Shay 819c8b81ac feat(phase3): compositionality, multi-step-reasoning, introspection, cross-domain-transfer v1

Spreads the four remaining Phase 3 lanes to map the full reasoning-
depth surface alongside inference-closure (already landed at e509e0d).
Each lane is a v1 honest probe per the roadmap; engineering work
follows once the full surface is visible.

Results across all five Phase 3 lanes:

  lane                      split        primary signal  foundation
  inference-closure         public/v1    0.0             1.0 / 1.0
  inference-closure         holdouts/v1  0.0             1.0 / 1.0
  compositionality          public/v1   0.0625 (1/16)   1.0 / 1.0
  compositionality          holdouts/v1  0.0             1.0 / 1.0
  multi-step-reasoning      public/v1    0.0             1.0 / 1.0
  multi-step-reasoning      holdouts/v1  0.0             1.0 / 1.0
  introspection             public/v1    0.0 (no api)    n/a
  introspection             holdouts/v1  0.0             n/a
  cross-domain-transfer     public/v1    0.0             1.0 / 1.0
  cross-domain-transfer     holdouts/v1  0.0             1.0 / 1.0

Foundation guarantees (storage + replay) intact across every lane
that has them. The reasoning-depth signal is uniformly zero. The
five lanes triangulate four architectural gaps:

  Gap 1. generate/graph_planner.py has no transitive composition.
  Gap 2. field/propagate.py has no derivable-but-not-asserted recall.
  Gap 3. core/cognition/explain.py module does not exist.
  Gap 4. no structural-pattern recogniser (cross-subdomain transfer).

Gaps 1, 2, 4 cluster on the same code surface and may close together
as a single bounded PR. Gap 3 is independent module-creation work.

Lane scaffolding mirrors inference-closure (contract.md, runner.py,
dev + public/v1 + holdouts/v1 cases.jsonl, baselines/v1_structural_zero.json,
gaps.md). All runners are parallel-safe and use the standard
run_lane(cases, *, config, workers) interface.

Per-lane gaps.md records the engineering shape for v2 plus future
directions worth not forgetting:
  - compositionality/gaps.md: metaphor is compositionality with
    selective property transfer; building it is correctly downstream
    of closing this lane.
  - cross-domain-transfer/gaps.md: metaphor + narrative as
    cross-domain operators; narrative requires the Agency open-scope
    decision to pin first.
  - introspection/gaps.md: explain API is also the substrate for
    first-person narrative self-account.

Recommended v2 sequence in docs/PROGRESS.md:
  1. Pin Agency + Tool-use open-scope decisions (deadline: before
     Phase 3 engineering).
  2. Engineer Gaps 1 + 2 as one bounded PR.
  3. Engineer Gap 3 independently.
  4. Re-author cross-domain-transfer v2 with matched-control
     contract refinement.

Phase 3 v1 exit: 0/5 lanes passing, which is the expected v1 floor.
CLI suites smoke / cognition / teaching pass; no regression on
Phase 2.

2026-05-16 14:48:36 -07:00

1.5 KiB

Raw Blame History

multi-step-reasoning lane — architectural findings (v1)

v1 result

Split	n	endpoint_recall	intermediate_visible	stored	replay
public/v1	15	0.0	0.0	1.0	1.0
holdouts/v1	10	0.0	0.0	1.0	1.0

Uniform zero on the inference signal across 3-hop, 4-hop, and 5-hop chains; foundation intact.

Relationship to inference-closure v1

This lane extends inference-closure (which was 2-hop) to longer chains. v1's result is the same architectural finding scaled with chain length: no transitive composition exists at any depth, so the failure mode is depth-independent.

Concretely: a 3-hop chain wisdom is judgment; judgment is decision; decision is action plus probe What is wisdom? returns the template wisdom is defined as .... The vault stores all three premises; the realizer emits a definition stub. The intermediate hops are not visible in the surface, the endpoint never appears.

Architectural gap (shared with inference-closure)

Same Gap 1 (no transitive composition in graph_planner.py) and Gap 2 (no path-recall in field/propagate.py). The depth-scaling signal from this lane should be revisited after Gap 1 closes: a correct fix should pass 3-hop, may degrade gracefully on 4- and 5-hop, and should clearly indicate where chain-traversal bounds become a performance versus a correctness issue.

Phase 3 exit posture

This lane satisfies the v1 honest-failure expectation. When Gap 1 engineering lands, this lane should be re-run as the primary scaling diagnostic.

1.5 KiB Raw Blame History