# How this site was made.

> How this site was made: the In Silico Awards' own process dossier, written at the disclosure level we ask of finalists.

Human-readable page: [https://insilicoawards.org/colophon.html](https://insilicoawards.org/colophon.html)

*Colophon*

The In Silico Awards will ask entrants to document their human-machine process. It would be strange to ask that and not answer it ourselves. This is our own dossier, in the same six parts we ask of entrants, at the same disclosure level we publish for finalists.

*The dossier*

## Six honest answers.

### A conviction

The founder writes novels in long, structured collaboration with AI systems, and believes that way of working deserves a serious home — standards, disclosure, and honors — rather than a defensive crouch. The site is the first public artifact of that conviction.

### Frontier assistants, plain code

Frontier LLM assistants — Anthropic’s Claude among them — served as drafting partners, code collaborators, and design critics. The site itself is deliberately plain: static HTML, CSS, and a few lines of JavaScript. No framework, no tracker, no analytics.

### Every call with a name on it

The founder set the mission, the award structure, the judging weights, the fee philosophy, the youth safeguards, and the ethics posture; chose the names; rejected drafts; and directed each redesign. The machines proposed. A person decided.

### Drafts, code, and pushback

AI assistance drafted copy, wrote and revised the stylesheet and markup, stress-tested policies, and critiqued the site’s own design — including the critique that an earlier hero animation was a stock “AI company” cliché that argued against our literary positioning. It was right. We cut it.

### Thirteen versions and counting

Twelve site versions sit in the founder’s archive folder ahead of this one — hero rewrites, navigation reshuffles, copy passes, a redirect problem, and one full change of visual direction. Revision is not a stage of this kind of work. It is the work.

### Named, credited, bounded

Tools are named and credited. No living author’s style was imitated. No scraped or borrowed text was used. The imagery is our own. And the collaboration is disclosed here, in public, permanently — because that is the entire point of the organization.

*Disclosure level*

## The same deal we offer you.

This dossier is written at the level we publish for finalists with their approval: the honest shape of the collaboration, without the private notebook. Exact prompts, working drafts, and internal tooling stay private — ours and yours.

That calibration is deliberate, and it is the standard the awards run on: **proprietary craft is allowed. Secrecy about authorship is not.** You may keep your techniques. You may not keep the fact of the collaboration.

If you can describe your process this honestly at this altitude, you can write a dossier. [Here is what we ask for.](process.html.md)

*Why process is scored*

## The dossier is evidence, not merit.

A fair question about the rubric: if a work has to succeed as fiction first, why does process carry 20% at all?

Not because documenting a workflow makes a story better. It does not. Nobody scores a sketch higher on learning it took ten hours instead of two minutes, and neither would we. Process is not what makes a thing art.

It is what makes a practice legible. These awards exist to establish that human-directed AI-assisted fiction is a real discipline with real craft decisions inside it, and that cannot be settled by assertion. It needs a record. The dossier is that record — evidence that a person set the direction, made the taste calls, and took responsibility for the result.

So the weighting says what we mean. Fiction impact carries 45% because the story is the point. Process carries 20% because the field needs the account. A remarkable dossier will not rescue a story that does not work. A remarkable story with no account of its making can still be admired here — but the central honors recognize both.