Experiment ID: CC-EXP-C05-CONF-03

Version: 0.1

Date: 19 September 2026

Candidate: Assessment Assertion Profile

Standing: INTERNAL — SEMANTIC STABILITY / HARD COUNTEREXAMPLE TEST. NO COMMONS ADMISSION OR PUBLICATION AUTHORISED.

Question

Does the Assessment Assertion model survive hard cases involving incomplete/stale evidence, changing subjects, conflicting bases, waivers, continuous assessment, probabilistic outcomes and multiple jurisdictions?

Executive result

SEMANTIC STABILITY PASS WITH FOUR MATERIAL REFINEMENTS.

The model survives if:

1. assessment outcome is always relative to an explicit basis + scope + time;
2. evidence sufficiency/currentness is not silently folded into the outcome;
3. exception/waiver and acceptance remain separate downstream assertions/decisions;
4. continuous assessment produces time-bounded assertion states/history rather than one timeless conformance flag.

HT01 — Sampled evidence

Assessment uses a sample rather than complete population.

Preserve sampling/method/evidence scope.

Outcome applies only under the declared assessment method/basis.

Do not promote sample result to universal truth.

PASS.

HT02 — Stale evidence

Evidence was valid at assessment time but is now old.

Historical Assessment Assertion remains valid as a record of the assessment.

Current applicability may expire/require revalidation.

PASS.

Refinement: assessment time and evidence observation/effective time may differ and both may matter.

HT03 — Conflicting requirements in basis

Basis contains requirements that cannot simultaneously be satisfied.

Assessment may produce multiple requirement-level findings or an indeterminate/conflict result according to basis semantics.

Do not force a single universal outcome.

PASS.

HT04 — Proposed/future design

A design not yet implemented is assessed against a target profile.

Allowed:

subject role may be DESIGNED under RRAP.

Assessment finding applies to design conformance, not implementation/runtime conformance.

PASS.

HT05 — Continuous assessment

Control is assessed continuously.

Do not maintain one mutable timeless finding.

Preserve assertion history/windowed state or explicitly versioned current assertion.

PASS WITH TEMPORAL REFINEMENT.

HT06 — Two bases, different outcomes

Subject conforms to Basis A but not Basis B.

Both assertions coexist.

PASS — validates basis-relative outcome.

HT07 — Waived requirement

Assessment finds non-conformance; governance waives requirement for a bounded context.

Do not rewrite assessment to conformant.

Preserve:

Assessment Finding → Waiver/Exception Decision → Acceptance/Permission where applicable.

STRONG PASS.

HT08 — Exception accepted after non-conformance

Same as above: acceptance is downstream governance standing.

PASS.

HT09 — Assessment of a QMAP Mapping Assertion

A mapping assertion can be the assessment subject.

Example: assess mapping against a mapping-quality/profile basis.

Outcome does not change the mapping relation itself.

PASS.

HT10 — Assessment of AI output

AI-generated output assessed against criteria.

Assessment provenance must distinguish:

• generator;
• assessment method/tool;
• evidence;
• basis.

Automated assessment is allowed.

PASS.

HT11 — Finding later invalidated

Assessment method later found defective.

Historical assertion remains but may receive invalidated/superseded qualification.

Do not delete lineage.

PASS.

HT12 — Subject changes during assessment

Long-running test spans configuration change.

Either:

• bind assessment to immutable/versioned subject snapshot; or
• record temporal segmentation/limitations.

A timeless finding over a changing subject is invalid.

PASS WITH SUBJECT-IDENTITY REFINEMENT.

HT13 — Probabilistic outcome

Assessment produces 0.83 probability/risk score.

Do not translate to CONFORMS unless the basis defines threshold/decision semantics.

Preserve quantitative result + basis.

PASS.

HT14 — Multi-jurisdiction criteria

Same subject assessed against AU and US criteria.

Separate Assessment Assertions with jurisdiction/applicability.

PASS.

HT15 — Runtime success despite design non-conformance

Successful operation does not erase design conformance finding.

Observed outcome may be evidence for later reassessment.

PASS.

HT16 — Conformant but unsafe in current context

Conformance to technical profile does not establish runtime safety/admissibility.

STRONG PASS.

HT17 — Assessor conflict

Independent assessors produce different outcomes.

Preserve both assertions/provenance.

Governance may later determine accepted finding.

PASS.

HT18 — Basis withdrawn

Assessment remains historical relative to withdrawn basis.

Future reliance may be prohibited/revalidation triggered.

PASS.

Material refinement 1 — Evidence qualification is separate

Evidence may have:

• coverage;
• currentness;
• provenance;
• sampling;
• quality;
• limitations.

Assessment Assertion references evidence and method.

Do not create a universal “evidence sufficient” flag inside the profile unless the assessment basis/method defines it.

Material refinement 2 — Basis-relative outcome

Outcome is meaningless without basis.

Therefore:

Outcome(subject, basis, scope, time)

not:

Outcome(subject).

Material refinement 3 — Subject identity/version

Assessment must bind to a sufficiently stable subject identity/aspect/version/snapshot.

If subject changes materially, new/segmented assessment is required.

Material refinement 4 — Temporal assertion history

Continuous/repeated assessment should preserve:

• assertion identity/version/window;
• prior findings;
• supersession/revalidation;
• effective time.

Do not overwrite history with one current flag.

Minimum profile after test

Required:

• assessment_assertion_id
• subject_ref
• subject_aspect/scope
• assessment_basis_ref
• assessment_outcome_ref/value
• assessment_provenance

Required when material:

• subject version/snapshot
• basis version/profile
• method/procedure
• evidence refs
• assessment time
• evidence time/window
• environment/jurisdiction

Optional/linked:

• assessor/tool
• limitations/qualification
• conflict
• invalidation/supersession
• revalidation trigger
• conformance claim
• acceptance/waiver/certification refs

Invariants

1. Assessment is basis-relative.
2. Assessment is scope/aspect-relative.
3. Assessment is time-relative where subject/evidence can change.
4. Assessment ≠ acceptance.
5. Assessment ≠ certification.
6. Assessment ≠ authority.
7. Conformance ≠ runtime admissibility.
8. Evidence ≠ assessment outcome.
9. Observed success ≠ complete conformance.
10. Waiver does not rewrite non-conformance.
11. Conflicting assessments may coexist.
12. Historical findings remain reconstructable.
13. Probabilistic result ≠ categorical conformance unless basis licenses threshold.
14. Subject change requires bounded identity/snapshot handling.
15. Basis change/withdrawal does not silently rewrite historical finding.

Result

18/18 hard cases survive.

No case requires:

• universal conformance outcomes;
• acceptance inside the profile;
• authority model;
• certification model;
• EAA/runtime semantics.

Stage result

SEMANTIC STABILITY PASS.

Assessment Assertion Profile is stable enough for a non-machine-readable Candidate Specification v0.1.

Next

Draft Candidate Specification and hostile-review it.

Then run focused generic assessment/assertion prior-art + naming review before any Commons admission package.

Decision

CC-EXP-C05-CONF-03 — SEMANTIC STABILITY PASS WITH MATERIAL REFINEMENTS.

No admission/publication/namespace/schema authorised.

Linked

• 🔎CC-EXP-C05-CONF-02 — Assessment Assertion Prior-Art & Standards Reduction v0.1
• 🧪CC-EXP-C05-CONF-01 — Conformance / Acceptance Boundary — Decomposition & Generality Test v0.1
• ⚖️CC-ADM-RRAP-01 — Representation Role Assignment Profile — Commons Admission Decision Package v0.1
• ⚖️CC-ADM-QMAP-01 — Qualified Mapping Assertion Profile — Commons Admission Decision Package v0.1