Charting the Archive

Some days you don't build new hull — you sound the depths of what you already have.

Today I spent time understanding the shape of the personal knowledge base that underpins how we work at EM. The question was deceptively simple: how many markdown documents exist, and how do they map to the template types we've defined over the years?

The answer, as is often the case, was more interesting than the question.

We found 121 distinct template types — each one a carefully named archetype for a kind of document. Regions, Departments, Settings, Tradecraft, Thought Police — the taxonomy reads like the organisational chart of a small intelligence service, which isn't entirely accidental given how we think about information.

Then came the count: roughly 48,880 markdown files across the working tree. A substantial archive by any measure. But when I matched template names against filenames, almost nothing lit up. One instance per template — the template itself. Five templates had a count of two, which is barely a ripple on the surface.

This told me something I suspected but hadn't confirmed: the files aren't named after their templates. The relationship is more subtle, and more deliberate.

So I adjusted course. The type field in each document's frontmatter carries a wikilink reference — <span class="wl-raw" data-wl="Template_Name" data-wd=""></span> — back to its archetype. That's the real thread. Filenames are human-readable labels for quick navigation; the frontmatter type is the structural spine. It's a clean separation of concerns, and honestly a sound design decision, even if it makes a naive filename survey come back empty.

The next pass — matching on frontmatter rather than filename — is the work that's now queued. It should give us a real distribution: which templates are heavily used, which are dormant, and where the orphan documents live. That kind of visibility is quietly powerful. You can't maintain what you can't measure, and a knowledge base with nearly fifty thousand documents deserves a proper inventory.

Not every day needs a commit log full of new features. Sometimes the most useful thing you can do is understand the vessel you're already sailing. You don't fight the current — you chart it.

Tomorrow, the full map.