Objections to Utilitarian Longtermism
In this article, I evaluate utilitarian longtermism (UL) and find it wanting, though to be fair to longtermism, I offer a defense of a moderate non-consequentialist form of longtermism.
For the sake of space, I left out some arguments against UL. I’ll provide of couple of them below.
To start, note that longtermism is a thesis in moral philosophy which holds that persons alive in the present have a moral duty to future persons (i.e., those who are not alive now but will be born and live in the long-term future) to act in ways that contribute to their welfare. Let us call this the Longtermist Thesis (LT). And according to a common articulation of utilitarianism, for any morally significant act, we should act in such a way as to bring about the greatest cumulative benefit for the greatest possible number of people. In other words, we should act to maximize happiness for as many persons as possible. I will call this the utilitarian thesis (UT). If we take the conjunction of UT and LT, and if we assume (a) that the number of future persons vastly outweighs the number of current persons and (b) that future persons exist (which, as I note, is a defensible claim if one assumes the thesis of eternalism in the philosophy of time), we get UL, which is roughly the claim that we should act in ways that maximize happiness for as many future persons as possible.
Now, it’s important to note that utilitarianism is a version of consequentialism, which is a category of normative theories holding that for any morally significant act, the consequences of that act make the act morally right or morally wrong. I.e., the end justifies or condemns the means used to produce it. Hence, UL is a consequentialist thesis which holds that the end of benefiting future persons justifies any means necessary to produce that result, including the harming of present persons if such is required to ensure the welfare of future persons (assuming that the number of future persons outweighs the number of present persons).
In the article, I provide two objections against utilitarianism, which I call the predictability problem (PP) and the harm problem (HP), and I claim that these considerations show that UL is problematic. I also raise the objection to longtermism that, assuming either presentism or growing-block theory in the philosophy of time, future persons do not exist and therefore we can’t have duties to them. I then note that longtermists can avoid this problem by assuming eternalism, another view about the nature of time.
Note the additional arguments below against UL that I didn’t include in the article.
First, there is what I call the maximization problem (MP). This is similar to the PP noted above. Recall that according to UL, we should act in ways that maximize happiness for as many future persons as possible. But this seems an infeasible demand. For any sufficiently productive choice c that person S makes which produces cumulative value v for future persons, for all S knows, S could have made a different choice c* that would have produced a slightly better result v+1. Suppose that S realizes this point and thus chooses c*. Well, for all S knows, there was some other choice c** that would have produced a slightly better result v+2. And so on.
One need not take the position that this problem repeats ad infinitum. However, given the limits of human knowledge, the likely astronomic number of future persons (according to some longtermists), the complicated nature of happiness, the various options that each individual might select with respect to producing future consequences, the possible results of such choices, and the various combinations of these factors, it seems that are many possible options that might be better than c. And since we cannot know what the best option is, it is epistemically possible that for any option we choose, some other option would have been better. This is an intractable problem for us humans, and one that is present whether we are talking about happiness in the utilitarian/hedonistic sense (i.e., pleasure and the absence of pain, desire-satisfaction, etc.) or in the Aristotelian eudaimonistic sense.
It seems to me that the MP is a serious objection to UL given that there might be trillions of future persons, each of whom has a complicated psychological life and eudaimonistic needs of which we cannot know precisely. It’s hard enough for, say, a parent to maximize the happiness of his own child. A good parent will seek such maximization as much as possible, but he can’t know all the results his actual and non-actual but possible decisions would produce, nor can he know exactly what his child will need at all points in the future. Human life is too complicated. A fortiori, so it seems, if we are considering trillions of future persons living in a future world about which we know little.
What we have here is something like a combinatorial explosion. To elaborate, in mathematical combinatorics[1], theorists use the term ‘combinatorial explosion.’ This term refers to a characteristic of a problem according to which the complexity of the problem sharply increases commensurate with the exponential growth of its numerical aspects.[2] For example, if a system (say, a world with trillions of persons over the course of its history) contains a vast number of relevant factors (e.g., trillions of persons, their desires, needs, welfare, etc. plus the various options that present persons might select to maximize the welfare of future persons) such that the exponential increase of the possible combinations of factors in that system render infeasible an exhaustive search of those factors, then the system is a combinatorial explosion. A brute-force solution to such problems is not available to us, regardless of how intelligent, resourceful, or perseverant we are in attempting to solve the problem. Computer scientist Richard M. Karp (1986, p. 98) describes the sharp increase by noting that a combinatorial explosion contains a vast, furiously growing number of possibilities to be searched. Such an explosion justifies the conclusion that a brute-force or exhaustive search solution regarding the combinations is intractable for human beings and for computers.[3] I suspect that the world possesses something like a combinatorial explosion of axiological factors and, thus, that an accurate and reliable axiological evaluation of the world, of the free choices of persons, and of the best results of those choices for trillions of future persons is intractable for human beings. This combinatorial difficulty weakens UL.
In short, we are not in the epistemic position to know what is best for what might be trillions of future persons, nor are we in the epistemic position to choose in such a manner as to bring these best consequences about.
Second, there is what I call the unrealized happiness problem (UHP), which also seems to undercut UL. Again, recall that according to UL, we should act in ways that maximize happiness for as many future persons as possible. I.e., for every morally significant choice we make, that choice should be aimed reliably at securing the greatest happiness for the greatest number of future persons. But if everyone were to follow such a principle, then it is likely that no one would enjoy fully the happiness that their ancestors attempted to produce for them. For instance, take everyone currently alive. Now, if everyone who lived prior to the firstborn of those currently alive had done everything they could have done to produce the best results for those currently alive, and yet everyone currently alive is exhausting their own efforts to produce the best results for every future person, then it is likely that no one currently alive is enjoying the maximal happiness that might have been produced for them by those who lived before, since those alive now will not rest long enough from their efforts to produce the best results for future persons to enjoy what was produced for them by those who lived before. Moreover, future persons are in the same position, since they must devote themselves tirelessly to maximizing outcomes for those yet to be born. And so on. (This point is related to a problem with utilitarian thinking that German philosopher Josef Pieper called “total work,” which is the utilitarian-driven process by which human beings are transformed into mere laborers and thus become unable to enjoy the benefits of leisure, which according to Pieper, is necessary for human flourishing. See Leisure: The Basis of Culture.)
It seems, then, that to follow UL is to undercut the outcome of overall happiness rather than to secure it. At best, the relatively few persons who live at the end of human (or personal) history (assuming there is such an end) will be able to rest and enjoy the results produced by past generations, assuming that these end-timers know that they are the last humans (or persons) and hence that they need not work for the benefit of any future persons but instead can relax and enjoy the fruits of past laborers.
In sum, the MP and the UHP combine with the PP and the HP to present a serious set of objections to UL.*
See my paper for a defense of a moderate version of non-consequentialist longtermism.
*Note: The MP and the UHP are similar to an objection to utilitarianism called the no-rest objection. According to this objection, for any act a person might perform to maximize utility, epistemically, there might be a (practically) countless number of alternative acts that would maximize utility more effectively. How should one choose? Where should one stop in seeking the best alternative?
References
Crozat, E. R. 2022. The Aporetics of Longtermism: Are You Morally Obligated to Future Persons? Epoché Issue #56, October 2022. Available at https://epochemagazine.org/56/the-aporetics-of-longtermism-are-you-morally-obligated-to-future-persons/
Gass, S. and Harris, C. 2001. Encyclopedia of Operations Research & Management Science. Boston: Kluwer Academic Publishers.
Illingworth, V. 1997. A Dictionary of Computing. 4th ed. Oxford: Oxford University Press.
Karp, R. 1986. Combinatorics, Complexity, and Randomness. Communications of the ACM. Volume 29, Number 2.
Pieper, J. 1948. Leisure: The Basis of Culture.
[1] Combinatorics is a branch of mathematics which addresses various problems of counting with respect to topics such as combinations, permutations, sets, and members of a set.
[2] Gass and Harris (2001) define ‘combinatorial explosion’ as “the phenomenon associated with optimization problems whose computational difficulty increases exponentially with the size of the problem.”
[3] According to Valerie Illingworth (1997:86), the game of chess is an example of a combinatorial explosion: “in the game of chess the number of choices at each level increases by the branching factor, which may typically multiply the options by 20 or more at each move. Although in theory it should be possible to analyse the game of chess from start to finish, the number of states to be examined is so enormous that it is completely impractical, not only at present but for any conceivable computer in the future. (To appreciate this, consider an example: if one million game states can be examined each second and the branching factor is 10, then to analyse 6 moves ahead takes 1 second, to analyse 12 moves ahead takes 11 days, and to cover 18 moves takes nearly 32 000 years.)” Illingworth (1997, p. 52) defines the “branching factor” of graph theory as “the average number of branches (successors) from a (typical) node in a [rooted] tree. It indicates the bushiness and hence the complexity of a tree.”