Why I think I am not a Rationalist
An exploration of the stereotypical rationalist
I am not a rationalist – at least, I think I am not a rationalist; it is honestly hard to say. What even is a rationalist? And why do I even care about whether I am or not?
Perhaps to answer this, I should consider an article that has been making the rounds on Substack: “Why I’m not a rationalist (clean version)” by
. In this article, the author discusses a group of individuals who have become collectively known as the “rationalist community.” Rohan describes them as “a group of people who are really fond of Bayes’ Theorem, AI, and invoking the name of a Carthaginian god of child sacrifice. Your typical stuff.” It’s obviously not what I would typically assume regarding “rationality”—nonetheless, it is clear that these particular people have become well-associated with the concept enough for a response on membership to be warranted.
The rationalists are further described as strange owing to trying to make “rationality” a core part of their identity. Mr. Ghostwind expounds this strangeness through a series of observations regarding the people involved. He describes them as “intellectual foodies,” using the accumulation of trivial knowledge as a means for inner-group identity and status, which leads to them being assigned an “unwarranted level of intelligence, because they have [he has] this manic ability to switch back-and-forth between a bunch of unrelated subjects when in conversation with others.” This is further expounded by a tendency toward being “disembodied,” wherein their source of that same trivial knowledge comes less from the exercise of activity in acquisition, and more so from reading and other targeted forms of “rationalist” advertisement.
The embodied knowledge that might be corrective to this condition is noted to be readily discarded within the rationalist paradigm due to a condition coined as “spreadsheet brain.” Embodied knowledge has a tendency to lack quantification, which makes it harder to interact with using technique—such as being operated on within a spreadsheet. Given that such embodied knowledge cannot be converted into a disembodied form, the rationalist is suggested to discard it, lending their analysis to be depersonalized and seemingly overconfident.
Mr. Ghostwind identifies the rationalists’ preference for disembodied knowledge with that of Utilitarianism, a system of ethics based around the concept of utility or quantification of value. This system allows for the rationalist to use what they have acquired within a spreadsheet toward acting out a moral and upstanding life in the style of the cynics—being shameless against the morals of the crowd. Utilitarianism grants the rationalist an exactness in figuring what to ethically do to help the world; however, the same discarded embodied knowledge lends the ethics system toward oddity. The rationalist is observed to get stuck up with hypotheticals comparing rivers of dead children against the suffering of a sun god—without any particular reference to the actual.
Within the oddity of the hypotheticals borne of Utilitarianism, there comes a great deal of reactive hostility—especially considering the rationalist either operating or seeking to operate the machinations of society. To protect from exterior criticism, termini technici—jargon—is developed and employed against the common language. The result is that the rationalist community is seen as “insulated” to the point of impenetrability. When a non-rationalist comes to try and explain some embodied phenomenon that goes against rationalism, it is deflected—at least until it is reinvented within the rationalist community.
The rationalists are thus viewed by Rohan as intellectually insecure and ironically self-defeating; anything rational that comes from the rationalists is seen as the result of “rationalizing”—being no better than translating a phenomenon into the acceptable language of a particular status-oriented tribe obsessed with quantification.
In total, Rohan views the rationalists as status-oriented, disembodied cynics whose morality lies in quantifiable thought experiments instead of real encounters. But whilst this may be good for laying out the normative view of what a rationalist is, it makes no positive claims on what the alternatives are—or why Mr. Ghostwind has a disdain toward being a rationalist.
notes this in his article, “‘Why I’m not a Rationalist’ is a Bad Article,” claiming that there are insufficient arguments answering why Rohan does not wish to be a rationalist.
Mr. Star points to a lacking of alternatives by Mr. Ghostwind—particularly ones that themselves don’t use the techniques invoked by rationalists—critiquing: “Half of the points are just calling rationalists pretentious, know-it-all, fat losers, and the other half is just arguments from ‘my opponent believes something.’” The rejection of rationalism on the grounds of ‘believing in using hypotheticals’ and ‘grappling with the question of how to make the world a better place’ is seen as extreme and inconsistent, whilst the combative language used offers little toward illuminating what is actually incorrect or untruthful regarding the rationalists.
Star does partially concede on particular problems—seeing the usage of jargon as a problem—but also notes that significant participants within the rationalist community have actively worked to make their points easier to grasp, particularly when compared to alternative academic affair. Ultimately, Star sees rationalism less as this disembodied quantification, but as a self-correcting movement in favor of truth that uses technique for the purpose of exploring the various pros and cons of moral positions.
Whilst Mr. Star’s critique stands, I do find Mr. Ghostwind’s description of the rationalist to be more compelling regarding people’s exterior experience of this group. One does encounter people who claim to high rationality whilst being utterly dismissive of alternative patterns of argument. It is these people that warrant criticism—though it is also the case that such people constitute a stereotype. And whilst stereotypes tend to be accurate, they often ignore particular sub-groups who avoid the main problems. To those sub-groups, if they be, I mean no disrespect and welcome dialogue on what is outside my vantage.
However, of that stereotypical rationalism, the exterior feeling has become one of disdain. The rationalist damns the pursuit of morality to being a machine, as the techniques of model construction from mathematics and physics are applied to ethics. The usage of these techniques leaves me with a bad taste on my tongue—they easily damn humans for being a little too human. As Dostoevsky said, “He is a man and not a piano-key!” It is from this perspective that I look at rationalism—for what is problematic, and what may be noble.
Starting with criticism, there is something ungirt about utilitarianism in the sense that it need not interact with, model, or assist human affairs. The operation of morality in ‘ethics’ is something done by humans at the behest of the supernatural—an immaterial calling that offers humanity a higher standard of being than that of animal or machine. To act ethically demands a person to be more than pig or calculator, but instead to focus emotions and follow ideals towards uplift. To disconnect from that, to try and position morality as purely algorithmic, comes across as holding preference to a simpler standard of being against the supernatural; It is to see the calculator as luckier in that it only needs to perform its simple function to be good. Utilitarianism seeks to have morality be calculus of preferences. It wants morality to be absolute. However, morality is not absolute.
To illustrate this point using a hypothetical: under ideal Utilitarianism, we should be able to quantify value, thus allowing us to model a map for moral actions. In the case that we can create such a map, we should also be able to create a machine that could take this map and conduct optimal moral actions.
However, it is known that machines require a set of computational resources to operate successfully. And it is also known that if the amount of computational resources required exceeds a specified amount that is conceivable to be obtained, it is declared trans-computational, being impossible to compute outside of a miracle.
It is further known that even very simple games—like chess—are trans-computational. If our moral machine were to enter a situation wherein the optimal moral action requires that it win a game of chess, it would be unable to. It would be even worse if it were to run into a situation wherein it must win an uncomputable game to conduct moral action.
Given the complexity of the universe, it is extraordinarily likely that optimal moral action would arrive at the need to win some simple trans-computational game. It is therefore the case that the quantification of value in practice is extraordinarily unlikely to offer moral absolution at all times. In fact, I would contend that since moral action lends itself to a high degree of inter-relation between one case and another, the existence of a trans-computational game would lend to the pollution of a great deal of hypotheticals.
Of course, I acknowledge that such a machine can most certainly approximate moral action—but approximation demands trade-offs between accuracy and computational resources. At the very least, the utilitarian must continuously contend with this approximation, so long as they base value on quantifiable elements. In truth, I have said little to a realist rationalist, except to remind them that morality is approximate—but to the idealist rationalist who seeks out absolutes, they will find that the negative results of algorithmic techniques apply to their morality just as much as the positive ones.
Furthermore, I strongly question a key element of common utility—namely, the property of enumerability: or being in one-to-one correspondence with the counting numbers, which thereby allows a moral outcome to be considered 10× better or 10× worse than another. The standard justification for utility being structurally numerical relies on applying the transitive property to moral statements—i.e., if one contends that inaction is better than weak action, and strong action is better than weak action, then it must follow that strong action is better than inaction.
While this transitive property may hold in trite or idealized cases, in real-world scenarios it is not unfounded for morality to collapse into a structure more akin to a game of rock, paper, scissors: strong action may be better than weak action, weak action superior to inaction, and inaction greater than strong action; Rock beats scissors, scissors beat paper, paper beats rock.
It is such the case that in hypotheticals contending between more than two moral options, the numerical tools that allow for evaluating moral outcomes begin to be exposed to questions regarding justification, especially when being used to establish a supreme moral preference. The usage of numbers quickly becomes obfuscatory in the implications of accepted moral axioms, which cuts into the exact purpose of utilitarianism. What is the point of utility-based analysis—or the application of technique to morality—if common situations quickly complicate the analysis to the point of confusing the would-be moral man?
Of course, none of this criticism is new to a seasoned utilitarian. It was already acknowledged at the inception of Game Theory by John von Neumann—whom is contended as a MANIAC rather than a rationalist, but is nonetheless an important figure. In his thoroughness, von Neumann proposed alternative frameworks1 with their own implications. Yet the point remains: when the use of technique becomes burdensome in answering critique in common situations, the whole approach begins to lose its plausibility.
However, even accepting that common utility is totally justified in being numerical, there remain issues regarding aggregation that I believe must be considered.
A well-known ethical thought experiment is the comparison of actively killing one person versus passively allowing five to die—the classic trolley problem. For the utilitarian, this is an ideal setup: the scenario can be reduced to a simple numerical exercise. One might frame it as: “Which is lesser? U(Intentional, 1) or U(Unintentional, 5)?” From here, it’s tempting to formalize utility as a function, say:
U(B, X) = a·B + X
where B represents the existence of intention and X the number of lives affected. Plug in the numbers, and voilà—a mathematical solution to a profoundly disturbing moral dilemma.
It’s here that utilitarianism begins to feel either overly reductionist or problematically permissive2. On one hand, in its minimalist form, it collapses morally salient distinctions—like intention—into fixed numerical weights. On the other hand, in its maximalist or “broad” form, it becomes so flexible that any utility function can be constructed:
U(B, X) = a·B + X,
or U(B, X) = B·X,
or U(B, X) = a·sin(b·B) + B·X²,
and so on. Without additional philosophical constraints, this system allows nearly any aggregation method—turning utility into little more than a placeholder for arbitrary formalization. The very pliability of utility undermines its claim to be a rigorous moral metric.3
This applies to a significant problem that I have with Shrimp Welfare, a cause commonly associated with utilitarian ethics. I could model a utility function that accepts the negative utility of killing a basket of humans and shrimp to be something simple like:
U(H, S) = –(H + n·S)
where H is the number of humans killed, S the number of shrimp killed, and n some adjustment factor for shrimp lives—or I can do something more complex, like:
U(H, S) = –(H + n·log(S)),
wherein shrimp lives in aggregate don’t matter significantly more than that of a single shrimp. This utility function even has the property that each additional shrimp life claimed is still negative!
The reduction of moral decision-making to a single number tends to push the real moral motivation back into a set of complex constraints on what constitutes a good utility function—constraints that are difficult to argue without writing mathematical notation or, at the very least, getting some AI to do it for you. Generally, people can smell something is up with the math based solely on the conclusions, but are otherwise left helpless and frustrated against the utilitarian—so much so that it would be preferable to find some way of “escaping” the confines of rationality than operating within it4.
Leaving aside the details of Utilitarianism, it is also worth addressing rationalism more broadly—particularly as it appears without utilitarian commitments, whether in the form of non-standard variants like rule utilitarianism or through the adoption of entirely alternative moral frameworks. Even in these cases, the rationalist may still be characterized as a disembodied cynic, seeking to construct a model of morality while rejecting strict quantification. To such rationalists, I pose the question: Is a singular, universal model of morality even possible?
I find value in the work of the epistemologist Jean Piaget, especially for his contributions to pedagogy. His studies in moral development, particularly in children, offer a compelling account of how moral understanding emerges over time and within specific cognitive constraints. Notably, Piaget shows that grasping even basic moral concepts—such as lying—depends on developmental stages that are passed through sequentially. While a rationalist might attempt to qualify each of these stages or integrate them into a broader framework, it raises serious questions about the feasibility of any universal moral model that does not account for the developmental and situated nature of moral cognition.
In fact, it is easier to conceptualize the existence of each developmental stage as its own moral model rather than imagining a universal one with the forces of transition being consistently redefined as one moves from stage to stage. The universal moral model, in a sense, is impossible to conceptualize without declaring oneself to be at the final stage—a claim that is anything but evident in our current paradigm. Even the sequential quality of developmental stages may itself be called into serious question as development unfolds.
The stereotypical rationalist, characterized in hubris regarding universal morality, encounters this problem again and again in facing people of normal morality. Rationalism’s understanding of ordinary morality appears thin, with little to say except that of the banality of evil, the animality of man, or the tastelessness of ordinary conversation. The rationalists are no doubt correct on the existence of these things, but their ever-presence does not hold observationally. There exist small contradictions that escape all communicable models. These local contradictions act as the dividing line between good moral behavior and that of evil. To understand them, one has to rely on close observation of particular men and their struggles, rather than aggregates.
This isn’t to say the rationalist hasn’t discovered some elements of normal morality that are important. Consider, for example, Long-Termism, wherein the improvement of behavioral and material conditions is built off of deferring pleasure today, particularly as it relates to the health and wellness of future generations. The commitment of rationalist thinkers to work on the long-term problems of the future against the immediate self-satisfying ones of today is noble, however, they do not hold a monopoly on this observation. Long-Termism has been the basis of religious order from the foundations of civilization, evident physically in the construction of cathedrals and mentally in the production and distribution of holy books. Long-Termism is evident wherein one sees civilization. As much so as the banality of evil, so too is the banality of long-term thinking.
However, whilst someone like Mr. Ghostwind might consider the reinvention of Long-Termism to be a mark against the rationalist, it is my humble opinion that reinvention is, in fact, a virtue. What is not a virtue is the lack of humility oft seen in the re-inventors, status-gaming their own reinvention. This has been epitomized by a rationalist sub-group known as the effective altruists, whose advocacy of Long-Termism has become associated with the scamming of our present society.
The virtue of reinvention comes not from being smart enough to arrive at a useful concept, but rather from striving to do good with what you have, where you can—regardless of whether it has been done before—all whilst holding the necessary humility to understand that such virtue is not exclusive to you, but in-waiting throughout normal society. The stereotypical rationalist account of normal morality fails insofar as it does not recognize those outside of them to be similarly capable of striving toward goodness—even if what it means to be good is different for them than for the rationalist.
Thus far, I have criticized rationalism for being absolutist, obfuscatory, pliable, disembodied, cynical, and hubristic. However, I also wish to recognize certain elements within rationalism that I believe are good, or that I wish to see more of.
For one, the rationalist has resisted the call of the edge case in falling into nihilism. Whilst the rationalist is certainly drawn to edge cases within hypotheticals, they maintain a commitment to truth such that exceptions do not collapse the entire construct. The rationalist believes in morality even against the dragon’s tail of exceptions. Simply put: the rationalist is not a nihilist.
Whilst my criticism—especially through moral pluralism—could be malignly interpreted as sympathetic to nihilism, it does no such thing. It merely points to the reality of development creating separate moral universes within normal society. I am continuously frustrated by the utter degeneracy of the empty-thinking nihilist who rejects any concept of reconciling values on a consistent basis. The rationalist is superior to the nihilist in that the rationalist can lose on their own terms. It is this quality of lossability that makes for a worthy opposition—and is minimal for discussing ethics, philosophy, and art. The rationalist deserves credit for not taking the nihilist’s stance, despite their shared appreciation for the edge case, which tends to causing messes.
Another element I find commendable in rationalism is one it shares—perhaps unexpectedly—with the cynics: an internal Dionysian spirit. At its best, there is something deeply disruptive about the rationalist, whilst nonetheless remaining festive over riotous. The number itself is profoundly Dionysian—intoxicating preexisting social order when properly applied. Mathematics, technology, and the sciences have long demonstrated this trait: overriding problems through the sheer force of descriptive clarity. The solutions found have had miraculous effect, albeit after a wild party within pages of arcane symbols schematizing the physical future.
Yet unlike the cynic, the rationalist is trained in the responsible repression of the Dionysian—often to the point that they can disguise themselves among the Apollonian literati. There is a form to their madness that has lent itself to trust—perhaps too much trust—amongst the civilization’s protectorate. The rationalist is given significant room to operate upon the social fabric, even being allowed to replace the infrastructure of core media which nudges the system forward. In this, the rationalist services humanity by providing optimality within the well-structured. Ultimately, it is the moral domain where the rationalist is chastised—for here, their calculus offends. The moral demands more than what the Dionysian spirit of the number can muster. The rationalist does not tread righteousness well, but when it comes to amoral problem-solving, none may have more virtue than the rationalist.
It is to these virtues that I return to my original statement:
“I am not a rationalist – at least, I think I am not a rationalist; it is honestly hard to say. What even is a rationalist? And why do I even care about whether I am or not?”
I am not a rationalist because I believe the application of exterior technique to morality is insufficient. I prefer an understanding rooted in pluralism, embodiment, humility, clarity, and the proper Dionysian. To me, it is the striving in-and-of itself that constitutes the ethical in hope that something is looking out for me and the remainder of normal society. The moral domain needs intuition, experience, and tradition to lead it, rather than technique, models, and hypotheticals.
The reason there is doubt regarding my identity in relation to rationalism is because I do see positives: in reinvention, in anti-nihilism, and in the disguised Dionysian nature of the rationalist. Critics like Mr. Ghostwind or supporters like Mr. Star might see those characteristics as inclusive of rationalism, and my critiques as non-disqualifying. Perhaps by some alternative, expanded definition, I would be considered one.
But to my mind, rationalism has become a negative movement—one that promotes obscure, quantified thinking among a select few who have taken charge of the future. The movement feels impenetrable to outsiders who haven’t dedicated significant brain-power – myself included -- to parsing the proper tomes, while also giving off concerning vibes that are likely to have real-world effect. Whether or not I am a rationalist by some technical definition, I remain uneasy from an exterior perspective. I think the techniques of rationality should stick to what they are good at, and avoid the moral, for fear of rationalizing controversy.
The alternative theory is based on having utility be represented by vectors, forming a graph. Ultimately, the question becomes one between that of cycles or ends that escape from those cycles. Interesting stuff.
I want to address a quick critique: It is the case that I push a lot of utilitarianism not on the basis of the linguistic, but the mathematics of utility. I have not read Bentham or Mill, but I understand that they were probably less mathematical than the post-Neumann ‘Game Theorist’ Utilitarians. I don’t spend time on things like impartiality, agent-neutrality, or happiness due to a desire to handle those particular problems on the basis of concrete believers as I lack full confidence in my understanding; I am more familiar with the limits on logic.
I believe that there might be some utilitarians who reject the idea of representing utility for humans as a utility function for similar reasons to those given. They would be more concerned about arguing the properties of morality rather than the representation of how moral an action is. They simply believe utilitarianism because of some desirable quality of the conclusions rather than the form. In such cases, I believe a more individual analysis is warranted, rather than something as abstract as what I provided.
Good stuff — appreciate the write up — look forward to reading future posts
Ethics is rationally based in contingent priorities. To the extent we subjectively share certain priorities, there are objectively better and worse ways to get them.