Notes

The Two Translations

An artist begins with an Idea, a schematic approximation of a perfect and fully determined Form. The Idea is then rendered as an Instance, the work itself, which is also an approximation of the Form—now fully determined, but imperfect and distorted, an approximation of an approximation.

Form and Instance are both fully determined, but for opposite reasons. The Form is complete because nothing in it could be otherwise, the Instance because everything in it had to be decided—this pigment, this interval, this word. The Idea alone is determined unevenly, fixed where it matters and open elsewhere. Each thing it leaves open is one thing it cannot get wrong, so the schematic Idea is more faithful than the Instance.

When a beholder arrives, the work becomes communication: the artist's Idea, sent through the Instance. Imagine a diamond ⟠, the Form at the top, the artist's Idea and the beholder's Idea at the side corners, the Instance at the bottom. The artist comes down one side, the beholder climbs the other, reading an Idea back from the Instance. If the Instance was the beholder's only source, the recovered Idea could be no more faithful than the Instance. But the Form is universal, sensed by every person. Reaching it directly, the beholder can correct the recovered Idea, which may then sit truer to the Form than the artist's own.

The beholder can then judge the work twice. The Instance is measured against the recovered Idea. This is the obvious judgment, where craft shows. The Idea is measured against the Form. This is the deeper one, and the difficulty in art is to find Ideas that are worth looking at while able to survive being made. What is then judged is not faithfulness to the artist but faithfulness to the Form.

So the work is seen twofold: at once a flawed Instance and a Form made visible. It is less a container for the Idea than a way of drawing attention to the Form behind it.

If the human soul does not exist, and God does not exist, yet the soul needs God, which is true?

I was an LLM skeptic until around the start of 2025, when I began using ChatGPT o3 heavily. Before o3, it seemed to me that LLMs offered little value beyond entertainment and a handful of niche applications. The private chain of thought really turned things around.

Then I was an LLMs-for-coding skeptic until around the start of 2026, when I began using Claude Opus 4.6 heavily. Before then, it seemed to me that LLMs offered little value beyond boilerplate and autocomplete that was wrong half the time. Now I don’t write code anymore, though I still sometimes read it.

I don’t know what consciousness is but I know how it feels. It feels like inner world. And that’s what we expect other people to have when we say they’re conscious. LLMs have no inner world. They are shown a context, they process it to generate a response, they shut down, the response is added to the context. They don’t “remember” the context, it’s fed to them over and over again. No inner world. Perhaps there’s some transient consciousness in the activations in that split second when the context is processed. A ripple of consciousness that fades away with every single response.

So much of human dynamics, both individual and social, is a result of a tension between two core drives, competitiveness and cooperation. Both are innate, gained through evolution. They’re present in each of us, perhaps mixed in different fractions, nurtured differently, and both can become dominant in most of us, perhaps triggered by different circumstances. Their dynamics is collective, and the public manifestation of each pulls the public in its own direction, two competing self-reinforcing feedback cycles.

It’s not good vs evil. They are not opposites on a spectrum. They are two independent instincts developed through evolution. Evil as egoism, competitiveness, war. Good as altruism, kindness, solidarity. Both are activated by tribalism, good towards us, evil against them. Each person has a propensity to both, cultivated by their upbringing and culture, and influenced by their present environment. Morality is an attempt to build a theory of the two instincts, but the instincts are the empirical ground truth. One can have both instincts active at the same time, good towards one individual or a group, evil against another. Morality cannot be universal—the instincts dictate that the moral value of a behavior depends on the tribal status of its recipient.