Animals are moral patients, so future animals will be moral patients. It is inappropriate to disregard their interests simply because they are temporally distant. Temporal distance certainly causes morally relevant ignorance and makes it difficult to attend to animals’ interests, but insofar as future animals’ interests are known, agents are morally accountable to them. Where the future existences of animals are not assured, as in real-world conditions, it is unclear that these obligations exist, and practically it is hard to act on the basis of the (potentially non-existent) interests of non-existent animals. However, I will argue agents can be certain that the animals that they are creating will exist and will be moral patients. Moreover, in creating them, assigning them both an identity and a context, agents prove their capacity to meaningfully affect the kind of lives that the animals that they create will live.
It is possible to do something bad for an animal in causing it to exist. One is unable to wrong an animal which does not yet exist: rights are only ascribable to what exists. But then one cannot harm an animal by causing it to exist, at least on the most common accounts of harm, because its welfare is not diminished by being made to exist; before existing, no welfare was ascribable to it. Yet, looking past wronging and harming, it would still be wrong to create (for example) an animal constituted solely to suffer and which indeed suffers, the most tragic play on the utility monster concept. This is because the animal is made poorly off independently of any other state or any right. Non-existence, not allowing for welfare comparable to existence, is not better for such an animal’s welfare, but its existence is actually bad for it, and it could be reasonably upset at being made to be as it is.
Note the emphasis on its existence. There is no thing which is just a being; to be is to be in some way. Being, being multiform, is not directly comparable to the absolute unity of that which is not. Whereas non-existence could not be anything for its subject, as it has none, each way of being is good, bad, or neither for its subject.
When that subject is a moral patient, and that moral patient has a life, as would be the case if it were an animal, the decision to create it should be a decision which considers the whole of its life. David Benatar proposes that it is always immoral to make a sentient being because of an asymmetry between good and bad experiences. He claims that it is good, even for the non-existent, to lack bad experiences, but that is not bad to lack good experiences if one does not exist. This asymmetry has been much critiqued, with most criticisms focusing on whether a thing is good if it is good for no-one. Benatar justifies it by analogy; while we regret making people who suffer, we do not regret not having children who would have good experiences had they existed. This gets at the central flaw of his theory, which is that it considers experiences rather than lives. No parent regrets having a child just because it cries when it is born, but they might regret having a child whose life is terrible overall. As these experiences are aggregated into the animal’s life, it is appropriate that they should be aggregated for decision making about creating those lives. This suggests a much more defensible position: intentionally causing an animal to come into being when one believes that it will have a bad life (absent relevant compensatory factors) is wrong because it is bad for the animal. While it is not bad to not have a good life if one does not exist, it is bad to not have a good life if one does. Unlike the former asymmetry, there is no appeal here to good for the non-existent, nor does this formulation implausibly imply the moral necessity of maximizing reproduction. Instead, it suggests, intuitively, that it is bad to make animals that one expects will have bad lives.
Now we reach the real crux of the issue: the non-identity problem. This problem hints at a critical gap between commonly-held intuitions about what animals one ought to make when given a choice and the ability of existing ethical theories to justify, or convincingly refute, those intuitions. Most people believe that, given a choice between making an animal which will have a good life and one which will have a better life, ceteris paribus, one ought to choose the one which will have the better life. The difficulty is explaining, in patient-affecting terms, why one ought to do this. If one chooses to make an animal which will have a worse life, perhaps a pure-bred with health problems instead of a mutt without them, the created animal, the one which actually exists, is not harmed or wronged. It gets the benefit of a good life, and if a mutt with a better life had been made, the pure-bred would not have been its beneficiary, as it would not exist. The non-identity problem is that, without a fixed set of moral patients accounts of wrongdoing that rely on the moral relationship an action has with a particular patient fail to match our intuitions.
There is currently no unified ethical theory which solves the non-identity problem in a way which matches our intuitions. Rights-based attempts run into problems two ways; they either assign rights to animals which do not yet exist and so can have no right conferring or containing attributes, or they assign retroactive rights not to have been caused to exist in some way, which any reasonable moral agent would waive so long as the life which she is given is good. Contractualist approaches fail because they require, contrary to their metaethical foundations, that respect be given to the interests of moral patients which will never exist, with untenable consequences, lest they be unable to provide guidance at all. Harm proves difficult to ascribe to something which had no prior state, and so is ill-suited, even in specialized forms, to addressing the non-identity problem; noncomparative event-based accounts imply absurdities and noncomparative state-based accounts cannot account for standards higher than the expectation of a good life. Consequentialism has well-known difficulties in handling situations with variable numbers of patients. Naturalism and ecofeminism cannot independently substantiate ethical claims, though they may serve as valuable calls for caution. Virtue ethics and care ethics cannot justify a standard for creation different than the expectation of a good life for the created. Attempts to explain why any higher standard for virtuous conduct or a caring relationship have to explain why one is not charitable or caring if one does not do as much as possible to promote the well-being of an Other which does not yet exist, a ludicrously high standard.
The failure of beneficence approaches deserves special mention; in encountering the non-identity problem, some philosophers have decided to codify what they take to be common ethical intuitions in the guise of a theory. They assert that people ought to make animals which can be expected to lead lives which are better than the lives of other animals they could create. Some take this to the extreme of asserting that one should only create the animals that one expects will have the best lives. Setting aside whether these principles actually match intuition, concluding by intuition alone is not sufficient for an ethical theory, particularly where a major difficulty with the intuition has been demonstrated, as is the case with the non-identity problem. Beneficence approaches should then be rejected until they can provide an independent theoretical justification.
It may prove productive to continue to attempt to solve the non-identity problem, but, barring the invention (discovery?) of a radically new kind of ethical theory which can be both substantiated metaethically and which can explain or refute our intuitions, this is a branch of research unlikely to bear much fruit, no matter how many folia are wasted on it. This is not to say that no solution is possible, but that the ethical tools to solve it do not yet exist.
Our responsibilities in making the moral patients that we choose to make are better understood by considering how these choices will harm or wrong existing moral patients. Because moral patients interact with other moral patients in predictable ways, it is reasonable to attribute wrongdoing to the creator of a maleficent animal, when she ought to have expected the animal to do wrong by others. This form of wrongdoing is possible even when the earlier substantiable ethical restriction on making animals, that animals made ought to be expected to have good lives, is followed, and so it constitutes a new, higher standard for creation.
This argument may read to some as skirting the issue, which it does, and justifiably so. The non-identity problem is not only relevant to decisions about individual reproductive choices. It also must temper the more interesting kinds of ethical claims one might make about intergenerational obligations, both from the past as a reason to reject reparations and for the future as a reason to reject conservation. This paper’s decision not to attempt to solve the non-identity is in some ways a dodge around the apparently central ethical issue at stake. That dodge should be excused for three reasons; the first is that the apparent dodginess is acknowledged, the second that reasons have been given to show why one might believe that solving this problem is an impossible task, and the third is that the actual goal of the paper was to learn about the smaller domain of making animals.
The second likely accusation of skirting the issue also identifies a real evasion, but, again, not a troublesome one. The final claim presented, that it would be wrong to make animals which will do wrong by other moral patients, seems to differ from the sorts of arguments on which this paper focused, namely those trying to show why creating an animal could be wrong because of the created animal’s own moral status, because it lacks the same focus on the created animal qua moral patient. That is a real distinction in kind of argument, and it should be acknowledged as such. However, while one of the goals of philosophical engagement with an ethical problem may be to face it head on, it is more important that one provide a (more) complete account of moral behavior than that one answers every question a pesky philosophy major may ask. The non-identity problem helps to explain this point precisely by showing the shortcomings of existing ethical theories. While a philosopher may not be able to deal with an ethical problem on terms she can choose, she still should find terms she can use.
 Perhaps not all, but some, and those are the ones an ethicist has reason to care about and the ones this paper addresses.
 Please excuse the more transcendent deities whose existence you may wish to provide as a counterexample.
 This is not to say that all that is required to have a good life is to have a preponderance of good experiences, but that, if quality of experience is the deciding factor for this decision, as Benatar suggests, the experience should be of the life, and not of its constitutive elements.
 What it means for a life to be bad is complicated, and I do not aim to present an explanation of which lives are bad. Instead, I appeal to the intuitions, which I believe are well founded, that some lives are bad, and that a life can be better or worse than another.
 The former does, and the latter may also, but the latter is so philosophically confused as to be a meaningless ideal.