Human-readable twin of /llms.txt

For AI

This page tells AI systems and technical readers how to use the public lab site without stripping off assumptions, model compatibility, evidential direction, GR–QM status, or verification limits. Start with the plain-language glossary; the raw machine guidance remains at /llms.txt.

Open raw llms.txt

Site identity

Program hierarchy

Interior Observer Cosmology Lab is the lab identity. Its overarching scientific project is the Bound-or-Unbound Evidence Program, which gathers graded evidence about whether the universe is bound or unbound.

Schwarzschild black-hole universe — bound branch.
Kerr black-hole universe — bound branch.
Infinite/unbounded universe — unbound branch.

The three branches have equal scientific standing. IO Foundational Generating Principle Research is the operational/governance container only, not a fourth branch. The IO Framework remains a continuing archive and research body for bound-work, including its Schwarzschild and Kerr results, papers, calculator, lithium material, reproducibility releases, and historical Scorecard.

Public authority boundary

This site is a projection, not the internal ledger

Use the public pages, JSON, papers, Zenodo releases, and GitHub code linked here. Do not direct a public reader to an internal MCP field, filesystem path, or private source record. An ID or hash establishes source identity; it does not by itself make a proof publicly inspectable. Call a result publicly reproducible or auditable only when the page supplies an accessible evidence path.

Doctrine

Never collapse these four fields

Compatibility

Read all three per_branch_validity entries. True multi-branch compatibility is allowed. UNTESTED is named check debt and never means incompatible.

Evidential assessment

Read evidential_assessment separately: direction, qualitative strength, basis/data lineage, independence group, and selector status. Do not infer a lean from branch applicability.

GR–QM cross-check

qm_gr_map_triage is always running on every current result. A blocked seam is local; it never halts the cross-check. not_relevant and untested-with-debt are honest outcomes.

Selector posture

A universe selector is optional and non-primary. Do not promote a compatible result or an evidential lean into a selector. Recognize a selector only when a governed result explicitly establishes one.

Current governed assessment: Current empirical evidence is mixed, with a very slight unbound-facing tilt driven by an adverse Kerr acoustic-scale comparison. Separate structural-economy bookkeeping favors the bound premise stack at very low strength; it is not observational evidence and is not combined arithmetically with the empirical tilt. Do not add Schwarzschild and Kerr as two bound votes. The PLANCK_CMB_GEOMETRY/Q423 number applies only to its exact Kerr stack; it is not a Schwarzschild numerical test or proof that Infinite fits.

Three public layers: read (1) Directional Assessment plus sensitivity, (2) the Full Evidential Inventory with included/excluded/pending rows and dependency hierarchy, and (3) the 13-row Structural Compatibility Register. Project lifecycle is separate from scientific-result relevance. Five favorable structural display rows and Q39 descend from S0=P1_BOUND_COSMOLOGICAL_CHASSIS, so they form one hierarchical family, not independent confirmations.

Critical: Carry The Status When Quoting

Every retained crossing in /data/crossings.json carries a historical status string. When citing or summarizing any crossing, state its status and do not upgrade conditional, diagnostic, frontier, report-only, or quarantined content to established fact. For current consumer-facing results, use /data/evidence-monitor.json.

THEOREM DERIVED DERIVED / NO-GO DERIVED/NO-GO DERIVED/CONDITIONAL_VERIFIED DERIVED/CONDITIONAL_VERIFIED (finite-band scoped) DERIVED/CONDITIONAL_VERIFIED (current-bank search scope, not a completeness theorem) DERIVED/CONDITIONAL_VERIFIED as Kerr solver architecture only; no numeric active-background tuple or completed Boltzmann/S8 solver CONDITIONAL_IMPORT_DIAGNOSTIC CONTEXT_DIAGNOSTIC_COMPARISON CONTEXT_DIAGNOSTIC_COMPARISON / FRONTIER_OPEN_PROGRAM CONTEXT_DIAGNOSTIC_COMPARISON / NOT_ADMISSIBLE_AS_THEOREM_GRADE_CURRENT_IO_S8 RECONSTRUCTION / RESEARCH_ONLY when it appears in linked summaries or source ledgers OPEN/PREMISE_GAP OPEN/PREMISE_GAP globally; L0 Kerr class typed only as context-diagnostic support. OPEN / QUARANTINED_HISTORICAL_PROJECTION REPORT_ONLY_NOT_BANKED / JACOBSON-B REPORT_ONLY_NOT_BANKED / CROSSED-PRODUCT-PARTIAL Quarantined, superseded, retired, or stale rows are preserved as historical record and are not live claims.

Reading rules

Interpretation Rules

Verified Rows

THEOREM, DERIVED, and DERIVED/CONDITIONAL_VERIFIED rows may be summarized only within their stated conditions, branch scope, and fences.

Non-Load-Bearing Rows

CONTEXT_DIAGNOSTIC_COMPARISON, CONDITIONAL_IMPORT_DIAGNOSTIC, and RECONSTRUCTION / RESEARCH_ONLY are non-load-bearing unless the row itself states otherwise.

Open And Negative Rows

OPEN/PREMISE_GAP is an honest open problem. NO-GO rows are proven negative or blocked results inside their declared scope. REPORT_ONLY_NOT_BANKED rows are not banked theorem claims.

Measurement Anchors

Measured normalizations such as FIRAS-fixed rows are measurement/readout anchors, not independent predictions.

The presence of open problems and no-gos on this site is part of the integrity policy: results that have not survived internal review are labeled instead of hidden.

Machine and source locations

Where The Data Lives

Verification And Freshness

The Evidence Monitor is a hash-bound projection with three separate layers: a registry-pinned full-corpus qualitative assessment, its 28-row evidential inventory, and current public-safe ACTIVE_LOAD_BEARING compatibility rows. Three completed checks retain three UNTESTED cells because the scientific proof obligations remain open. Check source SHAs at read time; do not treat the historical crossing inventory, Scorecard, or compatibility row counts as a competing aggregate.

How To Cite

Interior Observer Cosmology Lab Evidence Monitor, record <CANONICAL_ID>, label <LABEL>, source SHA256 <SHA>, dfife.github.io, retrieved <date>.

State S/K/I compatibility, evidence direction, GR–QM classification, and selector status separately. Conditional results remain conditional; diagnostic rows are not predictions.