John Broome's Weighing Lives provides a much-needed framework for the intriguing problems of population ethics. It is also an impressive attempt to find a workable solution to these problems. I am not sure that Broome has found the right solution, but I think he has done the ethics profession a tremendous service in tidying up the discussion. The framework he presents will make it possible for the participants in this debate to formulate their positions in a clear and precise manner. Even people who disagree with him will be helped by this framework, since they will now be able to show exactly where their views differ.

A crucial feature of Broome's framework is that he sees the fundamental problem of population ethics as an aggregation problem. The problem is to determine the value of a population (or, a distribution of well-being, as Broome prefers to call it). Each moment in a person's life is assigned a numerical value that represents how good that moment is for that person (momentary well-being). The problem is then to aggregate the values of the moments in all lives in a population into a general value of the whole population. More exactly, the problem is to find the right shape of the (continuous) value function that takes us from values of moments to values of whole populations.

For this framework to make sense we need to make some important assumption, the most important of which are:

1. (a) only well-being is relevant to the overall value of a population (welfarism);

2. (b) momentary well-being can be represented quantitatively;

3. (c) we can compare well-being both across times within the same life and across different lives;

4. (d) there is a distinction between personal value (individual well-being) and general value of populations (roughly, personal value is a value for individuals, general value is impersonal ethical value);

5. (e) betterness (both personal and general) is transitive (if A is better than B, B is better than C, then A is better than C);

6. (f) value (both personal and general) is fully comparable in the sense that for any two states of affairs either one is better than the other or they are equally good;

7. (g) general value is not relative (to times, populations, or anything else).

A main bulk of the book provides thorough and ingenious arguments for these assumptions, which, sadly, are often just taken for granted in the debate about population ethics. The remaining chapters deal with the exact shape of the general value function. One important claim Broome makes is that, whereas we can value the well-being of a person independently of the well-being of other people (an assumption Broome dubs ‘separability of lives’), we cannot value a person's well-being at a time independently of her well-being at other times (an assumption dubbed ‘non-separability of times’). Broome's main argument for the non-separability of times has to do with the value of longevity. He thinks that we value longevity in the sense that given a fixed total of time and level of temporal well-being, it is better for the time to be divided up amongst fewer lives rather than amongst more lives (p. 108). Now, if we valued times separately, we could not value longevity, since the length of a person's life is not something that shows up when we look at times separately; it only appears when we look at several times together.

Broome's final theory can be characterized as follows. The value of a life is calculated by taking the difference between the sum of momentary well-being in the life and a critical level v (p. 255). The general value of a population in a world is the sum of all these life values. The critical level v is supposed to be the level at which it is ethically neutral whether a life is created or not. That is, from an impersonal point of view, it is equally as good to create a life at this level as not to create the life. Broome calls it therefore ‘the neutral level for existence’. Broome explicitly denies that this level is the level of a life that is neutral for a person, an option that the classical total utilitarian would go for. Instead, he sets it at a positive value. This level is above the level of a life that is constantly neutral for people. A life is constantly neutral for a person just in case, at each time t in this life, it is equally good for the person that the kind of life he is leading at t is continued rather than not.

Broome is careful not to present this theory as the definitive answer to all problems. Rather, he presents it as a default theory worth taking seriously. He is also admirably honest about the problems with his theory. As Broome himself admits, his theory faces a serious dilemma. It seems bound to have either of two counterintuitive implications: the famous Repugnant Conclusion, or the less-well-known Sadistic Conclusion.

Broome's theory will imply the Repugnant Conclusion since it will have to admit that a population in which everyone lives a very good life is worse than a much bigger population in which everyone lives only just above the neutral level for existence (pp. 210–13). How problematic this conclusion is depends crucially on where the neutral level is set. If it is set at the level of a mediocre life, then it is especially disturbing. But if it is set at the level of a fairly good life it is less so. It does not seem that counterintuitive to say that a very large population of people with fairly good lives can be better than a much smaller one with very good lives.

However, if the neutral level is set this high, we have another disturbing implication, the Sadistic Conclusion (Future Generations: A Challenge for Moral Theory, forthcoming, and Interpersonal Compensation and Moral Duties to Future Generations: Moral Aspects of Energy Use, 1995). Broome's theory will now imply that a population with miserable lives may be better than a much bigger population with lives that are just below the neutral level (pp. 213–14). If the neutral level is the level of a fairly good life, we get the absurd conclusion that a population with miserable lives is better than a much larger population with good lives. In this case, the sadistic conclusion really deserves its name. There seems to be something sadistic about valuing the population of miserable lives above the population of good lives. The sadistic conclusion will of course lose some of its bite if the neutral level is set much lower than the level of a fairly good life, but then the repugnant conclusion will be more troublesome.

Broome's response to this dilemma is to invoke the vagueness of the neutral level (pp. 213–14). It is a vague matter exactly where the neutral level should be set. Instead of a precise cut-off point we have a vague borderline. If the borderline is wide, then we can make each of the conclusions less problematic. But Broome acknowledges that a wide borderline will have other ‘incredible’ implications (pp. 214, 202–5).

So, Broome's theory seems to be stuck between repugnance and sadism. Not a nice position to be in. The sadistic conclusion, which seems to be the greater ‘evil’ of the two implications, could be avoided if the neutral level was not set as positive. So, why does Broome insist that it should be positive?

One important reason is that doing this will give him a theory that captures the value of longevity. Broome introduces this value early on in his book, when he talks about a couple that can choose between extending their already existing child's life and having a new baby instead (pp. 8–9). Most people would find it obvious that they should extend the existing child's life instead of creating a new life, even if the distribution of temporal well-being would be the same no matter what they did. By setting the neutral value for existence as positive, we can capture the value of longevity in this sense. Applied to the couple's choice, this means that it is better to create the new child if the child's lifetime well-being is sufficiently above the neutral level. The difference between it and the neutral level must be greater than the loss of well-being suffered by the existing child. It might be seen as a virtue that Broome's theory captures our intuition about the value of longevity. But I doubt that our intuitions in the case of the couple's choice have very much to do with the value of longevity as Broome understands it.

First of all, it is crucial that if a life is not extended then there is someone who will be worse-off. This is crucially different from not creating the new child, since this will not make this child (or anyone else) worse-off. The value of longevity, as Broome understands it, does not take this into account, since it can be applied to cases in which only future contingent people are at stake.

Second, the intuitions about the couple's choice trade heavily on our commonsensical ideas about parental duties. We have a special duty towards our already existing children that does not apply to the children we have not yet created. The intuition seems therefore more deontic than axiological in nature.

Finally, Broome's theory will only give the intuitively right result in the couple's case if the neutral level of existence is set very high. Even if the new child would have a blissful life, many would think that this is not enough to make it better to let the existing child die, if the existing child will have a good future life. But if the level is set close to the level of a blissful life, then the sadistic conclusions will come back with a vengeance. It can be better to create some miserable lives than a huge number of near-blissful lives.

Apart from considerations about the value of longevity, Broome thinks that a positive level for existence is independently plausible:

It is not clear what Broome is hinting at here. He claims that a constantly neutral life may be as bad as a life lived in coma throughout, but what makes this life bad? In another context, Broome suggests that a life in a coma is bad because it is meaningless (p. 209). But one would have thought that a life that is meaningless is not just impersonally bad but also personally bad, bad for the person leading the life. But Broome cannot say this, since he claims that a life in coma is as good for the person as dying, not worse for her than dying. So, we get the odd result that a meaningless life is impersonally bad, but not bad for anyone.