The Digital Trail That Speaks For Itself

I took over a project this month from an agency charging a lot of money for not doing very much. The migration they’d spent five months on subsequently shipped in two weeks. Not because I worked harder — because every change was tracked, every commit dated, every decision in the trail. The client could see the difference without me explaining it.

That’s when I realised: this isn’t just a digital agency problem. It’s every profession that charges for opaque expertise.

What did your solicitor actually do last month? They kept a timesheet. But where’s the work? What did your accountant produce between February and the filing deadline? They’ll tell you they were “working on the accounts.” Show me. What did your SEO agency do with that monthly retainer? They’ll send a report. But the report is a summary of claimed activity, not a verifiable trail of actual work.

Jarndyce v Jarndyce

Dickens saw this in 1852.

Bleak House opens with a legal case — Jarndyce v Jarndyce — that has been running through the Court of Chancery for so long that nobody alive remembers how it started. Lawyers have billed it for generations. The fees have compounded. The case has passed through so many hands that the work itself has become untraceable. When the inheritance finally lands, there’s nothing left. The costs consumed the entire estate.

That’s the billing model for every profession that charges by the hour for work the client can’t verify. The legal system, the accounting profession, the agency world — they’ve all operated in the space between “we did the work” and “prove it.” Dickens wrote Bleak House because the Court of Chancery was a national scandal. The scandal wasn’t corruption. It was opacity. Nobody could see what was happening inside, and the people inside had no incentive to let anyone look.

A hundred and seventy-three years later, the mechanism hasn’t changed. What’s changed is that the trail is no longer optional.

The True Name

Le Guin gave us the vocabulary for this in 1968.

In A Wizard of Earthsea, everything has two names. The use-name — what the world calls it — and the true name, which is its actual nature. Knowing the true name of a thing gives you power over it. Not magical power. Accurate power. You see the thing as it actually is, not as it presents itself.

A monthly report is a use-name. It’s the surface presentation — “we optimised your site, we improved your rankings, we worked on your SEO.” It might be accurate. It might not. The client has no way to tell, because the use-name is all they see.

A git log is a true name. It’s the actual record of what happened: what changed, when, in what order, and why. You can’t fake a commit date. You can’t fabricate a diff. The trail is the work made legible — not a summary of claimed activity, but the activity itself, rendered in plain text.

Le Guin understood that the distinction between true names and use-names isn’t about honesty or dishonesty. It’s about legibility. The use-name serves a social function — it’s how things are known in public. The true name serves a structural function — it’s how things actually are. Both exist simultaneously. But the person who knows the true name sees what the person who only knows the use-name cannot.

AI reads true names. That’s what it does. Feed it the use-name — the monthly report — and it will summarise it politely. Feed it the true name — the version history, the commit trail, the actual changes — and it will tell you what happened. The gap between the two is where the uncomfortable questions live.

What This Means

Anti-Maxwell is one of the principles I built this practice on. I adopted it by choice. What I’m watching now is that principle becoming the market’s minimum expectation — not because transparency became a virtue, but because the technology made it unavoidable.

The professions that charge for opaque expertise — legal, accounting, consulting, agencies — have operated for centuries in the space Dickens described. The work happens behind closed doors. The client sees a report, an invoice, a summary. The trail stays inside. That model worked when the only way to verify the work was to hire another expert to check it.

AI changed the economics of verification. A client with Claude Code can search every page on their website and ask “what changed this month?” and get an answer in seconds. Not a summary. The actual changes. Dated, diffable, verifiable. The Jarndyce model breaks when the client can see inside the Chancery.

The businesses that already have the trail — the ones that committed every change, documented every decision, built the fond of faithful daily work — are the ones that answer the question without hesitation. The rest explain why they can’t.

Related: The Correction Loop · The Atomic Commit · Where Principles Come From · The LLM Test · Ingeniculture

If you want this kind of thinking applied to your business — here’s how I work with clients, or get in touch.