Public goods have recently received increasing attention by philosophers. In addition to work on the historical origins of the notion of public goods (Desmarais-Tremblay, 2017), their relevance to the thinking of particular political theorists (de Jongh, 2022) and the justifiability of particular public goods, such as the arts (Kessler, 2018), there have been a number of systematic attempts to develop a normative theory of public goods (see e.g., Kallhoff, 2011; Miller, 2004). Asking which public goods the state ought to provide, and how the benefits and burdens of their provision ought to be distributed, these contributions can roughly be divided into two camps. One group of authors has addressed these questions within political liberalism, remaining committed to the principle of state neutrality. By taking individual preferences as given, they note that many public goods promise net efficiency gains and ask according to what principles these benefits ought to be distributed (see e.g., Claassen, 2013; Cullity, 2008; Murphy & Nagel, 2001). In so doing, they see themselves as complementing existing theories of justice, which remain mostly silent on the allocation of costs and benefits of public goods (Miller & Taylor, 2018, p. 556). Another group of authors has moved beyond the principle of state neutrality, putting forward arguments in favor of the state provision of public goods that are at least in part independent of individual preferences, such as their potential to foster solidarity and to connect people (see e.g., Ferdman, 2018; Kallhoff, 2014; Kohn, 2020). These authors contend that the social value of public goods remains underappreciated, and that the state may be justified in providing them even if doing so does not result in efficiency gains and reject the view that efficiency is the central criterion by which to judge whether the provision of a public good can be justified. This philosophical interest has coincided with a broader sense of neglect of public goods, which has been voiced by journalists and public intellectuals, in particular in Britain and the United States (Irvin, 2012; Judt, 2010; Lehrer, 2020). Yet, despite the recent surge in attention, the philosophical and, more generally, the “scholarly literature on public goods is relatively thin” (Kohn, 2020, p. 2).1 This article aims to contribute to the philosophical thinking about public goods in three ways. First, highlighting an ambiguity in how the term “public good” is used, Section 2 proposes to distinguish between inherently public goods, which cohere to the economist's definition of a good that is non-rivalrous and non-excludable, and contingently public goods, which are non-rivalrous and, though in principle excludable, provided in a non-exclusionary form. Section 3 then draws on the notion of contingently public goods to develop two novel objections to a variety of benefit principles that have been proposed to govern the allocation of burdens and benefits in the provision of public goods. These principles are generally offered as complements to theories of distributive justice and aim to take care of an alleged blind spot resulting from the possibility of welfare gains from the provision of public good that are not required for initial conditions of justice. In engaging with these principles, the article operates at the level of ideal theory. It argues that because contingently public goods could also have been provided as club goods, the benefit principles, unless further qualified, demand unacceptable transfers to the wealthy as well as to people who object to their inclusive mode of provision. Section 4 then draws on the notion of contingently public goods to explore the potential of public goods in addressing existing social injustices. This section thus operates at the level of nonideal theory. It argues that providing public goods can be an alternative to transfer payments, or a universal basic income, in addressing injustices that are a result of economic inequality and that we have good reason to think that it can, at least at times, be more effective as well as politically more feasible. While this is true of public goods in general, several features set contingently public goods apart: they may at times be provided at little to no cost, their provision may require only more limited market interference (or even none), it can more powerfully express a commitment to status equality, and it may be particularly effective where injustices stem from unequal access to club goods. This section argues that there is an important, though frequently overlooked ambiguity in how the term “public good” is used in philosophical discourse. While some authors require a good to be in principle non-excludable in order to qualify, others merely require it to be provided in a non-exclusionary form. While there are benefits to each terminological choice, the section argues, it is important to keep them apart. Accordingly, it proposes to distinguish between inherently public goods and contingently public goods. The later sections draw on the notion of a contingently public good, attempting to show how it can advance philosophical thinking about public goods in the context of ideal as well as nonideal theory. Many philosophers have remained somewhat vague about how they understand the notion of public goods. For instance, Avigail Ferdman refers to what she calls the “standard economic understanding” according to which “public goods […] are material goods that the private market cannot provide efficiently” (2018, p. 662). Although she insists that there can also be nonmaterial public goods (2018, pp. 662–63), she otherwise relies on this characterization. Accordingly, public goods on her account constitute a market failure; but a market may fail to efficiently provide goods for various reasons and Ferdman does not specify which of these are defining of public goods. Even where explicit definitions have been put forward, however, they tend to be mutually inconsistent. The central rift is exemplified by the definitions proposed by Jonathan Anomaly (2015) and David Miller (2004).2 Anomaly writes that “[g]oods are public if they exhibit nonrivalry and nonexcludability” (2015, p. 109, italics original), invoking the standard definition of economic theory, where a good is said to be non-rivalrous if its consumption by one person does not diminish its availability to others, and non-excludable if it is infeasible to exclude some people from consumption. By contrast, Miller writes that by public goods, “I shall mean goods that are made available to everyone without charge, and that each person can enjoy without diminishing the opportunity of others to enjoy the same good” (2004, pp. 127–28). Miller's notion of a public good encompasses Anomaly's but is broader: Miller's criterion that a goods is provided to everyone without charge may be fulfilled because of the good's inherent non-excludability (think of the view of the starry sky) or because of a deliberate choice to provide it for free (think of museums in England). As a result, they will not only classify certain goods differently, but will find different questions pertinent, and even sensible: while both may ask whether the state should provide certain public goods, only Miller can sensibly ask whether it should provide certain goods as public goods. It is evident, then, that Anomaly and Miller are at risk of talking past each other, but less obvious which definition to prefer. Anomaly's definition fits with the classificatory scheme employed by economists, where public goods are contrasted with private goods (rivalrous, excludable; e.g., apples), club goods (non-rivalrous, excludable; e.g., pay-tv) and common-pool goods (rivalrous, non-excludable; e.g., fishing stocks). It also captures the market-failure aspect that Ferdman alludes to: since, by definition, no one can be excluded from the use of public goods, everyone has an incentive to contribute as little as possible. Due to this so-called free-rider problem, markets will not provide public goods at efficient levels.3 In conflating what economists call “club goods” and “public goods,” Miller appears to give up a valuable distinction. For him, a library qualifies as a public good so long as access is in fact offered to everyone free of charge. But as private collections like the London Library prove, markets can provide libraries as club goods. Even club goods typically involve a market failure: unless perfect price-discrimination is feasible, markets will not provide club goods at efficient levels. But this market failure is more limited—because people can be excluded, a market can at least in principle provide these goods. At the same time, Miller's account better captures the ordinary use of the term “public good”. It seems neither uncommon, nor unreasonable to speak of parks as public goods—but while they are often freely accessible to everyone, it is, as London's Belgrave Square Garden demonstrates, possible to restrict access to those willing (and able) to purchase a key. Because the definitions of a public good advanced by Miller and Anomaly each pick out important aspects, we have reason to preserve them both. But in order to avoid misunderstandings, we need to keep them distinct. To achieve this, I propose to distinguish between inherently public goods and contingently public goods.4 More specifically, I propose to define an inherently public good (along Anomaly's lines) as a good that is (i) non-rivalrous and (ii) non-excludable and a contingently public good (in Millerian spirit) as a good that is (i) non-rivalrous and (ii) as a matter of choice, provided in a non-exclusionary form. The qualifier “as a matter of choice” is important because in so far as inherently public goods are non-excludable, they are necessarily provided in non-exclusionary form, so the mere condition that certain goods are provided to everyone for free fails to mark the distinction. As Miller does not add this qualification, his notion encompasses both types of public goods. Moreover, I understand the term “non-exclusionary form” broadly: while a good can be provided in a non-exclusionary form by making it accessible to everyone free of charge, it can also be so provided if there is a token charge, yet one that every person can easily afford.5 The distinction between inherently and contingently public goods is important because it allows us to keep two things apart: the deliberate choice about the mode of provision of goods and the market failure resulting from the inability to exclude others. The distinction is close in spirit to, but not identical with, one that Vaughn Bryan Baltzly (2021) and Anomaly (2021) have recently drawn between public goods and publicized goods. For Baltzly, publicized goods are goods “whose ‘public’ character results only from a policy decision to make some (otherwise private) good freely and universally available” (2021, p. 376). Anomaly, discussing public health measures, characterizes as publicized “goods that are made public through the incentives created by government mandates” (2021, p. 2), where a good being public primarily means that its consumption imposes externalities on others (2021, p. 6). The notion of publicized goods, like that of contingently public goods, picks up on the fact that society can frequently decide whether to offer universal, free access to a specific good. Yet, unlike the notion of contingently public goods, it imposes no restriction on the extent to which a good is rivalrous. It is, for this reason, much broader: if a state provides apples for free, they become a publicized good, but not a contingently public good. As comes out in the discussion below, having the narrower, more precise term “contingently public goods” is valuable because it allows us to pick out those goods whose public provision promises efficiency gains. The remainder of the article is an attempt to show that the distinction between inherently and contingently public goods is valuable and that the importance of contingently public goods has been underappreciated. Three points should be noted at the outset though. First, the distinction is arguably better conceived of as demarcating a spectrum than a categorical divide: few goods are inherently non-excludable; instead, exclusion may be more or less costly. Second, whether some good is non-excludable, and for this reason inherently public, depends on the state of technology and may change6: if, in some dystopian future, people need to pay for access to biodomes, then clean air will have turned into a contingently public good. Finally, just as few goods are entirely non-rivalrous, few are non-excludable: the fact that I go for a run in the park does not diminish its availability, but if a crowd of people clogs pathways and occupies all picnic spots, this does diminish the park's availability (or value) to others—at the limit, the park becomes a rivalrous good. So, the criterion of non-rivalry should not be interpreted too strictly; instead, we may distinguish between public goods that are more pure or impure depending on their degree of rivalry—at a given level of consumption, or across a range of levels. Note, moreover, that the distinction between inherently and contingently public goods is orthogonal to several other distinctions that have been drawn within the debate on public goods. For example, Ferdman (2018) distinguishes between universal and nonuniversal public goods, where the former are valued by all whereas the latter are only valued by some. Even more prominently, some authors (see e.g., Claassen, 2013; Miller & Taylor, 2018) have drawn a distinction between essential (or necessary) and discretionary public goods, where the provision of the former, but not the latter, is required by justice.7 But the fact that some good is an inherently public good (or a contingently public one) neither affects whether it is universal, nor whether it is essential.8 This is explained by the fact that the classification of a good as inherently or contingently public is based merely on its structural properties, namely on whether it is excludable in principle or merely provided in a non-exclusionary form; but these structural properties evidently determine neither whether a good is valued by everyone, nor whether it is critical for justice.9 So, the terminological distinction advocated here is new and independent of the existing ones. At the same time, the different specifications can of course be combined; we can, for example, have a contingently public goods that is essential and universal. An important strand of the recent philosophical debate of public goods has operated at the level of ideal theory and has focused on the question of justice: according to which criteria should burdens and benefits of the provision of public goods be distributed to preserve conditions of justice (see e.g., Claassen, 2013; Miller, 2004; Murphy & Nagel, 2001)? This section puts forward two novel objections against the various benefit principles that have been proposed in response to this question. The objection from unequal access to club goods asserts that these benefit principles mandate unacceptable transfer payments to the privately wealthy. The objection from the intrinsic disvaluation of public goods asserts that the benefit principles, unless qualified, mandate unacceptable transfer payments to people who disvalue their inclusive mode of provision. Both these objections draw on the case of contingently public goods: in each case, it is the possibility of providing the respective public goods as club goods that generates a problem. More specifically, these objections draw on a specific subset of contingently public goods: those whose provision is not directly required by justice (and which are therefore also discretionary in the sense further explored below). The discussion of the demands of justice in the provision of public goods is motivated by two observations (see Miller & Taylor, 2018, esp. 562–64). On the one hand, there are some public goods that are required for justice. Consider the rule of law. Broadly understood as the effective restraint on arbitrary use of power through enforcement of legal rules and procedures, it qualifies as an (immaterial) public good. Moreover, it is presupposed by virtually all theories of justice: for what would be the point of allocating resources, if individuals were subject to the arbitrary exercise of power by others who could take them away (or impose physical harm)? On the other hand, there are goods whose provision is not required by justice, but generates a concern of justice: assuming an initially just society, how should the benefits and costs of their provision be allocated?10 Goods of the former type, “whose provision is itself a matter of justice”, are typically called “essential public goods” (Miller & Taylor, 2018, p. 564) or “necessary public goods” (Claassen, 2013, p. 273). Goods of the latter type, whose provision is not required by justice, are labeled “discretionary public goods” (Claassen, 2013, p. 273; Miller & Taylor, 2018, p. 564).11 Proponents of social theories of justice have not given much thought to discretionary public goods, and where they have considered them at all, they have typically endorsed procedural criteria.12 John Rawls, for instance, initially stipulated that the state may provide (discretionary) public goods only “when they satisfy Wicksell's unanimity criterion” (1999, p. 248), which makes their legitimate provision dependent on everyone's consent to a proposal specifying (i) the good to be provided and (ii) the tax schedule to pay for it (Wicksell, 1958). In his later work (Rawls, 2001), Rawls came to relax the unanimity condition, proposing instead a simple majority vote.13 As noted by Miller (2004), the problem with procedural approaches is that even if one thinks that they provide an account of legitimacy, they are silent on the question of distributive justice. But the surplus from the provision of discretionary public goods can be distributed in various ways—and not all appear (equally) just. In light of the weaknesses of procedural accounts, some authors have advocated substantive criteria (Miller, 2004; Murphy & Nagel, 2001). They propose to address the question of justice in the provision of public goods in two steps. First, we identify the set of essential public goods, whose provision is required by justice; for these goods, the distribution of costs is determined by our theory of justice. In a second step, we consider discretionary public goods, and it is at this stage that we need an additional principle to determine the distribution of costs and benefits. The most basic such principle is the benefit principle, which requires the provision of discretionary public goods to benefit everyone. But this principle is highly permissive; assuming self-interested voting, even the unanimity condition implies that the benefit principle is met. Two more restrictive principles have been advanced. Murphy and Nagel (2001) propose14 that every person should contribute to the cost of providing discretionary public goods in proportion to the benefits she receives; call this the proportional benefit principle. Miller (2004) proposes that discretionary public goods should be provided so as to equalize the net benefits individuals receive; call this the equal benefit principle. These principles have prompted objections. For example, Miller has criticized the proportional benefit principle as treating the state like an enterprise: insisting on contributions in proportion to benefits, he claims, contradicts the idea of a political community “whose foundational principle is that each member holds an equal stake” (2004, 144). Yet, Miller's equal benefit principle appears vulnerable to a related objection. In insisting on strictly equal benefits, one seems to adopt toward society the perspective of a jealous child. Among siblings, the equal benefit principle might have some bite: if one child does not like chocolate, but her brother does, she might insist that, if chocolates are the only sweets available, then, for reasons of justice, neither of them should get any. But at the level of society, insisting on a principle that prohibits the provision of public goods in circumstances where it would harm no one and benefit many seems to reveal a pathological insistence on an “equal share,” exposing a lack of generosity that can be expected toward fellow citizens. In addition, Miller and Taylor (2018, 568) have argued that the proportional benefit principle is incomplete. If two distinct sets of discretionary public goods satisfy the proportionality requirement, the principle does not tell us which one to choose.15 Assuming that we aim to maximize benefits, the equal benefit principle avoids this problem. But it is highly restrictive in other regards: unless benefits can be equalized, a good cannot be provided, even if everyone were to benefit from it. As Miller and Taylor note (2018, 569), this appears to render the equal benefit principle vulnerable to a leveling-down objection (though transfer schemes might help address this problem).16 Contingently public goods motivate two additional, and more fundamental objections, which apply to the various benefit principles simultaneously. First, there is the objection from unequal access to club goods: because contingently public goods are in principle exclusionary, they could be provided as club goods, but an unequal initial distribution of private resources implies unequal access to such club goods, which affects the distribution of benefits from their provision as public goods; as a result, the benefit principles mandate implausible transfers to the individually wealthy. This objection can be illustrated by reference to the case of a public swimming pool. A swimming pool that is accessible to everyone free of charge, or for a token fee, is an (impure) public good. We may initially think that all swimmers benefit from it. But suppose an individually wealthy swimmer argues that she does not benefit: because she is a member to a fitness club that offers pool access, she claims to gain nothing from a public pool. The proportional benefit principle, she holds, implies that she must contribute nothing, whereas the benefit principle and the equal benefit principle imply that she is even owed compensation. Of course, how the public pool affects the wealthy swimmer depends on the exact specification of the case. If her sole motivation for purchasing membership to the club was to gain pool access, she can cancel her membership once the public pool has been built and her benefit is captured by the money saved on the club membership. As a result, she benefits from the public pool and has to contribute to its costs of provision. Even in this case, however, part of her benefit may escape accounting. For if she had been willing to pay more for the club membership than was required, she was able to secure a surplus benefit privately, and this surplus benefit does not count toward her benefit from access to the public pool. By contrast, the entire benefit from pool access is taken into account for those who initially lacked all pool access. But, one might argue, if the initial distribution was just, then this is fine: while part of the benefit the wealthy gain from public goods escapes accounting, it is the part they had already obtained from access to the club good—so, the provision of a public pool maintains but does not increase their advantage. Next, however, consider a variation of the case. Suppose that the well-off swimmer decides not to use the public pool, but to keep her club membership. As she does not derive any benefit from the public pool, the equal benefit principle mandates a compensatory payment to the wealthy swimmer, provided there is a surplus benefit to building the public pool.17 The proportional benefit principle similarly justifies compensation if the public pool disbenefits the wealthy swimmer. Suppose that, once the public pool is built, some club members leave the fitness club, prompting a fee hike. If a wealthy swimmer decides to maintain her club membership, for example because she sufficiently values the other benefits of membership, then the provision of the public pool generates a disbenefit to her. Even according to the proportional benefit principle, she should thus make a negative contribution to the costs of providing the public pool, that is, she is owed compensation. These implications of the benefit principles appear puzzling. If an initial distribution of private goods is just, one might grant the well-off any benefit they obtain from purchasing club goods that are unaffordable to others. One might perhaps even accept that, if they prefer their respective club goods, they do not need to contribute to the costs of providing public goods. But it does not seem plausible to assign them claims to compensation. If a group of people is willing to carry all costs of making some good freely available to everyone, can they really be required to fund transfer payments to wealthy people who prefer to consume an equivalent club good, yet are not made any worse off? Now, it is important to be cautious here. If we make a judgment on whether some person can reasonably claim compensation in a particular case, this judgment will ultimately rest in part on our intuitions about this case (as well as our intuitions about relevant other cases, to which the systematic account that underpins our judgment extends). Since the argument here proceeds at the level of ideal theory, we need to ensure that we are not led astray by intuitions about nonideal cases. Perhaps, one might worry, we deem the claim to compensation unreasonable simply because, in the world we inhabit, it would be absurd to compensate those who are wealthy and already have access to club goods. But assuming an initially just allocation of private goods, might it not simply be correct that we need to compensate those who receive less benefit, or even a disbenefit, because they already had access to the relevant goods in the form of club goods? In my view, even under ideal circumstances, the demand for compensation is not plausible. The difference between cases where some are exempt from contributing and cases where they deserve compensation can be drawn by reference to the market as an alternative mode of provision. If those who favor a given public good could secure it by coordinating in a market, then they could provide it privately—in this case, the person with access to a corresponding club good would make no contribution, but they would also not be compensated. So, the outcome of the first type of case—no contribution by those with access to club goods—aligns with the market outcome, whereas the outcome of the second type of case—compensation to those with existing access to club goods—diverges from the market outcome. Of course, the mere fact that the hypothetical market outcome would involve no compensation does not imply that this is the correct view in a case where the state provides a good: the demands on the state differ from those on private market participants. But the divergence to the market-based outcome indicates that compensation is harder to justify than non-contribution, and especially where it is owed for a relative lack of benefit rather than for an indirect harm (due to costlier club goods). Note, moreover, that the state's involvement in the provision of the respective public goods may here be quite minimal—the state may simply provide an enforcement mechanism that allows those individuals who consent to chip in to collectively provide a public good that is then available to everyone. At least in such a case, even under initial conditions of justice, admitting the demand for compensation would effectively allow the better off to leverage their initial advantage to secure settlements from the worse off even though they already shoulder the entire cost of the state's provision of goods that are valuable and accessible to everyone. But this does not appear just—and even if the initial advantage attained through greater access to club goods was just. But suppose one were to reject this argument and insist that, because the state always must treat everyone equally, as soon as it gets involved in the provision of discretionary public goods, compensation is indeed owed. In this case, the argument from unequal access to club goods points to a more fundamental concern: relative to which baseline is equal treatment to be ensured? In addressing this concern, it is helpful to first summarize an insight by Claassen (2013). Claassen argues that Miller's equal benefit principle disadvantages people whose conception of the good life relies heavily on the consumption of public goods relative to market goods (2013, pp. 278–81). In a pure market economy, he notes, those who favor public goods find themselves at a disadvantage relative to those who favor market goods—few public goods are provided. But this disadvantage is preserved, if we introduce public goods, yet require that everyone benefits equally from their provision: for while everyone will now be better off, everyone will be equally better off, so relative positions remain unchanged. This casts doubt on the justifiability of the benchmark of a pure market economy. The objection from unequal access to club goods casts further doubt on this benchmark, but on different grounds. For it shows that, with a pure market economy as our starting point, part of the benefit that wealthier people obtain from the provision of contingently public goods escapes accounting—namely the part they had managed to already secure in a private market for club goods that serve as substitutes. The ben