The basic liberties: An essay on analytical specification

We characterize, more precisely than before, what Rawls calls the ‘analytical’ method of drawing up a list of basic liberties. This method employs one or more general conditions that, under any just social order whatever, putative entitlements must meet for them to be among the basic liberties encompassed, within some just social order, by Rawls’s first principle of justice (i.e. the liberty principle). We argue that the general conditions that feature in Rawls’s own account of the analytical method, which employ the notion of necessity, are too stringent. They ultimately fail to deliver as basic certain particular liberties that should be encompassed within any fully adequate scheme of liberties. To address this under-generation problem, we provide an amended general condition. This replaces Rawls’s necessity condition with a probabilistic condition and it appeals to the standard liberal prohibition on arbitrary coercion by the state. We defend our new approach both as apt to feature in applications of the analytical method and as adequately grounded in justice as fairness as Rawls articulates the theory’s fundamental ideas.


Introduction
While it is widely held that some liberties are more important than others, there has been considerable controversy about which liberties count as basic and which do not. Some theorists have held that such moral rights as the right to engage in sit-ins and mass picketing during strikes, or to participate in other forms of direct action, do not count as, and can take priority over, freedoms that liberals tend to regard as basic (Gourevitch, 2018;Raekstad and Rossi, 2021). The question has also arisen as to whether a right to an element of workplace democracy, or economic democracy, is basic by liberal lights (Clark and Gintis, 1978;Gourevitch, 2014;McLeod, 2018;O'Neill, 2008). Moreover, there is a prominent and ongoing controversy about whether certain laissez-faire economic freedoms qualify, as Tomasi (2012aTomasi ( , 2012b argues, as basic liberties. 1 Such controversies have important consequences for political philosophy and its applications. Like Arnold (2018) and , we are concerned with the underlying question, about which there is also controversy, of how to draw the distinction (that Flanigan, 2018 considers unfounded) between basic and non-basic liberties. That is: how are liberals and their critics to decide as to which liberties are, according to liberalism, the most important or basic? Focusing on Rawls's account of the basic liberties will enable us to recognize and address some intricacies that arise when attempting to answer this question.
There is also a wider context. Rawls's version of high liberalism can be seen as driving a wedge between classical liberalism and left liberalism via its favoured list of basic liberties. As Brennan (2020: 493) explains, Rawls's list excludes both economic liberties that classical liberalism includes, and social liberties that left liberalism includes. This shows that deciding what makes its way onto the list of basic liberties is not merely an exercise in 'list-drawing'. It is really a debate about the fate of liberalism and particularly about which version of liberalism, if any, it is fitting to accept. Our contribution in this article can thus be seen as an intervention in this debate, albeit, as we explain below, in a way more fundamental than merely arguing that one liberty or another makes it onto the list. : 290) writes that there are two phases involved in providing a defensible specification of the basic liberties. The first involves specifying a list of basic liberties under general headings. The second involves further specification of this list by determining the significance of different particular liberties that come under the same general heading and adjudicating over conflicts between them. For example, after Nickel (1994: 780), while in the first phase freedom of movement is recognized as a basic liberty, in the second phase it is recognized that certain particular liberties of movement (e.g. going on holiday) are much less important than others (e.g. attending a political rally).
Further, Rawls ( : 44, citing 1971Rawls ( : 61, cf. 2005: 450) remarks that 'the basic liberties are specified by a list'. Rawls ( : 45, cf. 2005: 292-293) distinguishes between 'historical' and 'analytical' ways of drawing up a list. Proceeding historically, 'we survey various democratic regimes and assemble a list of rights and liberties that seem basic and are securely protected in what seem to be [. . .] the more successful regimes' (Rawls, 2001: 45). Presumably, in doing this, we are guided by the functional role that the basic liberties play within the theory of justice as fairness, rather than by a pre-existent list, however incomplete, of putatively basic liberties. We are to examine democratic regimes and identify which liberties commonly play, or approximate to playing, that functional role within them. Proceeding analytically, 'we consider what liberties provide the political and social conditions essential for the adequate development and full exercise of the two moral powers of free and equal persons' (Rawls, 2001: 45). We can also mix the two methods, giving a hybrid method. 2 There is, however, an important ambiguity in the notion of specification. To clarify matters, let us distinguish between specification as process and specification as a result of that process. Specification as process is what Rawls (2001: 45) calls 'drawing up a list'. An actual list of the basic liberties itself is, or is part of, a specification as result. The historical, analytical and hybrid methods are ways of carrying out the process of specification. In our terminology, the process of specification of the basic liberties consists in the application, within an epistemic situation, of a method (whether analytical, historical or hybrid), in order to arrive at a list of the basic liberties.
We focus upon the analytical method. We distinguish between the method itself and the setting out of general conditions apt for deployment as part of any application of that method. These general conditions are ones that, under any just social order whatever, putative entitlements must meet for them to be among the basic liberties encompassed, within some just social order, by Rawls's first principle of justice (i.e. the liberty principle 3 ).
These general conditions are to be deployed in all particular applications, which tend towards variation in their outcomes across different democratic regimes, of the analytical method of specifying the basic liberties. The general conditions determine what it is to be a basic liberty in the sense that they concern the intension of the concept basic liberty, not its extension. By contrast, the ultimate aim of the process of specification is to identify the concept's extension. As we have previously observed, we can also use the word 'specification' to refer to the result of that process, namely a list of basic liberties (in the first phase, specified at a high level of abstraction and, in the second, more particularly specified and within 'a fully adequate scheme of equal basic liberties', as : 42) puts it). Our main concern is with the general conditions pertinent to each phase of any process of analytical specification of the basic liberties. Arrival at general conditions appropriate for application in analytical specification of the basic liberties is, we take it, a theoretical task that takes place as part of the setting up of justice as fairness as a theory. It is distinct from the task of drawing up a list of the basic liberties in that the latter, and not the former, takes place, with increasing particularity, over the four stages of the four-stage sequence (Freeman, 2007: 73;: 289-371; Wenar, 2017: §4.9). We reserve the term 'specification' for the process of carrying out this task, done using one of the three available methods, and for its result.
Our main task, then, is to render more precise the concept of basic liberty, rather than to settle questions about its extension. Fully settling the concept's extension is not a matter for philosophers only: rather, it is the same task as the provision of a full specification of the basic liberties. A full specification, as we understand it, includes a full list of particular basic liberties that feature within an overall scheme of liberty. Accordingly, the settling of the concept's extension must take place across both phases of specification, and the stages of the four-stage sequence. The task in which we are engaged is the determination of the general conditions apt to feature in any application of the analytical method. This task is prior to, and importantly distinct from, that of analytically specifying the basic liberties (in the first and second phases of specification). We aim to clarify what the analytical method is and to provide tools, in the form of the aforementioned general conditions, for its application.
In the literature, answers to the question of what it is for a liberty to be basic (as against mere lists of putatively basic liberties) do not abound (cf. Pettit, 2008: 201, 203). Rawls's account is primarily a 'functionalist' one: it concentrates mainly on the role that the basic liberties play, rather than dwelling, in detail, upon defining what they are. Moreover, his account of the analytical method itself, and of the general conditions pertinent to it, is relatively sparse. We aim to expand upon it, criticize it and improve upon it.
We interpret Rawls's account of what makes a liberty basic in terms of what we call the 'necessity reading'. Rawls defines as basic those liberties that are necessary for the full and informed exercise of what Rawls considers our two moral powers. We then turn to Rawls's general condition for use in the second phase of specification. This appeals to what Rawls depicts as the exercise of the moral powers in the two fundamental cases. We argue that, coupled with the necessity reading, invoking the two fundamental cases is too strong. It undergenerates: i.e. it excludes some particular basic liberties that it should not, such as certain forms of freedom of movement. We then discuss a new approach that seeks to resolve this problem. On the new approach, probability takes over the role that was occupied, on Rawls's approach and some variants of it, by modality.
We proceed to show, however, that the probabilistic approach remains extensionally inadequate. We argue, by appeal to the case of the freedom to produce or consume political satire, that it also under-generates. We go on to propose a remedy for this problem. This adds as a sufficient (but nonnecessary) condition upon a particular liberty's being basic that any restrictions upon it that did not promote, or which were not designed to promote, the weighting of liberties in a scheme of liberty would be arbitrarily coercive. From a specifically Rawlsian perspective, such restrictions would breach the liberal principle of legitimacy.

Rawls's proposal: A reconstruction
Rawls's conception of justice as fairness regards citizens as persons engaged in social co-operation who have 'two moral powers', namely the capacity to have a sense of justice and the capacity to have a conception of the good (Rawls, 2001: 18-19). The principles of justice concern the design of the basic structure of society, that is: the way in which the main political and social institutions of a society fit together into one system of social co-operation, and the way they assign basic rights and duties and regulate the division of advantages that arise from social co-operation over time. (Rawls, 2001: 10) The fundamental case in which the capacity for a sense of justice is exercised is in 'the application of the principles of justice to the basic structure and its social policies' (Rawls, 2001: 112). The fundamental case in which the capacity for a conception of the good is exercised is in 'forming, revising, and rationally pursuing such a conception over a complete life' (Rawls, 2001: 113).
The two moral powers and their exercise in the two fundamental cases feature importantly in the conditions that Rawls thinks a liberty must meet if it is to be a 'basic' particular liberty. As explained earlier, such a liberty is one that emerges from each of the two phases of specification as (what we call) a 'core' basic liberty. It is a particular case of the basic liberties identified (under general headings) in the first phase and it is identified, in the second phase, as being of substantial political significance.
Although Rawls does not pursue the matter in detail, he seems to intend that the analytical method should be apt to be invoked during both phases of specification. For Rawls, a liberty is basic if and only if it is necessary to the provision of 'the social conditions essential for the adequate development and the full and informed exercise of [people's] two moral powers [. . .] in [at least one of] the two fundamental cases' (Rawls, 2001: 112;our italics). 4 Necessity to the 'adequate development and the full and informed exercise' of the moral powers is Rawls's general condition for use in the first phrase of specification. The supplementation of this with the appeal to the two fundamental cases provides his general condition for use in the second phase. Evidently, it is not only the possession of the two moral powers that is essential to a person's being a free and equal citizen; it is also the exercise of them, in the two fundamental cases (Nickel, 1994: 773;Rawls, 2001: 43, 112). The basic liberties pertain to Rawls's first principle of justice, that: 'Each person has the same indefeasible claim to a fully adequate scheme of equal basic liberties, which scheme is compatible with the same scheme of liberties for all [. . .]' (Rawls, 2001: 42). This is why the exercise of our moral powers in respect of the two fundamental cases is 'essential to us as free and equal citizens' (Rawls, 2001: 45) and no trade-offs are to be made of basic liberties for other social primary goods (Rawls, 2001: 46-47).
Let us briefly consider three challenges to our account of the analytical method. First, perhaps the conception of the analytical method with which we are working, according to which any version of the analytical method employs a definition of the basic liberties, is one that Rawls would have considered alien. 5 We do not see, however, how the analytical method, as a general method (i.e. rather than by reference to a specific version of it) could otherwise viably be characterized. Even if Rawls did not intend to provide a definition, we do not see how that task can properly be regarded as dispensable.
Second, it could be asked whether the appeal to the two fundamental cases does indeed play a role in Rawls's account of the basic liberties. 6 While it is true that Rawls does not always mention the two fundamental cases in some relevant contexts (e.g. Rawls, : 45, 2005: 293), he does mention them in others (e.g. Rawls, : 112, 2005: 308). Philosophically, either the two fundamental cases are to be invoked in the first phase of specification or they become relevant only in the second phase; there does not seem to be reason not to invoke them at all given the specifics of the Rawlsian system. We believe, with Nickel (1994: 780-781) and in view of : 332-335), that the latter is the correct interpretation. The appeal to the two fundamental cases serves to help us to specify which particular liberties under an already selected (first-phase) basic liberty, specified at a higher level of abstraction, are of particular political importance (i.e. which are the core areas of that basic liberty).
Finally, one could argue that we read Rawls too strongly: it is the notion of enabling, not that of necessity, that is really relevant to understanding Rawls. In particular, after he gives the above general conditions, Rawls remarks that 'equal political liberties and freedom of thought enable citizens' to judge 'the justice of the basic structure of society and its social policies' and that 'liberty of conscience and freedom of association enable citizens' to form, to revise and rationally to pursue 'their conceptions of the good' (Rawls, 2001: 45;our italics). This is an interesting modal shift. Rawls's general condition (for the first phase of specification) appeals to the essentiality/necessity of the basic liberties to the full and informed exercise of the moral powers. The explanatory remarks just quoted appeal to the weaker claim that the basic liberties enable this. 7 The notion of enabling, however, cannot be the right notion to feature in an extensionally adequate definition of the basic liberties (cf. Brennan, 2020: 503;von Platz, 2014: 29). While all social primary goods enable the full and informed exercise of the moral powers : 307, cf. 2001: 57), not all social primary goods are basic liberties. Therefore, if enabling were to supplant necessity in the definition of the basic liberties, the resultant definition would be too permissive to be consistent with Rawlsian liberalism. No doubt, basic liberties, being social primary goods, have an enabling function, but that is true of all social primary goods and not only of the basic liberties (cf. Melenovsky, 2018: 434). While enabling is a necessary condition for a social primary good to qualify as a basic liberty (since the basic liberties are among the social primary goods), it is not a necessary and sufficient condition.

From modality to probability
We have no immediate quarrel with the proposed necessity reading as providing a general condition apt for application in the first phase of specification, fit to provide a means of identifying the basic liberties at a high level of abstraction (cf. Nickel, 1994: 783-786). Nevertheless, we contend, it provides an inappropriate general condition for the second phase of specification.
Is the presence of the basic liberties a necessary condition for the full and informed exercise of the moral powers in the two fundamental cases? A positive answer to that question is not, if we are to be charitable to Rawls, to be construed as suggesting that, for each individual citizen in a given society, the full and informed exercise of the moral powers in the two fundamental cases is possible only if all citizens possess the basic liberties (after Melenovsky and Bernstein, 2015: 50;cf. Melenovsky, 2018: 442-443). Highly developed moral and political sensibilities, including the capacity for a sense of justice and the capacity to have a conception of the good, may be present and exercised, even to a relatively high degree, in populations living under regimes in which the basic liberties recognized by Rawls are not afforded equally to all citizens: e.g. in the United Kingdom before universal suffrage, or in contemporary Cuba. Living under such a regime lessens the likelihood that people will be able, in a full and informed way, to exercise their moral powers in the two fundamental cases but it would not necessarily make it impossible for them to do so. While governmental refusal to afford equal basic liberties may place limits upon these two moral powers and their exercise in the two fundamental cases, it need not make these powers and their exercise in the two fundamental cases impossible for each and every citizen (cf. Arnold, 2018: §3;Brennan, 2020: 499-500).
Recall that for Rawls (2001: 112) a liberty is basic if and only if it is necessary to the provision of 'the social conditions essential for the adequate development' and/ or 'the full' and/or 'informed exercise of [people's] two moral powers [. . .] in [at least one of] the two fundamental cases'. Our interpretation of this is that a liberty is basic if and only if it is necessary to the provision of the social conditions that must be in place in order for it to be the case that every citizen naturally capable of doing so is not hindered, by arrangements relating to the basic structure of society and its laws, from exercising the moral powers in a full and informed way in the two fundamental cases.
Nevertheless, the appeal to necessity remains too strong. While the connection between the basic liberties and Rawls's conception of citizenship in a well-ordered society is indeed a necessary one, some liberties that ought, at least by Rawlsian lights, to qualify as particular basic liberties within any just social order are, as Arnold (2018) has argued, only contingently connected with the full and informed exercise of the moral powers in the two fundamental cases. Arnold (2018: §4.2) illustrates this point using various counter-examples to the necessity claim that are intended to establish that universal full and informed exercise of the moral powers in the two fundamental cases can be compatible with laws that deprive citizens of core liberal freedoms. Here, we summarize just one of them. For : 335), 'the liberty and integrity of the person' are 'violated [. . .] by denial of freedom of movement'; thus, freedom of movement is a basic liberty. Arnold's discussion suggests, however, that if a law were enacted that restricted people's freedom of movement to within their metropolitan areas, this would not make it impossible for every citizen to possess, and exercise in a full and informed way, the moral powers in the two fundamental cases (cf. Pogge, 2007: 87). While we think it very unlikely that, under such conditions, every citizen would be able to do this, we agree with Arnold that it does not seem to be impossible.
A possible move in response to the above argument, so as to secure the status of such particular core liberal freedoms as basic, might be to give up the appeal to the two fundamental cases (second phase) but to retain the necessity reading (first phase). An obvious problem with this suggestion is that it is not clear as to what might, for the purposes of adjudication between conflicting basic liberties, replace the appeal to the two fundamental cases. 8 Also, even if we find an answer to this question, another problem presents itself. In the second phase of specification, we are not only looking further to specify the basic liberties identified, at a higher level of abstraction, in the first phase; we must also employ a method of adjudication. Adjudication requires an account of the relative significance of different liberties. Here, Rawls proposes that: a liberty is more or less significant depending on whether it is more or less essentially involved in, or is a more or less necessary institutional means to protect, the full and informed exercise of the moral powers in one (or both) of the two fundamental cases. (Rawls, : 113, cf. 2005: 335) Before we discuss this quotation further, let us consider an ambiguity in it. A sentence of the form 'A is essentially involved in B' may be taken to mean either that A is essential to B (i.e. in this context, that B cannot happen without A) or (non-equivalently) that A cannot be A without being involved in B (i.e. if A exists or occurs then, thereby, B does, at least partly, too). We take the first reading to be the one that is apposite to interpreting Rawls: it accords with Rawls's own talk of the basic liberties being (either essentially or instrumentally) necessary for the full and informed exercise of the moral powers in the fundamental cases (Rawls, ]: 208, cf. 2001. For simplicity, let us set aside the reference to the two fundamental cases. Adding them to the present discussion would make no philosophical difference to our points. The idea is now that the weightings of particular basic liberties, within 'a fully adequate scheme' of basic liberties, once a list of basic liberties (specified at a high level of abstraction) is already in place, can be guided by measuring how necessary or essential they are to, or as institutional means to protect, the full and informed exercise of the two moral powers. This idea, however, is bizarre. Rawls, when providing his modal definition of the basic liberties, does not appeal to degrees of essentiality or necessity. In the remark about degrees of significance, he does so. A liberty, however, cannot (for example) both be essential (or necessary) for the adequate development and/or the full and/or informed exercise of the moral powers and not so essential (or necessary) to this. Following McLeod (2018: 247), degrees of significance cannot, consistently with Rawls's attempt to define the basic liberties, be spelled out in terms of degrees of essentiality (or necessity). 9 Distinguishing between the two phases of specification does not help resolve this problem.
While significance admits of degrees, this is not true of necessity (or essentiality). Even if there are various notions of necessity that differ in strength, when we are working with a single notion of necessity there are no degrees within it. While the significance of a liberty is a scalar property of the liberty, its necessity (or otherwise) for the full and informed exercise of, or as a necessary institutional means towards the protection of, the moral powers is a binary property: either a liberty is so necessary or it is not. If it is not, then it is at best contingently connected with the full and informed exercise of the moral powers.
Thus, supposing that some particular liberties are more significant than others, but that the division is not simply between those that are significant and those that are not, the proper way to cash this out cannot be in terms of degrees of necessity of the liberties in question to the full and informed exercise, or institutional protection, of the moral powers. Rather, if differences in degrees of significance of particular liberties are to be correlated with other gradated differences, then the appropriate appeal should be to the extent to which it is probable that, in the absence of a given particular liberty's taking on the functional role of a basic liberty, the full or informed exercise of (at least one of) the moral powers will, partly due to social conditions, be significantly impeded, stunted or atrophy (McLeod, 2018: 246-247).
We re-emphasize that this observation relates not to how an appropriate list of basic liberties is to be drawn up in the first phase but, rather, to the question of how, in the second phase, particular basic liberties are to be ranked in a fully adequate scheme of liberties. In defining what it is for a liberty to be basic, instead of appealing, as Rawls does, to a purported modal relationship between the basic liberties and the full and informed exercise of the moral powers in the two fundamental cases (or the institutional protection of liberties intrinsically related to this), we introduce the idea of mitigating against risk, above a certain threshold, to the full and informed exercise of the two moral powers. Accordingly, let us consider the suggestion that a liberty is basic if and only if the likelihood is above a certain threshold that, in its absence, and partly due to social conditions, the possession and/ or the full and informed exercise of one or both of the moral powers will be prevented, stunted or atrophied.
Let us explain how we think that this suggestion improves upon Rawls's account. For Rawls, the first phase of specification appeals to the supposed necessity of the basic liberties to the full and informed exercise of (or, in some cases, institutional protection of liberties intrinsically related to) the moral powers. In the second phase, and so as to identify the core areas of the basic liberties, he introduces the appeal to the two fundamental cases. Finally, his account of how adjudication between the resultant particular liberties works, within an overall scheme of liberties, uses the notion of degrees of necessity (essentiality). We replace the modal notion in Rawls's account, necessity, with the non-modal notion of probability. Thus, a basic liberty is a liberty in the absence of which the risk to the possession, and/or the full and informed exercise, of the moral powers rises above a certain threshold. Again by contrast with Rawls's account, this new general condition is apt for deployment in both phases of specification of the basic liberties: nothing needs to be added to it to cope with the second phase. A remaining task for the second phase, on both Rawls's account and according to our new suggestion, concerns the weighting of, and the adjudication between, the particular basic liberties within an overall scheme of liberties. Here our idea is simple. Adjudication proceeds using probability assessment above the specified threshold of risk. That is, we compare (ordinally or cardinally) the different risk levels to each other so as to measure the relative significance of different particular basic liberties. If the absence of a certain liberty would pose a higher risk to the possession of, and/or the full and informed exercise of, the moral powers than another, then the first of these liberties is more significant than the second.
We take it that our appeal to probability, in place of necessity, is no ad hoc move to save a broadly Rawlsian account of the basic liberties. If, in an account of the basic liberties, one wishes to retain Rawls's appeal to the moral powers then this appears to be the only way to proceed: at least, it seems to be the only plausible proposal on the table. Arnold's imaginary example provides a good illustration. His objection to Rawls's account withers away when necessity is supplanted, along the lines we suggest, with the appeal to a probability threshold.
Of course, a lot depends on how and where the threshold is set. It is neither crucial nor desirable that this question should be settled here and now. Our concern has been to arrive at general conditions apt for deployment in any process of analytical specification of the basic liberties. The identification of these conditions is a philosophical task that takes place outside of the four-stage sequence. Application of the analytical method, via which a list of particular basic liberties is drawn up, is a process that happens across the stages of the four-stage sequence and which, on both Rawls's own modal account and on its probabilistic counterpart, increasingly involves empirical enquiry. This is also true of threshold-setting. Now, according to some theorists (Mikl os, 2011;Mikl osi, 2008), the stages of the four-stage sequence see institutions, such as those constitutive of the basic structure of society, as determining moral content by carrying out, among other things, just the kind of specification job needed to make the above probabilistic approach work. Rawls's own discussion of the four-stage sequence further underlines this idea. It shows that specification and adjudication take place in an institutional setting and that the constitutional, legislative and executive stages interact: they feed material into each other with theorists going back and forth between them, as would be expected in a process of reflective equilibrium (Rawls, 1971(Rawls, : §31, 2005: 334-356).
To apply this framework to our case, what would happen, we surmise, is not so much that participants in these stages would literally discuss a threshold (although that can happen, e.g. in the work of the courts) but that they would come up with guidelines designed to reflect the abstract account given by the theorists, hand over their 'results' to the theorists (who could then apply them in the theory) and then repeat this process, going back and forth between the theorists and the other participants. 10 The emerging picture, then, is that of a process for setting the threshold that harmonizes with Rawls's ideas about the role of institutions with respect to justice.

The probabilistic approach: An outstanding problem
When discussing Rawls's modal definition of the basic liberties, in terms of their supposed necessity to the full and adequate possession and exercise of the moral powers in the two fundamental cases, we agreed with Arnold that some of the core areas of basic liberties that instantiate liberties, more abstractly specified, in Rawls's list of the basic liberties are not necessary in this way. Thus, the modal definition is extensionally inadequate because it under-generates when applied in the second phase of specification. Unfortunately, the probabilistic approach is vulnerable to counter-examples that show it to be extensionally inadequate in the same way. One such counter-example is the freedom to produce or to consume political satire.
Freedom of political satire presents a dilemma for the probabilistic approach. Suppose, for the first horn of the dilemma, that any satirical content can be conveyed non-satirically. (In doing this, suppose that the content of political satire can exhaustively be expressed, as, for example, some philosophers hold that the content of a metaphorical sentence can exhaustively be expressed, in literal language. 11 ) Suppose, further, that the probability is high that the content of any given piece of putative political satire that agents would otherwise be inclined to create will instead be expressed non-satirically should political satire be legally debarred. Other things being equal, it follows that freedom of political satire is not, on the probabilistic approach, a core area of the related basic liberty (namely freedom of speech) because (under these suppositions) the equivalence of expressions of satire (whatsoever their format, so including mere images) to literal language means that a ban on political satire would not, of itself, make it likely that the full and informed exercise of the moral powers (in the two fundamental cases) would be stunted or atrophy. Under our suppositions for his horn of the dilemma, even if political satire were legally debarred (supposing, contrary, we assume, to fact, that sufficiently precise legislation could be drafted, passed and implemented), the content of its message would, other things being equal, readily be effectively communicable, as no merely theoretical possibility, by non-satirical means. Accordingly, citizens' exercise of the capacity for a sense of justice would not be precluded from being full and informed: for neither full nor informed exercise of a moral power requires that for every manner of conveying facts and judgements about the justice of the basic structure and its laws citizens can legally have those facts and judgements conveyed to them in that manner. It is each citizen's possession of an adequate stock of relevant information, alongside access to that stock of relevant information that is possessed by the citizen's epistemic community, that matters here. It is not how the information is conveyed that matters. A restriction on the latter does not result, of itself, in compromise to the full and informed exercise of the relevant moral power in the relevant fundamental case. Now suppose, on the other hand, and for the second horn of the dilemma, that political satire can include satirical content that is not literally expressible. The case of a law banning political satire can then be contrasted with Rawls's discussion of the repression of 'subversive advocacy'. He writes: As Kalven observes, revolutionaries don't simply shout: 'Revolt! Revolt!' They give reasons. To suppress subversive advocacy is to suppress the discussion of these reasons, and to do this is to restrict the free and informed public use of reason in judging the justice of the basic structure and its social policies. And thus the basic liberty of freedom of thought is violated. : 346) On this horn of the dilemma, political satire can include rhetorical elements that outstrip reasons (since, for example, it can consist of an image alone). We presume that it is an essential feature of reasons that they can feature as premises in inferences (whether within theoretical or within practical reasoning). 12 We take it, therefore, that reasons can be stated in literal language. If reasons can be stated in literal language, while (on this horn of the dilemma) not all satirical content can so be stated, a law against political satire cannot be criticized (on this horn) for being a restriction on the free and informed public use of reason; for (on this horn) it is not that. It is, at best, a restriction on the free and informed use of rhetoric (i.e. in this case, the manner in which reasons are expressed). Given that the probabilistic approach to defining the basic liberties preserves, from Rawls's modal definition, the end relative to which the basic liberties are to be defined, namely, the full and informed exercise of the moral powers, the probabilistic approach is unable to secure freedom of political satire as a basic liberty.
The outcome of this dilemma is that the end relative to which Rawls defines the basic liberties is too narrow to encompass freedom of political satire. 13 It might be objected that the proper response to this dilemma is to bite the bullet and to accept that freedom of political satire simply does not count as a particular liberty that qualifies as basic. Freedom of political satire might still be apt for political protection, but whatever political protection it enjoyed would not be the special protection that is afforded to the basic liberties. We decline to bite the bullet. First, we hold that there are good reasons to take freedom of political satire to be a core area of freedom of (political) speech (and, more widely, of expression and, hence, of thought), which in turn is a core area of the group of basic liberties that Rawls calls 'political liberties'. 14 We think that the status of freedom of political satire as a core basic liberty is clear. 15 In many repressive regimes, and even in particularly sensitive cases in non-repressive regimes, the most effective method of social critique is sometimes through highly abstract satirical pieces: indeed, sometimes this is the only legally or politically available method.
Second, we believe that there are good reasons to think that freedom of political satire should be afforded the kind of protection that is afforded only to the basic liberties. The approach to political liberty that is embedded within Rawlsian liberalism already has readily available resources to accept this claim. On Rawls's view, if a liberty is basic then, within its 'central range of application', only restrictions upon it that promote the overall balance of basic liberties within a scheme of liberty can be justified (Rawls, 1971(Rawls, : 244, 2001(Rawls, : 111, 2005: 295; also Freeman, 2007: 65-72, 81-82). Our task now is to show that freedom of political satire indeed falls within the central range of application of freedom of expression (and thought).
Let us begin by contrasting political satire with non-political satire. The freedom to produce or consume non-political satire, for example about the consensual sexual antics of stars of 'reality' television, while also an instance of the more abstract liberty of freedom of expression, does not have the special protection afforded to the basic liberties. This is because, in the case of this type of satire, the realm (i.e. in this case, the subject matter) of the exercise of the more abstract liberty is largely irrelevant to the full and informed exercise of one or both of the moral powers. Notably, this sort of non-political satire does not concern 'the justice of the basic structure and its social policies' (Rawls, 2001: 112). The more important forms of political satire do not just ridicule or draw attention to unjust aspects of the basic structure and its social policies; they encourage their audience to do so too (at least attitudinally, if not out loud).
In accordance with its liberal nature, a Rawlsian account of freedom of expression must reckon, when assessing whether particular liberties that instantiate this freedom are to be afforded special protection, not only with the content of the expression but also with the multiplicity of ways in which that content may be expressed. Restrictions upon freedom of expression come in three broad forms: of the message (e.g. one is not allowed to criticize the head of state), of the medium (e.g. one is not allowed to criticize the head of state in print) and of the manner in which the message is expressed (e.g. one is not allowed to publish satire that lampoons the head of state). Crucially, freedom of expression is not only freedom to express the content of the message: it is freedom of medium, and of manner, too. We consider restrictions on the medium, or on the manner, that do not serve to promote the balance of particular liberties in an overall scheme of liberties to be, in the spirit of Rawls's high liberalism, just like restrictions on the expression of the content of the message, unjustifiable. That is, if there are many ways in which a particular basic liberty, such as freedom of political expression, may be exercised then the state cannot be within its rights arbitrarily to rule that only some of them, specified at an even lower level of abstraction, are allowed.
If a particular liberty is non-basic, then, in principle, there can be restrictions upon it that do not promote the overall balance of basic liberties within a scheme of liberty: this is so because of the particular liberty's realm. When a particular liberty is an instance of a basic liberty, then, while this does not of itself confer basic status upon that particular liberty, we suggest that the mere availability of an alternative manner of exercising that particular liberty is not enough to render it permissible for the state, for a reason other than the promotion of the overall balance of basic liberties within a scheme of liberty, to restrict that particular liberty. Since freedom of political satire is a particular instance of freedom of political expression, restrictions upon it could only be permissible for reasons of realm. There can be no such reasons, because the realm of freedom of political satire is exactly that of freedom of political, but non-satirical, expression. Political satire, even if it is not necessary for evaluation of the justice of the basic structure and its social policies, includes such evaluation in its characteristic subject matter. We conclude that freedom of political satire is on a par, in terms of its significance, with freedom of political but non-satirical expression.

Arbitrary coercion and the principle of legitimacy
Given that every version of the analytical method that we have so far discussed is extensionally inadequate, how are we to provide appropriate general conditions for the analytic method? We believe that progress can be made via appeal to the standard liberal injunction against arbitrary coercion. From a purely Rawlsian perspective, this injunction is embodied in a more specific principle that underlies Rawls's political liberalism, namely 'the liberal principle of legitimacy' which states that 'political power is legitimate only when it is exercised in accordance with a constitution (written or unwritten) the essentials of which all citizens, as reasonable and rational, can endorse in the light of their common human reason' (Rawls, : 41, cf. 2005: 137). 16 In our discussion of Rawls's modal approach to the provision of general conditions for analytical specification of the basic liberties, we focused on what we took to be an under-generation problem. From the various examples provided by Arnold, we used the case of freedom to move outside of one's own metropolitan area to illustrate this under-generation problem. Like Arnold, and Pogge before him, we did not argue that this freedom is a matter of basic liberty. Instead, we took it as rather obvious that it is, and, accordingly, that the failure of Rawls's modal approach thus to recognize it was a case of unintended failure of extensional adequacy.
In seeking to fix that problem, we suggested that Rawls's modal approach might be replaced by a probabilistic approach. That, we claimed, also suffers from an under-generation problem in that it does not provide for the inclusion of freedom of political satire within the basic liberties. In that case, we argued that freedom of political satire is a kind of freedom that should be regarded as basic by any liberalism worthy of the name. The failure of the previously discussed approaches to encompass that freedom is, we suggest, again an unintended failure of extensional adequacy.
If that is correct then an improved approach must, without resort to overgeneration, resolve this under-generation problem. Moreover, it is desirable to explain the basic status of freedom of political satire in a manner that does not rest only on a particular feature of that freedom, but that also generalizes to other cases, such as freedom of movement outside of one's own metropolitan area. Our argument, in the previous section, that freedom of political satire is basic appeals to features that are specific to its status as a form of freedom of political thought and expression. We will now establish its status as basic in a more general way, that encompasses other particular freedoms, and that is independent of the specific features of freedom of political satire to which our earlier argument appealed.
The source of the under-generation problem resides, we suggest, in a wider failing with the modal and probabilistic approaches to the provision of general conditions fit for deployment in analytical specifications of the basic liberties. In relating the basic liberties exclusively to the full and informed exercise of the two moral powers, either insufficient attention or no attention at all is paid to what must be regarded, by any political philosophy fit to brand itself as liberal, to issues of coercion and, in particular, to cases of what are, from any liberal perspective, arbitrary exercises of political power.
Accordingly, we propose that a general condition that is genuinely apt for deployment when specifying the basic liberties analytically will be along the following, disjunctive, lines. An entitlement is a basic right or liberty if and only if at least one of the following conditions holds: (i) the likelihood is above a certain threshold that, in its absence, and partly due to social conditions, the possession and/or the full and informed exercise of one or both of the moral powers will be prevented, stunted or atrophied; (ii) any legal restriction upon it that did not promote the weighting of liberties in a scheme of liberty would be arbitrarily coercive, i.e. an arbitrary exercise of political power.
Rawls operates with a specific understanding of what arbitrarily wielded political power is and of why it is unacceptable. In his thought, the principle of legitimacy serves to set out a necessary but non-sufficient condition upon the morally acceptable exercise of political power. By 'political power', he means nothing but coercion by the state: it 'is always coercive power backed by the government's use of sanctions' : 136, cf. 216). Moreover, 'in a democracy political power [. . .] is the power of the public, that is, of free and equal citizens as a collective body' : 216). The principle of legitimacy therefore imposes upon citizens 'the duty of civility -to be able to explain to one another', concerning fundamental questions, 'how the principles and policies they advocate and vote for can be supported by the political values of public reason' . : 446-447) depicts the principle of legitimacy, grounded in the principle of reciprocity, as follows: our exercise of political power is proper only when we sincerely believe that the reasons we offer for our political actions-were we to state them as government officials-are sufficient, and we also reasonably think that other citizens might also reasonably accept those reasons. This criterion applies on two levels: one is to the constitutional structure itself, the other is to particular statutes and laws enacted in accordance with that structure.
An exercise of state power will be arbitrary, on Rawls's account, then, when it breaches the principle of legitimacy thus stated: that is, when it constitutes a breach of the reciprocity that is integral to public reason and to his conception of citizens as free and equal parties to social co-operation. Indeed, : 447) writes that denial of a basic liberty normally involves a failure to observe the principle of reciprocity. Of course, liberal alternatives to Rawls's specific understanding of arbitrary coercion, of its wrongfulness, and to his principles of legitimacy and reciprocity, are available. We have therefore worded (ii) in a way that we intend to be neutral across different liberal accounts of these matters.
The adoption of our new, disjunctive condition, rather than the abandonment of the project of providing general conditions fit for use in any analytical specification, is vindicated by the fact that the principle of legitimacy, and the conceptions of reciprocity and public reason from which it derives, are all at the core of Rawls's political-liberal account of justice. Our approach, then, does not involve an ad hoc move. In particular, our new general condition, although disjunctive, is united in that the concepts in each disjunct that are most salient to the politicalliberal account of justice have the very same foundations. Rawls's conception of persons as 'free and equal beings with a liberty to choose' (Rawls, 1971: 256), although prior historically to Rawls's version of the principle of legitimacy and to the integration into his theorizing of considerations about the moral powers, underlies not only (i) but also (ii)'s injunction, manifested in Rawls's case by the principle of legitimacy, against arbitrary coercion. Rawls's writings clearly show that the principle of legitimacy, which justifies (ii) from a specifically Rawlsian perspective, is rooted in the fundamental ideas of the person (citizen) and of society, the exact two foundational reference points for the moral powers that are mentioned in (i) as well. 17 Moreover, there is good reason to read Rawls as suggesting that the principle of legitimacy is a constraint on reasoning about (and therefore, upon the determination of any fully adequate scheme of) the basic liberties. 18 In seeking to solve the under-generation problem that has exercised us, our invocation of the liberal hostility to the arbitrary exercise of political power, in Rawls's specific case articulated via the principle of legitimacy, is therefore by no means an ad hoc move.
Let us relate (ii) to the two under-generation cases previously discussed, namely freedom of movement outside one's metropolitan area and freedom of political satire. From a liberal perspective, a law that restricted people's movements to within their own metropolitan areas, and which did not promote the weighting of liberties in a scheme of liberty, would be wrong not because, or not primarily because, of prejudicial effects on the full and informed exercise of the moral powers. Rather, its wrongness would reside primarily in its constituting an arbitrary exercise of political power.
We think that (ii) captures exactly what would, from a liberal perspective, and independently of the particular features of freedom of political satire to which we appealed in the previous section, be wrong with a ban on political satire. Now that we have invoked (ii), doing so in Rawls's specific case via the principle of legitimacy, a further particular feature of freedom of political satire becomes salient. From a Rawlsian perspective, one of the most worthwhile aspects of freedom of political satire is that it is freedom to highlight, and to ridicule, exercises of political power that are improper, in that they are not founded on reasons that, on their merits, are sincerely believed to be ones that citizens might reasonably accept. More broadly, political satire highlights, and ridicules, the attitudes, beliefs and behaviours associated with the arbitrary exercise of political power.
One might still object to our proposal on the ground that condition (ii) renders condition (i) redundant because every liberty that meets (ii) also meets (i) but, as the case of freedom of political satire shows, not vice versa. This objection fails. Legislation that stipulates fixed-term parliaments puts a restriction on the exercise of the right to vote, and not to promote the overall weighting of liberties in a scheme of liberty. However, such legislation is not a case of arbitrary coercion. While (i), and not (ii), secures the right of universal suffrage, (ii), and not (i), secures freedom of movement outside of one's metropolitan area and freedom of political satire.
A final objection to our proposal is that, in dampening down the bump in the carpet concerning freedom of political satire, we have caused another bump to appear. : 364-365) writes that the right to advertise, while a form of freedom of speech, is not a basic liberty because, in the case of 'marketstrategic advertising', the right to advertise 'can be restricted by contract, and therefore [. . .] is not inalienable'. Does our lowering of the bar on what it is to be a basic liberty not, in admitting freedom of political satire, also have the undesirable result, for Rawls, that the right to engage in market-strategic advertising also turns out to be a basic liberty? It does not. Restrictions on market-strategic advertising, even if these are legally imposed (as opposed to just being contractually agreed), can be consistent with the principle of legitimacy and need not amount to arbitrary coercion. This is because reasonable and rational people may agree with : 365) that market-strategic advertising can be socially wasteful.

Conclusion
Both the question as to which liberties are basic and the nature of the analytical method of specifying the basic liberties are significantly under-theorized in the literature. This is despite the clear importance of the notion of basic liberties for liberal political theories.
We have emphasized, after Rawls, that any theory of the basic liberties must distinguish between two phases of specification: in the first, a highly abstract list of liberties is drawn up under general headings; in the second, this list is further specified into core areas of these liberties (that matter in 'real life') and adjudication takes place. It seems to us that much of the existing literature tends to entangle these two phases, insufficiently distinguishing between them.
In the framework provided by this distinction, we gave an account of Rawls's characterization of the analytical method of specifying the basic liberties as centring on a necessary connection between the basic liberties and the full and informed exercise of the two moral powers (first phase) in the two fundamental cases (second phase). We then argued that Rawls's appeal, in his definition of the basic liberties, to a modal relationship should be replaced instead by the appeal to a probability threshold. Using the case of freedom of political satire, we argued that even a probabilistic general condition under-generates. We remedied this by disjoining probabilistic considerations with a clause that makes direct appeal to the liberal injunction against arbitrary coercion.
While our proposed approach is motivated by purely analytical considerations, the concept of basic liberty that it involves is more inclusive, in terms of the rights and liberties that fall under it, than is the letter of Rawls's own discussion. The case of freedom of political satire is but one example among many. For example, while O'Neill (2008: 41-42) may be right that an element of workplace democracy is not necessary to the full and informed exercise of the moral powers, this does not settle the question of whether a right to an element of workplace democracy is basic (cf. McLeod, 2018). The same applies to the laissez-faire economic freedoms, such as freedom of contract, that Tomasi (2012aTomasi ( , 2012b regards as basic liberties. The question to ask about each freedom, provided that it is not the case that any legal restriction upon it that did not promote the weighting of liberties in a scheme of liberty would be arbitrarily coercive, is whether its lack would raise above the threshold risk to the full and informed exercise of the moral powers. Empirical psychology and social science therefore have even greater roles to play than Rawls thought. We hope to have assisted a broadly Rawlsian account of the basic liberties in defending itself against criticisms from thinkers both to the left and to the right of Rawls. As noted earlier, the discussion about the correct list of basic liberties is not merely a theoretical exercise in 'list-drawing'. It is also, and perhaps first and foremost, a debate about the fate of liberalism. While that fate is decided not merely by theoretical disputes but also by the many practical, and especially institutional, facets of liberal democracies (Benhabib, 2019), attaining a clear and nuanced understanding of the most important theoretical issues is crucial to deciding which liberalism, if any, to accept. We hope to have contributed to that effort.