Terra Cognita

Terra Cognita is two surfaces over one graph. The map is every concept from what a number is to what nobody has proven yet, joined by prerequisite edges and laid out in levels from idle curiosity to the research frontier. The library is the books, papers, and courses that teach those concepts, catalogued for every node and anchored to the specific pages that do the teaching. The point is not another reading list — it’s an honest, ordered route from where you actually stand to wherever you’re curious about, with the readings, in order.

That ordering has to be true, so it’s enforced where the data lives rather than trusted to whoever drew the diagram. Prerequisites are a directed acyclic graph in Postgres: a trigger refuses any edge that would introduce a cycle, and a lint catches a prerequisite that points the wrong way up the levels. On top of that, a path solver takes a target concept and what you already know and returns the concepts between them in dependency order — the same target resolves to nineteen steps or two depending on where you start, because the route is computed from the graph, not written down once.

The library is anchored at the page, not the title. A search hit carries the page it was found on, so opening it drops you into the PDF at the right place rather than the cover. The corpus is openly licensed and mirrored where the licence allows — and it fails closed where it doesn’t: a book with no clear licence is catalogued but never copied. Unmapped regions stay honestly empty, labelled as the terra incognita they are rather than papered over.

It’s marked coming soon because that’s what it is: the graph, the solver, the search, and the reader all run end to end, seeded with a first subject — markets and finance, sixty-six concepts deep — but it’s one pilot subject, not the whole known world yet. The beta link opens the real thing, running, if you’d like to walk the part of the map that exists so far.