On Axel Gosseries’ “What is Intergenerational Justice?”

Summary

What Is Intergenerational Justice? begins with a discussion of the nonidentity problem in intergenerational ethics and offers a strategy for solving this problem. To understand what the nonidentity problem is and why it is important, let us begin with a well-known case:

Risky Policy: Suppose that, as a community, we have a choice between two energy policies. Both would be completely safe for at least two centuries, but one would have certain risks for the further future. If we choose the Risky Policy, the standard of living would be somewhat higher over the next two centuries. We do choose this policy. As a result there is a … catastrophe two centuries later, which kills and injures thousands of people’. (Parfit 2010: 112)

As has famously been argued by Derek Parfit, our choice of energy policy affects who will live in the future. (Why is that? Policies have knock-on effects: they might affect where people live, whom they meet, and when they decide to have children and thereby affect the identity of those children.)

The nonidentity problem arises when we combine the identity-affecting nature of actions with the standard account of harm, which says that an action harms a person if and only if the person is worse off than the person would have been had the action not been performed. The standard account of harm implies that the risky policy does not harm future people, since they would not be better off had the safe policy been chosen. In fact, these future people would never have existed. Further, if the risky policy does not harm future people, choosing the risky policy does not wrong them, given the plausible assumption that we cannot wrong others without harming them. But this conclusion clashes with the widely held claim that choosing the risky policy does wrong future generations.

The nonidentity problem is admittedly abstract. But it threatens to undermine attempts to theoretically justify moral duties of justice to future generations. This is why many philosophers believe that the problem is worth thinking about.

Gosseries discusses three strategies to solve the nonidentity problem. Here, I focus on two: one which Gosseries rejects, and one which he accepts. I begin with the strategy Gosseries rejects. The ‘new grammar strategy’ rejects the standard concept of harm. Instead, it accepts a threshold concept of harm, according to which a person suffers harm when they are worse off relative to some absolute threshold (Gosseries 2023: 36–40). This solves the nonidentity problem: the risky policy causes future people to be badly off and thereby harms them. Further, the threshold concept nicely fits with sufficientarian accounts of distributive justice, according to which a distribution is just if everyone is better off than some absolute threshold (42). However, Gosseries rejects the new grammar strategy, as it is incompatible with most views on distributive justice; namely, those that are not sufficientarian (42). He gives the following example: Imagine a view that says that a policy A is just if and only if the least well-off person in the world where policy A is chosen is better off than the least well-off person in the world where the alternative policy B is chosen (42). If I understand this point correctly, a problem arises in cases such as the following: Assume that regardless of which policy is chosen, everyone in our hypothetical world will end up above the harm threshold. Further, assume that if we choose policy A, everyone will be very well off, but if we choose policy B, everyone will be only moderately well off. According to the view on distributive justice Gosseries describes, we should choose policy A. But we cannot explain this with reference to the threshold concept of harm since neither policy harms anyone.

The solution to the nonidentity problem that Gosseries prefers is the ‘full severance solution’ (Gosseries 2023: 48–50). This solution takes seriously the idea that duties of justice and duties not to harm are distinct. We can thus concede that choosing the risky policy might not harm future generations while insisting that in choosing the risky policy, we violate duties of justice to the future.

However, justice and harm are difficult to tear apart. The connection between justice and harm often seems to underlie moral thinking regarding intergenerational justice. As Gosseries notes, in climate justice, duties to the future are often seen as duties to compensate for harms caused by actions of past and present generations. This, very roughly, is what Gosseries calls the ‘rectificatory’ approach (120–126). On the rectificatory approach, descendants of past wrongdoers have special duties to compensate for climate-related harms. But Gosseries thinks that descendants of past wrongdoers have no special duties to compensate (beyond symbolic duties), since the descendants have not done any wrong themselves. Like all moral agents, they nonetheless have duties to compensate for unjust distributions. In this vein, Gosseries defends a ‘distributive’ approach to climate justice (129–32). On the distributive approach, duties of justice require us to ensure that a just distribution of benefits and burdens obtains across generations.

In sum, Gosseries argues that the tight connection between justice and harm is problematic in at least two ways. First, it gives rise to the nonidentity problem, undermining attempts to justify duties to the future. Second, it leads us to think of climate-related duties to the future in rectificatory terms as compensation for past harms when we should be thinking about these duties in distributive, forward-looking terms.

Nonidentity

I believe that the new grammar strategy might hold more promise as a potential solution to the nonidentity problem than Gosseries suggests. Recall Gosseries’s point that the threshold concept of harm fails to explain unjust distributions above the threshold. For example, a world in which everyone is sufficiently well off but some are much better off than others might seem unjust to some with egalitarian leanings. But, following the threshold concept of harm, there is no harm in this world. If injustice requires harm, this world is not unjust.

But the threshold concept might hold more explanatory power than this line of argument suggests. Arguably, the threshold concept offers an account of benefit as well as harm. Being better off than the threshold counts as a benefit, just as being worse off than the threshold counts as a harm. Further, as noted previously, distributive justice is concerned with the distribution of both benefits and burdens. So an unequal distribution of benefits can be unjust, even when nobody suffers harm. The threshold concept can explain these cases: some people enjoy more benefit than others, relative to the threshold. So, I suggest that the threshold concept can capture the injustice of unequal distribution of benefits by conceptualizing this case as an unjust distribution of benefits.

Interestingly, understanding the threshold concept as an account of benefits, as well as harms, loosens the connection between justice and harm (without, however, fully severing it), since injustice can result from unjust distributions of benefits as well as from the unjust distribution of harms.

If these considerations are on the right track, the new grammar strategy might be more promising than it initially seems.

Rectificatory Duties

My second point addresses Gosseries’s argument that climate justice is better thought of in terms of distributive justice than in terms of rectificatory justice. Recall that on the rectificatory view, descendants of past wrongdoers have special duties to compensate for the wrongdoings of their ancestors. In contrast, the distributive view says that duties of justice require us to compensate for present injustice, regardless of who is historically responsible for it. Consider:

Costless natural climate change: In World I, the climate is changing in ways that will lead to significant harm. The climate change in World I has entirely natural causes. Further, the people in World I have developed a cost-free way to fine-tune their planet’s temperature, and so prevent their climate from getting worse (adapted from Gosseries 2023: 127).

Are our duties to compensate for natural climate change equally strong as they are in the case of climate change caused by humans? As Gosseries argues, yes: the people in World I have duties to prevent their climate from getting worse that are at least as strong as duties to prevent the climate from getting worse (127):

Costly human impact climate change: In World II, the climate is changing in ways that will lead to significant harm. The climate change in World II is in large part due to human impact. The people in World II have developed measures to prevent their climate from getting worse, but at significant cost to themselves.

Gosseries assumes that intervention in World I is costless. What happens if we remove this assumption, but keep the rest of the case the same?

Costly natural climate change: In World III, the climate is changing in ways that will lead to significant harm. The climate change in World III has entirely natural causes. The people in World III have developed measures to prevent their climate from getting worse but at significant cost to themselves.

It seems to me that Gosseries’s intuition that the people in World I have stronger duties to prevent their climate from getting worse than the people in World II is driven by the idea that we have strong duties to prevent bad outcomes if we can do so at no cost to ourselves. But when we equalize the cost, things begin to look different. When we compare World II and World III, it seems, perhaps, less intuitively clear that there is no moral difference between the cases.

But why think there might be a moral difference? One might appeal to institutional or collective duties. Institutions can be held responsible over periods longer than individual lifespans. Might people today share responsibility for at least some of their ancestor’s actions in virtue of, for example, their shared citizenship? Alternatively, one might think that while I am not morally responsible for the actions of my ancestors, it can in some circumstances be appropriate for me to take responsibility for their actions. This does not mean that we are responsible for things we cannot control. It merely says that it can be appropriate to take responsibility for things we cannot control, including, perhaps, actions of people we stand in special relationships with. Now, I do not claim that these lines of argument would succeed. I am not confident that they would, and in any case, they require more argument. I merely suggest that once we hold costs equal, it seems less clear that there is no morally relevant difference between the cases.

The Relationship between Harm and Justice

The severance strategy is interesting and worth exploring. It can account for the plausible idea that duties of distributive justice and duties not to harm operate in different domains of ethics. But I suggest it faces the challenge of explaining how harm and justice interact within a broader system of ethics. If we allow that considerations of harm and considerations of justice can come apart entirely, the question remains how we can put them back together to arrive at an all-things-considered moral verdict.¹

I submit that harm has moral relevance. An action that constitutes harming is worse than an action that does not constitute harming, other things equal. But if harm and justice both carry moral weight, the nonidentity problem returns. Consider the following variation of the risky policy case:

Risky policy present: We have a choice between two energy policies. Both policies have benefits in the very short term but impose risks on the future. The first policy would be completely safe for one hundred years but imposes risks on more distant generations. The identity of these distant people depends on our policy choice. The second policy is similarly risky, but the risks would materialize within the current century and so would affect people who will exist regardless.

It seems that the first policy imposes unjust burdens on distant future generations whereas the second policy imposes unjust burdens on those who will live in the next one hundred years. Assuming that the first policy is unjust to future generations, and that the second policy is unjust to the present generation, both policies violate our duties of justice. However, if the nonidentity argument is correct, then only the second policy is harmful, whereas the first is not. The first policy is not harmful because the first policy changes the identity of those who will bear the risks that the policy gives rise to. Presumably, this makes a moral difference. We have strong duties not to harm others. So, when we accept the severance solution, there remain moral reasons against doing harm by choosing the second policy, which are not ‘matched’ by moral reasons against choosing the first policy.

One response is to bite the bullet. Perhaps our duties to future generations are indeed weakened by nonidentity considerations. This response might be compatible with Gosseries’s approach: Gosseries is careful to emphasize that intergenerational justice is only a part of justice, which is in turn only a part of ethics. So, he might be happy to agree that the nonidentity problem retains some of its bite, albeit outside of intergenerational justice.

We can push further, however. A practice often criticized by philosophers is to discount the well-being of future generations (Gosseries 2023: 142–48). But the nonidentity problem might justify discounting.² Here, the idea is that the standard account of harm captures a morally relevant feature of our actions: making others worse off than they would have been matters morally. If this feature is present when we harm present generations but not future generations, then this might justify discounting the well-being of future generations.

This argument is strengthened further when we recognize that the nonidentity problem does not just arise in cases of harm. It also arises in cases of beneficial policies, if we assume that benefitting someone means making them better off than they would have been (Unruh 2020). If our policies cannot benefit future generations, then we have no, or at least weaker, duties of beneficence to future people whose identity is affected by our actions than to present people. If this is so, then this would also give us a reason to benefit the present or near future (rather than the far future)—at least, unless we develop a new grammar of harm and benefit, as discussed earlier.

Conclusion

I shared some reflections on Gosseries’s severance strategy. I began by suggesting that harm-based considerations might have a role to play in solving the nonidentity problem and in explaining climate duties. In the last section, I took up the question how harm-based considerations combine with justice-based considerations. These questions are as puzzling as they are relevant.

Most of us care about future generations. We want our children, and our children’s children, to have good lives. But we have different views on what makes a good life, how best to secure it for future people, how much we should sacrifice for the future, and whether governments can legitimately enforce such sacrifices. Gosseries’s book draws out such complexities. It is written in the spirit that philosophy can help us think clearly about complex issues (Gosseries 2023: 178). Such clarity might help promote understanding in an increasingly polarized discourse and, further, guard against potential misuses of intergenerational justice talk, such as claiming to act in the interest of future generations as a rhetorical move to silence disagreement (3) or using intergenerational justice talk to divert from other pressing issues such as social justice (4).

Philosophy does not, and cannot, answer questions that are inherently political. But philosophy can show the complex nature of the questions, uncover assumptions we make in asking them, and illuminate the space of considerations that lead reasonable people to disagree. In this way, philosophy might help to promote a thoughtful and nuanced debate on the complex issues that arise when thinking about justice, within and across generations.