core/docs/analysis/universal-structure-and-field-symbol-coherence-gate-2026-06-04.md
Shay 3e2a52870d feat: Phase 2 — finite-entity grounding compiler + Phase 1.5 finding
The first comprehension->structure compiler: evals/deductive_logic/grounding.py
lowers a typed finite-entity problem (finite entities + unary predicates +
single-variable universal rules) into the propositional regime the ADR-0206
entailment operator decides, refusal-first with a closed typed reason vocabulary
(unsafe_symbol / unknown_entity / unsupported_predicate_arity / unsupported_
quantifier / malformed_case / empty_case) and collision-safe atom slugging.

finite_entity/v1/cases.jsonl: 8 cases with INDEPENDENT (oracle-derived) gold
(4 entailed, 2 unknown, 1 refuted, 1 refused) — chained rules, conjunctive
bodies, negative heads, inconsistent premises. 20 tests gate engine==oracle==gold.

This is the second diversity-panel domain (distinct comprehension, same checkable
substrate) — the universal-structure thesis validated on a different problem shape,
with the anti-overfit >=2-domain discipline now live.

Phase 1.5 finding recorded: a clean geometric/algebraic propositional decoder
agrees 716/716 with the oracle but is O(2^n) (enumeration-class), so logic is the
wrong first domain for field-as-reasoner; the wedge redirects to quantitative-
relational structure where the field is the natural non-redundant decoder.
2026-06-04 16:32:03 -07:00

243 lines
16 KiB
Markdown

<!-- CANONICAL | docs/analysis/universal-structure-and-field-symbol-coherence-gate-2026-06-04.md | 2026-06-04 | strategy/architecture-plan | the universal comprehension->structure->solve->verify spine, the binding-graph interlingua, and the field<->symbol coherence gate | verified: planning-only, no ADR number assigned; Phase 0 (INV-25) shipped -->
# Universal structure + the field↔symbol coherence gate
A plan for the most robust, clean, logical, and masterful path to **breaking
reading and inputs of any modality into a logical structure** — so the engine can
comprehend, articulate, and solve a problem in any field for which it has adequate
packs. GSM8K (math logic) is the first proving ground; the same spine must lift
GSM8K *and* be confirmed by independent problem-solving in other fields.
This document is planning-only. It assigns no ADR number, changes no serving path,
and introduces no capability claim. It records the architecture so later, gated
PRs can build it without re-deriving the rationale.
## 0. End goal
```text
input (any modality)
-> comprehended into ONE logical structure
-> reasoned over to produce a solution AND an articulated "understanding"
(what the problem is, the approach)
-> verified against INDEPENDENT gold
-> refuses when grounding is incomplete
```
Success = held-out GSM8K rises **drastically** with `wrong == 0`, **and** the
identical machinery clears an independent gold in ≥2 other fields.
## 1. The honest grounding (what exists today)
**The bottleneck is comprehension→structure, not arithmetic.** The GSM8K composer
is epistemically *unsafe* on real data (`resolve_pooled` = 2 right / 87 wrong on
holdout); bespoke per-shape readers do not compose. Capability is *real* only where
the conclusion is **checkable against independent gold** — deductive propositional
entailment (ROBDD) vs. an independent truth-table oracle: dev 200/200, holdout
500/500, `wrong = 0`.
**There are several disjoint problem-structure representations:**
- `generate/math_problem_graph.py::MathProblemGraph` — the GSM8K candidate-graph solver IR.
- `generate/derivation/model.py::GroundedDerivation` — the (unsafe-as-serving) derivation reader IR.
- `generate/binding_graph/SemanticSymbolicBindingGraph` (ADR-0132/0133/0134/0135) —
a **field-agnostic, unit-aware, provenance-carrying, acyclic, refusal-first**
semantic-symbolic DAG. It already unifies two domains: math (`bind_math_problem_graph`)
and proofs (`generate/proof_chain/builder.py`, ADR-0204 — "the binding graph's first consumer").
- `generate/proof_chain` + `generate/logic_canonical.py` (ROBDD) — the deductive decision IR.
- `generate/graph_planner.py::PropositionGraph -> ArticulationTarget` — the *articulation* IR (for saying the answer), separate from reasoning.
**The reasoning readers have zero field/algebra dependency.** `generate/derivation`,
`generate/proof_chain`, `math_solver`, `math_candidate_graph` import no
`versor_apply` / `cga_inner` / `holonomy` / `field.propagate`. The CL(4,1) field
engine participates in the **cognition turn loop** (capture field state, ratify
intent, recall, identity) — **not** in problem reasoning. Any claim of
"field-as-reasoner" is therefore currently *unverified*.
**The dormant lever.** The reliability gate (ADR-0175) already exposes
`checker="t2_precision"` ("widening past gold"), but it is inert: no genuinely
structurally-distinct second derivation exists, and nothing consumes it.
## 2. The synthesis — field comprehends, structure bridges, symbol verifies
The correct architecture is not "field OR symbol." It is a precise division of
labor with a **checkable handshake**:
- **Field/manifold engine = comprehension & constraint-resolution master (right-mode).**
A problem's entities/relations embed as geometric objects in CGA; solving =
propagating constraints via `versor_apply` until the configuration is *coherent*
(`versor_condition < 1e-6`). The converged field **is** the holistic understanding
apprehended as a whole, context-first, in CORE's *native* operations.
- **Binding-graph DAG = interlingua / corpus callosum.** The typed, unit-checked,
provenance-carrying form the field configuration is read **into and out of**: the
inspectable, checkable commitment where holistic apprehension hands off to analysis
and analysis hands corrections back. Not master, not mere servant.
- **Symbolic servants (parser, ROBDD, solver, oracle) = analyze & dispose.** They
verify the structure against independent gold, refute incoherence, articulate.
They never originate truth.
### 2.1 Why this is correct, not a compromise
**The field-derivation and the symbolic-derivation are two structurally-distinct
canonical decodings of the same problem. Their *agreement* is the `wrong = 0` gate.**
1. **It operationalizes the deepest axiom — "Truth is coherent."** Two independent
canonical decodings (geometric field + algebraic ROBDD) agreeing is coherence
across vantage points. Agreement admit; disagreement **refuse**. This is
exactly McGilchrist's "right apprehends, left analyzes, right re-integrates."
2. **It is the genuine second derivation that is currently missing.** The Tier-2
spine and `t2_precision` lever are inert because solver-vs-verifier shares
structure (decoration) and the two GSM8K readers never co-fire. **Field symbol
is that second derivation.** Activating the field as a reasoner is *what unblocks
the dormant capability lever* mechanism, not metaphor.
3. **It makes the field earn its role checkably (conforms to `wrong = 0` / no
decoration).** The field is load-bearing *only where its answer agrees with an
independent symbolic/gold derivation*. It can never serve an unagreed answer
disagreement refuses. This is INV-25 (Phase 0) generalized.
### 2.2 Conformance to the governing axes
| Axis | Conformance |
|---|---|
| Philosophical ("decoding, not generating") | Field & symbol are two canonical decodings of a reality that already is; agreement = it tracks the canonical form. |
| Axiomatic | Reasoning *is* `versor_apply` + `cga_inner` + `versor_condition` (the actual primitives); exact recall stays exact. |
| Ethical / `wrong = 0` | The field never serves an unagreed answer; refusal is the correction surface. |
| Mechanical sympathy | Field does vectorized geometric propagation (native strength, Rust-parity-ready); symbol does discrete checks. |
| Logical per "intelligence" | Holistic grasp + analytic check + refuse-on-incoherence is what integrated understanding *is*, with the non-negotiable twist that both modes must cohere or it declines. |
## 3. The earning path (the field must *prove* it reasons)
Today the readers have zero field dependency, so field-as-reasoner is a research arc
that earns its role on the **safest checkable micro-domain first**. A negative
result is an acceptable, honest outcome: where the field cannot earn agreement, it
**stays a servant** and the symbolic path carries that domain, refusing where it
cannot verify. **No unfalsifiable "the field knows" claim ever enters the codebase.**
**The wedge (Phase 1.5) — FINDING (2026-06-04): logic is the wrong first domain.**
The original wedge asked whether the field could decide *propositional entailment*
geometrically and agree with the ROBDD oracle. A bounded experiment settled it: the
cleanest geometric/algebraic encoding (the commuting-idempotent *function-algebra*,
each formula its function on the `2^n` minterms) **agrees 716/716 with the oracle
but is `O(2^n)`** it is truth-table enumeration in algebraic clothing, *not* a
genuine sub-enumeration reduction the way the ROBDD is. So a geometric decoder for
logic would either re-encode enumeration (adding no independence the oracle does not
already provide) or be decoration. **Propositional logic is combinatorial
(all-assignments), not geometric/metric** the field engine's native strength
(distance, incidence, proportion, betweenness) has no purchase there.
**Corrected wedge — quantitative-relational structure.** The field must first earn
its reasoning role where the structure is genuinely *metric*: quantitative-relational
problems (`A is twice B`, `A is 3 more than B`, part-whole, ratios) exactly where
GSM8K *comprehension* lives. There, the field derivation (solve the linear/metric
relation system by propagation) and the symbolic derivation (step-by-step arithmetic)
are two genuinely distinct decodings, and their agreement on the dataset answer is a
real second derivation. Logic keeps its independent gold (ROBDD truth-table oracle)
as a *symbolic* second derivation; the *field* earns its role in the metric domains.
This is the plan working as designed: the field earned nothing it did not deserve,
and we learned the right domain cheaply.
## 4. Phased plan
- **Phase 0 independent-gold discipline. SHIPPED.** INV-25
(`tests/test_architectural_invariants.py`): no capability claim without an
independent gold sharing no code with the SUT; deductive lane SHA-pinned
(`deductive_logic_v1`). The foundation every later phase rides on.
- **Phase 1 canonize the universal structure. SHIPPED.**
`SemanticSymbolicBindingGraph` is the documented problem-structure interlingua;
**INV-26** enforces its neutrality (the interlingua imports no field/eval/runtime,
and its core imports no domain reader only allowlisted bridges may). Deferred:
resolving the closed-vocab placeholder (`semantic_role="unknown"`) until a
load-bearing consumer defines it; the stronger "servant may not bypass consistency
checks" rule (the checkable neutrality half shipped; the rest stays doctrine until
a checkable form exists).
- **Phase 1.5 the keystone wedge. RUN (logic ruled out; wedge redirected).**
The propositional-logic wedge was settled by experiment 3): a clean geometric
encoding is enumeration-class, so logic cannot prove field-as-reasoner. The wedge
is redirected to **quantitative-relational** structure, where the field
(metric/proportion propagation) and the symbol (step arithmetic) are genuinely
distinct decodings. The dedicated quantitative wedge is the next field-reasoner
experiment.
- **Phase 2 comprehension compiler binding graph.** ** FIRST SLICE SHIPPED:**
the finite-entity grounding compiler (`evals/deductive_logic/grounding.py`) lowers
a typed finite-entity problem (entities + unary predicates + single-var universal
rules) into the propositional regime, refusal-first, gated by `engine == oracle ==
gold` the first reader proving a *different problem shape* compiles into the same
checkable substrate. **The diversity panel starts here** (see cross-cutting):
finite-entity is the second golded domain, so the compiler is validated against 2
structurally-distinct domains from its first commit and cannot overfit to one
shape. Remaining Phase 2: target the binding-graph interlingua directly and add
**field ⟂ symbol agreement** as the admission gate once the quantitative wedge lands.
- **Phase 3 activate `t2_precision`.** The fieldsymbol agreement *is* the t2
signal; wire the propose-loop consuming `checker="t2_precision"`. No invented
second derivation needed it is real now.
- **Phase 4 GSM8K on the universal structure** (the drastic-lift phase), gated by
fieldsymbol agreement + independent gold + held-out/sealed validation (never
train_sample the overfit firewall). Safe by construction.
- **Phase 5 cross-domain confirmation.** Systems/software arena (execution = a
third independent decoding; ADR-0199's first non-math arena), then a second
non-executable field with a proof-checker gold. The generalization proof.
- **Phase 6 multimodal grounding.** Connect the sensorium modality compilers
(`sensorium/compiler` `CompilerLike`/`CompilationUnitLike`) binding graph. A
modality becomes "just another reader" grounding into the universal structure;
stays afferent/gated per ADR-0198.
### Cross-cutting (every phase) — diversity panel, checkable budgets, adjoints, phenomenology
- **Structurally-diverse checkable panel (the anti-overfit instrument) woven in
from Phase 2, not deferred to Phase 5.** Multi-domain diversity is not a morale
tool bolted onto the plan; it *is* the experiment that confirms or refutes the
central thesis (that *different* subjects compile into the *same* universal
structure and are handled by the *same* solve/verify machinery). The defining
failure of this project was a single gameable ruler (the 50-case train_sample)
hiding unsoundness for weeks; the antidote is a panel of **structurally distinct**
domains, each with a **genuine independent gold** (the INV-25 bar a verifier
sharing no code with the engine). Adding a domain *without* real independent gold
is worse than useless: it is another gameable ruler. The cheap, strong starting
panel (logic / grounding / dimensional / execution / constraint, all with real
gold):
- **deductive logic** (ROBDD vs. truth-table oracle) shipped.
- **finite-entity grounding** (ROBDD/oracle gold; Phase 2).
- **units / dimensional analysis** *nearly free*: the binding graph already
carries dimensional algebra, so "does the answer's dimension check out" is an
independent structural gold.
- **systems/software** (execution = gold; Phase 5) the strongest verifier.
- **a small constraint/scheduling domain** (brute-force/SMT gold).
**The discipline (adopt now):** *a capability change must move 2 structurally-
distinct domains, or it is suspected overfitting until proven otherwise.* Make the
panel a first-class CI signal alongside GSM8K so progress is never read from one
flat number. Honesty note: a domain with no adequate pack yet simply refuses
everything that is a *coverage* signal (a breadth map), not a *capability*
signal. Capability credit counts only where there is a real pack **and** a real
gold **and** committed (non-refused) agreement.
- **Distortion budgets as first-class CI gates**, but *only checkable ones*:
grounding completeness %, unit-proof coverage, acyclicity, and **structural
perturbation stability** (paraphrase/reorder the input the binding graph and
answer are invariant, or it refuses). Regressions gate the build as hard as
accuracy. This is the antidote to metric-capture (the train_sample trap).
- **Adjoint-per-operator.** Every new transform ships with its verifier/correction
pass + a test that it reduces a *structural* error, not just a string.
- **Phenomenology harness.** Sample full structured articulations (the
"understanding": problem restatement + approach) for periodic trusted review;
qualitative depth as an explicit but *subordinate* signal (never a new truth
criterion that would re-open left-mode capture under a different name).
## 5. Risks & honest tensions
- **HIGH Phase 4 `wrong = 0`.** GSM8K serving breached it before. Mitigation:
agreement gate + independent gold + held-out/sealed + pre-push verification
(auto-merge pipeline).
- **MEDIUM field-reasoning may not pan out** in a given domain. Expected and
acceptable: it stays a servant there; the symbolic path carries it, refusing where
unverifiable.
- **MEDIUM McGilchrist over-reach.** "Field as master of reasoning" is realized
only through the *checkable* agreement gate; never asserted as unfalsifiable fact.
- **MEDIUM IR convergence cost.** Converge the disjoint reasoning IRs incrementally
(adapters first, direct targeting later), never a big-bang rewrite.
## 6. What this supersedes
The earlier 554/555/556 sequence is reframed: 554 (Tier-2 verifier) was inert
Phase 3 done *right* (fieldsymbol is the genuine second derivation); 555
(target-slot composition) was unsafe via the composer Phase 4 (comprehension on
the universal structure, gated); 556 (systems/software arena) Phase 5.