Who Thought of It

I've already written about which tools built this site. This is the harder question, and the one people actually mean when they ask: how much of Drithane did I think of?

The short answer is all of it, and that answer is doing more work than it looks like. So here is the long one.

What is mine

Every idea in this universe is mine. Not "mine after editing" — mine at the point of invention.

The Cosmic Cascade and its whole hierarchy. The Concordants and the rule that consciousness couples to the Etheric lawfully rather than magically. Trígrian's trinary suns and the convergent-evolution argument that makes its lioness a lioness without being one. The Federation's founding principle that a claim is valid without the claimant first proving itself familiar. The decision that there is no faster-than-light communication, only faster-than-light travel — which quietly determines half the politics in the setting. The rule that there is no pure evil here, ever, and that evil is emergent from structure and appetite rather than essence.

The premise. The characters. The plot. The theology. The names.

And the editorial rulings, which matter more than they sound: that contested material belongs in the Codex and settled fact in the Lore, that the Codex is canon-for-its-author rather than canon, that an overlay is neither. Those are the load-bearing decisions in a project like this, and I made them.

What is the machine's

A great deal, and none of it is ideas.

The engineering: the content-schema validation that fails the build on a malformed file, the scene-and-POV pipeline, the multi-domain content filtering, the deploy scripts. The consistency work — cross-referencing the lore against the glossary against the story bible and telling me where they disagree. The audits. The tidying. Some of the images, both generated and composited.

And historically, some prose. The early Codex passages were worked over in Sudowrite; the early scaffolding leaned on Copilot. I've said that elsewhere and it stays true.

That chapter is now closed. I have pulled the last of the Sudowrite material across — the ideas worth keeping — and archived the rest. The licence runs out and I won't renew it; it was a sunk cost that did its job, which was getting started. From here it is me, my photographs, and Firefly for the art that isn't photographic, with Claude Code working against the repository.

Salvaging that last batch was instructive in a way I did not plan. Almost everything in it that clashed had drifted for the same reason the two copies drifted: it had been written beside the repository instead of against it. A whole star system turned up describing a world the repo had already given to somewhere else. A band I had deliberately kept small came back famous. Nothing in there was wrong when it was written. It had just stopped being reconciled, and nobody noticed, because nothing forces you to notice until you go looking.

Where the line actually falls

Not at "who typed it." At who decided.

The most useful thing an assistant does here is not produce text. It is disagree with me accurately. It tells me that a new image contradicts the alt text that has been sitting under it for a month, or that the entry I am about to write depends on a fact three pages away that says the opposite. A machine that reads the whole corpus faster than I can is genuinely good at that, and I would not want to work without it now.

But every one of those catches ends in a ruling, and the rulings are mine. When an image turned up recently showing a creature that was not the lioness Trígrian's entry is built around, the machine flagged the mismatch. What it could not do was decide whether the fix was a new image or a new canon. That is not a limitation of the tool. That is the job.

The clearest case is the ideas I threw away. I spent a while on alternate timelines and reality overlays as standing canon, and talked myself out of it, because it would break the Survey Archive's single-record authority and the Cascade's one-way causality. Nothing generated that decision. A universe is defined at least as much by what you refuse to let into it, and refusal is not something you can delegate.

The obvious problem with this entry

This one is worth saying out loud, because it would be absurd not to.

A journal entry about how much of my work is mine is precisely the entry where an AI byline would poison the well. I know exactly how this reads if the disclosure is itself undisclosed.

So: the account above is mine, the judgements in it are mine, and I have gone over the words. If a sentence here is wrong about what I think, it is wrong because I let it stand, which makes it my error and not the tool's. That is the same standard I apply to every other page on this site, and I do not think this one gets a special exemption in either direction.

Why it matters here specifically

This site just added a fourth canon status — Chartered — for approved work by another hand. Writing that policy forced the question into the open, because the thing that separates the four statuses is not quality. It is who wrote it and where they stand: the Codex is a named in-universe source, an overlay is a lens laid across the record, a chartered work is new events by another author, endorsed but never binding.

A project that fusses that carefully about attribution inside the fiction cannot be vague about it outside. So the rule for the tooling is the same one that governs a chartered work, pointed the other way: the ideas bind the tools; the tools do not bind the ideas. The machine is fast, well-read, and often right about consistency. It does not get a vote on canon.

There is one place I hold the line absolutely, and it is not this repository. My photography is my own photographs — selected, cropped, and printed with whatever help I want, but never generated and never composited into something I did not stand in front of. Imperfect frames stay imperfect. Here, where the point is the ideas rather than the shutter, a composited sky is fine. There, it would be a lie.

Different work, different rule. Both worth stating plainly.