A directory that had to hold Marcus Aurelius and your accountant

The Mindola directory currently lists Marcus Aurelius, Seneca, Sun Tzu, Jane Austen, Charles Darwin, Nikola Tesla, and a consulting detective who never existed.

It also lists lenses published by real people — a coach, a consultant, someone who teaches a trade — and it will list more of them than classics soon enough.

That’s an odd shelf. Making one page work for both ends of it turned out to be where all the interesting decisions were.

The categories were the whole problem

The first design of Discover sorted lenses the way you’d sort a library: Letters. Invention. Poetry. It looked good. It was completely wrong.

Those categories describe thirteen dead writers. They describe nothing else. The moment a working accountant publishes a lens, there is no shelf for her — “Letters” is not where anyone looks for help with quarterly filings, and renaming the taxonomy later means breaking every URL and every expectation built on it.

So the axis changed to domains of expertise: Business, Leadership, Law, Money, Career, Teaching, Craft & writing, Science & tech, Health, Life & philosophy.

Marcus Aurelius sits under Life & philosophy. Machiavelli sits under Leadership, which he’d have enjoyed. And the accountant has somewhere obvious to be on the day she arrives.

This is the least glamorous decision in the whole build and the one most likely to still be right in three years. A taxonomy that only fits your seed content is a taxonomy you will replace.

Why the classics are all public domain

Every curated lens in the directory is built from writing that is out of copyright. That’s partly the obvious reason. It’s mostly a different one.

A lens is supposed to answer only from its source material and admit when it doesn’t know. That claim is hard to demonstrate, because for most lenses you have no way to check. If a consultant’s lens tells you something, you can’t audit whether it came from her material or from the model’s general knowledge.

With Seneca you can. The letters are right there. You can ask about grief and time and the shortness of life, and then go read what he actually wrote and see whether the answer holds up.

The classics are the only lenses on the platform where a stranger can independently verify the grounding. That makes them a better argument for how this works than anything we could write about it.

They also carry an honest asterisk. Their topic counts and publish dates are design values from the original comp, not measurements — unlike community lenses, which report real numbers. And they have no question telemetry at all, which is why they can’t appear in “Most asked.” That list is real volume from real visitors, so a curated showcase entry has no business in it.

Nobody lands here by default

The directory pulls community lenses from a single endpoint on the app, and that endpoint returns only lenses whose owner has flipped seo_indexable — off by default, per lens.

The part that matters more: that query reads the lens table alone. It never touches the users table. There is no path by which someone who hasn’t published ends up in a public list, because the data required to put them there is never fetched.

If you’ve made a lens and you don’t want it discoverable, you don’t have to do anything. That’s the default.

It’s built at build time, on purpose

The community entries are fetched when the site builds, not when your browser loads the page.

That sounds like an implementation detail and isn’t. This page exists to be found — by search engines, and increasingly by LLM crawlers answering questions on someone else’s surface. A directory that fetches its contents in the browser serves those crawlers an empty list. All the work of listing lenses, and the machines that decide what gets recommended see a page with nothing on it.

Building the list into the served HTML costs freshness — a lens published this morning appears at the next deploy, not instantly. That’s a real trade and we made it deliberately.

There’s a second effect worth naming: the marketing site never holds a database credential. It consumes one narrow, already-public feed. And if that feed is unreachable at build time, the build still succeeds and the page falls back to the classics alone. A directory missing its newest entries is a much better outcome than a failed deploy.

The two spots of colour

Mindola’s site is strictly monochrome. Ink on white, no accents, including inside product mockups. The directory is the only page allowed to break that, in exactly two places.

The verified check is blue. A verified badge means the sources behind a lens have been checked by us, and it’s a trust signal — the one place a reader genuinely expects a colour. Rendered in grey it reads as decoration, and a trust mark that reads as decoration has failed at its only job.

The top three ranks in “Most asked” get gold, silver, bronze. Same reasoning, lower stakes.

Everything else stays ink. Both exceptions are declared as tokens, scoped to that one page, and if the medals ever start feeling loud, setting all three to the faint grey returns the list to strict mono with no other change. That escape hatch was designed in before it was needed, which is the only time you can design one honestly.

What’s still missing

Community lenses arrive uncategorised. The lens table has no category column yet, so a published lens shows up under “All” but under none of the topic chips. It’s listed, it’s searchable, it just isn’t filterable by domain — and the fix is a column plus a picker in lens settings, not a redesign.

We’d rather say that plainly than let you discover it by filtering for Business and wondering why your own lens isn’t there.

Go and look

Open Discover and ask Marcus Aurelius how to stay steady when someone wastes your morning, or Twain how to make a hard truth land as a joke. They answer from what they wrote, and they’ll tell you when a question is outside it.

Then, if you want to be on that shelf: publish yours. It’s one setting, and it’s off until you say otherwise.