In their article, Kirill Danishevskiy and Martin McKee make two main claims: that (1) tobacco is illegal under Russian law and (2) that the correct policy response is to license addicts as ‘part of a strategy for the gradual but complete ban of tobacco sales over several decades’. This commentary will be primarily about the ethics of their proposal. However, I wish first to make some comments on their law and policy claims.

I have no expertise in Russian law or practices of legal interpretation. With that caveat, I confess I do not see that the authors have shown that tobacco is illegal. On the basis of what the authors write, it seems that the most they can claim is that Russian laws internally conflict in what they say about tobacco. The authors concede that Russian regulations of tobacco imply that it is legal. Their arguments for its illegality hinge upon interpretations of the right to health in the Russian constitution and Russian product safety law. However, constitutional rights usually require interpreting, and a right to health, in particular, will not be obvious in its meaning or scope. Resource constraints would occur in any society (except perhaps a society of superabundance, such as the one Marx envisaged in the higher stage of communism). It would thus not be practical to say that each citizen had a constitutional right to whatever would promote his or her health to its maximum. As the right must be interpreted, it will be open whether it is interpreted so as to exclude tobacco. That point undercuts the inference to the illegality of tobacco from product safety law. The authors say that tobacco must be covered because the law is exceptionless, and the law is exceptionless because it derives from the constitutional right to health. That right is, as I said, inevitably vague and may reasonably be interpreted to exclude tobacco. There is an obvious difference between, say dangerous toys and tobacco in that users know the risks of tobacco, which are inseparable from its use.

Consider now the authors’ suggestion to licence addicts. As a piece of policy analysis, theirs is the barest promissory note. They can have no idea whether the effects of such a policy would be beneficial overall. As they themselves point out, tobacco companies are pastmasters at evading regulations; perhaps they would evade the licensing policy too. They concede that corruption is a problem in Russia; a proposal to restrict sale to addicts would give rent-seeking state officials ample opportunity to take advantage of the vulnerable. They compare their policy with the criminalization of narcotics; but the war on drugs has hardly been a shining success and many hold it responsible for narco-states, mass imprisonment and a misallocation of resources to law enforcement and away from treatment of addiction (The Economist, 2009).

However, as I said, this will be mainly a commentary on ethics, and particularly questions about whether people ought to be allowed to smoke free from the state interference the authors recommend. Let us take the title question: ‘why is tobacco treated differently from other hazardous substances?’ Like hazardous substances, tobacco kills and causes disease. But in raising risks of death and disease, tobacco is also like base-jumping, rock climbing, hang-gliding, ballooning and so on. Should these risky activities also be banned? Do people violate their own alleged constitutional right to health when they knowingly run these sorts of risk? I assume the authors will agree that the right to health must be alienable, in the sense that people must be free to run some extra risks in their own lives. (The authors make nothing of second hand smoke and, like them, I shall leave aside third party effects, which may or may not make a good case for tobacco regulation.)

Unlike with some other hazardous substances, people choose to smoke. People are generally not forced or tricked into smoking by third parties. They know they are doing something hazardous. (If they do not, they should be informed.) However, the authors assume that smokers are addicted and their choices are therefore not autonomous. This leads us to a difficult question about the consumption of addictive substances: are people's choices voluntary enough that they should be left to make them unmolested by the state?

In the liberal tradition, people should be free from coercive interference if they are not harming others (Mill, Reference Mill1985; Feinberg, Reference Feinberg1986). The authors do seem to see the force of the idea, although I think they fail to see its scope. Towards the end of their article, they say of their licensing proposal that it ‘does not seek to make smoking illegal. There is no reason to do so if only the smoker is harming his or her self…’. I agree that people should not have their behaviour criminalized when they would harm only themselves, at least when their behaviour is sufficiently voluntary. However, the point goes beyond what people do to themselves and includes what they consent to others doing too. This point is of fundamental importance in liberal political philosophy and in past political battles. For instance, homosexual acts or interracial cohabitation should not be criminalized when between consenting adults. Our rights to run our own lives include the right to make links with other people who have the right to run their lives. When third parties are not harmed, people should be free to make these links as they wish. Consequently, if people should be allowed to smoke when they do not harm others because people generally should be allowed to do as they wish when they do not harm others, other people should be allowed to supply tobacco without being prevented by the state.

At this point, we must confront the addiction question over smokers’ choices. The authors are very brief here. They say that tobacco leads to addiction and that ‘addicted smokers are not acting on the basis of free will’. They say that many smokers want to quit. They claim that transactions are not enforceable at Russian law unless based on autonomy on both sides, and clearly mean that transactions between suppliers and consumers of tobacco are not autonomous on the side of the consumer. All these claims are far less straightforward than the authors let on.

Take first the claim about law. We may distinguish choices that are voluntary from choices that are autonomous. As a first approximation, voluntary choices are ones based on the will of the person. They are choices that are free from coercion, in the sense of irresistible threats, free from manipulation, paradigmatically from dishonesty, and made by someone who is not suffering from gross irrationality, such as that which would be caused by drunkenness. Autonomous decisions are ones that are not only voluntary but fulfil further criteria, such as being based on reflection or coherence with one's values. Does Russian law really require choices to be autonomous in this fuller sense? I doubt it. At least in Anglo-American law, voluntariness is sufficient. The sorts of misrepresentation, absence of disclosure, threats or failures of will that would vitiate a contract or excuse criminal behaviour are much more limited and tightly specified than the sorts of defect that would make a choice less than fully autonomous (Feinberg, Reference Feinberg1984, Reference Feinberg1986). The law has good reasons for this. One is that third parties cannot reasonably be expected to beware of people who are not living fully up to their own values, although they can be expected not to misrepresent or threaten and to beware of striking deals with people who are drunk. Another reason to insist on voluntariness rather than autonomy is to do with the claims of the individual. They have rights not to be interfered with just because third parties believe they know better what the individual's autonomy requires.

Are addictive decisions voluntary enough that they should not be interfered with? The relationship of addiction to voluntariness is difficult. Suffice to say that some experts believe that addiction is ‘a disorder of choice’ (Yaffe, Reference Yaffe2001; Heyman, Reference Heyman2009). Even heroin addicts tend to give up heroin after a few years. Heroin addicts with a lot to lose, such as a career as a pilot or physician, almost invariably give up. In the case of smoking, very large numbers of smokers gave up on finding out that smoking was harmful. Addiction, unlike mental diseases such as Alzheimer's, is affected by incentives. This is not to say that addiction to hazardous substances is good for people. However, we cannot assume that people must be acting against their free will when they harm themselves.

Smoking may be voluntary and yet not be autonomous. Smokers want to smoke. However, as the authors point out, many of them also want to quit. So some smokers have a conflict of desires. We can see how these smokers are not acting fully autonomously. On one view, doing what one wants is not enough for autonomy. The desires on which someone acts must also be ones that he or she reflectively endorses (Frankfurt, Reference Frankfurt1988; Dworkin, Reference Dworkin1988). If one is acting on desires that one reflectively rejects, then one is not acting fully autonomously. Note, however, that addicts who do endorse their desires (‘happy addicts’) are fully autonomous as far as this view goes. On this view, one could be addicted and autonomous, even if most addicts are not.

Let us suppose that we have a sound view of full autonomy; that addicts who wish to quit but find it hard or impossible are not fully autonomous; and that many smokers do indeed wish to quit and cannot. It follows that many smokers are not fully autonomous. However, what follows from that is by no means straightforward. In the first place, full autonomy at best picks out a character ideal. It is not the basis for drawing a line between permissible and impermissible state action. As I said earlier, the law has a good reason not to interfere with voluntary behaviour that happens to fall short of autonomy. A good deal of running our own lives involves going to hell in our own way. It would be intolerable if even a benevolent state were able to interfere with all the occasions on which we fail to live up to our own exalted values. Even if I would like to be a cultured ascetic, should the state step in to prevent me from taking a second helping of pudding or watching a soap opera instead of real opera?

The real importance in the argument of the desire to quit is that the state could help people do what they themselves want to do. I accept that restrictions on freedom can promote autonomy. People sometimes play tricks on themselves to overcome the likely defects in their decision-making. They may put their cigarettes somewhere where it is hard to get them, move their televisions away from their food, or put reminder notes around the house (Schelling, Reference Schelling1984). There are limits to what people can do for themselves, however. If they set the alarm clock for their early morning run, they can always turn it off when it rings. So people may welcome third party assistance. In the case of tobacco, some smokers would welcome efforts to help them quit by removing the temptations of seeing smokers or tobacco advertising in public places. Such efforts could plausibly be considered to promote their autonomy.

Two caveats are in order, however. One is that while some smokers may welcome having their freedom restricted, others would oppose it. The ones who oppose it would not have their autonomy promoted by the restrictions. I think the permissibility of restricting freedom would then turn in part on how many smokers did want to quit as against those who do not. In trying to find out the numbers, care should be taken to find out how many smokers want to quit at that time, rather than inflating the number by asking how many would want to quit at that time or in the future. The second caveat is that restrictions promote autonomy only when the subject of the restrictions welcomes them. Recall the man who prefers real opera to soap opera. If soap opera were banned, this man would do what he prefers, which is to watch real opera. Yet if this man opposes the banning of soap opera, the restriction would not promote his autonomy. In other words, we cannot count as autonomy-promoting just any means, no matter how draconian, that would give people what they say they want. If people do not welcome the draconian means, those means would conflict with their autonomy.

So now let us ask the question of the authors’ proposal: how many smokers, as a proportion of the total, would want to be prevented from getting cigarettes unless they were state-licensed addicts? If the answer is ‘not many’ then the proposal would be a wrongful infringement of people's freedom. But I do not know the answer and I do not suppose the authors do either. Maybe finding out is the next step.