1. Introduction
Reading Ernest Weinrib’s Reciprocal Freedom, I was visited by the familiar sensation of being pulled in opposite directions.Footnote 1 On the one hand, the book’s clarity of purpose and exposition, its well-developed argumentative lines and erudition, offered me a comforting, if complex, way to deal with the frustratingly unwieldy mess of precedents, legislative interventions, and private instruments and practices that constitute the explanandum of a theory of private law. On the other hand, I feared that the price for such comfort is accepting an excessive simplification of the normative landscape that lends meaning to the whole private law enterprise.
For one thing is clear: If Weinrib is right, several reasons (including moral reasons) that bear (at least prima facie) on the actions of judges, lawyers, and parties who occupy themselves with private law, would ultimately have no role in an appropriate normative account of their actions. After all, as he put it in The Idea of Private Law, corrective justice—the idea at the centre of private law—restricts the types of considerations that bear on the existence of a private right.Footnote 2 Thus, for instance, the goals of compensation and deterrence would not be acceptable justificatory considerations for a private right.Footnote 3 The first only applies to one side of the bipolar relation, and the second only applies to wrongdoers.Footnote 4 What is needed from a theory of private law, to put it in Weinrib’s terms, is an account of the “bilateral link” between the injured party and the injurer.Footnote 5 Reasons that apply only to one side of the equation do not count as grounds for private law rights.
This cull of non-bilateral reasons has been at the centre of an objection often raised against the Kantian approach. For instance, John Gardner has argued that the Kantian approach cannot account for certain types of public interest or public policy considerations that courts might be perfectly legitimated to bring to bear in tort cases.Footnote 6 For Gardner, from the fact that certain types of consequences (such as those predicated on the value of wealth-maximization) have little or no role to play in accounting for tort law, it does not follow that all types of general policy-based considerations should also be excluded. Indeed, it might appear that, from the Kantian requirement of correlativity (and its corollary bipolarity), it follows that all considerations pertaining to third parties, or the community as a whole, would also be necessarily excluded from the list of Kantian-right warranted reasons. At times, Kantians appear to think this themselves.Footnote 7 Reciprocal Freedom does much to explain how some of these prima facie reasons can be brought into the Kantian fold.
But not all such prima facie reasons could be cashed out in Kantian-right terms, and accordingly, a Kantian justification for private law would always leave a remainder of reasons which, regardless of their strength, would simply not be the type of reason that can bear on the existence of private rights. Excluding reasons by type is a serious move and a theory that goes down that path accepts the burden of explaining why the exclusion of these prima facie reasons is justified, regardless of their strength. I shall come back to that in Section 3.
My first concern, however, is to inspect the magnitude of this remainder against the background provided by the further unpacking of Weinrib’s position in Reciprocal Freedom. Clarity on this might help us understand whether Gardner’s examples of non-bilateral considerations that tort lawyers are allowed (indeed required) to take into account are indeed incompatible with Kantian right. It would also help to understand the conditions under which other non-bilateral reasons, beyond those envisaged by Gardner and others, might have a role to play in determining private rights.
Thus, clarity about this additional complexity of the Kantian argument is important for two reasons: First, as mentioned above, because it helps see how a Kantian perspective might be able to accommodate some non-bilateral reasons as bearing on the availability of private rights (including remedial rights). Secondly, and importantly, because these reasons exist in apparent tension with the requirement of bipolarity.
The first step in ascertaining what types of purely non-bilateral reasons are able to bear on the existence of particular private rights in a Kantian system, and how they relate to purely bilateral reasons (if there is such a thing), is to understand this extra layer of complexity and how it relates to the position of wrongdoers and injured parties. I turn to that problem in the next section.
2. Kantian sequenced argument
The most relevant development of Weinrib’s account of private law, as stated in Reciprocal Freedom, is a more detailed explanation of the Kantian argument’s sequential nature. The Kantian doctrine of right moves from what Kant takes to be truisms about the type of entity that humans are (self-determining, in contrast to objects), through a thought-experiment (the state of nature) that helps us understand the challenges we face in order to protect an aspect of humanity highlighted by those truisms (our autonomy), to the construction of conceptual and, further, institutional structures (in the civil condition) that would allow us to meet these challenges. Different types of consideration would be appropriate at different stages of the sequential progress, and, accordingly, reasons that are perfectly at home downstream (in the civil condition) are not allowed earlier in the conceptual itinerary. Importantly, it is not clear when, precisely, private rights (or aspects of private rights) emerge on this iter (and indeed it is controversial in Kantian scholarship). Thus, before we can evaluate which reasons bear on the existence (and shape) of private rights, we need a clearer grasp on (i) Weinrib’s account of the Kantian sequence, and (ii) Weinrib’s account of the moment when private rights emerge in the sequence.
This complexity in Kant’s account of private law is borne by his ambiguous use of the word ‘right’ (Recht). Recht, from a Kantian perspective, is not simply a Hohfeldian claim-right corresponding to a disjunctive duty to perform either the action required by a primary duty or the action required by the corresponding remedial duty. It refers both to a particular entitlement someone has within a set of entitlements and to the complete set of entitlements that derives from the principle of right (in relation to which a more natural translation would perhaps be ‘Law’).Footnote 8 As particular entitlements, Kantian rights must retain one feature commonly attributed to Hohfeldian rights: their special justificatory power. One way to express that power is to conceive the right as a protected reason, including not only a reason to do something, but also a reason not to act on reasons of a certain type.Footnote 9 But however this special justificatory power is cashed out, the particular right must be a part of a set of entitlements in which each particular right is compatible with every other right belonging to the set that ultimately derives from the principle of right.
The principle of right, famously, determines that an action is right if it can “coexist with everyone’s freedom in accordance with a universal law.”Footnote 10 So, if an action can in fact coexist with everyone’s freedom, one would not be under a duty not to perform it and, correspondingly, no one would have a right that the action not be performed; otherwise, one would have a duty not to so act and someone would have a corresponding right that the action not be performed. But a Kantian system of rights is not a list of entitlements that can be derived directly from the principle of right. True, all the rights in the system would ultimately find their ground in the imperative of coexistence between someone’s action and everyone else’s freedom, but many (perhaps all) rights that embody this co-existence cannot be derived directly from the principle.
This is an aspect of what Weinrib rightly saw as the sequential nature of Kant’s argument for his system of rights.Footnote 11 The point of departure, as mentioned above, is the distinction between objects (non-self-determining things) and persons (self-determining things).Footnote 12 The first step to be taken from this starting point is to try and work out what difficulties those self-determining entities would encounter as they realized their rationality by imprinting their will on each other and on the natural world (including their physical bodies). The most significant difficulty, Kantians believe, is that there might be a clash of determinations between the wills of two or more of those persons. This potential clash is at the core of the doctrine of right, and the Kantian state of nature is but a thought experiment that aims at getting the ball rolling in working out precisely the boundaries between personal wills. It is clear from this thought experiment (for Kantians) that each person’s will should have primacy over their own selves, including their bodies, and, accordingly, that each person should refrain from interfering with everyone else’s self-determination (roughly, Kant’s innate right).Footnote 13 As a corollary, people should also not interfere with objects that are in direct contact with someone else’s body (as that would necessarily imply some degree of interference with that other person’s body).Footnote 14
The thought experiment also reveals something about the relation we have with aspects of the physical world with which we are not in direct physical contact. For it is possible for a person’s will to impose purpose on objects. Accordingly, there is also the potential for a clash between the wills of persons. If I impose my will over a tract of land and you do the same, we need to find a way to establish whose will should prevail or, in other words, who should be the proprietor of the object. Regardless of the answer to that, Weinrib takes one thing to be clear in the pre-civil condition: What results from it is what Blackstone described as “that sole and despotic dominion which one man claims and exercises over the external things of the world, in total exclusion of the right of any other individual in the universe.”Footnote 15 The will of others is excluded from any form of control over the object. Property is not the only institution whose contours can be envisaged in the state of nature. It is also clear, for instance, that if two persons were to freely agree on how each would behave vis-à-vis a certain issue, their agreement would be binding on both.Footnote 16
But this thought experiment is not only a source of information about certain rights (or aspects of rights), that is to say, on the ways in which one person’s will could justly prevail over another’s. As we gain clarity about property, contract, and other aspects of the system of rights, it becomes progressively clear that what can be derived directly from the thought experiment is woefully insufficient. Thus, for instance, in the natural condition, there is clarity that agreements of wills create duties, but no clarity about what specific events count as an agreement of wills. Similarly, the principle of right helps establish (i) that ownership is to be understood as “the sole and despotic dominion,” etc., and (ii) that this type of entitlement stems from the agent’s exercise of the will to assert their primacy over the object; but the principle does not specify how precisely the will to control particular objects is manifested by an agent (i.e., should the will to own a tract of land be manifested by fencing it, by labouring it, by making a public verbal declaration, by entering the manifestation of the will in some sort of a public registry?).Footnote 17 Perhaps more importantly, if the imposition of my will were sufficient to generate a right, my unilateral will would be able to impose duties regarding the object on everyone else, and that, as Brudner remarked, would not be acceptable from a Kantian perspective.Footnote 18
So, the thought experiment reveals the need to supplement the normative guidance we gained directly from it. The need for those additional determinations calls for a larger perspective that cannot be provided by the bilateral will.Footnote 19 In the next stage of the sequence—the civil condition—an omnilateral perspective is called for to address the shortcomings identified in the previous stage. Although the omnilateral perspective adds to the normativity of contracts, property, etc., sequencing means that the latter is not fully independent from the former.
Precisely what the relation is between the normativities produced in each stage in the sequence is a vexed question. There are at least three ways, not mutually exclusive, in which insight gained in the first stage might relate to the determinations made from the subsequent omnilateral perspective. First, the state of nature might help identify several natural rights actually possessed by particular individuals. The best candidate for this type of right is perhaps the innate right that each person would possess in Kant’s system not to have their body subject to the will of another person. Brudner talks here of an “implicit omnilateral consent.”Footnote 20 This right requires no further determination.Footnote 21 There might still be a reason to enter into the civil condition to protect this right, but the civil condition does not add to our understanding of what the right requires from each person. Other rights would only be truly acquirable in the civil condition, as the ‘determinations’ produced from the omnilateral perspective would be necessary, for instance, to assign particular property rights to particular individuals, or to identify which agreements between persons are effectively binding (and hence a source of contractual rights). This is not, of course, just a practical matter. The civil condition bestows upon the right the omnilateral authorization without which my acquiring a right by merely asserting my will over an object would be an unjustifiable encroachment on your autonomy (as I would be creating a duty to you, without your consent).Footnote 22
Brudner believes that the acquired rights established in the civil condition, which are responsive to reasons that stem from an omnilateral perspective, completely replace the acquired rights that might appear to derive from the principle of right in the natural condition.Footnote 23 For Brudner, it would seem that whatever ‘acquired right’ could be identified in the first stages of sequencing (the state of nature) is primarily relevant because it allows us to see the value and point of the civil condition. Weinrib is not happy with this strict separation between the phases in the sequencing, but given that no one disputes that the omnilateral perspective is necessary for acquired rights to exist within the system of rights, we need another account of how the normativity of the state of nature might endure in the civil condition.
The second way in which there might be normative remnants of the state of nature in the civil condition does not concern particular entitlements, but instead properties of the rights that might only be fully acquired in the civil condition. Thus, in the first stages of the sequence we can see (i) that whoever turns out to acquire a property right in the civil condition will have the “sole and despotic dominion” over the object, to the exclusion of all others, and (ii) that this (property) right will have been originally acquired by an act of control, regardless of the determination made in the civil condition about how precisely this control is to be asserted. Similarly, whatever the specific conditions the civitas determines must be met for an agreement to count as a valid contract, the point of the determination is to track agreements of wills that will generate obligations to perform the actions the parties promised each other to perform. Furthermore, in the state of nature, it would be possible to understand, at least in outline, what would constitute a breach of those duties, the correlative violation of rights, and what would follow from it (i.e., the obligation to remedy the loss). Thus, it would already be possible to identify in the state of nature some properties possessed by private rights, even if the specific entitlements to those rights could not exist in the civil condition.
The third way in which the state of nature might be thought to shape the civil condition does not relate to specific rights (either full entitlements or properties of entitlements): It concerns the whole point of the system of rights. In the state of nature, we identify that the whole point of private law—and indeed of the state—is to allow for the greatest possible sphere of autonomy for each member of the relevant political community. Accordingly, autonomy-preserving considerations are either (i) the only reasons allowed to shape the normative landscape in our civil condition, or (ii) lexically prior to any other reason that might have a bearing on that normative landscape. Either alternative would imply the exclusion by kind of many reasons that bear at least prima facie on private (and indeed public) rights.
As mentioned above, Brudner is clear that the first type of ‘normative remnant’ (i.e., rights that exist fully formed in the state of nature and that migrate as they were into the civil condition) is restricted to innate right, and that all putative acquired rights that could be envisaged in the state of nature are not truly rights before we reach the last state of the sequence in the civil condition.Footnote 24 For Weinrib, that seems to be true in relation to at least some acquired rights, given that some (putative) acquired rights that might be envisaged in the state of nature might turn out not to be rights at all. The state of nature says nothing about whether it is sufficient, in order to acquire property, to assert control over land simply by fencing it off, or whether one must instead make a declaration asserting control in some sort of public register (etc.). The choice by the political community determines what mode of asserting control is necessary to generate a right to private property over that tract of land. Thus, presumably, no right to property over land can exist before the civil condition. A similar argument can be made in relation to particular rights derived from particular agreements and, as a corollary, in relation to particular tort rights/duties that stem from the violation of rights which themselves require determination. Thus, acquired rights can only exist at all in (and through) the civil condition.
But Weinrib is clear that there is a sense in which private right has a continued normative force in the civil condition. Much of the language used in Reciprocal Freedom is suggestive and metaphorical (the civil condition “does not inscribe its norms on a blank slate… [it] comes not to abolish the rights conceivable in the state of nature but to fulfil them,” etc.).Footnote 25 I suspect that one aspect of continuity is something along the lines of the second type of remnant identified above: Some properties of rights identified in the state of nature must be present in the rights the civil condition helps to ‘determine’. Thus, we might not know who (if anyone) has a right to this particular tract of land in the state of nature (even if we were to add to the state of nature knowledge of actions performed by persons attempting to assert control over that land); but we do know that, if and when this external object is in fact acquired, there needs to have been an assertion of control by the original owner (a ‘depositing’ of the owner’s will over the object) and we know that, after that assertion, they will have despotic control over the object to the exclusion of everyone else. We also know that interference with such control would create a duty to the person responsible for the violation to remedy the situation.Footnote 26 That excludes the possibility that, for instance, property allows other members of the relevant civitas to retain a significant (indeed any) measure of control over the external object acquired by the owner. It also seems to exclude that, in the civil condition, the community decides to assign property in a way that does not track the assertion of the first owner’s will. If the community decides to assign property based on, say, need or ability, without regard to the assertion of the will to control certain objects by some of its members, it would, presumably, not be respecting the type of autonomy whose roots lie deep within the previous stages of the sequence. Similarly, in relation to rights derived from an agreement of wills (a contract), some legal orders would introduce requirements for something to count as a contractual offer (or as the acceptance of a contractual offer) or impose a requirement that the agreement is made for some consideration, while others will not. The guidance provided by the thought experiment (i.e., the state of nature) is not sufficient to ascertain whether an agreement without consideration will be valid (and hence apt to create contractual duties).Footnote 27 For these determinations to be made, the agent must enter the civil condition.Footnote 28 But we would know that the content of the duty or duties stemming from the agreement would stem from the content of the agreement reached by the parties.
Weinrib also subscribes to the third normative remnant from the first stage to be found in the civil condition. As he puts it in the concluding section of chapter five in Reciprocal Freedom: “The root idea of a system of rights, that action should be consistent with the reciprocal freedom of everyone, thereby finds its place both in a state’s private law and in its public law.”Footnote 29 As I shall try to show in the next section, here lies the most challenging (and most typically Kantian) use of sequencing in Weinrib’s account of private law.
This brief analysis of the Kantian sequential argument and of the limits imposed by its earlier stages onto later stages will allow us to see more clearly how certain types of non-bilateral prima facie reasons might indeed bear on the existence of private rights.
3. The extent of the Kantian cull of reasons
One outcome of the sequential nature of the Kantian account of private rights is to provide two perspectives, from each of which a different type of reason bearing on private rights can be seen: the bilateral perspective in the first stage, and the omnilateral perspective in the civil condition. It also imposes a hierarchy between the stages, in that the latter exists to fulfil the normative project delineated in the first. To understand how the system excludes some considerations from the set of considerations that could bear on the existence and shape of private rights, it is worth considering each outcome separately.
The omnilateral perspective can accommodate reasons that do not concern only the holder of a right and the bearer of the corresponding duty. Take, for instance, the following putative reason: Allowing the owner to recover their property in a certain type of situation would have a detrimental effect in the legal system’s ability to guarantee all acquisitions. From the bilateral point of view concerning just the owner and one other person currently depriving the owner of exercising their will over the external object, the totality of acquisitions within a certain community cannot be discerned. The guarantee of all acquisitions concerns everyone in the community, and securing this common good is one of the primary functions of public right.Footnote 30 Weinrib’s discussion of the doctrine of ‘market overt’ is instructive in this respect.Footnote 31 In market overt, the owner is prevented (under some conditions) from recovering their property from a third party who, in good faith, has acquired the object from someone who, it turns out, did not have title over the goods they passed on. So, facts that register as legitimate reasons from an omnilateral perspective (the preservation of the whole system of acquisitions) might be sufficiently strong to outweigh considerations devised in the state of nature about the nature of a right (that the right over an external object is secured by someone asserting their will over it, in the right manner). The original owner of an external object that falls within the doctrine of market overt is deprived of their despotic dominion over the object to the exclusion of everyone else.
Notice, however, that reasons identified from this omnilateral perspective exist only to serve the primary (indeed, the exclusive) value identified in the state of nature: the preservation of the autonomous will against interference. In the civil condition, the determinations made by the community always aim at preserving each and every individual’s autonomy. It is at the altar of everyone’s autonomy that the owner (who obtained their property asserting their will to control the external object within the determinations of the civil condition) must sacrifice their legitimate control. But notice that the omnilateral perspective is also the owner’s perspective (the owner is, after all, part of the relevant omni). It is for the common good of preserving everyone’s autonomy (including their own) that the owner must relinquish control now.
If this type of reason—concerning the preservation of the conditions under which it is possible for everyone to be autonomous—is indeed something that can register within the Kantian system of rights, Kantians would have gone a long way towards addressing the concern identified above (the exclusion from private law of apparently legitimate policy considerations). Thus, at least some policy considerations must be allowed within a system of private law (e.g., those concerning the upkeep of a working legal system).Footnote 32 Although the fact that an institutional structure has the ability to promote or preserve a certain state of affairs (i.e., not disrupting a working legal system that guarantee’s acquisitions) might appear to be a consequentialist reason, there is a great and crucial distance between these types of ‘consequentialist’ reasons and the type of welfarist reason that Weinrib (and Kantians more generally) rightly reject. Welfarist reasons assume that value can be added across personal boundaries, so that an action φ might be deemed good because it produces the greatest aggregate value. However, enhancing aggregate value is perfectly compatible with producing no value whatsoever vis-à-vis one (or indeed most) person(s). The reason produced from the perspective of public right is grounded on the fact that it is valuable not only to all individuals, but to each individual equally. Value here is not added across the boundary between them. It is in the interest of every autonomous individual within the relevant community that acquisitions be guaranteed. It is even in the interest of our unfortunate owner whose property, legitimately and fully acquired, was lost to the doctrine of market overt. This adds a gloss to Weinrib’s earlier claim that the private law relationship must be understood exclusively from the standpoint of reasons that justify the correlative position of the parties bound in a particular right/duty pairing. Those reasons must indeed apply to both parties, as claimed in The Idea of Private Law, but vis-à-vis such reasons, they share their perspective with all other members of the political community.
Other reasons that apply only to the parties can only be seen from an omnilateral perspective that transcends them. Importantly, they could also bear on the existence of certain remedial duties and on the availability of certain tort remedies. Thus, there might be a public right not to be subject to a debt so overwhelming as to make the tort defendant into a slave of the plaintiff (and I suspect that much of our law of insolvency responds precisely to that concern). To put it another way, no one would agree to enter into the civil condition if the protection of the rights of others by the state would imply that they lose so much of their autonomy that they are reduced to the condition of working exclusively for the benefit of someone else for the foreseeable future.Footnote 33
It would appear, therefore, that the omnilateral perspective allows certain types of non-bilateral considerations to retain an explanatory (and a justificatory) role in relation to private law. It is less clear whether public rights can accommodate other prima facie reasons that push against personal autonomy. Take, for instance, reasons that militate against recognising ownership as a right that subjects what is owned to the “despotic” will of the owner, “in total exclusion of the right of any other individual in the universe.”Footnote 34 Some such reasons support excluding some objects from private ownership altogether, regardless of how enthusiastic an individual might have been in asserting control over them. There are many external objects excluded from ownership in every legal system, and at least some of the reasons that ground these exclusions are not easily captured in terms of ‘protection of autonomy’. A ban on private ownership of wild animals might be plausibly defended on grounds that they are necessary parts of the ecosystems to which they belong and such ecosystems, in turn, might be thought of as valuable because they contribute to an aspect of human well-being (i.e., our aesthetic appreciation of the wild, or the marginal benefits to air quality). The point of departure of the Kantian sequence (the distinction between self-determining entities and non-self-determining entities) puts the question of illegitimate interference with someone else’s control at the heart of private and public law. Law is concerned primarily (indeed exclusively) with the external world as something to be controlled and the problems that might ensue from conflicting claims to control. But the external world relates to us in other, more complex ways. It might be an object of aesthetic appreciation or a contributor to my quality of life (in ways not reducible to ‘fostering’ or ‘preserving’ my autonomy). Those appear to be perfectly good reasons for excluding certain goods from control, and they are not reducible to autonomy-type broader considerations. The point here is not that Kantians cannot allow that there are reasons to pursue certain ends. Mulholland has shown (to my satisfaction, at least) that Kant’s view could allow for a moral duty to pursue ends.Footnote 35 But those ends, within the system of rights, would ultimately have to be evaluated by the level of service they provide in protecting autonomy.
Additionally, there are reasons that militate in favour of restricting the “despotic” will of the owner to the exclusion of anyone else. Thus, the Land Reform (Scotland) Act gives everyone the right to be on land and inland water that belongs to others, for recreation, education and other purposes (under some conditions).Footnote 36 This right is accompanied by explicit duties imposed on the owner, including duties not to erect fences, walls, and hedges, duties not to carry out certain agricultural operations, and duties not to erect certain signs or notices.Footnote 37 This is not just a Scottish legislative idiosyncrasy, as the right appears to have roots in the law that precedes the statute and, in fact, Scottish courts have not shied away from enforcing such rights and duties.Footnote 38 Like the complete exclusion of objects from the regimen of private ownership, the so-called ‘right to roam’ is not easily assimilated to the value of autonomy.
Thus, while the sequential nature of the Kantian argument might allow private law to respond (i) to some types of policy consideration, and (ii) to some considerations that bear asymmetrically on the bilateral relationship between right-holder and their correlative duty-bearer, the Kantian perspective would still exclude all considerations that are not ultimately grounded on the protection of reciprocal freedom. This exclusion can take two forms, one radical, the other moderate. The more radical form excludes all reasons not solidly grounded on reciprocal freedom from the set of reasons that could legitimately bear on private (or indeed public) right. They simply would not be the kinds of reasons which could influence the existence or scope of private rights. The more moderate form of exclusion accepts that these reasons could have some influence in shaping private rights, but not if they come into conflict with reasons concerning reciprocal freedom. To use Rawls’ phrase, reasons grounded on reciprocal freedom would be “lexically prior” as far as normative accounts of private rights are concerned.Footnote 39
Hence, it turns out that the Kantian system of rights is more capacious than some of its critics (including John Gardner and me) have assumed in the past. Of course, the Kantian system still excludes many reasons by kind from bearing on private rights, regardless of their comparative strength. It follows that a relatively trivial violation of autonomy would always prevail over any other kind of normative gain, however great it might be. Some of those excluded reasons have been used by legislators and courts to justify restricting private rights (for instance, in the context of Scottish land law). These are often eudemonistic reasons concerning the advancement of some human good other than autonomy and are often very close to some of our moral intuitions. Although this exclusion of reasons by kind brings to the Kantian argument a heavy justificatory burden, it has not been my objective here to put pressure on Kantian attempts to face up to this burden.
Reciprocal Freedom is written squarely within the Kantian canon and, accordingly, does its argumentative work within the framework built by Kantian assumptions. My argument above had the modest aim of unpacking some of its potentialities and, in so doing, identifying more precisely some of the burdens it must discharge if it is to account for the reasons that apparently bear on private rights.
Acknowledgments
I would like to thank Ernest Weinrib, Euan MacDonald, Matthew Kramer, and audiences at Surrey and Edinburgh for helpful comments on a previous version of this paper (or, sometimes, on specific parts of the argument).
Open access
