Consider Eric, a warm-hearted individual who decided to donate some money one day. Upon research, he finds two interesting charities: one that supports children in distant countries to have a better education, and one that funds a struggling local bagel shop. Understanding that suboptimal sacrificesi are wrong, he struggles to decide.
Choices under such uncertain situations can be decomposed into the objective and subjectiveii “should”. Eric cannot grasp the objective
“should” easily, because there isn’t a definitive moral answer to which charity is better. However, given full knowledge of the situation (i.e. the charities’ utility and miscellaneous considerations), the comparatively
“better” should act as the objective imperative. If the moral view that he should donate to the bakery is objectively correct, it would be wrong for him to put in his money instead to educate children. However, the amount of good a charity can provide is difficult to measure numerically, as are the moral principles underlying each situation. This means it is unrealistic for Eric to obtain an objective “should” in a realistic situation. Individuals must then rely on patchy beliefs and determine decisions
through the subjective “should”. How to find the subjective “should”, then, is the problem. It is argued here that the best-expected choice-worthiness, or the potential “good” should be the sole consideration when it comes to the selection of charities because all other considerations of moral value or credence fail to find the intuitive choice in given
scenarios.
On Rival Theories
Traditionally, there are three theories or methods for obtaining an
optimum strategy. The first is proportional. One’s intuition when seeing Eric’s dilemma may be that he should split up his money and give portions to both charities. Generally, proportionality suggests he should donate k% of his resources to the moral principle he has k% credence in. Here, credence represents how much an individual believes the charities can achieve in proportion, similar to the conventional use of credence as probability of individuals believing a proposition to be true.
This seems very attractive, given that the two moral principles underlying each charity do not intersect or affect each other. That is, the support of
his local bagel shop does not promote nor impede efforts to improve
education in a foreign country. The first challenge to this model is when the moral principles are not independent. If the charities were to be a fertility initiative and contraceptive initiative, the split donation would mitigate each other’s effect. His proportional donation here would be
squandered, and clearly, this isn’t the right decision. When theories clash, proportionality may lead to the neutralization of one’ effects on the world, just like the fertility versus contraceptive scenario.
Finding a subjective “should” simply based on credence is also shaky when there is an involvement of empirical probabilities. An example is when comparing a charity that will guarantee a 5% literacy rate rise per donation against one that is a highly ambitious scheme which may raise literacy rates by 10% or nothing if it fails. Because you do not know whether the ambitious project will be successful, it would be extremely reckless to risk the chance of not improving the education of the 5% of children. Although a case may be suboptimal and less credited, the proportional argument supporting the reckless option is reckless itself.
The second theory, a possible antecedent to the first, is snappier. It states an action X is appropriate when X is permissible according to the moral theory an individual has the most credence foriii. The first issue with this
is when there are choices with equal credence. Choosing between them at random would be inconsiderate, so a fix could be that so long as X is permissible for any most credited choice, X is appropriate. This may be problematic too when the top choices collide like in the proportionality
example.
The deeper-rooted problem with this method is the process of
individuation. Suppose Eric does decide to credit the educational charity more, but he realizes that there are subtle variations and branches for the cause (e.g. providing primary education vs creating universities). This
distinction leads to the dilution of his credence on the original theory. In effect, this looks like instead of Table 1 which suggests that choice A is most appropriate, divisions of major choices can lead to a different result such as in Table 2, where B is the more appropriate choice according to the method of choosing the choice with highest credence.
Table 1
| Choice | Education – 60% | Bakery – 40% |
| A | Permissible | Impermissible |
| B | Impermissible | Permissible |
Table 2
Choice Higher education – 30% Primary education – 30% Bakery – 40%
| A Permissible | Permissible | Impermissible |
| B Impermissible | Impermissible | Permissible |
The choice of B is intuitively not very logical and helpful. In a real-life scenario, it is likely that individuals will be less capable of successfully determining the extent of individuation in their choices. Unequal
individuation leads to extreme cases where choices that have very small credence may trump more general charities that are subdivided many times. In Table 3, choice A is preferable to B, which seems rather
counterintuitive given the extremely low credence of some charity U in comparison to the subdivided C.
Table 3
| Choice | U – 2% | C1 – 1% … C98 – 1% |
| A | Permissible | Impermissible … Impermissible |
| B | Impermissible | Permissible … Permissible |
The third and final approach to including credence in choices of charity is through directly considering the individual’s credence for each option. If the individual believes an option to be most likely to be permissible, then that option should be opted for.iv This seems most reasonable as it gets rid of the possibilities of individualizing options.
However, the fundamental flaw shared among the second and third theory is that it is incapable of allowing trade-offs between one’s credence and the degree of choice-worthiness of options assigned by their theories.
Suppose that theories differ not only on the moral scale but also on the magnitude of their sponsored options. Imagine that a less plausible
charity works on a grave issue, while a slightly more plausible charity is about a comparatively minor issue.
Table 4
| Choice | X – 49% | Y – 51% |
| A | Permissible | Slightly wrong |
| B | Gravely wrong | Permissible |
According to the previous theories, choice B is more acceptable, as Y has more credence. However, given that choosing B would mean a grave fault on the principle for charity X, it seems more rational to outweigh the
small credential difference with the scales of the outcomes. This
introduction of a scale of outcomes is distinct from moral theory and
uncertainty and allows more weighing. An even better example would be when there are three options, two of which are gravely wrong for one of two principles and one that is only slightly wrong for both.
Table 5
| Choice | X – 50% | Y – 50% |
| A | Gravely wrong | Permissible |
| B | Slightly wrong | Slightly wrong |
| C | Permissible | Gravely wrong |
Option B is wrong, according to both principles. However, A and C seem both more radical and inappropriate compared to the milder option. The central question is therefore not about which moral theory to accept, but the impact each option has. Indeed, in cases like this, there is no moral theory that one should intrinsically accept. Instead, it’s about which
option it is appropriate to choose to minimize the harm done by neglection which all the theories of credence use.
Maximizing expected choice-worthiness of decisions
It is argued that because of the inconsistency in attempting to characterize moral tendencies (e.g. Eric’s attachment to his community and thus the
bagel shop), the expected choice-worthiness is useful for deciding appropriate charities. The simple principle is, that when a comparison can be drawn across the expected choice-worthiness of options, an option is appropriate iff the chosen option has the maximal expected choice-
worthinessv. In this way, an individual can treat their moral uncertainty as
an empirical uncertainty in their charity.
Although empirical uncertainties seem to be different from necessarily true moral truths, consider an uncertainty about a mathematical
probability. If one has a bet with another on guessing a random number from 1-10, the value the guesser puts in should be 1/10 of what the other bets. This seems trivial, but the ideal value obtained from the expected
utility theory acts as a sense of “ought” for real-life decision-makers. One believes he ought to put in 1/10 every bet as to follow the best-expected utility, assuming the friend is rational. Thus, if such empirical
uncertainties lead to similar effects on the subjective “should” of
individuals, it is not unreasonable to suggest that moral uncertainties
between charities should be considered just like empirical uncertainties.
This means that although the expected choice-worthiness (i.e. the “good” a charity can do) may not reflect moral considerations, comparing
empirical probabilities for each option grants more solid ground for uncovering the worthiest one. Even more, if an individual has no
knowledge of the nature of their uncertainty, and moral and empirical
uncertainties are treated differently, finding one’s own uncertainty could potentially change the decision completely. This seems weird since the
nature of the uncertainty should not affect the decision.
Thus, we have shown that moral concerns should be considered similarly to empirical uncertainties, where expected utility comes into play.
Because in all other methods, the involvement of moral theories has been unable to provide reasonable solutions in certain cases, maximal expected choice-worthiness provides empirical reasoning to a rational decision.
The main opposition to this argument is based on the overdemanding nature of choice-worthiness. There is a looming implication that
individuals should sacrifice most of their income for the poor, resonating with Singer’s message.vi For example, he argues that members of affluent countries are obligated to give a large proportion of their income to those living in extreme poverty and that failing to do so is as wrong, morally, as walking past a drowning child whose life one easily could save. Even though many would reject this argument, there is some credence that the conclusions are true: it is gravely wrong to have a child drown. Since
everyone agrees that it is at least permissible to donate to distant charities supporting such causes, the dominant argument is for everyone to donate to a cause that is strictly more impactful even though the impacts do not land on their proximity. There is a dominance of targets of donation, and
so not many bagel shops will ever be supported.
A response to this argument is that this depends on the credence of
individuals to Singer’s view. If it costs £500 to help a child out of poverty by giving them access to college, while the alternative is the survival of
one shop, the former is a much more rational choice according to Singer’s view. However, if the £500 could, with a 1/10 chance, have those effects on a child, the choice-worthiness difference between the first and second case must be at least ten times as great. There exists a tipping point where expected choice-worthiness overlaps when the expected probability
decreases on one side. Thus, if an individual has a lower credence on
Singer’s argument and the chances of a first option are low enough, it still is appropriate to donate to the second option.
It is difficult to argue against effective altruismvii given that the central purpose of charities is to reduce inequalities. The final remark on this is on the real-world application of such moral considerations. Support for more effective charities promotes a competitive environment on the grounds of effectiveness. That is, as charities see that more efficient
alternatives are funded, they would be inclined to become more efficient too. For individuals, the greater the causes they support, the greater the
feedback from charities. It is assumed that people feel better when their efforts convert to more good. Indeed, standards of effectiveness are harder to obtain for the less informed individual, as cross-industry and cross-national comparisons are difficult to evaluate. However, given the previous argument for the model for a rational choice, if there is a
comparable choice-worthiness scale, organizations should strive to
develop such measurements. And one may say, even if the bagel shop charity is less impactful, if its choice-worthiness (probability of it
expanding and giving back) is higher than an alternative, one should choose the likelier prospect of nice, tasty bagels.
i Horton, J. (2017). The All or Nothing Problem. The Journal of Philosophy, 114, 94-104.
ii Hedden, B. (2012). Options and the subjective ought. Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, 158(2), 343–360.
iii Gracely, E. J. (1996). On the noncomparability of judgements made by different ethical theories. Metaphilosophy, 27(3), 327–332.
iv Gustafsson, Johan & Torpman, Olle. (2014). In Defence of My Favourite Theory. Pacific Philosophical Quarterly, 95. 10.1111/papq.12022.
v Kim, J. William MacAskill, Krister Bykvist, and Toby Ord: Moral Uncertainty. Ethic Theory Moral Prac, 1057–1059 (2021). https://doi.org/10.1007/s10677-021-10216-1
vi Singer, P. (1972). Famine, Affluence, and Morality. Philosophy & Public Affairs, 1(3), 229–243.
vii Berg, A. (2018). Effective Altruism: How big should the tent be? Public Affairs Quarterly, 32(4), 269–287.