Introduction
Suppose you want to do as much good as possible. What should you do? According to members of the effective altruism movement—which has produced much of the thinking on this issue and counts several moral philosophers as its key protagonists—we should prioritise among the world’s problems by assessing their scale, solvability, and neglectedness. Once we’ve done this, the three top priorities, not necessarily in this order, are (1) aiding the world’s poorest people by providing life-saving medical treatments or alleviating poverty itself, (2) preventing global catastrophic risks, such as those posed by nuclear war or rogue artificial intelligence, and (3) ending factory farming.1(Singer, 2015) and (MacAskill, 2015) are the two original books advocating effective altruism. For a more recent articles setting out and defending effective altruism, see (MacAskill, 2018) and (Pummer and MacAskill, 2019).
These claims are both plausible and striking. If correct, they should prompt a stark revision of how we approach our altruistic activities. However, the project of determining how to do the most good—as opposed to say, whether we should do the most good—has only recently, within the last ten years, become the subject of serious academic attention. Many key claims have not yet been carefully scrutinised. This is a cause for concern: are effective altruists doing good badly?
In this thesis, I critique and develop some of the latest claims about how individuals can do the most good. I do this in three areas: the value of saving lives (preventing premature deaths), how best to improve lives (making people happier during their lives), and cause prioritisation methodology (frameworks for determining which problems are the highest priorities).2I note my usage of ‘improve lives’ is arguably non-standard. Ordinarily, improving lives would refer to increasing individuals’ well-being during their lives, rather than, less ecumenically, their happiness. As I focus specifically on happiness, where such specificity is required, as opposed to any other good that may constituent well-being (in whole or in part), the non-standard usage is more appropriate. In each case, I raise novel theoretical considerations that, when incorporated, change the analysis.
Roughly speaking, my main conclusions are (1) saving lives is not as straightforwardly good we tend to suppose, may not be good at all, and is not clearly a priority; (2) happiness can be measured through self-reports and, based on the self-reported evidence, treating mental health stands out as an overlooked problem that may be an even more cost-effective way to improve lives than alleviating poverty; (3) the cause prioritisation methodology proposed by effective altruists needs to be moderately reconceptualised and, when it is, it turns out it is not as illuminating a tool as we might have thought and hoped.
Before I provide some context to the chapters and outline their contents, I make three remarks.
First, the aim of the thesis is to see what follows, given particular moral theories, rather than to evaluate which moral theory is correct. Philosophers have focused extensively on the latter over the years, which is why I turn to the former to make my contribution; we will shortly see it reveals a rich bounty of interesting theoretical questions. Hence, this thesis is of the flavour ‘if X were true, then the surprising result is Y’ without arguing for the truth of any particular X.
Second, I expect the subject matter of the thesis to be of greatest interest to those (such as this author) who believe that, if we want to do the most good, the practical priority is to make people happy, rather than make happy people, or prevent the making of unhappy animals.3This paraphrases (Narveson, 1973) who at p. 70 states “morality is in favour of making people happy but neutral about making happy people”. I add a further reference to animals to accommodate the fact some think preventing the existence of unhappy animals is high priority. I focus on saving and improving lives and offer no argument here that those who already prioritise something besides these need to reconsider their position.
Third, the chapters have, for the most part, been written so each is comprehensible without having read the others; as I am not advocating a particular moral view or building towards a single conclusion, this seemed the easiest way to structure the thesis. The final two chapters are something of an exception to this—both develop ideas from earlier in the thesis and will be less understandable if they are read by themselves. Given this structure, this introductory chapter serves the important role of putting the different chapters into context and explaining how the arguments connect to one another.
There are seven chapters. Chapters 1 to 3 concern saving lives, chapters 4, 6, and 7 are about improving lives, chapters 5 and 6 relate to cause prioritisation. To explain the overlap, chapter 6 proposes a new approach to cause prioritisation and applies it in the case of improving lives. The context and contents of the chapters now follow.
Prima facie, saving a life does a great deal of good. Given the best estimates are that we can prevent a premature death for, on average, only a few thousand pounds if we donate to certain effective charities, it is easy to see why one would think saving lives is the most good you can do.4According to the analysis of charity evaluator (GiveWell, 2019a) However, on closer inspection, matters are not so straightforward.
Chapter one raises one complication: what if, when considering the value of saving lives, we account for the fact that most people are meat eaters? Peter Singer has famously argued that eating meat is wrong on the grounds of the animal suffering this causes.5(Singer, 1975) Singer has also argued that we would be morally required to jump into a shallow pond to save a drowning child even if this would ruin our expensive clothes.6(Singer, 1972) I argue that, assuming meat eating is wrong on the grounds of the animal suffering caused, it is plausible that meat eaters cause so much suffering that saving the life of an average stranger is bad. If we grant the compelling principle that we are not required to bring about the worse of two outcomes (or slight variants of this principle), then it follows that we are not required to save lives, even in cases where we can easily rescue someone. I do not seek to argue meat eating is wrong on grounds on animal suffering; this argument investigates what follows if we accept that view. The more general result of the argument is that, even if we do not think accounting for meat eating is sufficient to make saving lives bad, it will reduce the value of doing so by some amount. I note this is the only chapter in which I engage in normative issues, that is, about what we ought to do; all the other chapters are concerned solely with the value of outcomes.
Chapter 2 adds another wrinkle to determining the value of saving lives. Many people seem to accept the ‘Intuitive View’, that saving lives is good and—as the Earth is overpopulated—that averting new lives is good too. Although it is not widely recognised, the Intuitive View is in internal tension—another way to reduce population size is by not saving lives. This raises some questions, for instance, whether and how the Intuitive view could be true. I develop Greaves’ earlier analysis of this topic and then investigate both how probable the Intuitive View is and what the practical implications are of accounting for under/overpopulation.7(Greaves, 2015) I do this by first assuming Totalism is the correct theory of population ethics, and second assuming a Person-Affecting View. I argue that the Intuitive View is distinctly unlikely on Totalism: it is more probable that one of saving lives or averting lives is bad. Further, were the Intuitive View true, the value of each of saving and averting lives would be small—far smaller than we would expect—and thus both are relatively unpromising interventions if the aim is to do the most good. This result is problematic for Peter Singer who seems to hold the Intuitive View, Totalism, and that life-saving and life-averting charities are among the most-effective giving opportunities—this combined position is highly improbable.8See the recommendations made by (The Life You Can Save, 2019) a charitable organisation founded by Singer and named after his book of the same name (Singer, 2009) On a Person-Affecting view, the Intuitive View is (unsurprisingly) far more likely. If both the Person Affecting and Intuitive Views are true, the values of saving and averting lives would be in large range (how large is specified in the chapter). The general result is that we don’t know the values of saving or averting lives unless we know how under- or overpopulated the world is. I close by briefly arguing that it’s not obvious where the world is in relation to optimum population size (whether we think just this generation or all generations matter); hence it’s correspondingly unclear what the value of saving lives is on this wider analysis.
Chapters 1 and 2 consider some problems that arise when we account for the other regarding impacts of saving a life—the impact saving lives has on others (human and non-human). Many people seem to think saving lives is the most good we can do—the particular effective altruist suggestion is to save the lives of children in the developing world with cheap, effective health interventions. Those making this suggestion seem to be basing this primarily on the self-regarding value of saving a life—the value associated to the person whose life is saved. Chapter 3 takes a step back and asks whether, if we ignore these other-regarding effects, saving lives could still the most good we can do. I set out four commonly-held, but not exhaustive, views of the value of creating and ending lives (such accounts are a combination of a population axiology with an account of the badness of death). In each case, I claim it’s not obvious that saving lives is the most cost-effective option—either there is an alternative that seems similarly cost-effective or the view is sufficiently underdetermined that we cannot straightforwardly make the comparison.
Taken together, the conclusions of the first three chapters are that (1) accounting for the other-regarding effects of saving lives makes it far less clear that saving lives is good, or even if it is good, and (2) even if we ignore these other-regarding effects, saving lives is still not clearly a priority on any of a commonly-held range of views.
In chapter 4, the focus moves on to how best to improve lives, that is, to make people happier during their lives. Singer and MacAskill seem to suggest that the best way to do this is by alleviating global poverty.9(Singer, 2015), (MacAskill, 2015) While this suggestion is also, on its face, highly plausible, I argue this is not so clearly the case either.
Chapter 4, the longest in the thesis, starts with the observation that, while MacAskill and Singer seem to hold that well-being consists in happiness, they do not seem to make use of the ‘subjective well-being’ (SWB) literature from psychology and economics—where individuals provide self-reported measures of their moment-by-moment happiness and/or their overall life satisfaction. Instead, MacAskill and Singer rely on more conventional metrics for well-being such as income and Quality-Adjusted Life-Years (QALYs), a measure of health. I suggest four possible objections to using self-reports, rather than of any other metric(s), to determine what increases happiness: (1) happiness is not measurable through self-reports; (2) individuals’ scores are not interpersonally cardinally comparable (i.e. a one-point increase for one person on a 0-10 scale is equivalent to a one-point increase for anyone else); (3) there isn’t enough available evidence of such self-reports to use them to guide decision-making; (4) it is unnecessary to use such data as they wouldn’t change the priorities anyway. Chapter 4 addresses the first objection fully and the latter three partially. I argue that, despite the doubts, ‘subjective well-being’ (this umbrella term includes both happiness and life satisfaction) can be measured through self-reports. I explain why it’s broadly reasonable to treat the self-reports as interpersonally cardinally comparable even though we cannot be certain of this. I also argue that even if the scales are not interpersonally cardinally comparable that should does not present a sufficient reason not to use them. Regarding the latter two objections, I provide suggestive evidence that there is enough data on SWB to inform our actions and it may lead us towards different priorities: the clearest case is mental health, which stands out as a major cause of unhappiness and has not been considered an important issue by effective altruists so far. A more compelling reply to the final two objections is offered in chapter 7.
As chapter 4 revealed there is a new, potential priority if we want to make people happier, this prompts the question of what method, in general, we should use to determine which of the world’s problems are more important than others (more important in the sense that it allow us to do more good). MacAskill notes that typical priority-setting method used by effective altruists is the three-factor ‘cause prioritisation’ framework, on which problems are evaluated by considering their scale, solvability, and neglectedness.10(MacAskill, 2018) The obvious thought is we should just apply this method to the domain we are focusing on—improving lives. The problem is that, while the three-factor method seems to capture something important, no careful analysis has been done of the method and there are a number of open questions about how exactly it works. For instance, MacAskill implies it’s possible to evaluate problems via the framework prior to making quantitative cost-effective assessments of the solution to those problems. This is somewhat mysterious—how it is possible to evaluate a problem(/cause) prior to assessing the cost-effectiveness of particular solutions(/interventions) to that problem?
In chapter 5, I set out the cause prioritisation method, clarify its workings, and address some outstanding questions. I suggest the priority-setting process should be partially reconceptualised. The most important conclusion is that using the scale neglectedness-solvability framework to evaluate the cost-effectiveness of a problem (e.g. poverty) ultimately relies on how carefully we have estimated the cost-effective of particular solutions to that problem (e.g. particular poverty interventions). Hence, the cause prioritisation method turns not to offer much in the way of useful shortcuts for determining what the most important problems are: if we want to find out how to do the most good, we’ll have to carefully assess the many different ways we might solve the problems in front of us.
Having identified the limits of the cause prioritisation methodology in chapter 5, chapter 6 proposes a new method, ‘cause mapping’, in order to help organise our search for potential solutions to the problems we are interested in. In essence, cause mapping involves breaking the priority-setting process into a series of distinct steps and considering the relevant possibilities at each step in order to help us structure our thinking. I then apply the cause mapping approach to the question of how best to improve lives. The result is a ‘long list’ of options which, prima facie, seem to be highly promising ways to improve lives. In addition to mental health and poverty, I suggest pain, positive education (i.e. teaching resilience and life skills in schools), and drug policy reform stand out. As such, I set some further possibilities that have not yet been seriously considered by any effective altruists. What is required next is a careful empirical investigation of these options.
While I am unable to assess all these possibilities, I do examine one in some detail in chapter 7. As Singer and MacAskill draw their charity recommendations from GiveWell, a charity evaluator, I attempt a first-pass cost-effectiveness analysis using SWB scores, which compares GiveWell’s eight top recommendations to a developing world mental health organisation, StrongMinds. StrongMinds seems roughly on a par, in terms of cost-effectiveness, to GiveWell’s top-rated life-improving recommendations. Comparing life-saving to life-improving charities turns out to be highly sensitive to an unresolved methodological question (where on a 0-10 life satisfaction scale is the ‘neutral’ point that is equivalent to non-existence) and I am unable to draw conclusions here. The analysis in this chapter also allows me to decisively meet the latter two objections raised in chapter 4 about the use of SWB data, by showing there is enough available evidence SWB data to guide decision making and that it does indicate new priorities.
Have effective altruists been doing good badly? I do not go so far as to claim that— not least because it’s unclear how we would evaluate this question. However, in this thesis, I am able to set out a range of important, novel theoretical considerations. This challenges current thinking about how to the most good in several key areas and shows how those looking to do the most good can do good even better.
Download the PDF version to read the full thesis.