teal-sea / zeta-labstate of record · compiled 14 Aug 2026 · revision 9ebdea0 · source

Library · docs/doors/README.md

Entry-point guides

810 words · 91 lines · source

One page per way into the repository. A repository with several purposes does not get several repositories; it gets one verified core and several short entry points. The core is zeta/ and its tests, ontology/ and its seam. Each guide names an audience, states what they can do here, and gives the first command to run.

harness/ was a third; it was demoted on 2026-08-13 after being tested against the practice it was meant to improve. Its guide is kept and banners the verdict, because deleting the guide to a thing you decided against is how a lab forgets what it learned (harness/VERDICT.md).

The cost of adding a purpose is a guide page plus a test that the page's command still works. A purpose that will not pay that cost stays a document, not a directory.

(The directory is named doors/ for historical reasons; department code references these paths, so the name stays.)

The guides

GuideFor you if you want to…First command
learn.mdsee the classical machinery run at arbitrary precision.venv/bin/python scripts/06_tour.py
refute.mdtest a claim about the zeros against the control battery.venv/bin/python scripts/23_gate_3_battery.py
certify.mdwork in a regime where nothing is measured — Lean proofs and interval enclosurescd lean && PATH="$HOME/.elan/bin:$PATH" lake build
discover.mdrun the conjecture funnel and see its measured hit rate.venv/bin/python scripts/13_discovery_run.py --dry-run
adopt.md(demoted) the validation framework, and why it was frozen.venv/bin/python -m pytest -q -o addopts='' tests/test_harness_protocol.py tests/test_department_conformance.py

If you are new and unsure: refute.md — it is the one thing this repository does that a textbook, a notebook or a literature survey does not. If you are here for the validation pattern rather than for ζ: adopt.md.

Departments

A guide is for a reader. A department is a registered subject: the subject itself, plus the battery of controls that can refute claims made in that subject's name, plus a guide page. Departments are declared in harness/departments/ and audited by tests/test_department_conformance.py, which is parametrized over all of them — adding a department adds its audit automatically.

The admission rule: no department without a battery. Work whose claims nothing in this tree can falsify is not a department; it is exploratory, and exploratory work lives where nobody will mistake it for a result.

That is also the only thing the core asks. A department keeps its own guides, its own reading course, its own gallery — learn.md and the heat-equation chain in README.md belong to the zeta department, and the next department is entitled to the same room.

DepartmentSubjectBattery
zetathe Riemann zeta function and RH3 rivals, 2 decoys, 3 surrogates, 3 lesions
finitefieldcurves over F_p, where RH is a theorem2 rivals, 2 decoys, 2 surrogates, 3 lesions
compilerLLVM IR rewrites — belief in a transformation, separated from its size3 rivals, 2 decoys, 3 surrogates, 4 lesions
cronitercron schedule union semantics under #/W — the first foreign-domain subject, battery content authored independently2 rivals, 2 decoys, 2 surrogates, 3 lesions
refereethe verification machinery itself — hollow batteries detected, the audit's power measured, blind spots pinned3 rivals, 2 decoys, 3 surrogates, 4 lesions
statevalstatistical model evaluation — whether a reported improvement is skill or protocol3 rivals, 2 decoys, 2 surrogates, 3 lesions

Every department also carries declared detectors (power and specificity measured, floors stated), a scope (what a pass licenses), and a provenance record (who authored the battery content, and blind or not). Every claim outcome is paired with the battery's integrity grade — see docs/20, and run the whole thing:

.venv/bin/python -m harness.demo

See harness/README.md for what the four control roles mean; python -m harness.new_department <name> scaffolds the next one.

Exploratory areas

The other side of the admission rule. Exploratory work is work whose claims nothing in this tree can yet falsify — a normal and useful state, not a criticism. It is kept where nobody will mistake it for a result, appears in no department table, and gets no guide page.

AreaWhat it is
dossier/representing research state — intent, definitions, rejected alternatives, obligations, evidence — so an agent can resume. Four independent support axes, no aggregate
hunts/scoped exploratory studies; most fail, and the failures are recorded

Neither becomes a department by growing. Their negative controls are the zeta department's, and a department whose battery belongs to another department is not a department — the reason is recorded in docs/19 §6.