Pranav Bhave
Module 003 of 006 — all modules Current

Module 003 — Kernel of a receipt

When a governance receipt verifies, what exactly has been identified?

An execution is identified only up to the kernel of the receipt system's whole parse → canonicalize → digest pipeline. Receipt soundness is a ternary relation — Sound(C, Σ, P) — between the canonicalizer, the pathology alphabet, and the consumer set; it grows fragile as pathologies are discovered and consumers widen, so soundness does not persist by default.

01Observed / given

Receipts bind bytes: parse the artifact, canonicalize, digest, sign. Distinct inputs that canonicalize identically are indistinguishable to every downstream verifier — the kernel of the pipeline, not of any single stage.

02Visual experiment
Open the experimentHow a receipt loses a distinction — Ghost-Ark's kernel figure
03Feasible worlds

For one digest, the feasible set is every input in its kernel class. Which distinctions the kernel destroys is measurable; who needed those distinctions is a property of the consumer set, and it changes after the receipt is issued.

04Identification status

Identified up to the kernel; beyond it, underdetermined. The lab's own falsifier ledger keeps score: F2 stands CONFIRMED (2026-08-12) in its contracted form — a random sample of 64 real supply-chain attestation payloads carried zero pathology classes, so the collision claim reads "possible, not observed" — while F3 (that the consumer set is stable in practice, so antitonicity never bites) remains open and, in the repo's own words, the most under-attacked of the five.

05Result or bound

The thesis document at the bound commit states the ternary soundness claim and its two consequences: the kernel is a property of the whole pipeline (C1), and real canonicalizers — Ghost-Ark's included — have unintended kernel members (C2), as existence results.

06Witness / counterexample

Constructed collision inputs demonstrate existence; the census and mutation figures in the repository are drawn from a data manifest so a figure cannot drift from the number it draws.

07Claim envelope

GA-001

Supported within scope · Lab repository, public

Ghost-Ark is a verifier and measurement harness for the provenance limits of AI-governance receipts — a research artifact of the S2 Lab, Penn State — whose stated thesis is that receipt soundness is a ternary relation Sound(C, Σ, P): a receipt identifies an execution only up to the kernel of its whole parse → canonicalize → digest pipeline, so soundness does not persist by default as the pathology alphabet grows or the consumer set widens.

Scope
The repository README and thesis document at the bound commit.
Binding
00_THESIS.md @ 98c90d82
Re-worded 2026-08-23 to track the bound thesis precisely: soundness as a ternary relation, and the kernel as a property of the whole pipeline rather than of the canonicalizer alone.
Dimensions
visibilityPublicprovenanceLab repository, publicsupport roleDocumentmaturityIn development
Reviewed
2026-08-23 · window 120 days
Triggers
  • executable fires when the thesis document on the default branch changes
Non-claims
  • a verifying receipt does not establish that the governed action was safe, authorized, or semantically correct
  • kernel collisions in real canonicalizers are demonstrated as possible, not as prevalent — the repository's E12 sample found 0 of 64 real payloads carrying any pathology class
  • not hardened for deployment; not post-quantum secure

Rendered from the same registry as the ledger — one source, two views.

08Evidence
GA-001 @ 98c90d82 — Lab repository, public · reviewed 2026-08-23
09Falsifier

The executable trigger on GA-001 fires if the thesis document changes on the default branch. Empirically: an incidence measurement showing consumer-relevant divergence in the wild would move C2 from possible to prevalent — and its absence keeps the claim contracted.

10Non-claims
11Reproducibility
Ghost-Ark ships its verifier, collision corpora, and a standalone kernel probe "for people who do not use Ghost-Ark and have no reason to trust it."
12Next unknown

Incidence, not existence: does any real consumer population currently depend on a distinction some deployed canonicalizer destroys?