As might be expected of anything written by Sen, this is a book of immense depth, detail and range. The title itself is immediately suggestive: although it clearly echoes Rawls, the ‘idea’ indicates that it is not intended as a similar work of grand theory; it also echoes Collingwood (although imperfectly because it is not an account of the development of the concept). The tentative nature of the title indicates Sen's starting point – arguing from our understanding of instances of injustice and securing agreement on them, rather than developing an overarching theory and reasoning forward from that. To that end, a title such as The Idea of Injustice and its Remedies might have captured what he intended with more precision. This comes out in his own characterization of his enterprise:

As this is Sen's starting point, let us begin by considering it. In many ways Sen wishes to distance himself from Rawls and he does so partly by characterizing Rawls's approach as ‘transcendental institutionalism’ and his own view as a comparative approach. He states firmly that ‘a transcendental identification is thus neither necessary nor sufficient for arriving at comparative judgements of justice’. This means that we can simply live without what Rawls was offering in A Theory of Justice. In making his arguments, Sen captures something important about the way in which Rawls combines a focus on institutions with a certain sort of ideal theory which assumes that one starts from the ideal and works downwards, as it were. Sen is strong in his rejection of a top-down Platonism, in arguing that primary importance should be attached to the consideration of behaviour rather than institutions, and that one should start from remediable and observable acts of injustice rather than from the generation of a theory from which one then derives a view of particular acts of injustice. This is a point often made but still well worth making. In sum: we don't need a theory to spot injustice; it is a lot easier to spot injustice than to say what justice is; and the point of identifying injustice is to do something about it.

A critic might, however, plausibly deny that Sen is really any less theoretical than Rawls and argue that he is largely playing the same or a similar game. In other words, Sen's departure from the orbit of Rawls is far less radical and clear cut than Raymond Geuss's in Philosophy and Real Politics (2008). Geuss argues strongly against the Kantian stream in political philosophy in which politics is understood primarily as applied ethics. What Geuss finds objectionable in the claim that ‘Politics is applied ethics’ is where it is taken to mean that ‘we start thinking about the human social world by trying to get what is sometimes called an “ideal theory” of ethics’ (p. 6). Sen appears to be on Geuss's side in this respect, but in fact he remains caught somewhere between Rawls and Geuss. From a wider vantage point he and Rawls share rather more than they do not and many of the differences between them are differences of less consequence than Sen claims.

The Idea of Justice is ambitious in scope and it is not possible here to do more than summarize a few key points and to articulate the beginnings of a critique. The four parts of the book comprise considerations of the demands of justice, forms of reasoning about the materials of justice and public reasoning and democracy. Anyone familiar with Sen's previous work will not find many surprises here – in fact the book as a whole in many ways is a useful summary and drawing together of much of his work. The first part on the demands of justice focuses on reason, objectivity and impartiality and raises many interesting questions in relation to Rawlsian and other approaches. The second part focuses on how we might reason about justice in practice, and includes a discussion of the plurality of reasons which is of the utmost importance. The third part (dealing with what Sen terms ‘the materials of justice’) is perhaps most familiar as it deals with the capabilities approach: I certainly do not intend to criticize this here, as it is a helpful and persuasive approach to major issues in justice, rights and development quite independently of the other strengths or weaknesses of this book. By concentrating on what people are able to do rather than what they possess, or on the definition of needs or wants in an abstract sense, it illuminatingly cuts through many of the debates surrounding the subject. The final part of the book addresses public reasoning and democracy and is an interesting examination of the intertwining of democracy and public reasoning and their relation to justice, the distribution of goods (which includes a discussion of Sen's well-known position on famine) and human rights.

One of the key parts of Sen's argument is that rather than working from a single principle or criterion, we should consider the possibility of multiple grounds for reasoning and decision-making: ‘what is important to note here, as central to the idea of justice, is that we can have a strong sense of injustice on many different grounds, and yet not agree on one particular ground as being the dominant reason for the diagnosis of injustice’ (p. 2). This, of course, has to be distinguished from a sort of evidence-giving which amounts to no more than the accumulation of plausible but unrelated circumstantial evidence adding up to nothing. And this raises the issue of whether there has to be some sort of criterion which can be invoked to say that this, but not that, set of multiple grounds is reasonable or adequate. But, given his approach, Sen has impaled himself on the horns of a small dilemma here. Can he provide a criterion for coherence and for taking ‘plural grounding’ seriously? After all, merely noting that we often proceed this way in practice might be true, but it does not amount to epistemological justification. Or do we have to sense it on each occasion? Does it become a sort of moral intuitionism or particularism? One might argue the case for this, but at the same time it is open to rather obvious objections. And if on the other hand a criterion is to be given, whence do we derive it? Does it come from a theory, and if so, can and should Sen give us that theory? But he seems to have committed himself to denying that any such theory is possible.

Perhaps we can only answer this by digging a little deeper and asking what is Sen's alternative to the framework he rejects. It lies in what he terms a comparative approach. This approach has to be chosen because

One of the examples used in the book is that of three children and the just distribution of their access to a single flute: should it be on the basis of skill, poverty or reward for labour invested in making the flute? The example is used to good effect to start casting doubt on the ability of any overarching ideal theory to come up with answers to practical questions of justice. The issue, however, is this. Suppose we reject the view that a single overarching single principle theory can either be generated or ever be agreed upon in practice, where does this leave us? It appears that we are left with a form of ethical pluralism – if that is indeed what is implied by Sen's notion of multiple determining reasons. But here we have at least two problems. One is that although it is easy to reject the notion of an agreed-upon overarching Rawlsian theory, not only is it not easy to agree on an alternative, but the view that there is an alternative which might be easier to agree on is itself flawed, because the position it expresses is itself a contentious one. And, further, even if we suppose we can agree that in the absence of an agreed overarching theory we can accept that we have to rest content with a form of pluralism (there are good reasons for this conclusion because, as Sen notes, there is a logical asymmetry in that it is much easier to secure agreement on the identification of injustice than it is to create a coherent theory of justice tout court), we still have to account for the plural elements within the plurality itself. In other words, if we have multiple determining grounds, is there is a principle for inclusion in or exclusion from membership of this class of reasons? Sen does not address this point, which it is necessary to address even if one accepts his overall characterization of the way in which reasoning about justice should run.

In seeking to distance himself from Rawls and from what he terms the transcendental institutionalist view, Sen argues that ‘the distance between . . . transcendental institutionalism, and realization-focused comparison . . . is quite momentous’, and goes on to argue that:

The problem with the first approach, he contends, lies with, on the one hand, the fact that:

and, on the other, that:

So we have what might be called an impossibility and a redundancy thesis marching hand in hand to show that the Rawlsian approach – whatever its other merits and strengths (Sen admits many, and clearly admires both the work and the author immensely) – is unworkable.

The key point for Sen is that there is an overriding need to ‘focus on actual realizations and accomplishments, rather than only on the establishment of what are identified as the right institutions and rules’; this is a point he relates to the dichotomy between an ‘arrangement-focused view of justice’, and a ‘realization-focused understanding of justice’. The former proposes that justice should be conceptualized in terms of organizational arrangements, that is, ‘some institutions, some regulations, some behavioural rules – the active presence of which would indicate that justice is being done’ (p. 7). This is a vital point: Sen is primarily – perhaps solely – concerned with the necessity not to restrict the analysis of justice to ‘getting the basic institutions and general rules right’ and with the need to examine what emerges in a society, including the kind of lives that people can lead, given both the institutions and rules, and also other influences, including behaviour, that would necessarily affect human lives.

In making his point about the nature of theory, Sen repudiates a certain view which I earlier characterized as Platonism. He illustrates this with the example of the Mona Lisa, and states that ‘if we are trying to choose between a Picasso and a Dali, it is of no help to invoke a diagnosis . . . that the ideal picture in the world is the Mona Lisa’, because this is irrelevant to the choice between a Dali and a Picasso. Indeed, he suggests, ‘it is not at all necessary to talk about what may be the greatest or most perfect picture in the world, to choose between the two alternatives that we are facing’ (p. 16). The point is clear, although made in possibly a question-begging manner: to be precise, it assumes the aptness of the analogy. One can accept that the ideal of the Mona Lisa does not help us judge the relative merits of other paintings, but what really needs to be established is that the analogy holds for the goods of which Sen wishes to speak, and this needs to be demonstrated, not assumed.

That said, however, Sen is on strong ground in arguing that ‘transcendental theory simply addresses a different question from those of comparative assessment . . . [and] is of no direct relevance to the problem of choice that has to be faced’ (p. 17). This is because what is required is agreement, based on public reasoning, on rankings of realizable alternatives. He sets this in the context of a distinction drawn from Indian legal thought between niti and nyaya, where the former refers to organizational propriety and behavioural correctness, while the latter ‘stands for a comprehensive concept of realized justice’. For nyaya, institutions, rules and organization, although they are important, have to be assessed in a broader context, ‘which is inescapably linked with the world that actually emerges, not just the institutions or rules we happen to have’ (p. 20).

This focus on ‘realized justice’ is essential to Sen's approach, and to his rejection of any kind of a priori requirement for objectivity and the principles of justice. Sen agrees with Habermas and Rawls in so far as they link objectivity in ethical and political convictions to the ability to survive challenges from informed scrutiny from diverse quarters; however, he differs sharply in strongly asserting that ‘the principles that survive such scrutiny need not be a unique set’ (p. 45).

In conclusion, I would like to state a few criticisms of Sen's approach and questions that might be asked of it. The first is that, to follow on from a point implicit in Geuss's approach to politics and seemingly absent from Sen's, he might be charged with the accusation that he does not sufficiently consider the concept and exercise of power. While this criticism might not touch a work of transcendental institutionalism, it is certainly relevant to a work claiming practical political relevance such as The Idea of Justice.

Another criticism, which I advance very tentatively, is that Sen's approach is a comparative approach and this is why he relies on social choice theory. He avoids, however, its cousin, rational choice theory, because for Sen that sets too tight a constraint on what counts as rationality – both in respect of its behavioural assumptions and also in its assertion that only means and not ends can be subject to rational scrutiny. The difficulty is that it is not obvious that the social choice approach neatly fits the plural grounding approach that Sen advocates. The difficulty is that where social choice is about comparative rankings, we might argue that this only sits comfortably with preference orderings differing in degree and not in kind. However, in situations of real choice – the sort of situation that Sen would have us consider – differences in degree typically are found intertwined with differences in kind. In such instances social choice ranking might be accused of distorting the nature of the choices through a form of procrustean reductionism. Admittedly, it might be a better approach than ideal theory or rational choice theory (for different reasons) but it might not be supple or flexible enough to act as an appropriate bearer of our detailed moral and political reasoning in the areas with which we are concerned.

Another issue is this: the book is undeniably a fascinating read, but is the whole more than the sum of its often excellent parts? This might be unfair to Sen, but it does raise the issue of how much in a book of this sort one is justified in relying either implicitly or explicitly of the assumed validity of arguments already developed and deployed elsewhere. Is the book intended to stand alone or not? If it is, it seems to be wanting in a number of respects; if it is not, then this should be made clearer. This is especially important as the book is written in a style which eschews the technicalities of some of Sen's other works and presents itself as a more accessible guide to the topic. But if this is so, it raises some difficult questions about its relation to other works. However, in conclusion, if the question is whether the book is worth reading by anyone interested in ethics, politics, economics, welfare, capabilities, justice and development, the answer is an unequivocal yes. My comments and criticisms arise from engagement with an engaging work and I recommend others to engage with it too.