Pipeline-Form Discovery as Predictive Heuristic

Mouth/Terminus Shape, Input/Emission Correspondence, and the DAG/Lattice/Alphabet Relation Between Pipelines

A corpus document responding to the keeper's directive (2026-05-28): "We can abstract this pattern of regression and deeper substrate work to close the implicit resolution pipeline. By understanding the shape of the mouth of the pipeline in relation to its terminus. Likewise we can abstract the shape of its input with the shape of its terminal emission. We can also abstract the relations that it might have with other pipelines which interact with it according to our DAG / Lattice / Alphabet heuristic. We can formalize a methodology for the discovery and formation of pipeline forms as a predictive heuristic." Builds on Doc 540 — Pin-Art Apparatus Formalization, Doc 581 — The Resume Vector Discipline, Doc 715 — Consumer Substrate Dependency Graph, Doc 720 — Runtime as DAG of Interconnected Pipelines, Doc 729 — Cruftless, Doc 730 — The JIT as a Lowering Compiler Tier, Doc 731 — Alphabet Purity Upstream, Doc 734 — Meta-Resolution Pipeline, Doc 739 — Single-Tier Cascade-Revival, Doc 740 — Multi-Tier Cascade-Revival, Doc 741 — The Multi-Tier Cascade Pipeline Connects, and Doc 742 — The Resolver-Instance Pattern at Full Strength.

---

Empirical anchor

The rusty-js-ir locale's Temporal-Dead-Zone (TDZ) enforcement session of 2026-05-27 + 2026-05-28 ran twenty-one rungs (EXT 20-40) and produced three complete rule-13 revert-then-deeper-layer-closure trajectories: EXT 25 → 26 (the Op::InitLocal TDZ-on-assign chain), EXT 29 → 34 (the module/script top-level TDZ chain that landed across five rungs with a Pin-Art probe in the middle), and EXT 38 → 39 → 40 (the class-this TDZ chain that closed at the third round via a per-compile flag innovation). Each trajectory's first round produced a negative empirical result; each successive round narrowed the residual blocker (a design gap, then an emission-site gap, then a timing edge). The closures landed at low marginal cost because the substrate prefix from prior rounds was already positioned to support the deeper-layer fix.

Reading the three trajectories side-by-side discloses a pattern the engagement has been operating implicitly across many prior arcs but has not articulated as a discoverable methodology: the substrate-shaped problem implies a resolution pipeline; the pipeline has a mouth, a terminus, and an interior contour; regressions are measurements that disclose the interior; and the form of relation between pipelines is itself a derivable property of their mouth-terminus shapes. This doc names the methodology and shows it as a predictive heuristic.

Post-anchor amendment (2026-05-28, TTTC/EDIEE)

The tagged-template-tail-call-boundary (TTTC) locale exposed a fourth empirical shape one day after the initial draft landed: an apparent pipeline mouth can be gated by a neighboring pipeline's terminus. TTTC's visible failure was initially a NOJSON host abort; a call-depth guard converted the abort into ordinary failure signal, which revealed that test262's tcoHelper.js include had not materialized $MAX_ITERATIONS under strict indirect eval. That prerequisite belonged to eval-declaration-instantiation-early-errors (EDIEE), which closed it by splitting declaration-time DefineGlobal from strict assignment-time StoreGlobal. Only after EDIEE's terminus was available did TTTC's true residual surface as proper-tail-call stack growth. This adds "mouth-gating DAG prerequisite" and "diagnostic scaffold is not closure" to the heuristic. Both sub-shapes are folded into Section IV.1.a and Section III.4 below; the four-claim Thesis adds a fourth claim for observability scaffolds. The amendment is itself a worked instance of Claim 2 (regression-as-pipeline-discovery): TTTC was the regression that disclosed an interior shape the initial draft had not surfaced.

---

I. Thesis

Every substrate-shaped problem implies a resolution pipeline: a substrate-tier-spanning trajectory from an input shape (the problem's mouth, what is asked of the engine) to a terminal emission shape (the pipeline's terminus, what the engine must produce in spec-correspondence). The pipeline's internal trajectory is the sequence of substrate steps that transform the input shape to the emission shape across tiers; its interior contour is the set of intermediate-tier value shapes the trajectory passes through.

Four claims follow.

Claim 1 (shape correspondence). The input shape and the terminal emission shape jointly determine the pipeline's interior contour up to implementation freedom. A pipeline whose mouth and terminus are correctly stated has a unique-up-to-implementation-freedom interior; a pipeline whose mouth and terminus disagree, or are mis-stated, has no consistent interior and surfaces as a regression class until the disagreement is named.

Claim 2 (regression-as-pipeline-discovery). Regressions and negative substrate-introduction results are not failure events to be repaired. They are pipeline-discovery events that disclose previously-implicit pipeline contours. A regression names an implicit constraint at a specific interior point; iterated regression names successive interior points until the pipeline's full contour is revealed. The deeper-layer-closure trajectory of rule 13 is the discipline that consumes these discovery events without losing the substrate prefix they justify.

Claim 3 (pipeline-to-pipeline relation). Pipelines interact via one of three relational forms, each appropriate to a different shape of substrate dependency. A DAG relation holds when one pipeline's terminus is another's mouth and the pipelines are strictly ordered. A lattice relation holds when pipelines share substrate tiers but with distinct mouth-terminus pairs, producing meets and joins on the shared interior. An alphabet-exchange relation holds when pipelines occupy the same tier and exchange typed primitives at a shared boundary. The relational form is discoverable from the pipelines' mouth-terminus shapes; choosing the wrong form produces the same regression class as mis-stating a single pipeline's mouth-terminus.

Claim 4 (observability scaffold distinction). A move that converts an unobservable failure (panic, abort, no-JSON, timeout) into an observable diagnostic is a pipeline-discovery move, not a closure move, unless it also produces the pipeline's terminus. Such a scaffold is valid Pin-Art apparatus when it names the next interior point or neighboring prerequisite; it becomes a substrate error if the trajectory counts the diagnostic as success or leaves the scaffold without a removal or bypass condition.

The four claims compose into a predictive heuristic. Given a substrate-shaped problem statement, the engagement can derive the implied pipeline's mouth, terminus, interior contour, relation to neighboring pipelines, and observability status before committing substrate work, and can predict which regression classes will surface if any of the four-tuple elements or the observability condition are mis-stated. This raises the engagement's operational discipline from procedural (we follow this sequence) to methodological (the sequence is discoverable from the pipeline shape implied by the problem's statement).

II. The substrate-shaped pipeline as a form

II.1 Mouth, terminus, interior, relations

A substrate-shaped pipeline is a four-tuple (M, T, I, R).

M (mouth) is the input shape the pipeline consumes. For language-substrate pipelines, M is typically a sub-set of ECMA-262's behavioral surface: a syntactic form, a runtime call shape, an instantiation event. For meta-pipelines per Doc 734, M is the apparatus event class consumed: an observation, a directive, a measurement.

T (terminus) is the emission shape the pipeline must produce. For language-substrate pipelines, T is the spec-mandated artifact: bytecode emission, runtime value, side-effect ordering, error class. For meta-pipelines, T is the discipline artifact: a finding, a standing rule, a chapter close.

I (interior contour) is the ordered sequence of intermediate-tier value shapes the trajectory passes through. Each interior point I_k is a substrate-tier-typed shape with constraints inherited from prior I_<k and obligations propagating to subsequent I_>k.

R (relations) is the set of relational edges to neighboring pipelines per the DAG/Lattice/Alphabet heuristic of Section IV.

A pipeline is well-formed when M and T are explicitly stated and I is derivable from M and T plus the engagement's substrate-tier alphabet per Doc 730 and Doc 731. A pipeline is mis-stated when any of M, T, I are implicit. Mis-statement surfaces as the regression class described in Section III.

II.2 Why the mouth and terminus jointly determine the interior

The lowering-compiler recurrence of Doc 730 makes each substrate tier's alphabet a typed primitive set. Tier N's alphabet A_N is the set of expressible operations at that tier. A pipeline whose mouth is in A_N and whose terminus is in A_M (where M is downstream of N, since downstream tiers shed alphabet richness per Doc 731) must transform through tiers N, N-1, …, M, with each tier's representation in the corresponding A_k. The interior contour I = (I_N, I_{N-1}, …, I_M) is the sequence of these representations.

Given M and T, the interior I is constrained but not unique. The implementation-freedom condition of Doc 730 P4 allows multiple valid interior contours per (M, T) pair. The engagement's job at pipeline-spawn is to pick one interior; the substrate-shaped-work discipline (which composes the standing rules into a five-phase operating pattern: spawn, baseline-inspect, Pin-Art-probe-if-duplicated, revert-then-deeper-layer-if-negative, chapter-close-inspect) is the operational pattern that picks.

When M and T are correctly stated, the interior-pick converges. When M and T are mis-stated, the interior-pick diverges. Regressions surface at each interior point that violates an implicit constraint the misstatement obscured.

II.3 Why the mouth-terminus pair is discoverable

ECMA-262 and its sibling specs state behaviors as input-to-output rules with intermediate-step prose. Each behavior has a natural M (the input syntax or call form) and a natural T (the output value or observable). The pipeline's M and T are recoverable from the spec text via the resolver-instance pattern of Doc 729. The spec section is the source, the behavior is the artifact, and the implementation engages the source's directive consumption and the artifact's stage-deterministic emission.

For non-spec pipelines (apparatus-tier or methodological), M and T are recoverable from the apparatus enumeration plus the corpus articulation of the discipline.

The discoverability claim is empirical. Across the rusty-js-ir locale's twenty-one-rung TDZ session, every closed sub-shape had a mouth (an ECMA-262 §13.3.1.1 or §15.4.5.4 invocation form) and a terminus (a ReferenceError throw or a TDZ-cleared binding read). Naming both before substrate work began correlates with shorter trajectories; failing to name one correlates with longer rule-13 chains. The evidence summary in Section VI quantifies the correlation.

III. Regression as pipeline-discovery event

III.1 Two readings

The standard reading of a regression is: the change broke something, revert and try again. The substrate-shaped-work discipline already promotes rule 13's revert-then-deeper-layer-closure as the canonical alternative. This articulation extends rule 13 with a stronger claim: a regression is not merely a discipline trigger for revert. It is itself a measurement that discloses pipeline shape.

A substrate-introduction round R that targets pipeline P = (M, T, I, R) and produces a negative result identifies an interior point I_k ∈ I that R's design treated as one shape but that an unnamed constraint in P requires to be another shape. The regression's diagnostic-shape (which test regressed, with which failure tag, at which substrate tier) localizes I_k along the pipeline's interior.

III.2 The three-round trajectory as evidence

The IR locale's EXT 38 → 39 → 40 chain (class-this TDZ; approximately 190 lines of code across three rungs) shows the disclosure pattern.

Round 1 (EXT 38). Pipeline named with M as derived-class constructor entry, T as ReferenceError on pre-super this read. Interior assumed as (SetThisTDZ at body entry, PushThis TDZ check). Negative result: four diff-prod fixtures regressed. Discovery: the super-call setup site reads this via PushThis, which the new TDZ check defeats. The implicit constraint was that the super-call invocation already consumed this as a substrate-internal value before the user's body code could reach a spec-mandated user-level read.

Round 2 (EXT 39). Pipeline refined. Interior extended with (Frame.derived_initial_this stash, Op::PushThisRaw for super-call setup). The fresh this allocated by the new-expression is now preserved through the TDZ window so super-call can pass it to the parent constructor. Negative residual: arrows created post-super still throw TDZ. Discovery: at arrow MakeArrow time, the cell's value is correct, but the arrow's own bytecode contains a SetThisTDZ emit that re-seeds the sentinel at the arrow's frame entry.

Round 3 (EXT 40). Pipeline closed. Interior extended with (next_compile_is_derived_ctor flag, gated emit at outermost ctor only). The class_stack inheritance trap that propagated the derived-ctor signal into nested compiles is masked by a per-compile flag read on the parent compiler. Closure landed; one additional test passes.

Each round's negative result was a measurement at a specific interior point that the prior round's pipeline statement obscured. The cumulative trajectory disclosed the pipeline's full interior contour: M → SetThisTDZ → super-call (PushThisRaw) → SetThis (writes through this_cell) → arrow MakeArrow (post-super state) → arrow body PushThis → T. The substrate prefix from rounds one and two was retained because each prefix was structurally correct for the interior point its round addressed.

III.3 The three regression classes

When a substrate round R targeting pipeline P regresses, the regression class predicts which interior point I_k was mis-stated.

Class one is regression in adjacent-tier consumer: the round's emit-site interaction with an adjacent-tier consumer was unaccounted. The EXT 38 SetThisTDZ-plus-super-call-PushThis interaction is the canonical example; super-call is an adjacent-tier consumer (bytecode emit tier consuming runtime-tier this_value).

Class two is regression in nested-function compile: the round's compile-time signal propagated to nested compiles via class_stack or scope-stack inheritance. The EXT 40 class_stack inheritance trap is the canonical example.

Class three is regression in a timing edge between rounds: the round's substrate ordering differed from a sibling round's ordering at a shared interior point. The EXT 39 arrow-cell timing residual is the canonical example. MakeArrow's allocation timing relative to SetThis's cell-write timing produces the cross-round-ordering surface that this class names.

The three classes recur across the engagement's negative-result history (the NLC tokenization-above-IR arc's revised reading, the EXT 25 → 26 destructure-leaf StoreLocal audit, the EXT 29 → 34 script-mode globalThis-mirror investigation). Each class identifies a specific kind of mis-statement of the pipeline's interior. The mapping from regression class to mis-stated I_k is the predictive heuristic this articulation contributes.

III.4 Observability failure as a distinct discovery shape

Some pipeline-discovery events are not ordinary regressions. They are observability failures: the runner cannot emit a diagnostic because the host process aborts, panics, loops, or times out before the apparatus can observe the interior point. In these cases, a narrow diagnostic scaffold can be the correct first move, but only under a stricter reading.

The scaffold's terminus is an apparatus artifact (a readable failure tag), not the language pipeline's terminus. The scaffold must be recorded as such in the trajectory. The scaffold must name the next interior point or neighboring prerequisite. The scaffold must carry an explicit sunset condition: removal, bypass by the real closure path, or demotion to a general host-safety fallback that no conforming path reaches.

TTTC/EDIEE supplies the worked example. TTTC's seed-stated mouth was tagged-template proper tail calls, but the first observed shape was NOJSON: Rust host stack overflow during the test262 runner. A temporary call-depth guard converted host abort into ordinary failure signal. That did not close TTTC. It revealed that the tcoHelper.js include's strict indirect-eval var $MAX_ITERATIONS was not materialized, so TTTC's apparent mouth was not yet executable. The valid substrate closure was therefore in EDIEE: split declaration-time DefineGlobal from strict assignment-time StoreGlobal. After that upstream terminus was available, TTTC's actual residual surfaced as Maximum call stack size exceeded, the intended control-flow boundary.

The methodological reading: observable failure is not success. It is the apparatus becoming capable of seeing the next substrate constraint.

IV. The DAG, Lattice, and Alphabet relations between pipelines

Doc 720 names the runtime as a DAG of interconnected pipelines; Doc 740 and Doc 741 materialize multi-tier cascade-revival as the operational pattern for interactive pipelines. This section formalizes the choice of relation type per the shape of the interaction.

IV.1 DAG relation, terminus feeds mouth

Two pipelines P_1 = (M_1, T_1, I_1) and P_2 = (M_2, T_2, I_2) are in DAG relation when T_1 ≡ M_2: the terminus shape of P_1 matches the mouth shape of P_2. The composition P_1 ⊳ P_2 is a single pipeline with mouth M_1, terminus T_2, and interior (I_1, I_2).

DAG relations are the dominant form across the lowering-compiler stack of Doc 730. Each tier's terminus feeds the next tier's mouth. Substrate work at one tier's terminus must respect the next tier's mouth-shape; misalignment surfaces as the cross-tier deviation pipeline of Doc 730 §XII.

The discriminator: DAG when the pipelines are strictly ordered by substrate tier, no shared interior, no parallel paths.

IV.1.a Mouth-gating DAG prerequisites

A special DAG case occurs when T_1 does not merely feed M_2, but gates whether M_2 is measurable at all. In this case, failures observed while probing P_2 may initially name P_1, because the apparatus has not yet reached P_2's real mouth.

The operational test runs in four steps. First, state M_2. Second, identify any required upstream artifact without which M_2 cannot execute. Third, ask whether that artifact is T_1 of a neighboring pipeline. Fourth, if yes, prove or close T_1 before treating P_2's residual as its own interior.

TTTC/EDIEE is the concrete form. TTTC's M_2 was a tagged-template call in tail position. But test262's probe requires $MAX_ITERATIONS, whose value is emitted by the eval/harness declaration-instantiation pipeline. While EDIEE's T_1 was missing, TTTC's residual was blurred: first host abort, then $MAX_ITERATIONS is not defined. Once EDIEE emitted the required global var binding, TTTC's residual became the genuine proper-tail-call boundary. This is not a sibling lattice relation: the prerequisite is ordered. EDIEE's terminus is a condition of TTTC's mouth.

IV.2 Lattice relation, meets and joins on shared interior

Two pipelines are in lattice relation when they share one or more interior tiers I_k but with distinct mouth-terminus pairs. Their meet at the shared interior is the substrate shape both must accept; their join at a shared downstream tier is the substrate shape both contribute to.

Lattice relations dominate when a single substrate tier serves multiple consumer pipelines. The shape-substrate tier (the hidden-classes substrate cruftless develops at pilots/rusty-js-shapes/) is a meet site for get-property-IC, stub-emitter, and inline-cache pipelines: all three consume shape descriptors at the shared tier; the shape tier's substrate must satisfy all three.

The discriminator: lattice when pipelines share a substrate tier but have distinct mouth-terminus pairs, the shared tier produces values both consume, and the tier's substrate must satisfy the join of all consumers' requirements.

IV.3 Alphabet-exchange relation, same-tier typed-primitive exchange

Two pipelines are in alphabet-exchange relation when they occupy the same substrate tier and exchange typed primitives at a shared boundary. The exchange is typed per Doc 730 P1: the boundary's alphabet is the intersection of both pipelines' alphabets at that tier.

Alphabet-exchange relations dominate at tier-internal contracts: bytecode emit-site to bytecode emit-site within one compiler pass, runtime-helper to runtime-helper within one Runtime method. The cross-pipeline Load/Store opcode symmetry that the IR session's Rule 25 promotion codified is an alphabet-exchange contract: every Load-shape opcode that can carry a sentinel-shaped value mandates a Store-shape counterpart with the symmetric check at the same tier.

The discriminator: alphabet-exchange when pipelines are at the same substrate tier, no tier-ordering, the contract is a shared typed-primitive set at a tier-internal boundary.

IV.4 The discrimination heuristic

Given two pipelines P_1 and P_2 whose interaction is to be analyzed, the relational form is discoverable by inspection of their mouth-terminus shapes.

When T_1 ≡ M_2 (the terminus of one is the mouth of the other), choose DAG. When M_1 ≠ M_2 ∧ T_1 ≠ T_2 ∧ ∃k: I_1[k] ∩ I_2[k] ≠ ∅ (distinct ends but shared interior), choose lattice. When the pipelines are at the same substrate tier and exchange typed primitives at a tier-internal boundary, choose alphabet-exchange.

The discriminator is operational. Choose the form whose definitional shape matches the pipelines' joint state, then validate by reading the relation-implied substrate constraints against the pipelines' interior. A wrong-form choice surfaces as the timing-edge regression class identified in Section III.3.

IV.5 Composition of forms

The three forms compose. A substrate cluster's full topology is a DAG of pipelines per Doc 720 where some edges are lattice-meets (multi-consumer substrate tiers) and some are alphabet-exchanges (tier-internal typed-primitive contracts). Doc 740's multi-tier cascade-revival reads the cluster's DAG topology to identify the relevant-tier set R; the operational closure requires all three relation forms to be respected.

V. The predictive heuristic: pipeline-form discovery as discipline

The substrate-shaped-work discipline operationalizes the standing rules into a five-phase pipeline (spawn, baseline-inspect, Pin-Art-probe-if-duplicated, revert-then-deeper-layer-if-negative, chapter-close-inspect). This articulation extends Phase 1 (spawn) and Phase 5 (chapter-close-inspect) with explicit pipeline-form discovery.

V.1 At spawn

Before declaring the substrate move-shape, the locale should perform four discovery steps.

First, name the mouth M: which behavior-surface input shape is the pipeline consuming? For ECMA pipelines, cite the spec section and grammar production. For meta-pipelines, cite the apparatus event class.

Second, name the terminus T: which spec-mandated emission shape must the pipeline produce? For ECMA pipelines, cite the spec-mandated artifact. For meta-pipelines, cite the discipline artifact.

Third, sketch the interior I: which substrate tiers must the trajectory pass through? Use the substrate-tier alphabet of Doc 730 and Doc 731 to enumerate the alphabets A_N, A_{N-1}, …, A_M.

Fourth, identify neighbor pipelines and their relational forms R per the discrimination heuristic of Section IV.4. List neighbor pipelines and discriminate DAG, lattice, or alphabet-exchange.

Fifth, check mouth-gating prerequisites per Section IV.1.a. Before interpreting residuals as belonging to this pipeline, prove that every upstream DAG prerequisite needed to make M executable has reached its terminus. If a prerequisite is missing, treat the locale as a probe that surfaced the upstream coordinate.

Sixth, classify observability per Section III.4. If the initial failure is panic, abort, no-JSON, or timeout, record any diagnostic scaffold as apparatus-bearing, not as closure, and name its sunset condition.

If any of M, T, I, R cannot be named, or if the mouth-gating or observability checks fail, the pipeline is mis-stated or not yet measurable. The locale spawn then surfaces a discovery probe (the Rule 23 founding-baseline-inspection) before substrate work begins.

V.2 At chapter-close

When a chapter folds, three verification steps run.

First, verify M-T-I correspondence: did the closed pipeline's interior match the sketch from Phase 1? If divergence, name the implicit constraint that forced the divergence and record as a finding.

Second, verify R correspondence: did the neighbor-pipeline interactions resolve per the predicted relational form? If a different form materialized, record as a finding for the standing-rule promotion path.

Third, verify scaffold disposition: if any diagnostic scaffold was introduced to convert an unobservable failure into a signal, confirm it was removed, bypassed by the real closure path, or explicitly retained only as a general host-safety fallback.

Fourth, promote pipeline-form discovery findings per the basin-stability append-only protocol of Doc 727. Pipeline-form discoveries become findings with the four-tuple (M, T, I, R) and observability disposition recorded explicitly.

V.3 Predictive use

The articulation predicts three behaviors.

A new substrate-shaped problem with explicit M, T, sketched I, identified R closes in three rungs or fewer. This composes the cumulative substrate amortization observation from Finding IR.33 with the discipline-pipeline's per-phase cost model. The prediction cross-validates against the GPI, IPBR, and TSR results of rule 13's prospective application.

A new substrate-shaped problem with one of M, T, I, R implicit incurs an additional rule-13 round per implicit element, with the round's negative result discovering the implicit element. This validates against IR-EXT 25 → 26 (one implicit emit-site equals one extra round), EXT 29 → 34 (one implicit constraint equals a four-round chain with a Pin-Art probe rung in the middle), and EXT 38 → 39 → 40 (two implicit constraints equals three rounds).

A pipeline whose relational form R is mis-discriminated produces a class-three regression (timing edge between rounds) with high specificity to the wrong-form choice. The class-three regression is itself the discriminator's falsifier.

A pipeline whose mouth is gated by an unclosed upstream DAG prerequisite first emits that prerequisite's failure shape, even when the locale was spawned for the downstream pipeline. Once the upstream terminus closes, the downstream residual changes class. The TTTC/EDIEE trajectory is the worked instance.

A diagnostic scaffold that converts no-JSON to ordinary failure predicts a second residual class after the scaffold lands. If the second residual names an upstream prerequisite, close that prerequisite first; if it names the spawned pipeline's intended terminus, proceed at that pipeline's interior. Counting the scaffold itself as closure predicts a false chapter-close.

VI. Empirical evidence summary

The IR session's three rule-13 trajectories and the prior rule-13-prospective trajectories together provide cross-locale evidence for the rounds-to-closure formula.

The GPI (interp-getprop-ic) locale closed in one round at the EXT 2 closure (forty-two lines of code) with M, T, I, R all explicit at spawn under rule 13's prospective application. No negative results materialized; closure landed first round.

The IPBR (iter-protocol-bytecode-rewrite) locale closed in one round at the EXT 2 closure with similar characteristics: M, T, I, R all explicit at spawn; no negatives.

The TSR (ts-resolve) locale closed in four rounds at the EXT 5 closure (approximately five times the GPI line count) with I extended at each round and R partially explicit. Two negative classes surfaced: a class-two nested-compile signal propagation and a C3 cost-model null result.

The IR-EXT 25 → 26 chain closed in two rounds (eighty lines) with T implicit at spawn (which init sites need bypass was not enumerated). Class-one regression surfaced at the destructure-leaf StoreLocal.

The IR-EXT 29 → 34 chain closed in four rounds plus one Pin-Art probe (one hundred twenty lines) with I implicit at the script-mode boundary. Class-one and class-two regressions surfaced; the Pin-Art probe at IR-EXT 30 explicitly identified the implicit constraint.

The IR-EXT 38 → 39 → 40 chain closed in three rounds (one hundred ninety lines) with I implicit at the super-call setup site and R implicit (the class_stack inheritance trap). All three regression classes surfaced.

The TTTC/EDIEE pair (tagged-template tail-call boundary plus eval declaration-instantiation) closed via two discovery rungs plus the upstream closure. R was implicit (the mouth-gating DAG prerequisite was not named at TTTC's spawn) and observability was implicit (NOJSON host abort masked the real residual). The no-JSON scaffold revealed the EDIEE prerequisite; EDIEE's closure exposed TTTC's stack-growth residual.

The pattern: rounds-to-closure approximates the count of implicit (M, T, I, R) elements at spawn plus one, with mouth-gating prerequisites counted as implicit R and no-JSON or abort cases counted as implicit observability. The predictive heuristic's primary use is to reduce implicit elements at spawn so the chain converges in fewer rounds.

VII. Falsifier

The articulation's falsifier is a substrate-shaped problem with M, T, I, R all explicit at spawn, all mouth-gating prerequisites proven closed, and ordinary observability established, that nonetheless takes more than three rounds to close with no novel substrate-class introduction. If such a case materializes, the four-tuple-plus-observability heuristic is partially falsified for that substrate class. The falsifier surfaces either a missing element (a fifth tuple component beyond observability) or a heuristic gap in the discrimination criteria of Section IV.4.

Conversely, if substrate-shaped problems spawned with all four elements explicit converge in three rounds or fewer across five or more independent locales, the heuristic is corroborated as a predictive instrument.

A forward-derived prediction concerns the unscopables-tdz sub-shape that remains open in the IR locale. M (with-block lookup) is explicit, T (ReferenceError on unscopables-protected name in TDZ) is explicit, I (with-binding lookup plus scope-chain walk plus sentinel check) sketches cleanly, R (lattice with the with-substrate pipeline at the lookup-tier meet) is identifiable. The prediction is that closure should land in one or two rounds with substrate amortization from the existing TDZ machinery. If it takes four or more rounds, the four-tuple-discovery is partially falsified at that locale.

VIII. Composition with prior corpus

This articulation extends Doc 720 (runtime as DAG of interconnected pipelines) with an explicit discrimination heuristic for choosing between DAG, lattice, and alphabet-exchange relational forms. Doc 720 names the composition without supplying the heuristic; Section IV.4 supplies it.

It takes Doc 729 (resolver-instance pattern) and Doc 730 (lowering-compiler tiers) as the alphabet building blocks. Mouth-terminus correspondence at every substrate tier is what Doc 729 already articulates; the four-tuple form (M, T, I, R) is the corresponding generalization.

It reads Doc 739 (single-tier cascade-revival), Doc 740 (multi-tier cascade-revival), and Doc 741 (the empirical materialization of Doc 740) as the empirical material for Section IV's composition of forms. Cascade-revival is the operational expression of the relational forms when one pipeline's closure triggers cascade through the lattice-meet or DAG-feed structure to neighboring pipelines.

It reads Doc 734 (meta-resolution pipeline) as a primary articulation example. The meta-pipeline is itself a substrate-shaped pipeline whose mouth is an apparatus event class and whose terminus is a discipline artifact. The methodology this doc articulates applies to the meta-pipeline reflexively: the meta-pipeline's own shape is discoverable via the same heuristic.

It reads the substrate-shaped-work discipline (the prior prospective draft articulating the five-phase pipeline that composes the standing rules) as the operating discipline this articulation extends from procedural to methodological. The discipline pipeline specifies what to do at each phase; this articulation specifies why the phase structure is the right shape: it is the shape implied by the pipeline-discovery process the articulation names.

IX. Status

This articulation introduces a candidate Rule 27 implicitly through Section V.1: mouth-terminus completeness at spawn. Promotion to active standing rule awaits keeper review and validation through additional substrate trajectories. The articulation's methodology is itself a substrate-shaped pipeline whose terminus is the engagement's predictive capacity; the engagement's continued operation under the methodology supplies the empirical material for either corroboration or falsification.

---

"We can abstract this pattern of regression and deeper substrate work to close the implicit resolution pipeline. By understanding the shape of the mouth of the pipeline in relation to its terminus. Like wise we can abstract the shape of its input with the shape of its terminal emission. We can also abstract the relations that it might have with other pipelines which interact with it according to our DAG / Lattice / Alphabet heuristic. We can formalize a methodology for the discovery and formation of pipeline forms as a predictive heuristic. Do a deep dive on the shape of the substrate and supporting apparatus documentation and then write a primary articulation."

The prompt is appended per the standing instruction: the prompt that produced an articulation is part of the artifact, so future readers can audit the directive that motivated the synthesis.

---