1.1 The Ideal and Non-Ideal Theory Debate

On its surface, the ideal and non-ideal theory debate is a methodological debate about how a certain sort of political philosophy should proceed. Suppose you wish to theorize about justice, understood not as a personal virtue but as a “virtue of social institutions” (Rawls, Reference 76Rawls1999b: 3). Should you work on developing an “ideal theory” of what the perfectly or ideally just society would be like? Or should you engage in “non-ideal theory,” trying to understand the injustices we face in the real world and how to address them?

Once we dig deeper into the debate, however, we find that much of what is at issue doesn’t seem so methodological after all. To be sure, some aspects of the debate do touch on general issues of philosophical methodology, for example, concerning the role of simplifying assumptions or the value of practically irrelevant knowledge. But the debate more often turns on questions like: What is the logical structure of principles of comparative justice? How much short-term justice should we sacrifice for the promise of greater long-term justice? And what moral constraints are there on the pursuit of justice? These do not seem like methodological questions about how to do political philosophy, but rather like substantive questions within political philosophy.

The ideal and non-ideal theory debate can therefore seem bewildering, and it is easy to lose track of what everyone is arguing about, how it all fits together, and what is at stake. It certainly doesn’t help that different contributors use the terms “ideal” and “non-ideal” in different ways and – when they try to bring some order to the chaos – offer competing typologies of the different ways that theorizing can be ideal or non-ideal (e.g., Hamlin and Stemplowska, Reference Hamlin and Stemplowska2012; Valentini, Reference Valentini2012; see also Adams, Reference Adams, Roberts-Cady and Mandle2020: 75; Gaus, Reference Gaus2016: ch. 2; Laurence, Reference Laurence2021: ch. 1; Volacu, Reference Volacu2018: 891–894). Nor does it help that contributors often seem to talk past each other, focusing on claims of different strengths, and so making it hard to tell where exactly their disagreement lies. For example, we will see that non-ideal theorists are often primarily concerned with arguing that ideal theory lacks any strong sort of “priority” over non-ideal theory, whereas ideal theorists often aim only to show that ideal theory is not altogether worthless – that it is valuable and worth doing, its lack of priority notwithstanding.

My goal in this Element is to clear up this bewilderment, providing an opinionated guide to the ideal and non-ideal theory debate in political philosophy. Given its sprawl, I cannot hope to provide a detailed examination of every issue raised under its heading. Instead, I aim to accomplish three things. First, I provide a bird’s-eye view of the debate, and how its seemingly disparate aspects fit together. Second, I focus in more detail on those issues that are distinctive to political philosophy, giving relatively short shrift to more general questions of philosophical methodology; I also prioritize topics bearing on the relative merits of ideal and non-ideal theory, rather than intramural disputes about how each sort of theory is best done. And third, I give some sense of where the debate is currently at and what conclusions we can draw. I should therefore note from the outset that I am not a neutral third party, but rather a participant who has been largely critical of ideal theory (Barrett Reference Barrett2020; Reference Barrett2023a; Reference Barrett2023b; Reference Barrett2025a). Indeed, by way of preview, my basic take on these issues is that there is nothing special about ideal theory, and it deserves no pride of place in our theorizing, but it is rather something we might sometimes find helpful along the way – in much the same way that working through the details of various other thought experiments might be. Still, having registered this bias (which would surely come through even if I tried to hide it), I do my best to give each side a fair shake, ultimately arguing that while ideal theory should not enjoy the central role in the discipline it once had, it should not be abandoned or marginalized either: We should hope it remains one active research program among many.

In the remainder of this introductory section, I clarify how I use the terms “ideal” and “non-ideal” theory and what major questions I am concerned with, using Rawls’s famous discussion of ideal theory as a starting point. I then explain my basic strategy and plan going forward. To preview: I adopt a minimal conception of ideal theory as “theory that aims to characterize ideal or perfect justice,” and proceed by first examining each of three practical roles ideal theory might play, before finally considering the extent to which political philosophers should do ideal theory. In the conclusion, I also briefly consider related debates beyond political philosophy.

1.2 Rawls’s Conception of Ideal Theory

My subject is the ideal and non-ideal theory debate as it stands, and not its history – which many trace to debates between Plato and Aristotle in the ancient world. For the most part, I also focus on high-level ideas and arguments, rather than on the nuanced positions of specific thinkers (while providing ample citations for those wishing to investigate such nuances). However, as the current debate has largely arisen in response to Rawls’s account of ideal theory, I make one exception and begin with a discussion of Rawls – a discussion that is intended to clear some ground so that, afterward, we can leave exegesis aside and focus on the issues directly. Indeed, Rawls’s account of ideal theory combines several elements, only one of which I will treat – for purposes of this work – as providing the core contrast between ideal and non-ideal theory. His reasons for doing ideal theory have also framed much of the debate, and so, too, will frame much of our discussion.

When Rawls first introduces the idea of ideal theory in A Theory of Justice, he says that its goal is to describe the abstract principles that would regulate a “perfectly just society” (Rawls, Reference 76Rawls1999b: 8) or a “well-ordered society under favorable circumstances” (Rawls, Reference 76Rawls1999b: 215). A society is “well ordered” when it is characterized by a public conception of justice specifying fair terms of social cooperation, which effectively regulates its major social and political institutions, and which (almost) all individuals accept and comply with (Rawls, Reference 76Rawls1999b: 4–5). A society is under “favorable circumstances” when it is not subject to severe economic, technological, or cultural hardships that might compromise its ability to realize such principles (Rawls, Reference 76Rawls1999b: 132, 215–216). Notably, however, favorable circumstances are meant to be realistic: When doing ideal theory, we hold fixed general facts of human life (for example, that people are motivated by material incentives) and rule out principles that, in light of such facts, could not be stable (Rawls, Reference 76Rawls1999b: 434–441). Given Rawls’s focus on domestic justice, he also makes certain further simplifying assumptions, for example, that society is a “closed system isolated from other societies” (Rawls, Reference 76Rawls1999b: 7). In his later work, The Law of Peoples, Rawls applies a similar methodology to the issue of global justice, where he makes different simplifying assumptions, and rebrands ideal theory as the search for a “realistic utopia” – “realistic” because, as above, we hold fixed general facts of human life, but “utopian” because we search for the best principles consistent with these facts (Rawls Reference Rawls1999a, 11–19; see also Förster, Reference Förster2016: 336–340). To keep things simple, however, I focus on domestic justice here – on ideal theory as a theory of the ideally just society rather than of the ideally just world.

Rawls’s conception of ideal theory, then, combines six claims:

• Ideal justice: Ideal theory aims to characterize ideal or perfect justice.

• Abstract principles: Ideal theory aims to characterize the abstract principles that would regulate the ideally just society’s major social and political institutions.

• Full compliance: Ideal theory evaluates candidate principles given the assumption that they are fully accepted and complied with.

• Favorable conditions: Ideal theory evaluates candidate principles given the assumption that they are operating in favorable economic, technological, and cultural background conditions.

• Realistic conditions: Ideal theory evaluates candidate principles given realistic background assumptions about human psychology and social life.

• Simplifying assumptions: Ideal theory evaluates candidate principles given further simplifying assumptions.

We’ll come back to these claims shortly, but for now I briefly explain how they relate. The first two set the task of ideal theory: Ideal justice says that this is to develop a theory of ideal justice, and abstract principles specifies that this involves coming up with the abstract principles that would ideally govern our social and political institutions (rather than coming up with, say, those concrete institutions themselves). The remaining claims each specify further details about how to perform this task, but they imply different constraints on our theorizing and are motivated in different ways. The role of full compliance is the most contested, with some seeing it as falling directly out of abstract principles: To evaluate principles of justice as such, we must evaluate them on the assumption that everyone follows them (e.g., Simmons, Reference Simmons2010: 8). However, since it is also possible to evaluate principles in other ways, for example, given the degree of compliance we reasonably expect them to command, I prefer to interpret full compliance as a perfection assumption, selected as a normatively attractive feature the ideally just society must satisfy: If a society fails to satisfy full compliance, it fails, for that reason, to be ideally just. Next, favorable conditions is a scope assumption, reflecting a limit to the applicability of ideal justice: Ideal justice may not be realizable in certain highly unfavorable conditions, but this shouldn’t inform our theorizing about what ideal justice requires in societies that do meet preconditions for realizing it. Realistic conditions is, by contrast, a feasibility assumption. Although we are to assume that, in the ideally just society, principles are fully complied with, we rule out principles that could not feasibly achieve or maintain full compliance because, for example, this conflicts with human nature. Finally, simplifying assumptions does not concern normatively attractive features of society, limits on the scope of ideal justice, or concessions to feasibility, but further assumptions that are meant to make the task of ideal theorizing easier without substantially undermining the reliability of our conclusions.

When discussing ideal theory, it is common to refer to it as theory that “idealizes” in some way or another. However, the term “idealization” is highly ambiguous, as are other terms it is sometimes contrasted with, such as “abstraction” – where, roughly, “idealizations” are thought to involve false assumptions but “abstractions” merely to leave details out (O’Neill, Reference O’Neill1987; see also Hancox-Li, Reference Hancox-Li2017). For this reason, I avoid such terms here, preferring to delineate different theoretical assumptions by our reasons for making them. Perfection assumptions embody normatively attractive features an ideally just society must meet; scope assumptions reflect limits on the applicability of ideal justice; feasibility assumptions reflect concessions to feasibility, insofar as we think ideal justice is limited by some notion of feasibility; and simplifying assumptions are meant to make our theoretical task more manageable. There is nothing wrong with calling any of these “idealizations,” but to avoid confusion I refrain from that terminology myself.

1.3 Rawls’s Defense of Ideal Theory

So that’s Rawls’s conception of ideal theory – but why does he think we should do it? Rawls emphasizes the following reason: Ideal theory is worth doing because it is practically relevant and, in particular, because it bears on more directly practically relevant forms of “non-ideal theory” that concern “more pressing problems” of injustice and, more generally, “how we are to deal with injustice” (Rawls, Reference 76Rawls1999b: 8). Indeed, Rawls makes two strong claims here. First, ideal theory is not only relevant but prior to non-ideal theory, in the sense that “[n]onideal theory… is worked out after an ideal conception of justice has been chosen” (Rawls, Reference 76Rawls1999b: 216). As he clarifies: “[t]he reason for beginning with ideal theory is that it provides… the only basis for the systematic grasp of these more pressing problems” (Rawls, Reference 76Rawls1999b: 8). Second, this is the only reason (or, depending on one’s interpretation, at least one very significant reason) that ideal theory is valuable: “If ideal theory is worthy of study, it must be because…it is the fundamental part of the theory of justice and essential for the nonideal part as well” (Rawls, Reference 76Rawls1999b: 343).

Setting aside, for a moment, the extent to which ideal theory’s value depends on its purported priority over non-ideal theory, why should we think it has such priority? Rawls suggests three reasons. First, a conception of ideal justice serves as a benchmark for evaluating non-ideal societies as more or less just: “[e]xisting institutions are to be judged in the light of this conception and held to be unjust to the extent that they depart from it without sufficient reason” (Rawls, Reference 76Rawls1999b: 216). Second, our conception of ideal justice serves as a long-term target or “aim to guide the course of social reform” (Rawls, Reference 76Rawls1999b: 215; the terms “benchmark” and “target” are from Wiens, Reference Wiens2015a). Third, while Rawls is less explicit about this, his discussion of civil disobedience suggests that our conception of ideal justice also serves as a gauge of how to appropriately or permissibly respond to injustice, for example, of which means we may permissibly take when resisting it (Rawls, Reference 76Rawls1999b: 8, 319–343).

We can thus once again highlight six Rawlsian claims, this time about the value of ideal theory:

• Ideal value: Ideal theory is valuable and worth doing.

• Practical relevance: Ideal theory’s value derives exclusively (or at least in large part) from its practical relevance.

• Priority: Ideal theory is practically relevant because it is prior to non-ideal theory.

• Ideal benchmark: A conception of ideal justice serves as a benchmark of comparative evaluation.

• Ideal target: A conception of ideal justice serves as a long-term target for reform.

• Ideal gauge: A conception of ideal justice serves as a gauge of which responses to injustice are appropriate or permissible.

This time, the connection between these claims is straightforward. Ideal value asserts that ideal theory is valuable, practical relevance says that this is because of its practical relevance, and priority clarifies that ideal theory’s practical relevance derives from its priority over non-ideal theory. Ideal benchmark, ideal target, and ideal gauge then articulate three ways ideal theory may be prior (or at least relevant) to non-ideal theory: It may provide a benchmark of evaluation, a target for reform, or a gauge of propriety.

1.4 A Minimal Conception of Ideal Theory

Although Rawls’s initial discussion of ideal theory appeared in 1971 (in the original edition of A Theory of Justice), the ideal and non-ideal theory debate only really got going in the mid-2000s, with pathbreaking critical articles from Mills (Reference Mills2005) and Sen (Reference Sen2006) – the first casting ideal theory as problematically ideological and recommending that we instead focus on understanding the injustices we currently face; the second casting ideal theory as practically irrelevant and recommending that we replace it with a comparative approach focused on how to make things more just. Since then, the literature has exploded, and almost every detail of Rawls’s conception and defense of ideal theory has been criticized.

Beginning with Rawls’s conception, ideal justice is typically taken as definitive of ideal theory: Ideal theory is theorizing about ideal or perfect justice. However, his other claims have generated much disagreement. Some worry that abstract principles involve too abstract a focus on principles governing institutions, rather than on concrete institutions themselves (e.g., Gaus, Reference Gaus2016: 19–21; Hamlin and Stemplowska, Reference Hamlin and Stemplowska2012: 53; Sirsch, Reference Sirsch2024; Wiens, Reference Wiens2012: 48–49), or else too narrow a focus on principles regulating institutions, rather than on other outcomes or aspects of social life (e.g., Cohen, Reference Cohen2008: ch. 3; Sen, Reference Sen2009: 17–24, ch. 10) – though I here put aside this latter complaint and use “institutions” as shorthand for “institutions (or whatever other features of society matter to justice).” Some criticize full compliance and favorable conditions as being too idealistic or misleading (e.g., Brennan and Pettit, Reference Brennan, Pettit, Jackson and Smith2005: 258; Enoch, Reference Enoch2018; Farrelly, Reference Farrelly2007; Laurence, Reference Laurence2021: ch. 5; Levy, Reference Levy2016; Mills, Reference Mills2005: 169; Reference Mills2017: 76–77; Schmidtz, Reference Schmidtz2011: 777–778; Reference Schmidtz and Olsaretti2018: 339–341; Reference Schmidtz2023: ch. 11; see also Larmore, Reference Larmore2020); others criticize realistic conditions as making too great a concession to unsavory features of human psychology (e.g., Cohen, Reference Cohen2003; Reference Cohen2008: ch. 6; Estlund, Reference Estlund2011a; Reference Estlund2020: chs. 5–7). Finally, many debate the particular assumptions Rawls groups under simplifying assumptions, or the role of simplifying assumptions in political philosophy more generally (often calling these, as above, “idealizations”) (e.g., Appiah, Reference Appiah2017: ch. 3; Farrelly, Reference Farrelly2007; Hancox-Li, Reference Hancox-Li2017; Ismael, Reference Ismael2016; Leader Maynard, Reference Leader Maynard2025; Mills, Reference Mills2005; Reference Mills2017: ch. 5; Monaghan, Reference Monaghan2022; Robeyns, Reference Robeyns2008: 352–360; Schmidtz, Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 4; Stemplowska, Reference Stemplowska2008: 326–329; Táíwò, Reference Táíwò and Okeja2023; Uberti, Reference Uberti2014; Valentini, Reference 79Valentini2009: 351–355; Reference Valentini, Weber and Vallier2017).

Likewise, when it comes to the value of ideal theory, all six of Rawls’s claims have been the subject of dispute. Against ideal value and practical relevance, many have protested that ideal theory is worthless or even harmful (e.g., Anderson, Reference Anderson2010: ch. 1; Brennan and Pettit, Reference Brennan, Pettit, Jackson and Smith2005; Farrelly, Reference Farrelly2007; Gaus, Reference Gaus2016; Mills, Reference Mills2005; Reference Mills2017: ch. 5; Sen, Reference Sen2006; Reference Sen2009: 15, 96–102; see also Popper Reference Popper2013), while others have argued that the value of ideal theory (or political philosophy more generally) depends on something other than its practical relevance (e.g., Cohen, Reference Cohen2003: 243; Reference Cohen2008: 268; Estlund, Reference Estlund2011b; Reference Estlund2014; Reference Estlund2020; Swift, Reference Swift2008: 366–368; van der Vossen, Reference van der Vossen2015; Reference van der Vossen2020). But perhaps no claim has been as heavily criticized as priority, and especially the suggestion that ideal theory achieves priority through ideal target or ideal benchmark – ideas discussed at length shortly (Sections 2–3). Although less ink has been spilled on the topic, disagreements over ideal gauge represent a more recent strand of this debate (Section 4).

One difficulty we face trying to adjudicate all this is that the first set of issues – what is ideal theory? – cannot be neatly separated from the latter – what is the value of ideal theory? For example, if we endorse practical relevance, then we should evaluate claims like full compliance and realistic conditions in terms of how well they vindicate ideal theory’s practical relevance – and this might push us, for example, to reject the first claim while accepting the second. However, if we deny practical relevance and hold instead, say, that ideal theory is valuable because it gives us deep knowledge of justice, this might lead us to accept full compliance as characterizing the concept of perfect justice and to reject realistic conditions as a concession to unjust features of human psychology.

Debates over the plausibility of different ways of establishing priority similarly turn on how we understand the goal of ideal theory. For example, full compliance and abstract principles may be plausible if we are concerned with ideal benchmark, since we want to evaluate societies’ justice relative to the perfect standard of individuals and institutions fully complying with certain principles. However, ideal target may suggest rejecting both claims: If the ideal is to serve as a long-term goal, we must have some sense of not only the principles it would satisfy but also their institutional realization, and it’s far from obvious that we should treat the lofty standard of full compliance as a goal. Different simplifying assumptions may also be warranted depending on the function we want ideal theory to play – if we treat it as a benchmark for evaluating domestic justice, then the “closed society” assumption may be apt, but only the most rabid isolationist would treat a closed society as a target for reform.

For these reasons, I employ the following strategy in this Element. Rather than adopting Rawls’s full-fledged conception of ideal theory – or, for that matter, focusing on criticisms of ideal theory that rely on details of Rawls’s own favored ideal theory of justice (e.g., Farrelly, Reference Farrelly2007; Kang, Reference Kang2016) – I adopt a minimal conception of ideal theory, which embraces only the first of Rawls’s six claims:

• Ideal justice: Ideal theory aims to characterize ideal or perfect justice.

So, I don’t assume that ideal theory must concern abstract principles, employ full compliance, or so on. Rather, I use this minimal characterization of ideal theory to evaluate our six claims about ideal theory’s value, along the way considering how varying other features of our understanding of ideal theory (for example, whether we focus on abstract principles or assume full compliance) affects the plausibility of these claims. As it turns out, this approach will allow us to cover and draw connections between nearly every central issue in the ideal theory debate, while also setting aside some that are largely peripheral to these other issues, despite sometimes being discussed under the same heading. For example, I will have relatively little to say about simplifying assumptions, which we will see is largely disconnected from the core debate over the merits of ideal theory.

My minimal conception of ideal theory might seem to imply that non-ideal theory involves any theorizing that does not aim to characterize ideal or perfect justice. However, I restrict my use of the term “non-ideal theory” to theorizing about justice, focusing specifically on three types of non-ideal theory, each corresponding to a different question we face when confronted with an unjust society in need of improvement. First, there is theorizing about comparative justice – about what makes societies more or less just. While ideal benchmark claims that ideal theory is prior to comparative theorizing, critics disagree. Second, there is theorizing about what goals to adopt when pursuing justice. While ideal target holds that we should treat our conception of the ideal as a target or long-term goal for reform, many critics defend what I have elsewhere called a “problem-solving” approach (Barrett, Reference Barrett2020), which comes up with goals by trying to identify, diagnose, and find solutions to concrete problems of injustice (e.g., Anderson Reference Anderson2010: ch. 1; Schmidtz, Reference Schmidtz2011; Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 10; Sen, Reference Sen2006; Reference Sen2009; Wiens, Reference Wiens2012; see also Mills, Reference Mills2005; Reference Mills2017: ch. 5). Finally, there is theorizing about appropriate or permissible attitudes or actions to take in response to injustice. Here, too, we will see that critics deny that ideal theory serves the role claimed by ideal gauge, instead holding that to work out appropriate or permissible responses to injustice, we must study such injustice directly.

In defining ideal and non-ideal theory in these ways, I have been careful not to beg any questions about the relation between them. This is why I depart from the common practice (at least in certain circles) of holding that whereas ideal theory sets the goal, non-ideal theory “asks how this long-term goal might be achieved, or worked toward, usually in gradual steps” (Rawls, Reference Rawls1999a: 89). This oft-quoted definition of non-ideal theory is unhelpful here, since it presupposes the truth of ideal target and so the priority of ideal theory. My own definitions leave this open: If ideal theory is theorizing about ideal justice, and non-ideal theory involves theorizing about comparative justice, goals for reform, or appropriate responses to injustice, then ideal theory may, but need not, be prior or relevant to non-ideal theory.

Two clarificatory notes before we go on. First, while I have said that I will understand ideal theory as theorizing about ideal or perfect justice, one can always ask: ideal given what constraints? In general, I assume that ideal justice involves the upper limit of justice – the principles or institutions that are “most just” – but this leaves it ambiguous whether we mean the feasible upper limit of justice (and, if so, how to understand feasibility) or the conceptual upper limit (as the term “perfect” justice suggests). Here, I deliberately retain this ambiguity in my statement of ideal justice, since we will see that different ways of resolving it may be more or less plausible depending on the function we want ideal theory to play. Still, it is worth flagging this from the outset to avoid confusing the issue by assuming, for example, that all forms of ideal theory ignore feasibility altogether.

Second, going forward, I will often simplify my discussion by referring to defenders of ideal theory as “ideal theorists” and critics of ideal theory as “non-ideal theorists.” This shorthand departs from the definitions just offered, but it should always be clear from context when I am using the terms this way.

1.5 The Road Ahead

Here is how the rest of the discussion will proceed.

In Section 2, I begin by considering whether ideal theory provides a helpful or necessary benchmark for evaluating non-ideal societies as more or less just. I argue that while ideal theorists have not made good on the claim that ideal theory is necessary for this purpose, it is not altogether irrelevant either. Ideal theory can play some role when theorizing about comparative justice – just not a uniquely or especially helpful one.

In Section 3, I consider whether ideal theory provides us with a long-term target for reform. I argue that while ideal theorists have also failed to vindicate this role, they do point us toward the importance of focusing on the long run, in a way some non-ideal theorists overlook. I also similarly suggest that ideal theory can play some role in helping us to come up with targets for reform, but that it doesn’t play any special or unique role in this regard – and, indeed, that we shouldn’t treat our conception of the ideal itself as a weighty goal.

In Section 4, I turn to the possibility that ideal theory provides a gauge of how to appropriately or permissibly respond to injustice, bearing on what responses to injustice are permissible, how to appropriately acknowledge its victims, or what attitudes it makes fitting. This is a less well-trodden part of the debate, but I tentatively argue that the basic message is the same. Ideal theory can play some role in helping us to gauge appropriate or permissible responses to injustice, but ideal theorists face a heavy burden if they wish to claim that it plays an especially important role here.

By this point, we will have adjudicated debates over the various ways ideal theory might be practically relevant or have priority over non-ideal theory. In light of these results, Section 5 turns to an evaluation of the role ideal theory should play in political philosophy. Here, I consider further arguments of three types – concerning the practical or nonpractical value of ideal theory, professional obligations, and ideology. Although these arguments prove inconclusive, I suggest two main takeaways. First, there are good reasons to prefer non-ideal theory to ideal theory, and so to welcome political philosophy’s ongoing shift in that direction. But second, there are also good reasons to embrace methodological pluralism and so to prevent this shift from going too far. Ideal theory should not enjoy the central role in the discipline it once had, but it should remain one active research program among many.

Section 6 sums up our conclusions about the value of ideal theory and relates them back to Rawls’s other claims about what ideal theory is (for example, that it assumes full compliance). It then briefly considers the extent to which lessons from this debate generalize beyond justice or even political philosophy, examining connections with related debates (often going under the same heading) in other fields.

2.1 A Benchmark of Evaluation

There’s an old story that Charlie Chaplin once lost a Charlie Chaplin look-alike contest. I’m not sure if the story is true, but I like it, because it’s funny. The whole point of a Chaplin look-alike contest is to look like Chaplin. Judges should evaluate contenders by how far they deviate, in appearance, from Chaplin himself. But of course, Chaplin deviates not at all from himself. He is the ideal contender. That’s why it’s funny for him to lose.

The criterion judges should use in the look-alike contest has a distinctive structure. I call it an “ideal deviation” structure (Barrett, Reference Barrett2023b). Suppose that x and y are two contenders in the contest. Then, judges should evaluate contenders according to the following comparative criterion:

This criterion specifies one contender, Chaplin, as ideal or best. It then evaluates other contenders by how far they deviate from this ideal. An interesting feature of this criterion is that we can’t apply it except by reference to the ideal candidate. A judge who had never seen Chaplin would have no idea how to rank other candidates. The best way to figure this out would be to discover what Chaplin looks like. And judges can only explain their rankings of other contenders by reference to Chaplin himself.

Might justice be, in this respect, like a Charlie Chaplin look-alike contest? Suppose now that x and y are two societies, and consider the following comparative criterion:

If justice has this ideal deviation structure, then it might seem very important to work out what ideal justice is, since ideal justice serves as a benchmark relative to which we evaluate other societies as more or less just. As with the look-alike contest, this might give ideal justice epistemic priority: Perhaps someone without knowledge of ideal justice would have no idea how to comparatively evaluate other societies. It might also give it methodological priority: The best way to figure out how to evaluate non-ideal societies might be to work out what ideal justice is. Finally, it might give it explanatory priority: Maybe we can’t explain why one society is more just than another except by reference to ideal justice.

Ideal benchmark therefore suggests three ways ideal theory might be prior to non-ideal, comparative theorizing. And while we will see in a moment that the epistemic priority claim can be quickly dismissed, many do maintain the methodological or explanatory priority of ideal theory on these or similar grounds (e.g., Arvan, Reference Arvan2023: 223; Bertea, Reference Bertea2023: 8; Carroll, Reference Carroll2020: 456–458; Gilabert, Reference Gilabert2012: 46–47; Ismael, Reference Ismael2016: 23–24; Jubb, Reference Jubb2012: 239–242; Laurence, Reference Laurence2021: 76–84, 88–90; Matthew, Reference 74Matthew2017: 243; Reference Matthew2019: 559–561; Melenovsky, Reference Melenovsky2019: 434–435; Rawls, Reference 76Rawls1999b: 216; Robeyns, Reference Robeyns2012: 160; Shelby, Reference 77Shelby2013: 150, 153–154; Reference Shelby2016: 12; Simmons, Reference Simmons2010: 12, 23, 34; Stemplowska, Reference Stemplowska2008: 336–338). In this section, I provide an overview of surrounding debates. My main conclusion is negative: Ideal theory doesn’t plausibly enjoy any sort of priority over non-ideal comparative theorizing, at least given the current state of the literature. However, ideal theory is not entirely irrelevant either. Ideal theory may indeed provide a benchmark that sometimes helps us to make comparative evaluations of justice – but it is just one benchmark among many.

2.2 Priority, Three Ways

When non-ideal theorists criticize the alleged priority of ideal theory, they often interpret this priority in an untenably strong epistemic form.Footnote ¹ They assume that if ideal theory served as a benchmark of evaluation, this would imply that – much like in the Chaplin look-alike contest – we couldn’t make any comparative evaluations except by reference to the ideal. Critics then immediately dismiss this view, noting that we clearly can make comparative evaluations without any conception of ideal justice in mind, and that we do so all the time – for example, when we judge that slavery is unjust and that abolishing it improves justice (e.g., Anderson, Reference Anderson2010: 3; Appiah, Reference Appiah2017: 168; Huemer, Reference Huemer2016: 221–222; Sen, Reference Sen2006: 217–218; Reference Sen2009: vii–viii, 21–22).

Such criticisms are unfair, however, because ideal theorists do not assert the epistemic priority of ideal theory. Rawls himself, for example, holds that theorizing about justice involves a process of moving “back and forth” between (among other things) our judgments about ideal and about comparative justice – sometimes revising the former in light of the latter, sometimes revising the latter in light of the former, until all our judgments hang together in “reflective equilibrium” (Rawls, Reference Rawls1999a: 18). Our judgments about slavery, for example, serve as “fixed points” in our theorizing: Any plausible ideal theory must be able to accommodate the grave injustice of slavery and so its severe deviation from the ideal (Rawls, Reference Rawls2001: 29). So Rawls doesn’t embrace the epistemic priority of ideal theory, and this is not a charitable understanding of ideal benchmark more generally.

Instead, ideal benchmark claims that ideal justice provides a benchmark, “comparison with which defines a standard for judging actual institutions” (Rawls, Reference 76Rawls1999b: 199, my emphasis). The idea, then, is that we cannot articulate the criteria of comparative justice that explain our pre-theoretic comparative judgments without ideal theory, since such criteria have an ideal deviation form: One society is more just than another if and only if it deviates less from some specified ideal (compare Sen, Reference Sen2006: 218; Reference Sen2009: 98). So even if, epistemically, we can often make comparative judgments without ideal theory, ideal theory may ultimately be needed to explain the truth of these judgments, since it is needed to articulate the principles vindicating them (compare Laurence, Reference Laurence2021: 99–101). In this sense, ideal theory may be explanatorily prior to non-ideal theory.

From this explanatory priority, it is a small step to infer that ideal theory also has methodological priority, in the sense that non-ideal theory must be “worked out after an ideal conception of justice has been chosen” (Rawls, Reference 76Rawls1999b: 8). For while we often make pre-theoretic comparative judgments without reference to any ideal, the point of theorizing is to get beyond such “easy cases” – and this presumably requires us to invoke criteria of comparative justice. Thus, while we may make scattered pre-theoretic judgments of comparative justice without any conception of the ideal in mind, we can’t engage in systematic theorizing about this without recourse to ideal theory (e.g., Adams, Reference Adams, Roberts-Cady and Mandle2020: 83–84; Berg, Reference Berg2019: 506; Boot, Reference Boot2012: 9–12; Laurence, Reference Laurence2021: 99–100; Robeyns, Reference Robeyns2012: 160; Shelby, Reference 77Shelby2013: 156; Reference Shelby2016: 13; Valentini, Reference Valentini2011: 306–307). In this sense, ideal theory may be methodologically prior to non-ideal theory.

One way to resist the inference from explanatory to methodological priority is to embrace “methodological particularism” (Sinnott-Armstrong, Reference Sinnott-Armstrong1999: 9–11): Even if comparative evaluations of justice are ultimately explained by ideal principles, one may deny that we make progress on normative matters by seeking out and employing general principles rather than through case-based reasoning (e.g., Huemer, Reference Huemer2016: 231–232; Medina, Reference Medina2013: 11–12). Here, then, is a point where a more general question of philosophical methodology intersects with the ideal and non-ideal theory debate: If we generally don’t need principles to engage in normative theorizing, then we specifically don’t need ideal principles. However, this is not the only way to deny methodological priority: Rather than challenging the inference from explanatory to methodological priority, one might dispute ideal theory’s explanatory priority directly. Following my general practice of focusing on issues that are distinctive to political philosophy, I set particularism aside and consider this latter response here.

2.3 Second-Best Objections

We have seen that ideal theory is not plausibly epistemically prior to non-ideal theorizing about comparative justice, but that if comparative principles of justice have the following structure, this may vindicate ideal theory’s explanatory, and so methodological, priority:

Interestingly, then, whereas just above we found our first point where an issue in the ideal and non-ideal theory debate turns on a general question of philosophical methodology, we have now uncovered the first point that instead turns on a substantive question about justice. The case for the methodological priority of ideal theory depends on the substantive claim that comparative principles of justice have the above structure.

Well, do they? Of course, many comparative principles do not refer to any ideal: The criterion of whether one mountain is higher than another does not refer to the highest mountain (Sen, Reference Sen2006: 222; Reference Sen2009: 16, 102). But this doesn’t yet show that justice lacks this structure. After all, the criterion judges use in the Chaplin look-alike contest does have an ideal deviation structure, and so too does a criterion such as “distance from the highest mountain” (Barrett, Reference Barrett2023b: 35). So, whether justice has an ideal deviation structure is a substantive question, in the sense not only that it isn’t methodological but also that it doesn’t turn on any general claim about the logic of comparative evaluation. The question is whether ideal deviation principles make plausible principles of justice.

“Second-best” objections suggest that they do not. Ideal deviation principles are not plausible or helpful, such objections claim, but inaccurate and misleading (e.g., Barrett, Reference Barrett2020: 108–112; Brennan and Pettit, Reference Brennan, Pettit, Jackson and Smith2005; Estlund, Reference Estlund2014: 120–121; Reference Estlund2020: ch. 14; Goodin, Reference Goodin1995: 52–55; Sen, Reference Sen2009: 16; Wiens, Reference Wiens2016a: 142–144). The basic concern is that, in general, it is a mistake to assume that “second-best” options most closely approximate or resemble ideal or first-best options. Originally, the “general theory of second best” was proposed as an economic theorem, showing that if we cannot realize every “optimality condition” in a market, it isn’t always more efficient to realize more rather than fewer optimality conditions (Lipsey & Lancaster, Reference Lipsey and Lancaster1956). But the idea generalizes (e.g., Barrett, Reference Barrett2023b: 36–38; Estlund, Reference Estlund2020: ch. 14; Räikkä, Reference Räikkä2010; Wiens, Reference Wiens2020; Hammerton & Patra, Reference Hammerton and Patra2024). If my ideal drink is gin and tonic, I might nevertheless prefer rum and coke to either rum and tonic or gin and coke – even though the latter two options are descriptively closer to gin and tonic (they each share one common ingredient) than rum and coke (which shares none) (compare Sen, Reference Sen2009: 16). Or, returning to justice, suppose that because we assume the ideal realizes full compliance, it contains no corrective institutions (for example, those compensating victims of injustice). In this respect, abolishing existing corrective institutions would make our society deviate less from the ideal. But surely this is not a good reason to infer that abolishing our corrective institutions would improve justice – we need these institutions precisely because our society isn’t ideal (compare Estlund, Reference Estlund2014: 120–121).

Ideal theorists have two ways of responding to second-best objections. The first is to claim that they involve mischaracterizations of ideal justice. For example, perhaps it is a mistake to assume full compliance, precisely because we want our conception of the ideal to contain corrective institutions. The ideally just society, on this view, is more like a society with an ideal “immune system” preventing injustice from getting too out of hand, rather than an always healthy society where no one ever acts unjustly (Laurence, Reference Laurence2021: ch. 5). Since ideal justice, so conceived, involves corrective institutions, corrective institutions do not represent a deviation from the ideal.

One challenge for this first reply is that, in the case at hand, it may seem implausible to deny that an ideal society would satisfy full compliance. Now, if we were treating the ideal as a goal (Section 3), a partially compliant society might indeed be more suitable, since it is something we can more realistically achieve. But we are here considering the idea that ideal theory provides a benchmark of evaluation, and this might instead suggest that ideal justice should meet a standard of conceptual perfection unconstrained by considerations of feasibility – and, conceptually, some degree of noncompliance seems less just than none (e.g., Brennan, Reference Brennan2014: 71).

That said, there is a lively debate here about which realism or feasibility constraints are appropriate when theorizing about ideal justice, beyond mere conceptual possibility. For example, perhaps even ideal justice should be constrained by the dictum that “ought implies can,” so that we can rule out candidate principles surpassing human capabilities. A further question, though, is how to interpret this “can.” For example, if human nature precludes us from complying with a demanding conception of justice because we are too lazy, selfish, or prone to discrimination, does this imply that we “can’t” achieve it? On the one hand, we don’t typically think that if someone is too lazy or selfish to do something, they “can’t” do it in a sense that could block a requirement (Estlund Reference Estlund2011a; Reference Estlund2020: ch. 5–7; though see Anderson, Reference Anderson2010: 3–4; Berg, Reference Berg2023; Wiens, Reference Wiens2016b), nor do we think that uncertainty about our ability to prevent discrimination implies further uncertainty about whether such discrimination is unjust (Talisse, Reference Talisse, Weber and Vallier2017: 66–69). But on the other hand, the issue may be different when it comes to deep and inalterable facts about human nature (Southwood, Reference Southwood2015; see also Miller, Reference Miller2013: ch. 1), or to facts that are not vicious or “justice-tainting” but positive or neutral, for example, that we care about our children (Badhwar, Reference Badhwar2016). Or perhaps a relevant difference is that it takes many agents to realize the “ought” of justice, since what others are willing to do can, after all, constrain what I can do (Enoch, Reference Enoch2018; Schmidtz, Reference Schmidtz2011: 777–778; Reference Schmidtz2023: ch. 9; though see Estlund, Reference Estlund2020: 22–24), and since the relationship between individual and collective feasibility is complex – for example, collective feasibility depends not merely on what each of us can do alone but on our ability to coordinate (Barrett, Reference Barrett2025a, 695–6; Stemplowska, Reference Stemplowska2016a; though see Estlund, Reference Estlund2011a: 235–237; Reference Estlund2020: 126–129, chs. 11–12). And, anyway, it isn’t obvious that “ought implies can” applies to ideal justice, insofar as we are treating it as a benchmark of evaluation rather than something we should try to realize (compare Moen, Reference Moen2023: 294–295; Valentini, Reference Valentini, Weber and Vallier2017: 14). Perhaps we should instead think of ideal justice, for this purpose, as what we ought to achieve if we can (Cohen, Reference Cohen2003: 230–231; Reference Cohen2008: 251–253). Needless to say, this intramural debate about how best to do ideal theory gets very complicated.

Rather than digging further in, then, let us instead consider ideal theorists’ second response to second-best problems, namely, claiming not that we have incorrectly specified the ideal, but rather that we have measured deviations from it in the wrong way (e.g., Barrett, Reference Barrett2023b: 36–37; Boot, Reference Boot2012: 13–15; Gilabert, Reference Gilabert2012: 45–46; Stemplowska, Reference 78Stemplowska, Lippert-Rasmussen, Brownlee and Coady2016b: 286–287; Swift, Reference Swift2008: 372–377; see also Berg, Reference Berg2019: 504–506; Goodin, Reference Goodin1995: 53n.45; Hammerton & Patra, Reference Hammerton and Patra2024; Räikkä, Reference Räikkä2010: 213–217). For example, even if the ideal society would be a full-compliance society with no corrective institutions, it doesn’t follow that a non-ideal society deviates less in the relevant sense if it also lacks corrective institutions. Perhaps we should instead measure deviations in an abstract dimension concerned, say, with how much oppression a society has. We can then say that an unjust society with corrective institutions deviates less from the ideal in the sense that matters, because the amount of oppression in the society with corrective institutions is smaller, and so more similar to, the amount of oppression in the ideal.

While the first response (changing our conception of the ideal) can perhaps avoid some second-best problems, the second response (changing our deviation measure) is more general. After all, any time we encounter an apparent second-best problem but are confident in our specification of the ideal, we can infer that we must not be measuring deviations from it in a way that tracks justice. Now, it is always trivially possible to construct a deviation measure that corresponds to a society’s justice: A society deviates further from the ideal when it deviates further in its degree of justice. But such a measure lacks any explanatory power and so cannot vindicate the explanatory or methodological priority of ideal theory. The question, then, is whether we can construct a more informative measure that retains this connection to justice. As our earlier discussion suggests, this is unlikely to be a measure of institutional similarity, since institutions often interact in ways that generate second-best problems (see Section 3.1). Instead, if a deviation measure is to track and explain degrees of justice, it will have to track deviations in some dimension that fundamentally matters to justice: in the basic values or principles that explain why the ideal society is ideal. To take some simple examples: If what fundamentally makes the ideal society ideal is that it maximizes (some conception of) freedom, that it maximizes (some conception of) equality, or that it maximizes some weighted function of freedom and equality, then our deviation measure should count a society as deviating more from the ideal when it is less equal, less free, or does worse according to the weighted function of freedom and equality. It is only by going to this fundamental level that we can avoid second-best problems.

Incidentally, this also suggests that as far as ideal benchmark is concerned, ideal theory should aim to uncover abstract principles rather than concrete institutions. Specifying ideal principles, and measuring deviations from them in terms of features that fundamentally matter to justice, is precisely how to avoid second-best problems.

2.4 Redundancy Objections

So far, we have seen that the argument for the priority of ideal theory as an ideal benchmark relies on the claim that comparative principles of justice have the following structure:

For example, d must not be a measure of institutional similarity, but it might be a measure of equality, freedom, oppression, or some weighted function of these or other fundamental values.

While this view avoids second-best problems, it runs into another class of problems, which I call redundancy objections. Their basic gist is that, even if we can construct an ideal deviation principle of this sort, a conception of the ideal is ultimately redundant to the task of comparative evaluation (compare Sen, Reference Sen2006: 216–222; Reference Sen2009: 15, 96–102). Now, as we have seen, to articulate an ideal deviation measure, we must specify both a conception of the ideal and a particular dimension in which we measure deviations from it (e.g., Barrett, Reference Barrett2023b: 37–38; Sen, Reference Sen2006: 219–220; Reference Sen2009: 98–99). A redundancy objection arises when specifying each element. Consider each in turn.

The first redundancy objection focuses on the process of identifying something as ideal. Here’s one plausible model. We have various candidate conceptions of ideal justice before us. When evaluating which conception to accept, we compare them in terms of how well they realize our fundamental concerns – for example, which better realizes a concern with freedom, equality, welfare, or avoiding domination. Finally, we pick out the ideal as the candidate that best realizes these basic values (Wiens, Reference Wiens2015a: 437–440; Reference Wiens2025: ch. 3).

If this model is correct, however, ideal theorists face a problem (Wiens, Reference Wiens2015a: 442). For if we select the ideal by choosing, among various candidates, whichever best realizes certain basic values, then, it seems, we can simply use these basic values to make comparative evaluations directly: One society is more just than another when it better realizes these basic values. The ideal is redundant – a third wheel – in this exercise. To avoid this redundancy problem, ideal theorists therefore owe an account of how they can identify the ideal without invoking prior criteria that can also be used to make comparative evaluations. Otherwise, they lose any claim to explanatory or methodological priority: We use basic values both to pick out the ideal and to make comparative evaluations of justice, and there is no need to do the former before the latter (compare Boot, Reference Boot2012: 13; Hamlin and Stemplowska, Reference Hamlin and Stemplowska2012: 53–54).

Now, perhaps some alternative model of how to pick out ideal principles is possible (compare Erman and Möller, Reference Erman and Möller2022). But regardless, ideal theorists face a second redundancy objection. This objection grants that the ideal theorist has a conception of the ideal already selected, but focuses instead on how to measure deviations from that ideal. There are lots of ways to measure deviations, so an ideal deviation principle has to specify a particular deviation measure. But, this second redundancy objection goes, once our ideal theorist specifies this deviation measure, they’ve given away the game. For we can use the measure to make comparative evaluations directly; the deviation measure, not the ideal, does all the work (Barrett, Reference Barrett2023b: 38–46).

To illustrate, suppose we specify the ideal as a society with no oppression and say that a society is more just than another when it deviates less in its degree of oppression from the ideal. In spelling out this deviation measure, we develop a theory of how to aggregate different instances and types of oppression into an overall evaluation of which of two societies has more oppression overall, and so which deviates further from the ideal. Now, it seems, we can simply use this account of what qualifies as more or less oppression to make comparative evaluations of non-ideal societies directly: One society is more just than another when it has less oppression overall. This makes our reference to the ideal, oppression-free society, redundant. The fact that a society has less oppression explains why it is more just and so why it deviates less from the ideal; the fact that it deviates less from the ideal doesn’t explain why it has less oppression or why it is more just (see Section 4.2).

The burden, then, is on the ideal theorist to show how this second redundancy objection can be resisted. But it is hard to see how this burden can be met (Barrett, Reference Barrett2023b: 43–46). For consider any dimension d along which societies can deviate more or less from the ideal, in such a way that deviating further from the ideal always corresponds to greater injustice. In many cases, the ideal will be at one of the extremes. For example, dimension d may be the degree of oppression, such that in the ideal society d is minimized (at least to the extent that is feasible, after factoring in trade-offs with other relevant dimensions). Or if dimension d tracks the realization of a positive good, such as freedom, then d will be maximized in the ideal (given the same caveats). In either case, we do not need to refer to the ideal when making comparative evaluations of justice: What fundamentally matters is that there is less oppression, or more freedom, not that there is a smaller deviation in oppression or freedom from the minimally oppressive or maximally free ideal.

For an ideal deviation measure to avoid making the ideal redundant, then, it must deny that the ideal is at the extreme of some dimension. It must deny that – even after factoring in feasibility constraints and trade-offs with other dimensions – we should maximize or minimize something in some dimension fundamentally relevant to justice. Instead, our ideal theorist must invoke a “sweet spot” criterion on which there is some optimal, non-minimal or -maximal degree of d at the ideal, such that deviations from this sweet spot in either direction make things less just. For example, the principle might claim that exactly “5” oppression is ideally just, with either more or less oppression making things less just. However, this is a very peculiar sort of principle of justice, and I have never come across a plausible one. In fact, as I have suggested elsewhere, if we look at all the ideal deviation criteria that have been suggested in the literature, none are sweet spot criteria, and all make the ideal redundant (Barrett, Reference Barrett2023b: 39–43).

So far, then, the burden facing ideal benchmark remains unmet: Unless ideal theorists can come up with responses to these redundancy objections, we must conclude that a conception of the ideal lacks explanatory priority. Absent this priority, ideal theorists also lack an argument for ideal theory’s methodological priority. Justice, it would seem, is not much like a Chaplin look-alike contest after all.

2.5 Relevance without Priority

The two classes of objections we have discussed pose a dilemma.Footnote ² Either we do not measure deviations from the ideal in dimensions that are fundamentally relevant to justice, and we face second-best problems. Or we do measure deviations in dimensions fundamentally relevant to justice, and we face the charge of redundancy. Unless ideal theorists can find a way to slip through the horns of this dilemma, they have no grounds on which to assert the epistemic, explanatory, or methodological priority of ideal theory. But they can still retreat to a weaker view: Perhaps ideal theory is relevant to comparative theorizing, even if it lacks priority (e.g., Carens, Reference Carens2013: 301–303; Erman and Möller, Reference Erman and Möller2013: 30–31; Reference Erman and Möller2022: 536–537; Estlund, Reference Estlund, Sobel, Vallentyne and Wall2016; Reference Estlund2020: ch. 15; García Gibson, Reference García Gibson2016: 97–99; Gilabert, Reference Gilabert, Weber and Vallier2017: 117–118; Swift, Reference Swift2008: 372–378; Valentini, Reference Valentini2011: 307–309; see also Boot, Reference Boot2012).

This weaker claim is hard to dispute. If we accept anything like the method of reflective equilibrium, then of course theorizing about the ideal sometimes informs comparative theorizing. After all, any verdict we reach about ideal justice serves as a constraint: Whatever else is true of our comparative theory of justice, it had better pick out the ideal society as most just. For example, if the ideal society involves no oppression, this implies a conceptual constraint on our comparative criteria: They must pick out a society involving no oppression as their conceptual upper limit. Or, to take a more concrete example, suppose we develop a conception of an ideally just society in which race has no influence on one’s treatment or life prospects. This is an important data point when working out criteria of comparative justice, since it places a constraint on them: They must be able to explain why such a society is ideal.

Ideal theory may be relevant to comparative theorizing in other ways, too. For example, versions of ideal theory that assume full compliance may help us understand what justice requires by providing a simple scenario where we can work out the implications of candidate principles absent complications raised by noncompliance (e.g., Adams, Reference Adams, Roberts-Cady and Mandle2020: 78–79; Carroll, Reference Carroll2020; Ismael, Reference Ismael2016: 19–20; Leader Maynard, Reference Leader Maynard2025; Matthew, Reference Matthew2019: 566–569; Simmons, Reference Simmons2010: 8–9). Likewise, working out a conception of ideal justice may help us to better understand central normative concepts such as justice, freedom, or equality (e.g., Erman and Möller, Reference Erman and Möller2022: 537; Wiens, Reference Wiens2023: 787; Reference Wiens2025: ch. 7; see also Hamlin and Stemplowska, Reference Hamlin and Stemplowska2012; Stemplowska and Swift, Reference Stemplowska, Swift, Mandle and Reidy2013: 125). And with a better understanding of these concepts, we may go on to do better comparative work as well.

Ideal theory, then, is somewhat relevant to comparative theorizing. But how relevant? If our goal is to make comparative evaluations of justice, how much does ideal theory help? Some critics argue that ideal theory is not very helpful for this purpose, and may even be harmful, since focusing too much on the ideal may distort our comparative theorizing, in much the same way that attempting to fit a scientific theory to a small set of data points may lead to a distorted theory (e.g., Anderson, Reference Anderson2010: 4–5; Brennan and Pettit, Reference Brennan, Pettit, Jackson and Smith2005; Goodhart, Reference 72Goodhart2018: 38–44; Goodin, Reference Goodin1995: 45–50; Mills, Reference Mills2005; Reference Mills2017: ch. 5; Schmidtz, Reference Schmidtz2011: 775–778; Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 4). For example, though a color-blind ideal would imply a constraint on our criteria of comparative justice – they must explain why this is ideal – there are many criteria meeting this constraint, and focusing too much on the ideal may lead us to endorse one that most simply explains the ideal rather than one that better fits our convictions about a range of actual and hypothetical cases (Anderson, Reference Anderson2010: 4). Likewise, various common ideal-theoretic practices, such as assuming full compliance with (and universal acceptance of) principles of justice, may amount to assuming away the very problems to which justice is meant to be a solution (e.g., Freiman, Reference Freiman, Brennan, van der Vossen and Schmidtz2017: 302–303; Levy, Reference Levy2016; Schmidtz, Reference Schmidtz2011: 777; Reference Schmidtz and Olsaretti2018: 331–335, 340–341; Reference Schmidtz2023: ch. 11; see also Laurence, Reference Laurence2021: ch. 5; Larmore, Reference Larmore2020). So far from helping us to refine our concept of justice, this may mislead us, since we are trying to apply the concept where it doesn’t apply – namely, beyond the “circumstances of justice” (Rawls, Reference 76Rawls1999b: 109–112; though see Estlund, Reference Estlund2020: ch. 4).

Now, it is difficult to adjudicate such disputes or to articulate precisely to what extent ideal theory is useful to comparative theorizing. But the limited relevance of ideal theory nevertheless comes out clearly when we note that it has no more claim to relevance than theorizing about dystopia. Much like a conception of the ideal, a conception of dystopia may help us to develop our criteria of comparative justice by constraining our comparative criteria – in this case, because our comparative criteria had better be able to explain why they are maximally or at least very unjust. But, for that matter, so too can we develop our criteria by theorizing about all sorts of historical examples of more or less just societies, or even about science fiction scenarios: Any time we make any judgment about something being very just or unjust, or about something being more just than something else, this constrains our criteria of comparative justice, since they must be able to explain such verdicts. The same goes for ideal theory’s connection to normative concepts. Trying to apply these concepts to ideal scenarios may sometimes help us better understand such concepts and ensure the robustness or “counterfactual coherence” of our theorizing (Leader Maynard, Reference Leader Maynard2025). But the same role can also be played by our trying to extend these concepts to various historical or fantastical scenarios that have nothing to do with ideal justice at all.

All this suggests that ideal theory isn’t special. It enjoys no priority over comparative theorizing and at best plays a modest, ancillary role in such theorizing – just like it may sometimes be helpful to think about dystopia, or history, or science fiction, or anything else that helps focus our minds on social phenomena. Ideal theory may provide a benchmark relative to which we can evaluate other societies as less just. But, if so, it is just one benchmark among many – for example, we can also treat dystopia as a benchmark relative to which we evaluate societies as more just.

Here, ideal theorists might push back. We have been considering whether it is possible to vindicate ideal theory’s methodological priority by reference to its explanatory priority, but perhaps it is possible to make a more direct argument: Perhaps there is something about ideal theory that explains why it is especially helpful methodologically that has nothing to do with its ability to fundamentally explain our comparative judgments. For example, some cases seem to admit of easier intuitive or “eyeball” judgments of justice than others (Estlund, Reference Estlund, Sobel, Vallentyne and Wall2016). And maybe we just happen to be, psychologically, particularly good at making “eyeball” judgments of ideal justice. Now, this particular suggestion, even if true, wouldn’t vindicate any special role for ideal theory, which is meant to go beyond such intuitive judgments. But maybe some similar account could succeed.

I cannot hope to rule out the existence of such an account here, nor, for that matter, to have provided a definitive rebuttal of attempts to vindicate ideal theory’s methodological priority by appeal to its explanatory priority. But I do hope to have explained where the debate is currently at. Arguments for the explanatory or methodological priority of ideal theory over non-ideal theorizing about comparative justice face serious and unmet objections. Arguments that ideal theory is merely sometimes relevant to comparative theorizing seem unobjectionable. But if ideal theorists want to claim that ideal theory is not only somewhat but also especially relevant here – well, then, they have some explaining to do.

3.1 A Target for Reform

My favorite documentary of all time is the fourteen-part PBS series, Eyes on the Prize: America’s Civil Rights Movement. The series begins each episode with the titular folk song, which served as a rallying cry throughout the movement:

I have always found this song inspiring, as it casts up images of civil rights activists, facing daily dangers and indignities, fighting for the distant goal of racial justice. It urges them to “keep their eyes” on this goal, orienting themselves toward it, treating it as a long-term target for reform.

But “Eyes on the Prize” was not the only song of the civil rights movement. Perhaps even more famous and resonant was the gospel, “We Shall Overcome”:

While still acknowledging the need for change, the focus is now shifted toward the injustices in front of us. We are urged to overcome these injustices, rather than to orient ourselves toward some far-off ideal.

As it happens, the rival approaches embedded in these songs closely track two sides of the ideal and non-ideal theory debate. On one side, ideal theorists who embrace ideal target argue that we need ideal theory to provide a long-term target for reform. In order to pursue justice, we must keep our eyes on the prize of an ideally just society – so we better work out what that prize would be (e.g., Adams, Reference Adams, Roberts-Cady and Mandle2020: 81–82; Buchanan, Reference Buchanan2004: ch. 1; Erman and Möller, Reference Erman and Möller2013: 31–32; Reference Erman and Möller2022: 535–536; García Gibson, Reference García Gibson2016: 100–105; Gilabert, Reference Gilabert2012: 47–48; Korsgaard, Reference Korsgaard1996: 148; Laurence, Reference Laurence2021: ch. 3; Rawls, Reference Rawls1999a: 89; Reference 76Rawls1999b: 215; Robeyns, Reference Robeyns2008: 344–345; Reference Robeyns2012: 160; Shelby, Reference 77Shelby2013: 150, 155; Reference Shelby2016: 12; Simmons, Reference Simmons2010; Sirsch, Reference Sirsch2024; Stemplowska, Reference Stemplowska2008: 332; Reference 78Stemplowska, Lippert-Rasmussen, Brownlee and Coady2016b: 293; Stemplowska and Swift, Reference Stemplowska, Swift, Mandle and Reidy2013: 119–120; Valentini, Reference Valentini2012: 661–662). On the other, many non-ideal theorists advocate a “problem solving” approach: They argue that we shouldn’t treat the ideal as a goal, but should rather try to identify and diagnose the injustices in front of us, aiming to develop and pursue solutions that would overcome them (e.g., Anderson, Reference Anderson2010: ch. 1; Schmidtz, Reference Schmidtz2011; Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 10; Sen, Reference Sen2006; Reference Sen2009; Wiens, Reference Wiens2012; see also Mills, Reference Mills2005; Reference Mills2017: ch. 5).

This debate over ideal target is independent of the debate over ideal benchmark discussed in Section 2. Even if we don’t need the ideal to serve as a benchmark for evaluating what would make society more just, we may nevertheless need the ideal to serve as a long-term target whose pursuit we must trade-off against making society more just in the short term. In the literature, this point is often put in terms of the metaphor of a mountain range (Simmons, Reference Simmons2010: 34–35; see also Gaus, Reference Gaus2016: chs. 1–2). Suppose that height is analogous to justice, so that taller mountains correspond to greater justice. Then, if our goal is eventually to reach the greatest possible heights, we had better not just set off climbing upward from where we are, without knowing where the highest mountain is. For the easiest and most obvious route upward might take us away from this ultimate goal. And, correspondingly, if our aim is eventually to achieve ideal justice, we had better not just start trying to improve justice in the short term without knowing where ideal justice is. For the easiest and most obvious ways to improve short-term justice might take us away from this ideal.

Moving beyond the metaphor, there are two main dynamics generating a conflict between pursuing short-term justice through problem solving and pursuing the long-term target of ideal justice (Barrett, Reference Barrett2020: 108–111; see also Gaus, Reference Gaus2016: ch. 2; Reference Gaus2021). First, sometimes implementing change X improves things, and implementing change Y improves things, but X and Y interact in such a way that implementing both X and Y makes things worse. This is the same phenomenon generating the “second-best” problems discussed earlier (Section 2.3), but to take a new example: Some worry that implementing more generous welfare programs would improve justice, and that so too would having more open borders, but that implementing both would be disastrous (since the welfare programs could not survive the influx of immigration). Second, even if implementing both X and Y would produce a significant gain, sometimes one change interacts with the other in the sense that implementing X first makes it harder or impossible to implement Y later. For example, even if, contrary to the last example, it would be good to implement both a more generous welfare state and more open immigration policies, some worry that liberalizing our border policy first would reduce support for welfare state programs, preventing us from eventually achieving both goals (compare Carens, Reference Carens2013: ch. 12).

Elsewhere, I have called the first dynamic – changes X and Y are each good alone, but interact in such a way that they are bad together – “combinatorial complexity.” And I have called the second – changes X and Y are good together, but interact in the sense that implementing X sets back the achievement of Y – “transitional complexity” (Barrett, Reference Barrett2020). Together, these dynamics motivate ideal target in the following way. Due to combinatorial complexity, we can’t merely focus on isolated problems, whose “solutions” might combine in troubling ways, so we need a comprehensive conception of the ideal combination of features to treat as a goal. Due to transitional complexity, we can’t merely focus on the short term but must think about how implementing changes now furthers or sets back the achievement of future goals. Thus, pursuing long-term justice seems to require us to specify a maximally comprehensive and maximally long-term goal – and this is precisely how ideal target treats our conception of the ideally just society. Of course, few would claim that we should always sacrifice short-term justice at the altar of long-term justice, but so long as the long-term matters some, our conception of ideal justice must serve as an ideal target whose pursuit we trade-off against short-term justice. In this sense, ideal theory has priority over non-ideal theorizing about which goals to pursue, which cannot proceed responsibly without having such a long-term target in mind.

So ideal target has a lot going for it, and one can see why it has so many adherents. However, it faces two closely related objections, which I call the epistemic and moving target objections. In the rest of this section, I first develop these objections and then explain the inadequacy of a recent reply – namely, that they assume an overly ambitious conception of ideal theory – while also sketching an alternative, non-ideal approach to theorizing about long-term justice. Finally, and parallel to my discussion in Section 2, I consider how ideal theory may nevertheless play some helpful role when setting goals for reform. The upshot is that while we likely shouldn’t treat the ideal itself as a weighty goal, ideal theory may sometimes help us work out other targets to pursue.

3.2 The Epistemic Objection

In our previous discussion, we saw that if ideal theory is to provide an ideal benchmark, it should aim to identify abstract principles of justice rather than institutional particulars. We also saw that, given this aim, we should treat justice as a standard of conceptual perfection involving full compliance, or at most should factor in feasibility as it figures into what we “can” do in the sense relevant to “ought implies can” (Section 2.2). However, neither is true of ideal target. If we wish to treat ideal justice as a long-term goal, it is not enough to identify abstract principles of ideal justice – we need a fairly concrete sense of what it would take to realize those principles (e.g., Barrett, Reference Barrett2025a: 695; Reference Barrett2023a: 145; Gaus, Reference Gaus2016: 19–21; Sirsch, Reference Sirsch2024: 466–467; see also Cohen, Reference Cohen2010: ch. 4; Gilabert, Reference Gilabert, Weber and Vallier2017: 107–109; Hamlin and Stemplowska, Reference Hamlin and Stemplowska2012: 53). Likewise, it is not enough to imagine the most just society we can conceive of, or that is consistent with human capabilities. Instead, we need to uncover the most just society that is feasible in the sense that we can chart a path from here to there, where such a goal is unlikely to involve full compliance or conceptual perfection.

The point about feasibility is straightforward, since something is not a suitable goal for reform unless we can feasibly achieve it. But why do I suggest that ideal target mandates a focus on institutions rather than merely principles? Well, consider that proponents of the same principles often greatly disagree about how to realize them. For example, three diehard Rawlsians, who each agree with Rawls’s principles, might nevertheless disagree about their institutional realizations. One, who is optimistic about the power of markets and of philanthropy and entrepreneurship to correct market failures, might think the principles are best realized in a free market society. Another might claim the principles are best realized in a capitalist welfare state. A third might think they can only be realized given thoroughly socialist institutions. Obviously, these three thinkers are bound to disagree substantially about what far-off goal we should be pursuing, as well as what steps would constitute progress toward this goal. For example, does nationalizing healthcare bring us closer or further from eventually realizing Rawls’s ideal principles? Surely this depends on what institutions it would take to realize those principles.

So, on ideal target, the aim is not merely to identify abstract principles of justice, but also to identify their institutional realizations – at least in sufficient detail that they can orient our reform efforts. And it is here that ideal target runs into difficulty. For even granting that we can identify ideal principles, it is notoriously difficult to predict the effects of different institutional arrangements, especially when we are dealing with ideal institutions very different from anything we have ever tried. But if we cannot identify and work out how to achieve such ideal institutions, then we can’t treat them as a long-term goal, and ideal target can’t get off the ground (e.g., Barrett, Reference Barrett2020: 111–118; Reference Barrett2023a: 195; Gaus, Reference Gaus2016: ch. 2; Gaus and Hankins, Reference Gaus, Hankins, Weber and Vallier2017; Hendrix, Reference Hendrix2013: 132–134; Moen, Reference Moen2024: 99–100; Nili, Reference Nili2018: 98–99; Popper, Reference Popper2013: ch. 9; Wiens, Reference 80Wiens2015b: 466–467; see also Cohen, Reference Cohen2010: ch. 4; Herzog, Reference Herzog2012; Wiens, Reference Wiens2025: chs 4–5).

The epistemic problem is not merely that it is difficult to identify which institutions would be ideal if we could realize them – though this is certainly the case, given the very combinatorial complexity that motivates ideal target (Barrett, Reference Barrett2020: 115–116; Gaus, Reference Gaus2016: ch. 2; Gaus and Hankins, Reference Gaus, Hankins, Weber and Vallier2017; see also Wiens, Reference Wiens2025: chs. 4–5). It is furthermore that it is hard to know which institutions we could feasibly realize (Wiens, Reference 80Wiens2015b: 466–467), or which present steps would take us toward them (Barrett, Reference Barrett2020: 117). Social change is notoriously complex and hard to predict. One reason is that it is a strategic context in which one agent’s or group’s actions are responded to by other agents or groups, who have different goals. It is hard to predict the long-run outcome of competing efforts at reform and conservation. Another is that social change significantly depends on technological change, which itself resists reliable predictions. Indeed, we have both theoretical and empirical grounds for doubting the reliability of long-term predictions about social phenomena. Theoretically, most complex systems get harder to predict as we forecast further out: We are better at predicting tomorrow’s weather than next week’s (Barrett, Reference Barrett2020: 115; Gaus, Reference Gaus2016: 80; Reference Gaus2021: 10–13, 210–211). Empirically, society appears to be no different: Even the most talented forecasters (as measured by their track records), who sometimes exhibit a surprising knack for making relevant predictions as long as five to ten years out, do little better than chance trying to make similar predictions much further than that (Gaus, Reference Gaus2021: 211–213; see Tetlock, Reference Tetlock2006; Tetlock and Gardner, Reference Tetlock and Gardner2016).

The epistemic problem is therefore threefold: We cannot reliably identify which institutions would be ideal if we realized them, which institutions we could feasibly realize in the long run, nor what short-term steps would bring us toward them. Indeed, the problem grows worse once we factor in moral constraints: We must figure out not only which institutions we could feasibly realize from where we are but also which we could realize through a series of morally permissible steps, involving morally acceptable transition costs (e.g., Buchanan, Reference Buchanan2004: 61–63; Gilabert, Reference Gilabert2012: 52; Räikkä, Reference Räikkä1998: 33–38; Rawls, Reference Rawls1999a: 89; Simmons, Reference Simmons2010: 18–23; see also Moen, Reference Moen2022). And, in determining all this, we must only make use of information that we may permissibly gather and employ (Herzog, Reference Herzog2012: 274–275). Of course, none of this is to say that we can never have the slightest inkling whether a short-term change would further or set back the eventual achievement of ideal justice. But the question is not whether we can make such judgments with any reliability; it is whether we can make them with sufficient confidence that we should be willing to accept more certain sacrifices of short-term justice for the sake of furthering this long-term target (Barrett, Reference Barrett2020: 114–118; Gaus, Reference Gaus2016: ch. 2). Critics worry that such cases are, at best, few and far between. For most intents and purposes, we can ignore the ideal, since we can’t be sufficiently confident which steps would bring us toward it for this to weigh heavily in our decisions about which shorter-term goals to pursue.

3.3 The Moving Target Objection

The moving target objection picks up where the epistemic objection leaves off (e.g., Hendrix, Reference Hendrix2013: 134–135; Muldoon, Reference 75Muldoon2016: ch. 1; Nili, Reference Nili2018; Page, Reference Page2018: 11; Popper, Reference Popper2013: 150–151; Rosenberg, Reference Rosenberg2016: 62–70; Schmidtz, Reference Schmidtz2011: 783–785; Reference Schmidtz2023: ch. 21; see also Gaus, Reference Gaus2016: 84–87). It comes in two varieties. One remains broadly epistemic in its focus – it worries that our conception of ideal justice might change as we pursue it. The other is that ideal justice, or at least its institutional realization, might itself change in the course of its pursuit.

On either version, the core problem is this. Suppose that, despite the concerns just raised, we sometimes forego short-term improvements (or accept short-term losses) in justice, in order to take steps toward our current conception of ideal justice. Over time, our conception of ideal institutions is likely to change, as we learn more about justice and about how different institutions interact to produce their justice-relevant effects. Likewise, as background conditions change (for example, pertaining to technology or the climate), which institutions best realize ideal principles are likely to change too. So, the worry goes, if we pursue our current conception of which institutions are currently ideal, we are liable to set off in the wrong direction: If we forego short-term justice now for the sake of eventually realizing these institutions, we may later come to regret this when we realize these institutions never were, or are no longer, ideal. Furthermore, due to transitional complexity, pursuing our current conception of the ideal now may sometimes even set us on a path that makes it more difficult for us to achieve our future conception of the ideal: We may not only head off in the wrong direction but also get stuck there, or at least find it harder to make it back onto the right path.

The moving target objection compounds the epistemic objection. According to ideal target, ideal theory is prior to non-ideal theorizing about what goals to pursue, because our conception of ideal justice serves as a long-term goal for reform whose pursuit we must trade-off against short-term efforts to overcome injustice. The epistemic objection warns that we should not have much confidence in our judgments about what steps would take us toward ideal institutions. The moving target objection warns that if we do try to take steps toward our current conception of ideal institutions anyway, this conception is likely to change before we realize it. In effect, this further drives down our confidence that, in any given case where we are tempted to sacrifice relatively certain short-term justice for the sake of ideal long-term justice, this trade-off is worth it. Better to overcome the problems in front of us than to pursue a hazy, shifting conception of the ideal.

3.4 Aiming for Less

Ideal theorists rarely attempt to rebut such concerns directly. Instead, their more common response is to claim that such objections are beside the point, since they presuppose an overly ambitious conception of ideal theory. We have already anticipated one variant of this response: Ideal theorists only need to identify ideal principles (compare Estlund, Reference Estlund2020: 8–10). But, as we have seen, this is unsatisfying if the aim is to vindicate ideal target: We cannot treat principles as goals to pursue unless we have at least some rough sense of how to realize them institutionally. So more promising variants of this response maintain an institutional focus while downgrading its aspiration in other ways. For example, instead of aiming to provide a complete description of an ideally just society, ideal theorists might aim only to describe parts of it – that is, to describe features the ideal society would have to realize, no matter what else is true of it (e.g., Berg, Reference Berg2019; Carroll, Reference Carroll2022: 540–543; Robeyns, Reference Robeyns2012: 161; Sreenivasan, Reference Sreenivasan, Williams, Nagy and Elster2012: 245–247). In this sense, ideal theory can be “incomplete” (Berg, Reference Berg2019). And this is consistent with ideal target, since our conception of the ideal only need be specified in enough detail to guide reform efforts.

Unfortunately, incomplete ideal theory remains a tall task. While certain features of the ideal are obvious – for example, it wouldn’t contain slavery – these are not relevant here, for we do not need ideal theory to identify them (see Section 2.2). Our judgment that the ideal wouldn’t contain slavery derives from our judgment that slavery is a grave injustice that we can feasibly eliminate – not from any analysis of what would be ideally just. And given the complex interactions between social features that motivate the search for a comprehensive conception of the ideal, it is hard to see how ideal theory can get much beyond such easy cases, identifying nonobvious features of the ideal without identifying its other features. After all, our goal is not to come up with features of society that seem ideal in isolation (see Robeyns, Reference Robeyns2008: 344) – which, due to combinatorial complexity, might very well fail to be part of a comprehensive ideal (Simmons, Reference Simmons2010: 22) – but rather to come up with features of that comprehensive ideal. And features that might seem like they would be part of the ideal might prove problematic in combination with other nonobvious features of the ideal, or once we revise our understanding of what is feasible. To return to a previous example, we might think that a conception of the ideal wouldn’t contain any compensatory institutions (say, affirmative action programs compensating for background inequalities), since there would be no injustice to compensate for. But it might turn out that fully abolishing background inequalities is infeasible, such that compensatory programs would be part of the ideal after all. Furthermore, since it doesn’t always make sense to pursue features of the ideal as short-term goals – even if the ideal would lack compensatory institutions, that doesn’t suggest we should abolish them now – the view still runs into the problem that it is difficult to work out steps toward far-off goals. It also still faces the moving target problem: As background conditions change over time, so may the institutions that are ideal in those conditions, suggesting that even if we are right about what institutions would be ideal given current circumstances, this might change.

Ideal theorists are, to be fair, sensitive to such problems, and for this reason they downgrade their aspirations still further. They insist that they don’t need to make any strong claims to certainty or finality – they can recognize that our (incomplete) ideal theory is always simply our current best guess, which is provisional, subject to change as our experiences with different arrangements teach us more about justice and about how to realize it, and as background conditions change too. Nevertheless, they claim that a provisional account of ideal theory is still useful. Ideal theorists can simply do their best to outline their conception of ideal justice in as much detail as they can, and in so doing can treat this conception as a provisional goal – a goal that may be dynamically revised over time, but that nevertheless informs our pursuit of long-term justice (e.g., Berg, Reference Berg2019: 517; Carroll, Reference Carroll2022: 543; Herzog, Reference Herzog2012: 285; see also Gilabert, Reference Gilabert2012: 43–44; Reference Gilabert, Weber and Vallier2017; Valentini, Reference Valentini2011: 312–314).

Now, recognizing that a conception of ideal justice need only be highly uncertain and provisional certainly makes ideal theory epistemically easier to pull off. But unfortunately, this move doesn’t really help to vindicate ideal target. For once we acknowledge that our current conception of ideal justice is likely to change as we attempt to realize it, we run right back into the problem that it makes little sense to treat such a conception as a long-term goal for reform. The motivating idea of ideal target is that pursuing our conception of ideal justice is the best way to promote long-term justice. But it is not clear why pursuing an uncertain, provisional conception of the ideal is an effective strategy for promoting long-term justice, worth doing at the expense of short-term justice. In retreating to this position, ideal theorists rightfully acknowledge that our conception of ideal justice is a moving target: They are “upfront that the target may need revision” such that “[w]hen the landscape shifts, there is a need to reconsider and possibly revise [its] specification” (Carroll, Reference Carroll2022: 546–547). Unfortunately, however, merely acknowledging that one is chasing a moving target does not solve the moving target problem, namely, that we have little reason to pursue a long-term target that we expect to revise before we get there: “The method of first establishing an ultimate political aim and then beginning to move towards it is futile if we admit that the aim may be considerably changed during the process of its realization. It may at any moment turn out that the steps so far taken actually lead away from the realization of the new aim. And if we change our direction according to the new aim, then we expose ourselves to the same risk again” (Popper, Reference Popper2013: 150).

Admittedly, as I have argued elsewhere, non-ideal theorists who endorse a problem-solving orientation are often insufficiently concerned with long-term justice (Barrett, Reference Barrett2020: 105–111). But the question becomes whether there is an alternative, more effective strategy for pursuing long-term justice than either putting our heads down and focusing on the problems in front of us or chasing our current best guess about how to achieve ideal justice. And fortunately, while I cannot fully make the case here, I believe that there is. Rather than trying to make progress toward some (incomplete, uncertain, and provisional) conception of the ideal, we might pursue long-term justice in a more open-ended way, implementing reforms that make our society more “progressive”: better at getting better, or more conducive to further improvements in general, though not to the achievement of any antecedently specified ideal (Barrett, Reference Barrett2020; see also Barrett and Buchanan, Reference Barrett, Buchanan, Sobel and Wall2023). For example, when deciding what goals to pursue, we might factor in not only how just their pursuit would make things in the short term, but also how much we’d learn by pursuing them. After all, social reform is a context, much like any other where we can “learn by doing,” involving a trade-off between exploiting our current knowledge of how to promote justice, and exploring the space of possible institutions, thereby learning more about how to promote justice going forward (Barrett and Buchanan, Reference Barrett, Buchanan, Sobel and Wall2023: 130–134). Of course, learning from our mistakes is only valuable if we can correct them, so we must also be concerned with flexibility or option value (e.g., Hamlin and Stemplowksa, Reference Hamlin and Stemplowska2012: 59; Wiens, Reference Wiens2012: 65–66; Reference 80Wiens2015b: 472; see also Gilabert, Reference Gilabert2012: 47, 52; Reference Gilabert, Weber and Vallier2017). Certain goals maintain our flexibility and leave options open – for example, we can put in place policies with sunset clauses, reduce the number of veto points in the political process, or establish frameworks that conduce to ongoing, flexible experimentation. Others are more likely to lock us in – for example, because they include eternity clauses, make political procedures for changing the status quo more onerous, or generate unusually powerful interest groups with a stake in maintaining them.

Unfortunately, even this is too simple, since both increased information and greater flexibility can be double-edged swords: Put toward the wrong ends, more information about how institutions work and a greater ability to alter them can be dangerous (Barrett and Buchanan, Reference Barrett, Buchanan, Sobel and Wall2023). So the key when pursuing long-term reform may be to focus not only on the extent to which particular institutional changes promote learning and flexibility but also on the background institutional frameworks in which social change occurs. And here, I have suggested that our goal should be to establish both better learning mechanisms by which we are better able to learn from our experience with different institutions, as well as better selection mechanisms that tend to stabilize features of our society that solve problems of injustice while allowing for more flexibility and experimentation with areas that need reform (Barrett, Reference Barrett2020: 120–123). This, however, is not the place to get into the details of this intramural dispute over how best to do non-ideal theory, which quickly grows complicated. Instead, it is enough to note that there are various ways non-ideal theorists can attend to long-term justice without trying to describe a conception of ideal justice to serve as a long-term target for reform: They can focus on considerations such as how much information or flexibility various short-term institutional reforms promise to generate or provide, as well as how to improve background frameworks for reform so that institutional change tends to trend in a positive direction rather than a negative direction in general.

Once we factor in such considerations, the plan of pursuing our current best guess about ideal justice (even with the intention to change directions once our information improves) reveals itself as a poor strategy for promoting long-term reform. The point here, to be clear, is not that we should never pursue less certain longer-term goals over more certain shorter-term goals, or that we should never pursue provisional goals knowing that we might sometimes change directions before realizing them – these would be much too strong conclusions to draw. Rather, the point is that once we factor in the epistemic and moving target problems, and once we recognize that different sorts of arrangements are more or less conducive to the pursuit of long-term justice in the face of them (because they facilitate more learning or maintain more flexibility, for example), a chasm opens up between our current conception of the ideal and the best target for us to aim at insofar as we wish to pursue long-term justice. In this sense, our current conception of ideal justice need not be an ideal target after all.

To make this vivid, consider that it is sometimes better to pursue even a goal that couldn’t possibly be ideal, but that would nevertheless conduce to greater long-term justice, than to pursue a goal that has some chance of being ideal but which would leave us less likely to achieve other goals if it turned out not to be, or if our pursuit of it went wrong. For example, it may be that an experimentalist society in which different jurisdictions experiment with different institutions will necessarily be non-ideal, since some jurisdictions will try out unjust institutions (compare Bennett, Reference Bennett2016). Still, this experimentalist society may be a better target for reform if it has a higher chance of improving over time and of producing a reasonably high degree of justice in the long run. This is analogous to well-known cases in ethics where it is better to pursue an option with a high chance of being pretty good than one with a small chance of being ideal. If a doctor must give a patient one of two drugs, where one has a small chance of completely curing their disease but a high chance of failing to help and causing serious side effects, while the other will very likely help considerably but without any chance of either completely curing it or producing negative side effects, then picking the second drug is typically best – even though it precludes us from obtaining the ideal outcome of fully curing the patient (compare Jackson, Reference 73Jackson1991: 462–463). Or, to take another, perhaps closer analogy: Suppose that one of several new drugs will stop a pandemic in its tracks, but we do not know which, and the others may cause severe adverse reactions. Here, we should likely run some experimental trials first, in order to determine the drugs’ safety and efficacy, rather than simply implementing our current best hunch about which drug will prove best. And this remains true even if we know that running the trials will cut us off from the ideal outcome of implementing the best drug immediately, since the trials will take some time and will postpone the end of the pandemic.

The upshot is that long-term justice is likely better pursued through flexible institutional experimentation and through working out and aiming to realize better mechanisms of engaging in experimentation and gathering relevant feedback, rather than through trying to work out steps toward our current best guess about what institutions this experimentation might eventually identify as ideal (Barrett, Reference Barrett2020; Reference Barrett2023a: 145; Barrett and Buchanan, Reference Barrett, Buchanan, Sobel and Wall2023; Dewey, Reference Dewey1927; Muldoon, Reference 75Muldoon2016; Popper, Reference Popper2013: ch. 9; see also Anderson, Reference Anderson2006; Mill, Reference Mill and Robson1977; Müller, Reference Müller2019). Since we can never have much confidence about which arrangements would be ideal in the long term, or which shorter-term steps would bring us in their direction, we should rarely treat our best guess about this as our long-term goal, with the hope that once our minds change, we will be well positioned to shift directions. Instead, we should factor in long-term justice into our selection of goals by considering how much we would learn from pursuing them, how doing so would affect our flexibility, and so on. At best, our judgments about the extent to which implementing shorter-term goals would constitute steps toward the ideal should carry some mild weight, becoming relevant in cases where there is a tie (or something close) between other considerations. If the ideal is to serve as a target, it can, at most, serve as one target among many.

3.5 Relevance without Priority

We have been focusing on ideal target, on which the ideal serves as a long-term goal for reform.Footnote ³ If this view succeeded, it would imply a sort of methodological priority of ideal theory over non-ideal theorizing about goals for reform: We can’t responsibly theorize about what shorter-term goals to pursue until we know what ideal justice would be. Now that we have seen how difficult it is to sustain this view, however, we face a further question: Might ideal theory be helpful when thinking about what goals to pursue, even if it doesn’t vindicate priority?

As with our discussion of ideal benchmark (Section 2.5), it does seem plausible that ideal theory can help with this task. Indeed, ideal theorists have emphasized two ways ideal theory might help us work out better goals for reform or solutions to problems in our world (over and above its ability to help us work out comparative criteria). First, ideal theory might help to expand our understanding of the space of possibilities, helping us come up with new and creative solutions (e.g., Gilabert, Reference Gilabert2012: 47; Reference Gilabert, Weber and Vallier2017: 114–117; McKean, Reference McKean2016; Rawls, Reference Rawls2001: 4; see also Bertea, Reference Bertea2023: 257, 260–261). For example, in thinking through a market socialist ideal in which a free market sets individuals’ pretax income, redistributive taxation equalizes posttax income, yet individuals remain motivated to work by an egalitarian “ethos” (Carens, Reference Carens1981; see also, Cohen, Reference Cohen2008: 189–196; Reference Cohen2010: 62–65; Estlund, Reference Estlund2011a: 214–224; Reference Estlund2020: 115–117), we may come to recognize the possibility of using similar ethos-based solutions to solve more immediate problems. Second, since the concrete problems we face – the ways our society is less just than it could be – are themselves often hidden from view or difficult to understand, ideal theory might help us to identify or appreciate such problems, thereby guarding against a complacent acceptance of the status quo (e.g., Estlund, Reference Estlund2020: ch. 13; Gilabert, Reference Gilabert2012: 47; Stemplowska, Reference Stemplowska2008: 332–333). If, for example, we spell out a conception of an ideally just society free of gender oppression, then comparing this society to our actual society may help us notice how our current gender relations fall short.

These two facilitating roles of ideal theory may seem uncontroversial enough. But against each role, critics have warned that ideal theory may obstruct more than it facilitates. With respect to the first role, though theorizing about ideal justice is one way to expand our understanding of the space of social possibilities, critics argue that ideal theory is liable to focus us on less relevant regions of this space: on unrealistic rather than viable solutions (e.g., Anderson, Reference Anderson2010: 3–4, Schmidtz, Reference Schmidtz2011: 775–778; Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 4; Wiens, Reference Wiens2012: 48–53). For example, though the earlier variant of market socialism is certainly interesting, the history of attempts to structure an economy without relying on material incentives should give us pause and make us consider the various ways that actual societies have dealt with their problems. More generally, we often develop more promising solutions by beginning with a careful analysis of the causal workings of problems we face and by brainstorming various ways to fix them – or by engaging in cross-historical and -societal surveys of other past and present societies with similar problems – than by imagining an ideally just society free of such problems altogether. And in this regard, ideal theory may serve more to distract than to facilitate. We may explore more relevant regions of the space of social possibilities by focusing directly on our problems and by exploring the regions of this space that hew close to social actualities than by attempting to figure out the ideal points in this space.

Second, though it is again possible that comparing an ideally just society to our own helps us notice or better understand ongoing problems, critics claim that, in practice, focusing on the ideal is more likely to blind us to them. For one thing, we often develop our conception of the ideal in response to problems we have identified. Indeed, some argue that this is not only typically but necessarily true of moral and political thinking (Anderson, Reference Anderson2010: 3; Dewey, Reference Dewey1922: 260–261; see also Appiah, Reference Appiah2017: 168–169; Muldoon, Reference 75Muldoon2016: 121; Herzog, Reference Herzog2012: 284–285; Sen, Reference Sen2009: vii–viii; Wiens, Reference Wiens2012: 53–59). But even if a conception of the ideal sometimes helps us appreciate problems in our current society, it strains plausibility to claim that turning away from our actual, problem-ridden society and focusing instead on trying to identify an ideally just, problem-free society is generally a better way of identifying and understanding our problems. For example, if the ideal would contain no oppression, then ideal theory will involve no theorizing about oppression and so will leave us ill-equipped to identify or understand subtle and insidious forms of oppression pervading actual societies. The widespread practice of ideal theory might therefore not only obstruct theorizing about justice but also serve as an ideology that obscures and so helps to stabilize actual injustice (Mills, Reference Mills2005; Reference Mills2017: ch. 5).

We return to concerns about ideology later (Section 5.4). For now, I simply note that while ideal theorists raise good points about certain practically relevant roles ideal theory can play here, non-ideal theorists also raise strong criticisms against the idea that we should give ideal theory too much weight or any pride of place in our theorizing about which goals to pursue. Once again, then, we reach the conclusion that ideal theory may be relevant to non-ideal theory, but not especially so. In this case, while ideal theory can perhaps sometimes provide a goal we should put some limited weight on pursuing, and while it may also help us to come up with other viable goals – through helping us to brainstorm new solutions or notice new problems – it does not enjoy any central role in the non-ideal task of working out suitable goals for reform.

4.1 A Gauge of Propriety

One way to respond to injustice is to deny its existence. This is the tack taken by Professor Pangloss in Voltaire’s Candide, who insists that we live in the best of all possible worlds, despite his many experiences to the contrary:

In the grip of a theory, Pangloss is unable to learn from his experiences, and so to acknowledge that anything is worse than it could be. For him, no question arises about how to respond to the imperfections of our world.

For those of us who do acknowledge such imperfections – and specifically, such injustices – further questions do arise. Some of these have come up already: How should we evaluate what would be better or worse (Section 2), or what goals we should pursue insofar as we aim for (long-term) improvement (Section 3)? But in addition to such broadly teleological questions, we also face broadly deontic ones – questions about which responses are permissible or appropriate. And here, a familiar thought occurs. While a Panglossian belief that we have already achieved the ideal certainly doesn’t help us respond to injustice, perhaps a theory of what such an achievement would consist in does. Regardless of whether the ideal serves as a benchmark of evaluation or a target for reform, perhaps it serves as a gauge of propriety (compare Moen, Reference Moen2024; Stemplowska, Reference 78Stemplowska, Lippert-Rasmussen, Brownlee and Coady2016b: 294; Stemplowska and Swift, Reference Stemplowska, Swift, Mandle and Reidy2013: 118).

On the most straightforward version of ideal gauge, ideal justice constrains what means we may permissibly take when confronting injustice because our means must, in some sense, “prefigure” the ends we pursue. For example, if our ideal is democratic, then our social movement should be internally democratic, too. Such a view is sometimes defended on instrumental grounds (e.g., Raekstad, Reference Raekstad2018). But in the present context it is more often given a deontological gloss: When pursuing greater justice, we need means that are not only effective but also permissible, and the permissibility of means may depend on their conformity or resemblance to ideal justice (e.g., Korsgaard, Reference Korsgaard1996: 148; Laurence, Reference Laurence2021: 92–95; Taylor, Reference Taylor2009: 489–490; see also Moen, Reference Moen2022; Reference Moen2024).

Unfortunately, this view runs into well-known counter-examples, resembling the second-best problems discussed earlier (Section 2.2). For example, it is a mistake to infer from the fact that the ideal would be peaceful that we should only ever use peaceful means (even in resisting, say, outright despotism), or from the fact (if it is one) that the ideal would be color-blind that color-conscious policies are impermissible even as transitional steps. As a result, defenders of this view typically concede that the means we take should reflect the “spirit” but not the “letter” of our ideal theory: They should “pay due respect to those fundamental values… that animate it” (Taylor, Reference Taylor2009: 490; see also Korsgaard, Reference Korsgaard1996: 148; Shapiro, Reference Shapiro2003: 340, 354–355). If so, however, this view now seems to run into a version of the redundancy objection (Section 2.3), since it is the fundamental values underlying our selection of a conception of justice as ideal that also set constraints on what means we may take: Ideal justice does not explain which means are permissible, but rather, both ideal justice and the permissibility of means are explained by the same more fundamental values (compare Kang, Reference Kang2016: 50–53; Stemplowska, Reference 78Stemplowska, Lippert-Rasmussen, Brownlee and Coady2016b: 294). Of course, as in the case of ideal benchmark, ideal theory may be somewhat relevant here in reflective equilibrium (for example, it may help us to better understand these fundamental values), but there is nothing special about its ability to play this role.

In the rest of this section, I therefore investigate another, subtler way that ideal theory might provide a gauge of propriety. Earlier, we saw that it is difficult to maintain ideal benchmark, on which we need to appeal to how far non-ideal societies deviate from the ideal in order to evaluate which of two non-ideal options is more just than the other (Section 2). Regardless, however, we certainly do need to invoke the ideal if our goal is simply to gauge how far (or in what ways) we deviate from the ideal. And, some ideal theorists have suggested, the extent and manner of deviations from the ideal may determine the absolute severity of injustice (Estlund, Reference Estlund, Sobel, Vallentyne and Wall2016: 15–16; Reference Estlund2020: ch. 13). This raises an intriguing possibility: Even if absolute severity isn’t so relevant when making comparative evaluations of justice, it may inform the appropriateness of our responses to injustice – in at least three ways.

First, it is common to think that we may use more violent or disruptive means when resisting more severe forms of injustice. For example, perhaps in a “nearly” just society only nonviolent civil disobedience is permissible, whereas more radical tactics become permissible in a “severely” unjust society (Rawls, Reference 76Rawls1999b: 322). If so, ideal theory may prove essential for determining which means of resisting injustice are permissible after all, since it is essential for determining the severity of present injustice (Laurence, Reference Laurence2021: 80–84).

Second, how far or in what ways we deviate from the ideal may determine appropriate responses to victims of injustice and, in particular, how to appropriately acknowledge the injustice done to them. It may be impossible to acknowledge the depth and severity of this injustice without appreciating how far or in what ways we deviate from the ideal, and so without an ideal theory (Adams, Reference Adams2023; see also Gilabert, Reference Gilabert, Weber and Vallier2017: 103).

Third, it may be fitting to adopt more extreme attitudes toward more severe injustice – for example, to lament it to a greater degree. If so, and if the severity of injustice depends on its deviation from the ideal, ideal theory may determine which attitudes are fitting. And this may in turn be relevant to what is required for virtue, insofar as virtue involves having fitting attitudes (Estlund, Reference Estlund2020: ch. 17; see also Estlund, Reference Estlund2011a: 233).

These lines of arguments are newer and have received less discussion, so my treatment of them will be briefer, more tentative, and will depart further from the existing literature. I first explain why these views controversially presuppose the denial of a thesis I call “the priority of injustice,” then provide a new argument that we should accept this thesis and so reject ideal gauge, and then address objections. Finally, I suggest a familiar lesson: Even if ideal gauge doesn’t vindicate any priority or especially important role for ideal theory, ideal theory may nevertheless be somewhat helpful in gauging whether certain responses to injustice are appropriate.

4.2 The Priority of Injustice

The core idea of the ideal gauge views under consideration is that ideal theory determines the severity of injustice, where the severity of injustice is in turn practically relevant. For now, let’s set aside the practical roles that the severity of injustice may play and consider whether ideal theory is really needed to gauge such severity. Ideal gauge rests on the following assumption: Ideal theory is prior to theorizing about the severity of injustice, because if we wish to gauge how severely unjust the status quo is, we must determine how far it deviates from ideal justice. As before (Section 2.2), this priority can be given either an explanatory or a methodological interpretation, with the former plausibly implying the latter: The severity of injustice is explained by its deviation from the ideal, so (at least when it comes to hard cases) we need an ideal theory to reach judgments of severity.

Many non-ideal theorists, however, endorse a rival view, on which injustice is prior to ideal justice (e.g., Goodhart, Reference 72Goodhart2018: 27–28; Levy, Reference Levy2016: 327–328; Schmidtz, Reference Schmidtz2011: 774–775; Reference Schmidtz2023: ch. 10; Shklar, Reference Shklar1990: ch. 1; see also Sen, Reference Sen2009: vii–viii; Wiens, Reference Wiens2012). On the simplest version of this view, there are many discrete injustices in the world of various severities, and ideal justice consists in their complete absence. Something isn’t unjust because, and to the extent that, it deviates from ideal justice; rather, something is ideally just because it lacks any injustices of any severity. This explanatory reversal also naturally pairs with a methodological reversal, on which we can understand injustice, and its severity, without ideal theory. To work out how severely unjust the status quo is, we must study the injustices within it. We don’t need to first work out an ideal and then examine the size of the gap.

Regardless of which way we go on the priority question, one surprising feature of both views just sketched is that each implies that ideal justice is not something positive, but rather the absence of anything negative: The best we can do with respect to justice is getting the severity of injustice down to zero. By contrast, it might seem intuitive that there is a difference between an ideally just society and a neutral or “not unjust” society – for example, perhaps a society is unjust to the extent that it realizes oppression, such that a neutral or “not unjust” society lacks any oppression, whereas an ideally just society would not only lack oppression but also manifest positive relations of equality (compare Anderson, Reference Anderson1999: 316–317). One reason to prefer this latter view is that, otherwise, we seem forced to conclude that the nonexistence of any beings capable of injustice is most just – it is better not to have society at all, and so no injustice, than to have a society that contains even the mildest forms of injustice. But it may seem implausible that, at the bar of justice, we should prefer nonexistence to anything short of the ideal (Schmidt and Barrett, Reference Schmidt, Barrett, Greaves, Barrett and Thorstad2025: 475).

Thankfully, both views can be modified to avoid this result. For example, ideal theorists might claim that the severity of injustice depends on the size of its gap with a “not unjust” society, where ideal justice is a higher bar. On this view, it is not strictly speaking an ideally just society, but rather a not unjust society, that serves an ideal gauge function – but the view is otherwise similar. Likewise, the priority of injustice view might claim that there are both negative injustices (such as oppression or discrimination) and positive justices (such as equal relations or freedom), with ideal justice consisting not merely in the absence of any injustices, but also in the achievement of the most positive justices possible or the best balance of justices over injustices.

For simplicity, then, I largely set aside such complications and talk as if ideal justice merely involves the absence of injustice. Such a view is compatible with claiming either that ideal justice is explanatorily (and so methodologically) prior to injustice, as ideal gauge requires, or instead that injustice is prior to ideal justice. So, both views are possible – but which is more plausible?

4.3 The Ideal Shift Test

To answer this question, I propose the “ideal shift test.” The basic idea is to contrast two ways our evaluation of the status quo’s position relative to the ideal might change. First, we might come to uncover more injustice in the status quo. For example, perhaps we previously thought that the status quo was merely economically unequal, but we now discover the existence of slavery, or that relations previously thought to be innocent are (according to our improved moral understanding) actually exploitative. Call this a “status quo shift.” Second, we might come to endorse a loftier conception of ideal justice. For example, perhaps we previously thought that ideal justice required equality of opportunity in the sense that people’s life prospects do not depend on anything but their natural talents and voluntary choices (Rawls, Reference 76Rawls1999b: 57–58, 73–78), but we now accept a more demanding conception on which natural talents are also morally arbitrary: People’s life prospects should depend only on their voluntary choices (e.g., Cohen, Reference Cohen2010: 17–19). Call this an “ideal shift.” Now, it is clear that status quo shifts increase the severity of injustice. But our question is: Do ideal shifts always increase the severity of injustice in the same way?

This is a good test of our rival views about the relative priority of ideal justice and injustice, because if the severity of injustice depends on its deviation from the ideal, the answer must be “yes”: Since both status quo shifts and ideal shifts always increase the gap between the status quo and the ideal, they must always increase the severity of injustice in exactly the same way. However, if injustice is explanatorily prior and we can make judgments about severity without reference to the ideal, then ideal shifts needn’t increase the severity of injustice, at least in all cases. Rather, it is open to us to interpret such shifts along the following lines: Whereas we previously thought some hypothetical society would be ideal because it would eliminate all injustice, we now see that the society in question would in fact contain some residual injustice, and so wouldn’t be ideal after all. In other words, if injustice is prior, ideal shifts don’t necessarily change our judgments of severity, since we can hold fixed our judgment that the status quo’s degree of injustice is (say) “7,” interpreting an ideal shift as showing that the hypothetical society we previously thought had an ideal level of “0” injustice would in fact have a severity of (say) “1.” But this interpretive move is again impossible if severity depends on its deviation from the ideal, since this view implies that we cannot make sense of a degree of injustice of “7” except by reference to its gap with the ideal.

Thus, if we find that ideal shifts don’t always seem to increase the severity of injustice in the same way that status quo shifts do, this vindicates the priority of injustice and so defeats ideal gauge. And I tentatively propose that we can indeed find cases where ideal shifts fail to alter our judgments about the severity of injustice, when we plug our judgments of severity into the practical functions touted by ideal gauge. In particular, let us assume for now that the first function holds, on which more severe injustice licenses the use of more violent or disruptive means. Now, suppose I start off thinking that violence is impermissible when resisting injustice in my society, and I undergo a status quo shift: I come to realize that my society has more injustices in it, and so is more severely unjust, than I had previously realized. In some cases, this may rightfully lead me to infer that violence is permissible after all – say, in combatting the slavery I have just discovered. By contrast, suppose we instead hold fixed everything else that is relevant to the permissibility of means – for example, what concrete injustices we are combatting and what the effects of violent and nonviolent strategies would be – and I undergo an ideal shift, coming to accept a loftier conception of ideal justice on which people’s life prospects would ideally depend only on their voluntary choices, rather than (as I previously thought) also on their natural talents. Here, it seems implausible that this ideal shift alone should occasion any change in my thinking about whether violence is permissible, and so in the severity of injustice. Differences in the facts on the ground or in the likely effects of different strategies might very well determine whether violence is permissible. But whether we accept a lofty conception of ideal justice on which we are extremely far from ideal justice, or a somewhat less lofty conception on which we are only very far, does not, by itself, seem at all relevant to whether violence is permissible here or now (Barrett, Reference Barrett2023a: 145–146).

At least at a first pass, then, the ideal shift test favors the priority of injustice thesis. Status quo shifts plausibly increase the severity of injustice, which in turn plausibly affect such practical questions as which means are permissible. But ideal shifts don’t necessarily have the same power. Ideal gauge fails the ideal shift test.

4.4 Competing Intuitions

I now consider two objections to this line of argument. The first appeals to a competing intuition that the extent of our deviation from the ideal does seem to determine whether violence is permissible; the second to (potentially) competing intuitions about the other practical functions highlighted by ideal gauge.

First, and contrary to our earlier discussion, it is a common and intuitive view that whether we are in a “nearly” just society makes a difference to whether violence is permissible. For example, a popular Rawlsian position has it that in a nearly just society, which is close to ideal justice, only nonviolent resistance is permissible. But in a more deeply unjust society, violence may be a legitimate means of resistance; the difference between the “civil disobedient” and the “militant” is that only the former sees their society as nearly just (Rawls, Reference 76Rawls1999b: 322).

However, even granting this intuition, on closer inspection, it doesn’t really vindicate ideal gauge’s claim that it is proximity to the ideal that matters. For it is crucial to Rawls’s conception of a nearly just society that it is a constitutional democracy, in which Rawls’s ideal principles of justice are widely accepted and satisfied “more or less” (Rawls, Reference 76Rawls1999b: 310) – including those principles specifying that all should have “political liberty (the right to vote and to hold public office),” “freedom of speech and assembly,” “liberty of conscience and freedom of thought,” “freedom from psychological oppression and physical assault,” and “freedom from arbitrary arrest and seizure as defined by the concept of the rule of law” (Rawls, Reference 76Rawls1999b: 53). And in such a regime, there is a plausible explanation of why only nonviolent resistance is permissible: Nonviolent resistance is an effective means of social reform, and violent means are rarely effective and often counter-productive. So, general conditions on when violence is permissible, standardly understood to at least involve necessity and proportionality, imply the impermissibility of violent resistance in such a society: Violence is not necessary, since there are other means of reform available, and it is not proportional, since it is unlikely to successfully bring about ends whose moral importance is proportionate to the moral costs of violence. In other words, while it may indeed be plausible that in societies Rawls labels “nearly just,” only nonviolent resistance is permissible, this is not because such societies are nearly just, but rather because such societies are specified in such a way that general restrictions on violence, applied in this context, render it impermissible. Someone with a completely different theory of justice who denied that a society meeting these conditions qualified as nearly just, but who agreed that in such societies nonviolent resistance was an available and effective means of promoting social reform, would similarly accept the impermissibility of violence within them.

So much for conflicting intuitions about permissible means – what about other functions proposed by ideal gauge? Consider first the view that ideal theory is relevant because it determines how to appropriately acknowledge victims of injustice. Specifically, in many contexts, it is important that we socially or publicly acknowledge injustices done – for example, when it comes to ensuring victims’ self-respect or that we stand in appropriate relations to those victims. And, it might seem intuitive that, in order to acknowledge the injustice done to victims of injustice, we need to make reference to ideal standards of justice that are not being met, since an understanding of how far and in what ways we deviate from ideal justice is essential to understanding and expressing the severity and type of injustice to which victims are subject (Adams, Reference Adams2023).

Does this view fare any better with the ideal shift test? To check, let’s adapt our previous example. Suppose we live in a society embodying a rigid caste system in which members of a certain race lack any opportunities to socially advance, as they are relegated to a second-class citizenship and to lives of poverty and degradation. Such a society fails to clear even a very low bar of equality of opportunity, such as: It is unjust for one’s life prospects to be severely constrained by one’s unchosen social identity. Now, the question is: Do we need an ideal theory to acknowledge this injustice? For example, does it again matter whether we accept an ideal theory on which one’s life prospects shouldn’t depend on anything except one’s talents and voluntary choices, or an ideal theory on which one’s life prospects should depend only on one’s voluntary choices?

Well, suppose we come to abandon the former view for the latter. And we appeal to this ideal theory to acknowledge the injustice done to members of the oppressed group: We publicly acknowledge that they are experiencing the injustice of having their life prospects depend on something other than their voluntary choices, and that this is a deeper injustice than we previously acknowledged before we adopted our loftier ideal theory. Does this successfully acknowledge the injustice these victims are facing? No – if anything, it seems downright disrespectful. After all, most members of the privileged caste are also presumably victims of injustice in this sense, since nearly everyone has their life prospects depend on more than exclusively their voluntary choices. So members of the lower caste can rightfully complain that if we appeal to this ideal standard to articulate the injustice they are subject to, this fails to capture the depth and distinctiveness of this injustice: It casts their injustice in the same terms as the injustice faced by privileged individuals. Indeed, while exactly how we should acknowledge injustice is a complex and contextual affair, depending on history, salience, the relationship between the victim and the acknowledger, and so on, I would conjecture that, at least as a rough heuristic, if we wish to appropriately acknowledge a particular sort of injustice, we should not appeal to the highest standard – the ideal standard – that is relevant, but rather to the lowest standard. For example, it does a much better job to acknowledge injustice in the case at hand to point out that these individuals have their life prospects severely constrained on account of their unchosen social identity, since this picks out a deeper and more specific sort of injustice that not everyone experiences.

The ideal shift test, therefore, again seems to vindicate the explanatory and methodological priority posited by the priority of injustice view: To appropriately acknowledge injustice, we must study the depths of the injustices in front of us, rather than contrast them with some ideal. But what about the third version of ideal gauge, on which our deviation from the ideal instead determines the degree to which it is fitting to lament injustice (Estlund, Reference Estlund2020: ch. 17)? Granting that it is fitting to lament more severe injustice to a greater degree, do ideal shifts plausibly increase the fitting degree of lamentation? I must admit that when I perform this exercise, I lack much of an intuition either way. But a good reason to deny this version of ideal gauge is that we have already seen that the severity of injustice does not plausibly depend on our distance from the ideal in the other two cases we have considered, and presumably severity should work the same way in all three cases. This is not only for the generic reason that it would be less parsimonious and so seem unmotivated to deny this, but also because such a denial would implausibly detach our actions from our attitudes, implying that when an ideal shift occurs, this renders a greater degree of lamentation fitting – even when we shouldn’t acknowledge this to victims of injustice or employ different means of resistance. We should thus reject this version of ideal gauge along with the others, absent some independent argument for thinking severity works so differently in this case.

4.5 Relevance Without Priority

The ideal shift test casts doubt on ideal gauge, and so undermines the third and final attempt of ideal theorists to establish the priority of ideal over non-ideal theory. However, as before, it would be too quick to conclude that ideal theory is entirely irrelevant when theorizing about appropriate responses to injustice – and not only because my arguments in this section have been tentative, intended only as suggestive rather than conclusive. One reason is that, in reflective equilibrium, ideal theory can sometimes impact our judgments about severity or about what injustices we currently face. For example, in thinking through what would be needed to eliminate all injustice, we might sometimes come to better understand the injustices that exist and so what responses the status quo permits – and this may be especially common in cases where ideal shifts are more radical, occurring not merely between more or less lofty conceptions of ideal justice within the same family (as in the cases we have considered so far), but between conceptions of ideal justice (say, egalitarian and libertarian theories) that recognize entirely different sorts of considerations as relevant. Indeed, even if ideal shifts do not, all by themselves, affect which responses to injustices are appropriate, it is entirely possible for them to do this through an indirect route. As we saw when discussing ideal benchmark, sometimes our understanding of the ideal is relevant to our understanding of injustice or of how best to rank non-ideal societies as more or less just (Section 2.5). And when ideal shifts do occasion such ranking shifts, this can in turn shift our views about what responses are permissible or appropriate. Likewise, we have seen that ideal theory can sometimes help us to come up with better goals for reform (Section 3.5). And a different understanding of what goals we should be pursuing can certainly impact, for example, what means we may permissibly take when pursuing these goals, suggesting another indirect route through which ideal theory may affect which responses to injustice are appropriate or permissible.

However, rather than lingering on the familiar ways that ideal theory can be relevant to working out comparative criteria or goals for reform, and so indirectly relevant to gauging appropriate responses to injustice, let us consider some more specific ways ideal theory may bear directly on the sort of considerations raised by ideal gauge. For my own part, I can think of two sorts of cases where an ideal theory may be not only relevant but necessary for adjudicating questions about the appropriateness or permissibility of certain responses to injustice.

First, it might seem that the permissibility of means depends on ideal theory, at least in the special case where it is an open question whether our society has already achieved ideal justice, such that we need an ideal theory to determine whether we have done so (compare Gilabert, Reference Gilabert2012: 47; Valentini, Reference Valentini2011: 305–306; see also Stemplowska and Swift, Reference Stemplowska, Swift, Mandle and Reidy2013: 117). After all, no means are permissible in pursuit of making our society less just. However, such a function is not relevant in the real world, where Pangloss’s view remains manifestly ridiculous: It is obvious that we are not already at the ideal. That said, a similar function may sometimes be relevant when it comes to acknowledging injustice. For even if our society is clearly non-ideal, it is nevertheless sometimes an open question whether certain individuals are victims of injustice. For example, suppose a privileged class suffers no disadvantage whatsoever that cannot be traced either to their natural talents or to their voluntary choices. Whether members of this class are subject to injustice might indeed depend on whether ideal justice requires that disadvantages trace only to voluntary choice. Thus, we find a surprising result: Ideal theory may be necessary to acknowledge injustice after all, but only when it comes to acknowledging injustices faced by relatively privileged members of society for whom nothing but an ideal theory could determine whether they are victims, since lower standards of justice do not settle the matter.

Second, even if ideal theory is not needed to determine how much it is fitting to lament injustice, it may bear on other fitting attitudes. Indeed, just as it is fitting to lament injustice, it may be fitting to hope for ideal justice. And so, at least as long as our conception of ideal justice is not literally “hopeless” (compare Estlund, Reference Estlund2020: 118–121; Gaus, Reference Gaus2021: 13–17), ideal theory may inform, in this way, what attitudes are fitting.

In the case of ideal gauge, ideal theory may therefore prove necessary for certain tasks after all. But how important are these tasks – how important is it to appropriately acknowledge injustices faced by the privileged or to hope for ideal justice (under some more specific description)? When working through the details of the various functions ideal theory might play, it is easy to lose sight of the broader methodological debates these are meant to feed into, and once we bring these into focus, it seems obvious that these functions matter little. Given all the other tasks political philosophy can pursue, it would be bizarre to claim that we should structure the discipline around the aims of working out how to acknowledge privileged victims or the appropriate objects of hope – or even, for that matter, the degree to which the status quo is lamentable, embracing a sort of “political philosophy as lamentation” (Miller, Reference Miller2013: ch. 10). When it comes to such methodological debates, other considerations are more important than the question of whether ideal theory is at all relevant to non-ideal theory or even necessary for answering certain very specific questions within it. We turn to such considerations now.

5.1 The Practice of Political Philosophy

Engraved on Karl Marx’s tombstone is the epigraph: “The philosophers have only interpreted the world, in various ways. The point, however, is to change it.” Many non-ideal theorists, regardless of their stance on Marx’s own favored changes, have taken these remarks to heart, criticizing ideal theory for its perceived lack of practical relevance. The point of political philosophy, they seem to think, is to change the world – or at least to inform our thinking about how to change it for the better.

So far, we have considered several defenses of ideal theory sharing this orientation. These have aimed to show that ideal theory is indeed practically relevant, since it serves as a benchmark of evaluation, a target for reform, or a gauge of propriety. However, we have found these responses lacking: At best, they show that ideal theory is somewhat practically relevant, not that it is especially relevant or enjoys priority. But there is another strand of the ideal and non-ideal theory debate that challenges this orientation. It suggests that “the question for political philosophy is not what we should do but what we should think, even when what we should think makes no practical difference” (Cohen, Reference Cohen2003: 243; Reference Cohen2008: 268). Ideal theory might be valuable and worth doing, then, regardless of its practical relevance, since political philosophy should simply aim to discover truth.

Taking the debate in another direction, some have argued that if the question is what sort of work political philosophers should do, we need to look at the practice of political philosophy as a whole and not only at the relevance or value of particular instances of it. For example, perhaps we should endorse a professional norm requiring that political philosophers do practically relevant work because they are the only people paid to think about justice full time (Frazer, Reference Frazer2016). Or perhaps the historical dominance of ideal theory serves as an ideology, in the pejorative sense that it helps to sustain injustice by distracting us from more pressing problems (Mills, Reference Mills2005; Reference Mills2017: ch. 5).

In this section, I survey these aspects of the ideal and non-ideal theory debate, focusing on the question of whether and to what extent political philosophers should do ideal theory. I begin by considering whether ideal theory might be valuable and worthwhile due to its nonpractical or intellectual value, before turning to questions of professional norms or obligations, and then to the topic of whether ideal (or non-ideal) theory might be problematically ideological. After summarizing these debates, which I find largely inconclusive, I end by briefly mounting my own case for ideal theory, which has surprisingly little to do with any of the other issues discussed in this Element, but instead depends on the value of methodological pluralism.

5.2 Nonpractical Value

The basic upshot of our discussion so far is that while ideal theory might be somewhat practically relevant, it is hard to see why it is especially relevant to more directly practically relevant questions of non-ideal theory – concerning what is more just than what, what goals we should pursue, and how to appropriately or permissibly respond to injustice. If our goal is to arrive at practically relevant conclusions, ideal theory is not entirely useless. But nor is it especially useful or valuable.

However, a traditional perspective holds that philosophy aims at truth (or knowledge, or understanding, or so on), and not at practical relevance. Philosophy aims to uncover deep and intrinsically valuable truths about perennial topics like beauty, justice, and the good; it does not (necessarily) aim to inform action (e.g., Cohen, Reference Cohen2003: 243; Reference Cohen2008: 268; Estlund, Reference Estlund2011b; Reference Estlund2014; Reference Estlund2020: ch. 16; Swift, Reference Swift2008: 366–368; van der Vossen, Reference van der Vossen2020; see also Erman and Möller, Reference Erman and Möller2022: 537). Indeed, this is built right into the etymology of “philosophy”: Philosophers are lovers of wisdom, not activists.

Of course, this is a controversial stance, especially when the subject under investigation is justice. Philosophers may be lovers of wisdom, but Aristotle, for example, believed that theorizing about justice was an exercise in practical wisdom, which aims at action (Laurence, Reference Laurence2021: 52, ch. 6; see also Farrelly, Reference Farrelly2007: 844; Miller, Reference Miller2013: 34; Mills, Reference Mills2005: 170; Reference Mills2017: 77; Uberti, Reference Uberti2014: 208–209; Sen, Reference Sen2006; Reference Sen2009: ix; Valentini, Reference 79Valentini2009: 333–337). The idea that political philosophy must be practically relevant or “action guiding,” and so cannot merely aim at intrinsically valuable truths as such, is more often asserted than argued for. But one recent argument holds that theorizing about justice must be practically relevant because “justice” is a normative concept that implies reasons for action (Laurence, Reference Laurence2021: 62). This account, however, is unsatisfying. For it is consistent with theorizing about justice being about reasons that it is sometimes only about the reasons of hypothetical agents in hypothetical circumstances. When we say something should be practically relevant, by contrast, we presumably mean that it should inform the reasons of actual agents (Barrett, Reference Barrett2023a: 143; see also Laurence, Reference Laurence2019; Reference Laurence2021: ch. 4; Stemplowska and Swift, Reference Stemplowska, Swift, Mandle and Reidy2013: 125–126; Swift, Reference Swift2008: 381).

Regardless, even if we grant that truth can be intrinsically valuable, and so that uncovering truths about ideal justice might be a valuable exercise for political philosophers to engage in regardless of its implications for practice, there is a further question of how valuable this is. Famously, not all truths are equally valuable: One does not gain much of value by memorizing ever more digits of π. Intuitively, it might seem important to understand what ideal justice is, since “important things are valuable to understand” (Estlund, Reference Estlund2011b: 413–415). But it is not clear whether ideal justice should really count as important in this sense, unless it derives this importance from its practical relevance – and nor is it clear just how valuable knowledge of ideal justice would be even granting its importance (compare Barrett, Reference Barrett2023b: 699; Frazer, Reference Frazer2016: 185; Gaus, Reference Gaus2016: 11–18; Laurence, Reference Laurence2021: 176).

Here is not the place to resolve longstanding disputes over whether truth (or so on) is intrinsically valuable, and if so, to what degree. Instead, I simply make two points about why such considerations do little to resolve our current debate. First, even if we are concerned with arriving at truths about justice, it is far from obvious that we should aim at truths about ideal justice. Why isn’t it at least as important to understand the more directly practically relevant truths non-ideal theory aims to uncover, which also happen to be more practically relevant? If the arguments of the last few sections are successful, then ideal theory at best plays a modest, ancillary role in helping us to uncover these truths. Non-ideal theory is generally better suited to the task. Thus, it would seem that non-ideal theory dominates ideal theory as far as value is concerned: It is more practically relevant than ideal theory and at least as valuable on nonpractical grounds. This makes it hard to see how to construct an argument from nonpractical or intellectual value to the greater (all-things-considered) value of ideal theory.

Second, however, it is similarly hard to see how such considerations really resolve the question at hand. For we do not generally think that academic or other work is only worth doing insofar as it maximizes practical (or nonpractical) value – for example, we do not generally criticize metaphysicians, mathematicians, or violinists for doing work that doesn’t bear on how to promote justice (Swift, Reference Swift2008: 367). Furthermore, if one insists that ideal theorists must choose work that better promotes justice, then presumably they shouldn’t switch to non-ideal theory: They should quit academia and become, say, politicians or activists (Estlund, Reference Estlund2020: xiv). Considerations of practical and nonpractical value therefore fail to vindicate any central role for ideal theory within the discipline – but nor do they tell us much, on their own, about what role ideal theory should play more generally.

5.3 Professional Obligations

Perhaps we can do better by turning from value to obligation. Professional ethics is the branch of ethics concerning the obligations individuals have in virtue of their professions, or the norms that should structure such professions. What does this suggest about the sort of work political philosophers should do?

One line of argument here is that taking our professional obligations seriously implies that political philosophy must remain practically relevant after all (Frazer, Reference Frazer2016). When we think about the professional norms that should prevail in a discipline, it is rarely a good idea to ask those in it why their work is valuable. For it is entirely predictable that those working within a profession will tend to intrinsically value its aims and to cite those aims when justifying the profession’s norms. Instead, if we wish to discover which norms should be in place in a profession, or what aims it should pursue, it is more fruitful to ask what role the profession should play in the broader division of labor. And here, we might think that political philosophy has not lived up to its “vocation” or role in the “division of civic labor” unless it is continuous with ordinary political discourse (Frazer, Reference Frazer2016: 178, 186–191). If political philosophers wish to get paid to study justice, then political philosophy must not devolve into an esoteric discipline, rather than one that informs and provides guidance on matters of public concern.

This is a neat argument, since it directly addresses our earlier challenge from mathematicians and violinists: Though political philosophy cannot live up to its role in the division of civic labor without remaining relevant to ordinary political practice, we don’t assign a similar role to these other professions, so the same criticisms don’t apply. However, the argument’s key weakness is that it again assumes a controversial view on the “vocation” or purpose of political philosophy, which others argue should simply be to seek the truth about politics or justice. And defenders of this alternative view worry that veering too far in the direction of pursuing practical relevance has its own dangers. In particular, insofar as political philosophers see themselves as explicitly trying to improve the world, they find themselves in the role of activists. But once one is thinking like an activist, this introduces all sorts of biases into one’s work, which may undermine one’s ability to successfully pursue the truth. The goal of seeking the truth may therefore be undermined by aiming at practical relevance, such that political philosophers may instead have professional obligations not to aim at it explicitly (van der Vossen, Reference van der Vossen2015).

Unfortunately, then, our appeal to professional obligations reproduces rather than resolves our previous conflict. Before, the dispute was about whether instances of political philosophy should be evaluated by their practical or nonpractical value; now, it concerns whether political philosophy as a discipline should be structured by norms that further practical or nonpractical aims. Insofar as political philosophy orients itself toward truths that may be of limited practical relevance, it risks becoming discontinuous with ordinary political discourse and deliberation and so failing to live up to its role in the division of civic labor. Insofar as it orients itself toward practical relevance, it risks becoming activistic and so running into biases and social dynamics undermining the pursuit of truth. Shortly, I will suggest that this tension can be overcome. But first, let us consider one alternative attempt to push past our impasse: an appeal to the idea that some sorts of theory are problematically ideological.

5.4 Ideology

Until recently, ideal theory was – at least according to one common narrative – dominant in academic political philosophy. At the same time, the discipline was made up largely of privileged folk, with an interest in maintaining this privilege (Mills, Reference Mills2005: 171–172; Reference Mills2017: chs. 8, 10). These two observations set the stage for a famous “ideology critique,” which holds that the dominance of ideal theory is an ideology “in the pejorative sense of a set of group ideas that reflect, and contribute to perpetuating, illicit group privilege” (Mills Reference Mills2005, 166; Reference Mills2017: 73; see also Geuss, Reference Geuss2008: 53–54, 90–94; Goodhart, Reference 72Goodhart2018: 52–57):

So, the thought goes, when we engage in ideal theory and focus on ideal justice rather than the pressing injustices in front of us, we are “abstracting away from realities crucial to our comprehension of the actual workings of injustice in human interactions and social institutions, and thereby guaranteeing that the ideal… will never be achieved” (Mills, Reference Mills2005: 170; Reference Mills2017: 77). Ideal theory, therefore, distorts our understanding of justice, distracts us from actual injustices, and so helps to maintain an unjust status quo.

This objection, if successful, would allow us to bypass our debate about the practical versus nonpractical values or aims of political philosophy: If ideal theory is ideological, this both undermines its practical aim (by forestalling the pursuit of justice) and its nonpractical or intellectual aim (by distorting our understanding of justice) (Adams, Reference Adams2021: 676). However, the objection is also controversial, as it relies on several speculative claims (e.g., Adams, Reference Adams2021: 685, 689–690; Boettcher, Reference Boettcher2009; Stahl, Reference Stahl2024; Táíwò, Reference Táíwò and Okeja2023; see also Shelby, Reference 77Shelby2013: 149–153) and seems to overlook cases where ideal theory has been used in an effort to help remedy injustices faced by marginalized groups (e.g., Arvan, Reference Arvan2023: 231–232; Matthew, Reference 74Matthew2017; Shelby, Reference 77Shelby2013; Reference Shelby2016; Táíwò, Reference Táíwò and Okeja2023: 198–200, 209; Valentini, Reference 79Valentini2009: 343–347). Indeed, to drive this point home, ideal theorists propose a rejoinder. An insistence on non-ideal theory may itself be ideological, because the idea that work must be judged by its practical relevance may reflect and perpetuate ideologies of capitalism (given the focus on immediate practical upshots) and managerialism (given the drive to evaluate work by easily measurable impacts), which in turn may lead to the devaluing of work in the arts and humanities and more generally to the spread and reinforcement of oppressive structures (Adams, Reference Adams2021). Or perhaps an insistence on non-ideal theory is ideological simply because it discourages work of a more radical or “utopian” nature and so risks focusing us only on marginal, incremental improvements to the status quo, leading us to overlook ways in which the status quo is fundamentally unjust and in need of radical reform (Walters, Reference Walters2025). Either way, non-ideal theory may be ideological, too.

Such arguments are, of course, empirically speculative and normatively controversial – indeed, they are offered not so much to convince us of the ideological nature of non-ideal theory, but rather to diffuse the argument that ideal theory is ideological by showing that, if arguments of this sort are permitted, parallel arguments can be constructed in either direction, and both sorts of theory seem in symmetrically bad situations (Adams, Reference Adams2021: 687, 689–690; Walters, Reference Walters2025: 75). However, even without trying to determine the plausibility of any of these accounts, we can note that they share an interesting feature in common: In order to get off the ground, each of these critiques assumes that the sort of theory claimed to be ideological is dominant in the discipline (Adams, Reference Adams2021: 691; Mills, Reference Mills2005; Reference Mills2017: ch. 5; Walters, Reference Walters2025: 91–93; see also Estlund, Reference Estlund2020: 17–18). It is the apparent dominance of ideal theory that raises risks about it blinding us to injustice. It is the apparent dominance of non-ideal theory that risks reinforcing pernicious ideologies that imply only work with an immediate practical payoff is worth supporting, or leading us to overlook cases where radical reform is needed. All such concerns would be significantly mitigated if political philosophy were pluralistic, such that neither ideal nor non-ideal theory were seen as uniquely valuable or the only game in town (Walters, Reference Walters2025: 91–93). This brings me to my own proposal.

5.5 A Case for Ideal Theory

In my view, the best case for ideal theory depends surprisingly little on the sort of arguments canvassed in this Element. Instead, it relies on the simple thought that political philosophy, like any other knowledge-producing discipline, benefits from a healthy degree of methodological pluralism. It is good for some political philosophers to do ideal theory and others to do non-ideal theory. For that matter, it is good for some to do more abstract work and others to do more concrete work, for some to work with formal models, others with empirical data, and others with more traditional methods of philosophical analysis. We are working on difficult, complex questions that have occupied our minds quite literally for millennia. It is good to have many well-developed methodologies, research programs, and approaches to answering them.

I cannot develop this position in detail here, nor can I address related questions concerning, for example, the relationship between academic political philosophy and the political thought of ordinary citizens (e.g., de-Shalit, Reference 70de-Shalit2020; see also Sen, Reference Sen2009: Part IV) or social movements (e.g., Ypi, Reference Ypi2012: ch. 2). But for present purposes, a brief sketch, focusing only on academic political philosophy, will do. My sketch proceeds in four steps. First, we may invoke the “basic research” defense of ideal theory (e.g., Estlund, Reference Estlund2011b: 407–411; Reference Estlund2020: 311–313; Frazer, Reference Frazer2016: 170–180; see also Talisse, Reference Talisse, Weber and Vallier2017). Here I don’t mean that ideal theory is “basic” in the sense that it is “fundamental” or that its conclusions can be justified on similarly strong grounds as the conclusions of basic scientific research (see Muldoon, Reference Muldoon2021: 595). Rather, I mean only that it can be defended in a similar way that basic research (for example, in fundamental physics) can be – namely, that while it is difficult to predict which particular instances of basic research will prove important, such research often proves itself quite valuable down the line. For, we have seen that ideal theory is sometimes practically relevant (Sections 2.5, 3.5, 4.5), and that it may perhaps sometimes yield important (but practically irrelevant) truths (Section 5.2). But it is hard to predict when ideal theory will yield important practical upshots or nonpractical insights, and so, for this reason, it is good to have some people doing it – for there to be an active research program in ideal theory. And for familiar reasons relating to specialization and the division of labor, it may also make sense not only for political philosophers sometimes to do ideal theory, but also for there to be some who only or predominantly do so.

Our second step notes that there are also benefits from there being some political philosophers who care more about practical relevance and others who care more about truth. For this allows us to capture the above insights from both perspectives about the vocation of political philosophy and the dangers the alternative vocation poses. Political philosophy as a discipline must remain practically relevant, if it is to remain continuous with ordinary political thought and able to guide and inform actual politics. It is thus important to have some political philosophers pushing us in this direction. However, at the same time, there are reasons to worry about an explicit focus on activism, which may bias work away from the truth. We thus benefit from having some not concerned explicitly with practical relevance, but only with truth – to help correct for these biases. And this difference in priorities will likely bring differences in focus: While those who care more about practical relevance may incline toward non-ideal theory, those who care more about truth may lack this inclination.

Third, there are often epistemic benefits to having a diversity of methodologies and approaches – and in particular to having a diversity of perspectives, heuristics, and predictive models available to solve problems (e.g., Page, Reference Page2008: chs. 6–8, 13), including problems that arise when theorizing about politics and justice (e.g., Gaus, Reference Gaus2016: ch. 3; see also Barrett, Reference 68Barrett2025b). But more specifically, in the case of ideal theory, having a diversity of methods can address concerns about ideology. If ideal theory isn’t dominant, and many work on non-ideal theory, there isn’t a worry about ideal theory masking injustice. And if ideal theory isn’t marginalized, and ideal theorists aren’t required to justify their work by its practical relevance, then we don’t face even a prima facie concern about the perpetuation of ideologies that devalue work lacking immediately actionable and measurable payoffs, or about an exclusive focus on non-ideal theory blinding us to cases where there is a need for radical reform (compare Walters, Reference Walters2025: 91–93).

As a final step, we can note that even if, at the current state of the debate, non-ideal theory seems to have the upper hand – in virtue of it having greater practical relevance and at least as great nonpractical or intellectual value – this could prove wrong. For example, my evaluation of the existing literature could be mistaken, new arguments in defense of ideal theory could arise, or research programs in non-ideal theory could prove fruitless. In general, we cannot assume that the approach that seems best right now will prove best in the long run, so it is a good thing that others see these matters differently and adopt different approaches. We benefit from having many different research programs active in the discipline, and we should generally be glad that others favor and choose to work on approaches that we don’t favor ourselves (compare Lakatos, Reference Lakatos, Worrall and Currie1978).

All of this provides a case for having some ideal theory – but how much? I am not sure. I don’t know the ideal balance between ideal and non-ideal theory in the discipline, and doubt that anyone does. Thankfully, by this point the reader hopefully agrees that we don’t need to know what would be ideal in order to make progress. Putting my own cards on the table: To the extent that I am on the “non-ideal” side of the debate, it is because I believe, first, that ideal theorists have made overly strong and unsupported claims about the practical relevance or priority of ideal theory (as I have argued throughout), and second, that, at the current margin, the push toward a greater emphasis on non-ideal theory has been, and continues to be, a good thing. Non-ideal theory is my preferred research program, and I hope it continues to develop and to attract more adherents and prestige. But while I cast my own vote for non-ideal theory, I would hate for things to become a one-party system. A healthy competition – for publications, awards, graduate students, and the like – benefits all. The dominance of any one approach does not.

Here, a residual question may occur to the reader: Setting aside the extent to which political philosophers should do ideal or non-ideal theory, how much should they engage in methodological debates about ideal and non-ideal theory? On the one hand, some argue for greater attention to these and related methodological questions in political philosophy (e.g., Floyd, Reference 71Floyd2022). But on the other hand, some worry that the assumption that political philosophers must first sort out methodology before engaging with more substantive issues is itself ideological, since it prevents us from ever getting around to pressing problems of injustice (Walters, Reference Walters2025: 88–91). For my own part, I certainly agree that methodology is important, but also deny that it should have any priority. It is good to have an ongoing, healthy methodological debate in political philosophy, running in parallel to other debates of a more substantive nature. Whether political philosophy has recently swung too far in the direction of methodology, however, is an issue on which I do not consider myself well placed to judge.

6.1 The Road Traveled

Ideal theory, on my minimal conception, involves theorizing about ideal or perfect justice. For a long time, ideal theory enjoyed great prominence in the discipline, and this was justified on grounds of its methodological priority over non-ideal theory. But ideal theory’s position has eroded in the face of a sustained critique of its alleged priority, leaving ideal theorists on the defensive – retreating to the claim that ideal theory is somewhat valuable because it is either somewhat practically relevant or has nonpractical value.

In this Element, I have canvassed this debate as it has unfolded over the past couple decades. In Sections 2–4, I argued that, at present, there is little reason to hold that ideal theory enjoys any priority over non-ideal theory – though it is also hard to deny that it is sometimes, to some extent, relevant. Ideal theory does not plausibly serve as a necessary benchmark for evaluating what is more just than what, but it sometimes helps us work out comparative evaluations (Section 2). Ideal theory does not provide us with a weighty goal that should orient our quest for long-term justice, but it can sometimes help us to come up with other goals (Section 3). And while the literature on ideal theory’s ability to serve as a gauge of propriety is less developed, I have again tentatively argued that it can – at best – sometimes, to some extent, help with this (Section 4).

In Section 5, I considered what this suggests about whether political philosophers should do ideal theory. There, I suggested that an argument for ideal theory’s greater value is lacking, and that considerations of professional obligations and ideology don’t settle the matter either. But I also made a case for ideal theory on grounds of methodological pluralism. It would be unhealthy for ideal theory to dominate the discipline, but nor should it disappear. At the current margin, shifting toward non-ideal theory is likely valuable – but this shift could one day go too far.

In light of this, let us return to our six Rawlsian theses about the value of ideal theory, so that we can see how they have fared:

• Ideal value: Ideal theory is valuable and worth doing.

• Practical relevance: Ideal theory’s value derives exclusively (or at least in large part) from its practical relevance.

• Priority: Ideal theory is practically relevant because it is prior to non-ideal theory.

• Ideal benchmark: A conception of ideal justice serves as a benchmark of comparative evaluation.

• Ideal target: A conception of ideal justice serves as a long-term target for reform.

• Ideal gauge: A conception of ideal justice serves as a gauge of which responses to injustice are appropriate or permissible.

Our discussion suggests that ideal value is to some extent vindicated, but that practical relevance is controversial – some think ideal theory has great nonpractical or intellectual value, too. We have also seen that priority is extremely difficult to sustain, and that ideal benchmark, ideal target, and ideal gauge are all true only in a weak sense: Ideal theory sometimes facilitates these roles, but it is not essential, or especially helpful, when it comes to executing them.

6.2 Rawlsian Ideal Theory Revisited

With our discussion of ideal theory’s value complete, we can also return to Rawls’s six claims about how to conceive of ideal theory:

• Ideal justice: Ideal theory aims to characterize ideal or perfect justice.

• Abstract principles: Ideal theory aims to characterize the abstract principles that would regulate the ideally just society’s major social and political institutions.

• Full compliance: Ideal theory evaluates candidate principles given the assumption that they are fully accepted and complied with.

• Favorable conditions: Ideal theory evaluates candidate principles given the assumption that they are operating in favorable economic, technological, and cultural background conditions.

• Realistic conditions: Ideal theory evaluates candidate principles given realistic background assumptions about human psychology and social life.

• Simplifying assumptions: Ideal theory evaluates candidate principles given further simplifying assumptions.

I have treated ideal justice as definitive of ideal theory, only discussing Rawls’s five other features in passing. What have we learned about them along the way?

The main lesson is that whether we should embrace such features depends on our specific theoretical goals. If our aim is to specify an upper limit on justice relative to which we make comparative evaluations, as on ideal benchmark, then it makes perfect sense to theorize about abstract principles (Section 2.3). But if we instead wish to specify a goal for reform, as on ideal target, then this is insufficient, and we must also specify concrete institutions, at least in rough outline (Section 3.2). The same goes for full compliance. Conceptually, perfect justice may involve full compliance, making it an appropriate assumption if we are treating the ideal as a benchmark – though even this is controversial, since justice may be subject to an “ought implies can” proviso, and full compliance may be beyond our capabilities (Section 2.3). But if we are treating the ideal as a goal, then there is no reason to assume that ideal justice must involve full compliance (Section 3.2). (In each case, it is less clear what ideal gauge requires, since the position is less developed and invokes a plurality of functions, so I pass over such questions here.)

What about Rawls’s other assumptions? Although it hasn’t come up much, I earlier suggested that favorable conditions is best understood as a scope assumption, representing Rawls’s idea that ideal justice may not be realizable in sufficiently unfavorable conditions (Section 1.2). But we also saw that some critics worry that ideal theorists assume excessively favorable conditions that surpass this scope in the other direction: The concept of justice may not apply if conditions are too favorable (Section 2.5). This debate depends less on our theoretical goals, but which favorable conditions we should assume does sometimes display such dependence. For example, if we are treating the ideal as a goal, then conditions shouldn’t be so favorable that we can’t realistically achieve them. A similar constraint doesn’t arise, however, if we are treating the ideal as a benchmark of evaluation.

As for realistic conditions, we have seen that Rawls intends this as a feasibility assumption, and once again the appropriateness of different feasibility assumptions depends on our theoretical aims. Given ideal target, we need to place stringent conditions on what counts as realistic – we shouldn’t pursue goals that aren’t feasible, in the sense that we could never realistically get from here to there (Section 3.2). But it is less clear whether we must assume realistic conditions if we are concerned with identifying the conceptual upper limit of justice, as we might – concerns about “ought implies can” aside – given ideal benchmark (Section 2.3).

Finally, simplifying assumptions have played little role in our discussion. This is because questions about them are not unique to ideal theory or political philosophy: All theorizing simplifies, and which simplifying assumptions we should adopt depends on our theoretical aims (e.g., Carens, Reference Carens2013: Appendix; Carroll, Reference Carroll2020; Ismael, Reference Ismael2016: 29; Leader Maynard, Reference Leader Maynard2025; Schmidtz, Reference Schmidtz and Olsaretti2018; Reference Schmidtz2023: ch. 4; Uberti, Reference Uberti2014: 208; Valentini, Reference Valentini, Weber and Vallier2017; see also Weisberg, Reference Weisberg2007). Good simplifying assumptions make it easier to theorize without undermining the reliability of our conclusions – though we often face a tradeoff here between decreasing difficulty and decreasing reliability. Rawls’s most notorious simplifying assumption, that we live in a closed society, may be appropriate if we are concerned exclusively with domestic justice and if (controversially) our theories of domestic and global justice are independent. Otherwise, it is likely inappropriate. Indeed, just about any simplifying assumption can be appropriate or inappropriate depending on our theoretical goals. For example, theorists do not typically assign names to the members of hypothetical societies. This is an obviously acceptable practice in most cases, though even it runs into trouble if we are, say, trying to investigate the biases individuals face on account of their names.

If we wish to avoid talking past each other, it is therefore important to keep track of our theoretical goals. It makes little sense debating whether we should focus on principles or institutions, whether to assume full compliance, what concessions to feasibility or “realism” we should make, or what simplifying assumptions to employ, until we have specified what function we hope our theory will play. Participants in the ideal and non-ideal theory debate often make claims about whether certain assumptions are “deficient” or “appropriate” without further qualification. We would typically be better served if we made sure always to include relevant qualifications, asking whether certain assumptions are deficient or appropriate given our particular theoretical aims – though sometimes we might also wish to inquire, of course, into the value of these aims themselves (Section 5.2).

6.3 Beyond Justice

The ideal and non-ideal theory debate began as a response to Rawls’s conception of ideal theory, and so with a focus on justice. But there are two ways the debate can and has been expanded. First, remaining within political philosophy, many of the issues we have discussed may generalize beyond a concern with justice to a concern with other virtues of social institutions. Although Rawls held that justice was the “first virtue” of social institutions, he acknowledged that “a complete conception defining principles for all the virtues of the basic structure, together with their respective weights when they conflict, is more than a conception of justice; it is a social ideal” (Rawls, Reference 76Rawls1999b: 9). Most of the questions we have asked about ideal justice can similarly be asked about such a “social ideal,” or about the other virtues it comprises, for example, about ideal democracy (e.g., Estlund, Reference Estlund2009: ch. 4; Fung, Reference Fung2005; Tomlin, Reference Tomlin2012). Indeed, some have recently argued that political philosophy should focus less dominantly on justice and become more pluralistic in this respect, too. For example, “political realists” criticize mainstream political philosophy for not focusing enough on legitimacy, as well as for its “moralistic” tendency to import pre-political ideals into (what realists regard as) the autonomous political domain (e.g., Geuss, Reference Geuss2008; Rossi and Sleat, Reference Rossi and Sleat2014; Sleat, Reference Sleat2015; Williams, Reference Williams2005) – a debate that is sometimes run together with the ideal and non-ideal theory debate, but which I have set aside here since realists (rightfully) insist that it concerns different issues (e.g., Hall, Reference Hall2015; Rossi, Reference Rossi2019; Rossi and Sleat, Reference Rossi and Sleat2014: 690; Sleat, Reference Sleat2014). Others argue that political philosophy should focus more on beneficence or charity (e.g., Barrett, Reference Barrett2025a; Buchanan, Reference Buchanan1987; Goodin, Reference Goodin2017).

The second way of expanding beyond justice involves leaving political philosophy altogether behind. And this is the direction much recent literature has taken, with debates over ideal and non-ideal theory now extending to a staggering number of fields and subfields in philosophy. However, it can sometimes be hard to tell whether such debates are continuous with the political philosophy debate, or rather separate debates using the same terms differently. “Ideal” is, after all, notoriously ambiguous, and when we cannot understand it as referring to an ideal society, it is not always clear what it means.

In closing, then, let us briefly consider how the ideal and non-ideal theory debate in political philosophy relates to these other issues and domains. The form of ideal theory that perhaps most closely resembles ideal theory in political philosophy is that employed by normative ethical theories focused not on ideally just institutions, but on the ideal moral code, as realized by informal social rules. Here, two versions of ideal theory are prominent. Ideal rule consequentialism says that we should conform to the rules contained in the ideal moral code – understood as the code that would have the best consequences if it were adhered to (e.g., Hooker, Reference Hooker2000). Ideal contractualism also says to conform to the ideal moral code, but understands this as the code that nobody could reasonably reject in favor of any other code (Scanlon, Reference Scanlon2000). Notably, however, non-ideal versions of both theories are also available. For example, non-ideal (or “actual”) rule consequentialism says we should comply with existing rules when these rules have better consequences than their absence (Miller, Reference Miller2009). Similarly, non-ideal contractualism (as we might call it) says to abide by existing rules that nobody rejects as worse than nothing (compare Gaus, Reference Gaus2010; Reference Gaus2016: ch. 4.3). It is also possible to take the same orientation toward our moral code that many non-ideal political philosophers take toward our institutions – asking how it is suboptimal and how we should reform it.

The distinction between ideal and non-ideal theory also shows up in more individualistic ethical theories. For example, some virtue ethicists claim we should emulate the ideally virtuous agent: An action is right if it is what an ideally virtuous agent would do (e.g., Hursthouse, Reference Hursthouse1999: ch. 1). But others suggest that non-ideal agents sometimes do better to avoid acting like vicious agents (e.g., van Zyl, Reference van Zyl2011). Likewise, act-consequentialists often distinguish between what the ideal moral agent would do – maximize good consequences in every case – and what non-ideal agents like us should do, given our inability to live up to this ideal: follow decision procedures or rules of thumb that tend to produce good consequences over the long run (e.g., Hare, Reference Hare1981: ch. 3). Similar issues come up in debates over “actualism” versus “possibilism.” If an ideal moral agent would promise to do something and follow through on it, but I know that, given my flaws, I won’t follow through, should I still make the promise (e.g., Jackson and Pargetter, Reference Jackson and Pargetter1986)? Possibilists say that I should still make the promise; actualists say that I shouldn’t.

Beyond ethics, a related contrast has been drawn concerning epistemic and practical rationality. An ideally rational agent might be someone who, for example, unfailingly updates their credences in line with the strictures of Bayesian epistemology, and who always chooses options that maximize expected utility. Studies of non-ideal rationality might then ask how actual, non-ideal agents should behave given limitations that prevent them from achieving this (or some other) ideal, and how this relates to what ideal agents would do (e.g., Staffel, Reference Staffel2019; see also, Begby, Reference Begby2021: ch. 3; Carr, Reference Carr2022; Greco, Reference Greco2023: ch. 8; McKenna, Reference McKenna2023; Thorstad, Reference Thorstad2024). Or, analogous to the non-ideal theoretic turn from justice to injustice, they might focus on specific “epistemic vices,” trying to better understand these and how they can be avoided or ameliorated (e.g., Cassam, Reference Cassam2019; Tanesini, Reference Tanesini2021).

In all these cases, ideal theory retains a structural similarity with ideal theory as I have understood it throughout: Although they do not try to work out what ideal or perfect justice would be, these ideal theories are concerned with what the ideal or perfect moral code, virtuous agent, or rational agent would be. In this sense, we can generalize our contrast between ideal and non-ideal theory as follows:

Here, the normative properties in question include justice, moral rules, virtue, and rationality. Others draw the same distinction between ideal and non-ideal standards of knowledge (Pasnau, Reference Pasnau2013). Ideal theory is concerned with ideal or perfect versions of these properties. Non-ideal theory is concerned with non-ideal versions of these properties, or with their non-ideal variants (as in the case of focusing on injustice rather than justice, on irrationality rather than rationality, or on vices rather than virtues) – and especially with what to do in light of failures to achieve the ideal.

Importantly, however, not all uses of the ideal and non-ideal theory distinction fit this schema. For example, in ethics, the contrast between ideal and non-ideal theory is perhaps most commonly drawn by reference to full compliance: Ideal theory asks what an individual should do conditional on everyone else doing what they should; non-ideal theory asks what an individual should do when others fail to comply with the demands of morality. A standard application is that ideal theory might suggest that each of us should contribute some modest amount to charity to solve the problem of global poverty, with our non-ideal theory asking what an individual should do given that so many others fail to do this. Am I required to “pick up the slack” and do whatever it takes to solve global poverty? Or do I merely have to contribute what I would have to contribute if I lived in an ideal world where everyone else did what they should (e.g., Murphy, Reference Murphy2000)?

Although in the context of political philosophy, the full versus partial compliance distinction may track a distinction in ideal properties – between ideal and non-ideal justice, assuming that ideal justice requires full compliance – here there is no similar property to appeal to. Rather, the contrast concerns whether we are assuming ideal or non-ideal background conditions. To see the difference, suppose we hold fixed that we are doing ideal theory in the ideal properties sense, since we are focused exclusively on what an ideally moral agent would do. Still, we can ask what an ideal agent would do given ideal background conditions (where everyone else does what they should) or given non-ideal background conditions (where others don’t do what they should). So we are drawing the contrast between ideal and non-ideal theory in a new way:

We have just seen this contrast in the case of ethics, but it has also been prominent in versions of the ideal and non-ideal theory debate that have arisen in theoretical (as opposed to practical) fields. For example, there is now a program of non-ideal philosophy of language, which attempts to theorize about language assuming various morally or politically non-ideal background conditions, such as inequality and prejudice (e.g., Beaver and Stanley, Reference Beaver and Stanley2019; Englehardt and Moran, Reference Englehardt and Moran2025; Keiser, Reference Keiser2023; Mühlebach, Reference Mühlebach2024; though see Cappelen and Dever, Reference Cappelen, Dever, Khoo and Sterken2021). A similar contrast has emerged in social ontology, between ideal theories that assume ideal, harmonious background conditions and non-ideal theories that assume conflict, oppression, and power inequalities in the background (e.g., Burman, Reference Burman2023). And, returning to ethics and political philosophy, this same contrast has also been employed by both virtue ethicists (e.g., Battaly, Reference Battaly, Alfano, Lynch and Tanesini2020; Bell, Reference Bell and Tessman2009; Tessman, Reference Tessman2005) and democratic theorists (e.g., Lepoutre, Reference Lepoutre2021; Wenner, Reference Wenner, Weber and Vallier2007) who are respectively interested in what virtue or democracy requires given unjust or otherwise non-ideal background conditions.

With these two contrasts in place, we can now draw out some general lessons. To begin, the ideal and non-ideal theory debate in political philosophy likely bears little on debates going under the same name in theoretical domains, or that more generally employ the ideal background conditions distinction. The core issue in political philosophy concerns an ideal normative property, justice, and whether a conception of that ideal property informs various forms of non-ideal theory – about comparative evaluations of justice, suitable goals for reform, or appropriate responses to injustice. It is not clear how we would even formulate analogous questions in theoretical domains, where investigations do not center on some normative property that can be ideal or non-ideal. While some insights may carry over – for example, about second-best problems – these are really separate debates.

However, when it comes to normative domains where the ideal and non-ideal distinction does concern ideal properties, the debate is more continuous. For similar questions do arise. For example, does our conception of ideal rationality serve as a benchmark for evaluating what counts as more or less rational (e.g., Staffel, Reference Staffel2019)? Should we treat the ideal moral code as a long-term target for reform (e.g., Copp, Reference Copp1995: 200)? And does our conception of ideal virtue affect the appropriateness of different responses to our own moral lapses (e.g., Johnson, Reference Johnson2003)? Of course, it might very well turn out that different answers to these questions are appropriate in different domains. But the contours of the debate are likely similar, with more insights carrying over.

That said, one distinctive feature of debates about ideal and non-ideal theory in the case of ethics and rationality is that, as we have seen, they commonly assume ideal theory is practically relevant because we ought to emulate or conform to the ideal: to act (or believe) as the ideally moral or rational agent would act (or believe), or to conform to the moral code that would ideally be realized in our society. Now, criticisms of such views are widespread, since there are many cases where what we should do seems to depend crucially on our failure to achieve the ideal. For example, an ideal moral code might prevent there from being any violence, but it would be a mistake to infer that one can never use violence in the real world – what about self-defense? (Parfit, Reference Parfit2011: 312–320; see also Pogdorski, Reference Podgorski2018). Likewise, if we fall short of ideal virtue, then presumably we should sometimes do things to improve our virtue or to compensate for our failures. But an ideal agent would never do these things since they have no room to improve and no failures to compensate for (Johnson, Reference Johnson2003). So it looks like we shouldn’t emulate the ideal after all.

I happen to find such objections compelling, but now is not the place to explore them, nor the many rejoinders to them that have been offered. Instead, I simply note that they highlight a fourth possible function for ideal theory, in addition to it playing the role of an ideal benchmark, an ideal target, or an ideal gauge. And perhaps we will uncover still further possible functions of ideal properties, which have been similarly neglected in political philosophy.

Summing up, then, we seem safe in drawing two conclusions. First, the debate over ideal and non-ideal theory in political philosophy is discontinuous with, and so at best tangentially relevant to, debates going under the same name in fields like philosophy of language and social ontology, where the distinction between ideal and non-ideal theory concerns background conditions. Second, the political philosophy debate is more relevant to debates about ethics and rationality, where the contrast often concerns ideal versus non-ideal properties. But even here, we are skipping steps if we simply import our conclusions from the political philosophy debate, especially if the ideal might play other candidate roles in these domains beyond those standardly discussed in political philosophy.