Duy Nguyen

"We limit ourselves to the attainable, and this means the renunciation of all other potentialities. One man loses a valuable piece of his past, another a valuable piece of his future."

— C.G. Jung, The Stages of Life

What is possible in a lifetime?

I've always asked this question. Like David Deutsch.

In The Fabric of Reality, he opens with a thought he had as a child and never gave up on — that every modern child is told the same thing: in ancient times a single learned person could know everything that was known, but we have moved past such a world; there is now too much knowledge for any one mind.

Like him, I found this surprising and disappointing, and simply did not want it to be true.

Deutsch spent most of the book dismantling it, and showing me his answer.

Knowing, he insisted, is not memorizing — he had no wish to track "the fall of every sparrow." He meant understanding: having the right explanations. One simple theory, he writes, can cover "an infinity of indigestible facts."

An AI researcher cannot recall every training technique, but understands the bitter scaling behind Chinchilla and test-time compute. New theories absorb their predecessors, just like Newton subsumed into Einstein — "fewer, deeper and more general." As for Deutsch, it was Popper, Dawkins, Turing, Everett. As for me, I'd try to weave it through this tiny space I occupy on the web. But the essence is: as knowledge grows, its structure gets simpler, not harder, to hold.

"The proposition that I refused to believe as a child," Deutsch writes, "is indeed false, and practically the opposite is true. We are not heading away from a state in which one person could understand everything that is understood, but towards it."

I grew up in Vietnam, then Oregon, and eventually arrived at Stanford.

At Stanford, and maybe in life, we inherit other people’s ambitions and desires before noticing they are not ours. Girard called this mimesis — the way desire is often borrowed before it is chosen, then rationalized as our own. For me, that borrowed desire took the shape of computer science.

Computer science trained me in what Strauss called the knowledge of homogeneity: systems, repeatability, decomposition, the instinct to make things "identical", "interchangeable", "measurable". At Stanford, this often means being trained into a kind of technocracy — learning how to optimize before asking what is worth optimizing for.

But studying AI/RLVR scaling radicalized me (you can see where the future is headed). It loosened the shackle I did not know I was wearing: that I had to dedicate myself to one narrow form of competence to matter.

Strauss thought we were stuck between two kinds of knowledge - homogeneity, the knowledge of mathematics, and heterogeneity, the knowledge of ends, purposes, and qualitatively different goods. The danger is to absolutize one side - to become a technocrat who reduces life to what can be measured, or a mystic who abandons the discipline of reality.

In that future, I see the needs of inheriting both.

The duality is not abstract to me anymore.

Homogeneity, to me, is the knowledge of means: mathematics, models, mechanism, optimization, the power to make things repeatable and scalable. Heterogeneity is the knowledge of ends: purposes, judgment, desire, education, the soul, the things that aren't interchangeable.

Postmodernity trains us toward the first, and AI made the first astonishingly powerful but also less sufficient as a life. If competence becomes something we can amplify, automate, and distribute, then the harder question is not merely how to do more, but what is worth doing.

That is where philosophy returns. Not as an escape from technical reality, but as the discipline that refuses to "absolutize" either pole. Maybe moderation is not cowardice, but the virtue that controls philosophical speech (saying enough for the right reader to continue).

This is the duality I want to think through: how to inherit the power of homogeneity without surrendering the soul to it, and how to pursue the knowledge of ends without abandoning the discipline of reality.

The world rewards curiosity. The gentlement scientist model is so back. Play long-term positive-sum games. Shoot beyond the stars.