Using terms that Isaiah Berlin made famous — but acknowledging a break with his usage — Jason Brennan and David Schmidtz begin from a distinction between negative and positive liberty. You have the negative liberty to do X just when no one opposes your doing it, whatever opposition consists in; you have the positive (or effective) liberty to do X when, in addition, you have the capacity to do X, whatever capacity is thought to require. They denounce the myth, as they see it, that these two conceptions of liberty pair off with two conceptions of the job of government, respectively right-wing and left-wing. The conceptions do not pair off in this way, so they say, because it is false — a “bad” assumption — that the job of government is to promote either form of liberty “in a direct way.”
This assumption is false, allegedly, because in order to know “how valuable negative liberty is” or “how valuable positive liberty is,” we need to do some empirical investigation: we “need to get off the armchair and go check.” The authors assume that the success of government or society is to be measured by how far people enjoy “happier or healthier or wealthier lives” — in short, “better lives.” They argue, as I read them, that it would therefore be philosophical hubris to ascribe to government the job of promoting either form of liberty directly; that ascription could only be based on the empirically vulnerable claim that the promotion of such liberty will give people better lives.
Do the authors think that the job of government, then, is to promote better lives for people directly: this, rather than promoting either form of liberty? It appears not, for at one point they give expression to a deep skepticism about government that would be inconsistent with assigning any such grand goal. “In the real world, to give government officials the power to do x is to hope that officials will use it to do x, knowing that, no matter who actually ends up holding such office, the person in question will duly pay lip service to doing x, then will use the power for purposes of his or her own.”
What goals might we assign to government, then, assuming that we do not embrace anarchism and despair of government altogether? The answer comes after a discussion of how often governments promise something — “guarantee” something — without actually delivering it, for at that point their skepticism moderates. “We want government to issue guarantees only to the extent that the guarantees work.” And how can we know which guarantees work? By recourse to empirical research. “We have to check how well legal guarantees work in our world.”
It transpires, then, that the main claim made by Brennan and Schmidtz is that in assigning a job or role to government — a set of guarantees that it should underwrite, a set of policies that it should pursue — we have to rely on empirical research. And with that claim I am wholly in agreement. On the very first page of the book on republican freedom that they are kind enough to cite, I make the point that the issue of which are the best policies and institutions inevitably depends on “empirical questions” as well as on “ideas as to how things ought ideally to be” (Pettit 1997, p. 1). It is for this reason that in a recent review article on the neo-republican approach to freedom Frank Lovett and I used the subtitle: “A Normative and Institutional Research Program” (Lovett and Pettit 2009).
But while I agree strongly with Brennan and Schmidtz that the project of policy-making and institution-building — and by extension the project of normative political argument — ought to be informed by empirical research and modeling, I cannot see why they think this undermines philosophical argument about the merits of adopting one or another conception of freedom as a primary concern of government. I cannot see why there is supposed to be no worthwhile debate remaining about the political merits of negative or positive freedom, or indeed of the freedom as non-domination that I associate with the long republican tradition of thought.
Do they think that empirical knowledge alone is going to determine what policies or programs — in their terms, what guarantees — government should underwrite? Presumably not, since empirical knowledge is certainly necessary for telling us what policies are likely to be reliably implementable or feasible, but it’s not going to tell us what policies are desirable. The desirable is a subset of the feasible and, while we need empirical research to give us information on what is feasible, we need a normative criterion of assessment in order to determine the more specific issue of which among the feasible policies are desirable.
Perhaps what they mean to suggest, however, is that as we assess the policies of a government, or prescribe for the policies it should implement, we need only form views about the desirability of now this policy, now that, without any single abstract criterion of assessment. We can say that this or that policy is desirable, they may be implying, without providing an abstract basis or criterion by reference to which it is desirable: without arguing that it is desirable insofar as it promises to do well by the cause of negative or positive liberty, or even more abstractly, promises to do well by the cause of making people’s lives better. They may be more or less particularistic about political evaluation, in order words, denying that when we say a particular policy is desirable we have to cite a salient property, or a salient set of properties, in virtue of which it is desirable.
Their claim may just be that there are a myriad features that matter in policy-making and that in recommending a particular policy we may draw, case by case, on whichever of that legion of properties seem relevant, refusing to be tied to invoking anything so general as freedom. They may think of political evaluation on a model that certainly seems to apply to aesthetic valuation, where we pick the artworks we like, and can usually manage to find some features to explain our likes, but may have no general sense of the aesthetic features that matter to us.
This pragmatic point of view will have attractions for many people but I think there is good reason to resist it, or at least to think that it does not provide an adequate view of what political assessment and argument require. When we engage in political evaluation, we do so in the course of deliberating with our fellow citizens — in actuality or imagination — about the respective merits of the institutions and policies we are recommending. But that means that we are required to find a basis of evaluation that will make sense to them generally, assuming that we are not just addressing a particular coterie of interest or opinion. And, more strongly again, it means that we are required to find a basis of evaluation that will make general sense, as a matter of general acceptance; no one will think that it is irrelevant or think that others will think it is irrelevant, and so on. To aspire to address the full membership of the society in the political recommendations presented is implicitly to claim that there are reasons that no one can dismiss which argue in support of those recommendations; people may vary in the weight assigned to those reasons, but no one can regard them as irrelevant. This is what the publicity of political assessment and argument requires.
So why shouldn’t the reasons invoked in assessments be legion and why shouldn’t they figure, now with this feature to the fore, now with that other one? The reason is simple. The more numerous the reasons invoked in public debate, the less manifest it is going to be that they are reasons that truly count as relevant across different sectors of society. That is why there is such pressure to find common starting points for public discussions in a few relatively abstract, and presumptively irresistible considerations or values. The formation of a society as a community of deliberation — a community where it is not just the numbers that matter — depends on there being values of this kind that people can put forward, however unsuccessfully, as matters that are of universal concern.
What values are going to available for deployment in this role? One value that may sometimes be available is the higher-order consideration that a policy is a good compromise, suiting this group for one set of reasons, that other group for other reasons, and so on. But such values can only be relevant under circumstances of fortuitous convergence. The values that will make a stronger claim for consideration are routine, purportedly irresistible values like welfare or prosperity or equality or justice — or, of course, freedom. The debate about how best to understand one or another value of this kind, then, will inevitably connect with the question of how important that value is in the assessment of policies and how important it ought to be as a criterion of policy-making. The segregation of issues for which Brennan and Schmidtz argue is a will-o-the-wisp.
Let me conclude with some observations on why I am attracted to the idea of freedom as non-domination; this may be the best way of explaining my point of view. I think that any value that is put forward as an important criterion in evaluating political institutions and programs ought to satisfy a number of constraints. Ideally, it ought to have historical resonance in the culture of the society; it ought to be something that no one is likely to dismiss in the abstract as irrelevant; and it ought to be a value with palatable implications — palatable, at least on reflection — for the constitution and direction of the state.
Freedom as non-domination is the sort of freedom you enjoy when you are not subject to the will of another agent or agency. You are your own man or woman; you live on your own terms — you are sui juris, in the old Latin phrase. This ideal has deep roots in our culture and, on the face of it, no one is likely to deny its political relevance. A good formulation appears in the eighteenth century tract series Cato’s Letters: “Liberty is, to live upon one’s own terms; slavery is, to live at the mere mercy of another” (Trenchard and Gordon 1971, Vol 2, 249-50).
I think that the idea of freedom in this sense is very interesting and that it has been lost to contemporary consciousness, partly because of the suggestion that negative and positive freedom exhaust the field. It is not equivalent to positive freedom, since it requires the absence of domination, as negative freedom demands the absence of interference. And it is not equivalent to negative freedom, for two reasons. You may be subject to someone’s will and yet not suffer interference, as when the person stands over you, able to interfere should they wish. And you may be interfered with without being subject to the interferer’s will, as when that agent — like Ulysses’s sailors — interferes on your terms: that is, by your own will.
If freedom as non-domination engages me, however, that is not just because of its interesting conceptual architecture. An even more important consideration is that there is promising work to be done on elaborating its political implications. What would it be for a constitution and state to promote freedom as non-domination? That is the central question in what I think of as the neo-republican research program and it is now at the center of a growing body of work.
This work, needless to say, has to be guided by empirical investigation and modeling and on that count Brennan and Schmidtz can have no objection. But would they still think the guiding question misguided, on the ground that it does not separate out analytical and policy-making issues? Would they say that I am moving illicitly from a debate “about how to use language,” as they nicely put it, to a debate about “how to use the police.” I look forward to hearing what they have to say.
Philip Pettit is the Laurance S. Rockefeller University Professor of Politics and Human Values at Princeton University.
 Brennan and Schmidtz contrast the direct with the indirect promotion of a value: say, its promotion by means of “a basic institutional framework.” I do not know what significance this may have for them (see footnote 2); it may be clear from their book, which I haven’t yet had the time to read. Since they are apparently better disposed towards the indirect promotion of a value like freedom, the divergence between us may not be so great as this response makes it seem.
 Since this point applies equally with the “direct” and “indirect” promotion of something, I do not see why that distinction is so important to them; see footnote 1.
 The most extreme form of particularism in any area of evaluation would say that there are no properties that figure always on the pro side of assessment, or always on the con side; in other words, there are no reliably good or reliably bad features (for a critique, see Jackson, Pettit and Smith 1999). But what Brennan and Schmidtz are supporting, given their absence of comment on this issue, is likely to be the less radical view.
Jackson, F., P. Pettit, et al. (1999). Ethical Particularism and Patterns. Particularism. B. Hooker and M. Little.
Lovett, F. and P. Pettit (2009). “Neo-Republicanism: A Normative and Institutional Research Program.” Annual Review of Political Science 12.
Pettit, P. (1997). Republicanism: A Theory of Freedom and Government. Oxford, Oxford University Press.