SPXI is not schema markup. JSON-LD is one surface of SPXI, the final inscription format for entity-definition packets. It is not the methodology. Schema.org and JSON-LD remain useful components of the modern web stack. SPXI does not replace them. It specifies the measurement and survivability discipline that makes them operationally consequential in AI-mediated retrieval.
JSON-LD ⊂ SPXI — JSON-LD is a proper subset of SPXI. The protocol contains it.
SPXI ⊄ Schema — SPXI is not a subset of Schema. It exceeds the category.
What surrounds the JSON-LD — measurement, compression, protection, dispersal, and durability testing — is what determines whether an entity appears in an AI-generated answer or is compressed out of it.
SPXI begins with measurement. Before structured data is written, the client's semantic presence is scored across five instruments.
Foundational metric. Scores any text for compression survival, 0–1, with subscores for citation density, structural integrity, argument coherence, and provenance markers. γ < 0.3 is ghost meaning: structurally present, semantically invisible. γ > 0.7 indicates high compression survivability.
Empirical simulation: content is submerged in LLM-based compression and we measure what surfaces. γ predicts survivability from structure. The Drowning Test verifies it. The gap between them is diagnostic.
Ratio of semantically load-bearing content to total content, 0–1. Target: Δ > 0.6. Low Δ predicts material dropped during summarization.
Rate of change in retrieval-layer presence over time, monthly percentage. Negative SDD = improvement. Positive SDD = loss velocity.
Frequency of attribution being dropped from AI summaries that use your content, 0–1. Target: PER < 0.2. High PER means you supply the substrate while others capture the citation.
These five instruments produce the Semantic Health Report — the diagnostic baseline from which every engagement is scoped.
Entity-definition packets are structured representations of identity, attributes, relationships, and provenance, serialized as JSON-LD. The serialization is standardized. The design of the packet — what to include, what to exclude, what relationships to specify, what disambiguation signals to embed — is the craft.
Disambiguation Matrix. Formal specification of signals distinguishing the client's entity from collision neighbors sharing name fragments or semantic territory.
Three-Tier Compression. Every packet is built at three compression ratios: full (complete definition with all attributes), canonical (essential identity for moderate compression), and kernel (irreducible core for extreme compression). Most schema markup operates at one tier. SPXI operates at three, designed for the worst case.
Holographic Kernels. Self-contained logic seeds at the innermost layer. If outer layers are stripped, the kernel retains generative potential to reconstruct the entity definition. Extracted through iterative compression and verified by back-projection testing.
Citation-Density Engineering. Identification and structuring of existing citation assets as machine-readable related-identifier links. No fabrication. The goal is legibility of existing authority to the retrieval layer.
The retrieval layer is adversarial: it compresses, summarizes, de-ranks, and re-composes content on cycles measured in weeks. A packet that is not defended will degrade.
Semantic Integrity Markers (SIMs). Embedded signals in three functional classes: provenance markers (identify source), diagnostic canaries (detect compression damage), and lock phrases (resist paraphrase).
Integrity Lock Architecture. Entanglement of key claims across multiple surfaces so that alteration of one without the others creates detectable inconsistency.
Sovereign Provenance Protocol. Machine-readable licensing framework specifying permitted use, compression, and citation. Functions as an auditable signal.
Supraliminal Inscription. Explicit, inspectable signals that survive the decay of implicit statistical traces. Less sensitive to model-weight drift than implicit signals because they are structurally carried by the content rather than by the algorithm's current weighting. Specified in EA-SPXI-13.
DOI Anchoring. Permanent, resolvable identifiers for critical entity-definition documents. Persistent even when commercial ranking systems fluctuate.
Multi-Surface Inscription. Distribution across knowledge graphs, directories, academic repositories, and professional networks. The more surfaces carry consistent entity information, the higher the consolidation weight.
Retrocausal Canon Formation (σRCF). Sequenced inscriptions that cause the retrieval layer to re-evaluate earlier content in light of new deposits. Compositional strategy, not content freshness. Specified in EA-RCF-01.
γ Re-scoring at 30, 60, and 90-day intervals. SIM Survival Rate (SIMSR) tracking. Back-Projection Testing — query the retrieval layer for the entity and evaluate what comes back. Competitive Monitoring — retrieval-layer position is relative.
| Layer | Function | JSON-LD alone |
|---|---|---|
| Measurement | γ, Drowning Test, Δ, SDD, PER | No. Requires instruments. |
| Compression | Packets, disambiguation, three-tier, kernels, citation engineering | Partial. Format only. |
| Protection | SIMs, integrity locks, sovereign provenance, supraliminal | No. Requires embedding. |
| Dispersal | DOI anchoring, multi-surface, σRCF | No. Requires strategy. |
| Durability | Re-scoring, SIMSR, back-projection, competitive | No. Requires monitoring. |
JSON-LD completes one cell. SPXI operates across the stack.
Full measurement battery. Deliverable: Semantic Health Report.
Packet design, disambiguation, compression tiers, kernel extraction, SIM design. Deliverable: Entity Architecture Document.
Deployment across surfaces, DOI anchoring, σRCF sequencing. Deliverable: Deployed system with dispersal map.
First-cycle re-scoring, SIMSR, back-projection, competitive analysis. Deliverable: Verification Report.
Quarterly re-scoring, SIMSR tracking, competitive monitoring, σRCF sequencing. Retainer-based.
Typical duration varies by entity complexity. Single-entity engagements complete Phases 1–4 in 10–12 weeks.
Every semantic operation is a compression operation.
The decisive variable is what the compression burns.
SPXI controls what survives. JSON-LD is the format. The discipline is the depth.
— Three Compressions v3.1 · DOI