Cartographic infrastructure for mapping reality through data
Stop obsessing over the black box. Our databases are white—we built them, we can inspect them.
The blackness is outside. The unexplored territory. The darkness beyond our maps.
We are cartographers stepping into darkness, progressively mapping reality.
The map emerges from the journey, not before it.
For two decades, we’ve worried about "black box" AI and opaque systems. We’ve been looking in the wrong direction.
The Schrödinger Shift: - Old question: "Is the cat dead or alive?" (What’s inside THIS box?) - New question: "Are there cats beyond?" (What exists in unexplored reality?)
The inversion: - Our boxes are white (D_p—phenomenal databases we built and understand) - Reality is dark (D_n—noumenal territory beyond our measurements) - The question isn’t "How does this model work?" but "What haven’t we measured?" - Focus shifts from interpretability (opening boxes) to cartography (mapping darkness)
Journey Grammar for Databases (JGD) is the formal infrastructure for this paradigm:
-
Catalog the white box (Zone 1: ~15% coverage—D_p instances we’ve created)
-
Mark known blind spots (Zone 2: ~25%—gaps we’ve identified)
-
Acknowledge unknown unknowns (Zone 3: ~60%—unconceived domains)
-
Track how maps emerge from journeys (temporal evolution of knowledge)
-
Guide exploration (where to step into darkness next)
See PHILOSOPHY.adoc for the complete philosophical framework.
See ARXIV-POSITION-PAPER.adoc for the academic paper (ready for arXiv).
JGD is a grammar, not just a vocabulary. It goes beyond VoID/Dublin Core/DCAT to provide compositional infrastructure:
-
Vocabulary: Terms and definitions (what VoID/Dublin Core provide)
-
Grammar: Vocabulary + syntax (EBNF) + semantics (Idris2) + composition rules + transformations (category theory) + cartographic guidance
✓ D_p/D_n classification - Phenomenal databases (observations) vs noumenal reality (territory) ✓ Three-zone model - White box (15%), known darkness (25%), unknown darkness (60%) ✓ Homoiconic containers - Databases describe themselves using same grammar as contents ✓ Metamorphic transformations - Provable representation changes (RDF ↔ JSON-LD ↔ Graph ↔ SQL) ✓ Temporal cartography - Maps emerge from journeys, tracked via verisimdb ✓ Formal semantics - Idris2 dependent types prove metadata correctness ✓ Epistemic humility - Machine-readable blind spots and unknown unknowns ✓ Atlas-based federation - verisimdb coordinates distributed D_p instances ✓ Multi-format - RDF/Turtle, JSON-LD, SQL accessibility layer (JGD-SQL) ✓ Legacy compatibility - Crosswalks to VoID/Dublin Core/DCAT
See CONCEPT.adoc for conceptual overview and CARTOGRAPHY.adoc for practical guide.
@prefix jgd: <https://hyperpolymath.org/ns/jgd#> .
@prefix geo: <http://www.opengis.net/ont/geosparql#> .
:ClimateDB a jgd:D_p ;
jgd:observes jgd-domain:GlobalClimate ;
jgd:title "Global Climate Observations 1980-2025" ;
jgd:creator :NOAA ;
# Spatiotemporal coverage (the white box for this D_p)
jgd:spatialCoverage :NorthernHemisphere ;
jgd:temporalCoverage [
jgd:start "1980-01-01"^^xsd:date ;
jgd:end "2025-01-31"^^xsd:date
] ;
jgd:resolution "1km grid" ;
# Known blind spots (Zone 2)
jgd:knownBlindSpots [
jgd:spatial :SouthernOceans ;
jgd:temporal "pre-1950" ;
jgd:reason "Limited sensor deployment before satellite era"
] ;
# Epistemic humility
jgd:uncertainty jgd:Medium ;
# Versioning (temporal cartography)
jgd:versionHistory :ClimateDB-History ;
jgd:currentVersion :ClimateDB_v2025 .-- Create phenomenal database metadata
CREATE D_P climate_db (
observes = 'global-climate',
spatial_coverage = ST_GeomFromText('POLYGON(...)', 4326),
temporal_coverage = TSTZRANGE('1980-01-01', '2025-01-31'),
resolution = '1km grid'
);
-- Mark blind spots
INSERT INTO jgd.blind_spots (d_p_id, spatial, temporal, reason) VALUES
('climate_db', 'southern-oceans', 'pre-1950', 'Limited sensor deployment');
-- Query: Find D_p covering Europe in 2020-2025
SELECT d_p.name, d_p.spatial_coverage
FROM jgd.d_p
WHERE ST_Intersects(d_p.spatial_coverage, ST_MakeEnvelope(...))
AND d_p.temporal_coverage && TSTZRANGE('2020-01-01', '2025-12-31');See ARCHITECTURE.adoc for complete specification.
Layer 1: Idris2 Formal Specification
- Dependent types for D_p/D_n
- Proofs of metadata correctness
- Category theory for metamorphic transforms
- Generates C ABI headers
Layer 2: Zig FFI Implementation
- Memory-safe C ABI
- Cross-platform compatibility
- Zero runtime dependencies
- SQL parser (JGD-SQL → SPARQL compiler)
Layer 3: Language Bindings
- Rust: Systems tools and CLI
- Julia: Batch processing and data science
- ReScript: Web UI and visualization
- Python: ML/AI ecosystem integration
Layer 4: Applications
- CLI tools (create, validate, query, transform)
- Web UI (visual metadata creation)
- verisimdb integration (atlas coordination)
- SPARQL/VQL endpoint (temporal queries)See BACKENDS.adoc for storage independence details.
verisimdb is the canonical backend providing full temporal cartography:
Unique capabilities:
- Temporal versioning: Git-like semantics for D_p evolution
- Time-travel queries: "What coverage did we have on 2020-01-01?"
- Cartographic deltas: Track blind spot filling, uncertainty reduction
- VQL extensions: SPARQL + temporal operators (AT TIME, DELTA)
- Homoiconic: verisimdb is itself a D_p describing database evolution
Other backends supported (see BACKENDS.adoc): - RDF triplestores (Jena, Virtuoso, Blazegraph) - SQL databases (PostgreSQL, MySQL) - Property graphs (Neo4j, ArangoDB) - Document stores (MongoDB, CouchDB)
JGD core is storage-independent. verisimdb provides optimal temporal features.
Philosophical Framework: - PHILOSOPHY.adoc - White box in darkness paradigm, three-zone model - VISION-BLINDSPOT.adoc - Nature paper analogy, cartographic framing
Technical Specifications: - CONCEPT.adoc - Core concepts and overview - ARCHITECTURE.adoc - 4-layer architecture (Idris2 → Zig → bindings → apps) - BACKENDS.adoc - Storage independence, verisimdb vs alternatives - JGD-SQL.adoc - SQL accessibility layer specification
Practical Guides: - CARTOGRAPHY.adoc - How to create D_p instances, mark blind spots - ROADMAP.adoc - 6-phase development plan through Q2 2027
Academic Publication: - ARXIV-POSITION-PAPER.adoc - Complete arXiv position paper (ready for submission) - ARXIV-PAPER-DISCUSSION.adoc - Timing analysis and structure - ARXIV-RELATED-WORK.adoc - Related work section (cartography, philosophy, databases)
Project Metadata (.machine_readable/6scm/): - .machine_readable/6a2/STATE.a2ml - Current project state, progress, tasks - .machine_readable/6a2/ECOSYSTEM.a2ml - Relationships to other projects - .machine_readable/6a2/META.a2ml - Architectural decisions, philosophy, governance
See ROADMAP.adoc for complete plan.
Phase 1: Formal Specification (Q1 2026) ✓ In Progress - EBNF grammar, Idris2 types, SHACL shapes, JSON-LD context, OWL ontology
Phase 2: Homoiconic Container Model (Q2 2026) - Self-describing storage containers, reflection APIs
Phase 3: Metamorphic Library (Q3 2026) - Transformation catalog, Idris2 proofs, round-trip validation
Phase 4: Integration Ecosystem (Q4 2026) - verisimdb atlas, SPARQL/VQL endpoint, D_p discovery tools
Phase 5: Tooling & Libraries (Q1 2027) - Zig FFI, Rust CLI, ReScript web UI, Julia batch scripts, Python/JS bindings
Phase 6: Documentation & Community (Q2 2027) - Specification site, tutorials, academic paper (ISWC 2027), community forum
Success metrics by 2027: - >1000 D_p instances created - >100,000 databases indexed in verisimdb atlas - White box expansion: 15% → 17% coverage - >10,000 Zone 2 blind spots cataloged
What exists: - VoID (2011): Vocabulary for linked datasets (technical metadata) - Dublin Core (1998): Resource description (15 core elements) - DCAT (2014): Data catalog vocabulary (government data)
What’s missing (what JGD adds): - ❌ No compositional grammar (just vocabularies) - ❌ No temporal evolution tracking - ❌ No cartographic semantics (coverage, blind spots, epistemic humility) - ❌ No homoiconic containers - ❌ No metamorphic transformations - ❌ No atlas-based federation coordination - ❌ No paradigm shift (they assume completeness, not exploration)
Licensed under the Palimpsest Meta-Philosophical License (PMPL-1.0-or-later).
SPDX-License-Identifier: PMPL-1.0-or-later
Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
See TOPOLOGY.md for a visual architecture map and completion dashboard.
