Answers are grounded in this paper and cite their sources.
What methods should political philosophers use? One debate—henceforth just the debate—is about ideal versus non-ideal theory and asks questions such as how far should political philosophy make idealizing assumptions about reality? And how much utopian thinking is necessary or even desirable in political philosophy? One way to further advance the debate is to produce more abstract methodological arguments. But I think we should also probe how far different first-order theories in political philosophy are actually exposed to the challenges identified in the debate or how they might be adjusted to avoid them.1 My contribution here is of the latter type. Specifically, I argue that consequentialist political philosophy is a powerful position because it avoids various standard objections that have been raised to both ideal and non-ideal theory. To make this case, I proceed as follows. In Section 2, I outline a generic version of consequentialist political philosophy. According to this version, consequentialism in political philosophy is about social practices, has a comparative structure, and focuses on expected consequences. In Section 3, I introduce the debate. I first tease out different dimensions along which a theory can be ideal or non-ideal. In Section 3.1, I discuss arguments that have been raised against ideal theory. For example, Amartya Sen and others argue that utopian thinking is unnecessary, often unhelpful, and sometimes counterproductive when we compare real-world options (Sen, 2006, 2009, chap. 4). Others argue that idealized models often exclude important real-world concerns and contain blind spots. I show how consequentialism avoids those charges by focusing on expected goodness and comparative judgments and by eschewing what I will call a "privileged model" (Section 3.2). I also discuss how consequentialism encounters its own epistemic problems and offer several responses. In Section 4, I discuss charges raised against non-ideal theory. For example, Cohen (1997, 2008) argues that the real-world constraints built into non-ideal theories compromise their normative principles and bias them toward the status quo. I show that consequentialism circumvents these challenges. While consequentialism is uncompromising in its axiology, its comparative structures allow us to adjust how ideal or non-ideal we want our theories to be. This flexible yet unified structure, I argue, has independent advantages too. I conclude in Section 5. Note that my arguments also apply to theories that, while not fully consequentialist, assign significant weight to instrumental modes of institutional justification. Also, I use "political philosophy" throughout but mean for this to be interchangeable with "normative political theory." Consequentialism has a long and eminent tradition. The most prominent consequentialist political philosophers are the classical utilitarians Bentham, Mill, and Sidgwick (Bentham, 1876; Mill, 1848, 1871; Sidgwick, 1887, 1891, 1907). But consequentialist political theorizing can already be found before and outside of Western philosophy, for example, by Mozi and the Mohists in ancient China (Lum, 1977).2 As a rough distinction, we can stipulate that ethics raises normative questions about individuals and their virtues, motives, and actions, whereas normative political philosophy asks normative questions about phenomena involving social practices. This would include social philosophy, which focuses primarily on informal social practices, as well as political philosophy (narrowly conceived), which asks normative questions about institutions that include formal rules and norms (Hindriks, 2019). I here take a broad understanding of "political philosophy" that also includes social philosophy (but not much hangs on this). Practice Consequentialism: Of any two (partly) practice-constituted options available to a set of agents A, one is morally more choiceworthy to the extent that it brings about more impartial good in expectation.3 Let me spell out several features of this definition. First, the view is comparative. Its deontic criterion is not defined by what is maximally best, ideal, or ideally just. Instead, it ranks options by how "morally choiceworthy" they are, by which I mean by how much all-things-considered moral reason one has to choose them. I later show that this comparative version has important advantages when responding to concerns around ideal and non-ideal theory (particularly in Section 3.2). Moreover, a comparative version is generic and allows for different versions of consequentialism. For example, it allows both maximizing versions (as the option with the best consequences is "most choiceworthy") as well as "scalar" formulations of consequentialism.4 Second, my definition includes a variable A for an agent. A is understood broadly and can include any agent involved in influencing social practices. For example, A could be a parliamentarian, president, or policymaker, or even a broader public pushing for institutional measures. A includes both individual and collective agents (the agent variable is discussed again in Section 4.3). Third, it is common to distinguish between "objective" and "subjective" versions of consequentialism. Objective versions focus on "fact-relative" consequences and rank options by the consequences they will actually have. Subjective consequentialism, in contrast, assesses options based on how much good they do in expectation. Subjective consequentialism comes in at least two varieties: belief-relative versions focus on the expectations constituted by an agent's actual beliefs, whereas evidence-relative versions focus on the expectations it is rational to have given the evidence available. I here assume a subjective and evidence-relative version, for two reasons. First, there are good (prima facie) arguments for it. For example, an evidence-relative version is more action-guiding than a fact-relative one. When we enquire, ex ante, which policy option to choose, evidence-relative consequentialism is more likely to give us an answer, given that we do not know how facts will turn out.5 It also answers more pertinent questions than a belief-relative version. Imagine a consequentialist president withholding the COVID-19 vaccine, falsely believing that its medical risks outweigh its benefits. Applied political philosophers are probably more concerned with finding the correct evidence-relative judgment ("making the vaccine available is more choiceworthy") than with what the president ought to do given his false beliefs. Finally, belief-relative versions struggle with collective, corporate, and multiple agents: whose beliefs should be pivotal? And how, if at all, should we aggregate different people's empirical beliefs? Second, I here assume an evidence-relative version to make my task manageable. But note that variations of my arguments likely still work for fact-relative and belief-relative versions (although how is beyond my scope).6 Finally, I stipulated that consequentialist political philosophy is about ranking practice-constituted options. Such options can include (i) specific public policies (e.g., vaccination policies or tax reforms); (ii) largely informal social practices (e.g., informal gender norms, norms around inclusive language); and (iii) fundamental and broad institutions, such as questions like: What is distributive justice? What is democracy? Should we have direct or representative forms of democracy?7 Clearly, there is great consequentialist value in having social practices and institutions, as many of them make the world much better than it otherwise would be. Practices typically also come with immanent or practice-internal normativity.8 For example, the rule of law, democracy, and football matches include internal norms and normative expectations about how participants should behave. This, in part, constitutes what those social practices are. Consequentialist political philosophy, as I argue elsewhere, should focus on the normativity internal to the social practices that bring about a lot of good (Schmidt, 2023a). While a consequentialist might ultimately only care about bringing about good outcomes, she must still set great store by the institutional principles required to sustain valuable practices and institutions. For example, assume a given consequentialist believes that legal justice and the rule of law are not intrinsically valuable. But she will still recognize that those practices are extremely valuable in bringing about good outcomes. Accordingly, she will endorse the norms and institutional principles necessary for legal justice and the rule of law. So, a consequentialist political philosopher can then write about those norms and institutional principles because she would regard them as necessary or at least helpful in sustaining social practices that do good. Note that I do not assume any particular theory of the good. For example, total utilitarianism would be one way to fill in the view. But my definition allows for a much broader range of views, including theories with a wider theory of wellbeing or with non-welfarist axiologies that contain other personal goods (such as knowledge, freedom, and achievement) or impersonal values (like biodiversity). It also allows for non-utilitarian ways to aggregate personal goodness into overall goodness, such as prioritarianism. Instrumentalism: Of any two (partly) practice-constituted options available to a set of agents A, A has weighty pro tanto moral reason to choose the option that, in expectation, does more impartial good.9 Here, instrumental reasons weigh heavily but can be overridden by non-instrumental concerns. Henceforth, I only write about consequentialism, but the arguments also apply, mutatis mutandis, to instrumentalism: an evidence-relative and comparative version of instrumentalism can avoid many of the worries that have been raised about both ideal and non-ideal theories. I now introduce the debate and show how consequentialism has a plausible take on the central arguments in it. The ideal versus non-ideal distinction is complex, as several debates run concurrently (Stemplowska & Hamlin, 2012; Valentini, 2012). Here is a first approximation: when we ask "how ideal" our theorizing should be, we ask how far such theorizing should be empirically constrained, that is, constrained by facts about the actual or nearby possible worlds. However, two distinct senses of "ideal theory" are sometimes run together. First, in the limiting case, ideal theories enquire what a "perfect" or at least perfectly just society would look like. The institutional principles (or rules of regulation) that regulate such a society are then the correct and core normative principles. Call this ideal theory as utopianism.10 Anti-utopian theories, in contrast, do not bother with what is ideal but assess which feasible options are better or worse (or less unjust). Second, "ideal theory" can also mean that theorists make unrealistic assumptions so they can build a normative model. Such models might require simplification, idealization, and abstraction. Call this ideal theory as model remoteness. For example, David Gauthier argues, roughly, that some normative principles are justified because agents who act like rational choice agents (or "constrained maximizers" in Gauthier's case) and have non-tuistic preferences would agree to follow such principles (Gauthier, 1986). To construct his case, Gauthier works with a simplified and abstract model of rational agency and social interaction. Or consider Rawls. In A Theory of Justice, Rawls seeks to formulate principles of justice for the basic structure of society. For this, he devises a model with many unrealistic assumptions. For example, he assumes countries are fully closed and that all people are able-bodied, in good health, and compliant with all laws (Rawls, 1971).11 Now, utopianism and model remoteness are often related: some theorists use models remote from our actual world to formulate institutional principles that would result in ideal societies, or at least in societies far better than those in nearby possible worlds. But utopianism and model remoteness can come apart. For example, if you adopt a picture like Gauthier and develop principles based on self-regarding individual rational behavior, you might not end up with anything close to an "ideal society." Gauthier would agree that a hypothetical society of altruistic angels would be better. Similarly, while Rawls assumes people have the two moral powers, he also assumes their altruism is limited. For example, the "strains of commitment" that participants would be willing to take on constrain which principles of justice we can choose.12 So, while Rawls idealizes in the model remoteness sense, such idealization is not meant to support a far-fetched utopian ideal. Before surveying arguments against ideal theory, let me add two comments. First, some such arguments contend that a given methodology results in theoretical shortcomings. Other arguments are more practical and contend that a given methodology makes theories insufficiently action-guiding. Often enough, however, theoretical and practical concerns relate. In normative political philosophy, we typically expect theories to provide both (some) practical guidance (telling us what to do) and a theoretical justification (why we should do it). Even for more applied questions, theorists are interested not only in "picking the right option" but also in providing justificatory reasons for it. For example, imagine a black-box AI system that would reliably tell us which option is most choiceworthy but would never reveal justificatory reasons for it. While useful, such an AI system would not count as a normative theory. Second, because theories can be ideal or non-ideal in different ways and to different degrees, methodological challenges are really about theories that are either too ideal or that idealize in problematic ways (Stemplowska & Hamlin, 2012). Unnecessary: utopianism is unnecessary. When choosing between feasible options in institutional design, we can sometimes say which one is better without knowing what the ideal society would look like (Sen, 2006, 2009, chap. 4).13 I don't need to taste the perfect dish to know that apple strudel tastes better than gruel. Similarly, we do not need a theory of the best to judge what is better. Unhelpful: sometimes moving from ideal theory to comparative judgements about real-world options does not work, because the theory's ideal principles provide little guidance in cases too different from the ideal model. Theoretically Unhelpful: some ideal theories are sometimes structurally unable to provide comparative judgements. When non-ideal phenomena, like criminal justice, are outside a theory's core, an ideal theory might not contain a path from the ideal towards a non-ideal approximation procedure to assess options from a feasible set (see (Valentini, 2009) for a discussion). This problem gets compounded when real-life situations are "non-ideal" in several dimensions. For example, Rawls' theory might simply contain no answer on how we can approximate Rawlsian principles in a country marked by various "non-ideal" conditions at once, such as sovereignty conflicts, systematic gender inequality, widespread health problems, significant immigration, speciesist exploitation, and widespread noncompliance. Relatedly, in real situations, ideal theories often fail to guide us when confronted with trade-offs (Farrelly, 2007; Goodin, 1995a; Sen, 2009, 98–101; Wiens, 2020). Epistemically Unhelpful: social reality is both complex and unjust. We cannot rationally judge which of two options is more inclined to bring us closer to the ideally just long-term institutional arrangement, as social reality is just too complex. The ideally just outcome is mostly unhelpful as a guide for social reform (Barrett, 2020). Epistemically Unhelpful could allow that ideal theories contain within them a theory for comparative judgments—based on the options' propensity to bring us closer to the ideal institutional arrangements—but that epistemic complexity makes such judgments practically unviable. Bad Rankings: the best can be the enemy of the good: ideal theorising might risk giving us bad rankings when choosing among feasible options. A common dictum, often expounded by an uncle during the holidays, is that communism is a good idea in theory but does not work in practice. However, trying to approximate communism—or whatever is ideal—in the real world might make things worse rather than better. One much-discussed problem is the "failure of approximation": by using the criteria for the best outcome to approximate a good outcome in the real world, we end up with a worse outcome (Stemplowska, 2016; Wiens, 2015, 2020). Blind Spots: abstracting away empirical facts, ideal theories often assume away normatively important issues from their theories. Thereby, they arrive at principles that fail to fully account for important real-world issues. Blind Spots raises an issue for model remoteness in three ways. First—and related to Theoretically Unhelpful—treating "non-ideal" issues as being outside its core, ideal theory might fail to guide us on how to approach such issues in the real world. Second, which issues to exclude in ideal settings is itself a normative question. Ideal theorists might argue that all scientists must abstract and idealize when building a model. However, in normative theory, excluding certain real-world issues from a theory's core is rarely normatively neutral. For example, excluding bad health, structural gender inequality, or the domination of non-human animals from our theory's core and treating those as peripheral issues requires a normative justification. Finally, excluding such issues from the core model might simply give us the wrong institutional principles. If normative principles fail to capture important real-world problems—or treat those as peripheral—are they the right principles? Ideological Capture: by including blind spots in their normative theorising, ideal theories can produce principles that hide rather than draw attention to real-world problematic power relationships, injustices, and so on. Thereby, ideal theory can help reproduce status quo problems and privileges (See (Mills, 2005) for example).14 I now argue that consequentialism has good answers to all the above challenges.15 Several features of consequentialism, as described in Section 2, allow it to meet the challenges posed to (overly) ideal theory. I first describe two such features and then address those challenges one by one. First, consequentialism is "non-ideal" about normative institutional principles. In Section 2, I argued that consequentialist political philosophy is primarily about practices and institutions that bring about good outcomes. Institutional principles tend to come in when one wants to characterize the practice-internal normativity of such practices. And what practice-internal normativity such practices will have—say the practice of democracy—is in large part determined by empirical considerations. So, institutional principles are thus not "ideal a priori principles" but are always connected to, and contingent on, empirical evidence about the effects and practice-internal normativity of real-world practices. Second, consequentialism is essentially comparative. Rather than starting out by enquiring which institutions are fully just or unjust, or good or bad simpliciter, consequentialism tells us which ones are more choiceworthy than others.16 Now, compare consequentialism with views in political philosophy, such as Rawls or Gauthier, that rely on model remoteness. As mentioned earlier, such views derive normative principles through an abstract and idealized model. Such Call such theories privileging theories, because they privilege a particular model relative to which they formulate their core principles. For example, Gauthier develops a model of interactions between rational agents. When we assess real-world institutions, we assess them against the principles developed in this privileged model. Or consider Nozick. Nozick posits principles for the just acquisition and transfer of property. Imagining a hypothetical scenario of an original acquisition along with principles for just transfer and compensation, we can then assess real-life property distributions against this privileged hypothetical model. Finally, Rawls has a privileged model of society as a system of fair cooperation between free and equal persons. The hypothetical contract behind the veil of ignorance between able-bodied persons within a closed country is a vehicle for articulating institutional principles for this privileged model. Consequentialism, in contrast, does not have a privileged model. The standards consequentialism uses to assess outcomes and options do not require such an idealized model. The above two features help consequentialism meet the objections raised to ideal theory. Start with Blind Spots. Blind Spots was the charge that by formulating institutional principles relative to an idealized and remote model, ideal theories structurally fail to give sufficient importance to important real-world issues. Given its two features outlined above, consequentialism avoids this challenge. First, being non-ideal about institutional principles forces consequentialism to focus on expected effects. And when estimating expected effects, we must consider all axiologically relevant effects, that is, all factors that influence the expected value of an option. And such effects will include "non-ideal" factors like structural gender inequality, disability, ill health, partial compliance, disagreement about justice and the good, human irrationality, migration, speciesist exploitation, and so on. Accordingly, we cannot rank institutional set-ups by ignoring such factors, because they all influence how likely and good different outcomes will be. Non-ideal factors do not come in after the theorizing is done, they are built into the theorizing itself. Second, privileging theories have a remote model of society relative to which they formulate their core institutional principles, which are then used as standards to assess options. Inevitably, a remote model must idealize and abstract away from important real-life normative concerns, which means core principles will always either entirely neglect those concerns or treat them as peripheral. Consequentialism, in contrast, does not formulate any core institutional principles relative to a privileged remote model—it is essentially comparative. Because it does not require a remote model for its evaluative standards, consequentialism also avoids neglecting important real-world issues. Its focus on expected goodness, its absence of a privileged remote model, and its comparative structure give consequentialism a structure that does not build blind spots into its theoretical structure. Note that I do not argue that its structure guarantees that consequentialist theorists will not have blind spots. Epistemic accidents happen, and they happen for all theories in political philosophy: whatever your favorite theory, the complexity of the real world, the fallibility of human beings and their vested interests, and our limited time and resources virtually ensure you will miss something relevant. A consequentialist theorist, for example, might miss important empirical considerations. Or she might come to learn that her favorite axiology was mistaken because it excluded important real-world issues. A good theoretical structure alone does not guarantee good theorists, but it is an advantage all the same.17 In Bad Ranking, the worry was that ideal privileging theories might give us bad rankings for options far from the privileged model. This does not apply to consequentialism, because it is essentially comparative. There is no privileged model nor can any "ideal scenario" change the rankings of two options. In Unnecessary, Sen argued that we do not need a theory of the ideal to make comparative judgments. This is not a problem for consequentialism, because consequentialism is comparative and thus does not require a theory of the ideal. While the above is already sufficient, I think consequentialism offers an even stronger response to Unnecessary, once we appreciate that Sen's argument has a problem. I show, first, what that problem is and, second, how consequentialism can fix it. The problem with Unnecessary is that we still need a theory. Without a theory, our comparative judgment would be merely context-dependent and intuitive. To be clear, such a judgment might still be epistemically justified even without any deeper theory. However, remember that I earlier mentioned that normative theorizing is not just about "picking the right option" but also about providing justificatory reasons for it. So, when we do political philosophy, we typically aim to provide a theoretical justification for a comparative judgment. For this, we need a theory. Now, when we provide a theoretical justification of why one option O1 is better than another option O2, we must say something more "general" about the reasons that apply to those options. And when we provide justificatory reasons, those reasons must apply more generally.18 So, we do need principles that go beyond O1 and O2. For example, imagine you say that current incarceration rates in the United States are too high and that it would be better if they were lower. This seems intuitively right. So, arguably, your judgment is epistemically justified without requiring a deep theory. Yet when we try to provide a theoretical justification for this comparative judgment, it seems we need to invoke more general principles. Call this the Generalist Argument. Now, someone might argue that the Generalist Argument brings us right back to ideal theory: how are we going to get general principles for a just society unless we build a somewhat idealized model? As the positive sciences show, idealized and abstract models often provide general explanations that go beyond highly context-specific judgment. When an economist builds a model of a market, for example, she must abstract from various context-specific features. Should we not say the same for political philosophy? And, if so, does this not again require a privileged model?19 Being both generalist and comparative, consequentialism shows that the Generalist Argument does not lead us back to ideal privileging theory. Because its axiology is both generalist and comparative, we can rank two options under consequentialism while also providing a deep generalist theory to justify this ranking. For example, when we say that the US should have lower incarceration rates, we need not refer to principles of justice that would apply in a privileged model, let alone to some utopian society. Instead, we point to axiological features of a possible world with lower incarceration rates and point out how that possible world would be better than our actual world. At the same time, these justificatory reasons apply generally and can thus suitably feature in a theoretical justification. So, consequentialism is a generalist theory that provides justifications for comparative judgments without requiring utopianism or some remote idealized model. Finally, Unhelpful was the worry that some ideal theories are not particularly helpful for evaluating institutional options in our unjust society. Theoretically Unhelpful contended that some ideal theories lack a "theoretical path" from the ideal to comparative judgments, particularly when our world differs from the ideal along so many different dimensions. Consequentialism sidesteps Theoretically Unhelpful. First, as explained above, with their connection to expected value, consequentialist comparative judgments must always consider all real-world factors that influence expected value. So, it does not start with an ideal model and then "approximates" in the real world, but starts with the real world right away. Second, and most importantly, consequentialism provides comparative judgments between non-ideal options because its structure is comparative rather than approximative. So, unlike approximative theories with a privileged remote model, real-world comparative judgments are baked into consequentialism's very structure. Both of these factors mean consequentialism does not require a privileged model to derive the standards with which to compare options. It has those standards without any privileged model and thereby circumvents Theoretically Unhelpful. Two worries have been expressed to me about this line of argument. Addressing those will further clarify my argument. First, imagine, for the sake of argument, that consequentialism ended up recommending socialism as the best institutional arrangement because it contains the highest well-being of all possible institutional arrangements. Would consequentialism then not recommend that we evaluate options based on their propensity to lead to socialism? Would consequentialism not be approximative after all? Of course, one could respond that a decision-making procedure that single-mindedly "aims for the best" would not be endorsed by consequentialism: even if socialism is uniquely best, we should also consider the "costs" along the way and the probabilities with which our actions lead to very bad outcomes (such as totalitarian capture). But even leaving such worries aside, the central point is that the socialist approximative heuristic would still just be a decision-making heuristic and not a standard with which to evaluate options at a deeper level. "Socialism" does not replace the theoretical standard with which we assess options—the standard is given by consequentialism. "Aiming for socialism" would play a role quite different from
No takes yet. Share an insight, caveat, or question.
Andreas T. Schmidt (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: