There are two ways one can go wrong with regard to the non-aggression principle (NAP).

One way to go wrong is to treat the NAP as a rigid, out-of-context principle that can be applied fairly mechanically with little attention to other values or to the details of the situation.

The other way to go wrong is to reject the NAP, or to downgrade it to the status of a defeasible presumption or rule of thumb.

Have You Ever Heard of Plato, Aristotle, Socrates? Morons!

From the eudaimonist perspective I favour (see here and here), the content of justice stands in reciprocal determination with the content of the other virtues. That means that virtues like prudence and benevolence play a role in determining the content of justice, but also – via a process of mutual adjustment – that justice plays a role in determining the content of virtues like prudence and benevolence.

The function of virtue is to guide action; since action is unitary (i.e., one cannot act in opposite ways at the same time), virtue needs to be unitary as well. So it’s not just a matter of “good luck” that the requirements of the virtues are mutually consistent. Rather, the nature of virtue demands of us that we dialectically construct its requirements so as to be mutually consistent, starting from the raw material of prima facie virtue-judgments that may not initially be consistent at all. (As I’ve noted elsewhere, the fact that “all but the hardiest deontologists generally try to show that their favoured policies will in fact have good consequences, while all but the hardiest consequentialists generally try to show that they’re not committed to morally outrageous conclusions,” suggests that most professed deontologists and consequentialists are actually, to their credit, crypto-eudaimonists.)

Does the eudaimonist approach tell for or against the correctness of the slogan “Let justice be done though the heavens fall” (Fiat justitia ruat cælum, or Fiat justitia pereat mundus)? The answer to that question is not straightforward.

If justice were independent of beneficial/harmful consequences in the Kantian manner (setting aside some thorny issues of Kant interpretation), then the slogan would be straightforwardly correct. If instead justice were dependent on beneficial/harmful consequences but not vice versa, in the utilitarian manner, then the slogan would be straightforwardly wrong. But when the determination is reciprocal, matters are more complicated.

On the eudaimonist view, the prima facie badness of the heavens falling will need to be taken into account both in shaping the principles of justice and in determining their application to particular situations; this lessens the likelihood that justice will indeed require sacrificing the heavens. So if the slogan means that concern for the heavens falling should play no role in determining what’s just, the slogan is wrong.

But lessening the likelihood that justice will not require the heavens falling is not the same as guaranteeing the possibility’s elimination, since our assessment of the badness of the heavens falling likewise needs to be informed by the prima facie content of justice, which is heavily deontological. (In any case, a purely consequentialist approach is incoherent, since direct consequentialism has bad results (and so is self-refuting) while indirect consequentialism is conceptually unstable.) Hence it is conceivable that in some instance justice might require the heavens falling; and if it should do so, we would be obligated to let the heavens fall – and in that sense the slogan is perfectly right.

Since rights are a matter of justice, one upshot of the eudaimonist approach is that any defensible theory of rights will have to be informed by considerations of prudence and benevolence. And this means that consequence-oriented considerations will have to play a role in determining the content of rights. But inasmuch as reciprocal determination is a two-way street, rights will also play a role in determining what count as a better or worse consequence. That is why, for eudaimonists, one is better off, in self-interested terms, dying justly than living unjustly. (For why this doesn’t mean that once you’ve committed an injustice you have no more reason to live, see here.)

Here I disagree, by the way, with George Smith’s recent contention that an act can be unjust and yet not morally wrong. Justice is a proper part of virtue, not an alternative to it. As Rothbard writes, “tort or criminal law is a subset of ethics.”

Is the non-aggression principle an attractive rights thesis from a eudaimonist perspective? Well, strictly speaking, the non-aggression principle – in its standard formulations, at least – is not a rights thesis at all. But a right has two components – an obligation (on the part of others, to treat the right-holder in a certain way) and a permission (for the right-holder, or her agent, to force others to treat her that way). The non-aggression principle is simply the obligation component of the right of self-ownership, or in other words the right not to be aggressed against. And there is indeed a eudaimonist case to be made for that right. As I have argued elsewhere:

On an Aristotelian virtue-ethical account, right action is action that expresses the attitudes and dispositions appropriate to a flourishing human life, where the latter is conceived as a life that gives primacy to the exercise of distinctively human capacities. … Hence, the best life for a human being is one that navigates between the extremes of subhuman and superhuman. …

Aristotle identifies the distinctively human capacity for reason and speech as the basis of our being naturally political animals, for it enables us to pursue our goals through discussion with one another. Moreover, Aristotle famously regards logos, reason or speech, as the essential trait around which a flourishing human life must be organized. This, it seems, is why Aristotle regards it as an essential component of a truly human life to deal with others politically, i.e., through reason and discourse – i.e., as conversation partners. …

To deal with others by force is to act in a subhuman manner, like a beast of prey; we live a more human life (and therefore, in Aristotelian terms, a better life) to the extent that our relations with other people embody reason and persuasion rather than coercion. Therefore, the need to avoid the bestial type of vice gives the virtuous agent reason to accept an obligation to respect other people as ends in themselves, rather than to treat them as mere means to her own ends. If this high-level human end places a constraint on the pursuit of lower-level, animal ends, so be it. … [W]hat legitimates the [permission component] is the need to avoid the corresponding godlike type of vice, the pure pacifist position that requires the virtuous agent to cling to cooperation even when the other party abandons cooperation and resorts to aggression. The saintlike commitment to turn the other cheek accords less respect to one’s own material needs than they deserve. [The NAP] can thus be seen as striking an appropriate balance – a Golden Mean – between subhuman aggression and superhuman pacifism.

Moreover, the eudaimonist has good reason to treat the NAP as indefeasible. After all, if I aggress against you only a little, or only occasionally, I am still treating you as a legitimate object of nonconsensual use, and thus choosing, to that extent, a less over a more human life.

But given the reciprocal-determination thesis, the eudaimonist’s good reason to treat the NAP as indefeasible would not be a decisive reason to treat it as indefeasible if consequentialist considerations posed a sufficiently strong objection. Happily, the conduct favoured by consequentialist considerations is close enough to NAP that very little adjustment of our prima facie conceptions of benefit and harm needs to be done to authorise the NAP’s indefeasibility.

But the NAP is also fairly abstract, and there are different ways of rendering it more concrete. A variety of moral considerations, some consequentialist, constrain the ways in which its generality can reasonably be specified. The contours of what counts as aggression are not infinitely malleable; but absent the contribution of further values (which, as David Gordon notes, can include convention) they are not infinitely specific either. (The same applies to determining the boundaries of the person.) In some cases the principle will require not using someone else’s property without her consent; at other times (i.e., in emergency situations where the threat is great and the use temporary and minimal) it may merely requires compensating the owner after the fact for unconsented use. That the application of the principle is consequence-sensitive does not mean that consequences override the principle; nor could such emergency situations license a permanent and systematic agency of coercion such as the state.

Eudaimonism might be thought to run afoul of Bryan Caplan’s “fundamental rule of philosophical reasoning,” namely: “don’t use the obscure to argue for the obvious. It’s silly to say, ‘Murder violates man’s nature, so murder is wrong,’ when you can just say, ‘Murder is wrong.’” Perhaps eudaimionism does violate Bryan’s rule; but I disagree with Bryan’s rule. In philosophy, the goal of reasoning is (ordinarily) not just to show that something is true, but to explain why it is true; and there is nothing wrong with explaining the obvious by the obscure. It’s obvious that apples fall and that the moon doesn’t, while Newton’s laws are comparatively obscure; but that doesn’t mean that Newton’s attempt to derive the behavior of these objects from his laws was somehow wrongheaded. And as in physics, so in ethics: if we want to know not just that but why murder is wrong, we should be prepared to ground murder’s wrongness in something not quite so obvious. (This is part of why I disagree with Jason’s review of Gary’s book, but that’s another story.)

Where Matt Went Wrong; Some More of Matt’s Greatest Mistakes; and Who Is This Matt Person Anyway?

And now I come at last to Matt Zwolinski’s critique of NAP and the debate that has ensued over it.

Matt raises a number of objections to NAP; but most of them seem to be objections only to applying NAP in a simplistic and context-insensitive manner. I agree wholeheartedly that NAP should not be applied in a simplistic and context-insensitive manner; consequences and conventions play a partial (though not a total) role in determining what counts as aggression. Once this is recognised, Matt’s examples appear a lot less intractable than he seems to suppose.

Of course Matt essentially grants this, remarking that “ingenious folk“ can figure out how to “tweak, tinker, and contextualize the NAP in a way that makes some progress in dealing with the problems”; but, he insists, “there comes a point where adding another layer of epicycles to one’s theory seems no longer to be the best way to proceed,” where “what you need is not another refinement to the definition of ‘aggression’ but a radical paradigm shift.”

Maybe so; but why think this time has been reached? An old paradigm should be abandoned for a new one only if the new one does as good an explanatory job as the old one. But I can’t see that any new paradigm does as good a job as NAP at capturing the Lockean idea that people are not “made for one another’s uses.”

But if the NAP needs so much context and nuance (“epicycles,” in Matt’s phrase) to be applied correctly, why not abandon it in favour of some lower-maintenance principle? Well … like what? I seriously doubt that there’s any defensible principle that doesn’t need context and nuance in order to be applied correctly. Embracing the NAP because one mistakenly takes its application to be simple and automatic, and rejecting the NAP because its application turns out not to be simple and automatic, strike me as two forms of the same mistake. There is no principle that can do our thinking for us, so it’s a mistake to make its ability to do so a criterion for accepting a principle.

I also have to quarrel a bit with Matt’s summary of NAP as holding that “aggression against the person or property of others is always wrong, where aggression is defined narrowly in terms of the use or threat of physical violence.” As George points out, this isn’t quite accurate; libertarians generally use the term “force,” which is usually interpreted (and not only by libertarians) as being broader than “violence.” Moreover, in order to read the non-aggression principle as failing to prohibit fraud and the imposition of unreasonable risk, Matt has to interpret “physical violence” very narrowly; but in order to read the non-aggression principle as prohibiting all pollution, he has to interpret “physical violence” very broadly. I don’t think he’s entitled to have it both ways. (Incidentally, Rothbard himself deals – whether satisfactorily or not – with some of the questions that Matt raises about pollution, in “Law, Property Rights, and Air Pollution”; see for example his discussion of how one homesteads pollution easements simply by being alive. I won’t deal with Matt’s starving-children example, because Matt has already declared my reconciliation of children’s positive rights with the NAP “thoughtful and intelligent.”)

One of Matt’s objections to NAP is that “[t]he libertarian armed with the NAP has little need for the close study of history, sociology, or empirical economics.” Well, maybe or maybe not. For some libertarians a close study of history, sociology, or empirical economics may have gone into the justification of NAP. Moreover, moral justification, even if not purely deontological, has a strong deontological component (and in any case, economics is not solely empirical).

The NAP is sometimes described as the “non-aggression axiom,” and this term probably encourages critics to think that the principle is intended to be non-derivative. But that’s certainly not how Rothbard intended the phrase, since he provides it with a grounding in neo-Thomistic natural law. As I explain elsewhere:

[T]he term “axiom,” … is sometimes taken to imply that the prohibition of aggression enjoys a special epistemic status analogous to that of the law of non-contradiction, e.g., that it is self-evident, or knowable a priori, or a presupposition of all knowledge, or that it cannot be denied without self-contradiction. … [But] there is a broader sense of “axiom” in which a foundational presupposition of a given system of thought counts as an axiom within that system of thought even if it rests on some deeper justification outside that system; for example, Isaac Newton described his fundamental laws of motion as “axioms” within his deductive system of mechanics, yet regarded them as grounded empirically. In this sense non-aggression might legitimately be regarded as an “axiom” of libertarian rights theory regardless of what one takes its ultimate justification to be.

Both Matt (in his original post) and Julian Sanchez (here) seem to think that if what counts as aggression is determined by normative considerations, then a normative prohibition on aggression becomes tautological. That might be true if what counts as aggression were determined solely by normative considerations; but instead what’s driven by normative considerations is the specification of a notion whose generic outline is neither solely normative nor infinitely malleable.

Despite the title of his original post, Matt says later on that what he rejects is not the NAP itself, but the NAP’s status as “the sole and absolute criterion of libertarian justice.”

But I’ve never been able to figure out what the word “absolute” means when talking about rights. Every right is absolute if you define precisely enough what it’s a right to. Likewise, no right is absolute if you describe it vaguely enough.

And “sole” is not much better, as there’s no one right way to individuate criteria of justice. Various considerations should be taken into account in applying the non-aggression principle. Should we treat these considerations as built into the principle (in which case it will be “sole”), or as external constraints on applying the principle (in which case it will not be “sole”)? Well, you can slice it either way as far as I can see; it’s like worrying about whether an hour consists of two 30-minute intervals or three 20-minute intervals.

I do think that in order to acknowledge people’s dignity and not regard them as resources for others’ uses, we do have to treat NAP (properly interpreted) as indefeasible; so if that’s what Matt means, count me as an NAP absolutist. And NAP’s indefeasibility in turn means that, ultimately, the right not to be aggressed against is the only right we have; or, to put it another way, we may speak of ourselves as having many rights, but every one of those rights is some manifestation of the right not to be aggressed against. So in that sense I’m in the “sole” camp too.

But what that means, then, is that there is a way of regarding NAP as the “sole” and “absolute” criterion of justice, while at the same time treating its application as context-sensitive and informed by broader values. The NAP can be a plumbline and a Lesbian rule at the same time!

Though I Walk Through the Vallier of the Shadow of Death, I Will Fear No Evil

Kevin Vallier argues that we should abandon the NAP because it is “not a foundational principle of justice,” that is, it does not constitute “an ultimate or non-derivative explanation of the normative force of a suitably broad range of considered normative judgments.”

But I’m not sure why we should care whether NAP is foundational so long as it’s true. The claim that lead is heavier than balsa wood isn’t foundational either, but that hardly seems like a reason to scrap it. Moreover, I’m not sure that there are any foundational normative principles in Kevin’s sense. As an ethical constructivist, I think that all-things-considered norms are constructed out of prima facie norms. The prima facie norms can’t be foundational, since many of them will be revised or abandoned in the process of reflective equilibration, and those that survive will owe most of their authority to the equilibrative process; and the all-things-considered norms can’t be foundational either, since their normative force is derived from the process as well.

Kevin argues that libertarianism really needs four principles – self-ownership, acquisition, transfer, and rectification – and that the latter three cannot be derived from the first. Well, I would say it depends what “derived” means. I’ve argued that self-ownership (or NAP) places stringent constraints on what principles of acquisition and transfer (see here, here, here, and here) and principles of rectification (see here) we can coherently defend. Surely using self-ownership to narrow the range of options counts at least as deriving from self-ownership the disjunction of the remaining options, even if further narrowing will require additional norms and/or convention. (Thus the NAP is not “parasitic ” on a theory of property rights, since it has to be used in specifying property rights.)

Incidentally, several would-be defenders of Rothbard (I’ve been reading so many contributions to the debate that I’ve forgotten who) have rushed to rescue him from the charge of deriving all of libertarianism from self-ownership, urging that for Rothbard at least a homesteading principle is needed in addition. But I’m not sure Rothbard would welcome the defense. He does say that “[f]rom these twin axioms – self-ownership and ‘homesteading’ – stem the justification for the entire system of property rights titles in a free-market society” but he immediately adds in a footnote that strictly speaking, “rather than being two independent axioms, the homesteading principle really follows from the single axiom of self-ownership.” Broadly speaking, I think he was right.

He Sank Beneath Your Wisdom Like a Stone

As a free bonus, let me close by saying something about eudaimonism and drowning children.

Matt thinks we have a duty to rescue drowning children. Bryan thinks we don’t. For Bryan, such assistance is morally praiseworthy but supererogatory.

As a eudaimonist, I reject the category of the supererogatory. The right action (or, usually, the disjunction of equally acceptable right actions) lies in the Aristotelean mean between too much and too little, both of which are wrong. The notion of supererogation gets its plausibility from ethical systems that accept the possibility of conflicts between morality and self-interest, or between one person’s self-interest and another’s; if the demands of morality can require self-sacrifice, then it makes sense to suppose there might be a level of moral demand that asks too much, that requires an unreasonable amount of self-sacrifice. But if such conflicts are impossible, then the demands of morality can never be too high, can never require any genuine sacrifice, and so there is no need to place them beyond the call of duty.

It remains true, of course, that fulfilling one’s moral obligations will be easier in some circumstances and more difficult in others; and such difficulty will often make moral failures more excusable. But that does not mean that difficult obligations are anything less than morally required.

That doesn’t mean, of course, that every instance of helping others is either obligatory or forbidden. As noted above, in most cases there are a variety of equally acceptable options that fall within the virtuous mean, and in any given case some may involve offering aid while others do not. So a person who offers aid may be virtuously doing something she was not obligated to do; but that does not mean she was doing something more praiseworthy than her other permissible options. (She may well be doing something more worthy of gratitude than her other permissible options, but that’s not quite the same thing.)

I think we do have a duty to help others. For one thing, it would be an offense against libertarian thickness if other people mattered so much that all aggression against them was forbidden, yet so little that indifference to others’ fate could be permissible. But I also think the duty to help others is an imperfect duty; that is, it’s a duty to help some others some of the time, but not everyone you could conceivably help at every opportunity. This is because the content of benevolence is partly informed by the content of other virtues, including prudence.

Now ordinarily you get to pick and choose the occasions for satisfying your imperfect duty; but if circumstances confront you with a case where helping is easy, the need is severe, and you are the person best placed to help (in other words, the drowning child case, as ordinarily presented), then helping in that particular case may become obligatory. Given the demands of the other virtues, however, such an obligation cannot generalise to Bryan’s case of “Drowning Children as far as the eye can see, 24/7.” Sorry, Peter Singer.

Needless to say, the fact that saving the drowning child is sometimes morally obligatory does not mean that it should ever be legally compulsory. “Good Samaritan” laws treat the potential rescuer as a mere means to her potential beneficiaries, thereby insulting the dignity of both of them.