One of the most commonly discussed principles of moral and political philosophy -- if not the most commonly discussed -- is Mill's Harm Principle. Introduced in Chapter 1 of his famous polemical essay, On Liberty, the Harm Principle is set out in the following way (all quotes are from a Wordsworth classics edition - page numbers may vary across editions):
The object of this Essay is to assert one very simple principle...That principle is, that the sole end for which mankind are warranted, individually or collectively, in interfering with the liberty of action of any of their number is self-protection. That the only purpose for which power can be rightfully exercised over any member of a civilised community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant.
Thus stated, the Harm Principle seems like a robust, and in many ways attractive, statement of classical liberal or libertarian political philosophy. "Don't tread on me!", it seems to be saying. Governments cannot be too zealous or overreaching in moralistic and paternalistic interventions into individual behaviour. They must take a step back and allow us all space to experiment and grow as individuals.
Those familiar with Mill's essay, and the subsequent philosophical and legal debates it has invoked, will, however, know that the Harm Principle is anything but 'one very simple principle'. For starters, it is not clear that it is really 'one' principle. Mill asserts multiple versions of it in On Liberty and qualifies it in numerous ways. Furthermore, it is not obvious that it is 'simple' in its application. What exactly counts as 'harm'? Is 'harm' a necessary or sufficient condition for government interference?
It's not possible to do justice to all the nuances of the philosophical debate about the Harm Principle in what follows. Instead, I will take a more modest approach. I will try to highlight some complexities in Mill's formulation of the principle. I will then discuss some of the particular problems that arise in relation to the concept of harm and consider whether the inability to formulate a fully satisfying theory of harm undermines the credibility of the principle. In short, and following an argument put forward by Anna Folland, I will suggest that the Harm Principle is credible, despite the vagueness of the concept of 'harm'.
1. Mill's Formulation of the 'Very Simple Principle'
Several important questions arise from Mill's formulation of the Harm Principle. The first, and in some ways most crucial, is whether it states a necessary and/or sufficient condition for interference with individual liberty. The quoted passage above provides strong support for the idea that it states a necessary condition for interference. He talks about the 'sole end' and the 'only purpose' for which interference is warranted.
But does the principle also state a sufficient condition for interference? The initial formulation isn't clear on this point, however, Mill's subsequent discussions suggest that it does not. He analyses when, exactly, we are morally justified in interfering with harmful conduct. He thinks there are some kinds of harmful conduct (e.g. lying) that do not warrant interference. In those analyses he leans into his more general utilitarian philosophy, asking whether the benefits of interference outweigh potential costs. So, in other, words, the Harm Principle, for Mill, works as an initial filter for interventionist policies. First we ask whether we are intervening in conduct that is harmful to others. If the answer is 'yes' (and only if it is 'yes'), we ask additional questions about the costs and benefits of the proposed intervention.
This is a sensible view. If all conduct that was harmful to others justified external interference, the Harm Principle would, arguably, be overinclusive and would be anathema to many liberals and libertarians. We will explore this is more detail below when we look at issues that arise with the concept of 'harm'. But, sensible as it may be, this interpretation of the Harm Principle raises problems when considered in light of Mill's broader philosophy. Mill is not a rights theorist. He does not believe in absolute individual rights. As he himself puts it:
I forego any advantage which could be derived to my argument from the idea of abstract right, as a thing independent of utility. I regard utility as the ultimate appeal on all ethical questions; but it must be utility in the largest sense, grounded in the permanent interests of man...Those interests, I contend, authorise the subject of individual spontaneity to external control, only in respect of those actions of each, which concern the interest of other people.
What is he saying here? In essence, he is saying that the Harm Principle is justified by the principle of utility. A broad, non-interventionist stance, serves the principle of utility -- the greatest happiness of the greatest number -- in the 'largest sense'. It could be that, in particular cases, utility is served by intervening in individual behaviour even when that behaviour is not harmful to others, but it is better to adopt a general rule against intervention since, over the long haul, this is more likely to undermine the greatest happiness of the greatest number. As far as I am aware, Mill does not provide a more detailed utilitarian argument in defence of this blanket policy of non-intervention. It sounds plausible to me, and I would like to believe it, but it might be wrong.
A second question that arises is: 'To whom does the Harm Principle apply?' The obvious answer is 'the government'. The government is not justified in introducing policies that interfere with individual behaviour, unless that behaviour causes harm to others. And, certainly, most contemporary policy-related debates about the Harm Principle, focus on government intervention. But Mill clearly intends the principle to apply more generally. One of the key ideas in On Liberty is that there can be a 'tyranny of the majority' and that minority views and minority lifestyles need to be protected against interference by busybody majorities. So, for Mill, interference by our social peers and social majorities is just as much a problem as government interference. Indeed, so much so, that one of the duties of an effective government is to provide for the liberty of minorities.
A third question that arises is: what counts as a liberty-undermining interference? Mill gives some guidance on this. Right after the most-quoted passage stating the 'very simple principle', he says:
[Man] cannot be rightfully compelled to do or forbear because it will be better for him to do so, because it will make him happier or because, in the opinions of others, to do so would be wise or even right. These are good reasons for remonstrating with him, or reasoning with him or persuading him, or entreating him, but not for compelling him, or visiting him with any evil in case he do otherwise.
From this, it seems that liberty-undermining interference is limited to coercive interference. This would include threats of punishment or imprisonment for failing to do something, but also, perhaps, other threats of harm. Remonstrating with people or trying to persuade them through reasoned debate is fine, even when this concerns conduct that does not harm people. This might be interference but it is not liberty-undermining interference. This concession is interesting insofar as it permits some busy bodying interference by majorities in the form of public education campaigns. So, for example, a government education campaign intended to discourage people from smoking or consuming alcohol would, on Mill's reasoning, be acceptable so long as it doesn't amount to coercion.
A more complex edge case of interference would be that of 'nudging'. This an idea that comes from the work of Cass Sunstein and Richard Thaler. In essence, it involves the use of techniques for pushing (nudging) people to make decisions that serve their own welfare (or that of the general population) that bypass rational cognitive faculties. For example, placing healthy snacks in someone's eyeline nudges them to favour health ysnacks over unhealthy ones. This doesn't involve reasoned dialogue but it doesn't involve coercive interference either. Where this stands within Mill's framework is open to dispute and there is, of course, a very extensive and detailed debate about it in the philosophical and legal literature.
A fourth question that arises is: does the principle apply to all people, irrespective of age or status? Mill is clear on this point:
It is perhaps, hardly necessary to say that this doctrine is meant to apply only to human beings in the maturity of their faculties. We are not speaking of children, or of young persons below the age which the law may fix as that of manhood or womanhood.
This is a reasonable, mainstream position. Of course, the age fixed by the law is, always, going to be somewhat arbitrary. If the capacity to make decisions for oneself is largely a function of cognitive ability and emotional intelligence (etc) then there will, undoubtedly, be people officially designated 'children' by the law that ought to be treated as equivalent to adults, and vice versa. We may be able to overcome this problem by applying functional capacity tests -- i.e. assess whether people have the capacity to make decisions for themselves in certain domains -- but these can be difficult to employ in practice. A bright-line cutoff between youth and maturity, for all its arbitrariness, is often the simplest approach.
What is probably less well-documented is the fact that Mill thought the maturity restriction could apply to entire civilisations as well as individuals:
For the same reason, we may leave out of consideration those backward states of society in which the race itself may be considered in its nonage...Despotism is a legitimate mode of governance in dealing with barbarians, provided the end be their improvement, and the means be justified by actually effecting that end...there is nothing for them but implicit obedience to an Akbar or a Charlemagne, if they are so fortunate as to find one.
This will, no doubt, reek of racism and cultural prejudice to the modern reader. Mill is suggesting that entire peoples can be so backward and immature that they need the helping hand of a benevolent dictator. That said, I think there may be something to the point he is making. Hobbes, after all, makes a similar argument, namely, that liberty is, in some sense, a luxury for societies that have achieved a level of prosperity and security. I agree with that, to a point, and Mill himself alludes to a constant tension between security and liberty in his writings. What I don't agree with is the suggestion that despotism is the route to achieving the level of prosperity and security needed for discussions of liberty to become salient. Indeed, my reading of social history is that openness and freedom, combined with strong and stable government, are often drivers of prosperity, not luxuries to be enjoyed after prosperity is obtained.
Mill places some other, miscellaneous constraints on the Harm Principle. For instance, at one point he suggests that people might be legitimately compelled to do things that benefit other people:
There are many positive acts for the benefit of others, which [a person] may rightfully be compelled to perform; such as, to give evidence in a court of justice; to bear his share in the common defence, or in any other joint work necessary to the interest of society of which he enjoys the protection...
There is a certain common sense to these examples, and implicit in some of them is the tension between security and liberty to which I just alluded. Nevertheless, I find it hard to reconcile the general claim being made with a strict interpretation of the harm principle. Why can I be legitimately forced to do things to benefit other people if my failing to do so does not harm them? The answer to this might lie in what exactly we mean by 'harm' in the context of the harm principle. It is to this thorny topic that I now turn.
2. What is harm? Comparative Accounts
One thing Mill does not do is provide us with any general theory of harm. This might seem odd given that harm is the central concept in the Harm Principle. But perhaps it is good to avoid overly abstract thinking about harm. There are paradigmatic cases of harm to others, there are paradigmatic cases of harmless conduct, and there are borderline or difficult cases. So, for example, if my hobby is going around smashing people's fingers with a hammer, then clearly my conduct is harmful to others and I can justifiably be stopped. If I like to stick needles in my own fingers, in the privacy of my own home, then clearly my conduct is harmful to no one other than myself and I should not be stopped. If I play loud music, late into the evenings, and this upset my neighbours, then this is more borderline case. We can have an argument about whether it really counts as harm, and whether it can be justifiably stopped.
Many philosophers don't like having fuzzy or contestable concepts at the heart of our ethical theories. They seek theoretical and abstract purity. They want a general theory or account of harm that tells us whether or not specific conduct falls foul of the Harm Principle. You can understand why. A general, abstract theory of harm could be used to assess novel or controversial cases of harm and give us a clear answer as to whether it can be justifiably interfered with or not. No surprise then that many philosophers have attempted to provide a general theory of harm that we can plug into Mill's principle.
Some, however, have argued that no general and satisfying theory of harm exists. Anna Folland, in her article 'Mill's Harm Principle and the Nature of Harm', examines these critiques in some detail. The critics all follow the same strategy. They introduce some general theory of harm -- oftentimes one that has intuitive appeal or support from other debates -- and argue that it is under or over inclusive in some crucial respect. In other words, they argue that the theory would identify conduct as harmful that most people agree should not count as harmful or, vice versa, would fail to identify cases as harmful that should count as harmful.
Let's consider three examples. The first is the Temporal Comparative account of harm (TCA) (definitions taken from Folland's article):
TCA: An event e harms a subject s if, and only if, e makes s worse off (in terms of well being) after e than s was prior to e.
Assume I was healthy and well-functioning yesterday. Today, a man in the street punched me in the face and gave me a black eye. Clearly, I am worse off now, after the punch in the face, than I was yesterday, before the punch. The punch has harmed me.
Sounds sensible, right? Not so fast. Critics argue that the TCA doesn't deal appropriately with some cases. Suppose that your child has contracted a virus and has a terrible sore throat. You could, if you so decided, eliminate their pain by giving them some medicine, which you have freely to hand. On the TCA, you do not harm them by failing to give them the medicine. But, arguably, the failure to intervene in this case is a kind of harm. Similarly, imagine a case in which a child suffers from some debilitating illness because her mother smoked throughout her pregnancy and both parents have smoked, in her presence, from the moment she was born. Their actions clearly harm her, but they don't make her worse off than she previously was since the harmful actions coincided with her conception and birth.
Counterexamples like this lead people to reject the TCA and favour alternative theories. One such alternative is the 'baseline from Mankind' comparative account (MCA):
MCA: An event e harms a subject s if, and only if, e makes s worse off (in terms of well-being) than the normal well-being level of mankind.
This avoids the counterexamples to TCA, but faces some counterexamples of its own. For instance, if someone is extremely wealthy, relative to the average person, then stealing their money so as to bring them down to the average level of wealth would not count as harm, on the MCA. This is counterintuitive. Similarly, if someone is extremely attractive (admittedly more a subjective quality), then disfiguring them so as to make them more averagely attractive would not count as harm. This is also counterintuitive. This makes the MCA a non-runner.
This leads to one final theory of harm, the Counterfactual Comparative theory (CCA):
CCA: An event e harms a subject s if, and only if, s would have been better off (in terms of welfare over her lifetime) in the absence of e.
This is a very popular theory of harm. It pops up in numerous applied ethical debates. For example, I have seen people propose it as the theory that explains why death is harmful. It also avoids the pitfalls associated with the two previous theories. But it faces problems of its own.
The main one is that if harm is defined by comparison to what would have happened in some counterfactual possible world, then we have to pick an appropriate reference class of possible worlds. But, depending on how we select the reference class, conduct can be deemed harmful or non-harmful in somewhat arbitrary ways. For instance, if I, as a parent, fail to provide private music lessons to my child, am I harming them? Intuitively, most people would probably say no. But perhaps there is counterfactual possible world in which they do much better, in terms of welfare, if they receive the private music lessons. So my failure to pay for them is a harm if I compare this world with that world. In short, weird things start to happen when you start comparing this world with counterfactual ones. If you choose counterfactual worlds in which people do much better than they do in the actual world, then they harmed by virtually everything that is happening to them in the actual world. Conversely, if you choose counterfactual worlds in which people do much worse than they do in the actual world, they are not harmed.
There are some other theories of harm, but these three are a representative sample of the ones debated by philosophers and each fails to provide a fully satisfying underpinning for the Harm Principle. What do we conclude from this? Critics conclude that the Harm Principle must fail. Folland, in her analysis, is not so quick to draw this inference. As she points out, critics are adopting something like this argument:
- (1) In order for the Harm Principle to be acceptable, it must be grounded in some fully satisfying theory of harm.
- (2) The only plausible candidate theories of harm are the TCA, MCA or CCA (...etc)
- (3) Neither the TCA, MCA nor CCA (etc) provides a fully satisfying theory of harm.
- (4) Therefore, the Harm Principle is not acceptable.
But there are a number of ways to reject this argument. One way would be to reject premise (2) and argue that there are other plausible candidate theories of harm or that, perhaps, some mishmash of theories could be fully satisfying (it's not either/or). Another way would be to reject premise (1) and argue that the fate of the Harm Principle does not rest on articulating a fully satisfying theory of harm. Folland suggests this might be right response to critics. She does so on the grounds that harm features in many ethical debates and principles and yet rarely do we require those that deploy the concept to articulate a fully satisfying theory of it. It is prima facie plausible that harm is a meaningful ethical concept that delineates between acceptable and non-acceptable conduct. That we have some problems providing a fully satisfying theory of it does not undermine its use in the Harm Principle.
I think this is correct. I have long been uneasy with the argumentative standards employed by moral (and other) philosophers, which seems to suggest that we should reject normative principles if they fail to provide intuitively satisfying results across all possible worlds. I am not sure any principle could satisfy such a demand.
3. What is harm? Miscellaneous problems
There are other problems with the concept of harm. One concerns the distinction between harm to self and harm to others. This is central to Mill's principle. He thinks harm to self cannot provide a justifiable basis for coercive interference. He thinks harm to others can. Some argue, however, that this formulation is not quite right. The crucial distinction is not between conduct that is harmful to self versus harmful to others. The crucial distinction is between conduct that is consented to versus unconsented to. If I harm you, but you consent to being harmed, then it would be wrong to coercively interfere with that choice. Although Mill avoids this formulation, it is consistent with his other views, such as his claim that freedom of association is one of the fundamental forms of personal liberty:
from this liberty of each individual, follows the liberty, within the same limits, of combination among individuals.
A related, and perhaps more serious problem, concerns the causal relationship between self harm and harm to others. Sometimes conduct that is, primarily, harmful to the self is also, indirectly, harmful to others. If I want to drink myself into oblivion, you might argue that this is my right, as a free person. But, of course, my decision to do so might harm my family. It might deprive my children of care and resources they need to survive. So can we coercively interfere with the decision to drink?
Mill has a long discussion of this issue in On Liberty, focusing on the arguments of temperance activists and prohibitionists (looking to ban the sale of alcohol). He rejects an outright ban, arguing that drinking is a private pleasure within the sphere of personal liberty. But, he accepts that if drinking becomes harmful to others, coercive interference may be justified. He also argues that those occupying certain social roles -- e.g. police officers, surgeons -- may be justifiably banned from drinking, at least while on duty.
Another problem concerns the gravity of harm needed to justify coercive interference. Some people are unwilling to accept that any and all harms meet the threshold needed to justify coercive interference. Perhaps the most elaborate discussion of this issue can be found in Joel Feinberg's, multi-volume work on the moral limits of the criminal law, which is, in effect, an extended discussion and application of Mill's principle. It's impossible to capture the nuances of Feinberg's position in a short summary, but the gist of it is that he distinguishes harm from hurts and offensive conduct. A harm, for Feinberg, is something that sets back your life interests. So it is a reasonably serious kind of injury to your ongoing life plans and personal welfare. A hurt is a more trivial and temporary kind of injury, e.g. a graze upon your knee. Offensiveness is mental displeasure or distress caused by the conduct of others. For Feinberg, harms justify coercive interference, but hurts and offensiveness do not. But Feinberg ties himself up in knots about this, eventually conceding that some kinds of offensiveness, if they are sufficiently persistent and serious, might justify coercive interference, partly on the grounds that they end up being harmful if they are persistent and serious.
Others counter argue that Feinberg's attempt to distinguish between different levels of harm is unnecessary. For instance, Turner argues that the Harm Principle ought to rest on an expansive definition of harm. Virtually all harm to others, no matter how trivial, raises a prima facie case for coercive interference. Whether coercive interference is then justified depends on whether it passes the 'greatest happiness' test: i.e. do the benefits of the coercive interference outweigh its costs. Since all coercive interference is itself harmful, this is a difficult threshold to cross in the case of relatively trivial harms. Less intrusive and coercive policies will almost always be preferable.
The foregoing is a summary of some of the key challenges facing the Harm Principle. Mill's claim that it is a 'very simple principle' is both true and misleading. It is true insofar as it is easy to state the principle, and it is intuitively appealing. Nevertheless, it is misleading insofar as its practical application raises a number of complexities. Still, the mere fact that it can be contentious and that there can be difficult edge cases is not, in itself, a reason to reject the principle. One can see why it remains popular to this day.