The Rules of Rescue

Front matter

Cost, Distance, and Effective Altruism

THERON PUMMER

Cost, Distance, and Effective Altruism

Contents

Acknowledgments ix

Introduction: Non-Consequentialism and Methodology | l; 2.

Requirements to Rescue and Permissions Not to 19 Numbers Count 39 vii

Introduction

NON-CONSEQUENTIALISM AND METHODOLOGY

I’ it wrong to let an innocent stranger die when you can rescue them at no cost? What if you could save either a few strangers or many others, at no cost to yourself either way—would it be wrong to save the few? Or what if, to save the lives of two strangers, you had to sacrifice your life, limbs, time, money, or personal projects—would it be wrong not to save them at such great costs to yourself? Is it ever wrong to save a few strangers at great cost to yourself rather than save many others at no greater cost to yourself?

These questions are clearly relevant to emergency rescue cases in which you can save more or fewer nearby people who are drowning, imperiled by hurtling boulders, or trapped in burning buildings. They are also relevant to constantly occurring cases in which you can use time or money to help distant people in need of food, shelter, or medical care. For example, by donating to charity, you can prevent distant strangers from dying of malaria. Depending on where you donate, you can provide more help or less.

This is a book about the moral reasons and requirements to use time, money, and other resources in ways that help others. I take it

The Rules of Rescue. Theron Pummer, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/0s0/9780190884147.003.0001 that you are not always morally required to provide the most help possible. First, there are moral constraints on doing so. You are not morally required to harvest an innocent non-consenting person’s organs to save two others, just whenever this is what would help the most. Youre morally required not to do that. Second, there are moral permissions not to help the most. You are not morally required to sacrifice your life to save two strangers, just whenever this is what would help the most. Had Edgar Wilson known he was going to drown in the River Thames while rescuing two boys, it would have been permissible not to jump in—such heroism is praiseworthy but beyond what is required.’

In accepting constraints and permissions, I’m rejecting act consequentialism, the view that you are always morally required to do whichever available act has the best outcome (I am likewise rejecting act utilitarianism, the view that you are always morally required to do whichever available act brings about the most well-being across all individuals). There are other versions of consequentialism, such as rule consequentialism. But since I doubt these other versions of consequentialism do much better in terms of accommodating constraints and permissions, I am drawn to zonconsequentialism instead.’

Non-consequentialism isn’t the view that consequences do not matter. It’s just the view that consequences are not all that matter. Clearly, there are strong moral reasons to prevent significant harms from befalling strangers. It is morally wrong to allow these harms to occur when you could prevent them at no cost. Consequentialism does not have a monopoly on beneficence.

What is more, plausible versions of non-consequentialism hold that, in a range of cases, it is morally wrong not to prevent significant harms like suffering and death even when doing so involves incurring substantial personal costs. Plausible versions of non-consequentialism also hold that, in a range of cases, it is morally wrong to help a lesser number of strangers rather than comparably help a greater number of other strangers. I argue that these versions of non-consequentialism are not restricted to nearby emergency rescues but also apply to using time and money to help distant strangers (by volunteering, donating to charity, and making a difference with your career).

In earlier work, I argued that it is—in a significant range of cases—wrong to donate to less cost-effective charities rather than to more cost-effective charities, even when it is morally permissible not to donate (e.g,, it’s wrong to donate $6000 to a charity that on average saves one life per donation of this size, rather than donate $6000 to a charity that on average saves two lives per donation of this size, even when it’s morally permissible not to donate this sum of money).? However, my argument took two important claims for granted: first, that it is wrong to save a lesser number of strangers rather than a greater number of other strangers and, second, that there are no relevant disanalogies between nearby rescue and distant rescue via charitable donation. In this book, I strengthen my previous argument by defending (qualified versions of) these claims. Moreover, it may seem paradoxical that it could be wrong to provide some help rather than more help when it’s permissible to provide zo help. I further strengthen my previous argument by dissolving apparent paradoxes generated by this claim.

Overall, this book is my attempt to articulate why each of us should be thinking about how to help more, whether by shifting our career, volunteering, or donating.‘ It is also an attempt to provide a defensible non-consequentialist picture of the moral reasons and requirements to engage in effective altruism, construed as the project of using time, money, and other resources to help others the most.’ Clearly, act consequentialism supports engaging in effective altruism. But—contrary to the impression shared by many effective altruists and their critics alike—act consequentialism is only one of many very different moral views that support effective altruism. This matters for two reasons. First, most people reject act consequentialism. Whether or not effective altruism can find alternative philosophical support makes a difference to how many would be willing to join the cause. Second, act consequentialism is implausible. Effective altruism would lack plausible support if it lacked alternative support.

In providing a non-consequentialist picture of the moral reasons to use time, money, and other resources to help others the most, I do not present a fully formulated version of non-consequentialism as an alternative to act consequentialism’s criterion of permissibility. Instead, I present my arguments within a theoretically minimal framework. This gives the arguments greater flexibility, allowing readers to fit them into their own favorite theories.

At various points throughout the book, I do endorse specific versions of non-consequentialism that I lean toward. But often this won't impact my main arguments or conclusions. For example, in Chapter 1, I defend “autonomy-based permitting reasons” in addition to “cost-based permitting reasons.” And in Chapter 2, I suggest you're not required to toss a coin to determine which of two strangers to save (it is permissible just to save either, without first tossing a coin). Other non-consequentialists may reject autonomy-based permitting reasons or accept a requirement to toss a coin to determine which of two strangers to save. Readers who favor different versions of non-consequentialism can feel free to transpose my arguments from the point of departure, seeing where they lead. Many such variations can occur within the same broad picture according to which there are strong moral reasons to use time and money to help others the most.

Throughout the book, I appeal to intuitive claims about cases. Take the claim that you aren’t morally required to harvest an innocent non-consenting person’s organs just whenever this would have the best outcome. This is an intuitive claim in that it, considered independently, seems correct. We can find a claim intuitive without accepting it. For instance, an act consequentialist can agree that the above claim about the organs case is intuitive, even though they do not accept it. But it is a price of accepting act consequentialism that it is inconsistent with this intuitive claim (it is a further question whether it is worth paying the price to avoid other, greater ones).

Not all intuitive claims should be accommodated. Some intuitive claims can be debunked, particularly if they reflect biases or misunderstandings. Some intuitive claims are inconsistent with others so that at least one must be incorrect. Some are more intuitive than others. And some are found intuitive by a greater proportion of those who understand them.

What’s more, accommodating intuitive claims about cases is not the only thing that matters when deciding between competing moral views. Suppose two views equally capture intuitive claims about cases, but whereas one view just consists of a list of these many claims, the other derives them from a few basic principles. Parsimony favors the latter view, with fewer unexplained elements. Or suppose both views derive all these intuitive claims from the same number of basic principles, but one view’s principles are more intuitive as principles than the other view’s principles. Intuitiveness at this more general level matters too.

Readers need not accept all the claims I’ve just made about intuitions. I invite those who disagree to approach the book with their own favored methodologies. The ways I rely on intuitive claims throughout the book are compatible with most approaches that aren’t entirely dismissive of intuitions. And substantive moral views simply cannot be defended without relying on intuitions or judgments at some level.°

Many of the cases to which I appeal are imagined and thinly described. Consider the following case from Chapter 1:

Costly Rescue: A stranger faces a deadly threat. You can either do nothing, allowing them to die, or you can, at great cost to yourself, save their life.

When considering cases like this, complications beyond those mentioned are to be set aside. For example, the person you can rescue is an innocent, not a villain. They have a serious interest in staying alive. Saving them does not involve harming, lying, or stealing. Beyond losing your legs, there are no negative side effects of saving them. You are aware of all these details. We can suppose, for the sake of concrete illustration, that a boulder is hurtling toward the stranger. If you do nothing, it will crush them to death. You can save this person’s life by putting your legs in the path of the boulder. There is no other way they can be saved.

Even when supplemented with this sort of concrete illustration, Costly Rescue remains rather stylized. I use clean cases like this for the same reason scientists use controlled experiments. In the real world, there are a lot of potentially relevant factors buzzing about. We need to minimize the “noise” introduced by factors other than those we're interested in testing. If we fail to control for confounding factors, it will remain unclear whether our experimental results are explained by the factor we were meant to be testing or instead by one or more of the factors correlated with it.

Were Costly Rescue more realistic, it would involve uncertainty about whether attempting to save the stranger by putting your legs in the path of the boulder would backfire, resulting in you or bystanders being crushed to death (along with the stranger). Such uncertainties could well make it seem morally permissible not to put your legs in the boulder’s path to protect the stranger. But what we are trying to test with Costly Rescue is just whether you are morally required to incur a great cost (like the loss of your legs) when that would prevent a much larger harm (like death) from befalling a stranger. To consider the case properly, we need to make it somewhat unrealistic. We need to suppose you're sufficiently certain that putting your legs in the path of the boulder would save the stranger’s life, crush your legs, and have no further “noteworthy” effects. Once we have controlled for confounding factors in this way, it may still seem you're morally permitted not to put your legs in the boulder’s path to save the stranger. Then what we are finding intuitive is that a great cost to you can permit you not to save a stranger’s life, even when it’s much smaller than the harm of death. Other imagined cases used in this book are similarly controlled (see the Glossary of Cases).

Despite the advantages of using imagined cases, there are some dangers. One sort of danger is that an imagined case, used in an attempt to control for confounding factors in the real world, ends up introducing its own confounding factors. For example, many imagined cases in ethics involve victims being put in harm’s way by diabolical agents, when there is an accidental or non-agential source of harm in purportedly analogous real-world cases. Many imagined cases resort to magic or science fiction, when this may blur our picture of what potentially relevant factors are present. But these dangers aren’t reasons not to use imagined cases. They are reasons to take care when doing so. When using imagined cases, I try to avoid introducing confounds, and I try to make it easy for readers to get an imaginative grasp of what potentially relevant factors are present. Even if some of my cases are complicated, they are at least understandable.

In addition to these epistemic dangers of using imagined cases, there are dangers of practical inapplicability. The way two factors interact in the absence of other factors (in a “laboratory” setting) may differ from the way they interact in the presence of other factors (in “nature”). This can make it far from straightforward how to take the results of our controlled experiments and apply them to a messy world. Relatedly, in attempting to apply the lessons of thought experiments to the real world, we need to check the empirical details to make sure there is a reasonably good analogy between the designed case and the naturally occurring case in question. Ethicists may be so accustomed to thinking about ethically difficult problems in theory that they overlook ethically easy solutions in practice—there’s no need to shove anyone in front of a runaway trolley to save many others, when you could instead just press the emergency stop button! Again, the remedy isn’t to abandon imagined cases but to use them with care.

In sum, I frequently appeal to intuitive claims about imagined cases (as well as appeal to intuitive general claims or principles). I have addressed some concerns about the use of such cases, though I am afraid my brief remarks here won’t convince those who are skeptical. I hope skeptics of the method of imagined cases can nonetheless get something out of the book.’

In a nutshell, then, here’s the plan. Chapter 1 sets up the book’s basic framework of reasons and how they serve to determine whether an act is required or permissible. Chapter 2 defends the intuitive claim that it is wrong to save a lesser number of strangers rather than a greater number of (at least twice as many) other strangers. Chapters 3 and 4 defend the claim that it can be wrong to save the lesser number rather than the greater number, even when it’s permissible to save no one. These chapters dissolve two apparent paradoxes that are generated by this claim (Chapter 3 shows how the claim does not imply that if you are not going to save the greater number, then you are required to save no one; Chapter 4 shows how the claim allows that you can be praiseworthy for saving the lesser number, even when it is wrong to do so). The first four chapters defend a set of claims about the ethics of rescue in nearby emergencies. The remaining four chapters explore the extent to which these “rules of rescue” carry over to helping distant strangers by using time and money (assuming these resources rightfully belong to you). Chapter 5 shows how several differences between nearby rescue cases and distant rescue cases fail to capture the sense that it’s wrong not to help in the former but permissible not to help in the latter.

Chapter 6 develops a view of how the frequency of opportunities to help over the course of your life affects reasons (not) to help on specific occasions. Chapter 7 turns to special connections, their ability to make it permissible or even required to save fewer people rather than more, and their interaction with considerations of frequency. Chapter 8 draws together the book’s main argument: that core claims from Chapters 1-4, about requirements to help the most and permissions not to help the most, carry over to a significant range of real-world cases in which you can help using time and money. (This chapter includes a discussion of how to modify the book’s main argument, when we drop the assumption that the time, money, and other resources in your possession rightfully belong to you.) It is argued that a significant proportion of us are required either to be effective altruists or else to provide no less help over our lives than we would have done if we did the minimum required as effective altruists.

More detailed chapter summaries can be found below.

Chapter 1: Requirements to Rescue and Permissions Not to

I introduce the book’s basic framework of requiring reasons and permitting reasons. Requiring reasons serve to make acts required. Permitting reasons serve to make acts permissible (without serving to make acts required) by serving to prevent requiring reasons from making acts wrong. An act is required—wrong not to do—when there is most requiring reason overall to do it and no sufficiently strong permitting reason not to do it. A permitting reason is a kind of “defeater”: when it’s sufficiently strong, it prevents the balance of requiring reasons from making an act wrong. I then proceed to defend two main claims: first, that there are (strong) requiring reasons to rescue strangers from (large) harms and, second, that there are cost-based and autonomy-based permitting reasons not to rescue, which can make it permissible not to act in accord with the balance of requiring reasons to rescue. Taken together, these two claims explain how it can be wrong not to help in cases in which rescuing a stranger is costless to you (Costless Rescue) but permissible not to help in cases in which rescuing comes at great cost to you (Costly Rescue). Finally, I discuss the compatibility of these two claims with three competing views of rights to be rescued—the no-rights view, the cost-sensitive view, and the cost-insensitive view.

Chapter 2: Numbers Count

I defend a view about how the numbers count. First, I claim that (in Costless No-Conflict) it is wrong to save one stranger’s life at no cost to yourself when you can instead save both this stranger’s life and another’s at no cost to yourself. I then defend the claim that (in Costless Conflict) it is wrong to save one stranger’s life at no cost to yourself when you can instead save the lives of two other strangers at no cost to yourself. I defend a view of how the numbers count according to which there is nonetheless an important difference between conflict cases and no-conflict cases. While there is more requiring reason overall to save 4 and C than there is to save A (as in Costless No-Conflict) and more requiring reason overall to save Band C than there is to save A (as in Costless Conflict), there is an individualist permitting reason to save A in the latter case but not in the former case. While this individualist permitting reason to save A is not sufficiently strong to make it permissible to save 4 in this case, there is a sufficiently strong individualist permitting reason to save 4’s life instead of saving B’s life and C’s finger. I further show that this view is compatible with the plausible claim that (in Bored v. Joyful) you are not required to save one stranger’s life over another’s just because the first stranger’s remaining life would be of much higher quality than the second stranger’s would be. And I show that this view of how the numbers count does not imply the counterintuitive “fully aggregative” claim that there is more requiring reason overall to save billions of people from very mild pain than there is to save one person from very intense agony.

Chapter 3: The All or Nothing Problem

I argue that (in Costly No-Conflict) it is wrong to save one stranger’s life at great cost to yourself when you can instead save both this stranger’s life and another’s at the same great cost to yourself. This is so even when it is permissible to save neither stranger. There is a natural explanation of these claims. There is significantly more requiring reason overall to save the greater number than there is to do either alternative. But while there is a sufficiently strong permitting reason to do nothing, there is no sufficiently strong permitting reason to save the lesser number. I then turn to the “all or nothing problem” that the plausible claim that it is wrong to save the lesser number and the plausible claim that it is permissible to do nothing may together seem to imply the implausible claim that, if you are not going to save the greater number, you are required to do nothing. I present four possible solutions to this problem and reject all but one. According to my solution, although it is plain old wrong to save just one stranger, it can be conditionally permissible to do so—you may save just one given that you are not going to save both. I argue that my solution to the problem similarly applies to conflict cases (like Costly Conflict) in which you can save one stranger’s life at great cost to yourself or instead save the lives of two other strangers at the same great cost to yourself.

Chapter 4: Praiseworthiness

I consider a new objection to the claim that (in cases like Costly No- Conflict and Costly Conflict) it is permissible to do nothing, wrong to save the lesser number, and permissible to save the greater number. The objection is that, if it is wrong to save the lesser number (and you lack an excuse for doing so), then you are overall blameworthy for saving the lesser number. But this is contrary to the intuition that (at least in Costly Conflict) you are worthy of praise, not blame, for heroically saving a stranger. In response, I argue that not all unexcused wrong acts are overall blameworthy, even if all are blameworthy to some extent. Some unexcused wrong acts are less responsive to the balance of requiring reasons than is required and yet seem overall praiseworthy (saving the lesser number in Hand). Other unexcused wrong acts are more responsive to the balance of requiring reasons than is required and yet seem overall praiseworthy (saving the lesser number in Costly Conflict and in 10 Plus Conflict). Not only can you be praiseworthy overall for performing such wrong acts, but you can be more praiseworthy overall for performing such wrong acts than you would be for performing permissible alternatives.

Chapter 5: Distant Rescues

In Pond, you can save a drowning stranger at the cost of your new clothes. In Charity, you can donate to a malaria charity that saves on average one life for every $3000 they receive. In Pond v. Charity, you can either save a drowning stranger at the cost of your $6000 or instead donate this money to a malaria charity that saves on average one life for every $3000 they receive. Some argue that, due to one or more differences between cases like Pond and cases like Charity, it is wrong not to help in cases like Pond, permissible not to donate in cases like Charity, and permissible (if not also required) to help the drowning stranger in cases like Pond v. Charity. In this chapter, I focus on differences with respect to distance, salience, uniqueness, injustice, and diffusion. On the basis of several “clean” cases—which bracket various complications—I argue that these factors do not make the relevant sort of moral difference (“risky diffusion” is an exception). Whether taken individually or in combination, such factors would not make it the case that, while it is wrong not to help in cases like Pond, it is permissible not to donate in cases like Charity. Nor would they make it the case that it is permissible to help the drowning stranger in cases like Pond v. Charity.

Chapter 6: Frequent Rescues

While cases like Pond are rare, cases like Charity are very frequent. To isolate frequency, I examine Frequent NearPlus, a case in which individual opportunities to help are like Pond with respect to distance, salience, uniqueness, injustice, and diffusion—yet arise very frequently. In this imagined case, it seems you are not required to take every individual opportunity to help, even if you are required to take some. I develop a view of requiring reasons and permitting reasons that explains this plausible claim. The idea behind this view is that “lifetime” features can amplify (cost-based and autonomy-based) permitting reasons not to save strangers. Crucially, it is the presence of permitting reasons that does the explanatory work, rather than the absence of requiring reasons. In Frequent NearPlus, there is an equally strong requiring reason to save each stranger. I then turn to a variant of this case (Rare/ Frequent NearPlus), asking whether it can be permissible not to respond to a frequently occurring opportunity to help even when it would be wrong not to respond to an otherwise similar rarely occurring opportunity to help. I argue that, while frequency cannot itself make this kind of moral difference, considerations of cost and autonomy correlated with frequency can. Such considerations make it permissible to save the lesser number in a range of cases like Pond v. Charity.

Chapter 7: Special Connections

The focus of this chapter is whether, when, and how special connections—including personal relationships, projects, and commitments—enhance requiring reasons or permitting reasons to save others. First, I look at some of the ways in which reasons can be enhanced by special connections, in accord with the kinds and degrees of these connections. Cases like Pond and Charity can differ with respect to special connections, as can different Pond-like rescues and different Charity-like rescues. Take a familiar real-world case in which you can donate to a charity to which you have a special connection or donate the same amount to a more cost-effective charity to which you have no special connection. Next, I show how lifetime features can amplify otherwise insufficiently strong permitting and requiring reasons to save a lesser number of people to whom you're specially connected over a greater number of strangers, making it permissible or even required to save the lesser number (I use Frequent Friend v. Strangers to show this). Finally, I distinguish between responsibly acquired special connections and non-responsibly acquired special connections and show how the former (but not the latter) can increase the cost you are required to incur in helping others over the course of your life.