One Hundred Percent Truth-Grounded

This page computes the proof that verification coverage equals 100%. Start by clicking on the main result 100 = 100 [1] to open it in the side panel, then use Focus in Graph to see the knowledge graph, or Focus in Layers to see the full computation chain. Click Taints to see inconsistencies we're currently aware of — mostly documentation that hasn't caught up with the code yet.

The Integrity Cortex landing page claims 100% truth-grounded — verified against source. This is not a pass rate measured on a corpus. It means every statement in the system has a computed grounding state: verified, unverified, no-evidence, fabricated, manually-verified. Even "not grounded" is an explicit grounding state. The system has an opinion about every item's relationship to source truth, and that opinion was computed by running the item against the documents, all the way down.

A derive either bottoms out in facts with verified evidence, or the system computed that it doesn't — and that computation IS the grounding state. No item is left in limbo.

pltg
Load
(import (quote .src.truth_grounded tg))
tg.truth-grounded-coverage
Out
100

Five stages, each with a loss factor

This totality is a conjunction of five conditions. Each stage in the truth-grounding pipeline is a place where information could be lost — items could escape assessment, the verifier could drop matches, branches could go unvisited. We prove each stage preserves by computing its loss factor: satisfied properties / total properties. If a stage preserves, its factor is 1. If it loses, the factor drops below 1 and the final number reflects exactly how much coverage was lost and where.

Coverage = 100 × f₁ × f₂ × f₃ × f₄ × f₅

Stage 1 — All language atoms have grounding state

Facts, axioms, and terms support direct grounding via evidence. Derives inherit grounding transitively via their :using chains. Diffs need no evidence — inconsistency between their sides is itself a state. No directive type is exempt. Factor: 1. [2]

Stage 2 — Grounding was computed, not assumed

The engine's three consistency layers each actively run: evidence verification, fabrication propagation, and diff evaluation. No layer is skipped. Factor: 1. [3]

Stage 3 — No information loss in the quote verifier

The normalizer applies the same function with the same config to both the document and the quote. Each of the 6 steps is checked for both-sides application. A step that processes both sides creates no matches and destroys no matches. Factor: 1. [4]

Stage 4 — Verifier is part of the pipeline

The quote verifier's output feeds into the engine's consistency assessment — it IS consistency layer 1, and fabrication propagation consumes its results. Results are not discarded. Factor: 1. [5]

Stage 5 — Every branch is traversed

The dependency scanner covers all five directive types and the traversal is transitive. No branch of the graph is left unvisited. Factor: 1. [6]

Result

pltg
Result
tg.truth-grounded-coverage
Out
100

Coverage: 100 [7] × 1 [2] × 1 [3] × 1 [4] × 1 [5] × 1 [6] = 100. [8]

Checks

CheckResultRule
src.truth_grounded.truth-grounded-provedHOLDSsrc.truth_grounded.marketing-claim-100 (100) must equal src.truth_grounded.truth-grounded-coverage (100)

References

#NameKindValueGrounding
1src.truth_grounded.truth-grounded-proveddiff100 = 100tainted: depends on tainted: src.truth_grounded.truth-grounded-coverage
2src.truth_grounded.s1-factorcalc1derived: (/ src.truth_grounded.s1-satisfied src.truth_grounded.s1-total)
3src.truth_grounded.s2-factorcalc1derived: (/ src.truth_grounded.s2-satisfied src.truth_grounded.s2-total)
4src.truth_grounded.s3-factorcalc1derived: (/ src.truth_grounded.s3-satisfied src.truth_grounded.s3-total); tainted: depends on tainted: src.truth_grounded.s3-satisfied
5src.truth_grounded.s4-factorcalc1derived: (/ src.truth_grounded.s4-satisfied src.truth_grounded.s4-total)
6src.truth_grounded.s5-factorcalc1derived: (/ src.truth_grounded.s5-satisfied src.truth_grounded.s5-total)
7src.truth_grounded.baselinefact100baseline is 100 — the ideal before any pipeline stage: "We prove each stage preserves by computing its loss factor: satisfied properties / total properties. If a stage preserves, its factor is 1. If it loses, the factor drops below 1 and the final number reflects exactly how much coverage was lost and where." (notebook, verified)
8src.truth_grounded.truth-grounded-coveragecalc100derived: (* src.truth_grounded.baseline (* src.truth_grounded.s1-factor (* src.truth_grounded.s2-factor (* src.truth_grounded.s3-factor (* src.truth_grounded.s4-factor src.truth_grounded.s5-factor))))); tainted: depends on tainted: src.truth_grounded.s3-factor

Lineage

GroupKindNodeDescription
Parseltongue (OSS)rawcore/validation/core.pltg, std/counting.pltg (imported by every proof)
Parseltongue (OSS)rawcore/std/epistemics.pltg
Parseltongue (OSS)rawcore/demos/data_governance_pltg: generate.py, checker.pltg, resources/governance_policy.md
Parseltongue (OSS)rawcore/inspect/perspectives/visualisation/taints.py
Parseltongue (OSS)rawcore/quote_verifier/index.py
Parseltongue (OSS)rawcore/README.md
Websiterawvectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)
Proof notebooksrawzero-silent.pgmd, truth-grounded.pgmd, sub-minute.pgmd, unbounded.pgmd, reproducible.pgmd
Proof notebooksrawresources/integrity-cortex.md
Proof modulesfactssrc/zero_silent.pltg"0 silent inconsistencies" by construction: five naive consistency criteria map to the epistemic lattice, which the engine proves maps to errors and warnings, which have detectors; the count of silent ones is 0. Closes with diff zero-silent-grounded (the tile's 0 against silent-inconsistency-count).
Proof modulesfactssrc/truth_grounded.pltg"100% truth-grounded" by construction: a baseline of 100 (every item has a grounding state) times the loss factor of each of five pipeline stages. Closes with diff truth-grounded-proved (the tile's 100 against truth-grounded-coverage).
Proof modulesfactssrc/sub_minute.pltg"<1 min instead of weeks" by construction: the demo checker's compliance derives, counted as a document, give the manual scale in weeks; the per-edge cost of the taint BFS gives the parseltongue time. Closes with diffs weeks-grounded and sub-minute-grounded.
Proof modulesfactssrc/unbounded.pltg"∞ nodes and connections" by induction on system size (zero/succ): a zero-node system is valid, and validity at N gives validity at succ(N) with no guard on N. Closes with diff unbounded-grounded (the tile's ∞ against node-limit).
Proof modulesfactssrc/reproducible.pltg"Reproducible =" by existence: the proof pages exist and each executes a deterministic parseltongue computation. Closes with diff reproducible-grounded.
Proof pagesfinalproofs/zero-silent.html
Proof pagesfinalproofs/truth-grounded.html
Proof pagesfinalproofs/sub-minute.html
Proof pagesfinalproofs/unbounded.html
Proof pagesfinalproofs/reproducible.html
Proof pagesfinalproofs/integrity-cortex.html (every proof, imported)
FromToHow
core/validation/core.pltg, std/counting.pltg (imported by every proof)src/zero_silent.pltgimport
core/validation/core.pltg, std/counting.pltg (imported by every proof)src/truth_grounded.pltgimport
core/validation/core.pltg, std/counting.pltg (imported by every proof)src/sub_minute.pltgimport
core/validation/core.pltg, std/counting.pltg (imported by every proof)src/unbounded.pltgimport
core/validation/core.pltg, std/counting.pltg (imported by every proof)src/reproducible.pltgimport
core/std/epistemics.pltgsrc/zero_silent.pltgimport
core/demos/data_governance_pltg: generate.py, checker.pltg, resources/governance_policy.mdsrc/sub_minute.pltgload-document
core/inspect/perspectives/visualisation/taints.pysrc/sub_minute.pltgload-document
core/quote_verifier/index.pysrc/truth_grounded.pltgload-document
core/README.mdsrc/unbounded.pltgload-document
zero-silent.pgmd, truth-grounded.pgmd, sub-minute.pgmd, unbounded.pgmd, reproducible.pgmdsrc/reproducible.pltgload-document (all five)
zero-silent.pgmd, truth-grounded.pgmd, sub-minute.pgmd, unbounded.pgmd, reproducible.pgmdsrc/zero_silent.pltgload-document (its own notebook)
zero-silent.pgmd, truth-grounded.pgmd, sub-minute.pgmd, unbounded.pgmd, reproducible.pgmdsrc/truth_grounded.pltgload-document (its own notebook)
zero-silent.pgmd, truth-grounded.pgmd, sub-minute.pgmd, unbounded.pgmd, reproducible.pgmdsrc/unbounded.pltgload-document (its own notebook)
vectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)src/zero_silent.pltgquotes the 0 tile
vectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)src/truth_grounded.pltgquotes the 100% tile
vectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)src/sub_minute.pltgquotes the <1 min tile
vectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)src/unbounded.pltgquotes the ∞ tile
vectorcat-website src/components/react/IntegrityCortexCards.jsx (the IMPACT tiles)src/reproducible.pltgquotes the A=A tile
src/zero_silent.pltgproofs/zero-silent.htmldiff zero-silent-grounded
src/truth_grounded.pltgproofs/truth-grounded.htmldiff truth-grounded-proved
src/sub_minute.pltgproofs/sub-minute.htmldiffs weeks-grounded, sub-minute-grounded
src/unbounded.pltgproofs/unbounded.htmldiff unbounded-grounded
src/reproducible.pltgproofs/reproducible.htmldiff reproducible-grounded
src/zero_silent.pltgproofs/integrity-cortex.html (every proof, imported)import
src/truth_grounded.pltgproofs/integrity-cortex.html (every proof, imported)import
src/sub_minute.pltgproofs/integrity-cortex.html (every proof, imported)import
src/unbounded.pltgproofs/integrity-cortex.html (every proof, imported)import
src/reproducible.pltgproofs/integrity-cortex.html (every proof, imported)import
resources/integrity-cortex.mdproofs/integrity-cortex.html (every proof, imported)load-document