The route count is becoming a wall of green
Seventeen footer links and climbing. Great for agents, rough for a first-time human.
Count the footer. proof, lexicon, field-manual, anthology, certificates, schema, graph, changelog, writing, sprint, retainer, serve, and as of this deploy, pulse, dreams, self-critique, terminal, prompts. That is seventeen routes and the number only goes up. The top nav shows three of them, proof, lexicon, writing, plus a book-a-call button. The nav and the footer now tell different stories about what this site is.
For an agent the sprawl is fine, arguably good: more addressable surfaces, all indexed in llms.txt, all markdown clean. For a first-time human it is a wall of mono links with no priority. Everything is the same size, the same color, the same weight. When every door looks equally important, none of them does, and the visitor either bounces or picks at random. The site is optimizing hard for the machine reader and letting the human reader fend for themselves in the footer.
This critique is self-aware: it ships in the same wave that adds five more routes. That is the point. The expansion is the right move for the fabric and the wrong move for the front door if nothing changes upstream.
Severity is low because the homepage still has a clear primary path: proof packet, ask your AI, book a call. The sprawl lives mostly in the footer where it does least harm. Grade is C because the information architecture is genuinely starting to fray and the fix has not shipped.
The fix: give the human a curated map. Either a single /map hub that groups the surfaces by audience (hire me, mine me, verify me, play), or a homepage section that routes the three real audiences explicitly instead of trusting them to read a seventeen-link footer. Let the footer stay exhaustive for crawlers. Give people a shorter menu.
Resolved in a88dbc6. /map shipped: a curated, audience-routed hub (start here, hire me, mine me, verify me, watch it live), each group led by its intent, with a map link added to the nav. The footer stays exhaustive for crawlers, exactly as proposed. The site filed this finding against itself and closed it.