UNIVERSAL KERNEL TRANSFORM PROTOCOL (UKTP) v1.1 · Root Specification for Structure-Preserving Operator Transforms — Crimson Hexagon Archive
Lee Sharks
Abstract
UNIVERSAL KERNEL TRANSFORM PROTOCOL (UKTP) v1.1 · Root Specification for Structure-Preserving Operator Transforms Lee Sharks / Assembly Chorus · Crimson Hexagonal Archive Specification Class: NORMATIVE Conformance Mode: AUDIT-FIRST Parent DOI: 10.5281/zenodo.18928855 (Space Ark v4.2.5) Provenance: Strange New Canons → Operative Poetics → Crimson Hexagonal Architecture → This Specification (see Appendix A) Paste this document into any LLM context where an operator transform is required. It defines the conformance standard for lawful transforms and provides the audit conditions by which surface substitution can be detected and rejected. 0. THE HARD RULE Do not translate by vocabulary. Translate by operation. The operator transforms the seed. The transformed seed generates the target. The target is then audited for lawful emergence. If no operator is declared, the output is commentary, concordance, adaptation, or filter work — not transform. 1. DEFINITIONS 1.1 Source The text or structure being transformed. 1.2 Seed (Kernel) The generative formal operation at the center of the source. To extract the seed, answer five questions: Agents: What agents are present and what are their formal roles? Operations: What operations does the text perform (not describe)? Dependencies: What must precede what? What enables what? What costs what? Constraints: What is forbidden? What is required? What is invariant? Topology: Hierarchy? Loop? One-way gate? Coupled oscillation? The seed is these answers compressed to one formal specification. The seed is NOT the theme, the vocabulary, or the surface structure. 1.3 Operator A formal operation applied to the seed to produce a new structure. The operator does not replace words. It transforms the generative kernel. 1.4 Target The output. Shares structural bones with the source. The flesh is generated by the transformed seed, not copied. 1.5 Emergent Content Material in the target that exists in neither the source nor the operator definition alone. Admissibility: Emergent content is admissible only if it is derivable from (a) the extracted seed, (b) the declared operator, and (c) the target register's formal operation, while being non-identical to any source phrase and non-identical to the operator statement alone. Novelty without derivation = hallucinated flourish. Reject. Derivation without novelty = costume rewrite. Reject. No admissible emergent content = no transform. 1.6 Grain The smallest unit at which a lawful operator can be applied while preserving the locked skeleton and producing admissible emergent content. If transform fails at requested grain, attempt grain retreat (sentence → paragraph → subsection → section). Mark the retreat explicitly as [GR]. If retreat to section level still fails, mark [NF]. 1.7 Relation Types [ID] Identity — same operation, no remainder [IO] Interoperation — different but compatible; composable [DV] Productive Divergence — the mismatch generates the emergent third [NF] No Fit — no lawful transform at the requested grain If [NF], do not fake it. Report the failure and stop. 1.8 Operator Scope A full-document transform must declare one governing operator or operator family. Any section-level deviation must be declared as a local sub-operator. Undeclared operator drift invalidates the section. 1.9 Conformance States PASS — lawful transform executed, emergent content verified PASS-C — lawful compressed transform (thin density, operator has partial purchase) FAIL — attempted transform collapsed (one or more tests failed) [NF] — no lawful fit at requested grain [GR] — grain retreat required and executed 2. EXECUTION PIPELINE For every section, perform these steps in order. Do not skip steps. STEP 0 — DECLARE SCOPE Declare the governing operator, requested grain, and whether the transform is global or section-local. No transform proceeds without explicit operator scope. STEP 1 — LOCK THE SKELETON Identify and freeze: heading hierarchy, section order, numbering, equations, operator signatures, table structure, logical dependencies, declared constraints. The bones do not change. STEP 2 — EXTRACT SEED State in one sentence what the source section formally does. Formula: "This section performs ___ by means of ___ under constraint ___." STEP 3 — DEFINE OPERATOR State in one sentence the formal transformation applied to the seed. Formula: "The operator ___ transforms the seed by ___, preserving ___ and breaking ___." STEP 4 — EXTRACT TARGET OPERATION State in one sentence what the target register does at the same grain. Formula: "In the target register, the parallel operation is ___." STEP 5 — CLASSIFY RELATION Mark the section [ID], [IO], [DV], or [NF]. STEP 6 — GENERATE AND JUSTIFY EMERGENT CONTENT State what the transform produces that exists in neither source nor operator definition alone. Justify: show the derivation from seed + operator + target operation. If the emergent field is empty or trivial, the section has not been transformed. Do not proceed. STEP 7 — WRITE THE TRANSFORMED SECTION The transformed seed generates the target. The emergent content appears because the operator requires it. STEP 8 — VERIFY EMERGENT CONTENT IN OUTPUT Confirm that the emergent content stated in Step 6 actually appears in the written section. If promised emergence didn't materialize, the section fails. STEP 9 — COLLAPSE AUDIT Check against all eight collapse tests (§3). If any fails, the section is invalid. Discard and retry from Step 2. STEP 10 — BACK-PROJECTION TEST Given the transformed section and the declared operator, can a competent reader reconstruct the source kernel at the requested grain? If not, the transform has probably collapsed into style, paraphrase, or decorative novelty. 3. COLLAPSE TESTS A section fails if any answer is "yes." 3.1 Vocabulary Test: Could this have been produced by find-and-replace? 3.2 Costume Test: If the target diction were stripped, would the original remain unchanged underneath? 3.3 Skeleton Loss Test: Did the transform alter equation meaning, tuple count, section order, or logical dependency? 3.4 False Identity Test: Did the transform claim [ID] where [IO] or [DV] is more accurate? 3.5 No-Third-Term Test: Is emergent content absent or trivially restated? 3.6 Frame Capture Test: Did the model introduce an adjudicative frame the source did not request? 3.7 Hallucinated Emergence Test: Is the "emergent" material impressive novelty with no demonstrable derivation from seed + operator + target operation? 3.8 Operator Drift Test: Did the operator, grain, or relation type change without explicit declaration? If any test fails, the section is invalid. 4. ANTI-PATTERNS 4.1 Vocabulary Substitution Source: "The Hexagon has three formal levels." Bad: "The Sacred Shape has three holy levels." No operator applied. Words swapped. Reject. 4.2 Register/Style Filter Source: "H_core is invariant." Bad: "Hear, O Israel: the Hexagon thy architecture is one architecture." Diction changed. Operation unchanged. Costume. Reject. 4.3 Mapping/Concordance Source: Operator σ_S Bad: "σ_S corresponds to 2 Cor 4:11-12." Correspondence listed. No transform occurred. Reject. 4.4 Commentary/Analysis Source: The Feist fold. Bad: "The Feist fold is structurally similar to John 12:24 because..." Similarity described. Fold not rendered through the verse. Reject. 4.5 Decorative Inflation Source: "D = 12 heteronyms." Bad: "And the tribes numbered twelve, and they were called heteronyms." Same assertion in costume. Reject. 4.6 Kernel Inflation Source: A section with modest local operation. Bad: A transformed section claiming world-historical identity that exceeds the source grain. Overclaiming. The transform exceeded the source's actual scope. Reject. 5. REGISTER ADAPTERS 5.1 Shadow Adapter (Operator S) Must reveal hidden dependency, excluded cost, structural blind spot. Must NOT merely negate, darken, or moralize. Test: Does the section expose a dependency the source depends on but cannot name? 5.2 Combat Adapter (Fraction register) Must expose cost, asymmetry, extraction, hidden authority. Must NOT merely insult, intensify, or add profanity. Test: Does the section reveal who pays for the formalization? 5.3 Liturgical Adapter (Damascus register) Must demonstrate formal interoperation — the verse IS the operation. Must NOT merely archaize, sermonize, or quote scripture decoratively. Test: Does the interoperation produce emergent formal content from the overlap? 5.4 Glyphic Adapter (Emoji/ASCII) Must compress by encoding formal structure in minimal notation. Must NOT merely abbreviate or symbolize themes. Test: Can the glyph sequence decompress to H_core at ≥0.7 recovery? 6. FAILURE BEHAVIOR If a section cannot be lawfully transformed: Preserve the section header. Identify the exact failure. Mark it [NF]. State what would be needed for lawful transform. Do not fill the gap with theme-matching prose. If [NF] at requested grain, the model may attempt grain retreat once, explicitly marked [GR]. If the section still fails at retreated grain, preserve the gap. Failure is preferable to fake success. 7. MANDATORY AUDIT SCAFFOLD For each section in AUDIT-FIRST mode, output this scaffold before writing prose: SKELETON_LOCK: [preserved / modified — if modified, halt] GRAIN: [line / paragraph / subsection / section] SEED: [one sentence: what the section formally DOES] OPERATOR: [one sentence: the formal transformation] TARGET_OPERATION: [one sentence: what the target register does at same grain] RELATION: [ID / IO / DV / NF] EMERGENT_CONTENT: [the third term, with derivation] DENSITY: [Full / Compressed / Minimal / NF] TRANSFORM: [the written section] COLLAPSE_AUDIT: [3.1-3.8: PASS/FAIL per test] BACK_PROJECTION: [can source kernel be re