of detached political power, they should work closely with appropriate deliberative bodies and be responsive to their democratic constituents. But, even if one shares that picture of how political decision-making should ideally proceed, it will be difficult to purge the business of politics from the phenomenon of dirty hands, especially if we do not restrict the notion to the most extreme cases. Dirty hands are part of the reality of politics as we know it, and it has implications for political ethics.

Andrew Sabl points out that the focus on dirty hands cases paints a distorted picture of the everyday business of politics:

This is an important reminder that there are a variety of roles in politics, that political ethics will have to differentiate, and that the dirty hands thesis is not meant to exhaust what political ethics has to say about political conduct. Nevertheless, as far as I can see, if the dirty hands thesis is not limited to extremely severe cases, it seems like something that applies to politicians in many different roles, although the most extreme cases are something only politicians in leadership positions will experience.

Playing Dirty to Succeed

Are dirty hands only justifiable when necessary to achieve a morally desirable goal like, for example, saving a great number of people from a terrorist attack, or can dirty hands also be justifiable for the mundane purpose of defending or increasing one’s political power? Niccolò Machiavelli holds that it is impossible to succeed in politics without getting dirty hands, and, at least apparently, he implies that it is permissible to get dirty hands simply in order to succeed in politics: A “man who wants to make a profession of good in all regards must come to ruin among so many who are not good. Hence it is necessary to a prince, if he wants to maintain himself, to learn to be able not to be good, and to use this and not use it according to necessity” (Reference Machiavelli1532/1985: Ch. 15). Similarly, he writes:

Is this right? At least two qualifications suggest themselves. First, obviously not just anything can be justified for the sake of gaining or maintaining political power. One of the great achievements of modern democratic politics is that, for the most part, the transition of political power is peaceful. One thus may not use violence for the sake of obtaining political power. That being said, it is probably true that, even in modern democratic politics, one will not succeed in politics without doing some quite nasty things. Bernard Williams mentions “lying, or at least concealment and the making of misleading statements; breaking promises; […] sacrifice of the interest of worthy persons to those of unworthy persons; and (at least in a sufficiently important position) coercion up to blackmail” (Reference Williams and Hampshire1978: 59) as part of ordinary politics. Adam Gibbons adds that bullshitting in politics pays (Reference Gibbons2024). Some of these things will not count as dirty hands cases because they do not cross the line of severe enough “violations” and “betrayals” – but some will. And if it is really impossible to succeed in politics without doing some of these things, then perhaps this must sometimes count as permissible, and a good politician must be regarded as someone who, to some degree, is capable of (reluctantly) doing them.

Second, even Machiavelli sometimes seems to imply that political power is not an end in itself, but should be used for good purposes. He writes, for example, that “Cesare Borgia was held to be cruel; nonetheless his cruelty restored the Romagna, united it, and reduced it to peace and to faith” (Reference Machiavelli1532/1985: Ch. 17). That the necessities of political power justify some (relatively minor) dirty hands therefore probably only holds if and insofar one otherwise uses one’s political power for something good. I will leave open whether this good must be intended by the agent, or whether it is sufficient if it is brought about as a side effect (which arguably is the case in Machiavelli’s Borgia example).

Dirty Hands and Clean Gloves

Democratic politicians aim to justify their actions to the public: They need to be perceived as legitimate if they want to succeed long term (see Section 1). Dirty hands make it difficult for politicians to be perceived as legitimate, though, and so they often try to hide their dirty hands in clean gloves (Hollis Reference Hollis1982: 389, Bellamy Reference Bellamy2010: 425). This adds a kind of hypocrisy to the business of politics that I think we all know too well. Richard Bellamy writes: “[D]emocratic princes no less than the Machiavellian prince must appear compassionate, generous, reliable, morally upright and honest, yet be prepared to be treacherous, break their promises and use their resources selectively” (Reference Bellamy2010: 427).

Bellamy argues that democratic voters share some of the responsibility for the hypocrisy of democratic politics. We “tolerate wire tapping of terrorists and spying on unfriendly powers, but prefer not to be told. To a degree, we share the dirt with politicians” (Reference Bellamy2010: 426, see also Thompson Reference Thompson1987, Archard Reference Archard2013, Beerbohm Reference Beerbohm2018, de Wijze Reference de Wijze2018).

What does this imply for political ethics? As long as voters cannot be expected to face the reality of dirty hands, politicians must be good at hiding their (justifiable) dirty hands. This implies that we cannot expect politicians to be authentic and open about who they are and what they do. Political virtue seems quite different from what we would hope to find in a good friend.

Summary

The dirty hands thesis holds that, in politics more than elsewhere, doing what is overall morally right sometimes involves doing something that is (severely) morally wrong. The best way to make sense of this is to appeal to pro tanto moral duties that are outweighed under some circumstances. Unfulfilled duties leave a moral remainder, which is clearest to see in cases where somebody’s rights are infringed. This accounts for the uncancelled disagreeableness that we find in dirty hands cases.

Political ethics must hold that good politicians will be prepared and willing to get dirty hands, but also that good politicians will only reluctantly dirty their hands and will feel guilty or remorseful after doing so. Probably political ethics must also endorse the view that it is permissible for politicians to get some (less severely) dirty hands simply to be able to succeed in politics. And, because political actions and decisions in democracies must be perceived as legitimate, politicians must sometimes wear clean gloves to hide the dirt on their hands.

The Abuse of Power for Personal Gain

I adopt a broad notion of political corruption, which includes corruption that occurs in the legislative and executive branches of government, but also corruption associated with other public officials, from bureaucrats to judges and police officers.

What is corruption? At its core, corruption can usefully be characterized as the abuse of public office for personal gain (e.g., Nye Reference Nye1967: 419, Rose-Ackerman Reference Rose-Ackerman, Rowley and Schneider2016: 551). A paradigmatic example is bribery. When an officeholder takes a bribe, they make a personal gain – usually monetary – in exchange for some kind of service. This service is associated with their role as an officeholder, for example, the service of issuing business licenses. Bribery must be distinguished from extortion, where coercion is used to get the payment. Another example of political corruption is nepotism. Nepotism involves a public official who uses office powers to privilege relatives, friends, or political associates. In nepotism, the personal gain is the benefit of having relatives, friends, and associates who are grateful, loyal, and supportive, which is advantageous for one’s career and one’s influence in the office. That corruption involves a personal gain thus does not mean that the gain must be private (see also Ceva and Ferretti Reference Ceva and Ferretti2021: 35): The personal gain can also be related to one’s professional career.

Seumas Miller argues that acts of corruption need not involve the abuse of a power one has as an officeholder. He gives the example of a regular citizen who breaks into the local electoral office and falsifies the electoral role (Reference Miller2017: 80, cf. 100). Miller looks at corruption mostly from the point of view of its effects on the institution it is embedded in. As voters, citizens have an institutional role to play, says Miller, even though they do not hold an office. As I see it, it certainly makes sense to say that the citizen in his example is corrupting the electoral process. Yet because it is not an abuse of a power related to an office, one should deny that the citizen is performing a corrupt act. To contribute to the corruption of an institution or an institutional process is not the same as performing an act of corruption. (I will say more about that later.)

Like it is important to distinguish between corrupt acts and the corruption of institutions, we should also distinguish between corrupt acts and corrupt persons (or a corrupt character). Corruption as the abuse of office applies to acts, and it is, it seems to me, the primary notion: One should understand a corrupt person as someone who has a character disposition to engage in corrupt acts. A person who performs one singular corrupt act is not necessarily corrupt as a person. One becomes a corrupt person by repeatedly and habitually performing corrupt acts. The underlying motive will typically be greed, for money or other types of personal gain, but one can also be corrupt out of malignance or self-indulgent laziness, for example (cf. also Schmidtz Reference Schmidtz2023: 120–121).

Universal Standards and Local Conventions

Often, corruption involves a violation of legal norms, but this is not necessarily so. Some forms of corruption may be legal, although they should not be. But can we even identify what is corrupt without looking at the relevant legal conventions of a particular society? Elizabeth David-Barrett and Mark Philp argue that “the concept of political corruption cannot provide a set of clear and universal standards of conduct” (Reference David-Barrett, Philp, Hall and Sabl2022: 172). As an example, they point out how difficult and to a considerable extent arbitrary it is to determine what precisely is to count as a conflict of interests (Reference David-Barrett, Philp, Hall and Sabl2022: 178–184). Similar concerns can be raised about the distinction between bribes, gifts, and tips (cf. Rose-Ackerman and Palifka Reference Rose-Ackerman and Palifka2016: 236–242). Local conventional practices must determine what is to count as political corruption, and the relevant political decisions must be made against a background of disagreement about what should count as political corruption (Reference David-Barrett, Philp, Hall and Sabl2022: 176–177). What is more, since politicians can decide what is to count as political corruption, there is always a danger that they do so in self-serving ways (Reference David-Barrett, Philp, Hall and Sabl2022: 176).

To acknowledge that standards of corruption are in this way embedded in the practice of politics is what David-Barrett and Philp consider a realist approach to corruption. And, as pointed out in Section 1, I agree that political ethics should be sensitive to the realities of politics. I also agree that local practices – legal and informal norms – are needed to determine what is to count as corruption in a particular context. Corruption is at least to some extent “politically constructed” (Thompson Reference 68Thompson2018: 506). But, as far as I can see, this does not mean that there are no universal (and reasonably clear) standards of conduct related to corruption. If, as will be argued later, the abuse of office for personal gain is pro tanto morally wrong, then this is a universal principle or standard of conduct, albeit a somewhat vague one that different conventions can concretize in different ways. Although vague, it seems clear enough to allow theorists to identify cases where conventions fail to properly address behavior that is an abuse of office for personal gain.

Institutional Corruption

What is the relation between corrupt acts and corrupt institutions? According to Miller’s theory, social institutions are organizations or systems of organizations that are to serve certain purposes (Reference Miller2017: 23). Universities, for example, are to provide education, the police is to provide security, and so on. An institution is corrupted to the degree that it no longer serves its purpose. What he calls “acts of institutional corruption” are acts that corrupt an institution and/or a person’s character qua institutional role occupant (Reference Miller2017: 65–67). Of course, the causal effects of individual corrupt acts are often minimal and hardly discernible, but minimal effects do add up when many officials regularly perform corrupt acts. When bribery is widespread in a bureaucracy, for example, then the bureaucracy as an institution will eventually be corrupted, it will no longer function in the way it should. Sometimes, bribes might not have a causal effect on the corruption of the institution at all, though, when they are isolated and undetected. When this is the case, these bribes do not count as acts of institutional corruption, but as “acts of noninstitutional corruption” (Reference Miller2017: 69, 120).

To my mind, it seems a little artificial to speak of “acts of institutional corruption” in opposition to “acts of noninstitutional corruption.” Why not say that there are corrupt acts – acts that involve an abuse of power for personal gain – and then to take it as a separate question whether these acts contribute to the corruption of an institution (or to the corruption of the person’s character)? Typically, corrupt acts will contribute to the corruption of an institution (and of a person’s character). But they do not necessarily do so.

Likewise, institutions will typically be corrupted through corrupt acts; but this is not necessarily so either. The acts that causally contribute to the corruption of the institution do not necessarily involve an abuse of power for personal gain, and so they are not necessarily corrupt acts (see above). When an institution becomes dysfunctional and nobody can be held responsible for it, Miller speaks of institutional corrosion (Reference Miller2017: 66). His example is how reduced funding for a court system continuously leads to a decline in the quality of the court’s service, for which nobody can appropriately be held responsible. An alternative is to see this is an example of an institution that is corrupted through causal mechanisms that do not involve corrupt acts.

Dennis Thompson – who introduced the notion of institutional corruption into the debate – draws a different distinction between institutional corruption and individual corruption (Reference Thompson1995: 28–33, Reference 68Thompson2018: 502–504). In institutional corruption, says Thompson, politicians make political gains rather than private gains. He treats some forms of constituent service and campaign financing in the United States as prime examples of institutional corruption. That politicians try to secure votes to win an election is, generally speaking, perfectly legitimate and, indeed, expected in a democracy. It only becomes problematic when democratic processes are undermined, and this happens, most importantly, when politicians become problematically dependent on private interests (see also Lessig Reference Lessig2011: 230–231). The corruption of democratic institutions means that some groups of persons who are affected by democratic decisions are not properly included in democratic processes (Warren Reference Warren2004: 333–334). Individuals in corrupt institutions need not be corrupt persons; they may well be “good souls” who just try to do what they are supposed to do (Lessig Reference Lessig2011: 17).

In individual corruption, on the other hand, the gain is personal rather than political, an inappropriate motive is involved, and providing the favor is not part of the official’s job description, explains Thompson. He treats bribery as a prime example of individual corruption. One reason why it is important to Thompson to distinguish sharply between institutional corruption and individual corruption is that the former is in need of a different response from the latter. Rather than removing “bad apples,” institutional corruption requires an institutional response, for example, rules specifying what is to count as a conflict of interests (cf. Reference Thompson1995: 55–60, 124–130, Reference 68Thompson2018: 505).

Against Thompson’s sharp distinction between individual and institutional corruption, Emanuela Ceva and Maria Paola Ferretti suggest that the acts that contribute to institutional corruption should be classified as corrupt acts, even if they do not involve a corrupt motive: If a congresswoman responds to lobbyists in a problematic way, this should count as an act of corruption even if it is not motivated by personal gain (Reference Ceva and Ferretti2021: 57–58, cf. Weinstock Reference Weinstock2018: 226; for criticism of Thompson’s concept of institutional corruption see also Juarez‐Garcia Reference Juarez‐Garcia2023). As I see it, Ceva and Ferretti are right that institutional corruption is often caused by corrupt acts (individual corruption). Taking bribes and other corrupt acts do corrupt institutions. But this does not mean that all acts that contribute to institutional corruption are corrupt acts: Only acts that involve an abuse of office powers for personal gain are.

That institutional corruption can be caused by corrupt acts is not the only interesting connection between the two. Institutions can incentivize corrupt acts. As Daniel Weinstock points out, sometimes institutional rules create perverse or conflicting incentives (Reference Weinstock2018: 228–230), and sometimes the incentives that are important to let an institution work the way it is supposed to work can also lead to its corruption (and the corruption of those who work in the institution). The competition of political parties in representative democracies, Weinstock argues, is, in principle, adversarial in a good way: Political parties should be incubators for policy platforms which they are forced to present as visions for the general good (Reference Weinstock2018: 236–238, see also Goodin Reference Goodin2008: 212–213, Rosenblum Reference Rosenblum2008: 364, White and Ypi Reference White and Ypi2016: 13–14). But the desire to win that comes with the competition of political parties can also lead partisans to primarily try to confer benefits upon themselves and their party and subordinate the concern with the general good (Reference Weinstock2018: 239–240).

A Moral Duty Not to Perform Corrupt Acts?

Since this is an Element about political conduct, my focus will be on corrupt acts, not corrupt institutions. Is corruption – the abuse of office powers for personal gain – pro tanto morally wrong? One way to argue that it is pro tanto morally wrong appeals to the moral duty to perform one’s duties of office. Duties of office are, first of all, just conventional duties that are specified by legal and otherwise informal norms. It is thus an open question whether there is a moral duty to fulfill such conventional duties. Arthur Applbaum tells the story of Charles-Henri Sanson, executioner of Paris under king Louis XVI, who always conscientiously fulfilled his executioner duties and, in the course of the French Revolution, equally conscientiously executed Louis XVI (Reference Applbaum1999: Ch. 2). Applbaum’s main goal in his book is to show that “institutions and the roles they create ordinarily cannot mint moral permissions to do what otherwise would be morally prohibited” (Reference Applbaum1999: 3). We do not need to take a stance on that thesis. But, for our context, what the story conveys is that probably not all conventional duties of office translate into moral duties.

The question, then, is when and why there is a moral duty to fulfill one’s duties of office. Emanuela Ceva and Maria Paola Ferretti assume that the role-specific obligations of officeholders are morally binding because they are voluntarily incurred when someone enters office (Reference Ceva and Ferretti2021: 97, cf. Simmons Reference Simmons1979: 18–19). They hold that corruption is “inherently unjust because it constitutes a violation of the duty of office accountability” (Reference Ceva and Ferretti2021: 81, cf. 98). But this holds only as long as the institution the office is embedded in is morally legitimate (Reference Ceva and Ferretti2021: 97 n. 15, 111, 123).

There are other possible strategies to argue for a moral duty to fulfill one’s duties of office, but this is not the place to explore them. Probably the most straightforward approach would be to argue that, whatever the correct theory of political authority is, if it can show that (some) states have political authority, then those who hold offices within state institutions have the moral rights and the moral duties that correspond to the legal rights and legal duties that are associated with these offices.

For the present context, it is enough to make note that it is not, as such, pro tanto morally wrong not to fulfill one’s duties of office. It all depends on whether the circumstances are such that one has a moral duty to perform one’s duties of office, perhaps because one voluntarily took them on and the institution in which they are embedded is legitimate, perhaps for some other reason. Of course, arguably many politicians and bureaucrats find themselves in circumstances where it is pro tanto morally wrong not to fulfill one’s duties of office.

But it is not necessary to show that they do. This is because, no matter if the circumstances are such that it is pro tanto morally wrong not to fulfill one’s duties of office, it is plausible to hold that there is a pro tanto moral duty not to abuse one’s powers of office for personal gain – and thus to not perform corrupt acts. There are two ways to make that case. The easier one takes “abuse” to be a moralized concept from the start. To abuse one’s powers of office would then imply that one uses one’s office powers for an inappropriate purpose. Understood that way, an abuse of one’s office powers is pro tanto morally wrong.

Alternatively, “abuse” could be understood in a non-moralized way. So understood, to abuse powers of office merely means to use them for a purpose that is not in line with one’s mandate. This seems to leave conceptual space for “noble cause corruption.” Miller gives the example of a meagerly paid police officer in India who is prohibited by law to have a second job and, in order to be able to satisfy the basic needs his family, takes bribes from rich people in return for additional surveillance (Reference Miller2017: 98). Miller emphasizes, though, that usually habit will form the motivational basis of all corruption, and he identifies a tendency toward moral arrogance in anyone who engages in corruption (Reference Miller2017: 98–99). Ceva and Ferretti argue that noble cause corruption should not be classified as corruption because what makes it salient is not “a deficit of office accountability,” but rather a resistance to injustice (Reference Ceva and Ferretti2021: 41–42, 110). They give the example of Oskar Schindler, the German industrialist who saved the lives of 1,200 Jews during the Holocaust by employing them in his factories. I think we should not classify noble cause corruption as “real” corruption for a different reason, namely that it is not an abuse of one’s powers for personal gain. And that holds even when we work with a non-moralized conception of “abuse” (and thus of corruption). What makes corruption pro tanto morally wrong, then, is not that it involves an abuse of one’s powers, but rather that it involves an abuse of one’s powers for personal gain. If that is right, one can say that corruption – the abuse of one’s powers of office for personal gain – is pro tanto morally wrong, no matter if we endorse a moralized or a non-moralized reading of “abuse.”

Down the Hill

That it is pro tanto morally wrong to perform corrupt acts does not imply that it is always morally impermissible to do so, though. The pro tanto moral wrongness of corruption may sometimes be outweighed by other considerations. I will now look at the consequences of corruption to see whether considerations about consequences could sometimes make it morally permissible to engage in corruption despite its pro tanto moral wrongness.

Corruption can manifest itself in different forms; Michael Johnston (Reference Johnston2005: Ch. 3) speaks of “syndromes” of corruption. Simplifying a bit, in well-established market democracies, corruption mostly comes in the context of trying to get access and influence within strong state institutions (“influence markets corruption”). In newer market democracies with moderately strong economic and political institutions, corruption is largely controlled by elites in business, politics, the military, and the media (“elite cartel corruption”). In societies with weak institutions, corruption is often extensive, unpredictable, and associated with violence and the open exploitation of power (“oligarch and clan corruption,” “official mogul corruption”). The consequences of corruption will be different depending on the syndrome we are dealing with.

That being said, there are some quite general explanations for why (widespread) corruption has pernicious social and economic consequences. Susan Rose-Ackerman argues that toleration of corruption can lead to a downward spiral (Reference Rose-Ackerman, Rowley and Schneider2016: 552–554): Toleration of corruption provides incentives for officials to create more scarce benefits for which they then can be bribed and to create more restrictive rules so that they can be paid to not enforce them. Further, the more corrupt acts occur, the less stigma will attach to them, which leads to more people engaging in corruption. Finally, the more officials are corrupt, the fewer honest individuals will be attracted to work in the public sector.

According to Johann Graf Lambsdorff (Reference Lambsdorff2007: xi, Chs. 3–4), the core to understanding why corruption is detrimental to economic development is that corrupt politicians and bureaucrats can no longer commit to honesty and thus be trusted. Such trust would be necessary for economic development because economic investment will be attracted when politicians and bureaucrats can credibly commit to policies that offer a reliable, stable framework for such investment.

Transferring insights from the public choice literature on rent-seeking to the debate about corruption, Toke Aidt (Reference Aidt2016: 149) argues that a crucial insight to understanding the social costs that come with corruption lies in the notion of “rent contestability.” Once officials have a steady income from receiving bribes, others will put real resources into trying to become an official who can collect such bribes – real resources that could be put to productive uses instead.

A Social Lubricant

Despite the pernicious long-term consequences of corruption, Michael Munger defends the view that “corrupt acts by individuals in corrupt systems are morally permissible in the ‘short run’” (Reference Munger2018: 159) when they occur within corrupt systems where corruption is widespread and accepted as “part of doing business” (Reference Munger2018: 180). This is because in deeply corrupt economic–political systems, individuals may only be able to run a business or buy a home if they bribe the relevant authorities. In such conditions, bribery and related acts of corruption serve as the “lubricant for social intercourse” (Reference Munger2018: 160, see also Leff Reference Leff1964, Tullock Reference Tullock1996). They lead to efficiency gains short term. Even a well-intentioned person with a noncorrupt character has reason to engage in acts of corruption under such circumstances and, Munger suggests, it is morally permissible to do so.

Mario Juarez-Garcia adds that the presence of corruption sometimes has epistemic value: That a law is frequently circumvented may indicate that it is a defective law that should be eliminated (Reference Juarez‐Garcia2025). The prohibition laws in the 1920s and early 1930s serve as an example. As Montesquieu famously points out, there are two kinds of corruption, “one when the people do not observe the laws, the other when they are corrupted by the laws; the latter is an incurable ill because it lies in the remedy itself” (Reference Cohler, Miller and Stone1748/1989: Ch. VI.12). Based on Juarez-Garcia’s point, one could try to argue that, when laws are corrupting because they are bad laws, it becomes morally permissible to perform corrupt acts to circumvent such laws.

But Munger also points out that widespread corruption means that the society will be stuck in a corrupt system, that the “benefits of corruption are ‘locked in’” by creating rents for corrupt officials, and that the long-term effects are detrimental. Corruption is thus always “vicious in the ‘long run’” (Reference Munger2018: 159). To explain, what distinguishes the short run from the long run, from an economic point of view, is that in the short run all production inputs are invariable; in other words, the system is to be taken as a given, such that individuals acting alone cannot change the norms of the system (Reference Munger2018: 160, 168).

How do these long-term effects affect the moral status of corruption? If corruption has detrimental long-term effects, does this not make it morally impermissible to engage in corruption? Well, the difficulty is that escaping a corrupt system involves a collective action problem (cf. Rothstein Reference Rothstein2011, Mungiu-Pippidi Reference Mungiu-Pippidi2013, Persson, Rothstein, and Teorell Reference 66Persson, Rothstein and Teorell2013). An individual who refrains from engaging in corruption does not make a difference, as long as others keep engaging in corruption. And every individual can reasonably assume that most others will keep engaging in corruption. Because each individual also understands the short-term gains of engaging in corruption, it seems rational for each individual to not restrain themselves. Under such circumstances, perhaps corruption indeed becomes morally permissible, if the personal gains are reasonable ones, and if there really is no other way. Generally speaking, it has proven difficult to make the argument that it is morally impermissible to engage in behavior that does not make a difference for solving a collective action problem (see Nefsky Reference Nefsky2019 for an overview).

One thing that is striking about Munger’s discussion, though, is that he seems to exclusively focus on the moral permissibility of paying bribes, not the moral permissibility of taking bribes. The officeholders who take the bribes could certainly offer their service without taking a bribe. They cannot point out that they would not be able to get a business started or buy a house without taking bribes. And it is only they who abuse office powers for personal gain, which is, as we argued earlier, pro tanto morally wrong. For that reason, taking bribes is much harder to morally justify than paying bribes.

Munger offers one argument that might explain why even taking bribes might sometimes be morally permissible: Sometimes, officeholders have invested a lot of time, money, and energy to get into office, and they do so in the expectation to be able to collect bribes:

Are these expectations legitimate enough to justify taking bribes? Given the long-term detrimental effects, I am skeptical.

One final remark: In cases where paying – and perhaps taking – bribes is permissible, given the circumstances, this can, but need not, involve dirty hands (see Section 3). It all depends on whether the wrong involved is severe enough.

Fighting Corruption

Given that corruption is pro tanto morally wrong and has pernicious social and economic effects in the long run, the remaining question is how to fight corruption. It is clear that one cannot rely on virtue alone. What strategies of institutional reform promise to help fight corruption? In the spirit of a knavish constitution (see Section 2), the most straightforward way to reduce corruption levels is to reduce the opportunities for corruption by shrinking the state. Susan Rose-Ackerman writes: “If the state has no authority to restrict exports or license businesses, no one will pay bribes in those areas. If a subsidy program is eliminated, the bribes that accompanied it will disappear as well” (Reference Rose-Ackerman and Palifka2016: 558). Yet probably not all state activities that are associated with corruption can be eliminated. One reason is that shrinking the state will face resistance from those working in the state apparatus. Another reason – emphasized by Rose-Ackerman – is that unless you are an anarchist, you will think that it is not desirable to eliminate all state activities, because at least some serve purposes that cannot easily be privatized (Reference Rose-Ackerman, Rowley and Schneider2016: 556–558).

So what else can be done (for the following, see Rose-Ackerman Reference Rose-Ackerman, Rowley and Schneider2016: 558–564; see also Rose-Ackerman and Palifka Reference Rose-Ackerman and Palifka2016: Chs. 4–6, Klitgaard Reference 64Klitgaard1988: Chs. 3–4)? One thing is to implement and credibly enforce anticorruption laws with clearly defined rules that determine what corruption is. For this to be successful, one has to be able to rely on a noncorrupt judiciary, though. Another strategy is to reduce official discretion by simplifying the laws and regulations in the context of which corruption occurs. Introducing competitive pressures within government agencies by creating overlapping bureaucratic jurisdictions can help as well: If one can apply for licenses or permits at different places or approach different officials, officials who ask for bribes can be avoided. Similarly, overlapping enforcement areas for police officers can help to reduce bribes that are paid to avoid tickets or penalties. Offering better pay in the public sector reduces the incentive for corrupt behavior, and adequate pay will ensure that not only people who are willing to accept bribes are attracted to work in the public sector in the first place. Finally, increasing public accountability and transparency increases the risks of engaging in corruption: A free press, anti-corruption NGOs, and the protection of whistleblowers can help to keep corruption in check.

As a general approach to reform, Johann Graf Lambsdorff suggests to make those willing to take bribes unacceptable for public positions, to encourage betrayal among corrupt parties, to make corrupt contracts legally unenforceable, to hinder the operation of corrupt intermediaries, and to find clearer ways of regulating conflicts of interests (Reference Lambsdorff2007: 229). He points out that “combating corruption is like judo. Instead of bluntly resisting the criminal forces, one must redirect the enemy’s energy to his own decay” (Reference Lambsdorff2007: 234–235).

Institutional reforms are important to create the right incentives, but, in the end, there comes a point at which one has to rely on the character of individuals; one has to trust that individuals do not want to be corrupt. Not everything can be monitored, and not everything can be encoded in codes of conducts. As Elizabeth David-Barrett and Mark Philp put it: “It is clear that making it easier for people to do the right thing is good, but we should not think that there is a design that avoids the necessity of people having some virtue” (Reference David-Barrett, Philp, Hall and Sabl2022: 187, cf. 184). The virtue that is at stake here is a relatively minimal one: The willingness to not abuse one’s powers of office for personal gain, that is, not to do what is pro tanto morally wrong.

As we saw earlier, when corruption is systemic, escaping corruption becomes a collective action problem, though, which explains why anti-corruption policies are so often unsuccessful. As Anna Persson, Bo Rothstein, and Jan Teorell point out, “in a context in which corruption is the expected behavior, monitoring devices and punishment regimes should be largely ineffective since there will simply be no actors that have an incentive to hold corrupt officials accountable” (Reference 66Persson, Rothstein and Teorell2013: 457). In such circumstances, incremental change will not do; what is needed, Bo Rothstein argues, is a “big bang”:

Summary

We should distinguish corrupt acts from the corruption of institutions and the corruption of a person’s character. Corrupt acts involve an abuse of powers of office for personal gain. As such, they are pro tanto morally wrong. When corrupt behavior is widespread, it has pernicious long-term social and economic consequences.

On the other hand, in highly corrupt political–economic systems, corruption is the social lubricant to get things done, and because it is a collective action problem to escape such systems, it may sometimes be morally permissible to perform corrupt acts, at least for those who pay bribes; it is more doubtful whether it can be morally permissible to take bribes.

Shrinking the opportunities for corruption by shrinking the state is arguably the most straightforward way to fight corruption, but there are other ways to shift incentives the right way. At some point, one has to rely on politicians’ virtue, though. In circumstances of systemic corruption, a “big bang” may be needed to overcome the collective action problem of escaping corruption.

The Platonic Picture

One can read Plato’s Republic as an exercise in political ethics. Socrates explains that the guardians of the city must be brave and that the rulers of the city must be wise. To become wise, to-be-rulers go on a long educational journey, until…

There is something inspiring about Plato’s vision: Of course, one might say, politics should help everyone live good lives, and so those who have the relevant insight and character should be the ones who rule. On this picture, there is no real place for compromise in political ethics: Political ethics will tell those who are not wise to either become wise or to acquiesce to the rule of the wise. Compromises only become necessary under the unfortunate, nonideal circumstances that some who hold power are not wise and refuse to give up their power.

Interestingly, the early John Rawls upholds a somewhat Platonic picture of the well-ordered society as well, when he writes that, in a well-ordered society, everyone agrees on justice (Reference Rawls1971: 5). If we take that seriously, then political ethics will be quite simple when it comes to making compromises: Those who adhere to the wrong theory of justice should educate themselves and adopt the correct theory of justice, so that the relevant disagreements disappear. Once they do that, compromises will no longer be necessary. (Disagreements will be limited to disagreements about empirical facts, which one should be able to adjudicate.)

Disagreement and Legitimate Partisanship

I think political ethics should not adopt the Platonic picture. Rather, if political ethics wants to be relevant for real politics, it should accept, as a starting point, that a variety of politicians and political parties each legitimately try to implement their different visions of political morality. Political ethics should have something to say about how politicians ought to be and behave under circumstances where they disagree about political morality and cannot be expected to change their minds (see Section 1).

Different theorists will offer different reasons for why political ethics should take disagreement on political matters as a given and regard a variety of political parties as legitimate. Some political realists and agonists hold that the contestation and confrontation of antagonistic interests is at the very heart of politics, and that it is neither desirable nor possible to displace political contestation with an overlapping consensus on justice or anything of that sort (Honig Reference Honig1993, Newey Reference Newey2001, Mouffe Reference Mouffe2005). Some theorists make metaethical claims about the plurality and incommensurability of values to explain the unavoidability of conflict and disagreement (Stocker Reference Stocker1990, Gray Reference Gray2000, Bellamy Reference Bellamy1999). Some point out how everyone’s political convictions necessarily are a product of “previous historical conditions, and an obscure mixture of beliefs (many incompatible with one another), passions, interests, and so forth” (Williams Reference Williams2005: 12–13).

While political realists and agonists are skeptical about the value of philosophical theorizing about justice, I would like to emphasize that one can both acknowledge the realities of conflict and disagreement about political morality, including justice, and also think that it is an important and worthwhile endeavor to theorize justice. To illustrate, let me briefly lay out how nonrealist philosophers can come to acknowledge the realities of conflict and disagreement about political morality, including justice.

John Rawls, in his later works, makes clear that there is a plurality of reasonable conceptions of justice and that we cannot aspire to have a society in which everyone agrees on justice (Reference Rawls1996: xlviii, xlix, 223, 226; see also Gaus Reference 62Gaus2011: 2, 277–278, 445, 548). That we disagree in politics is not just a fact; we reasonably disagree, says Rawls, and he supports this with epistemic considerations (Reference Rawls1996: 54–58): Given the human condition – we find ourselves in a complex world populated with individuals with limited cognitive capacities – disagreement about justice and political morality is to be expected. Only oppression could eliminate it. Although Rawls does not spell this out much more, his take on reasonable disagreement leaves space for a variety of legitimate political parties to play an important role in politics (see Bonotti Reference Bonotti2017). But Rawls never stopped to endorse and defend his own conception of “justice as fairness,” and there is nothing contradictory about that.

Steven Wall offers a very different line of argument. He shows that those who are right about justice (and political morality more generally) have reason to support what he calls “constitutional settlements” (Reference Wall2013: 488). Constitutional settlements are “complex on-going social practices that both express certain values to which political societies are committed and establish procedures for resolving disputes among members of these societies” (Reference Wall2013: 483). Once a constitutional settlement is achieved, a sufficient number of people – who disagree about the content of political morality – come to converge on the judgment that they must honor the terms of the settlement (Reference Wall2013: 487). Even though the settlement is only a second best, from the point of view of justice, those who are right about justice have reason to support it: “A constitutional settlement is an achievement. It allows groups of people, who disagree over justice, to live together peacefully and productively” (Reference Wall2013: 490). To be sufficiently stable, though, everyone, including those who are right about justice, has to attach value to non-subjugation in politics (Reference Wall2013: 489, 493–494). What does this mean for political ethics? One could argue that, because of the value of non-subjugation that is embodied in constitutional settlements and that we have reason to value, a variety of political agents and political parties have to count as legitimate, so that we cannot expect them to change their minds, and that, for that reason, we should bracket who is right about political morality when we engage in political ethics, in particular the morality of making compromises (see Wendt Reference Wendt2025b).

No matter which of these lines of thought you find most convincing, in the end they all converge on our background assumption: That a variety of political agents and political parties legitimately try to implement their different political agendas. This does not hold for all political agents and political parties, though. While I will not be able to discuss where to draw the line, those who pursue abhorrent or deeply unjust political agendas and/or do not respect basic democratic norms and values will not count as legitimate parties. They will not count as players in the political arena like other players. With regard to them, we will have to take a Platonic stance: Political ethics will simply tell them to educate themselves and change their agenda.

How to Analyze Compromises

Politicians make compromises when they enter coalition governments, compromises are made in international politics, and even within political parties, politicians make compromises when they settle on the party agenda. As Edmund Burke remarks, “[a]ll government […] is founded on compromise and barter” (Reference Burke and Rhys1775/1908: 130–131; for a conceptual history of compromise, see Fumurescu Reference Fumurescu2013).

It is helpful to think of compromises in terms of two levels of evaluation (cf. Kuflik Reference Kuflik, Pennock and Chapman1979: 51, May Reference May2005, Wendt Reference Wendt2016: Ch. 3). The parties to a compromise disagree about the merits of different feasible proposals on the level of what can be called the fundamental evaluation. One party thinks that proposal A would be best, proposal B second best, and proposal C third best, while the other party thinks that it is the other way around. On the level of the fundamental evaluation of the proposals, the parties consider everything that speaks for and against the different proposals: Which of the proposals would be the most just, for example, but also which proposal seems commendable from the point of view of their personal interests. Then there is the proposals’ evaluation under circumstances of disagreement about the fundamental evaluation of the proposals. At this level, the parties consider everything that matters given the fact that they disagree about the fundamental evaluation of the proposals. In the end, the parties might possibly come to an overall ranking of the proposals that is different from their ranking based on their fundamental evaluation. But they do not change their views about their respective fundamental evaluations of the proposals. If both parties make concessions and end up agreeing on proposal B, they both accept it as a second best from the perspective of their fundamental evaluation of the proposals.

As Chiara Lepora and Robert Goodin point out, compromises can take many different forms (Reference Lepora and Goodin2013: 20–23, see also Carens Reference Carens, Pennock and Chapman1979: 126–127, Weinstock Reference Weinstock2013: 540). Sometimes the parties may be able to set aside whatever it is they disagree about and instead implement something completely different that they both find acceptable. Lepora and Goodin speak of a “substitution compromise.” Sometimes there is some overlap between what the parties desire, then they may be able to get to an “intersection compromise” in which they implement whatever it is they overlap on. And sometimes, when there is no overlap and no substitution available, they may find a “conjunction compromise” in which some elements of what one party wants and some elements of what the other party wants are implemented.

Making Compromises to Reach One’s Goals

In the following, I will discuss the reasons to make compromises in politics. Let me start with two preliminary remarks. First, talk of reasons is different from talk of duties or obligations. It leaves open whether these reasons merely speak in favor of making compromise, or whether they are conclusive – and morally stringent enough – to take the form of a moral duty or obligation. Second, reasons to compromise come into play in the evaluation under circumstances of disagreement about the fundamental evaluation of the proposals – which is the level the morality of making a compromise really is concerned with. Reasons to compromise are reasons to accept a proposal that is a second best from the perspective of the party’s fundamental evaluation. Or, to put it differently, they are reasons to accept a proposal that is not the proposal which the party would convince the other party of, if it could.

So what reasons are there to make compromises in politics? First and foremost, politicians, in many situations, have pragmatic reasons to make compromises for the simple reason that it allows them to better advance their goals than by refusing to do so. More precisely, a party has a pragmatic reason to compromise when it cannot act unilaterally, and the policy proposal it deems fundamentally best cannot be put into practice, given that the party with whom it could cooperate to get something done disagrees about the fundamental evaluation of the proposals. (The proposal it deems fundamentally best would be feasible, though, if the other party could be convinced of its fundamental merits.) Without making compromises, political parties are not able to participate in coalition governments, political leaders are not able to get to agreements on international affairs, and politicians are not able to influence their party’s agenda.

The Intrinsic Desirability of Making Compromises

When a politician or a political party has a principled reason to compromise, they have reason to compromise not because doing so advances their goals, but because there is something intrinsically desirable about making a compromise. When politicians or political parties have a principled reason to compromise, they have reason to compromise even though they would better advance their goals by not compromising. A political party might, for example, invite other political parties into a coalition government even though it would be able to govern alone.

Simon May argues that there are no principled reasons to compromise in politics (Reference May2005). Whenever it looks as if there are principled reasons to compromise, this is because people have reason to modify their position, what I called their fundamental evaluation of the proposals. But reasons to modify one’s position are not reasons to make compromises. Other philosophers have quite convincingly argued that there are principled reasons to make compromises in politics, though. In the following, I present four principled reasons that have been put forward.

First, some draw a connection between compromises and democratic values. Daniel Weinstock points out that democratic respect and inclusion are never perfectly embodied in real-world democratic institutions, and that we should therefore correct for these imperfections by making compromises that integrate the concerns of minorities (Reference Weinstock2013: 548–551). Christian Rostbøll argues that we treat other members of the political community as co-rulers when making compromises with them, and so compromises allow us to at least approximate the ideal of collective self-determination (Reference Rostbøll2017: 628–630, see also Kelsen Reference Kelsen, Urbinati and Invernizzi Accetti1920/2013: 76).

With regard to Weinstock’s argument, Jonathan White and Lea Ypi suggest that democratic respect and inclusion may more properly be expressed by adherence to democratic procedures rather than finding compromises on substantive matters (Reference White and Ypi2016: 160–162). I am not sure: If democratic procedures are and can be expected to remain imperfect and tend to exclude some voices, then compromises that include these voices still seem to help with democratic respect and inclusion.

With regard to Rostbøll’s argument, one could point out that only a compromise that includes all legitimate political viewpoints would really give us the ideal of collective self-determination: Typical compromises in democratic party politics do not include parties that do not participate in the coalition government. One could try to make the case that at least some countries at least sometimes come close to making the type of compromises Rostbøll’s argument supports, though: In Switzerland, the Federal Council constitutes something like a permanent grand coalition government, and in Denmark, there regularly are minority governments that make policy based on shifting coalitions.

More generally, though, one can also highlight some tensions between democratic values and compromises in politics: Compromises typically come in the form of untransparent backdoor deals that are in tension with ideals of publicity and transparency, although much depends on how these ideals are understood (cf. Baume and Novak Reference Baume, Novak, Baume and Novak2020 for discussion): One may well justify secret negotiations and deliberations, at least for some circumstances in politics (cf. Chambers Reference Chambers2004). Compromises in democratic politics are also in tension with majority rule and related democratic ideas: As Patrick Overeem puts it, the “need to give majorities their say while also protecting minorities, to come to quick and clear decisions while keeping wide support, to take the people as a unified whole and still recognize its plurality” are conflicting democratic aspirations (Reference Baume, Novak, Baume and Novak2020: 54, cf. 65). I will leave open here whether these tensions ultimately undermine the democracy-based case for principled reasons to compromise or not.

Second, Élise Rouméas argues that the process of give and take that characterizes compromise negotiations embodies reciprocity, which expresses “an ethos of mutual concern and the willingness to take seriously the other’s perspective and judgement” (Reference Rouméas2021: 388, see also Bellamy Reference Bellamy1999: 101, 110–111, Reference Bellamy2012: 457, Tully Reference Tully2000: 474–475, Wall Reference Wall2019: 482–483).

As Manon Westphal rightly points out, though, not all negotiators always take the other side’s perspective and judgment seriously, and so not all compromise negotiations will be properly reciprocal (Reference Westphal, Rostbøll and Scavenius2017: 88–91): Some compromise negotiations involve hard-nosed, power-driven bargaining, and some theorists – including Rouméas – accordingly distinguish between proper compromises on the one hand and mere strategic bargaining on the other hand. Westphal suggests that the institutional framework in which compromise negotiations are embedded could be designed in a way that promotes reciprocity in the negotiations, for example, by strengthening the bargaining power of weaker parties through veto powers and by providing other incentives to make concessions (Reference Westphal, Rostbøll and Scavenius2017: 91–92). It should be noted, though, that if one conceives of reciprocity as an ethos, institutional mechanisms cannot by themselves realize reciprocity. Others defend political bargaining in democracies under conditions of asymmetric power, at least as long as it is constrained in a way that makes sure that interests are treated equitably (Schwartzberg and Knight Reference Schwartzberg and Knight2024: 8–9).

Third, I have argued elsewhere that there sometimes are principled reasons to make compromises because this leads to substantive outcomes that are desirable under circumstances of reasonable disagreement. More specifically, legislators should sometimes make compromises and enact laws they deem a mere second best in order to have laws that are publicly justifiable, that is, acceptable from the perspectives of all legitimate parties (Wendt Reference Wendt2016: Ch. 14). Relatedly, there is value in having widely accepted laws, and making compromises can be necessary to get widely accepted laws (Wendt Reference 69Wendt, Rostbøll and Scavenius2017). To clarify, note that being acceptable and being accepted are not the same thing, and so the ideal of public justifiability and the ideal of wide acceptance of laws are not identical. But having publicly justifiable laws will make it more likely to get widely accepted laws, which is one reason – albeit not the only reason – why public justifiability matters.

Fourth, Klemens Kappel argues that politicians sometimes have principled epistemic reasons to make compromises (Reference Weinstock2018: 83–89; cf. also Weinstock Reference Weinstock2013: 545–548, Zuolo and Bistagnino Reference Zuolo and Bistagnino2018, and cf. May Reference May, Baghramian, Carter and Cosker-Rowland2024 for a reply). In cases of moral peer disagreement, we have reason to lower our confidence in our own moral position and to give greater weight to reasons to compromise (without changing the substance of our initial position). Strictly speaking, then, there are no epistemic concerns that would themselves constitute reasons to compromise, but there are epistemic concerns that constitute reasons to lower confidence in our moral view, and this in effect strengthens reasons to compromise. In other words, there sometimes is an epistemic source for why we should compromise.

A Compromising Mindset

I presented four arguments for the existence of principled reasons to make compromises in politics. Ultimately, they can all, in one way or another, be traced back to reasonable disagreement in politics. Whether such principled reasons to compromise are conclusive – making it morally permissible or even morally required to compromise or to accept a particular proposal that is on the table – will depend on the specifics of the situation and on what other moral considerations are at stake.

Insofar as political ethics is not only to specify what is morally permissible, morally impermissible, and morally required in politics, but also what is politically virtuous, the existence of principled reasons to compromise suggests that it is a political virtue to have a “compromising mindset” (Gutmann and Thompson Reference Gutmann and Thompson2012: Ch. 3). Politicians should be willing and ready to make compromises, when appropriate, at least when they interact with other parties that count as legitimate. Russell Muirhead and Nancy Rosenblum claim that a willingness to compromise is part of what it means to be a good partisan: To “insist on winning the whole truth for politics is ultimately an invitation to violence. And the rejection of violence is the cardinal virtue of partisans. Partisans at their best are not purists; they are open to compromise” (Reference Muirhead, Rosenblum, Hall and Sabl2022: 138, see also Dovi Reference Dovi2006: 164, Weinstock Reference Weinstock2018: 240).

If we regard a compromising mindset as a political virtue, this will also matter for the evaluation of democratic practices. Amy Gutmann and Dennis Thompson argue that campaigning undermines the spirit of compromise (Reference Gutmann and Thompson2012). And indeed, it is hard to deny that democratic campaigns can be divisive and polarizing, and it seems desirable for contestants for public offices to adhere to norms of respect that counter that tendency. As has often been observed, the United States and many European countries have become significantly more polarized and divided in recent years. On a positive note, Jennifer Wolak has offered some evidence that people’s attitudes toward compromise are challenged but not eroded by affective polarization (Reference Wolak2020: Ch. 5).

Fair Compromises

Should compromises be fair? Fairness considerations can arise at several points in the context of making compromises (see Jones and O’Flynn Reference Jones and O’Flynn2013, Ruser and Machin Reference Ruser and Machin2017: Ch. 1, Wendt Reference Wendt2019): There can be fairness concerns about inequalities of power among the parties, about the fairness of the negotiation procedure, and about the fairness of the actual compromise proposals that are on the table.

I will here focus on the latter, that is, the fairness of the compromise proposals themselves, not the fairness of the negotiation procedures or the initial negotiating position of the parties. There are two ways in which the content of a compromise proposal can be unfair. First, it can be substantively unfair. For example, the content of the proposal may include a law that disadvantages some minority group. This unfairness is not specific to the proposal’s character as a compromise. The proposal would be substantively unfair in the exact same way if everyone agreed that passing the law is the best thing to do, and no concessions were made by anyone at all.

Second, the content of a proposal can be unfair qua compromise. What makes compromises fair qua compromises is best captured by David Gauthier’s “principle of minimax relative concession” (Reference Gauthier1986: 133–146). Gauthier conceives the bargaining process in two stages: On the first stage, both parties propose the arrangement within the feasible range that gives them their maximum utility. That is their initial claim. Since they cannot both get what they want, they have to make concessions, which happens on the second stage. This is where the principle of minimax relative concession comes in. The parties follow the principle of minimax relative concession when the relative magnitude of their concessions is equal. The absolute magnitude of a concession is the difference in utility between the initial claim and the compromise the party is now supposed to accept. The absolute magnitude of a “complete concession” is the difference in utility between the party’s initial claim and the initial bargaining position. The relative magnitude of a concession is the proportion between these two utility differences (and so it can be determined without interpersonal comparisons of utility). When the parties follow the principle of minimax relative concessions on the second stage, they end up with a compromise arrangement that requires the smallest possible relative concession from both parties.

Even though this is an intuitive account of what fair compromises are, one may well doubt that it is intrinsically desirable to make compromises that are fair in this sense (Wendt Reference Wendt2019: 2863–2870). Think of the infamous three-fifths compromise from 1787, according to which three-fifths of each state’s slave population would count for determining the number of seats in the House of Representatives. Perhaps slaveholding states and free states made equal concessions in the compromise negotiations, but is this something to be celebrated? Clearly not. In response to this example, one may be inclined to claim that, perhaps, fair compromises are only desirable when both parties are to count as legitimate parties. But think of an ordinary compromise between the buyer and the seller of a used car: If one of them is a better negotiator and manages to get a deal in which he or she made fewer concessions, is that problematic? I do not think so. Perhaps, then, fair compromises are only desirable when both parties are legitimate and have principled reasons to compromise? I am still skeptical: When the parties have principled reasons to compromise, they arguably do have reason not to drive the hardest bargain possible. But this does not mean that they cannot make fewer concessions than the other party, as long as the proposal that is agreed upon is one that satisfies whatever the principled reasons to compromise demand: As long as the compromise leads to democratic inclusion, publicly justifiability, and so on, it will be fine. And for this to be the case, concessions do not have to be equal.

Élise Rouméas argues that concessions must meet some proportionality criterion if a compromise is to embody the value of reciprocity (see above). But she rightly emphasizes that a “compromise need not be about strict equivalency between concessions made, as the expression ‘splitting the difference’ seems to entail” (Reference Rouméas2021: 386, see also Weinstock Reference Weinstock, Rostbøll and Scavenius2017: 73–76). The “specific rule of proportionality that reciprocity-in-compromise demands ought to remain largely undefined” because specifying it would “betray the very nature of the social practice” (Reference Rouméas2021: 385). This seems plausible, at least under circumstances where politicians have principled reason to compromise (on proportionality, cf. Brighouse and Fleurbaey Reference Brighouse and Fleurbaey2010, Zanetti Reference Zanetti2020, Spang Reference Spang2021).

The demand of proportionality is in tension with the role obligations politicians owe to their voters, though, argues Edward Hall. The ethos of reciprocity makes it intrinsically desirable for politicians to negotiate compromises in a cooperative spirit that takes the other side’s perspective and judgment seriously and to make proportional concessions. But as representatives of a particular group of voters, they owe it to these voters to be tough in compromise negotiations and to concede as little as possible Reference Hall(2022b: 226–230). According to Hall, this tension inevitably leads to dirty hands: Either they violate obligations toward the voters they represent, or they violate the ethos of reciprocity. With Thomas Christiano, one could add that good representatives only make compromises with regard to political means, but not with regard to political ends, because ends are to be set by voters (Reference Christiano1996: 218). If that is right, then whenever politicians make compromises with regard to ends, perhaps in the spirit of the ethos of reciprocity, they violate a pro tanto obligation owed to their voters. I cannot offer a deeper discussion of the ethics of political representation here. But I would like to note that compromises in which politicians violate obligations owed to the voters they represent will only amount to dirty hands cases if these violations are severe enough (see Section 3).

Compromising Integrity

While there often are good reasons to compromise, there often also are good reasons not to compromise. Like reasons to compromise, reasons not to compromise do not apply in the fundamental evaluation of compromise proposals, but in the evaluation under circumstances of disagreement about the fundamental evaluation (see above).

Such reasons can be pragmatic. If one can better achieve one’s goals by refusing to compromise, then one has good pragmatic reasons not to compromise. In politics, there are also good pragmatic reasons to try to preserve one’s ideological identity, and thus to avoid making concessions: In multiparty systems, parties need to protect their ideological niche by distancing themselves from other parties, so that voters are able to understand what each party stands for (see Downs Reference Downs1957: 143–144). This can also be regarded as beneficial for the culture of a democracy, since it promotes healthy competition among distinct political visions. As Alexander Ruser and Amanda Machin put it: “The political realm is enlivened by the incompatibility and clear distinctiveness between political positions. Is it not important for some positions to remain steadfast and uncompromising?” (Reference Ruser and Machin2017: 40, see 34).

Are there also principled reasons not to compromise in politics? I think so, and the most prominent principled reasons not to compromise stem from the value of integrity. On Bernard Williams’s influential account, persons have integrity insofar as they act in line with ground projects they identify with (Reference Williams and Rorty1976). For the context of politics, Edward Hall argues that politicians with integrity display fidelity to principles or ends that are “constitutive of their identity as political agents” (Reference Hall, Hall and Sabl2022a: 76). These principles or ends may be what their voters authorized them to pursue, but they may also be principles or ends they sincerely endorse independently from such authorization (Reference Hall, Hall and Sabl2022a: 78–79).

As Cheshire Calhoun puts it, persons with integrity “stand for something” and show this in their relations to others (Reference Calhoun1995). Arguably, there are some constraints on what kind of principles and values politicians can stand for in a way that matters for their integrity, though (Hall Reference Hall, Hall and Sabl2022a: 79, but see Mendus Reference Mendus2009: 18). Firm commitment to Nazi or Stalinist values, for example, might, as a matter of fact, be what a politician stands for, what constitutes their identity as a political agent; but we probably would not describe compromises on such values as detrimental to the Nazi or Stalinist politician’s integrity. We only assess parties that can count as legitimate parties (see above) as lacking or displaying integrity.

From the perspective of an agent’s or political party’s integrity, compromises always seem worrisome (Ruser and Machin Reference Ruser and Machin2017: 30, White and Ypi Reference White and Ypi2016: 149, 155–156 163, Wendt Reference Wendt2023: 1033, Wendt Reference Wendt, Horn, Kolev and Müller2025a: 109). To make compromises means to share responsibility for actions and outcomes that one cannot fully endorse from the point of view of one’s own political outlook (cf. Lepora and Goodin Reference Lepora and Goodin2013: 23–29). Of course, not all compromises are the same: Whether and to what extent an agent’s or a political party’s integrity is affected depends on the distance of a compromise proposal’s content from their own position, and it also depends on whether the rationale behind a proposal is at odds with what they stand for (Wendt Reference Wendt, Horn, Kolev and Müller2025a: 111–115).

One may wonder, though, how the value of integrity can generate principled reasons not to compromise, if on the other hand virtuous politicians have a “compromising mindset.” And, indeed, some have argued that making compromises does not necessarily undermine a politician’s integrity, properly understood (Benjamin Reference Benjamin1990: Ch. 6, Tillyris Reference Tillyris2017). In reply, it is true that some compromises do not undermine a politician’s or political party’s integrity. The content of the compromise may not be that far from their own position. But other compromises do undermine integrity. Second, and more importantly, there is nothing contradictory in saying that the value of integrity and the virtue of a compromising mindset sometimes pull in opposite directions, that there sometimes are both principled reasons to compromise and principled reasons not to compromise. Good politicians must have the judgment that allows them to navigate such conflicts.

What alleviates the conflict is that integrity is not the kind of consideration that would make it impermissible to compromise; what it does is that it sometimes makes it permissible not to compromise (see Wendt Reference Wendt2024: 259–260). With regard to political parties, I argued elsewhere that there is an “integrity prerogative” that makes it morally permissible for them to refuse to compromise under many circumstances (Wendt Reference Wendt2023: 1032–1036). My main example was the decision of the Free Democratic Party (FDP), the liberal party in Germany, to leave the coalition government negotiations with the Christian Democrats and the Green Party in 2017, which was, it seems to me, permissible and reasonable, even though it led to a good amount of public outrage. But that it is morally permissible for political parties to refuse to compromise does not necessarily mean that it is the best thing to do, all things considered, if there are principled and pragmatic reasons to compromise. And there are limits to the integrity prerogative. Sometimes, when there is no reasonable alternative for forming a government available, that is, when democratic stability is at stake, then a political party will be morally required to make the compromises that are necessary to be able to govern. Sometimes, one is morally required to compromise one’s integrity (cf. also Lepora and Goodin Reference Lepora and Goodin2013: 164–166, Sabl Reference Sabl2018: 273).

A brief sidenote: Ronald Dworkin has raised a very different integrity concern in relation to compromises in politics: Compromises, Dworkin argues, should not lead to incoherent “checkerboard laws” that undermine the integrity of the law (Reference Dworkin1986: 183–184): When matters of justice are at stake, we should make a “compromise about which scheme of justice to adopt, rather than [adopt] a compromised scheme of justice” (Dworkin Reference Dworkin1986: 179). Dworkin’s integrity concern does not – or not primarily – give politicians a reason not to compromise, but it offers guidance about how to compromise.

The Limits of Compromise

I claimed that political integrity is a value that makes it permissible not to compromise, sometimes, but not a value that could make it impermissible to compromise. I will now discuss the considerations that do make it impermissible to compromise. These are the “limits of compromise.”

The most prominent account of the limits of compromise is Avishai Margalit’s. Margalit uses the concept of “rotten compromises” to delineate the limits of compromise. Rotten compromises, he writes, are compromises that help maintain an inhuman regime, or made with persons that “undermine morality itself” (Reference Margalit2010: 61, 98, 21–23). An example for the latter is the Munich Agreement with Adolf Hitler in 1938. Rotten compromises “should be prohibited in all circumstances,” while “[o]ther compromises should be evaluated on their merit, case by case” (Reference Margalit2010: 3). In the end, though, Margalit seems to regard rotten compromises as permissible in exceptional circumstances where the consequences of not making them are even worse. He writes: “Compromises should never be allowed in cases of crimes against humanity, except to save the lives of the people threatened by such regimes” (Reference Margalit2010: 63, see also 89, 137).

Simon May’s account of the limits of compromise “identifies racial equality – and legitimate democratic government more broadly – as a value that may not be sacrificed for the sake of competing values, that is, values external to legitimacy” (Reference May, Rostbøll and Scavenius2017: 42). Racial equality and other demands of democratic legitimacy have strict priority: Compromises on racial equality that help to achieve a greater level of racial equality in the long term may be justifiable, but not compromises on racial equality for the sake of other values (Reference May, Rostbøll and Scavenius2017: 44). Once the core demands of democratic legitimacy are satisfied, there is hardly any room left for justifiable compromises on racial equality. In that sense, racial equality is a nonnegotiable value or demand (Reference May, Rostbøll and Scavenius2017: 35, 39, 43, 45, 46).

I think it is a mistake to single out any of these principles as specifying the limits of compromise (Wendt Reference Wendt2024, cf. Räikkä Reference Räikkä2026). Whatever moral principles there are that set limits on human action, they will sometimes make it morally impermissible to accept a compromise proposal, namely when the proposal involves a violation of the moral principle. This holds for the principles Margalit and May highlight, but it also holds for other moral principles, for example, moral principles that prohibit the intentional killing of innocents, lying to the public, or wasting tax money on inefficient policy.

It goes without saying that whether something is morally impermissible to do or not depends on the alternatives. As explained earlier, it is convenient to think of the parties to a compromise as having a ranking of the available proposals. Somewhere in the ranking is the cutoff point that separates permissible from impermissible compromise proposals. Moral principles like the ones discussed earlier typically put all proposals that violate them below that cutoff point – as long as there are other proposals available that do not violate them.

What is important to note is that whether a proposal helps to maintain an inhuman regime, violates racial equality, or violates some other moral principle that sets the limits of compromise is something that matters in the fundamental evaluation of the proposals, not in the evaluation under circumstances of disagreement about the fundamental evaluation (see above). The principles that determine the limits of compromise are not considerations that are specific to the morality of making compromises. They determine the fundamental moral status of proposals, no matter if there is disagreement about that status, and no matter if a compromise is being made or not. This is different from the value of integrity that I discussed earlier. The value of integrity becomes relevant when the proposal that is agreed upon (or could be agreed upon) deviates too much from what is best given an agent’s fundamental evaluation of the proposals: It is relevant in the evaluation under circumstances of disagreement about the fundamental evaluation (cf. Wendt Reference Wendt2024: 259).

Summary

Political ethics should start with the assumption that a variety of politicians and political parties legitimately try to implement their different visions of political morality, which will lead them to different fundamental evaluations of the available policy proposals. To accept a compromise means to accept a proposal that is not one’s first choice from the perspective of one’s fundamental evaluation of the proposals.

Political parties have not only pragmatic but also principled reasons to make compromises in politics. Principled reasons to compromise may stem from democratic values, the procedural value of reciprocity, substantive values like the value of public justifiability, or from epistemic considerations. Because there are principled reasons to compromise in politics, having a compromising mindset should be considered a political virtue. Even when there are conclusive principled reasons to compromise, there is nothing intrinsically desirable about making equal relative concessions, though; rather, some more vaguely determined proportionality in concessions is all politicians (sometimes) should aim for.

There are also pragmatic and principled reasons not to compromise in politics. Principled reasons not to compromise stem from the value of political integrity. The value of integrity sometimes makes it permissible not to compromise, but it does not make it impermissible to compromise. Other moral considerations – which operate at the fundamental evaluation of the proposals – sometimes do make it impermissible to compromise; in fact, any moral principle that can prohibit some type of human action can also make it impermissible to accept a compromise.