At the heart of the debate over distributive justice is the question of whether people have full private property rights over their holdings (or might readily acquire such rights). Libertarians famously answer this question in the affirmative: they maintain that at least some people have unilaterally appropriated natural resources such that any non-consensual use or redistribution of those resources would be wrong. They, thus, reject all rival principles of distributive justice that permit such redistribution. Given that the successful appropriation of private property negates all rival principles in this way, a crucial question for those interested in distributive justice is whether such appropriation has, in fact, taken place.

So what are the necessary and sufficient conditions of converting unowned natural resources into private property? John Locke famously answered this question by positing what has become known as the Lockean proviso: a person has the normative power to unilaterally appropriate natural resources ‘at least where there is enough and as good left in common for others’ (2005, § 33).Footnote 1 This proviso has proven popular among libertarians, though there is significant intra-libertarian debate over what counts as ‘enough and as good’. Left-libertarians maintain that the appropriator must leave enough such that each non-appropriator could appropriate a share that is, in some specified sense, equal to the appropriated share.Footnote 2 By contrast, right-libertarians endorse some version of the following principle, which, following Attas (2003), might be called the negative principle of appropriation:

NP Person P has the power to unilaterally appropriate some natural resource R by φ-ing just in case, for any non-appropriator Q, such appropriation does not leave Q worse off.Footnote 3

The reason that it was only claimed that right-libertarians endorse some version of NP is that NP is ambiguous, with there being multiple ways of interpreting a number of its elements. This point has been made most extensively by Attas, who notes that there are five elements of NP that require additional specification if the principle is to have determinate content. First, one must specify the respect in which Q is no worse off in the appropriation world relative to the comparison world, i.e., one must specify the currency of comparison (pp. 356–358).Footnote 4 Second, one must specify exactly what it is that must not leave Q worse off (e.g., the mere ownership of R vs. that and the act of φ-ing) (pp. 358–360). Third, one must specify the relevant set of non-appropriators from which Q is drawn (e.g., the set of all existing persons vs. the set of all persons who either exist or will exist) (pp. 360–363). Fourth, one must specify the counterfactual baseline relative to which Q must not be left worse off (e.g., the world where R goes unappropriated vs. the world where it is appropriated by someone else) (pp. 363–368). And, finally, one must specify how persistent Q’s non-worsening must be (e.g., Q is left no worse off at the time when P φ-s vs. many years in the future irrespective of Q’s post-appropriation choices) (pp. 368–371).

For each of these elements of NP, Attas argues that there is a weak specification of that element that makes it plausible that NP can be satisfied. However, he maintains that right-libertarians cannot make use of these weak specifications, either because the specifications conflict with other core libertarian commitments or because they are independently implausible. By contrast, there are stricter specifications of each element that are more acceptable. The problem for right-libertarians is that a strict specification of NP would be satisfied in only a small number of possible worlds at best—i.e., in most worlds (including ours), no one has the power to unilaterally appropriate unowned resources. Thus, Attas concludes that right-libertarians cannot appeal to NP to ground the power to appropriate.

There are, however, three plausible ways of specifying NP that Attas either does not consider or does not adequately address. The purpose of this paper is to introduce these specifications and argue that they, similarly, are either unacceptable, do not support the right-libertarian position, or go unsatisfied in most (if not all) possible worlds. It, thus, concludes that there is no available interpretation of NP that might ground the existence of the kind of property rights favored by right-libertarians, thereby restoring the possibility of permissible egalitarian redistribution.

The paper proceeds as follows. The subsequent section—labeled ‘The Non-Subjection Specification’—introduces a novel currency of comparison proposed by Bas van der Vossen (2021): Q must be no more subjected than she would be absent appropriation. While this proposal avoids Attas’s objections to other currencies of comparison, this section argues that this specification of NP is not satisfied in any world where exhaustive appropriation has occurred. Rather, it will only be satisfied if latecomers are left an equal share of natural resources to appropriate. It, thus, supports the aforementioned left-libertarian position rather than the right-libertarian one.

The next section (‘The Systematic Specification’) considers the popular interpretation of NP that holds that one must consider whether Q is worse off in the world where P appropriates as part of a larger system of private property vs. the world where there is no such system. While Attas presents a brief objection to this specification in his paper, this section argues that he is a bit too quick, and the proposal is more plausible than he gives it credit for. However, it then goes on to present a more detailed explanation of why this interpretation of NP is unacceptable.

Finally, the section labeled ‘The Compensation Specification’ considers a specification of NP whereby Q is judged to be no worse off if appropriation benefits her in a way that fully compensates her for any associated harms. Attas argues that this interpretation is incompatible with other libertarian commitments regarding the nature of property rights (namely, that people do not forfeit property rights in virtue of inefficient use). However, this section argues that these conflicting commitments could be tolerably abandoned by libertarians who wish to preserve NP. Thus, some other argument is needed if one wishes to reject NP. This section goes on to provide such an argument, contending that appropriation fails to compensate Q—at least, given a standard notion of compensation. While there is an alternative sense of compensation that does allow for appropriation to compensate Q, the section contends that an NP that employs such an interpretation is implausible.

Given that the three proposed specifications of NP are either implausible, fail to support right-libertarianism, or go unsatisfied, the paper concludes that right-libertarians must find some other way of specifying NP if the principle is to ground the existence of inegalitarian property rights. However, given the difficulties presented below, it is not clear that such a specification of NP can be found. Thus, this paper provides at least prima facie reason for abandoning right-libertarianism in favor of either left-libertarianism or some non-libertarian principle of distributive justice.

The Non-Subjection Specification

NP insists that appropriation cannot leave any non-appropriator Q worse off relative to some non-appropriation baseline. But, worse off in what respect? A natural answer to this question is that Q’s level of welfare in the appropriation world must be at least as high in the baseline so as to compensate Q for the loss of her moral permission to use natural resource R. However, Attas objects that libertarians cannot interpret NP in this way for two reasons. First, libertarians generally deny that welfare considerations can ground people’s moral requirements (2003, p. 360).Footnote 5 And, second, having welfare-based compensation as a necessary condition of appropriation ‘amounts to a duty to improve owned resources beyond their natural state’—a duty that libertarians would similarly reject (p. 358). Further, Attas argues that the most plausible rival specification of NP’s currency of comparison—namely, the extent to which Q is at liberty to use resources comparable to R—will limit appropriation in a way that right-libertarians cannot accept (p. 358). Thus, he concludes that one cannot specify the currency of comparison in a way that renders NP compatible with other core libertarian commitments.

However, in a recent paper, Bas van der Vossen (2021) has proposed an alternative currency of comparison that seemingly avoids Attas’s objections. Van der Vossen concurs with Attas’s view that welfare is an inappropriate currency of comparison for libertarians (2021, pp. 187–188). According to van der Vossen, the point of Locke’s proviso (as articulated by NP) is to ensure that people are not subjected, where a person is subjected with respect to some right to the extent that her ability to exercise that right depends on the choices of others (p. 193). To be a bit more precise, van der Vossen contends that there are three components to non-subjection: having the Hohfeldian moral permissions that we need to ‘live our lives’, having an immunity from the loss of those permissions, and having the de facto opportunity to do the permitted things (p. 191).Footnote 6 Additionally, he seemingly takes non-subjection to also entail an immunity from the loss of select Hohfeldian powers in addition to permissions. Specifically, van der Vossen takes people to be subjected when others can deny them either the de facto or de jure opportunity to exercise their right to appropriate resources (pp. 191–192). Given that appropriation is an exercise of a Hohfeldian power (as it establishes new rights), a requirement that others lack the de jure authority to limit the use of this power seemingly entails that all non-subjected persons have a Hohfeldian immunity from the loss of this power—i.e., the absence of such an immunity entails subjection.

Van der Vossen contends that the Lockean worry about excessive initial appropriation is a worry about subjection: over-appropriation allows appropriators to prevent latecomers from exercising their right to acquire property. If, for example, a cartel has appropriated all existing resources, then latecomers would only be able to acquire property if the cartel agreed to yield certain holdings (p. 193). Given that latecomers’ ability to exercise their right entirely depends on the cartel’s choices, they are unacceptably subjected by the cartel’s prior initial appropriation.

To preclude the possibility of such subjection, van der Vossen posits a specification of NP that has non-subjection as its currency of comparison. Specifically, he contends that appropriation can occur only if that appropriation leaves Q as non-subjected as she would be in a world without appropriation (p. 194).Footnote 7 One advantage of this specification of NP is that it seemingly avoids both of Attas’s objections to welfare-based interpretations of NP. More provocatively, van der Vossen contends that this interpretation can be satisfied even in a world where all resources have already been appropriated so long as there is an adequately competitive labor market. Given such a market, latecomers’ ability to acquire property would not depend on someone else’s choices: while perhaps one property owner might decline to hire a latecomer, there will be countless competitors who would be willing to give her resources in exchange for her labor. Given that latecomers’ ability to acquire property does not depend on any other particular person’s choice, they are no more subjected than they would be absent appropriation (p. 201). By contrast, in a monopsonistic labor market where a single person has appropriated all the resources, latecomers would depend on that person’s acquiescence to acquire property (p. 200).

While specifying NP in terms of non-subjection avoids Attas’s objections to welfare-based NPs, the problem with this proposal is that it appears to fail on its own terms. Specifically, the non-subjection NP turns out to be incompatible with exhaustive appropriation: a proper subset of persons cannot exhaustively appropriate all available resources while still having all of their respective appropriations satisfy van der Vossen’s proviso. And this is true even if those appropriators establish a competitive labor market where latecomers’ ability to acquire property does not depend on any single person’s will.

To see why, begin by noting that natural resources are appropriated sequentially, where appropriators can be assigned labels that designate the order in which they appropriated. Let ‘Z’ stand for the first latecomer to arrive on the scene after exhaustive appropriation has occurred. Let ‘Y’ then stand for the last person to appropriate, with ‘X’ referring to the last person to appropriate prior to Y, ‘W’ referring to the last person to appropriate prior to X, etc. For simplicity, consider a three-person world composed of X, Y, and latecomer Z, where X and Y have exhaustively appropriated all available resources. In such a world, there are three possible situations that might obtain:

Case 1

X and Y independently refuse to hire Z.

Case 2

X refuses to hire Z and Y offers to hire Z (or vice versa).

Case 3

X and Y independently offer to hire Z.

In Case 1, Z is straightforwardly subjected vis-à-vis her right to acquire property.Footnote 8 She cannot acquire property via initial appropriation due to the fact that all existing natural resources have already been acquired. And she cannot acquire property via just transfer due to the fact that X and Y are unwilling to hire her. Thus, Z lacks any way of exercising her right to acquire property and is therefore subjected. Given that Z is subjected in Case 1—but, by hypothesis, is not subjected in the non-appropriation world—it follows that Y’s appropriation violated the non-subjection NP and, thus, could not have actually occurred.

Similar remarks apply to Case 2. Unlike in Case 1, Z is able to acquire property in this case via just transfer. However, given X’s (or Y’s) refusal to hire Z, Z’s ability to acquire property through transfer is entirely dependent on Y’s (or X’s) will: if Y (or X) were to change her mind and refuse to hire Z, then Z would not be able to exercise her right to acquire property. Thus, Z is subjected in this case as well, where this, in turn, implies that Y’s posited appropriation could not have occurred due to its violation of the non-subjection NP.

What about Case 3, where both X and Y are willing to hire Z? In this case, Z is not subjected, as her ability to acquire property does not depend on any single person’s will. Even if Y changed her mind and refused to hire Z, Z could still acquire property by working for X instead. Similarly, if X were to change her mind, Y would still hire Z. Thus, Z is not subjected because she is able to acquire property no matter what any other person decides. Given that Z is not subjected in the case, it follows that Y’s appropriation satisfies the non-subjection NP, with Case 3 thereby seeming to instantiate van der Vossen’s contention that exhaustive appropriation is compatible with non-subjection.

However, this conclusion is a bit too quick. While it is true that Z is not subjected in Case 3, it turns out that Y is subjected in this scenario, as her power to appropriate is a function of X’s will. To see why, note that X is able to decide whether Case 2 or Case 3 obtains: by simply changing her mind and refusing to hire Z, she can transform Case 3 into 2 wherein the non-subjection NP precludes Y from appropriating.Footnote 9 Thus, while Y is not precluded from appropriating any resources in Case 3, this is only because X allows Y’s power to appropriate to remain intact (by choosing to hire Z if given the opportunity). Given that Y’s power to appropriate depends on X’s will in this way, Y is subjected in Case 3.Footnote 10 In other words, X’s appropriation realizes a world where Y is subjected (because Y’s ability to appropriate depends on X’s willingness to hire Z). However, this runs afoul of the non-subjection NP. Thus, to avoid contradiction, one must reject the assumption that X successfully appropriated her posited portion of resources.

The foregoing argument has shown that, in a competitive three-person labor market comprising two buyers and one seller, exhaustive appropriation by a proper subset of persons is incompatible with the posited non-subjection NP. But what about larger labor markets where there are millions of buyers such as one sees in advanced capitalist economies today? Perhaps surprisingly, a very large number of labor buyers does not eliminate the incompatibility: the non-subjection NP still precludes exhaustive appropriation irrespective of the size of the labor market.

To see this, consider a new case—call it Case 4—that is identical to Case 3 except for the fact that an additional appropriator/labor buyer W has been added to the scenario. Given this addition, there are two possibilities: either W is willing to hire Z, or she is not so willing. If the latter possibility obtains, then this scenario is transformed into Case 3, as W functionally removes herself from the labor market. Without W to hire Z, it remains the case that Y’s power to appropriate is a function of X’s will, as X refusing to hire Z would still make Y’s appropriation a violation of the non-subjection NP and thereby preclude Y from appropriating. Thus, X would still be unable to appropriate in this scenario. Alternatively, if W is willing to hire Z, then Y is no longer subjected. However, now X is subjected, as her ability to appropriate becomes a function of W’s will: because X’s appropriation violates the non-subjection NP if W chooses not to hire Z, W is able to preclude X from appropriating by making this choice. This, in turn, entails that W’s appropriation violates the non-subjection NP and, thus, could not have occurred.

Informally, one can see that there is a pattern here that will persist irrespective of how many people are added to the scenario: each added person will, conditional on successful appropriation, have a choice as to whether or not to hire Z, with either decision leaving someone subjected. Thus, no matter how many people are added, exhaustive appropriation will violate the non-subjection NP.

More formally, call an appropriation scenario non-exhaustive if it is one where a would-be appropriator E is precluded from appropriating available resources because such appropriation would subject a later would-be appropriator L—and, more specifically, would subject her because it would make her own appropriation a violation of the non-subjection NP. To illustrate this notion, Case 4 is non-exhaustive in the sense just defined: W is precluded from appropriating because such appropriation would make any further appropriation by X a violation of the non-subjection NP. Further, note that as the label suggests, a non-exhaustive appropriation scenario is one where exhaustive appropriation is incompatible with the non-subjection NP, with the latter implying that some portion of natural resources must go unappropriated.

Next, take any arbitrary non-exhaustive appropriation scenario and expand it by adding a new first-arriving would-be appropriator A. Call A an enabler if she can make a choice that—when paired with her own successful appropriation—would enable E to appropriate by making it such that this appropriation no longer subjects L. Note that if enabler A does not make this choice, then the expanded scenario is (trivially) still non-exhaustive: it is still the case that E is precluded from appropriating due to this action subjecting L. Suppose that, instead, A does make the enabling choice. This choice eliminates L’s conditional subjection and, thus, enables E to appropriate. However, this also entails that E is subjected, as her ability to appropriate depends on A’s willingness to make the enabling choice. Given this subjection, A is precluded from appropriating because doing so would subject a later would-be appropriator—and, more specifically, it would subject her because it would make her attempted appropriation a violation of the non-subjection NP. In other words, the expanded appropriation scenario is still non-exhaustive. Thus, adding an enabler to a non-exhaustive scenario will always preserve the non-exhaustive character of the scenario.Footnote 11

Finally, note that (a) Case 4 is a non-exhaustive case, and (b) to add an additional appropriator to the scenario—or any expanded version of the scenario—is just to add an enabler (as her decision regarding whether to hire Z determines whether later would-be appropriators are able to appropriate in compliance with the non-subjection NP). Given that adding an enabler preserves the non-exhaustive character of a non-exhaustive scenario, it follows that the expanded appropriation situation will remain non-exhaustive. And this will continue to be true no matter how many additional appropriators are added, as adding an enabler always preserves the non-exhaustive character of the scenario. Even in a competitive market with thousands of employers willing to hire a single laborer Z, exhaustive appropriation by those appropriators would still violate the non-subjection NP. Thus, contra van der Vossen, a competitive labor market is not sufficient for non-subjection; by his own lights, some resources must be left unappropriated for latecomers to independently acquire.

Given this result, it turns out that van der Vossen’s proposed NP actually commits him to something proximate to the left-libertarian position. Recall from the Introduction that the signature left-libertarian commitment is that appropriators must leave behind enough resources such that each non-appropriator has the opportunity to appropriate a share that is equal in value to each appropriator’s share. By contrast, the foregoing argument has established that each non-appropriator must be left with a share so that exhaustive appropriation does not obtain (in violation of the non-subjection NP). But what size must this share be? Presumably, when van der Vossen posits that persons have a right to acquire property, he is asserting that persons have a right to acquire some non-nominal quantity of property; otherwise he would have to implausibly maintain that the person who is denied access to all resources except for a single pebble is not subjected. Further, to avoid moral arbitrariness—i.e., arbitrarily assigning a right to one person but not another when there is no relevant property difference to ground this difference—it must seemingly be maintained that each person has a right to acquire an equal quantity of property. If this is correct, then the non-subjection proviso is satisfied only if early appropriators leave latecomers adequate and equal shares of resources to appropriate. Finally, note that early appropriators’ shares will exceed the set-aside shares if they have appropriated a quantity in excess of that which they (and all others) have a right to acquire.Footnote 12 Thus, if appropriation is constrained by people’s rights to acquire resources, then the non-subjection proviso will be satisfied only if each latecomer is left a share of resources equal in value to the share of resources appropriated by early arrivals. In other words, the non-subjection NP will support a left-libertarian rather than a right-libertarian position. While this result does not vindicate those who deny the existence of property rights altogether, it nonetheless represents an important result in the debate between right-libertarians and egalitarians.

The Systemic Specification

In addition to specifying the currency of comparison, a complete interpretation of NP must also specify what it is, exactly, that must not leave Q worse off. The most obvious specification is that it is the appropriation of R that must leave Q no worse off—i.e., the relevant comparison is between the world where R is appropriated and the baseline non-appropriation world. However, the more popular view among libertarian defenders of initial appropriation seems to be that it is the entire set of entitlements that must not leave Q worse off.Footnote 13 While perhaps removing R from the commons leaves Q worse off by precluding her from permissibly using it, this systemic specification of NP allows libertarians to dismiss this fact as irrelevant and, instead, point to the many ways that Q benefits from the productivity gains made possible by a system of private property.

In his argument against the viability of NP, Attas quickly rejects this proposal, arguing that this specification is ‘completely off the point. The proviso is a requirement of particular appropriations. Particular appropriations have to involve counterbalancing gains in order to be justified… [Thus,] a promise of increased benefits of the general system. … cannot justify [P] owning [R]’ (2003, p. 359). This objection correctly notes that (a) it is the act of appropriation that must be justified by NP, and (b) appealing to the benefits of the whole system of property seems like a way of justifying the system rather than the particular act.Footnote 14 However, given the popularity of the systemic specification of NP, it is worth considering why proponents of this specification might maintain that the entire system does play the appropriate justificatory role. Only after denying these apparent grounds for affirming the systemic specification can the specification be persuasively rejected.

To defend the conclusion that the benefits of a system of private property justify P’s appropriation of R, the proponent of the systemic NP might advance the following mereological argument. This argument begins with the following mereological premise: if some action or state of affairs is justified, then every part of that action or state of affairs is justified. For example, if a surgeon is justified in carrying out lifesaving surgery and one step of that surgery is cutting open the patient’s chest, then cutting open the patient’s chest is justified. Analogously, an act of appropriation is a part of a broader set of appropriations and resulting property rights. Thus, if this set is justified, it follows that every individual act of appropriation that composes it is also justified. Finally, given that the complete set leaves Q no worse off, it follows that the set is justified, which, in turn, implies that P’s act of appropriation is justified.

To assess this argument, begin by noting that the mereological premise is plausible only if a particular account of parthood is affirmed. Consider a modified version of the surgery case wherein the surgeon embeds a small toy car inside the patient’s chest during the operation. Clearly, the fact that the lifesaving surgery was justified does not justify this particular action. Thus, to preserve the truth of the mereological premise, it must be the case that the implanting of the toy car does not qualify as a genuine part of the surgery. But what could ground this exclusion? The apparent answer is that unlike cutting open the chest, implanting the toy car is not a cause of the patient’s life being saved. Granted, the implant is a cause of the state of affairs that results from the surgery, as the implanted car is part of that state of affairs. However, it is not a cause of the justificans of that state of affairs. This suggests that the mereological premise is true only if one accepts the following (partial) analysis of parthood: a sub-state of affairs is a part of a justified state of affairs only if it is a cause of the latter’s justificans.

This account of parthood creates the following problem for the proponent of the systemic NP. The reason for favoring a systemic specification of NP is that it seems like it is more easily satisfied than a non-systemic specification: while a single act of appropriation is unlikely to be to Q’s benefit, an entire system of private property might be. However, if one accepts the mereological argument with its correlative account of parthood, then a systemic NP will be satisfied if its non-systemic counterpart is satisfied. The mereological argument asserts that P’s appropriation is justified if the appropriation is a part of some system and that system does not worsen Q’s position on net. Further, the posited account of parthood entails that P’s appropriation is a part of the system only if it causes Q’s position to not be worsened on net—i.e., P’s appropriation satisfies a non-systemic NP. Thus, the mereological argument for the systemic NP entails that P’s appropriation is justified if it satisfies a non-systemic NP. Given this result, libertarians cannot appeal to the mereological argument to defend a systemic NP if they want to gain any dialectical advantage vis-à-vis a non-systemic specification of NP.Footnote 15

The Compensation Specification

The final specification of NP to be considered is one that is endorsed by almost all proponents of NP (irrespective of the other specifications they endorse).Footnote 16 This specification maintains that the non-worsening condition of the NP is satisfied if Q is left no worse off—or, typically, better off—when one considers the net causal effect of appropriation on her well-being.Footnote 17 This causal assessment is most naturally rendered in counterfactual terms: is Q better off, all things considered, in the world where appropriation occurs or in the world where it does not? Libertarians who endorse the compensation specification maintain that in almost all circumstances, Q is better off in the appropriation world. While appropriation diminishes Q’s liberty, they argue that it results in significant productivity gains and improvements being made to natural resources that end up benefitting her on net.Footnote 18 Thus, the compensation specification is able to vindicate most purported acts of appropriation.

As noted in the section labeled ‘The Non-Subjection Specification’, Attas objects that libertarians cannot accept this specification because it ‘amounts to a duty to improve owned resources beyond their natural state’ (2003, p. 358). This characterization is not quite correct, as it suggests that P might appropriate R but then subsequently wrong Q by failing to adequately improve R and make it available to Q. However, given that NP is a constraint upon appropriation, the compensation NP would actually imply that the P who fails to adequately improve R fails to appropriate R at all. Thus, the more precise way of stating Attas’s objection is that the compensation NP is incompatible with the libertarian view that property rights are counterfactually robust such that, if one owns a thing, one’s rights over that thing obtain irrespective of what one does with it (excluding, perhaps, the use of that thing to harm others).Footnote 19 If one accepts this robustness premise, then it follows that the P who owns R, improves it, and trades it to Q would have still owned R in the possible world where she did neither of these things. This result contradicts the compensation NP’s implication that P would not own R in that world—a conclusion that makes this specification seemingly unacceptable to libertarians.

While it is true that many libertarians take property rights to be counterfactually robust, many others might be willing to abandon this commitment in favor of the compensation NP. Nozick, for example, explicitly rejects this robustness in favor of a compensation NP, arguing that a person’s ownership of a resource might be invalidated in the counterfactual world where that resource is both precious and scarce such that ownership of it would amount to a monopoly (1974, p. 180). Similarly, Eric Mack posits that property owners cannot rightfully use their property in a way that unduly limits others from exercising their world-interactive powers (though he insists that this is not a limitation of their ownership rights) (1995, p. 188; 2002, p. 246). More generally, many libertarians’ enthusiasm for markets is grounded in the tendency of those markets to generate wealth. Thus, one can imagine many libertarians affirming that people would not have private property rights over land if they chose to let that land lie fallow (or rights over resources if they elected to leave those resources permanently untapped). On this view, property ownership would obtain only if the property in question were used in an adequately productive—though not necessarily maximally productive—way. Such a position would be fully compatible with the compensation NP, thereby sidestepping Attas’s objection to the position.

While this concession allows the compensation NP to avoid Attas’s objection, there is a further problem with this specification, namely, that it is not satisfied by any attempted act of appropriation. Strictly speaking, the compensation NP is satisfied only if P’s initial appropriation compensates Q by enhancing her well-being even as it diminishes her liberty. However, when someone appropriates some resource, the act makes a moral change to the world rather than a physical one. Presumably, such a change lacks causal power; thus it would be unable to produce any benefits whatsoever to compensate excluded parties. This, in turn, implies that every attempted appropriation will violate the proposed NP because it will not cause anyone to be better off in the required way.Footnote 20

If the compensation NP is never satisfied, then initial appropriation has never occurred and there are no existing private property rights. Given this result, the proponent of the compensation specification might retreat to a revised version of the compensation specification: initial appropriation occurs just in case P’s act of φ-ing (vs. the associated initial appropriation) leaves Q better off than if that act had not occurred. Given that φ-ing is a physical (vs. normative) act, it follows that, unlike appropriation, P’s φ-ing could have a causal effect on Q’s well-being. Thus, this revision avoids the objection presented immediately above.

One problem with this account is that most of the posited physical acts of initial appropriation do not leave people better off. For example, many proponents of initial appropriation propose that one must merely stake a claim of some kind to appropriate. However, it is hard to see how the mere staking of a claim could leave others better off. A further, more serious problem is that even if one grants that an act is justified if it compensates those to whom the act must be justified, it is not P’s φ-ing that needs to be justified. Rather, it is the appropriation of R. Thus, it would be inappropriate for the compensation specification to be defined in terms of the effects of φ-ing on Q.

Granted, if P φ-ing entailed the appropriation of R, then the benefits of the antecedent would plausibly justify the consequent. Here one might appeal to analogous cases such as the person who will live only if a surgery is performed; in this case, the benefits of living seem to justify performing the surgery because surgery is a necessary condition of those benefits being realized. However, to assert that there is such a logical relation in the case of appropriation would be to beg the question, as what is at issue is whether or not P φ-ing does, in fact, entail that P has appropriated R. To put this point formally, note that the proponent of the revised compensation specification contends that there is at least some situation where, if P φ-s, then P appropriates R. To defend this conclusion, she maintains that (1) if P φ-ing benefits Q, then P appropriates R by φ-ing, and, in this case, (2) P φ-ing benefits Q. However, as just noted, (1) is plausible only if appropriation is a necessary condition of P φ-ing, i.e., (3) P φ-s only if P appropriates R. Because (3) is identical to the posited conclusion, any appeal to the revised compensation NP to ground appropriation will beg the question.

This objection suggests a third possible revision of the compensation specification: the relevant counterfactual comparison is between the world where the act of appropriation does not occur and the world where appropriation occurs and there is full compliance with the rights generated by the act of appropriation. On this view, if Q is better off in the latter world, then appropriation does, in fact, occur, and property rights are established. Such a proposal has a number of advantages. First, it does not require positing that moral changes have causal effects, as the comparison world is one where people are stipulated to act in accordance with moral changes. Nor does it have to make implausible claims about the effects of the physical acts of appropriation. Additionally, it seems to adequately express the motivating idea of the compensation specification, namely, that everyone is better off in a world where people are governed by established property rights. For these reasons, this revised compensation specification seems to be the most attractive version of the compensation specification presented so far.

Unfortunately for proponents of the compensation NP, this revision suffers from three problems of its own. The first problem is that the notion of full compliance is ambiguous and requires additional specification if the compensation NP is to have determinate content. Consider the case where P appropriates R and thereby establishes a set of claim rights against non-appropriator Q. When assessing the benefits accrued by Q, is the relevant counterfactual the closest possible world where Q respects P’s rights relative to R? Or is it the closest possible world where Q respects all established property rights? Or the world where Q complies with all moral requirements including, but not limited to, the obligations imposed by property rights? Alternatively, perhaps the relevant world is that where not only Q complies, but also all persons comply with P’s rights over R/all persons’ property rights/all moral requirements. The fact that there are so many ways to specify the notion of full compliance is not an insoluble problem, as the proponent of the compensation specification may well have the theoretical resources to argue for one of these specifications over the others. However, there is at least a challenge here that the proponent of the compensation specification will have to overcome. For these purposes, no suggestion will be given about how to best specify the notion of full compliance, as all of the just-mentioned specifications will be vulnerable to the remaining objections.

The second problem for this revised version of the compensation specification is that it will be vulnerable to certain objections that have been raised in the ideal/non-ideal theory literature. For example, Huemer (2016) argues that it is inappropriate to derive moral theories from facts about what would happen under conditions of full compliance. On his view, the relevant question for outcome-minded political philosophers is what the effects will be of state enforcement of a particular policy (2016, p. 225).Footnote 21 Thus, what follows from people fully complying with some moral or legal regime is simply not relevant to the justificatory project of political philosophers. For example, when defending the justice of enforcing drug laws, it would not make sense to appeal to the effects of full compliance with those laws, as there will not be such compliance in the world where those laws are enforced (p. 225). While Huemer does not apply this argument to the case of initial appropriation, it would seemingly function as an objection to the revised compensation NP: appealing to the benefits that follow from people’s full compliance with property rights fails to adequately justify the relevant justificandum, namely, the enforcement of those rights.

There is a final objection to the third revision of the compensation specification. This objection holds that the compensation NP is an applied instance of a false general proposition: that, because some obligation would benefit a person under conditions of full compliance, people have the power to non-consensually impose that obligation. The apparent falsity of this proposition is most obvious in cases of what might be called paternalistic obligations. Given the harms associated with diabetes, many people would be better off in the counterfactual world where (a) they were obligated to not eat sugar in excess of some fixed quantity, and (b) they complied with this obligation. However, it does not follow that anyone has the power to oblige these people to restrain their sugar intake in virtue of this fact. More provocatively, fully compliant people would be better off in the world where they were obliged to obey the commands of a well-informed, benevolent dictator.Footnote 22 But, again, it does not follow from this fact that a sufficiently well-informed, benevolent person has the power to impose obligations on others. Thus, the counterfactual provision benefits under conditions of full compliance does not ground a power to impose obligations.Footnote 23

This conclusion does not, by itself, entail the negation of the revised compensation specification, as one might posit that, while counterfactual benefits do not ground the power to impose any kind of obligation, they do ground the power to impose property-related obligations. In other words, while the more general proposition is false, the proposition that restricts the obligations in question to those correlative of property rights is true. But what would explain this difference? To avoid arbitrariness, there must be some posited reason that the power to impose property-related obligations has a less-demanding sufficient condition from the power to impose other sorts of obligations. However, it is difficult to see what that reason might be. Why would the conditions sufficient for imposing an obligation to not consume sugar differ from those sufficient for imposing an obligation to not use some object or resource? Absent a compelling answer to this question, the revised version of the compensation specification of NP should be rejected along with the more general proposition that counterfactual benefits ground a power to impose obligations.


Given the number of terms that appear in NP and the number of ways of defining each of them, it does not seem possible to provide a single argument demonstrating that all possible specifications of NP will be inadequate. Thus, skeptics of NP must take on the challenging task of considering each possible specification of NP and arguing against it separately. Attas’s paper is the most comprehensive existing contribution to this effort, having considered over a dozen possible specifications of NP. However, as noted above, there are at least three seemingly plausible specifications that he either fails to consider or does not adequately address. This paper has attempted to fill in these gaps by demonstrating why the proposed specifications fail to support to the right-libertarian project of proving that there are existing (or readily attainable) property rights. Specifically, it has argued that the necessary conditions proposed by the various specifications are either unacceptable or fail to obtain. Thus, right-libertarians must find some novel specification if they want to appeal to NP to ground the existence of property rights.