The Non-​Aggression Principle Can’t Be Salvaged—and Isn’t Even a Principle

First, the viability of a principle has to be judged in the context of what role it is meant to be playing in your ethical thinking. If the NAP is understood as akin to “tell the truth”—one among many rules of thumb that establish rebuttable presumptions for guiding your daily conduct—then it is perfectly servicable. May I respond to an insulting remark by throwing a punch? I may not. But that is not really how libertarians usually want to use the NAP: Rather, it is meant to serve as a kind of master principle from which other more concrete rules may be derived, and against which competing forms of social order may be evaluated. And for that purpose it is not particularly useful. When we are not looking for rough guides to our day-​to-​day conduct, but engaged in the rather more leisurely activity of theory-​building, there is less practical reason to rely on simplified models because “the math is easier,” and more to the point, we are most likely to be wrestling with precisely the kind of hard cases that arguably constitute exceptions to the rough rule, or deciding between variant forms of the very concepts on which the NAP relies.

If the NAP were truly a universal, exceptionless master principle—if the apparent counterexamples were either bullets we could bite or cases where the principle would yield the right answer after all once properly understood—then we could employ it directly even at the highest theoretical level. But if, as Jason seems to concede, we are allowing that the exceptions are really exceptions, then we are implicitly relying on some other higher-​level principles—respect for autonomy or hedonic utility or Kantian universalization or contractualist agreement—to limn the boundaries. But at the theory building level, the appropriate thing to do is to refer directly to those deeper principles. To extend the initial metaphor, it is precisely when you are trying to do fundamental physics that you end up needing to step back from the approximate truths of classical mechanics. The NAP is no help deciding the questions you’re attempting to answer at this level, because as Zwolinski notes, it’s parasitic on theories of property and coercion that reside at this same level of abstraction. You can’t resolve a philosophical debate between a classical liberal and a socialist by appealing to the NAP, because each can claim their view is consistent with that principle given their theories of property: The state is not “aggressing” on an individual “property owner” if in fact The People ultimately own (or have some kind of share right in) all property, given the normatively loaded way “aggression” is used here. The appeal of the NAP lies in its apparent simplicity and intuitive plausibility (tautologies tend to be intuitively plausible), but it’s typically deployed in a way that amounts to a kind of shell game: I argue that socialism must be rejected on the grounds that it violates this one simple moral principle, and hope my interlocutor doesn’t notice that I’ve essentially begged the question by baking a theory of strong property rights incompatible with socialism into my conception of “aggression,” when of course libertarian property rights are ultimately backed by the threat of (individual or state) violence as well.

This gets us to the deeper problem with the Non-​Aggression Principle: It is not really a principle at all, but a tautology gesturing in the direction of a theory. Suppose I tell you that I have, at long last, discovered the simple principle from which all morality derives. Assuming you suppress your initial skepticism about this bold claim, you are likely to be disappointed if I then reveal that my principle is something like “One ought not to commit wrong actions,” or “Never violate the rights of others.” The latter is not completely content free—it specifies that it is the moral claims of individuals that matter, rather than compliance with some natural or divine law unmoored from human consequences—but neither principle really gives us the concrete guidance we had expected, even at a highly abstract level: These are formaltruths given the meaning of the normative terms “right,” “wrong,” and “ought,” not substantive guides to which actions or social orders satisfy these conditions. This is slightly less obvious in the case of the NAP because the partly-​descriptive term “aggression” seems to add a substantive component. Yet this amounts to little more than the recognition that we are creatures made of matter whose moral entitlements are ultimately guaranteed or made effective by the application or (more often) implicit background threat of physical force in the last resort. A “right” is a claim that you are justified in ultimately resorting to (direct or indirect) force to make effective; whether or not a particular act counts as “aggression” or “initiation” of force depends on whether the right enforced is a genuine one. But all the real action is in the definition of rights; invoking the NAP adds nothing. It is tantamount to saying “only enforce rights that are really rights.” To establish your right over (say) your car just is to establish that I ought not to take or use it without your permission (perhaps barring extraordinary circumstances, the parameters of which will tend to be implicit in the argument establishing the right). It is neither necessary nor illuminating to add the additional premises that taking what you have a right to counts as “aggression,” and that one ought not to aggress.

Now, it is surely true, as Jason suggests, that any simple principle is going to be highly incomplete and tied to a larger theoretical apparatus. The utilitarian principle of “the greatest good for the greatest number” does not tell us whether to maximize aggregate or average utility, or whether to concieve of utility in purely hedonic and subjective terms, as a measure of preference satisfaction, or something else. The Rawlsian Difference Principle, which stipulates that social inequalities are justified only to the extent they benefit the worst off group in society, does not in itself tell us how to understand “worst off”: For Rawls, it is determined by a laundry list of “primary goods,” but one could just as easily apply a variant difference principle that used hedonic utility instead. The Millian Harm Principle—individuals may be restrained only to prevent harms to others, not to themselves—notoriously runs into a thicket when we try to specify precisely what counts as a “harm.” It is no objection to a principle that it depends on theory. But it is an objection if the principle is being made to do load-​bearing work—treated as a fundamental “axiom” rather than a presumptive rule derived from higher-​order principles—that conceals those dependencies. The utilitarian principle—though “gappy” and therefore susceptible to a wide array of variant interpretations—really is “axiomatic” in some sense: The basic idea that one ought to weigh the interests of all equally—and then act in the way that (or follow the rule that) generates the greatest net of benefits minus burdens—really is fundamental in utilitarian theory. It is not quite that there is nothing further to say about why it might be plausible or not, but we are pretty close to the core intuition at the heart of any moral theory that one ultimately either shares or doesn’t. The Harms Principle and Difference Principle, by contrast, are transparently derivative rules, the upshot or result of a lengthy preceding argument in light of which they are to be interpreted, fleshed out, and applied to particular circumstances.

The NAP as deployed in much libertarian rhetoric tends to trade on an ambiguity between these two types: It often masquerades as the former, even though it only really has content if understood as the latter. We then often end up being misled into thinking we can defend a particular form of sociopolitical order just by appealing to the simple and intuitively plausible NAP, without noticing that the highly intuitive formal NAP only gets substantive content by choosing a specific cluster of contestable conceptions of “aggression,” “property,” “coercion,” “consent,” and so on. Yet the choice between these conceptions ultimately just is the choice between competing sociopolitical orders—so arguments of this form are circular.

None of this is to deny that the NAP may be a perfectly servicable way to sum up what libertarianism is all about to the fellow on the next stool at the bar who’s never heard of it. But if we are doing political philosophy, we are probably better off leaving it behind: It is a principle that does no independent work, but tends to muddy the waters by appearing to. If we are interested in building sound theories, rather than winning quick debates against inattentive opponents, we should hope to do more than that.