Deconfusing Effective Altruism: The Philosophy

Davidmanheim

In practice, the widely-endorsed goal that Effective Altruism has, of doing good, is entangled with a number of things which are less universally lauded - from the community to the implied philosophical positions. That makes it especially valuable to separate claims and assumptions from approaches, to allow individuals to consider and evaluate them separately. This is part 1 of my attempt to disentangle and clarify some parts of the overall set of claims that comprise effective altruism, in this case, the philosophy - not to evaluate them, but to enable clearer discussion of the claims, and disagreements, to both opponents and proponents.

To do this, I will first note that there are at least two very different things meant by Effective Altruism. The first is a somewhat narrow definition of a philosophy, and the second is a social movement, and community. This post will focus on specific parts of the first, more narrow definition, discussed below.

Evidently, the ideal division of labor is for there to be a core philosophy, and an interpretation of that philosophy by the community. Unfortunately, however, research is value-laden in ways that make the interpretation central to the epistemic task of effective altruism. In fact, MacAskill attempts to build a definition based on two desiderata, the first being to ensure the definition matches community practice, and the second being a “ concept to be broad enough to be endorsable by or useful to many different moral views, but still determinate enough to enable users of the concept to do more to improve the world than they otherwise would have done.”

To accomplish this, however, I think we need to add a third component to the definition of Effective Altruism, a normative component^[1]. This component is Richard Chappell’s idea of Beneficentrism, that “helping others is deeply important, and should be amongst one’s central life projects.” The new normative component is still, I think, broad enough to fulfill the first half of the desiderata, but it leaves out the part which is determinative, and allows the value-laden part of EA to be separate^[2].

Normative Effective Altruism Should be Near-Universal

As Richard Chappell argues, “most people don't really seem to treat promoting the general welfare as an especially important goal.” This is unfortunate, and there is discussion among ethicists about whether beneficent acts are morally required, or just an ideal^[3] - but in either case, the principle of beneficence, that we ought to do things for the good of others, can be seen as part of basic and nearly universal morality. In ethics, “beneficence connotes acts or personal qualities of mercy, kindness, generosity, and charity. It is suggestive of altruism, love, humanity, and promoting the good of others.” (Beauchamp 2019) And this principle is widely accepted as normative - but not necessarily a priority. This additional facet is, I think, what Richard Chappell’s idea of Beneficentrism addresses.

Beneficentrism is the claim that “helping others is deeply important, and should be amongst one’s central life projects.” This is a claim primarily aimed at the global middle-and upper class; the case for higher relative prioritization is stronger when the competing priorities are luxuries rather than necessities. Beneficentrism is also a stronger claim than the principle of beneficence, in that it not only ought one to do good, but they should prioritize it over at least some other things. As Chappell points out, beneficentrism is sufficient to serve as a moral foundation for Effective Altruism, once we address the additional requirement of effectiveness. And it should be unsurprising that very few critics claim that Effective Altruism is wrong for asking people who are in the highest income percentiles globally to prioritize beneficence to at least a moderate extent.

The Effectiveness Criteria Should be Widely-Shared

Having claimed that the “altruism” of Effective Altruism should be near-universal, the question turns to the question of effectiveness. Effectiveness is in some ways an epistemic requirement, rather than a moral claim directly. Despite that, EA as a moral theory insists on effectiveness. I view this as widely shared - very few people actually object to effectiveness in principle.

To reinforce the fact that this is not as widely shared a view as beneficence, it is useful to note that there are alternatives. For example, in some views, beneficentrism could require no criteria of effectiveness at all. Virtue ethics can have the view that, while it is critical to prioritize beneficence, the doing of a positive act is sufficient regardless of impact. Alternatively, moral acts could depend on whether you know about the effectiveness of an action, rather than the expected outcome. If true, this would lead to a subjective beneficentrism - a requirement to do what seems good^[4]. Alternatively, it could lead to a morally irrelevant beneficentrism, where only actual consequences matter, and knowledge short of perfect prediction is irrelevant. And because all admit that perfect knowledge is impossible, a claim that partial knowledge about impacts is morally irrelevant is a claim that there can be no moral benefit to research.

The Epistemic Task

Returning to the criteria of effectiveness, MacAskill defines effective altruism as having two components^[5]. First, there is an epistemic task of determining “how to maximize the good with a given unit of resources, in impartial welfarist terms” - two qualifications we will not address in this post - and second, there is a component of doing that good, which is beneficence, mentioned earlier^[6].

The epistemic task, according to MacAskill, seeks to “maximize the good with a given unit of resources,” and per MacAskill, effectiveness is an important criteria for helping others - though not necessarily the only criteria. It is also worth noting that the combination of beneficence and the epistemic task imposes a limit on the epistemic task. That is, given finite time and resources, there is a trade-off between investigation and activity, and effective altruism as a project must balance them - too much focus on philosophy takes resources away from the practical project of doing good, while too little ends up failing to be effective^[7]. Given that, critics of Effective Altruism should rest assured that it will always be possible to disagree with effective altruists for both putting too much effort into analysis, and for making choices without considering other concerns or models of impact.

But the contents of the task are still unclear. What should be done to pursue effectiveness?

Historically within the movement, the methods used for this determination have been strongly associated with analytic philosophy, econometric methods, Randomized Controlled Trials, and Bayesian reasoning. If used appropriately, even critics of effective altruism agree that these are the available and appropriate tools to pursue the goal of maximization. At the same time, there have been discussions of whether these tools are misapplied, used to exclude other valid sources of information relevant to the project, or applied in overly mechanistic ways that fail to account for known shortcomings.

Regardless of that reasonable criticism, however, even the tools used are not fundamental to the epistemic project, and if other tools were shown to work better, they would need to be adopted. At the same time, the tools have clearly sociologically influenced the practice of effective altruism quite strongly, and the various shortcomings of the methods as they are applied is a valid target for criticism of the practice of Effective Altruism.

In any case, we now have a basis for effective altruism as an obligation, and in this post I will not identify the target of the beneficence, nor discuss the sociology of the movement and the consequences. Even without addressing those, however, we can make several observations about the nature of Effective Altruism, the obligations it imposes, and how it differs from or comports with other philosophies.

Obligations and Comparative Philosophy

Once we have defined what the moral tasks of effective altruism are, we can consider the degree to which the stated priority is obligatory, and what form it takes. We will consider two forms, one personal, the other monetary.

The monetary claim is by far the simpler one. To the extent that beneficentrism isn’t totalizing, and is only a priority, the level of priority can be considered as a purely fiscal decision - how much money should be spent on effective charity^[8]. Following the suggestion of many Effective Altruists, I would personally endorse the Schelling point of giving 10%, but if someone chooses to give 1% or 50% of their income, or time, they are still prioritizing helping in a way completely consistent with the definition of effective altruism.

The personal claim is that the obligation or voluntary choice for effective beneficence includes choice of career and other behavior - especially questions of diet. Regarding career choice, 80,000 Hours, a central Effective Altruist organization, provides guidance for doing the most good with one’s career. One assumption or assertion underlying their analyses is that because careers have widely varying degrees of impact, the choice is subject to the same analysis, and at least part of the decision one’s job and career should be the extent to which it comports with Effective Altruism.

Interpreted one way, in my view incorrectly, the obligation of beneficence is outcome-focused. If choosing between a career that might save 10,000 lives, and one that only provides a standard income, the only way to accomplish 10% of the possible good is to have an effective career. This view, however, makes the personal claim very different from how the fiscal claim was understood. By focusing on the outcome, it comes close to either optimizing beneficence, or to the claim that the obligation of beneficence is a high enough priority to dominate most decisions. And it seems that this is not how effective altruist organizations and leaders view the issue.

However, in the most extreme version, we find that an outcome-focused prioritization of beneficence effectively requires utilitarianism - which is in some sense totalizing beneficentrism^[9], that is, giving exclusive priority to beneficence.

At this point, it seems important to note that no part of MacAskill’s claim implies or entails essentially any of the utilitarian philosophical commitments which many (in my view mistakenly) assume are required by a (non-totalizing) effective altruism.

In fact, as many Effective Altruists have noted, the claims noted so far are compatible with almost any other worldview^[10]. It is not consequentialist, though it does require viewing consequences as a component of an ethical evaluation - that is, the actual benefit, as understood via the epistemic task, matters. Optimizing beneficence is also not utilitarian, though it’s certainly required by standard utilitarianism. It’s not a deontological system, nor is it virtue ethics, or divine command theory, but, omitting the epistemic task, effective or optimizing beneficence is often implied by or is at least compatible with all of them. It’s even (minimally) compatible with ethical egoism, in that an egoist may care about others, albeit in that case selfishly.

I will also add that nothing about the definition of effective altruism requires that “the good” which is being promoted is hedonic, or even measured economic welfare^[11]. This, among other things, allows effective altruists to reject utilitarianism, and it means that utilitarians can reject the views of some effective altruists about prioritization. That is, the two may be largely overlapping, but they are distinct.

Conclusion

The basic ideas of Effective Altruism are individually simple, but the specific choice of claims to accept yields different concrete outcomes. Unfortunately, the name “Effective Altruism” and the affiliated community includes only a small portion of the people who would embrace the claims. This essay is an attempt to identify the core of the most universally acceptable parts of Effective Altruism.

Doing this, however, creates a critical motte-and-bailey to avoid, where community members promote a worldview, or donations to a specific charity, or starting on a specific career path, and when questioned, respond with claiming that the philosophical position is unassailable^[12]. This ignores the real impact of community and auxiliary beliefs on the choices of causes and charities made within the effective altruism community - none of which have been discussed anywhere in this essay.

Conversely, many criticisms of effective altruism attack the philosophy of Effective Altruism, rather than the community, on the basis of philosophical positions that aren’t actually related. Many are even based on ideas or philosophies that are soundly rejected by the community - for example, criticizing Effective Altruism as promoting anything from financial fraud, to wealth inequality, to racism. Hopefully, criticism of Effective Altruism can differentiate between these, and provide clearer reasons for disagreement.

In summary, Effective Altruism as a philosophical goal, in the general sense used here, can be one of many priorities people have^[13]. As a philosophy, it does not need to include community involvement at all. And the relatively small tent of central effective altruists organizations, conferences, and the somewhat insular community around these should not be conflated with the very inclusive philosophy and broader movement. For that reason, the community should be careful to clearly evaluate its various beliefs, and be especially wary of embedding various (far less obviously beneficial) views into its claims to be doing good.

Citations

Nielsen, Michael. “Notes on Effective Altruism” June 3, 2022 https://michaelnotebook.com/eanotes/

Chappell, Richard Y. “Beneficentrism: Utilitarianism minus the controversial bits” https://rychappell.substack.com/p/beneficentrism

Beauchamp, Tom. "The Principle of Beneficence in Applied Ethics", The Stanford Encyclopedia of Philosophy (Spring 2019 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/spr2019/entries/principle-beneficence/>.

MacAskill, William. “Defining Effective Altruism” Effective Altruism Forum, 2019. https://forum.effectivealtruism.org/posts/9wYa8BqSTMcx9j2tK/defining-effective-altruism

^{^}
This directly opposes MacAskill’s ideal, but I think it rescues the movement from something which is fundamentally confused.
^{^}
Note that the current attempt to decompose EA is very much incomplete. There are other parts of effective altruism that would benefit from similar decomposition, from the way that socialization is promoted, to the naivety of some approaches to policy change, to a lack of epistemic humility, to discussions about how to evaluate uncertainty and even which specific intervention types should be considered effective. All of these are worthy of discussion, and each has received a greater or lesser amount of attention already - but they are different discussions, and so I will leave them to the side.
^{^}
Bernard Gert is a proponent of this claim, for example.
^{^}
In this view, being informed about effectiveness could change your moral obligation, but you do not have a proactive duty to find that information. Note that this is different from claims that there is a limited value in additional information, and easily obtained information is sufficient to discharge your obligation of beneficence.
^{^}
Despite his preeminence in the community, his definition isn’t universally accepted within the community, though in this case it seems to be based on widespread agreement about community thinking about what the philosophy is. For that reason, I am using it as definitive of at least the philosophical claims, while agreeing that it does not capture the community’s other beliefs, which, again, are mostly out of the scope of this post.
^{^}
The second component also addresses the social movement, which, again, will not be addressed in this post.
^{^}
Those effective altruists who are focused on economic evaluation will recognize this as a generalized value-of-information question, while those focused on longtermism and artificial intelligence will recognize it as an explore/exploit tradeoff. These are different framings of the same issue.
^{^}
We will, for now, consider this independently of the tradeoffs between the epistemic and beneficent tasks.
^{^}
The other requirement for utilitarianism is impartiality, which is not discussed in this post. And for group-level decision making, I personally remain convinced that something like utilitarianism is almost inescapably correct - but aside from making that statement, I’ll avoid digressing into this point.
^{^}
Of course, this does not prove that effective altruism is socially compatible with those theories, or the social norms of the proponents of those theories. But that will be addressed below.
^{^}
I agree that most EAs adopt some version of hedonic or preference definitions, but not all do. The consensus view, however, is not the same as the philosophical claim. Fundamentally, the only criterion that any definition of “the good” does need to follow is measurability - that is, there must be a basis of comparison between different types of good that are contemplated. But this is necessary for any decision process. That is, if we have any decision for philanthropic decisionmaking, it must have some criteria that are being used. So if measurability is restrictive, the fault is in the need for some decision procedure, not the specific one employed - whether hedonic, preference based, or otherwise.
^{^}
This is distinct from Nielsen’s “EA-Judo,” where criticisms of EA-in-practice are turned into discussions of how EA-in-practice should change because people accept EA-the-intellectual-project. In that case, there is substantive discussion of how practice should change. More worrying, and unfortunately not uncommon, is the motte-and-bailey of dismissing the criticism of EA-in-practice by retreating to defense of EA-the-intellectual-project.
^{^}
In fact, my view is that it’s better if it is not totalizing.

Effective Altruism Forum
EA Forum