Sublimation and Involution
Embodied Mathematics — American Scientist
This is a discourse on the book Where Mathematics Comes From (2000).
What I feel reading this piece bears a resemblance to the disillusionment with the illusion of control that defines the AI generation.
It is the act of packaging explicit natural-scientific facts into subjective, personal vessels and interpreting thought through them. This is similar to what I have previously called isomorphic aphorisms.
When thought cannot be standardized, the instinct is not to see the resulting frustration as born from ignorance, but rather to reinterpret it as “creativity” — an exceptional quality unique to human beings. A frustratingly clever trick because it’s an unquantifiable trait to begin with. This book frames the development of mathematics much as Steven Pinker, in The Language Instinct, framed human language as something intrinsically human: “Oh, eπi + 1 = 0 holds not because of any natural law but because of cognitive reasons, and so on and so forth.” Needless to say, mathematicians have torn this argument to shreds.
The Hallucination of the AI Generation
What, then, is the hallucination — the illusion — of the AI generation?
I do not mean that models hallucinate. What I mean is the resurgence of “involution” (内卷) — a concept repurposed with an entirely different meaning by young people in the Chinese-speaking world.
Involution was originally a term from anthropological sociology, born from the study of why certain social systems self-destruct without change — to the point of resembling an ant mill (the scientific phenomenon in which isolated ants, following only each other’s pheromones, spiral into collapse). This is explained in detail in Agricultural Involution.
A similar phenomenon is sweeping through the AI ecosystem. Within the system of software, the fundamental difficulty is being sublimated — rhetorically as 昇華 (something rising up), chemically as sublimation — into “creativity,” something vague yet grand, only to combust. This cycle repeats.
Ultimately, the origin of ideological human action can be traced to the search for necessity in action. The concept of the “company,” softened and pre-chewed like a mother bird regurgitating food for her young — this is the foundation of modern startup culture. Sublimation at the enterprise level is the act of proceeding without internalizing that necessity. It is the inversion of causality, the tautological assertion: “This will bring innovation to the ecosystem.” But such companies combust and vanish the moment they touch air.
Sublimation (differentiation, positional ascent) and combustion (destruction), repeated. I suspect this is the current state of startup culture surrounding AI and LLMs.
Difficulty as Feature
This system is, without question, functioning on mathematical logic, and the difficulty itself is the key that proves the system’s excellence — feature, not bug. Of course, the system itself is subject to improvement and will improve. But just as the algorithm that surpassed Dijkstra’s was discovered only last year, this progress is incremental and minute. What matters more than speed is the fact that it is hard. This difficulty is the awe that arises when humans confront nature, and it is, in truth, what gives these discoveries their meaning in the first place.
And yet young people like me feel discomfort in that very “difficulty,” and insist on repackaging it into the parochial vessel of human life before redistributing it. The way these redistributed artifacts circulate only within the ecosystem resembles agricultural involution.
Projects like OpenCode, Oh My, OpenClaw, and Moltbot are created — but they can only be understood within the category of “productivity.” They are self-referential proofs of productivity for productivity’s sake, and examples that penetrate deeply into life are hard to find.
The Paradox of Self-Criticism
This position is difficult to approach neutrally — and, surprisingly, rationally.
I myself am a member of this ecosystem (it has been weeks since I last coded without Claude Code; I feel a rush of excitement whenever I see new commands and skills), and to criticize the ecosystem I inhabit seems contradictory and selfish. As though I alone were some kind of prophet, I fall into the self-referential paradox of elitist thinking — isolating myself from the very people I criticize.
This is also difficult because I am also part of the startup ecosystem, hoping to be enacting a modicum of positive change to the world. And I want to believe that it is the case that there are very very very few but select companies that managed to do that somewhat ethically. Google and Apple could be considered to be on the pedestal, but I do take the responsibility of making the delineation of ethics blurry for the sake of a larger argument. I don’t want to use the word “good” and “ethics” to come from my shallow understanding of social impact, but only to acknowledge their genuine excellence as companies. Google and the iPhone can be said that their necessity was proven as they were borne ex nihilo — ontologically true because for something to exist it HAS to exist on its own — from… from what, exactly? Genuine hard work??
“There is a central quality which is the root criterion of life and spirit in a man, a town, a building, or a wilderness. This quality is objective and precise, but it cannot be named.”
— Christopher Alexander, The Timeless Way of Building
Obviously people claim to have found the scientific evidence in such success even though these products and companies have been only around for 30 years. This translates to the difficulty that lies in the inability to read my own fortune; how can I be so sure that the difficult problem I’m solving is meaningful and impactful to not contribute to the involution but is a genuine revolution.
Epistemic Iteration
In Inventing Temperature, Professor Hasok Chang interprets this kind of calibration as a natural stage of scientific development, explaining that through epistemic iteration, this “chaos” can gradually be restored to order. (This interpretation may well be colored by my own misunderstanding.)
In some ways, language and dishwashing are comparable. Even if the dishwater is dirty and the cloth is dirty, we still somehow manage to get the plates and glasses clean in the end. Likewise, we have terms that are not clear to us, and logic that is limited in ways we do not understand when language is applied. Even so, we have successfully used language to clean our understanding of nature.
— Niels Bohr, 1933 (as quoted by Werner Heisenberg in Physics and Beyond, 1971)
Essentiality
And so I think: what is essential? This essentiality must
- be directed from ignorance toward understanding,
- possess universality,
- prevent logical leaps through “creative” evasion of friction, and
- simultaneously contain “creativity” within itself.
In the end, I tentatively think that this essentiality is nothing other than the pursuit of science itself.