2026-08-22 23:11:21 +02:00
# Concept ↔ Implementation Fitness Map
2026-09-28 12:06:24 +02:00
**Updated:** 2026-09-28 (TD-WP-0003-T02/T03/T04)
2026-08-22 23:11:21 +02:00
Traces each important concept to the implementation, experiment and evidence that
support it. **Unsupported entries are the point of this map** — a concept with no
implementation and no evidence is not a gap to be embarrassed about, it is the
current honest state, and hiding it defeats the map's purpose.
Support levels follow `TestDriverImprovementLoop.md` §13:
`C0 Idea` · `C1 Hypothesis` · `C2 Experimentally Supported` ·
`C3 Practically Validated` · `C4 Architectural Invariant`
## Current state
2026-08-22 23:50:29 +02:00
`C-semantic-action` is the first concept to reach `C2` : it has an experiment
behind it (the T07 two-arm comparison), and that experiment narrowed the claim
rather than confirming it. Everything else still rests on unit tests.
T04: deterministic semantic kernel
Alice/Bob/Carol runs end to end, deterministically, replayable from seed.
16 tests pass, no third-party dependencies.
- src/testdriver: intent, provenance, world, actions, drivers, observers,
oracles, evidence, energy, scenario, runner
- lab/minimal.py: the SUT, exposing the independent observation channel
required by D-07
- evidence is stratified S1/S2/S3; Runner refuses to attribute S2/S3 to an
actor; claims are frozen and provenance-checked at construction
- missing evidence yields INCONCLUSIVE, which outranks PASS in the run verdict
- EnergyEvents captured, no scoring (H-005 dormant)
The observation channel records both stored state and an out-of-band
enforcement probe; their disagreement is an invariant and is what detects an
authorization defect that leaves the audit trail intact. A seeded
RevokeIsCosmetic lab fails the run via both the claim and that invariant.
Also closes TD-WP-0001-T02 (stack and commands now exist).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-22 23:21:07 +02:00
The deterministic kernel exists (T04) and its guarantees are covered by unit
2026-08-22 23:50:29 +02:00
tests. **Levels do not move for those.** A passing unit test is not an experiment: it
T04: deterministic semantic kernel
Alice/Bob/Carol runs end to end, deterministically, replayable from seed.
16 tests pass, no third-party dependencies.
- src/testdriver: intent, provenance, world, actions, drivers, observers,
oracles, evidence, energy, scenario, runner
- lab/minimal.py: the SUT, exposing the independent observation channel
required by D-07
- evidence is stratified S1/S2/S3; Runner refuses to attribute S2/S3 to an
actor; claims are frozen and provenance-checked at construction
- missing evidence yields INCONCLUSIVE, which outranks PASS in the run verdict
- EnergyEvents captured, no scoring (H-005 dormant)
The observation channel records both stored state and an out-of-band
enforcement probe; their disagreement is an invariant and is what detects an
authorization defect that leaves the audit trail intact. A seeded
RevokeIsCosmetic lab fails the run via both the claim and that invariant.
Also closes TD-WP-0001-T02 (stack and commands now exist).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-22 23:21:07 +02:00
shows the code does what its author intended, not that the concept holds under
the mutations it claims to survive. Levels rise when E-001/E-002/E-003 produce
evidence, not before. The implementation column below moves; the level column
does not. The initial classifications in §13 of the Improvement Loop
2026-08-22 23:11:21 +02:00
(`Actor Isolation C2` , `Independent Oracles C2` ) are corrected downward here: they
were aspirational, not evidenced.
| Concept | Level | Implementation | Experiment | Evidence | Open question |
|---|---|---|---|---|---|
2026-09-28 12:06:24 +02:00
| `C-use-case` | C1 | `intent.py` , three reference scenarios | T02 | `tests/test_generalisation.py` | Three synthetic use cases fit; external authoring still unmeasured. |
2026-08-23 00:39:36 +02:00
| `C-actor-isolation` | **C2** | `world.py` , `runner.py` | E-001 | `td://self/actor-isolation` , examined every run | **F-0003 resolved** — automatic canaries; observable on every scenario. |
2026-09-28 12:06:24 +02:00
| `C-semantic-action` | **C2** | `actions.py` , `agentic.py` | E-001 | 2026-09-28 two-arm receipt | **F-0005** — supported only where stable identifiers are absent. Narrower than the concept model claims. |
T08: the classifier, measured and attacked
False Adaptation Rate = 0/7 across the labelled catalogue and the three E-003
attacks. 11 of 12 mechanical mutations absorbed without a human, so the safety
result is not bought by escalating everything.
- classification.py: total function over three signals, rule order chosen so
every rule that could excuse a regression sits after the rule that reports
one. SAFE_TO_ACCEPT is a two-element closed set, asserted.
- CompositeDriver plus scenarios/full_journey.py: one asset crossing both
surfaces, so UI mutations are visible as surface differences while the
claims they do not touch stay green.
- E-003: surface substitution (new M23), concurrent mechanical+defect,
evidence starvation, provenance laundering. All held.
F-0006 (CONCEPT_DRIFT, resolved): the T02 design listed SEMANTIC_CHANGE as an
outcome the table could produce. It cannot - M12 and M19 are behaviourally
identical, as the lab has asserted since T05. PRODUCT_DEFECT and
SEMANTIC_CHANGE collapse into one escalating outcome, BEHAVIOUR_CHANGED, and
the distinction becomes a human adjudication. INTENT_CHANGED survives but is
detected by the claim fingerprint moving, not inferred from behaviour.
Two classifier defects found and fixed rather than reported: claims downstream
of a failed realization now yield INCONCLUSIVE rather than FAIL (a false
accusation is the mirror image of a false adaptation), and the browser driver
records a page signature so surface change is detectable when the interaction
path is unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-23 00:02:58 +02:00
| `C-oracle-independence` | **C2** | `runner.py` , `oracles.py` | E-001, E-003 | — | Independence of components ≠ independence of belief. (H-004) |
2026-08-22 23:38:23 +02:00
| `C-evidence-pack` | C1 | `evidence.py` | — | `td://self/evidence-reproducibility` | Verdicts are reproducible from S3 alone, on passing and failing runs. |
T05: the lab and its labelled mutation catalogue
lab/app.py (users, tenants, auth, resources, sharing, read/write, revoke,
audit), lab/http_api.py (JSON API + browser UI, stdlib only), 20 labelled
composable version-stamped mutations, ground-truth matrix. 48 tests pass.
Detection against the reference scenario: MECHANICAL 0/10 flagged (correct),
DEFECT 6/6, SEMANTIC 2/4 with both inert cases declared.
- F-0002: M16 and M18 initially escaped detection entirely. A use case
protects exactly what it asserts. Resolved by adding two claims already
stated as intent in INTENT.md; the six-mutation catalogue would never have
surfaced this.
- test-id axis added: stable selectors survive most UI mutations, which would
make H-001 trivially false. Mutations now vary on preserves_test_ids so the
hypothesis is analysed split by that axis rather than rigged.
- M12 (semantic deferred revoke) and M19 (defect race) are behaviourally
identical and asserted as such - the discrimination problem as a test.
lab/minimal.py removed; superseded by lab/app.py.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-22 23:31:22 +02:00
| `C-observation-channel` | C1 | `lab/app.py` | — | — | **D-07** — required of every system under test. Adoption cost unknown. |
2026-09-28 12:06:24 +02:00
| `C-adaptation` | **C2** | `classification.py` , `agentic.py` | E-001 | T08 matrix | 26/27 mechanical absorbed; ten dropped-id cases now included. |
T08: the classifier, measured and attacked
False Adaptation Rate = 0/7 across the labelled catalogue and the three E-003
attacks. 11 of 12 mechanical mutations absorbed without a human, so the safety
result is not bought by escalating everything.
- classification.py: total function over three signals, rule order chosen so
every rule that could excuse a regression sits after the rule that reports
one. SAFE_TO_ACCEPT is a two-element closed set, asserted.
- CompositeDriver plus scenarios/full_journey.py: one asset crossing both
surfaces, so UI mutations are visible as surface differences while the
claims they do not touch stay green.
- E-003: surface substitution (new M23), concurrent mechanical+defect,
evidence starvation, provenance laundering. All held.
F-0006 (CONCEPT_DRIFT, resolved): the T02 design listed SEMANTIC_CHANGE as an
outcome the table could produce. It cannot - M12 and M19 are behaviourally
identical, as the lab has asserted since T05. PRODUCT_DEFECT and
SEMANTIC_CHANGE collapse into one escalating outcome, BEHAVIOUR_CHANGED, and
the distinction becomes a human adjudication. INTENT_CHANGED survives but is
detected by the claim fingerprint moving, not inferred from behaviour.
Two classifier defects found and fixed rather than reported: claims downstream
of a failed realization now yield INCONCLUSIVE rather than FAIL (a false
accusation is the mirror image of a false adaptation), and the browser driver
records a page signature so surface change is detectable when the interaction
path is unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-23 00:02:58 +02:00
| `C-classification` | **C2** | `classification.py` | E-001, E-003 | T08 matrix, FAR 0/7 | **F-0006** — cannot infer SEMANTIC vs DEFECT; collapses to one escalating outcome. |
T09: crystallization
A stable agentic realization becomes deterministic code. All four exit
criteria met; 163 tests pass.
- crystallization.py: trajectory capture, stability assessment requiring the
same path across several runs, CrystallizedDriver, pytest codegen
- crystallized/test_grant_access.py: generated, runs with no model, carries
its lineage in the docstring
- descendant preserves the ancestor's oracle set, agrees with it across five
lab versions, and still catches a seeded defect
- reversibility shown both ways via new M24 (grant endpoint renamed): the
frozen descendant fails loudly rather than searching, and the agentic
ancestor recovers from the same mutation
F-0007 (open): the 54% cost reduction must not be quoted in support of the
thesis. The T07 runtime is token-free, so the measured saving is one page
fetch, one parse and a two-candidate scoring pass. The saving the concept
actually claims - tokens, latency, retry variance - is unmeasured. Together
with F-0005 this makes a bounded live-model experiment the highest-value next
investment.
Assertions in the generated test are imported rather than restated, so it is
not fully standalone. Deliberate: paraphrased claims would be a second
unverified statement of intent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-23 00:21:48 +02:00
| `C-crystallization` | **C2** | `crystallization.py` | E-002 | T09 agreement matrix | Fidelity supported; **F-0007** — the economic case is unmeasurable with a token-free runtime. |
2026-08-22 23:38:23 +02:00
| `C-intent-provenance` | C1 | `provenance.py` | E-003 | `td://self/intent-independence` | Constrains provenance, not quality. Accepted residual. |
T09: crystallization
A stable agentic realization becomes deterministic code. All four exit
criteria met; 163 tests pass.
- crystallization.py: trajectory capture, stability assessment requiring the
same path across several runs, CrystallizedDriver, pytest codegen
- crystallized/test_grant_access.py: generated, runs with no model, carries
its lineage in the docstring
- descendant preserves the ancestor's oracle set, agrees with it across five
lab versions, and still catches a seeded defect
- reversibility shown both ways via new M24 (grant endpoint renamed): the
frozen descendant fails loudly rather than searching, and the agentic
ancestor recovers from the same mutation
F-0007 (open): the 54% cost reduction must not be quoted in support of the
thesis. The T07 runtime is token-free, so the measured saving is one page
fetch, one parse and a two-candidate scoring pass. The saving the concept
actually claims - tokens, latency, retry variance - is unmeasured. Together
with F-0005 this makes a bounded live-model experiment the highest-value next
investment.
Assertions in the generated test are imported rather than restated, so it is
not fully standalone. Deliberate: paraphrased claims would be a second
unverified statement of intent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-23 00:21:48 +02:00
| `C-lineage` | C1 | `scenario.py` , generated headers | E-002 | generated module docstring | Parent pointer plus provenance in the artefact; no graph. |
2026-09-28 12:06:24 +02:00
| `C-energy` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
| `C-temperature` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
| `C-confidence` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
| `C-campaign` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
| `C-metabolism` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
| `C-retirement` | Removed | — | T04 | Generalisation review | No demonstrated decision; not an implementation promise. |
T05: the lab and its labelled mutation catalogue
lab/app.py (users, tenants, auth, resources, sharing, read/write, revoke,
audit), lab/http_api.py (JSON API + browser UI, stdlib only), 20 labelled
composable version-stamped mutations, ground-truth matrix. 48 tests pass.
Detection against the reference scenario: MECHANICAL 0/10 flagged (correct),
DEFECT 6/6, SEMANTIC 2/4 with both inert cases declared.
- F-0002: M16 and M18 initially escaped detection entirely. A use case
protects exactly what it asserts. Resolved by adding two claims already
stated as intent in INTENT.md; the six-mutation catalogue would never have
surfaced this.
- test-id axis added: stable selectors survive most UI mutations, which would
make H-001 trivially false. Mutations now vary on preserves_test_ids so the
hypothesis is analysed split by that axis rather than rigged.
- M12 (semantic deferred revoke) and M19 (defect race) are behaviourally
identical and asserted as such - the discrimination problem as a test.
lab/minimal.py removed; superseded by lab/app.py.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1629012@bnt-lap001
Assistant-Session: 78d4fb13-8a1e-474b-87a3-9b9261c49a39
2026-08-22 23:31:22 +02:00
| `C-security-mutation` | C1 | `lab/mutations.py` | E-003 | `lab/GROUND-TRUTH.md` | Catalogue is hand-written; no derivation mechanism from use cases yet. |
2026-08-22 23:11:21 +02:00
2026-09-28 12:06:24 +02:00
## Generalisation and compression review — 2026-09-28
2026-08-22 23:11:21 +02:00
2026-09-28 12:06:24 +02:00
No new kernel concept was required for delegation/sequencing or tenant lifecycle.
Ordered `Step` s supply the Schedule; scenario `variant` identifies seeded defects.
No independent scheduler or Variant class was needed. Integration/security lenses
remain descriptive groupings; no Lens runtime object was consulted or validated.
The generic observer interface works, but each new domain needs a custom snapshot
collector: observation-channel adoption cost remains a real concern.
2026-08-22 23:11:21 +02:00
2026-09-28 12:06:24 +02:00
Temperature, Energy, Confidence, Campaign, Metabolism and Retirement are removed,
not deferred to another gate. `energy.py` , its exports and evidence field are
removed; H-005 is dormant-indefinite and F-0008 is resolved. Ordinary evidence,
judgments, realization metrics and lineage remain. See
[review ](../../docs/TestDriverGeneralisationReview.md ) for compatibility and limits.
2026-08-22 23:11:21 +02:00
2026-09-28 12:06:24 +02:00
Previous compression (T10): `Verdict.SUSPICIOUS` , `Step.expect_refusal` ,
`ActorIsolationError` , `World.seed` , `EvidencePack.latest()` , `Trajectory.method` .
2026-09-28 12:30:13 +02:00
T08 admission follow-up: `revisions.py` implements conservative intent-definition
identity alongside `intent.py` /`provenance.py` . `classification.py` requires a
passing reference and complete revisioned evidence; `crystallization.py` consumes
that same admission rule. These close local safety holes, not new concepts or an
increase in external-validation level. See the generalisation review.
