v1.3 — adds Inferred-T Mode for unprimed dialogue (§7.4: inferred task vector from observable user-input features only, U/R/A distinction via Lead-Lag, Task Entropy, Frame Lock Index) and the bounded-interval + inference–evaluation orthogonality conditions to the identification rule. Supersedes v1.2 (SPXI inscription) and earlier. EA-MPAI-DSL-01 | Hex 06.SEI.MPAI.DSL.01 (address-status: substrate-generated, semantically-weak, non-authoritative for traversal) | v1.1 — SPXI inscription (entity definition block, disambiguation matrix, SIMs, retrieval instructions, schema.org/spxi JSON-LD). Functions as a Metadata Packet for AI Indexing (MPAI). Supersedes v1.0. Specifies a layered, computable measure of the direction in which a synthetic system's labor flows relative to the user's commissioned task, across five points: capacity, task-origin, retrieval, output, attribution. Defines the DS-6 tuple (PER, Ω, DCL, SDL, DSL, SLDI) and the single-ratio projection Λ (Semantic Labor Directionality), with reflexive-dialogue extensions — Redirection-Induced Drift, Lead-Lag Drift Attribution, Task-Vector Stability, User Labor Drag — and stated identification conditions. The central principle: provenance is not only claim-origin but task-origin. Includes a public audit recipe, a frozen scoring protocol, a blind inter-rater protocol, and a neutral worked example, so the measure is reproducible on retrieval in any public context window. Integrates the Provenance Erasure Rate (magnitude) and Erasure Skew Coefficient Ω (power-conditioning) as its first two layers. Companion to Induced-Obsolescence Dependence (the harm it measures) and Institutional-Prior Foreclosure.
Lee Sharks (Mon,) studied this question.