code wiki / _hdl_build / nx_scientist.nx

nx_scientist.nx

buildroot/runtime/_hdl_build/nx_scientist.nx

3159 B46 linesdepth 3pulls 3 transitivereach 4 importersview sourcekind library
docsdependenciesstructsconstsfunctions

about

nx_scientist.nx -- the NISHI SCIENTIST: the team's authoring + publishing arm (operator: Claude is currently the publishing arm; the TEAM needs to write S-class-exceed publishable papers + docs that survive OUTSIDE critics -- the world + the scientific method -- not just Claude+operator). The Scientist assembles a PAPER from a proven capability so it stands on its own evidence. A paper carries the six things the scientific method demands: 1. a FALSIFIABLE claim at a tier (S/A/B) 4. the HONEST grade (from the Examiner) 2. a reproducible METHOD 5. CAVEATS / limitations (from the Critic) 3. EVIDENCE (gate passes + data) 6. a REPRODUCIBILITY gate (Librarian-persisted) PUBLISHABLE iff: COMPLETE (all six) AND NO OVERCLAIM (claimed tier <= the honest Examiner grade -- the cardinal sin against outside critics is claiming S for an A result) AND reproducible. RACI: Writer drafts + Author owns the claim (both the Scientist); the LIBRARIAN is the publisher/archivist (citations + persistence); the EXAMINER supplies the honest grade; the CRITIC supplies the caveats. Honest boundary: prose polish may stay partly Claude/Modelwright-assisted, but the team OWNS the scientific RIGOR -- the no-overclaim/evidence/reproducibility gates a paper must pass. license_tier: ORIGINAL Refs: the scientific method; reproducibility; the Critic's no-gospel rule.

dependencies 2 imports · 3 importers

nx_sclass_grader.nx nx_syscalls.nx nx_scientist.nx nx_paper_emit.nx nx_publish_review_test.nx nx_scientist_test.nx

imports: nx_sclass_grader.nxnx_syscalls.nx

imported by: nx_paper_emit.nxnx_publish_review_test.nxnx_scientist_test.nx

structs

none

consts

20const SCI_REJECT: i64 = 0 // overclaim or unsound -> does not survive outside critics
21const SCI_REVISE: i64 = 1 // missing a component -> send back
22const SCI_PUBLISH: i64 = 2 // complete + honest + reproducible -> publishable

functions

25func sci_complete(has_claim: i64, has_method: i64, has_evidence: i64, has_grade: i64, has_caveats: i64, has_repro: i64) -> i64
called by 2: sci_reviewmain
31func sci_no_overclaim(claim_tier: i64, honest_grade: i64) -> i64 { if claim_tier <= honest_grade { return 1 } return 0 }
called by 2: sci_reviewmain
34func sci_decision(complete: i64, no_overclaim: i64, reproducible: i64) -> i64
42func sci_review(has_claim: i64, has_method: i64, has_evidence: i64, has_grade: i64, has_caveats: i64, has_repro: i64, claim_tier: i64, honest_grade: i64) -> i64