A non-axiological approach to population ethics

This senior philosophy thesis is among my produest works. It seeks to untangle some difficult quandaries that are a product of consequentialist framings of population ethics. Though this was not my intention, it also strictly bounds the badness of existential risks.

CONTENTS

• INTRODUCTION
• 1 FOUNDATIONS
  • 1.1 The No Reason Principle
  • 1.2 Neutral Axiologies
  • 1.3 An Alternative to Teleological Ethics
  • 1.4 Person-Affecting Views
  • 1.5 Impartiality
  • 1.6 Non-Identity Problem, Revisited
• 2 FRAMEWORK
  • 2.1 The Nature of Moral Permissibility
  • 2.2 The Independence of Irrelevant Alternatives
• 3 CATALOG OF REASONS
  • 3.1 Second Child
  • 3.2 Lives Not Worth Living
  • 3.3 Necessary People
  • 3.4 Repugnant Conclusion
• 4 MISSING COUNTERPART RELATIONS
  • 4.1 The Problem with Necessary People's Wellbeing
  • 4.2 Reasoning Under Superposition
  • 4.3 Averagism and Axiology
• CONCLUSION
• BIBLIOGRAPHY

INTRODUCTION

Since Derek Parfit brought population ethics to the attention of moral philosophy, philosophers have searched for a satisfactory population axiology: a theory that determines what makes one state of affairs better than another when they differ over the numerosity, identity, and wellbeing of their constituent members. This project has been difficult. Existing theories have struggled to accommodate core principles like the Procreation Asymmetry, the Person-Affecting Intuition, and the Non-Identity Intuition. My thesis strays from the quest for axiology. Instead, we aim to plot a course for a normative population theory to support the aforementioned principles. We will create a framework for judging the permissibility of different outcomes and providing an account of the moral reasons underlying procreation scenarios.

In Chapter 1, we will discuss extant theories in population ethics and motivate our own approach, which affirms the Impartiality Principle while adopting Johann Frick’s bearer-regarding account of reasons. In Chapter 2, we will introduce the moral reasoning framework through which we will analyze population ethics problems. In Chapter 3, we will study problems that involve tradeoffs between various moral claims brought forth by population members and demonstrate the versatility of our moral reasoning framework in accommodating a wide range of intuitions. Finally, in Chapter 4, we will clarify how our theory handles cases that seemingly require a counterpart relation, which we leave undefined. Our goal is not to provide a definitive guide to ranking possible worlds, but to reduce population ethics quandaries down to problems that existing ethical theories can handle.

1. FOUNDATIONS

Population theories have generally fallen in two groups: neutral axiologies and person-affecting views. Neutral axiologies adhere to what Gustaf Arrhenius calls the principle of neutrality, whereas person-affecting views do not.

The distinction concerns the kinds of facts used to judge populations. Unlike person-affecting views, neutral axiologies do not consider facts pertaining to identity. It does not matter whether Alice exists in both worlds or alternatively Alice exists in one world and Bob exists in the other, so long as their welfare is held constant.

Both families of views have separate troubles with upholding the trifecta of principles below. We will explore why, and set the stage for our own theory, through an analysis of their theoretical merits and shortcomings.

1. Create Nobody
2. Create Fivey, a person with a good life
3. Create Tenny, a person with a great life

1.1 The No Reason Principle

The No Reason principle is the common thread between the Procreation Asymmetry, Person-Affecting Intuition, and Non-Identity Intuition. Given its significance to the goal of accommodating the aforementioned intuitions, it’s worth a brief discussion on why it is compelling.

Consider the following choice scenario, meant to illustrate the appeal of the person-affecting intuition.

If we deny No Reason, the sheer amount of wellbeing that comes from creating trillions of happy lives compels us to pick Adam & Eve. This has troubling societal implications. Instead of addressing issues like hunger and poverty, we may want to channel trillions of dollars into a long-term project of terraforming and populating Mars. It is also out of line with our everyday moral priorities. We do not lose sleep over unpopulated planets. But we do have a profound sympathy for those experiencing acute hunger or those dying before they could complete their life projects. A reasonable mind can be indifferent to the existence of extraterrestrial life, but only a madman would not wish that there was less hunger in the world!

What’s more troubling is when we consider the offsetting potential of creating new lives. Consider Stuart Rachels’s Strong Thesis.

If true, the prevention of a possible happy life would be worse than the termination of an extant happy life, as the unborn person would have more positive experiences denied than the already existing person. We could imagine a modification to the trolley problem scenario where the onlooker can create five people by diverting the trolley to kill somebody tied to the tracks. Sacrificing one person to create five people does not seem nearly as justifiable as sacrificing one person to save five people. How can mere possibilities have a moral claim over actual people? Discounting the wellbeing of possible people does not solve this problem. Given a sufficiently large addition to the population, such a view will prioritize the outcome that adds new people at the expense of bettering or saving existing lives. It would also be challenging to motivate a particular discount factor for potential people’s wellbeing (Should we value their wellbeing at half that of an existing person? One-third?).

The Strong Thesis (or a weaker version) would have radical implications for contraceptives, abortions, and general reproductive freedoms. The hardships of child-rearing likely pale in comparison to the goodness accumulated over the course of a happy life. Though Rachels argues that procreation can be considered a supererogatory duty, it’s unclear whether the demands of parenthood counterbalances the benefits of procreation. Perhaps it can forestall an obligation to create children for extremely unwilling parents, but those seem the exception rather than the norm.

That said, many people do feel like creating happy lives is a good thing. Rachels suggests that “it would be good for God to add ten billion flourishing people to a distant part of the cosmos” (Rachels 2002, p. 103), contra to our earlier case. He cites Bennett, who has a “liking for rich, organic complexity” and argues that “it would be better for more happy people to exist because then there would be more rich, organic complexity” (Bennett 1978, p. 64).

I acknowledge that many people have a preference for a bustling universe, but I believe that Bennett’s desire for a universe teeming with life is an aesthetic one, not a moral one. Indeed, he acknowledges that his “pro-humanity” (p. 65) stance does not rest on any principle (p. 65). Those who prefer Adam and Eve judge it to be a better world relative to their personal tastes and value system, where values include non-moral ones, and are equivocating on the moral and non-moral use of “better.”²

1.2 Neutral Axiologies

By and large, neutral axiologies deny the No Reason Principle. These include totalism, averagism, critical level theories, and variable value theories.

Totalism and averagism suffer from several philosophical challenges. Since totalism relies solely on the total amount of wellbeing in a world, and does not consider the distribution of wellbeing, it can favor worlds with high inequality. A world where Alice and Bob have wellbeing levels³ 1 (a life barely worth living⁴) and 10 (a great life) respectively is better than a world where Alice and Bob both have a wellbeing level of 5 (a good life). In addition, totalism entails the Repugnant Conclusion.

In essence, a dystopian world with quintillions of people living out meager existences will contain more total wellbeing than a utopian world with billions of people living lives of great happiness and fulfillment. This is widely considered to be counterintuitive and the other three neutral axiologies are motivated in part by a desire to avoid the Repugnant Conclusion. Averagism judges a world by the average wellbeing of its population and thus correctly considers the dystopian world far less choiceworthy than the utopian world. However, averagism faces its own problems. It is a context-sensitive theory—a great life can have negative contributory value and a terrible life can have positive contributory value depending on the current average wellbeing of the world. As such, the goodness or badness of a child’s birth can depend on seemingly unrelated facts, such as the quality of previous lives or lives in faraway places. Parfit argues:

There are even stronger objections to Averagism; we will revisit them in a future chapter.

In general, neutral axiologies require just two pieces of information to determine the moral desirability of a world: population size (n) and average wellbeing ($wb_{avg}$) . Through some function of these parameters, the world is given a score which measures the intrinsic degree of goodness of that particular state of affairs. Supporters of neutral axiologies are typically consequentialists and endorse the Normative-Axiological Link.

For all the neutral axiologies, adding a life with sufficiently high wellbeing increases a world’s score, which (due to the Normative-Axiological Link) generates a strong moral reason in favor of doing so. Since these theories are indifferent to whether the score increases because existing people’s wellbeing increases or because new bearers of wellbeing are added, they are teleological in nature. According to Johann Frick, the teleologist is somebody who believes that “the unique appropriate response to what is good or valuable is to promote it, ensuring that as much of it exists as possible” (p. 19). Teleologists like Moore characterize the good as simply what ought to exist (Moore 1903). Since this is usually identified as human wellbeing (= welfare = happiness), they believe that all our moral reasons are rooted in the effort to bring about more wellbeing.⁵

The theories’ teleological nature signals trouble for No Reason. Since creating a happy life contributes to the total amount of good in the world, we have a moral reason to do so. To affirm No Reason, we must either abandon teleology or restrict the scope of individuals whose wellbeing ought to be promoted. We adopt Frick’s view which employs the former approach, while the person-affecting views that we will later explore opt for the latter.

1.3 An Alternative to Teleological Ethics

Heart of Morality

Utilitarians are often accused of misidentifying the heart of morality. Their theories place wellbeing as the end. This has the effect of subordinating human beings to serve as the substrate of what is actually valuable—wellbeing. Singer (1993, p. 121) writes, “the total version of utilitarianism regards sentient beings as valuable only insofar as they make possible the existence of intrinsically valuable experiences like pleasure.” This does not seem right. We want more human wellbeing in the world because we care about people, and we want things to go well for the thinking, feeling creatures that inhabit our world.

As we’ve discussed, there are two ways of increasing the amount of wellbeing in the world. We can increase the amount of wellbeing among existing people, or we can create new people who will live good lives. Fixed population utilitarianism can avoid the brunt of the objection by claiming that their commitment to bringing about the best state of affairs stems from a concern for people; they wish to maximize wellbeing out of a desire to improve people’s lives.⁶ But this line of reasoning is not valid in variable population cases, where we have the option to create new bearers of wellbeing.⁷ Any pro tanto reason to create a happy person must be derived from an appreciation of the contributory value of their wellbeing, not from a desire that their life goes well. Reasons that are derived from somebody's existence can’t also function as reasons to bring them into existence.

The teleological ethicist’s outlook towards wellbeing falls prey to Goodhart’s Law: when a measure becomes a target, it ceases to be a good measure. A theory that aims to maximize human wellbeing invites the loophole of creating more people. This is misaligned with the true goals of morality, which is not to create as much happiness as possible, but to ensure that as many people as possible have their interests met.

Bearer-Regarding Reasons

Johann Frick argues that our moral reasons to confer wellbeing on people are “bearer-regarding” reasons. Since they are ultimately rooted in a person’s moral status, they are conditional on existence. He attempts to place the Procreation Asymmetry in the broader phenomenon of moral values exhibiting an innate asymmetry through a comparison to promise-making.

Most of us believe that we have a moral reason not to make a promise that we won’t be able to keep. (Compare: we have a moral reason not to create a life that will unavoidably be not worth living). By contrast, we do not think that we have a reason to make a promise just because we will be able to keep it (Compare: we do not think we have a moral reason to create a new life, just because that life will be worth living). Keeping a promise does not seem to add any moral value to the world that must be taken into account when deciding whether to make that promise. (Frick 2014, p. 25)

Our attitude towards moral virtues like promise-keeping, justice, and equality is not one of maximal instantiation. We do not think that making and keeping tons of promises is a way to improve the world. Similarly, we do not care to create more people just so they can treat each other justly. Our reasons for instantiating a value are conditional on the existence of somebody who has a claim to that value. When we say that promise-keeping matters, it is because we have an obligation to somebody that is derived from the promise we made to them. Similarly, we have a moral reason to promote wellbeing among actual people, derived from the moral status granted by way of their existence as a sentient being. We also have moral reason to avoid creating an unhappy life—doing so would result in a world where a person lacks happiness, a deeply lamentable situation. In contrast, since possible people do not have moral status yet, we have no moral reason to promote their wellbeing by creating them.

The bearer-regarding account of reasons is a persuasive way to affirm No Reason and will shape our approach to addressing contingent people when we begin constructing our moral reasoning framework. In the following section, we will reflect on the person-affecting views to establish the groundwork for the Impartiality Principle, the other major pillar of our theory.

1.4 Person-Affecting Views

Person-affecting theories incorporate the Person-Affecting Restriction, the idea that moral claims necessarily involve a reference to humans. In slogan form, it goes: an outcome can only be better or worse than another if it is better or worse for somebody. The world where Bob is punched is worse than the world where Bob doesn’t get punched because it violates Bob’s desire not to be attacked.

The person-affecting theorist’s approach is to limit who counts as a “somebody”—a person for whom an outcome can be better or worse—and consequently whose welfare matters for moral judgements. The general idea is that an outcome can’t be better (or worse) for someone if they don't exist in the alternative. A nonexistent person can't act as the recipient of an existential boon.

Person-affecting views differ on the requirements for moral status. Suppose that Bob happily exists in Outcome 1 but does not exist in Outcome 2. Presentists, Actualists, and Comparativists argue that Outcome 1 is not better for Bob because he is not a currently existing person, a person who does or will exist, or a person who does or will exist in all possible outcomes respectively. Since Outcome 1 is not better for Bob (and likewise Outcome 2 is not worse for Bob), Outcome 1 and 2 are not better or worse than each other, and so there’s no moral reason to bring about Bob.

This provides a means to justify the No Reason principle. The Person-Affecting Intuition follows from the Person-Affecting Restriction, but problems arise for both the Procreation Asymmetry and the Non-Identity Intuition. The first half of the Procreation Asymmetry prohibits the creation of unhappy lives. But we can employ a parallel argument to come to the conclusion that there is no reason to prevent the creation of a suffering individual. Suppose that Bob now unhappily exists in Outcome 1 but does not exist in Outcome 2. Outcome 2 is not better for Bob since he is not an existing/actual/necessary person, so there’s no reason to bring about Outcome 2 over Outcome 1 (and vice versa).

What do Person-Affecting views say about the Non-Identity Problem?

If Fivey exists in one world and Tenny exists in another, neither world is better or worse for Fivey or Tenny. The person-affecting notion of harm is individualistic. In order for somebody’s welfare to be lower in one outcome, we have to presuppose that they exist in both. Outcome 1 is not bad for Fivey, since she wouldn’t exist otherwise. It’s also not bad for Tenny, since she doesn’t exist. Same goes for Outcome 2. Since neither outcome is better or worse, we have no moral reason to bring either about. This runs contrary to our intuitions. Most people would agree that we should create Tenny, since they would live a better life than Fivey. In fact, our treatment of Fivey and Tenny should parallel the Same-Person Case.

Since Outcome 2 is better for Bob than Outcome 1, we have a moral reason to choose Outcome 2.

The Neutral Axiologies sidestep the Non-Identity Problem by not considering facts of identity when evaluating outcomes. Outcomes are judged based on their distribution of welfare without regard to whether particular persons are advantaged or disadvantaged. This has often been interpreted as a shortcoming, as it gives rise to viewing people as fungible receptacles of value. But I believe that there’s a way to remove the identity constraint while not running afoul of the fungibility criticism and preserving the spirit of the person-affecting restriction.

The person-affecting restriction considers actions as good or bad only if they are good or bad for someone. Traditionally, this is taken de re—“someone” is used to pick out a particular individual. Consider Parfit’s delayed pregnancy case.

It seems that Sarah made the wrong choice in choosing to have a child now. She left her child with a bad start in life. But consider that her current child would not have existed had she waited ten years. If she waited, her child would have been a different person entirely, born into a different decade and possessing a distinct genetic profile. Since Child Now and Child Later are “different people”, the person-affecting restriction is silent on who we have more reason to create. Neither outcome is bad for someone de re.

But there is a sense in which having Child Now is bad for someone—it’s bad for Sarah’s child! Since Sarah is set on having a child at some point in her life, some particular person has to fulfill the role of Sarah's future child. Sarah has a welfare-related reason, grounded in her future child’s wellbeing, to delay her pregnancy until she is in a better place to care for them. This de dicto reading of the person-affecting restriction collapses the Non-Identity Problem while preserving the intuition that people constitute the foundation of moral claims and that potentially nonexistent people don’t have standing to make a moral claim for their existence.

Typically, when we care about the interests of others, we care about their interests de re. If we want our friend to be happy, abandoning our unhappy friend in favor of a new, happier one would not be considered a valid way of achieving that goal. What we mean by “we want our friend to be happy” is that we care about the happiness of the particular individual that “friend” refers to. It seems to me that while we typically care about existing people de re, we value future people’s interests de dicto. This is because we haven’t formed a personal relationship with a particular person and are unaware of their individuating characteristics. When Sarah says that she cares about the wellbeing of her future child, she is directing her empathy towards her current conception of that child, not towards the individual itself (the individual doesn’t exist!). To that end, if there are competing possibilities as to which particular individual will assume the role of her child, she can unobjectionably say that she wants to create the happier one because she wants her future child to be as happy as possible.

1.5 Impartiality

Our de dicto interpretation of the person-affecting restriction leads us to the Impartiality Principle.

My formulation of Impartiality is similar to that of Ralf Bader, who equates it to permutation invariance, where “permuting the identities of the members of the distribution, whilst holding fixed the structure or value profile of the distribution, does not affect the betterness ordering” (Bader 2022, p. 14). It is also a consequence of Neutrality. Like the neutral axiologies, we only care about the distribution of welfare among people. But we do so in the person-affecting spirit. We want to do what is best for future people; it’s just that we are okay with varying the identities of the future people in the process.

Our feelings towards future people can be characterized as an attitude of general care and concern. They deserve to live flourishing lives by virtue of their (eventual) existence as a highly sentient being. To that end, we have responsibilities to look out for their interests and embark on future-oriented projects. Check out our reasoning in the following scenario.

Notably absent from our moral reasoning are facts of identity. We don’t even know these children’s parentage, much less their likeness as an individual. This does not detract from the crux of the matter: their existence confers moral status, and with it, a reason to care for their wellbeing.

It may be unavoidable that we employ an identity-less kind of reasoning around future people. We can’t hold an attitude of care and concern for a particular future person due to the fragility of procreation. Minute changes to the time of conception will result in a different sperm/egg pair and a genetically distinct individual. We don’t consider ourselves the same person as our siblings (or even twins), despite having the same parents and sharing a similar environment. As such, Alice’s child had she conceived at 5pm is distinct from her child had she conceived a second later. But the precise time of conception doesn’t seem to be a morally relevant fact!

Other facts of identity don’t seem of consequence either. Consider the following story.

Bob Jr. and Charlie Jr. would be very different people. Their genetic identity differs considerably. In addition, their respective fathers would instill contrasting values, which affects their religious beliefs, personalities, and friendships. Perhaps Bob Jr. would grow up in Thailand and live as a Buddhist monk, while Charlie Jr. would grow up in the United States and work as a software engineer. None of this provides a reason to choose Bob Jr. over Charlie Jr., and vice versa.

What matters is that we are bringing into existence a person with a certain wellbeing level. Their wellbeing level encapsulates all that is morally relevant, including their future experiences and genetic identity.⁸ While we can’t ascertain somebody’s exact wellbeing level, we can reason about it under expectation. It does not seem possible to reason about an expected identity. Even if we could, or if we had knowledge about a person’s identity, those facts are only valuable insofar as they shape our knowledge of their wellbeing.

1.6 Non-Identity Problem, Revisited

We are now equipped to handle the Non-Identity Problem. There are two intuitions at stake.

Through the bearer-regarding account of reasons, we find that in choosing Nobody, there is no reason to feel regret for the absence of Great’s happiness. Likewise, there is no reason to create Great solely for their happiness. Therefore, it is permissible to not create Great.

Through the Impartiality Principle, we find that there is a beneficence-related reason to choose the outcome that’s best for the wellbeing of the person that will exist. That Great and Good are different people has no moral bearing—one of them will inevitably exist. Since Great has a higher level of wellbeing than Good, Great is the better choice in terms of the interests of the future individual.

2. FRAMEWORK

We will now develop a framework for analyzing population ethical problems using the Impartiality Principle and the Bearer-Regarding View. I posit three categories of people for the purposes of moral reasoning.

• Actual People: People who already exist or will exist regardless of, and not as a result of, our choice.

• Contingent People: People whose existence is conditional on our choice. These are people who do not exist in one or more of the outcomes.

• Necessary People: People that will exist as a result of, and regardless of, our choice. These people can be thought of as the human beings that are created no matter what. There are as many necessary people in a scenario as the minimum number of humans created across all outcomes.

We will apply the Impartiality Principle to necessary people, the Bearer-Regarding View to contingent people, and conventional ethical principles to actual people.

My usage of the terms “actual”, “necessary”, and “contingent” diverges from how they are used in the literature.⁹ These designations are meant as useful abstractions for moral reasoning and do not carry any metaphysical weight. There are no actual, contingent, and necessary people per se; they are only defined within and relative to a choice scenario. The designations are properties of a scenario, not properties of an outcome of a scenario or individuals within an outcome. Though it will often be convenient to speak of the contingent or necessary people in an outcome, this is technically incorrect. For example, suppose that Alice and Bob are deciding whether to create Charlie. This choice scenario contains two outcomes—one where Charlie exists and another where Charlie does not exist. As such, the scenario contains one contingent person. While it is acceptable here to refer to Charlie as the contingent person, this may not always be the case. Keep in mind that while we will sometimes refer to particular people as bearing a designation, there is often no fact of the matter as to which particular individual has which designation.

Since these designations are defined relative to a scenario, the same individual can function as an actual person in one scenario and a contingent person in another. Suppose that in Scenario A, Beatrice is deciding whether to have a child. Additionally and unrelated to the decision at hand, Alice is pregnant and will give birth to a child. In Scenario A, Beatrice’s child ‘is’ a contingent person while Alice’s child is an actual person. Scenario B takes place a week later. Alice has second thoughts about parenthood and is deciding whether to abort her child. Now her child is a contingent person. My “scenario-relative” definition differs from the actualist’s view, which holds that Alice’s child will either exist in the real world or they won’t. If Alice decides to keep (abort) her child, she is an actual (nonactual) person in both A and B.

Let’s apply our designations to the Non-Identity Scenario. Alice is choosing between

1. Fivey: Create Fivey, a person who will have a wellbeing level of 5

2. Tenny: Create Tenny, a person who will have a wellbeing level of 10

3. Nobody: Create no new life

This scenario contains one contingent person. It’s permissible for Alice to bring about Nobody, since the happiness of a contingent person does not generate a moral reason to create them, in accordance with the Bearer-Regarding View.

What about the permissibility of Outcome 1? Intuitively, our rationale might go something like this. Given a choice between creating Fivey or creating Tenny, Alice should create Tenny out of a consideration for her future child’s wellbeing. (This is in accordance with Impartiality. The future child’s identity does not matter; their wellbeing does.) So if Alice wants a child (that is, she does not opt for Nobody), and all else is held equal, she has an obligation to choose the outcome that is all things considered better for her child’s wellbeing. In our reasoning, we are effectively reducing the option set to $\{\text{Fivey, Tenny}\}$ and employing the necessary person distinction. Somebody (that is, Alice’s child) will exist regardless of the outcome chosen. We have a general outlook that whomever one day roams the Earth does so happily. To that end, we have a duty to choose what’s best for the de dicto child, which is Outcome 2.

2.1 The Nature of Moral Permissibility

To formalize our reasoning and generalize the results of the two-choice scenario back to the three-choice scenario, we will first have to take a detour through the nature of moral permissibility.

Moral permissibility, I contend, is the product of incommensurability¹⁰ among moral claims. There are a wide variety of reason-generating values: freedom, equality, wellbeing, self-preservation, promise-keeping, justice, truthfulness, fairness, etc. These values are qualitatively distinct. While we can speak of somebody being more truthful than another person, we can’t say that somebody is more truthful than another person is fair. They are on different scales! This makes it difficult—but not impossible—to interchange between the reasons generated by competing values. Consider the Burning Building scenario, inspired by Shelly Kagan (1989).

You have a wellbeing and altruism-related reason to go into the burning building and save the two people. But you also have a self-preservation related reason to not enter the burning building. Since the tradeoff between these reasons is imprecise, either is permissible.

On the other hand, moral impermissibility stems from the existence of an outcome that we have all-things-considered greater reason to bring about. This happens when one outcome instantiates the same value to a greater degree, or when the difference in magnitude between values is so stark that one dominates the other. In the Burning Building example, if you choose to run into the building, you have a moral obligation to open both doors; it is impermissible to rescue one person but not the other. At the point where you are inside the building and have already opened one door, you no longer have a self-preservation related reason to not open the other door. Since there is a wellbeing-related reason to save two people rather than one person, and no other countervailing considerations, you have an all-things-considered reason to save both people. What if the burning building scenario is modified such that hypothetically, a million people would be saved by entering the building and opening a door? It seems that the moral reason to save a million lives is so great that it outweighs one’s right to self-preservation. Alternatively, if the risk was small enough (say, the only personal cost was getting a mild first-degree burn), the reason from altruism dominates the reason from self-preservation. For one final example, take the values of beneficence and not causing harm. While there may not be a clear answer as to whether you should sacrifice one person to save two lives, you should probably sacrifice one person to save a billion lives.¹¹

Between Fivey (Outcome 1) and Tenny (Outcome 2), Alice has a welfare-related reason rooted in the interests of the necessary person to choose Tenny. Since we assume that all other morally relevant facts are held constant between outcomes, we can say that Outcome 2 morally dominates Outcome 1. There is no aspect where Outcome 1 has an advantage over Outcome 2. As a result, Alice has a moral obligation to choose Outcome 2, rendering Outcome 1 impermissible.

2.2 The Independence of Irrelevant Alternatives

To generalize this result back to the three-way scenario, we will employ the Independence of Irrelevant Alternatives Principle for Moral Reasons (IIAP).

IIAP is grounded in the Internal Aspects View of Outcome Goodness, a commonly accepted perspective that states that the degree of goodness of a possible world is determined independently of the alternative worlds available.

While Temkin argues that outcomes are compared with one another on a cardinal scale of goodness, in our account, outcomes are compared based on the moral reasons generated by each relevant moral ideal. The metric by which outcomes are scored still depend only on the outcome’s internal features. But since the reasons generated by different moral ideals can be incommensurate with one another, two outcomes can also be incommensurate with one another with regards to choiceworthiness. As discussed earlier, this gives rise to moral permissibility.

Moral permissibility obeys contraction consistency, the second half of the Independence of Irrelevant Alternatives.

An option is permissible if it is not dominated by any other option in the option set. Since the goodness of an outcome does not depend on other options, removing $O_3$ cannot decrease $O_1$'s choiceworthiness nor increase $O_2$'s choiceworthiness. As such, $O_1$ remains permissible. However, moral permissibility does not obey expansion consistency. A permissible option can become impermissible if a new option is added that dominates it.

Moral impermissibility works the other way, obeying expansion consistency but not contraction consistency.

An option is impermissible if it is dominated by another option in the option set. Since the dominating option ($O_2$) is not affected by an expansion of the option set, $O_1$ remains impermissible. Adding outcomes to our option set may also render $O_2$ impermissible, but it will not make $O_1$ more choiceworthy than $O_2$.

This is the piece we need to generalize the impermissibility of creating Fivey in the two-way scenario between Fivey and Tenny to the three-way scenario between Fivey, Tenny, and Nobody. Since $O_{Fivey}$ is impermissible when the option set is ${\{O_{Fivey},O_{Tenny}\}}$, it is still impermissible when the option set is $\{O_{\text{Fivey}}, O_{\text{Tenny}}, O_{\text{Nobody}}\}$ due to expansion consistency.

What if we start with $\{O_{\text{Fivey}}, O_{\text{Nobody}}\}$? Both options are permissible.¹² But permissibility is not guaranteed to survive expansions of the option set. The addition of $O_{Tenny}$, which is strictly more choiceworthy than $O_{Fivey}$, renders $O_{Fivey}$ impermissible.

3. CATALOG OF REASONS

Having applied our moral reasoning framework to the Non-Identity Problem, we will demonstrate its versatility and effectiveness on other population ethical issues. In this chapter, we will explore scenarios that involve tradeoffs between the moral claims generated by actual, necessary, and contingent people. Our aim is twofold. We will continue to clarify and fill in our theory, while stress testing its ability to provide intuitive answers to a range of procreation scenarios. The expectation is that our moral reasoning framework can remove the peculiarities from population ethics problems and simplify them to problems that conventional moral theories can handle.

Our choice of morally significant factors is responsible for the “last-mile” work in rendering moral judgment. In the following examples, we will appeal to moral principles from a variety of ethical perspectives in our account. Our catalog of reasons contains:

• Contingent person’s happiness (does not generate a reason)

• Contingent person’s suffering (equivalent to existing person’s suffering)

• Necessary person’s wellbeing (equivalent to existing person’s wellbeing)

• Reproductive freedoms

• Ability to fulfill a standard of care to potential children

These inputs to our moral framework will determine the balance of reasons for each scenario. They are flexible; others may use a different catalog of reasons and reach similar conclusions.

3.1 Second Child

The act of having more children can negatively impact the wellbeing of existing children. Despite this, it is generally held as acceptable to expand one’s family.

What reasons support the permissibility of having a second child? Neutral axiologies and person-affecting views consider the second child’s positive wellbeing to be an offsetting force to the first child’s decrease in wellbeing, but we’ve independently established that a contingent person’s wellbeing lacks moral force.

Our catalog of reasons contains three factors that may counterbalance Derek’s decrease in wellbeing. For one, Alice and Bob’s increase in wellbeing can generate a reason to create the second child. Alice and Bob might derive a strong sense of fulfillment and self-worth from having a large family. Or they might live in a community where the norm is for children to take care of their parents when they are elderly and Derek alone can’t provide for them in old age. In such circumstances, it’s possible for the second child to have a net positive impact on the wellbeing of actual people.

I’m somewhat doubtful that the broad permissibility of procreation stems from an affirmative response to the empirical question of whether having more children increases existing people’s wellbeing. While having children can bring a sense of fulfillment and purpose, it comes with significant monetary and time cost. Even if it’s in somebody’s interest to have children, their wellbeing and quality of life could take a huge hit—consider a mother who works long, grueling hours to support her child.¹³ In such cases, the additional responsibilities that come with another child actually decrease parental wellbeing. Alternatively, suppose that Alice and Bob find out about a surprise pregnancy and are ambivalent on whether to keep the second child (their wellbeing would not be affected). Even though they don’t have a strong desire to have another child, it still seems permissible for them to decide either way.

This can be explained by appealing to Alice and Bob’s procreative rights. The Supreme Court recognized the right to procreate as “one of the basic civil rights of man” and concluded that preventing individuals from producing offspring constituted a deprivation of a basic liberty (Skinner vs. Oklahoma). Ronald Dworkin argued that this right is grounded in individuals’ autonomy to shape their lives the way they see fit, especially around important decisions that are “touching the ultimate purpose and value of human life itself” (1994, p. 158). This right generates a moral reason in favor of whatever Alice and Bob decide that supersedes our welfare-related concerns for Derek.

Like all factors in our catalog of reasons, the right to procreate is not absolute. It is certainly weaker than the reason to not bring about a miserable life. We may want to qualify it further by setting forth a standard parents must meet in order to bring somebody into the world.

The precise standard of care is determined by both objective and contextual components. All children should be entitled to lives worth living, caring custodians, safe home conditions, and other basic needs. Beyond that, what parents owe to their child depends largely on their particular circumstances. Parents in wealthy countries should be able to support their child’s education through secondary school, but parents in developing countries typically have no corresponding duty. While it would still be unacceptable for those in extreme poverty to bring about children if they could not provide adequate nutrition, they do not have to provide the same quality of diet as a middle-class family in America, and so on.¹⁴

The Standard of Care Principle takes a contractualist perspective on the permissibility of having children, allowing individuals to exercise their reproductive rights only if they are capable of meeting mutually recognized and agreed upon obligations to their children. Though having the second child would reduce Derek’s wellbeing (because he wouldn’t be able to attend university), Alice and Bob are still able to meet the standard of care for both Derek and his future sibling; they do not have an obligation to pay for Derek’s university tuition. As such, their qualified reproductive freedoms outweigh Derek’s decrease in wellbeing.

In the previous example, Alice and Bob’s autonomy outweighed Derek’s wellbeing, allowing them to choose either option. (If Alice and Bob wanted a second child, it would be impermissible for us to prevent that out of a concern for Derek’s wellbeing). We will now illustrate two instances where reproductive freedoms are overruled by other factors.

When the wellbeing of tens of thousands of Greek soldiers and the fate of Greek civilization itself rest on Peleus’s future son, Peleus has a moral obligation to procreate. The inverse scenario can also occur.

Again, the wellbeing of the tens of thousands of Thebeans outweighs Laius’s right to procreate, this time rendering it morally impermissible for him to have a child.

My account of the general permissibility of procreation is but one approach, and the moral claims I employed are not crucial to my project. The purpose of this section is to show that we are able to add color and incorporate nuance into scenarios. Whereas previous theories operate on the level of bar graphs encoding basic population statistics (quantity and identity of people at various wellbeing levels), we can accommodate the kinds of reason-generating factors present in the broader moral discourse. Our normative approach provides a way to move beyond the welfare-centric model of population axiologies. Through our framework, we’ve centered our moral reasoning around the wellbeing, preferences, rights, and duties of actual people, couched in the familiar language and debate of conventional ethics.

3.2 Lives Not Worth Living

Any population theory must address the permissibility of creating individuals with lives not worth living. Although we have a very strong moral reason against doing so, it is not an absolute prohibition.

In this situation, it seems likely that the benefit to existing individuals surpasses the threshold required to justify the existential harm done to the Unhappy Altruist. Ultimately, the question of whether it is impermissible, permissible, or obligatory to create the Unhappy Altruist rests on the more general Harm-Beneficence Asymmetry—the difference in strengths between the reason to not cause a certain amount of harm and the reason to bestow an ‘equal’ amount of benefit.

That being said, existential benefits (creating new happy people) cannot ever offset existential harms—contingent people’s happiness does not have justificatory force. In Le Guin’s short story The Ones Who Walk Away from Omelas, Omelas is an idyllic city where the populace lives in prosperity and joy. However, their happiness comes at a cost—a single child is confined to a dark basement, living in a state of perpetual filth and misery. The child's suffering is necessary to maintain the happiness and prosperity of the city. All the citizens of Omelas learn about the existence of the child when they are of age, and they must reckon with whether it is ethical to live in such a society. Their question is slightly different than the one we will be asking: is it permissible to bring about the city of Omelas?

The natural consequence of our view is that absent external considerations, it is wrong to create Omelas.¹⁵ This is largely due to the No Reason principle that underlies our attitude towards contingent peoples’ wellbeing. This claim might strike some as unintuitive and excessively suffering-focused, but such a reaction may arise from a misunderstanding. We are not arguing that it is always wrong to sacrifice one person’s wellbeing for the greater good of society. (My catalog of reasons permits this, though yours may not). If the scenario was modified such that there are one million and one already existing people all at wellbeing level 0, and we have the option to move them into Omelas, where we would select one person to live in misery and one million people to live in bliss, there is a great moral reason to do so. As such, we are not condemning the city of Omelas itself. We are merely opposed to its creation if the entire populace, including the miserable child, wouldn’t have existed otherwise. To put it another way, we’re not in favor of creating Barren Rock (a world with no life) from Omelas, but we're also not in favor of creating Omelas from Barren Rock.

We present these cases mostly for the sake of theoretical completeness. It’s very rare to find a real-world situation where creating a miserable life has positive side-effects or a situation where one person’s happiness is dependent on one another’s misery. Since we have little information ahead of time about a person’s future quality of life, we can only reasonably expect a life to not be worth living if there’s a high risk of a severe genetic disorder or if the surrounding environment is extremely unfavorable (e.g. being born in a war-torn region). As such, it seems unlikely that there are actual scenarios where it’s permissible to create an expectedly miserable life.

3.3 Necessary People

Since a scenario containing necessary people will inevitably result in the creation of future lives, we must give them the same consideration as future actual people. Unsurprisingly, a version of the Second Child example where Derek is not yet realized yields the same results as the original.

It shouldn’t matter that the first child is yet to be born—we can infer from our options that they will eventually exist, which grants them standing to make moral claims upon us. It’s worth noting though that a mix of necessary people and contingent people in real-world scenarios is uncommon. The decision-making process typically resembles the original Second Child example, where the existence of each child is determined separately.

Since necessary people’s interests are just as strong as those of existing people, sometimes there is reason to trade off the wellbeing of existing people for the wellbeing of necessary people. Consider the following case.

Our framework treats the y-people and z-people under the unified banner of necessary people. Since the necessary people fare much better in Alternative B, we have a strong moral reason to choose it over Alternative A, even though the latter is slightly better for the presently existing people.

The necessary person distinction also provides the natural answer to the Actualist’s Dilemma.

Moral actualism says that an action’s goodness or badness depends on its effect on the wellbeing of actual people—those who presently exist or will exist in the real world. It should be noted that our definition of ‘actual people’ differs from that of actualists. While we use this term to refer to individuals within a specific decision-making context, actualists apply it in relation to reality itself. All actualists endorse the following.

Actualism is susceptible to normative variance: the permissibility of an action depends on whether it was chosen. Since the right action can only be determined after an action is chosen, it prevents the theory from offering moral guidance in choice scenarios. It also puts us in a strange dilemma. If the outcome of a@ is that Misery exists, then Misery, as a member of S@, matters while Moremisery does not. As such, we ought to have brought about Moremisery instead! But the same can be said if the outcome of a@ is that Moremisery exists. The upshot is that any option we choose is wrong, and the alternative is right. Hare remarks,

Under our framework, Misery and Moremisery coalesce into a single subject for moral reasoning. Since the option to bring about Misery is better for the necessary person, it is more choiceworthy. Our theory does not give rise to normative variance because the designations are narrowly tailored and made with the scenario as the reference frame; they cannot change based on its outcome.

Finally, we will explore situations where necessary people and contingent people face the prospect of unhappy lives.

The necessary person generates a moral reason to bring about Outcome 2, as it is significantly better for their wellbeing (-5 vs -10). However, the contingent people in Outcome 2 each generate a moral reason against that outcome, as it is not in their interest to be brought into existence with a life not worth living. Ultimately, the moral reason to avoid creating four moderately bad lives outweighs the moral reason to improve the wellbeing of the necessary person. That being said, we can imagine a case where it is permissible to bring a contingent person with a life not worth living into existence in order to improve a necessary person’s wellbeing:

Although determining the exact amount of additional beneficence necessary to warrant the creation of an unhappy life is challenging, a number does exist. That has been a prominent (though not necessary) theme in my account so far.

3.4 Repugnant Conclusion

We will now apply our moral reasoning framework to the seminal problem of population ethics: the repugnant conclusion. Derek Parfit’s formulation goes as follows.

The original Repugnant Conclusion is aimed at population axiologies, which attempt to rank possible worlds. It's not immediately clear how we should translate a question of betterness between hypothetical worlds into the questions that our moral framework seeks to address. We do not seek to establish a total ordering over hypothetical worlds; our objective is instead to provide guidance in choice scenarios that affect the future population. For our purposes, the Repugnant Conclusion is under-defined. It is not referring to a choice situation taking place in a reference world.

We will reframe the Repugnant Conclusion in three different contexts. The first interpretation asks whether God, at the dawn of the universe, should create World Z (the outcome with an unimaginably large population whose members all have lives barely worth living) over World A (the outcome with ten billion happy people). The second interpretation addresses whether a policymaker has a moral reason to guide a World A society towards World Z. The third interpretation addresses a situation where overpopulation naturally arises and brings the world towards World Z.

Interpretation 1: God

This scenario does not contain a base population whose interests we have to consider. Since at least one billion people will be created regardless of what outcome is chosen, there are one billion necessary people whose wellbeing we have a moral reason to maximize. The necessary people fare considerably better in Outcome 1 (where they have a welfare level of 10) than in Outcome 2 (where they have a welfare level of 1). There are also ninety-nine billion contingent people, whose wellbeing has no contributory value. In sum, the one billion necessary people generate an overwhelming reason to create Outcome 1, strong enough to outweigh God’s procreative autonomy and render the Repugnant Conclusion world impermissible.

Interpretation 2: Policy

The policy would negatively impact the wellbeing of the base population, which is a moral reason to not enact it. The policy would also create nine billion people with lives worth living, but again the happiness of contingent people does not generate a moral reason to bring them about. In the absence of other considerations, it is morally impermissible to pass the policy.

Interpretation 3: Natural

In this scenario, individuals are exercising their personal procreative freedoms to have children. The decrease in actual people’s wellbeing generates a moral reason to prevent the Repugnant Conclusion world from occurring. But what does the base population want? What if it’s in their interest to bring about World Z? They could hold an aesthetic preference for a bustling planet, place great importance on having large families, or follow a religion that emphasizes the importance of procreation to attaining salvation. In such a case, it’s theoretically possible for the repugnant conclusion to go through. People often forego their wellbeing for their interests, and that is to be respected. If some legitimate political mechanism establishes that everybody can continue to have as many children as they wish, World Z can permissibly occur.

There remains an asterisk to the permissibility of World Z: the Standard of Care Principle. Under this principle, if parents are unable to provide for their child’s basic needs, it is impermissible to bring them about. Since the wellbeing of future people would reach the level of just barely worth living, future parents may not be able to fulfill the standard of care required to beget offspring. This moral factor operates as a natural blocker to the Repugnant Conclusion in the spirit of the Critical Level Theory. If we include it in our catalog of reasons, World Z would be impermissible even in a world where it is the base population’s interest.

It is very unlikely that existing people would ever prefer World Z. But what if they’ve found themselves in an unoptimal Nash equilibrium? Perhaps they don’t want to live in the Repugnant Conclusion world, but since everybody else is procreating, they might as well also. This seems like a realistic avenue to overpopulation. In this case, government action that’s backed by a legitimate political mechanism can be taken to curtail everybody’s freedoms for the common good. China in the 1970s implemented a broad range of policies designed to control population growth. This included minimum ages for marriage, but the most infamous and restrictive was the One Child Policy. Such a policy, as part of an effort to prevent overpopulation, could be justified in the wellbeing of actual people.

Our moral framework naturally lends itself towards blocking the Repugnant Conclusion world. However, our three different formulations of the Repugnant Conclusion emphasize that we cannot discern from chart-form (the bar graphs used to represent a population’s wellbeing levels in two possible worlds) whether we have moral reason to bring about a population. People’s interests and designations play a crucial role in our moral judgments.

4. MISSING COUNTERPART RELATIONS

4.1 The Problem with Necessary People's Wellbeing

We have left unaddressed a major consideration for scenarios containing a mix of necessary and contingent people. In endorsing the Impartiality Principle, we have made the radical decision to not include a counterpart relation in our moral theory. Typically, the counterpart relation—the piece that marks individuals across worlds as the “same” for the purposes of moral reasoning—is just the identity relation. Other variations also exist. Christopher Meacham (2012) proposed a Saturating Counterpart Relation based on harm minimization that relaxes the personal identity constraint. Our theory’s absence of a counterpart relation enables necessary person-style reasoning, but it presents some complications when calculating the strength of the reason generated by the necessary person’s wellbeing. We will take up this problem now.

Let's first introduce some terminology. In this chapter, we will construct example scenarios as pairs of outcomes, where a population outcome is represented as an unordered set of wellbeing levels. For example, the outcome $\{3, 5\}$ contains two people, one with wellbeing level 3 and the other with wellbeing level 5. We will refer to individuals within populations by the outcome they belong to and their wellbeing level. $A_5$ indicates the person in Outcome A with wellbeing 5. Lastly, $R_{(x \text{ vs } y)}$ denotes the amount of moral reason to create somebody with the higher wellbeing level instead of the lower wellbeing level, or equivalently the moral reason to not create somebody with the lower wellbeing level instead of the higher wellbeing level.¹⁶ The 'vs' operator is unordered: $R_{(x \text{ vs } y)} = R_{(y \text{ vs } x)}$. For example, $R_{(6 \text{ vs } 5)}$ is the weight of the reason to create somebody with wellbeing level 6 instead of wellbeing level 5. When evaluating two worlds X and Y, where X: $\{5\}$ and Y: $\{6\}$, $R_{(6 \text{ vs } 5)}$ counts in favor of Y, and $R_{(6 \text{ vs } 5)}$ counts against X.

Reasoning about a necessary person’s wellbeing level has been relatively straightforward thus far. If Outcome A contains one person with wellbeing level 5, and Outcome B contains two people with wellbeing level 6, Outcome B is better with respect to the necessary person’s wellbeing. This scenario has one necessary person with wellbeing level 5 or 6 and one contingent person with wellbeing level 6.

Likewise, between Outcome A and D, it's clear that Outcome D is worse with respect to the necessary person's wellbeing.

But it is not always clear what the necessary person's wellbeing is. Suppose our choices looked like this.

We know that the necessary person has wellbeing level 5 in Outcome A, but what about Outcome E? What wellbeing level should we attribute to the necessary person, and what wellbeing level should we attribute to the contingent person? Depending on our answer, Outcome E could be better or worse with respect to the necessary person’s wellbeing.

One potential solution is to say that in situations where one outcome dominates the other with respect to the necessary person’s wellbeing (that is, the necessary person is guaranteed to be happier in the former outcome versus the latter), there is reason to bring about the former. Likewise, there is reason to not bring about an outcome where the necessary person is definitively worse off. And in situations like Outcome A vs Outcome E, the two worlds are incommensurate. I find this unappealing because it results in too much incommensurability and unduly curtails our theory’s predictive capacity. Consider the following scenario where Outcome F has a million people at wellbeing level 100 and one person at wellbeing level 4.

It seems that Outcome F is better for the necessary person. They will almost certainly have a wellbeing level of 100. In fact, there’s only a one-in-a-million chance that they end up with a lower wellbeing level than the one they would’ve had in Outcome A, and the difference between wellbeing level 5 and 4 is slight. Our reasoning here is reminiscent of the veil of ignorance, where individuals are asked to design a just society without knowing what position they will occupy. If somebody has to exist no matter what, and they do not know which life they will end up living, which world would they want to be born into? They would likely choose Outcome F. It is a low risk, high reward proposition—almost all of the probability mass lies under obtaining a wellbeing level of 100.

The necessary person designation draws upon a similar intuition: if some number of individuals have to exist no matter what, which world should we expect to be best for them? This question is difficult to resolve since oftentimes there is no fact of the matter on the necessary person’s wellbeing level. We can’t point to a particular person in both outcomes that is the necessary person; the necessary person is nothing more than an abstraction used for moral reasoning. However, the undefined nature of the necessary person's wellbeing level does not prevent us from incorporating it in our moral reasoning.

4.2 Reasoning Under Superposition

There is another domain in which one must reason about undefined properties: quantum mechanics. Quantum systems can be in multiple different states at once. When an electron is in a superposition of positions, there is no fact of the matter about its location; its exact position is undefined. In the double slit experiment, a beam of electrons is fired at a screen through a plate containing two openings, and each electron simultaneously takes both paths through the plate, exhibiting wave behavior. While the electron is traveling, its position is undefined. Nevertheless, we can mathematically represent its position using a wave function—an equation that contains the possible states and coefficients that determine the relative probabilities of finding the electron in each state (e.g. $\frac{1}{2}\lvert \text{Path A} \rangle + \frac{1}{2}\lvert \text{Path B} \rangle$). We can use this wavefunction to reason about the probabilities of the resulting position measurements .

We can apply the concept of superposition to necessary people’s wellbeing levels. I propose that the necessary person’s wellbeing level is in a superposition of all wellbeing levels in the population, and each possible wellbeing level generates a moral reason. Although these moral reasons may vary in strength, they are given equal consideration and weight because each population member has an equal claim to the necessary designation. Our reasoning should incorporate parity among all candidates.

Let’s apply the superposition principle to our original scenario and quantify the strength of the moral reason to choose Outcome B.

The reason generated by the wellbeing of the necessary person has strength $\frac{1}{2} R_{(6 \text{ vs } 5)} + \frac{1}{2} R_{(6 \text{ vs } 5)}$, which sums to $R_{(6 \text{ vs } 5)}$, the moral reason to create somebody with wellbeing level 6 instead of wellbeing level 5. This checks out. Conversely, the balance of reasons in choosing Outcome A comes out to $\frac{1}{2}(-R_{(6 \text{ vs } 5)}) + \frac{1}{2}(-R_{(6 \text{ vs } 5)}) = -R_{(6 \text{ vs } 5)}$. Note that we scale each R by the inverse of the population size to ensure that the quantity of Rs do not impact the strength of the moral reason. (If Outcome B contained another contingent person with wellbeing level 6, the reason generated by the wellbeing of the necessary person should remain $R_{(6 \text{ vs } 5)}$.)

Now let's modify the original scenario by increasing the wellbeing of a person in Outcome B to 7.

The reason generated by the necessary person in favor of Outcome C has strength $\frac{1}{2} R_{(6 \text{ vs } 5)} + \frac{1}{2} R_{(7 \text{ vs } 5)}$. We have a stronger reason to bring about Outcome C than Outcome B, and this aligns with our intuitions: Outcome C is better than Outcome B in terms of its population's wellbeing. However, the necessary person no longer has a defined wellbeing level. It is now in a superposition of two wellbeing levels. Note that it would be a mistake to interpret the $\frac{1}{2}$ coefficient as a fifty-fifty chance that the necessary person possesses a wellbeing level of 6 or 7 respectively, analogous to how it's mistaken to think that there's a fifty-fifty chance for an electron in a superposition of locations to be in one place or the other. The coefficient's purpose is to normalize the overall reason generated by the necessary person.

The previous two scenarios had an unambiguously better outcome for the necessary person's welfare. Let's revisit the situation where the necessary person could be better or worse off depending on whose wellbeing level they were linked to.

The balance of reasons to choose Outcome E comes out to $\frac{1}{2} R_{(7 \text{ vs } 5)} - \frac{1}{2} R_{(5 \text{ vs } 3)}$. If we assume that $R_{(5 \text{ vs } 3)} = R_{(7 \text{ vs } 5)}$, we should be indifferent to bringing about A or E from the necessary person's consideration. Alternatively, we can adopt a prioritarian view where there is greater reason to improve the wellbeing of the worst off. It's plausible that wellbeing is subject to diminishing returns. If so, $R_{(5 \text{ vs } 3)}$ has a greater magnitude than $R_{(7 \text{ vs } 5)}$, and so there is more moral reason to choose Outcome A. The prioritarian view complements a risk-averse attitude. We can view Outcome E as a gamble—there's a risk to the wellbeing of the person we're definitely creating, and it would be more bad to lose two units of wellbeing than good to gain two units of wellbeing.¹⁷

The prioritarian intuition is, in my view, insufficient to explain the imperative to avoid creating miserable lives.

Outcome J is more choiceworthy than Outcome A based off the wellbeing levels of the populations: $\frac{99}{100} R_{(20 \text{ vs } 2)} >> \frac{1}{100} R_{(2 \text{ vs } 1)}$. But it seems like Outcome K should be less choiceworthy than Outcome A because its population includes somebody with a miserable life. That said, a singular wellbeing delta of two separates K from J. The effect of diminishing marginal returns on welfare should not be strong enough to flip the choice-worthiness of outcomes by itself. What's at play is a qualitative difference between $R_{(2 \text{ vs } 1)}$ and $R_{(2 \text{ vs } -1)}$. It is the difference between an existential harm and an existential benefit. While having a life not worth living is a lamentable affair, having a life barely worth living is still a good thing. This causes $R_{(2 \text{ vs } -1)}$ to have a significantly greater magnitude than $R_{(2 \text{ vs } 20)}$. Despite this fact, it's unclear if Outcome K is worse than Outcome A for the necessary person. While $R_{(2 \text{ vs } -1)} > R_{(2 \text{ vs } 20)}$, is it the case that $\frac{1}{100} R_{(2 \text{ vs } -1)} >> \frac{99}{100} R_{(2 \text{ vs } 20)}$? A 100x multiplier is in effect. Instead, I contend that Outcome K's impermissibility is rooted in a welfare-related reason generated by contingent people. We have thus far ignored contingent people in our moral calculus because their happiness does not count in favor of an outcome. But their suffering does count against outcomes. Each contingent person generates the following moral reason: $\frac{99}{100} R_{(20)} - \frac{1}{100} R_{(-1)}$. Since $R_{(20)}$ (the moral reason to bring into existence somebody with wellbeing level 20) is 0, and there are ninety-nine contingent people, as a whole they contribute $-\frac{99}{100} R_{(-1)}$ to Outcome K's balance of reasons. When the reasons generated by the contingent and necessary people are combined, the overall balance of reasons for Outcome K is approximately $R_{(20 \text{ vs } 2)} - R_{(-1)}$. There is much more reason to avoid creating a miserable life than there is to essentially increase somebody's wellbeing level from 2 to 20, rendering Outcome K impermissible.

So far we have only explored cases with one necessary person. Our framework also generalizes to larger populations.

A separate moral reason is generated for each member of the cartesian product of the two sets and normalized to one so that the overall strength does not increase with the number of reason components. The moral reason to choose Outcome L is has strength

$\frac{1}{3}(R_{(5 \text{ vs } 4)} + R_{(5 \text{ vs } 5)} - R_{(5 \text{ vs } 6)}) + \frac{1}{3}(R_{(6 \text{ vs } 4)} + R_{(6 \text{ vs } 5)} + R_{(6 \text{ vs } 6)})$ while the moral reason to choose Outcome M has the opposite strength. Each reason component corresponds to a wellbeing delta between the two outcomes. We can reduce the equation to $\frac{1}{3}(R_{(5 \text{ vs } 4)} + R_{(6 \text{ vs } 4)})$, which is positive, indicating that Outcome L is more choiceworthy from the perspective of the necessary person.

Writing out these equations can be cumbersome, especially when dealing with larger populations. Fortunately, it’s fairly easy to tell from inspection which outcome we have more moral reason to bring about and so this long-form reasoning almost never needs to be employed. In the above scenario, Outcome M can only make the necessary peoples’ wellbeing worse as $M_4$ reduces their aggregate wellbeing level. This thorough account of the formalism is mostly for the sake of theoretical completeness. In reality, procreation is generally permissible, and a sort of Averagism can approximate the extended calculus in most scenarios.

4.3 Averagism and Axiology

Our efforts to circumvent the counterpart relation bears some resemblance to the field of population axiology. We have taken up the task of determining which world we have the most reason to bring about given only the wellbeing levels of their populations. In fact, when we are considering the moral reason generated by necessary people, our account operates similarly to a prioritarian version of Averagism. We seek to choose the outcome that maximizes the wellbeing for a minimal number of people. This is extensionally equivalent to maximizing the average wellbeing. Almost all philosophers have rejected Averagism, for good reason. While we cannot address all of its objections, we will discuss a few to show that our account salvages some of Averagism’s appeal while comfortably avoiding its pitfalls.

It follows from the objective of maximizing necessary peoples’ wellbeing that Outcome P is more choiceworthy than Outcome Q.

Most people believe that it’s counterintuitive for one person experiencing a slightly higher welfare level to be preferable to many people experiencing a high welfare level. But such a complaint only applies to Averagism as a population axiology. We make the limited claim that Outcome P is (marginally) better for the necessary person, not the claim that it’s all things considered more choiceworthy than Outcome Q. The small impact on the necessary person's well-being means that other considerations, like the desire of billions of parents to have children, can easily outweigh it.

Averagism entails a more problematic conclusion: it is sometimes better to add people with miserable lives. Adding miserable lives will improve the average wellbeing if the average wellbeing is currently even lower than the wellbeing of the miserable people.

It is obviously impermissible to have children who will grow up to suffer so that we can increase the average wellbeing. Fortunately, our semblance of averagism does not permit adding people with lives not worth living. This is because while Averagism operates on raw wellbeing levels, our theory operates on the strength of moral reasons to bring about people with certain wellbeing levels. As a result, we factor in the asymmetry between conferring a miserable life and conferring a happy life—there is much more moral reason to not do the former than there is to do the latter. This is illustrated by the following scenario.

Under Averagism, S and T are equally choiceworthy—$S_{(-5)}$ "cancels out" $S_5$. But under our theory, $R_{(-5 \text{ vs } 0)} >> R_{(5 \text{ vs } 0)}$ and so there is a much greater reason to bring about T with regards to the interests of the necessary person. As discussed previously, we accommodate the prioritarian intuition through the same mechanism. (The Averagist must rank A: $\{5\}$ as equally choiceworthy with D: $\{3, 7\}$.)¹⁸

The last problem with Averagism that we will address is the violation of the Mere Addition Principle.

Since the additional person in B brings down the world’s average wellbeing, Averagism says that B is less choiceworthy than A. Greaves remarks,

Her framing of the Mere Addition Principle suggests that what we are doing is adding one person with a great life to an extant population. In other words, B is the result of adding a contingent person to A. In such a context, our theory says that B is not (intrinsically) worse than A—they are equally choiceworthy with regards to the wellbeing of the contingent person, though other factors may swing the balance one way or another.

We can alternatively construe the Mere Addition Principle in a way that involves the wellbeing of necessary people.

Now Outcome B is worse than Outcome A on the basis of the wellbeing of necessary people. Though this construal seems unintuitive (how is it different from the previous one?), I reckon that any prima facie puzzlement is due to our tendency to pair the A100s with the B100s based on index and conceptualize B99 as an “extra” person tacked onto the world. But remember that the outcomes are unordered sets—it would be equally valid to write B: $\{100, 99, 100, 100\}$ and so we cannot see B99 as a ‘mere addition.’

In reality, it’s rare for the creation of multiple individuals to be commingled in the same decision. In most cases, procreation decisions are made independently of one another. When both outcomes contain only one person, or the same number of people, we don’t need a counterpart mapping between the two worlds or a framework to adjudicate the wellbeings of necessary and contingent people. It’s only when our options involve creating two irreducible, different-sized packages of people that we face this problem. I contend that the standard reading of the Mere Addition Principle views B99 as an addition to an already-existing population, not as an inseparable part of the outcome. Our handling of the standard reading is unobjectionable, whereas our handling of the latter reading—which does imply that B is worse than A in one respect—is prima facie puzzling only because B99 is not treated as a ‘mere addition’. It is difficult to come up with realistic examples of such a situation where the existences of a hundred people are inextricably linked.

CONCLUSION

In our search for an acceptable population theory, we have strayed from the voluminous discourse on proposals, refinements, and defenses of various population axiologies. Instead, we present a moral reasoning framework for judging the choice-worthiness of procreation scenarios. Drawing upon the Impartiality Principle and Frick’s Bearer-Regarding View, we distinguish between three types of people in these scenarios for the purposes of moral reasoning. This approach allows us to work within conventional ethics and remain flexible towards the normative theory we use to provide moral claims. The contingent person designation encapsulates our intuitions around the No Reason Principle, setting us up for a satisfactory response to the Procreation Asymmetry and the Person-Affecting Intuition, while the necessary person designation handles the Non-Identity Problem. Much of our framework’s novelty and functionality is owed to the missing counterpart relation. While the Neutral Axiologies also work without a counterpart relation, we are able to employ factors beyond the intrinsic welfare of population members in our decision-making process. This requires postulating some additional principles to resolve ambiguous cases that arise when the wellbeing level is undefined due to the lack of a mapping across worlds.

My thesis sketches a possible course for handling perplexing population ethics problems while still preserving our commonsensical intuitions around procreation. Unfortunately, due to the breadth of subject matter covered, we have only been able to touch upon certain aspects of the theory that deserve a more thorough discussion. Nevertheless, my hope is that the central framework we’ve advanced has been compelling and that future work can further develop our non-axiological approach to population ethics problems.

BIBLIOGRAPHY

Arrhenius, Gustaf. Future Generations: A Challenge for Moral Theory. 2000.

Bader, Ralf M. "Person Affecting Utilitarianism." The Oxford Handbook of Population Ethics, edited by Gustaf Arrhenius et al., Oxford University Press, 2022.

Bennett, Jonathan. "On Maximizing Happiness." Obligations to Future Generations, edited by R.I. Sikora and Brian Barry, White Horse Press, 1978.

Chappell, Richard Yetter. "Value Receptacles." Noûs, vol. 49, no. 2, 15 Apr. 2013, pp. 322–332.

Douglas, William Orville, and Supreme Court Of The United States. U.S. Reports: Skinner v. Oklahoma, 316 U.S. 535. 1941. Retrieved from the Library of Congress.

Dworkin, Ronald. Life's Dominion: An Argument about Abortion and Euthanasia. Alfred A. Knopf, 1994.

Frick, Johann David. "Making People Happy, Not Making Happy People": A Defense of the Asymmetry Intuition in Population Ethics. 2014.

Greaves, Hilary. "Population Axiology." Philosophy Compass, vol. 12, no. 11, Nov. 2017.

Hare, Caspar. "Voices from Another World: Must We Respect the Interests of People Who Do Not, and Will Never, Exist?" Ethics, vol. 117, no. 3, Apr. 2007, pp. 498–523. Accessed 9 Dec. 2020.

Huemer, Michael. "In Defence of Repugnance." Mind, vol. 117, no. 468, 1 Oct. 2008, pp. 899–933.

---. "Transitivity, Comparative Value, and the Methods of Ethics." Ethics, vol. 123, no. 2, 2013.

Kagan, Shelly. The Limits of Morality. Oxford University Press, 1989.

Le Guin, Ursula. The Ones Who Walk Away from Omelas. 1973. Mankato, Minn. Creative Education, 1993.

Meacham, Christopher J. G. "Person-Affecting Views and Saturating Counterpart Relations." Philosophical Studies, vol. 158, no. 2, Mar. 2012, pp. 257–287.

Moore, George Edward. Principia Ethica. Cambridge University Press, 1903.

Narveson, Jan. "Moral Problems of Population." The Monist, vol. 57, no. 1, 1973.

Parfit, Derek. Reasons and Persons. Oxford University Press, 1984.

Rachels, Stuart. "Is It Good to Make Happy People?" Bioethics, vol. 12, no. 2, 1998.

Roberto, Fumagalli. "Eliminating 'Life Worth Living.'" Philosophical Studies, vol. 175, no. 3, 2018.

Singer, Peter. Practical Ethics. Cambridge University Press, 1993.

Temkin, Larry S. Rethinking the Good: Moral Ideals and the Nature of Practical Reasoning. Oxford University Press, 2012.