Supports modeled structural-burden, event-history, retained-time, women’s lane visibility, and sensitivity analysis claims. It does not support diagnosis, return-to-play, clinical injury prevention, or worker surveillance.
LAKANA Sovereign Systems · reviewer surface
Evidence Posture
LAKANA separates architecture thesis, formal abstraction, simulation evidence, empirical support, source-governed research packages, prototype posture, and deployment claims. This page defines how strong each evidence type is allowed to sound.
Evidence ladderSimulation ≠ field validationEmpirical support ≠ certificationNo medical diagnosisNo emergency-service certificationNo official weather authority
Evidence ladder
Each claim must stay inside its evidence tier.
1 · Architecture thesis
Allowed wording: designed to, proposes, separates, constrains.
What it supports
Local-first design, non-extractive doctrine, scoped release, authority contraction, custody events.
Artifacts
Architecture pages, interface table, governance overlay, proof routing, constitutional language.
Not allowed
Claims of deployed safety, certified privacy, or guaranteed outcomes.
2 · Formal abstraction
Allowed wording: defines, models, gates, bounds.
What it supports
TSARO states, NICOLE events, authority gates, interface contracts, fail-closed behavior.
Artifacts
Architecture interface, claim matrix, threat model, engine equations in source stacks.
Not allowed
Assuming math abstraction has been validated in field hardware.
3 · Client-side simulation
Allowed wording: demonstrates, visualizes, illustrates.
What it supports
Public-safe behavior of the integrated engine and solo proof surfaces.
Artifacts
Proof Theater, Integrated Engine, SOS/CivOS, SSI, W-X/WX-Ag public simulations.
Not allowed
Deployment proof, responder certification, medical clearance, crop prescription.
4 · Monte Carlo evidence
Allowed wording: simulation-internal estimate, modeled result, under assumptions.
What it supports
SSI event-history pathways, SOS/CivOS architecture performance, W-X/WX-Ag scenario behavior.
Artifacts
SSI J2/v94/v95 stacks, SOS v71 corrected source bundle and figures, W-X/WX-Ag v10 record.
Not allowed
Causal field efficacy, clinical incidence, emergency-service replacement, official forecast.
5 · Empirical support
Allowed wording: empirically connected, support-resolved, calibrated where stated.
What it supports
Dataset presence, trace ingestion, support-rate resolution, empirical/proxy lane status.
Artifacts
W-X/WX-Ag weather/SOC/soil support tables; SOS GUIDE and temporal proxy lane metadata; SSI empirical bundles and public tables.
Not allowed
Full external validation or real-world effectiveness unless a field protocol produced it.
6 · Source-governed package
Allowed wording: reproducibility package, candidate manuscript, manifest-bound.
What it supports
Research review, artifact inventory, SHA256 lineages, claim registries, source bundles.
Artifacts
SSI J2 119-artifact suite, SOS v71 preprint bundle, W-X Zenodo submission package, reviewer packages.
Not allowed
Peer-reviewed acceptance unless actually accepted.
7 · Prototype / pilot roadmap
Allowed wording: needs validation, planned, next test.
What it supports
Bench tests, red-team tests, partner pilots, responder workflow studies, privacy/legal review.
Artifacts
Validation roadmap, threat model, architecture interface, future pilot protocols.
Not allowed
Completed field validation unless documented.
Research branch posture
What each research stack can currently support.
Supports local-first emergency-state simulation, degraded transport, TSARO/NICOLE governance, and bounded responder sessions. It does not support emergency-service certification, live rescue guarantees, or production hardware claims.
Supports environmental truth-state, deterministic agronomic proxy, empirical-support resolution, stale/replay handling, and public-safe scenario results. It does not support official weather authority, extension-grade agronomy, or yield guarantees.
Evidence-to-language rules
Approved public language patterns.
For doctrine and architecture claims: local-first, non-extractive, fail-closed, scoped release, user supremacy, non-release logging.
For client-side pages and proof theater behavior: visible state transitions, authority contraction, release path, custody events.
For Monte Carlo results: survivability, retained time, burden, event histories, environmental stress, and sensitivity outputs.
When a dataset support check, trace, or source inventory is connected, but do not call that field validation.
For manuscripts and bundles unless peer review or DOI registration status supports stronger wording.
For hardware, responder, clinical, agronomic, legal, and privacy claims outside the current public evidence stack.
Red-team boundary: simulation evidence can be strong and still not be field validation. Strong LAKANA language should be narrow, source-governed, and explicit about what it does not claim.