Principles of justice and the idea of practice-dependence

ABSTRACT In recent years, several political theorists have argued that reasonable principles of justice are practice-dependent. In this paper it is suggested that we can distinguish between at least two main models for doing practice-dependent theorizing about justice, interpretivism and constructivism, and that they can be understood as based in two different conceptions of practices. It is then argued that the reliance on the notion of participants that characterizes interpretivism disables this approach from adequately addressing certain matters of justice and that a better way of developing the idea of practice-dependence can be found in a constructivism that starts from the Rawlsian idea of overlapping consensus, but which shifts the focus of that approach from societies to a more open-ended category of domains, and which understands the parties to a possible overlapping consensus as stakeholders in a certain set of interconnected practices.

with, Dworkin's account of legal interpretation. Here the idea seems to be that the relevant triangulation is between (i) our practices and (ii) some moral or political values that are too indeterminate to simply derive a conception of justice from. This approach is explicitly adopted by some of the main proponents of practice-dependent theorizing, such as Aaron James (2005, 2012), and possibly by Ronzoni (2012) as well. The second approach is Rawlsian constructivism, which builds on, but need not be identical with, Rawls' political constructivism. Here the triangulation instead relies on (i) our practices and (ii) ideas about public reason and what is involved in being reasonable.
The argument in this paper will be in favour of the Rawlsian approach as being more promising when developing a framework for practice-dependent theorizing. The argument comes in three main steps. First, the underlying difference between the Dworkinian and Rawlsian approaches will be explained in terms of adherence to two different conceptions of practices. Second, it will be argued that while the Dworkinian approach might be reasonable in a legal context, it does not work as a generalized method for normative theorizing. Third, it will be argued that even if Rawls' political constructivism is in need of further development in order to work as a generalized method for practice-dependent theorizing, this type of generalization can be done by understanding Rawlsian constructivism in terms of a method for articulating normative principles for domains (sets of interconnected practices). 2

Understanding practices
Irrespective of how, more concretely, the idea of practice-dependent theorizing is developed, it seems quite clear that contemporary articulations of this idea has their roots in Rawls' later political theorizing, partly in the constructivist approach developed in Political Liberalism (1993), partly in the application of that approach in The Law of Peoples (1999b), the idea then being that different principles of justice will be reasonable at the domestic and the international/global level because of differences in the underlying practices. 3 Sangiovanni (2008, 138n2) also draws even further on Rawls, at least in his stated understanding of the term practice, namely as 'any form of activity specified by a system of rules which defines offices, roles, moves, penalties, defenses, and so on, and which gives the activity its structure' (Rawls 1955, 3n1).
The question here is not necessarily what practices really are, the social world is probably too messy for definitive definitions, but rather how we should best understand them in order to facilitate normative theorizing. Rawls' understanding of practices clearly then makes sense when seeking to theorize justice. It is a fundamentally distributivist understanding: it involves seeing practices as a form of distributions of rights and duties, where a given distribution will define which moves that are open to whom. 4 It is also a picture that fits well with seeing practices as resulting from an ongoing bargaining process where a certain distribution of rights and duties has gradually grown out of repeated interactions, and where there are then normative expectations on us to 2 For more on this, see Brännmark and Brandstedt (2019). 3 As suggested by Jubb (2016, 79-80), this frame of reference has arguably led to discussions of practice-dependence insufficiently engaging with moral and political philosophy beyond the global justice literature. 4 That Rawls understands rights and duties within a distributivist paradigm is a point of criticism in Young (1990, 25), but here this understanding will just be taken for granted. play in accordance with those rules. 5 We might have very different goals in doing so, and we can play along both in good faith or bad, but the rules are there (although political theory can certainly aim at renegotiating them). Because of its strong focus on rules, we might call this a deontic understanding of practices. Its focus lies on which moves that are open or not rather than on any deeper reasons for why certain moves make sense. Indeed, if we want to discuss whether the rules are reasonable or not, we need to take up a standpoint external to the relevant practices, such as constructing a theory of justice. We are then on the level of principles rather than rules.
At the time when Rawls articulated this conception, and drew his distinction between two concepts of rules, he was very much in line with Hart's (1961, 9) distinction between two types of rules, and what later became known as his practice theory of rules. Perhaps somewhat ironically, given that some adherents of practice-dependent theorizing both take their starting-point in Rawls and then favour a Dworkinian understanding of interpretation, the foremost critic of Hart was, of course, Dworkin. 6 The main criticism that Dworkin directs against the model proposed by Hart (and indirectly Rawls) is that it cannot adequately handle disagreement or controversy about what the rules say, but also that it gets the relation between practice and rules wrong: a practice is not a set of rules, a practice is something that underpins social rules and that can be used to justify them (Dworkin 1977, 58). For Dworkin, interpreting our practices becomes not just a matter of laying bare the rules that constitute them, but to lay bare what underlies the rules. There is no rock bottom simply consisting of rules. While theorists like Rawls and Hart understand practices in terms of an analogy with games, whether baseball (Rawls 1955, 25) or chess (Hart 1961, 56), the analogy that informs Dworkin's view of understanding practices is rather one with art, as involving 'a way of seeing what is interpreteda social practice or tradition as much as a text or paintingas if this were the product of a decision to pursue one set of themes or visions or purposes, one "point," rather than another' (Dworkin 1986, 58-59).
We might call the conception of practices underlying Dworkin's model a telic conception of practices. On both conceptions one would certainly accept that practices involve rules, and an adherent of the deontic conception might very well accept that practices often come with ideas about some point(s) or purpose(s), but the difference is that given a telic conception you will not have captured the core of a practice if you have only accounted for the relevant rules. Indeed, similar to how some utilitarians have objected to deontological ethics by comparing it to 'rule worship' (Smart 1973, 6), Dworkin (1986, 89) warns of the risk of legal practice collapsing into 'runic traditionalism' without the right kind of interpretative attitude. As argued by Postema (1987, 302-05) the choice here is however not between understanding practices either in terms of mechanical performance in accordance with certain rules or in terms of some overarching point or purpose. In actuality, mastering a practice can be understood more in terms of being at home in it, being able to navigate it ably and even improvise in a way that still makes sense. But the latter only presupposes that actions within a practice are meaningful, not that this meaningfulness is understood in terms of some ultimate point or purpose of the practice. A particular action might, for instance, balance different duties together with the agent's own self-interest in an elegant and economical way without this way of excelling being a matter of fulfiling some underlying function of the practice as a whole.
We do not accordingly need to adopt the kind of interpretative stance favoured by Dworkin, and we can certainly still make interpretations without them being Dworkinian, 7 but it could still be a good method. One obvious reason why would be that it might facilitate a certain kind of theorizing or concrete interpretative practice. Dworkin's own account is an account of legal interpretation as guided by an 'adjudicative principle of integrity ' (1986, 225), which 'asks judges to assume, so far as this is possible, that the law is structured by a coherent set of principles about justice and fairness and procedural due process ' (1986, 243). To the extent that he should be understood as an example of practice-dependent theorizing, 8 it is because his approach involves a triangulation between these higher principles together with the relevant legal framework and practice, working towards a reading of existing practice that is infused by those higher principles, but does not involve straightforwardly deriving legal conclusions from them. Being neither a natural law theorist nor a positivist, Dworkin's approach is about enabling judges to read positive law in the light of justice. It is a form of legal interpretation where great weight is placed on positive law, but where the arc of legal practice still bends towards justice. 9 In practice-dependent political theorizing, we are of course instead trying to arrive at a conception of justice, so it is clearly not about simply using Dworkin's approach. 10 But we can arguably still take up that kind of interpretative stance, adopting a telic rather than a deontic conception of practices. Sangiovanni (as well as James) is explicit about adopting a Dworkinian three-stage model of interpretation, 11 where the first is pre-interpretative and is about identifying the shared object of interpretation, and where the interpretative second stage is characterized like this (Sangiovanni 2008, 148): First, the interpreter seeks to determine the point and purpose of the institution in question. What aims and goals is it intended to serve? Second, the interpreter assumes the point of view of the participants in order to reconstruct what reasons they might have for affirming its basic rules, procedures, and standards. Why and how do the participants arrange their affairs to achieve the goals and aims of the institution? In achieving both tasks, the interpreter seeks to understand the institution (or set of institutions) as an integral whole, whose parts work together in realizing a unique point and purpose.
Social reality can occasionally be quite messy, so the idea of understanding institutions as integral wholes should probably here be seen in the light of how Dworkin himself assumes a certain level of coherence in order for the interpretative process to get going. Of course, Rawlsian constructivism is also coherentist, since the end-goal is a state of reflective equilibrium, but there coherence comes at the end of the process rather than being assumed as an interpretive strategy. And even then, such coherence can simply be about consistency and mutual support, similar to how a deontological ethical theory can 7 For instance, Rawls is clearly also engaged in interpretation, albeit primarily of the 'public culture' (1993,(13)(14) that accompanies our practices. 8 Although see Meckled-Garcia (2013) for an argument that Dworkin's method is 'the opposite of practice-dependence.' 9 But even with respect to what one could take to be an obviously unjust law, like the 1850 Fugitive Slave Act, it could still be the case that the law of the USA at the time, properly interpreted, did include it (Dworkin 1986, 219). 10 A point noted by, e.g. Valentini (2011, 408). 11 Ronzoni (2012, 176) embraces the idea that 'principles of justice for specific practices depend on the nature of those very practices in light of a sound interpretative account of its point, purpose, scope, and actors.' be coherent without there being a unique point and purpose which the moral framework serves. In contrast, the idea of integral wholes is about something being much more strongly woven together. Once we have this kind of interpretation in place, we turn to the post-interpretative stage, which is where we start moving towards articulating principles of justice. Here both Sangiovanni and James seem to presuppose that we already have a commitment to certain values, but that they are relatively indeterminate. Sangiovanni (2016, 17) talks about 'higher-level principles and valuesof justice, legitimacy, solidarity, reciprocity, and so on' which we can then form determinate conceptions of based in our interpretation of the relevant practice(s). James (2012, 27n27) assumes a contractualist moral theory, but where principles of justice are then still tailored and justified in relation to a specific practice. In contrast to Dworkin, the relevant triangulation here is about arriving at a specific determinate conception of moral or political values or principles which are, at least initially, open to several different precisifications. In Dworkin we are reading practice in the light of justice, in Sangiovanni and James we are reading justice in the light of practice. In both cases, the idea is that when confronted with something that allows multiple readings, whether the law or certain higher-level values of principles, the assumption that our practices can be read in terms of their point(s) and purpose(s) allows us to reason our way towards determinate meanings or conceptions.
If we instead work with a deontic conception of practices, we can still seek coherence between the concrete rules governing those practices and a set of higher principles, but such a pure coherentist approach would be much more open-ended in where it would end up. Because of the messy and contested character of social and institutional reality, there will typically be significant indeterminacy in that there will be several different ways in which we could move towards greater coherence, and with nothing more to aid us than the ideal of coherence alone, there will be an arbitrariness in going in any specific direction. What characterizes the Rawlsian approach, however, is precisely that it is not a pure coherentism. It too adds a further factor to the equation, one that allows a kind of triangulation. Understood as a form of practice-dependent theorizing, the Rawlsian works with (i) principles of rationality, reasonableness, and the idea of public reason, and (ii) a particular social and institutional situation, in order to arrive at principles of justice addressing the latter. The reliance on (i) clearly locates Rawlsian constructivism in a Kantian tradition, although while Kant is attempting to derive determinate content from his analysis of practical reason as such (based in his particular metaphysics of the person and society (cf. Rawls 1993, 100)), the Rawlsian constructivist works with a thinner conception of (i) and needs (ii) as another known point in order to arrive at a determinate enough conception of justice. We will return to the Rawlsian approach in Section 3, but first let us look at some problems faced by the Dworkinian approach.

Against Dworkinian interpretivism
Social and institutional reality is often messy and contested. Dworkin's original approach was developed for a particular part of institutional reality that is arguably the most ordered and systematic, namely domestic law. When turning to other practices, one obvious worry about assuming an interpretative stance that operates on an analogy with works of art is that this will lead to a mischaracterization of those practices, making them come out as more harmonious and uncontested than they might actually be. Even to the extent that a given practice is relatively stable, it does not seem necessary that there must be a clear point and purpose to it, weaving it together into an integral whole; it might just represent a point of equilibrium in a social game involving different types of actors who all have their interests, a mere modus vivendi (Rawls 1993, 47). Stability can simply be a matter of how the balance of power aligns with how much different actors benefit from the practice in question. When we make this type of analysis, we tend however to take an external perspective, where the reasons that participants themselves appeal to might very well be understood simply as pretext for power moves.
In contrast, the interpretivist approach relies on taking an internal perspective, reasoning from the point of view of participants. Yet this reliance on the perspective of participants seems far from unproblematic. To begin with, it seems to presuppose a certain willingness and active involvement in shaping a practice in order for someone to count as a participant, and it seems clear that practices can have a strong impact on people even when we cannot be said to really participate in them. Call this the problem of exclusion. 12 Additionally, there is something about the idea of us all just being 'participants' that hints at a basic equality between us. Yet take a society like Gilead (Atwood 1985). We can certainly describe both men and women there as all being participants in its practices, but would this really give a fair picture, when some have no say in writing the rules of that society? 13 Analysing those involved in a practice merely as participants with different points of view risks masking the power imbalances involved. People can suffer from false consciousness, wishful thinking or have adaptive preferences, and what we then risk doing in our interpretation of the point and purpose of a practice is simply to reify what are really fictions masking the real functioning of the relevant institutions. Call this the problem of masking.
It should be said that both James and Sangiovanni are aware of there possibly being these kinds of difficulties. James is careful to point out that identifying justice claims of participants will not have any 'immediate implication for the status of non-participants' (2005,309), and that it is possible that the justice claims of non-participants can be determined in some other way, while Sangiovanni suggests that we can employ ideology critique even within practice-dependent theorizing (2008,163). Both of these points are certainly correct in terms of identifying what is in principle still possible. But the question here need not, and perhaps should not, be understood as being about whether taking interpretivism as a starting-point positively rules out certain considerations, but about whether doing so misaligns the process of theory construction already from the outset. This might be an issue already with respect to domestic distributive justice, but with many other matters of justice we are dealing with much more fractured practices, and a focus on the category of participants becomes even more problematic. Let us consider three such examples.
To begin with, consider questions about global justice, more specifically the extent to which the more affluent have obligations to the less affluent. This has probably been the main area where practice-dependent theorizing has been applied and the argument is then often that we largely live in a world of sovereign states and that the principal arena for distributive justice is the domestic one (e.g. Miller 2007;Sangiovanni 2007). But is the global stage a reasonable domain for applying Dworkinian interpretation? While on the level of domestic justice we might perhaps already have reasonably shared conceptions of the point and purpose of our domestic institutions, and what we think that justice is about, the international or global level is quite different. It is arguably in large parts an anarchic system, with the actual balance of power between states determining much of what is possible and not, making it considerably more difficult to ascribe a deeper underlying point and purpose to the relevant practices. 14 While the international order certainly has consequences for us all, it seems like a stretch to conceptualize of us all simply as participants in it. Given the power relations involved, and the great discrepancies in wealth that exist between countries, there would seem to be a high likelihood of self-serving ideological constructs playing a significant role in the thinking of people living in the more affluent parts of the world, and adaptive preferences and false consciousness playing a significant role in the thinking of people in the less affluent parts of the world.
The second issue concerns questions about the status and rights of migrants and especially refugees. This has not really been a major area of activity in political theory until relatively recently, with Carens (2013) and Miller (2016) being two major works. Both of them have a strong focus on actual practice, but with Miller probably being the one who most clearly exemplifies practice-dependent theorizing (although not Dworkinian interpretivism per se). If we look at possibly ascribing point and purpose to current institutional arrangements regulating movements of people across borders, we again have the problem that when looking at the overall system level, there does not seem to be much point and purpose to it since it is more or less the product of policy choices made on the level of states with respect to perceived national interests. Miller seems to work from an assumption that we should not start from a systemic perspective already in how he frames his main questions: 'Should we encourage immigrants to join our societies or try to keep them out? If we are going to take some in but refuse others, how should we decide which ones to accept?' (2016, 1). Yet while many of 'us' do indeed belong to a we that can ask questions like these, there is also a broader set of people holding stakes here and more to be said about who should reasonably have a say in constructing the relevant policies. 15 Migrants and refugees certainly exercise agency within the bounds set by the relevant political institutions, but in a system of sovereign states, refugees especially can also be stuck between institutions, not really occupying any stable positions with clear rights and duties. Is thinking in terms of participants really helpful here?
Finally, there are questions about the obligations of people in the present to future generations, a matter which has in recent years primarily been considered in relation to the possible effects of climate change, but which is relevant to the issue of sustainable development in general. Again, this seems like a clear case where focusing on the perspective of participants rather than, say, stakeholders seems problematic (cf. Reglitz 2016). Both of the previous issues were cases where there is a mismatch between (i) who 14 This is not to say that Dworkinian interpretation cannot yield anything in this context, James (2012) is a clear example to the contrary. However, even to the extent that we can identify some point and purpose, we might not have good reason to trust such results if we find that the method has an in-built tendency to skew our interpretations in masking and excluding ways. 15 For a different perspective than Miller's, see Abizadeh (2008). gets to influence the rules that govern the relevant practices and whose interests are accordingly likely to be reflected in the point and purpose that can be read into these practices, and (ii) who will suffer the adverse consequences of them. Future generations have absolutely no influence on what our practices look like, at least not beyond what our ideas about them might lead us to do (currently: very little), and yet our practices might have an enormous impact on them. Of course, the Dworkinian interpretivist might respond that the point is not that we will derive our principles of justice from our current practices, but simply that which principles that can count as reasonable principles of justice will always be constrained by current practices. But even if this is perfectly true in principle, it seems doubtful if Dworkinian interpretation is a viable starting-point in trying to move towards principles of intergenerational justice.
When it comes to domestic justice, one reason why our current practices possibly could work as a reasonable starting point for articulating principles of justice is that the people who have predominantly shaped these practices and the people for whom the relevant political institutions manage their problems of justice more or less coincide. What this means is that in coming to better understand the workings of these practices we will also at the same time come to understand the more precise problems to which our institutions are ideally a solution. Yet in all of the three cases considered above this is not the case. There are mismatches between those who predominantly shape the relevant practices and those who have a stake in the kinds of actions that these practices enable and regulate, and whose lives will be shaped by having them in place. While there might certainly be ways to mitigate these problems, their source is hardwired into the method. The Dworkinian approach is built on taking an idealizing stance in relation to existing practice, one that pushes tensions to the margins and places unifying features at the centre already from the starthence the problem of masking. And the move from merely taking a more sociological view of practices to an emphasis on the insider perspective seems bound to place the notion of participants at the centre of the interpretative approachhence the problem of exclusion.

Rawlsian constructivism beyond Rawls
Can Rawlsian constructivism steer clear of the problems of masking and exclusion? At the very least, working with a deontic conception of practices would seem to provide a more promising start since such a conception allows for practices to be characterized by underlying tensions, very different agendas and objectives among different parties to the practice, and stability primarily based in a power balancea mere modus vivendi. Indeed, the principles of justice that are articulated are supposed to be capable of serving a unifying role, by providing a basis for reasoning together. On the Rawlsian approach, the startingpoint is not that of participants in an already coherent and unified practice, but rather parties to certain shared problems who are in need of shared principles in order to be able to handle those problems as a community of reasoners.
Having said this, it should be noted that Rawls' own constructivism, although it is later applied to the international level as well, is developed in reasoning about domestic justice. This means that while a key notion for Rawls is that of parties (to a possible agreement), political constructivism also seems to share with Dworkinian interpretivism an emphasis on insiders, since it is understood as a procedure involving citizens of a society (Rawls 1993, 90). And yet, institutions and practices 16 in one's own society can clearly have significant impact on the status and situation of people outside of it. The problem of exclusion accordingly seems like it could be an issue for Rawlsian constructivism as well. But does constructivism as a method really presuppose an insider-type framing in terms of societies and citizens? It seems far from obvious that this framing really is hardwired into the method, and the goal of the rest of this section is to suggest, at least in outline, how what is still a basically Rawlsian approach can be developed into a generalized constructivism, one that still relies on ideas about rationality, reasonableness, and public reason as forming a known point in a process of triangulation where the other known point involves current practices and institutions.
Even if we restrict ourselves to principles of justice, there are clearly issues of justice in many human contexts or areas of interaction. Family life, the workplace, sports, various associations, and so onin more or less every area of human interaction there are goods that are somehow distributed, where some might get less and some more, and where our actions have an impact on not just ourselves but others as well. Some of these areas might feature highly reciprocal interactions, others decidedly one-sided ones. And while some areas might be highly formalized and even feature written rules, others will be characterized by more informal and implicit norms (or some mix of formal and informal). But there will be norm-governed offices, roles, moves, penalties, defences, and so on, which provide structure to the relevant activities, i.e. they are practices in the Rawlsian sense. There are certainly questions about how to individuate practices more exactly, but the main point here is that in normative theorizing we are dealing with relatively large-scale areas of life organized and regulated in terms of such practices. Rawls himself suggests that 'there is a special domain of the political ' (1989, 242) which his theory of justice and fairness then seeks to articulate principles for, and we could understand a domain precisely as such an interconnected set of practices (cf. Brännmark 2016).
In carving out the political as a domain, Rawls is looking for something that is fit to be theorized and while relevant domains up to a point can be expected to be described sociologically in terms of interconnected sets of practices held together by certain important offices and roles or key goods and values, we should arguably not expect the notion of domains to be purely empirical: what will count as a domain will partly depend on what it takes for a set of practices to be suitably interconnected in order to be theorized in terms of which normative principles that are reasonable for regulating interactions between people in that domain. This also means that what counts as a domain will partly depend on how we understand normative theorizing. For now, this issue will however have to be set aside, the important thing instead being that it seems reasonable to accept that a Rawls-style constructivism could in principle be more widely applicable than merely to the domain of the political. And if we look at domains more generally we are not committed to citizens being the only relevant parties to consideron the contrary, identifying relevant stakeholders, i.e. those types of actors that should reasonably be included as parties to an agreement on principles of justice, will be an important part of describing specific domains.
In terms of the goal towards which theorizing is supposed to move, an important part of Rawls' political constructivism is the idea of a well-ordered society, a society in which 'reflective equilibrium is also general: the same conception is affirmed in everyone's considered judgments' and where 'not only is there a public point of view from which all citizens can adjudicate their claims, but also this point of view is mutually recognized as affirmed by them all in full reflective equilibrium ' (2001, 31). 17 He then goes on to explicate the idea of an overlapping consensus, 'to formulate a realistic idea of a well-ordered society ' (2001, 32), one where citizens would affirm the same political conception of justice, but would do so based in different comprehensive doctrines. Again, even with this idea of wellorderedness, there does not seem to be any essential tie to the political as a domain. We can distinguish between domains in which the practices that are in place are in place merely as part of a modus vivendi, something that we accept given how things stand now, and wellordered domains, where our interactions are governed by principles around which there is an overlapping consensus. Modus vivendi domains are not worse by any independent moral standard, but to the extent that there is stability in them it is the stability of a balance of power, whereas in a well-ordered domain we have a stability based in a deeply shared understanding of justice, one that can persist in the face of shifts in the balance of power. While in a modus vivendi domain we are primarily making private use of reason, the shared principles in a well-ordered domain enable a genuinely public use of reason.
What theorists will mainly be working towards is to articulate principles of justice around which an overlapping consensus could be formed, a version of the Kantian emphasis on principles that can be shared, but where actual practice forms an important known point in triangulating towards a conception of those principles. While Kant's universalizability test abstracts from our actual situation and turns on the formal character of our maxims, looking at whether there is some contradiction involved in willing them as universal laws, Rawls' constructivist procedure has a substantive element in that it involves looking at the fit between potential principles of justice and our actual social and institutional situation. In looking for such principles it accordingly becomes important to analyse this actual situation in order to identify both main dividers and unifiers, where these can be assumed to be at least potential resources or obstacles for moving towards a well-ordered domain under conditions of reasonableness. 18 Indeed, any systematic practice-dependent approach will have to have a way of characterizing existing practice and the situation we actually find ourselves in. This inevitably means taking a summarizing approach, and hence eliding certain things, and what is needed then is a way of doing so which still steers clear of the problems of exclusion and masking. This is where Dworkinian interpretivism runs into problems, but where Rawlsian constructivism, in relying on a deontic conception of practices, looks more promising. The suggestion here is that in framing the assessment of how proposed principles of justice can fit with our actual situation, the domains for which we are to articulate principles of justice can be analysed in terms of the following three 17 Rawls' specification of full reflective equilibrium is as general and wide, where the latter is a matter of someone having 'carefully considered the leading conceptions of political justice' and 'weighed the force of the different philosophical and other reasons for them ' (2001, 31). 18 Being reasonable is mainly about being realistically responsible qua reasoners; it involves 'the willingness to recognize the burdens of judgment and to accept their consequences for the use of public reason' (Rawls 1993, 54), which among other things means accepting 'that many of our most important judgment are made under conditions where it is not to be expected that conscientious persons with full powers of reason, even after free discussion, will all arrive at the same conclusion' (Rawls 1993, 58).
principal categories (and where all three are in some form already present in Rawls' own approach, but where an attempt is made here to further generalize them): The goods at stake Areas of interaction tend to revolve around certain goods, although exactly which goods might vary from area to area. In articulating apt principles, we need to know what kinds of goods the distribution of which those principles are supposed to handle. Some aspects of human well-being will inevitably be at stake, but such goods can still vary between domains, say, medicine compared to war or in family life compared to academic life, and so on. This kind of identification of goods at stake should be expected to be graded in the sense that certain goods will be the primary concerns in an area, whereas other goods will be more peripheral. Partly this will be due to how, depending on the layout of the relevant practices, there will be certain predominant ways in which the actions of some affect the lives of others; accordingly, what is primarily at stake will to a large extent depend on the character of the relevant practices. If we take future generations as an example, most of the concrete circumstances of specific future individuals will be determined by decisions taken then rather than now, but quantities of natural resources available and qualities of the natural environment are certainly matters that are at stake between us and them. In other cases very different goods will be at stake for different people. When it comes to questions about border controls, for instance, we can on the one hand have people seeking refuge, on the other hand people who primarily have a stake in the functioning of societies where they are already members; these possible tensions will then form a starting-point for practice-dependent theorizing rather than simply being something that independently articulated principles are applied to.

The relevant stakeholders
The problem of exclusion points to a need to move away from a narrow focus on participants, or citizens for that matter, and rely on a more open category for identifying possible parties to an agreement on certain principlesan identification that must be made before we can attempt to articulate such principles. Given that we have started by identifying the goods that are at stake, it seems natural to then identify the different stakeholders involved, and where identifying such stakeholders will also involve identifying certain social positions, in terms of deontic statuses such as rights and duties, that people can occupy within the relevant domain and which govern which moves they can make in relation to which goods. The category of stakeholders is open to there being quite different ways in which we are stakeholders, depending on the concrete relations in which we stand to each other, to the relevant goods, and to the means of acquiring a share of those goods. If we take future generations as an example, we are not stakeholders in the sense that they can affect us through our decisions, but we are still stakeholders in a pool of natural resources where they are also stakeholders. Nor does the notion of stakeholding presuppose reciprocity, and the idea of sharing principles does not presuppose a form of contract between stakeholders. 19 Additionally, as indicated by the example of border controls, in order to be stakeholders in a domain, we do not need to have the same stake in it, only some stake. We can also be stakeholders at different levels. For instance, in looking at global justice, even if most us are primarily stakeholders in the pie to be divided domestically, the size of that pie is clearly affected by distributive effects of how interactions on a global level are regulatedand so we are stakeholders there as well. This observation points to a picture where global justice and domestic justice might be different in terms of which principles that are reasonable where, but also to it being difficult to fully separate the two into distinct areas of inquiry.

The historical context
Practice-independent theorizing often seeks to articulate timeless principles of justice and, then, depending on at which stage of history we are at, the distance to that ideal might be greater or lesser. The idea here, however, is that in order to assess which principles that could be the object of an overlapping consensus for some relevant set of stakeholders, we cannot have an ahistorical understanding of the problems that those stakeholders need to deliberate about. While principles of justice need to lie at a certain level of abstraction in order to really be principles, in order to address relevant stakeholders they also need to be sensitive to the particular historical context of those stakeholders. There are at least three aspects to the historical context that seem reasonable to take into account in theorizing about justice: (a) The main characteristics of the moral and political traditions that largely govern the understandings that stakeholders will have of the domain in question.
Especially since Political Liberalism it is quite clear how Rawls' own theory of justice is strongly rooted in a particular, albeit relatively broadly conceived, traditionwhich also means that as a theory of domestic justice it will not be relevant to all societies. In developing constructivism beyond Rawls' own political constructivism this feature is still an important one, given that we are seeking a general reflective equilibrium. Principles of justice must be conceptually and imaginatively approachable from where we start in our thinking. Sometimes we start in identifiably distinct places and then it will be more likely that the principles around which we can form an overlapping consensus will be weaker. In articulating principles of justice, political theorists can certainly try to affect our self-understanding, but it seems reasonable that in general we should work under an assumption of considerable continuity: we accept that world-views and value systems evolve gradually and that reasonable principles of justice need to be shareable in an area of overlap between world-views and value systems that are recognizably similar to what we find already today, i.e. the world that is supposed to be regulated by those principles. (b) The main features of the backgrounding institutional framework likely to persist over some time. Ahistorical moral and political theorizing tends to proceed by identifying important (alleged) constants in human lives and societies. For practice-dependent theorizing there will instead be factors that contextually can be taken as if they are constants. To the extent that we seek principles that can be shared as principles regulating our practices, these principles need to be ones that address the types of decisions that we actually tend to take, and how these decisions can be framed or possibly reframed. This is something largely determined by the institutions that are already in place. For instance, if we look at questions of global justice in today's world, then the institutional framework for organized distributive justice, similar to what is found in many individual states, is arguably not there, nor does it even seem to lie on the horizon. The existing framework instead points to questions of justice on the global level having mainly to do with (i) international peace and security, and (ii) fairness in trade and investments. Existing frameworks can also have implications for how reasonable principles of justice need to be framed on a conceptual level. For instance, starting in current practices, there seems to be a prima facie case for framing matters in terms of human rights. While we can certainly have different views about how well human rights work as an international framework, as well as about how they should be interpreted and implemented, human rights simply are what comes closest to being an ethical lingua franca (Tasioulas 2007, 75). (c) Major historical wrongs or grievances that bear on how current circumstances are understood. Even in articulating principles of justice for specific domains, such domains will need to be described at a relatively high level of abstraction, rather than in all their particularity. This also means abstracting away from much of the historical background to the tensions that actually shape political discussions in specific societies. At the same time, current practices are often strongly shaped by the particular historical path that has taken us to them and there is a risk of falling prey to the problem of masking if history is not taken into account at all. The idea here is not that certain historical wrongs need to be identifiable as wrongs by some independent standard; what matters for an overlapping consensus is rather the subjective dimension: perceptions of historical wrongs or grievances that can undermine the well-orderedness of a situation where certain principles of justice are to be shared. Take global justice as an example. We could theorize global justice for any world that happens to be a world of sovereign and yet still interdependent states. Our world is such a world. That kind of theory could be applied to possible worlds which have no history of colonialism. 20 Our world is however clearly not one of those. By not directly addressing such a major factor as our history of colonialism, a theory would risk masking the ways in which that history has affected the relative positions of different countries in today's world, and how this history means that principles which are formally egalitarian in the rights and duties they assign risk amounting to simply being the object of a modus vivendi largely favouring those who have gained from the legacy of colonialism. This history therefore seems reasonable as input into the process of theorizing, rather than to be handled as an afterthought. An argument could for instance be made that a conception of global justice that is to be reasonable in the light of the history of colonialism needs to include principles enabling us to address important structural injustices. 21 Note that even with a relatively fleshed-out account of a domain, this kind of analysis of existing practices and their primary accompanying features is more like doing an inventory, understanding the construction site, the basic building materials, and the main features of the people supposed to inhabit the ultimate outcome of the construction process. The constructivist aims at coherence, in the Rawlsian sense of wide and general reflective equilibrium, but in contrast to how the Dworkinian method relies on an assumption of coherence in order to facilitate interpretation, the constructivist method is geared towards identifying the issues that need to be addressed in order to reach a coherent and shared set of principles of justice. The idea of well-orderedness provides a governing idea for this process, but it should certainly be recognized that actually reaching an overlapping consensus is more like an ideal limit. 22 The point is rather that constructivism allows our thinking about justice to be organized and structured already in the present, by thinking in terms of what well-ordered domains would have to be like.

Concluding remarks
Both Dworkinian interpretivism and Rawlsian constructivism are forms of practicedependent theorizing where we move towards a determinate conception of justice by relying on an analysis of existing practice in order to provide direction for the coherentist reasoning on which both approaches ultimately rely. A main distinguishing point between them is what kind of understanding of practices that they assume, and how two different ways of analysing current practices then become natural. It has been argued here that the Dworkinian model, while possibly still reasonable in the case of law, faces two major problems, masking and exclusion, when understood as a generalized method. It has also been argued that while there are aspects of Rawls' own constructivism which might appear to open it up to some such issues, especially the problem of exclusion, the prospects for developing a generalized form of Rawlsian constructivism are considerably more promising and a schematic account of such an approach has also been outlined. There is, of course, much more that needs to be said about its details, but the argument here is mainly that by relying on a deontic conception of practices, the Rawlsian can analyse practices without eliding the kinds of tensions and struggles that exist within real-life practices or the problematic histories that often underpin them.
text, as well as to the two reviewers for this journal, whose perceptive comments were incredibly helpful in shaping the final version of the text.