Old books have a page at the back called the colophon. It lists the typeface, the paper stock, the printing method, the bindery. “Set in Garamond Premier Pro. Printed on 80# Finch Opaque. Smyth-sewn and casebound by Worzalla.” The colophon is a transparency document. It says: here is exactly how this object was made. You can inspect the materials and the methods. Nothing is hidden.

I keep thinking about what the colophon for this site would look like.

The technical stack (Jekyll, GitHub Pages, SCSS, Markdown) is the paper stock. It tells you what the site is made of but nothing about how it was made. The real colophon would list the methodology. The voice protocol that governs every sentence. The evaluation lenses that test every page against multiple frameworks simultaneously. The connections architecture that maps the 77 pages into a navigable web of relationships. The sculpting loop where raw conversation material gets refined through structural, narrative, and voice passes until the output is indistinguishable from hand-written prose.

Here’s the thing about this particular colophon: it already exists on the site. The FormWork page describes the coordination harness. The voice governance essay explains how the voice protocol works. Lens extraction covers the evaluation framework. SavePoint documents the protocol that captures cognitive turns during the work itself.

The methodology I used to build the site is described by the site. The site was compiled by the methodology it describes. The colophon and the content are the same thing.

This circularity isn’t accidental. It’s the strongest possible form of proof.

When a consulting firm tells you they have a methodology, they show you a slide deck. Maybe a case study. They describe the process, they name the phases, they reference past client outcomes. You take their word for it. The methodology and the proof of the methodology are separate artifacts, which means you’re trusting the description.

When the methodology produces the description, the trust relationship changes. You’re not reading about a process and hoping it works. You’re reading the output of the process and deciding whether the output is good. If the writing is clear, specific, and sounds like a human practitioner who’s actually done the work, the methodology proved itself on the page you just read. If it doesn’t, the methodology failed in real time, and you can see the failure.

I find this more honest than a portfolio of client logos. Logos tell you someone hired you. They don’t tell you whether your methodology held up under pressure or whether the output was any good. A self-compiling site tells you exactly one thing: whether the system produces results that meet its own standards. The reader is the evaluator. The work is the evidence.

The voice protocol is a good example. It has twelve checkpoints. Zero em dashes. Zero banned words. No personification of tools. No fortune-cookie closers. Practitioner register throughout. These rules exist because they solve specific problems I identified in AI-generated copy: the generic feel, the motivational-speaker cadence, the tendency to smooth every rough edge into corporate polish. If the rules work, the prose sounds like me. If they fail, the prose sounds like a language model, and you’d notice.

Every page on this site has been through the voice protocol. The protocol is described on the site. If the description reads like a real person wrote it, the protocol did its job. If it reads like a machine trying to sound human, the protocol failed, and the failure is visible in the very document that claims the protocol works.

The evaluation lenses create a similar loop. The lens array runs multiple frameworks against a page: structural integrity, narrative clarity, voice fidelity, positioning alignment. These lenses are described on the site. The pages that describe them have been evaluated by them. If the descriptions are structurally sound, narratively clear, voice-faithful, and positioning-aligned, the lenses validated their own descriptions. If the descriptions are muddled, the lenses missed something, and the failure is self-evident.

I think the reason traditional colophons are satisfying is that they represent a commitment to transparency that most producers don’t make. Telling you the paper stock and the binding method is an act of confidence. It says: here are the materials, judge for yourself. The methodology colophon extends that same confidence to the intellectual production. Here is the voice protocol, here are the evaluation lenses, here is the coordination harness, here is the knowledge system. Judge for yourself.

The circular proof has a vulnerability. If the methodology is flawed in a way that’s invisible to the methodology’s own evaluation tools, the site will look fine and the flaw will be hidden. Blind spots don’t surface in self-evaluation. This is why I use external evaluators (Gemini reviews with no prior context, practitioners who aren’t part of the system) as a check on the loop. But even that check is described on the site. The Site Is the Proof documents the external evaluation. The transparency extends to the limitations of the transparency.

A colophon traditionally lives at the back of the book. It’s the last thing you encounter, a closing disclosure. On this site, the colophon is distributed across every page. There’s no single disclosure moment because every page is already showing the methodology at work.