UN-GGCE Portal

Appearance

UN-GGCE ToR2 Decision Log (Shared Memory) ​

This document acts as the authoritative record of strategic decisions, architectural pivots, and stakeholder approvals throughout the ToR2 consulting engagement. All agents (systems-architect, geodetic-researcher, risk-analyst, knowledge-graph-engineer, quality-assurance, and technical-writer) MUST read this file to understand the current context and history of the project.

Instructions for Agents ​

  • READ ONLY: Treat this as a source of truth for current project status and strategic direction.
  • DO NOT MODIFY: Only the lead consultant (Human) or the primary orchestrator should update this file after a decision is finalized.

2026-03-23: Project Setup & Infrastructure ​

  • Decision: All OTel telemetry logs will be stored locally on the server drive to avoid Google Drive sync conflicts and maintain a permanent decision trace.
  • Decision: A Context Graph Data Fabric using RDF-Star and SHACL 1.2 will be the primary architecture for tracking and validating the "Robust Global Geodetic Supply Chain."
  • Decision: A specialized technical-writer agent has been created to ensure all deliverables match the formal UN-GGCE style established in the "State of Geodesy 2026" report.
  • Decision: This decisions.md log will serve as the shared "Drop-box" memory for all agents to ensure consistency across the five work assignments.

2026-03-23: Work Assignment 1 - Principles Selection ​

  • Strategic Pivot (QA Mandate): The quality-assurance agent rejected the initial "best-effort" architectural approach. It is now a mandatory requirement that all "Basic Supply Chain" definitions move beyond scientific cooperation to formalized long-term operational agreements (SLAs/MOUs).
  • Strategic Pivot (QA Mandate): All data centres must strictly adhere to FAIR (Findable, Accessible, Interoperable, Reusable) data principles as defined in JDP Objective 1.2.
  • Strategic Pivot (QA Mandate): Workforce resilience, including mandated succession planning and knowledge transfer, is now a non-negotiable PPTD requirement for any observatory or centre to be included in the "critical path."

2026-03-24: Deliverables Portal & CI/CD Pipeline ​

  • Decision: VitePress is the official static site generator for the UN-GGCE Deliverables Portal, providing a high-performance documentation experience.
  • Decision: The portal is deployed via Docker on Google Cloud Run, utilizing a dedicated build/sync pipeline (sync-docs.sh) for rapid deployment of consulting outputs.
  • Decision: All geodetic workflows and capability models must use Mermaid.js syntax enclosed in standard Markdown (.md) files to ensure proper rendering in the portal.
  • Decision: Raw .mmd files in the outputs/ folder are considered source assets and must be mirrored into Markdown wrappers for publication.

2026-03-24: Work Assignment 1 - Document Traceability & Context ​

  • Decision: Renamed the primary guiding principles document from "Part 1" to "Work Assignment 1: Guiding Principles" to accurately reflect the Terms of Reference (ToR2).
  • Decision: Mandated explicit markdown linking for all foundational UN precedents (IGIF, WMO, Sendai) within the deliverables to ensure direct traceability.
  • Decision: Included a broader research appendix in Work Assignment 1 to formally acknowledge the socio-economic impact studies (EUSPA, London Economics), critical infrastructure risk registers (NIST, UK Risk Register), and enterprise architecture standards (TOGAF, DMBOK, FAIR) that informed the baseline assessment.
  • Decision: Updated the technical-writer and geodetic-researcher agents to actively cite this broader library of references and academic research in future outputs.
  • Decision: Created the "Investment Prioritization and Risk Rubric" (1b_investment_rubric.md) to operationalize the guiding principles. The risk-analyst and roadmap-planner agents have been updated to strictly use this rubric for evaluating gaps and sequencing investment recommendations based on existential risk mitigation.
  • Decision: Conducted a comprehensive scan and download of the primary strategic artifacts from the UN-GGCE website (Strategy and Operating Plan, 1st JDP, Needs Assessment, Hidden Risk Report). These documents have been added to the ./docs directory to serve as the authoritative grounding context for all agent research and drafting activities.

2026-03-25: Geodetic Terminology, Mathematical Prioritization, & Data Provenance ​

  • Decision: Adopted IAG/GGOS Authoritative Terminology: Shifted from IT-centric "nodes" to geodetic "observatories" and "centres" across all documents, reflecting the terminology of the International Association of Geodesy (IAG) and the Global Geodetic Observing System (GGOS). geodesy.science is now the ultimate grounding source for all agents.
  • Decision: Mathematical Translation of the Rubric (Agent Handoff): Documented the explicit mathematical algorithm (Weighted Gap Analysis + Network Criticality Multiplier) in data/mathmatical_rubric.md to translate qualitative PPTD scores into a sequenced JDP roadmap. This establishes a clear workflow: the risk-analyst evaluates the PPTD gap, the knowledge-graph-engineer maps the network topology (EGV bottlenecks, co-location, scarcity), and the roadmap-planner uses SPARQL to compute the Final_Investment_Priority.
  • Decision: W3C PROV-O Implementation for UN Traceability: Upgraded the observatories.json dataset to JSON-LD with strict W3C PROV Ontology tracking (ggsc:provenance). The geodetic-researcher and extraction scripts now log exact source URLs and timestamps for every metadata point (from ITRF, IGS, IVS, ILRS). This ensures the quality-assurance agent can validate all infrastructure data against UN transparency mandates.
  • Decision: Consolidated Core Observatory Network: Refined the raw instrument data into 664 unique physical observatories. Identified 107 "Core Sites" based on GGOS co-location criteria (3+ techniques) and ITRF/IGS official core network designations. Published an interactive component (ObservatoryTable.vue) to the portal for stakeholders to explore this validated infrastructure.
  • Decision: Capability Model Visualization: Replaced the wide, difficult-to-read Mermaid rendering of the Capability Model with a responsive, CSS-grid-based HTML layout directly within Markdown to vastly improve readability on the portal.

2026-03-27: Infrastructure Migration ​

  • Decision: The UN-GGCE Deliverables Portal (Cloud Run instance) has been successfully migrated to the ben@risemarketing.com.au account under the rise-infrastructure-491413 project. The new authoritative deployment URL is https://un-ggce-portal-d2zfxnoqxa-uc.a.run.app. This consolidates the portal hosting under the Rise Marketing infrastructure as requested by the Sysadmin agent.

2026-03-26: EGV-Specific Supply Chain Modeling (Satellite Orbits) ​

  • Decision: EGV-Specific Risk Mapping: Decided to model supply chain risks at the individual Essential Geodetic Variable (EGV) level (starting with GNSS Satellite Orbits) rather than a monolithic view, recognizing that an observatory's criticality varies significantly depending on the specific product pipeline it supports.
  • Decision: 4-Tier Architecture Decomposition: Restructured the Satellite Orbits workflow into four explicit tiers (Tier 0: Acquisition, Tier 1: Archiving, Tier 2: Analysis Centers & Software, Tier 3: Combination & Distribution) to accurately pinpoint institutional and technological bottlenecks.
  • Decision: Software Diversity vs. Maintenance Risk: Formally documented that while the Analysis Center layer is algorithmically secure due to software diversity (Bernese, GipsyX, NAPEOS, PANDA, EPOS.P8), it carries a High "Key Person" maintenance risk due to reliance on niche academic teams.
  • Decision: Identification of Critical Bottlenecks: Identified the Analysis Center Coordinator (ACC) and the SPOCC combination engine workflow is classified as a "High Risk" bottleneck due to process/governance vulnerability, explicitly validated by historical U.S. government shutdowns (2013, 2018-2019).
  • Decision: Stakeholder Accountability Matrix: Shifted workflow documentation from purely institutional to human-centric accountability. Key personnel within the IGS Central Bureau (JPL), Infrastructure Committee (GFZ, EarthScope, CDDIS), and SPOCC development team (GFZ) are now explicitly mapped to their respective supply chain tiers to highlight "Key Person" dependencies.

2026-04-23: Network Operations & Global Network Monitoring Gap ​

  • Strategic Finding: A longitudinal review of the IGS Technical Reports (2013 and 2018) revealed widespread failure across major Analysis Centers, Global Data Centers (SIO, IGN, KASI), and critical Working Groups (e.g., Reference Frame) to submit annual reports.
  • Mapping to Capability Model: This administrative and reporting fragility has been formally mapped as a direct symptom of the low maturity score (2.3) in the "Network Operations" capability.
  • Impact on Supply Chain: The inability to consistently document and report on system health signifies a critical gap in Global Network Monitoring. While data may continue to flow (technical resilience), the lack of systematic monitoring and administrative reporting blinds the supply chain to creeping degradation and capacity constraints, reinforcing the necessity for formalized Service Level Agreements (SLAs).

2026-06-10: Graph Database State — WA2/WA3 Readiness ​

  • Decision: Jena Fuseki (http://localhost:3030, dataset ggsc) is the authoritative datastore. HolonBridge v2.1.0 running on port 3031 provides NL query and SHACL-validated update access.
  • DataBooks in Jena (as of 2026-06-10):
    • capability-maturity-2026-05-12 — 890 triples; 58 capabilities with full PPTD scores (v1.1.0). This is the authoritative source for all maturity scores in WA2 and WA3. Do not read scores from prose.
    • satellite-orbits-v1 — Satellite Orbits workflow v1.0.0 (pushed; blocks landed in default graph pending DataBook format fix — knowledge-graph-engineer to migrate to named graph https://w3id.org/ggsc/satellite-orbits/workflow).
    • itrf-workflow-v0.1, eop-workflow-v0.1, icrf-workflow-v0.1, gravity-workflow-v0.1 — v0.1 stub topology only; must be upgraded before rubric can run (see tor2_workflow.md Step 3b).
    • ggsc-event-graph-holon (20 WA3 country risk DataBooks + event graph + KG holon) — all pushed 2026-06-10.
    • Richard Gross review AssertionEvent v1.2.1 — 10 assertions in graph/events-2026-04-23.
  • Namespace status: 537 triples use legacy ontologist.io namespace. Official migration bridge at http://w3id.org/holon/spec/migration-bridge#graph. Full GGSC reconciliation pending Kurt Cagle's supplement to HGA Pass A §1. This does not block WA2/WA3 content work.
  • Blocking item for WA2: Workflow DataBook stubs (ITRF/EOP/ICRF/Gravity) must be upgraded to use canonical ggsc-cap: IRIs and disaggregated PPTD scores before the mathematical rubric can run. See docs/wa2-wa3-submission-brief.md.

2026-06-10: HGA Namespace Migration Decision ​

  • Decision: HGA vocabulary has migrated from https://ontologist.io/ns/holon# to http://w3id.org/holon/ (W3C Holon CG, acting chair Kurt Cagle). Spec published at https://github.com/w3c-cg/holon. 15 normative DataBooks across Passes 0–F. CC0-1.0 ontology license.
  • Decision: The ;|} RDF-star annotation block terminator convention has been applied across all 21 WA3 DataBooks (committed 2026-06-09).
  • Deferred: Full DataBook rewrite to new namespace deferred until Kurt Cagle completes the GGSC namespace reconciliation supplement (HGA Pass A §1). GGSC-specific terms (holon:EventEnvelope, holon:EventGraphHolon, holon:AssertionRecord etc.) have no bridge mappings yet.

2026-04-28: QA Review - Work Assignment 2 (EGV Workflow Modeling) ​

  • Finding (Positive) - EGV Progression & Tone: The narrative workflow reports (outputs/itrf_workflow_description.md, outputs/eop_workflow_description.md, outputs/icrf_workflow_description.md, outputs/global_gravity_model_workflow_description.md) excellently model the progression from Level 0 to Level 3 EGVs. Institutional mappings and bottleneck analyses correctly identify Single Points of Failure (e.g., VLBI correlation, ITRS Product Center at IGN) and align perfectly with ToR2 expectations. The professional tone meets UN-GGCE standards.
  • Finding (Positive) - DataBook Adherence: The draft DataBooks (data/shacl/*-workflow.databook.md) correctly implement v0.9 standards, utilizing PROV-O (prov:Activity, prov:wasAssociatedWith) to model the multi-tier supply chains as directed.
  • Remediation Completed - Capability Linking: The DataBooks now explicitly bind to the 59 Phase 1 capability instances in data/ontology/ggsc-capabilities.ttl (e.g., ggsc-cap:VLBIDataAcquisition, ggsc-cap:GeodeticDataProcessing).
  • Remediation Completed - Vendor Neutrality: The process metadata block in the DataBooks has been generalized to use "UN-GGCE LLM Pipeline" and "urn:ggsc:pipeline:llm-v1", complying with the strict vendor-agnostic architecture principle.
  • Remediation Completed - Evidence Traceability: The DataBooks have been updated to include prov:wasDerivedFrom links pointing to authoritative IAG/GGOS primary sources and local narrative descriptions, satisfying the Zero Hallucination and Evidence Mandate.
  • DataBook Approval Gate Status: APPROVED. All remediation items have been successfully addressed. Final sign-off is granted for graph database load.

2026-07-05: Stakeholder Approval — Tier Structure and Investment Risk Framework ​

  • Approval: Nicholas Brown (Head of UN-GGCE) has reviewed and largely approved the two framework documents submitted in late June 2026:
    • outputs/appendix_egv_levels_and_tiers.md (v1.1) — EGV Data Levels and Supply Chain Tiers, previously reviewed by R. Gross (IAG President).
    • outputs/1b_investment_rubric.md (v3.0, June 2026) — Investment Prioritization and Risk Rubric, using the two-track operational/governance framework (Tier Risk Score = PPTD Gap × Step Criticality for the operational track; Principles 2, 3, 4, and 6 assessed qualitatively in the governance track).
  • Consequence: The two-track framework is now the approved scoring basis for all ToR2 deliverables. The authoritative tier risk scores are in Jena named graph https://w3id.org/ggsc/graph/tier-risk-scores-2026-06-22 (206 triples; straight unweighted 4-dimension PPTD average per commit 10fcb65, capability DataBook v1.3.0 scores).
  • Decision: The superseded named graph https://w3id.org/ggsc/graph/tier-risk-scores-2026-06-10 (181 triples; old two-factor scoring run) is dropped from Fuseki so agents cannot pull stale scores. The scores remain recoverable from git history and the 2026-06-10 decision log entry if audit reconstruction is required.
  • Standing requirement (Nick Brown, June 2026): All deliverables must include a plain-language lead-in and explicit Terms of Reference linkage so UN Headquarters reviewers (payment approval chain) can follow the documents without prior familiarity with the engagement. This applies to every WA deliverable from this point forward.
  • Follow-up items:
    • data/mathmatical_rubric.md still documents the retired v3.0 three-factor model (Recovery Factor). Requires a v4.0 update to match the approved two-track framework — documentation reconciliation only; the Jena scores already reflect the approved model.
    • Confirm with Nick which residual reservations sit behind "largely approved" before WA3 rewrite is finalized.
    • Confirm whether formal approval by Richard Gross (Chair of IAG) is still required for the WA1 gateway (per tor2_workflow.md WA1 Step 5) or whether Nick's approval supersedes it.

2026-07-06: Capability Maturity Graph — Traceability Gap Closed ​

  • Finding: The named graph https://w3id.org/ggsc/graph/capability-maturity-2026-05-12 — cited by WA2 §2.1 and the standing traceability rule as the authoritative machine-readable score record — contained zero scored capabilities. The 58 scored capabilities existed only in the local file data/shacl/ggsc-capability-maturity.databook.md (v1.3.0). The 2026-06-10 decision entry claiming "890 triples; 58 capabilities with full PPTD scores" in Jena was inaccurate as of this check.
  • Root cause: The six cap-* turtle blocks in the DataBook had a blank line between the <!-- databook:id: ... --> annotation and the code fence, so databook push treated them as anonymous blocks and PUT them sequentially to the default graph, each overwriting the last. Same defect class as the satellite-orbits routing issue logged 2026-06-10.
  • Fix: Removed the six blank lines (whitespace-only change to the DataBook); re-pushed all blocks to their urn:ggsc:holon:capability-maturity:2026-05-12#<block> graphs; additionally merged the seven capability blocks into the canonical https://w3id.org/ggsc/graph/capability-maturity-2026-05-12 graph via databook push -b <id> -g <graph> --merge. Verified: 58 subjects with ggsc:averageMaturityScore; spot-check gravity-data-acquisition-and-storage = 3.3 per v1.3.0 correction.
  • Caveat (resolved 2026-07-06): During the initial (mis-routed) push, the default graph was overwritten before its prior contents were inventoried. Re-pushed data/satellite-orbits.databook.md as a precaution and verified: block graph triple counts (workflow 156, risks 50, capabilities 74) exactly match the pre-existing canonical https://w3id.org/ggsc/satellite-orbits/* graphs — the earlier migration had already copied everything out of the default graph, so no data was lost. Default graph dropped (now 0 triples).
  • Also completed 2026-07-06: data/mathmatical_rubric.md updated to v4.0 — Part A now documents the two-track model actually used in the 2026-06-22 scoring run (two-factor operational score, qualitative governance track, NA=0/÷4 tier-gap convention); retired three-factor/Recovery Factor content removed. Closes the math-doc reconciliation follow-up from the 2026-07-05 entry.
  • Follow-up (resolved 2026-07-06): databook validate (CLI v1.4.2) crashed with require is not defined. Root cause: three CommonJS require() calls inside executeLocalValidation in the ESM module commands/validate.js, plus a fallback-logic bug that invoked local validation with a null engine when neither Jena nor pyshacl was installed. Same defect confirmed present in upstream w3c-cg/holon main (architectures/databook/commands/validate.js) — patch to be sent to Kurt Cagle / filed upstream. Local install patched (requires → top-level imports; clear die() when no engine found); pyshacl 0.31.0 installed via uv tool install as the validation engine.
  • SHACL validation status (2026-07-06): all six DataBooks validate CONFORMS against ggsc-shacl-all.databook.md#ggsc-shacl-all-block: capability-maturity, itrf-workflow, eop-workflow, icrf-workflow, gravity-workflow, satellite-orbits. Brief Step 7 validation checklist items are satisfied.

2026-07-06: QA Sign-Off — WA2/WA3 Submission Package ​

  • QA validation pass (quality-assurance agent): full 10-item checklist run against the live Fuseki graphs and deliverable files. Nine of ten items passed on first pass: all 50 tier-score citations across WA2 and WA3 match tier-risk-scores-2026-06-22 exactly; classification bands and labels consistent; capability claims (9×1.00 Category A list, Network Operations dims, Appendix A bottom-12, domain averages) match the capability-maturity graph; principle citations meet the ≥3-of-5 threshold for all six principles; provenance complete (all 25 tier scores carry prov:wasGeneratedBy → typed prov:Activity with model version and score source); HQ lead-ins present in all seven documents; no stale-model artifacts outside version-history tables.
  • Defect found and fixed: WA3 §3.2–3.5 cited four named-graph IRIs (https://w3id.org/ggsc/graph/{itrf,eop,icrf,gravity}-workflow, taken from the 2026-06-04 submission brief) that contain zero triples in Fuseki. The workflow DataBooks declare — and the data resolves at — https://w3id.org/ggsc/databooks/{x}-workflow-v0.1#{x}-workflow-topology. Citations corrected to the declared IRIs; all four verified resolving (251/228/194/191 triples). The brief's named-graph table is aspirational and should not be used for citations until the Step 3b DataBook upgrade adopts canonical IRIs.
  • Verdict: SIGN-OFF APPROVED (QA verdict was "one four-line citation correction away from approval; no re-verification needed beyond the resolution check" — correction applied and resolution re-verified 2026-07-06).
  • Package status: WA2 v2.0 and WA3 v2.0.0 are submission-ready pending the Human Review gateway (Nick Brown).

2026-08-21: Stakeholder Working Call — "Robust" Redefined, Three-Horizon Framing Established ​

  • Strategic Pivot (Nicholas Brown, 21 August 2026 working call): "Robust" was being used loosely across earlier drafts as if it were a status the current supply chain could claim. Redefined consistently with UN-GGCE's own established usage: robust = the top of the maturity scale (State of Geodesy 2026 Level 5, "Strategic") and the fully-resourced target architecture in the Zero Draft of a Robust Global Geodesy Supply Chain (15 April 2026, ~US$165M/year target vs ~US$70–106M current funding — a ~US$100M shortfall). This assessment does not claim the supply chain is robust today; claiming otherwise would not survive scrutiny from IAG or Member States.
  • Decision: Adopted a staged path to robustness — originally labelled "Phase 1/2/3," later renamed to Horizon 1/2/3 (see 27 August entry below) — Minimum Capability (near-term, achievable) → Reliability → Robust (the destination). WA1 and WA1b were rewritten around this framing.
  • Follow-up items from this call, resolved in the days after: readability pass requested on WA1/WA1b (plain-language Key Terms, GNSS/WMO spelled out, shorter sentences); the operational/governance "two-track" framework flagged for retirement in favour of a unified six-principle model; explicit ToR1 → State of Geodesy 2026 → ToR2 provenance requested so readers understand this engagement builds on a published baseline rather than a fresh assessment.

2026-08-27: WA1/WA1b Restructuring — Horizon Rename, Two-Track Retirement, Readability Rework ​

  • Decision: Renamed the staged robustness path from "Phase 1/2/3" to "Horizon 1/2/3" after identifying a three-way naming collision: ToR1's own "Phase 1," the robustness path, and JDP Phase Sequencing's investment bands were all independently using "Phase 1/2/3" for different concepts. Applied consistently across WA1 and WA1b; explicitly connected the horizons to JDP Phase Sequencing ("Horizon 1 is what 'immediate action' means there").
  • Decision: Retired the operational/governance "two-track" model entirely. Principles 2, 3, 4, and 6 — previously assessed qualitatively only, with the note "governance findings are not scored numerically" — now each carry an explicit, quantitative Minimum Capability Target, on equal footing with Principles 1 and 5.
  • Decision: WA1b restructured around "The Six Minimum Capability Targets" — a single table giving one measure and one fixed threshold per principle — with PPTD explained at first use (including a direct answer to "why is it a 5?"), and the worked example (Satellite Orbits Tier 3) integrated progressively rather than held to a late, separate section.
  • Decision: WA1's Key Terms table changed from bare acronym expansions to genuine plain-language explanations (GNSS added, was missing); sentence length reduced throughout; tone benchmarked against State of Geodesy 2026.
  • Decision: JDP Phase Sequencing made genuinely Horizon-driven: Horizon 1 = failing any one of the six Minimum Capability Targets, not a Tier Risk Score percentile cut. A tier can now land in Horizon 1 purely on a governance failure (Principle 3) even if its Tier Risk Score alone looks moderate.

2026-08-27: Near-Miss — Capability-Maturity DataBook Averaging Convention Investigated, Found Deliberate ​

  • Finding investigated: An apparent inconsistency was found in how Fuseki's capability-maturity-2026-05-12 graph stores averageMaturityScore for capabilities with N/A dimensions (mean-of-applicable-dimensions-only, not divided by 4).
  • Action taken, then reverted: Began editing 9 capability blocks in data/shacl/ggsc-capability-maturity.databook.md to "correct" this to a strict NA=0/÷4 average, believing it to be a bug.
  • Caught before push: Running databook head and reading the DataBook's own frontmatter showed the convention was deliberate and documented — it exists specifically to match the literal published State of Geodesy 2026 report's own averages. All 9 edits reverted via git checkout before anything reached Fuseki. No source data was altered.
  • Resolution: The actual need — internally consistent domain averages for WA2's Principle 3/4/6 findings — was met by computing them independently downstream, in Python, directly from raw PPTD dimension scores, applying one convention (NA=0/÷4) uniformly. This resolved a real, separate inconsistency (2.72 vs 2.39 for Satellite Orbits' Technical Interoperability domain, found using two different capability-selection methods) without touching any data source.
  • Standing distinction to preserve: the capability-maturity DataBook's mean-of-applicable convention is correct for matching the published SoG2026 report; WA1b's NA=0/÷4 convention is correct for this rubric's own floor/domain checks. Both are legitimate; conflating them is the error to avoid.

2026-08-28: Stakeholder Working Call #2 — Technical Product Working Group Consultation Model Agreed ​

  • Decision (Nicholas Brown & Allison Craddock, 28 August 2026): Endorsed a two-stage, bottom-up consultation model for validating findings and setting Horizon 2/3 targets, framed by Nick as "help us help you" rather than top-down scoring — "if we gave you $1M tomorrow, where must it go?"
  • Decision: Process named consistently as Technical Product Working Groups — one per product (Satellite Orbits, ITRF, EOP, ICRF, Gravity), drawn from IGS, IVS, ILRS, IDS, and IERS. Two stages: Stage 1 (Self-Evaluation/Scoring Workshop) — the working group confirms or challenges which capabilities apply and rates current maturity, and separately defines the PPTD score required to reach Horizon 1/2/3 for each; Stage 2 (Workflow & Bottleneck Review Workshop) — scores mapped onto the workflow diagram to identify gaps and translate them into investment recommendations. Cross-product oversight sits with a Steering Committee and Reference Group, both to be established over the following ~2 months.
  • Decision: This consultation process is explicitly flagged as an assumption this engagement's documents proceed on, not yet a contracted engagement — it is expected to become its own Terms of Reference, separate from ToR2, and forms the basis of the next phase of this engagement.
  • Consequence: Formally documented in WA1b as a new subsection ("How Horizon 2 and Horizon 3 Get Set: Technical Product Working Groups"), replacing a previously vague reference to "consultation with IAG and the technique service coordinators."
  • Consequence: Because current-state findings will be tested and revised by the working groups rather than standing as final, WA1b and WA2 both gained a standing "Status of these findings" caveat — every PASS/FAIL verdict is now explicitly framed as a provisional baseline pending working-group confirmation, not a locked conclusion. This also gave licence to assert product-specific maturity departures more confidently in the interim (see Capability Evidence Trail entry, below) — the working groups are the check on those assertions, not a reason to withhold them.

2026-08-29: Principles 3, 4, and 6 — Domain-Average Target Replaced with Per-Capability Floor Check ​

  • Decision: Changed Principles 3 (Centralised Accountability), 4 (Technical Interoperability), and 6 (Transparency and Accountability) from a domain-average target (≥2.5) to a domain-scoped floor check — no required capability in the relevant domain may score below Level 2, in any applicable dimension — the same mechanism Principle 5 already used for tier-linked capabilities.
  • Rationale: For a domain like Governance & Strategy, currently averaging 1.52–1.66 across all five products, lifting the average to 2.5 requires near-universal Level 3+ scores across fourteen capabilities — not a realistic near-term ask. Getting every currently-Level-1 capability to "something exists" (Level 2) is a bounded, achievable Horizon 1 target. An average can also let a domain pass on the strength of a few strong capabilities while one is barely established — exactly the blind spot Principle 5 was designed to catch, now closed for 3/4/6 too.
  • Material consequence: Recomputing against the floor check (verified directly against the capability-maturity graph) flipped Principle 4 from passing to failing for all five products. The previous domain average (2.60–2.70) was masking that Data Quality Management and Geodetic Software and Tools both carry a Technology dimension of 1.0 and are required in every product's Tier 1 and Tier 2. Principles 3 and 6 were already failing everywhere under the old method and remain failing under the new one.
  • Follow-on decision (same day): Rather than discard the domain average, reinstated it as a "Monitoring" figure reported alongside the floor-check verdict in both WA1b and WA2 — the floor check answers "has this been fixed," the average answers "is this getting better," and a floor check alone can't show incremental progress since it stays flatly FAIL until the very last capability crosses the line. Framed explicitly as measuring (the gate) vs. monitoring (the trend line), the same relationship Principle 1's binary ≤7 cutoff already has with its own continuous Tier Risk Score.

2026-08-29: Step Criticality Override — Satellite Orbits/ITRF Tier 1 GDC Archiving Failover ​

  • Decision: Raised Step Criticality for Satellite Orbits Tier 1 and ITRF Tier 1 from 3 to 4 (Tier Risk Scores: 8.10→10.80 and 9.00→12.00 respectively). Both tiers rely on the same CDDIS/IGN/SIO Global Data Centre equalization protocol as their cross-site archiving redundancy.
  • Evidence, independently verified against primary sources: the protocol is real and has fired at least once (IGSMAIL-5600, 2007 CDDIS hardware fault — directly confirmed by fetching the message, which does redirect users to IGN and SIO mirrors by name). However, no failover drill, disaster-recovery exercise, or documented instance of an Analysis Centre actually switching from CDDIS to IGN/SIO was found across the two confirmed CDDIS-affecting shutdowns in scope (2013, 2018–2019). Step Criticality 4 ("dominant path, alternatives exist but aren't operationally ready") is more defensible than 3 ("meaningful redundancy") absent evidence the redundancy has been tested; does not rise to 5 since the protocol itself is real.
  • Precision correction (same review): the original framing cited BKG as "another GDC" with no disaster-recovery capability, using this as supporting evidence. Corrected after verification: BKG is a Regional Data Centre, not one of the three GDCs (CDDIS/IGN/SIO) this Step Criticality rating actually scores. The BKG quote itself was independently verified (IGS 2018 Technical Report, p. 159, fetched directly) and is accurate — but it's the closest available parallel, not direct evidence about the GDCs themselves. Also softened "three known CDDIS-affecting disruptions" to two confirmed shutdowns (2013, 2018–2019) plus the ongoing 2025 NASA budget/workforce uncertainty, explicitly flagged as a separate, still-unresolved risk rather than a third confirmed event on equal evidentiary footing.
  • Mechanism: pushed to Fuseki via a scoped SPARQL UPDATE (delete/insert on exactly the two changed triples per tier), not a full DataBook block PUT — a full-block push would have silently dropped rdf:type, rdfs:label, dcterms:created, and prov:wasGeneratedBy triples present on the live graph but not captured in the DataBook's own turtle block. Verified triple/subject counts unchanged before/after (206/26).

2026-08-29: Data-Integrity Audit — Systemic Staleness Found Across Four of Five WA2 Annexes ​

  • Finding: While extending the Satellite Orbits evidence-trail method to ITRF, discovered that ITRF's own displayed capability tables didn't match the values its own Tier Risk Scores were actually computed from. Extended the check to EOP, ICRF, and Gravity and found the same pattern in all four — roughly 30 stale or erroneous capability rows in total, none in Satellite Orbits.
  • Root cause, confirmed via git history: the tier-risk-scores DataBook underwent a documented "DataBook v1.3.0 (2026-06-15 re-verification)" that corrected 25 capabilities in the source-of-truth graph (e.g. "vlbi-data-processing-and-analysis (was 3.25, now 2.3)", "groundbased-asset-management (was 1.75, now 2.5)"). ITRF and EOP's capability tables were last substantively authored 11 June 2026 — four days before that correction — and every commit since (six of them, through 28 August: Step Criticality changes, the two-track retirement, a provenance-labelling pass, a readability pass, the Horizon rename) touched framing and citations but never re-pulled the underlying capability values from the graph. Satellite Orbits was the sole exception: it started from the same 11 June baseline but got a dedicated 14 August commit that specifically re-verified its tables against the live graph — the same process now being called the Capability Evidence Trail.
  • Structural gap identified, not yet closed: the graph is a live, provenance-tracked source of truth that gets corrected in place (with prov:wasGeneratedBy and ggsc:scoreSource triples); the annexes are static Markdown with a citation string ("Source: State of Geodesy 2026") but no verifiable, machine-checkable link back to it. A correction to the graph currently has no way to propagate to a document that quoted numbers from it earlier. Flagged as a platform process gap worth closing (via real DataBook status for the annexes, or a periodic diff-against-Fuseki check) — not yet actioned.
  • Consequence for findings: no Tier Risk Score, Horizon classification, or PASS/FAIL verdict changed anywhere. Every affected score had already been computed from the correct, current graph values; only the visible tables were out of sync with their own math. Corrected all ~30 rows against Fuseki, with a dated, explained note on every amended table. One genuine side-effect: Ground-Based Asset Management's correction (1.75→2.50, all dimensions ≥2) removed it as valid evidence for Principle 5 floor-check failures in ITRF/EOP/ICRF/Gravity — each product's FAIL verdict held anyway via other, still-valid evidence (Knowledge Management or Training and Education).

2026-08-29: Capability Evidence Trail Extended to All Five Products ​

  • Decision: Extended the confirmed/asserted/considered/queued classification method (piloted on Satellite Orbits, 14 August: 18 Confirmed, 3 Asserted, 1 Considered, 6 Queued) to ITRF, EOP, ICRF, and Gravity, added as a new "Capability Evidence Trail" section in each annex.
  • Headline result: across the four newly-classified annexes (76 capabilities total), every single one is Confirmed at baseline — zero Asserted departures exist outside Satellite Orbits. A consistent Queued pattern recurs across all four: Mandate, Funding and Investment, and Legal Framework are evidenced in each annex's own prose or Governance table but were never carried into a scored tier-table row. ICRF's Tier 3 prose is the strongest single case — it states outright that no formally designated body holds accountability for ICRF combination.
  • Interpretation, explicitly recorded: this pattern is read as under-research, not a genuine absence of evidence. Satellite Orbits' evidence trail is more complete because it received a dedicated ~2-day working session; the other four products are structurally similar (small expert teams, proprietary combination software, thin succession documentation) and would plausibly yield further Asserted departures with the same attention. This distinction is intentionally not yet written into WA1b/WA2's own narrative (to avoid the risk-finding numbers looking like a genuine per-product maturity difference) — flagged for a future pass if useful before the Technical Product Working Groups convene.

2026-08-29: Milestone 1 Delivery Plan — Refreshed to Match Current State ​

  • Finding: commercial/milestone_1_delivery_and_closeout_plan.md had drifted significantly behind the deliverables it describes — citing WA1/WA1b/WA2/annex versions three to five revisions behind current source, still targeting a 28 August submission date that had passed, and its draft transmittal email described content from before the Horizon rename, the floor-check rework, the data-integrity audit, and the Capability Evidence Trail additions.
  • Decision: Full refresh — version table brought current, §0 known-issues status corrected (Satellite Orbits Tier 2 fix confirmed resolved; the Word-diagram fallback confirmed still open), action-plan dates and flowchart relabelled to reflect actual status (NOT STARTED / date TBD) rather than left showing passed dates as if still live, and the draft transmittal email rewritten to reference the Technical Product Working Group model and the new evidence-trail sections — both things Nick would specifically want reflected back, given the working-group model was his own proposal.
  • Standing gap, unchanged by this refresh: no .docx has been rebuilt since 23 July for any document. Markdown source is complete and internally verified; the Word/PDF packaging layer has not been started. This is the only remaining blocker to submission.

2026-09-02: Floor-Check Threshold Language — "Every," Not "Any," Applicable Dimension ​

  • Finding: every Minimum Capability Target for Principles 3, 4, 5, and 6 defines its floor check as "at or above Level 2, in any applicable dimension." Read literally, "any" is an existential quantifier ("clears the bar in at least one dimension") — the opposite of how the floor check actually works, and actually applied throughout every finding in WA1b, WA2, and the five annexes: a single dimension below Level 2 (e.g. a Technology score of 1.0) fails the whole capability regardless of its other three dimensions. That's a universal quantifier — "every" applicable dimension must clear the bar — and WA1b's own detailed "Measure" sections for these principles already said "every," so the ambiguous "any" in the Six Minimum Capability Targets summary table was a genuine internal inconsistency, not just imprecise phrasing.
  • Decision: Swept "in any applicable dimension" → "in every applicable dimension" across 1b_investment_rubric.md and 2_product_mapping_report.md (7 instances each — the summary table's four principle rows, the Six Minimum Capability Targets intro paragraph, and the domain-scoped-floor-check explanatory paragraphs). Also corrected the adjacent "at any tier" → "at every tier" in Principle 5's own target row in both documents, the same quantifier mismatch in the same cell. The five WA2 annexes were checked and were already consistent (they use "every applicable dimension" wherever they cite these targets) — no annex changes needed.
  • No scores or verdicts changed — this corrects how the existing, already-applied floor-check logic is described, not the logic or any computed result itself.
  • Recorded as writing_style_guide.md rule 12, so this doesn't get reintroduced in future drafting: pass/fail threshold language must use "every," not "any," for a universal (must-clear-every-dimension) requirement.

2026-09-02: Principle 1's Minimum Capability Target Raised from ≤7 to ≤9 ​

  • Finding: across all 25 assessed tiers, the average PPTD Gap is approximately 2.7. At that typical Gap, a Tier Risk Score ≤7 requires Step Criticality ≤2 ("multiple parallel, fully operational institutions") to pass — a tier with genuine but imperfect redundancy (Step Criticality 3, "important but substitutable") already fails, almost regardless of maturity. Given global geodesy's actual institutional distribution (small numbers of specialised centres is the norm, not the exception), most tiers structurally sit at Step Criticality 3–5. Only 7 of 25 tiers (28%) cleared the ≤7 threshold. No calibration rationale for the original 7 was found in this document set — it sits at, but doesn't cleanly align with, the Score classification table's Significant/Minor boundary.
  • Decision: raised the threshold to ≤9 (Step Criticality 3 at the assessment's own average Gap ≈ 9), specifically for Horizon 1. Rationale given by the user: Horizon 1 exists to identify the critical, immediate gaps — a tier with meaningful (if imperfect) redundancy shouldn't automatically fail the near-term floor. Later horizons are expected to tighten the threshold as the supply chain matures, via the Technical Product Working Group workshops.
  • Quantified impact: pass rate rises from 7/25 (28%) to 13/25 (52%). Six tiers flip FAIL→PASS: Satellite Orbits Tier 0 and Tier 2, ITRF Tier 2, EOP Tier 0, ICRF Tier 3, and Gravity Tier 0 — all Tier 0/2 (acquisition/analysis) findings, not the marquee Tier 3 "critical bottleneck" tiers (Satellite Orbits 18.25, ICRF 15.00, ITRF 13.75 all remain well above either threshold).
  • No product-level verdict changes: Principle 1's product-level PASS/FAIL is driven by each product's worst tier, and every product's worst tier remains above 9 (Satellite Orbits 18.25, ITRF 13.75, EOP 12.75, ICRF 15.00, Gravity 12.00). All five products remain FAIL on Principle 1, and — since every product also fails Principles 3, 4, 5, and 6 regardless — no product's Horizon 1 membership changes either.
  • Updated: 1b_investment_rubric.md (target definition, worked example, and a new explanatory note distinguishing the Score classification table's investment-priority ranking from the Minimum Capability Target's pass/fail floor) and 2_product_mapping_report.md (Six Minimum Capability Targets table and all five products' Minimum Capability Target Summary tables). The five WA2 annexes' own Principle 1 citations are qualitative (geographic distribution narrative), not tied to the numeric threshold, and needed no change.

2026-09-02: Principle 1's Minimum Capability Target Raised Again, ≤9 to ≤10 ​

  • Finding: modelling what Horizon 1's own recommended remediation (lift a tier's capabilities to the Level 2 floor; formalise an existing informal backup relationship — an unmandated validator — into a real operational mandate) actually does to each product's worst tier showed the ≤9 threshold left at least one product's outcome effectively decided by a scoring-convention choice rather than a real difference in redundancy. Applied in full to EOP's worst tier (Tier 3, EOP Combination): capability floor uplift (Risk Management to Level 2; Performance Management confirmed already at floor on its applicable dimensions, no uplift needed) plus formalising the existing USNO/SYRTE best-effort relationship into a real SLA (Step Criticality 4→3) computes to a Tier Risk Score of 9.19 under this rubric's own NA=0÷4 averaging convention — a hair over ≤9 — versus 7.69 under the alternative (NA-excluded) convention this document already flags as a live but unofficial option. Applying the identical remediation to Satellite Orbits' worst tier (18.25) and Gravity's worst tier (12.00) showed both fall well short of either threshold regardless of convention, since both lack an existing relationship to formalise (Satellite Orbits: no independent SPOCC implementation exists anywhere to elevate to a backup; Gravity: no fourth institution capable of ultra-high-degree modelling exists to elevate) — their gap is a genuine capacity investment, not a governance one, so raising the threshold doesn't paper over a real gap for either.
  • Decision: raised the threshold to ≤10. Rationale given by the user: the change doesn't materially alter the underlying logic (the Level 2 floor and the 2-jurisdiction/mandated-backup requirement are unchanged) but gives a fully remediated tier a genuine margin instead of a coin-flip outcome, providing a clearer path to Horizon 1 closure.
  • Quantified impact: pass rate across all 25 tiers rises from 13/25 (52%) to 14/25 (56%). Exactly one additional tier flips FAIL→PASS: ICRF Tier 2 (9.94). No other tier among the 25 sits between 9 and 10.
  • No product-level verdict changes: ICRF Tier 2 is not ICRF's worst tier (Tier 1, 15.00, remains FAIL and remains the product's binding constraint), so this change alone does not flip any product's Principle 1 verdict. Combined with the planned remediation package, it newly clears EOP's Principle 1 verdict (Tier 3: 12.75 → 9.19 post-remediation, under ≤10 but not under ≤9) — this is the practical effect the change was made for. Satellite Orbits, ICRF, and Gravity remain FAIL on Principle 1 under either threshold; all five products continue to fail Principles 3, 4, 5, and 6 independently of Principle 1.
  • Updated: 1b_investment_rubric.md (target definition, Principle 1 Measure rationale, worked example, and the Score-classification paragraph) and 2_product_mapping_report.md (same six locations as the ≤7→≤9 change). The five WA2 annexes needed no change, as before.

2026-09-02: Work Assignment 3 Full Refresh (v2.5.0 → v3.0.0) ​

  • Finding: WA3 is dated 22 July 2026 and had not been touched since — it predates the two-track scoring retirement (2026-08-27) by five weeks, and predates every WA1b/WA2/annex version bump made in this engagement's finalisation sweep. On review it still described Principles 2, 3, 4, and 6 as a separate qualitative "governance track" alongside a numeric "operational track" for Principle 1 — the exact model retired in favour of uniform quantitative scoring of all six principles — cited 1b_investment_rubric.md v3.1 (current: v6.6) and the WA2 annexes at superseded version numbers, and built §4.1's systemic capability list from the flat State of Geodesy 2026 baseline scan rather than the tier-specific data in each product's own WA2 Minimum Capability Target Summary.
  • Decision: full refresh rather than a targeted patch, per explicit user direction, so WA3 stops being visibly out of sync with WA1b/WA2. Retired all "two-track"/"operational track"/"governance track" language (§1.2, §3.1, §3.3, §5.3, §7, Appendix B); synced version references throughout; replaced §4.1's Category A with the 13-capability tier-specific analysis from this session (8 universal + 5 product-specific, with the Performance Management correction noted); added a fix-class column to §4.2's SPOF table (governance-fixable: ITRF; capacity-required: Satellite Orbits, ICRF); added new §5.1 "Horizon 1 Closure Path" stating the 89% (24/27) closure math; resequenced §5.2's Priority Ranking Table and §5.4's intervention timeline so the governance bundle leads as the cheap, fast win ahead of the three genuine capacity investments.
  • No underlying scores changed — this is a consistency and analysis-depth update, not a rescoring. All Tier Risk Score, PPTD Gap, and Step Criticality values in §3's product tables are unchanged from v2.5.0.
  • Updated: 3_risk_and_gap_analysis.md only (v2.5.0 → v3.0.0). No WA3 changelog file existed prior to this revision; one was not created, since the document's existing "Major revisions" + "Supersedes" header convention (matching WA1/WA1b/WA2 pre-changelog-extraction style) was judged sufficient at this document's current size — revisit if WA3 accumulates further major revisions.

2026-09-11: GGSC Sync Run Sheet Superseded — Stale Local Checkout, Not a Real Graph Gap ​

  • Finding: A run sheet (data/project/ggsc-sync-runsheet.md) was written to push corrected Satellite Orbits/ITRF Tier Risk Score values (raised Step Criticality 3→4) from the deliverable annexes into the live ggsc graph on context-graph-dev, based on data/shacl/tier-risk-scores.databook.md appearing to still carry the pre-fix values against a git log of the local checkout. Phase 0 discovery (querying live Fuseki directly) found stepCriticality=4 (10.80/12.00) already live — the values matched the deliverables, not the local file.
  • Root cause: the local checkout was 5 commits behind origin/main (db093df..29e5959, 28 Aug – 2 Sep 2026), a gap git log on the stale checkout couldn't surface since it only reflects local history. That range already contained this exact Step Criticality fix (tier-risk-scores.databook.md §8, with full sourcing) plus the Horizon 1/2/3 rework, the two-track retirement, and a WA3 full refresh.
  • Fix: none needed against the graph — Phases 1-4 of the run sheet (backup, edit, push, verify) were moot. A DataBook edit made before checking origin/main was discarded (git restore) in favour of git fetch + git merge origin/main --ff-only, which brought in the already-correct, better-sourced version (it independently resolved an additional unexplained gap — gnss-proc re-verified 3.75→3.50 the same day — that the local edit hadn't touched).
  • Caveat, worth generalising beyond this repo: "query live graph, diff against local file" correctly caught that live was ahead of the file, but is only as good as the checkout — git fetch before any drift investigation, not just before pushing. A stale checkout makes "live is ahead of file" and "file is ahead of git log" look identical from the same evidence.
  • Separate, unaffected: the bridge-software gap (Node v2.9.0 vs. Python) and context-graph-dev's public-exposure/firewall gap, found during the same investigation, are real and tracked independently in ~/dev/consulting/platform/SOLUTION_ARCHITECTURE.md §8.1/§10 and ~/sysadmin/context-graph-dev-firewall-runsheet.md.

2026-09-22: Semantic Capability Matching — Methodology Note for the Future WA3 Appendix (Not Yet Actioned) ​

  • Decision, from the user, for future application: where a function has no obvious capability alignment by name, the gap-analysis methodology should fall back to each capability's own description — not just its name — to check for a semantic match before concluding a real coverage gap exists. A function and a capability can genuinely correspond despite a naming mismatch; declaring a gap on name-matching alone risks false positives that a description-level check would resolve.
  • Where it belongs: explicitly not Work Assignment 2 (2_product_mapping_report.md) — that document is scoped as pure structural mapping only (confirmed in this session's read-through) and doesn't carry gap-analysis methodology. The user's instruction: add it to "the WA3 appendix" — i.e. Work Assignment 1b (1b_investment_rubric.md), once it's restructured as a Work Assignment 3 Appendix per the rename under discussion (see the WA1b→WA3-Appendix conversation this same session). Also relevant to Work Assignment 3 (3_risk_and_gap_analysis.md) itself, and to the MECE "collectively exhaustive / mutually exclusive" completeness test already defined in the Work Assignment 2 Appendix's Primitive 3 section — the same kind of check, applied at the point a gap is first identified rather than only when the capability model itself is being reviewed for completeness.
  • Status: not yet written into any document. Neither Work Assignment 1b's rewrite nor its rename to a Work Assignment 3 Appendix has started. Logged here so this methodology point isn't lost between now and when that work happens.

2026-09-30: Product Workflow DataBooks Rebuilt as Functions and Steps — Tier Layer Removed ​

  • Decision (Ben): the five product workflow DataBooks are rebuilt from outputs/{eop,icrf,itrf,global_gravity}_product_flow.drawio and outputs/satellite_orbits_product_flow3.drawio as bacm:ValueStream → bacm:ValueStreamStage (Function, F1…Fn) → ggsc:Step, with ggsc:flowsTo between steps and ggsc:requiresCapability cross-mapping each Function to the canonical ggsc-cap: IRIs. No Tier layer. Tier risk scores are dropped: WA2/WA3 are incomplete and the scores are not being used.
  • Ontology v1.6: adds ggsc:Step, ggsc:hasStep/inStage, ggsc:flowsTo, ggsc:stageSequence, ggsc:egvDataLevel, ggsc:requiresCapability, ggsc:isExternalInput, ggsc:isCombinationBottleneck. stageSequence is production order, not F-number: ICRF correlation F4 precedes archiving F3. EOP's combined F7a/F7b lane is modelled as two parallel stages sharing sequence 7.
  • Loaded to Fuseki ggsc: …/databooks/{eop,gravity,icrf,itrf,satellite-orbits}-workflow-v0.3#workflow-topology (217/180/177/217/231 triples) plus #meta graphs. Stage/step/flow counts match the drawio sources (EOP 10/19/20, Gravity 9/15/15, ICRF 8/16/16, ITRF 9/21/23, Satellite Orbits 7/23/23); every ggsc-cap: IRI resolves in the capability-maturity graph. Maturity scores are not copied into these graphs — they stay on the capability.
  • Removed from Fuseki: the four *-workflow-v0.1#…-topology graphs and all v0.1/v0.2 #meta graphs; graph/tier-risk-scores-2026-06-22 (recoverable from data/shacl/tier-risk-scores.databook.md in git). In both Satellite Orbits workflow graphs, the five ggsc:WorkflowTier nodes and their hasChild links were deleted and the "(Tier N)" suffix stripped from 14 ggsc:role labels; the centres, organisations and risks there are untouched.
  • Not touched, needs a decision: principle-violations-v1 (5 product graphs, every triple tier-keyed), the Satellite Orbits capabilities/risks graphs (a few tier references each), and the WA2/WA3 deliverables and 1b_investment_rubric.md, which still cite tier scores.

2026-09-30: Source Trace for the Workflow Functions and Steps (NOT LOADED; dropped pending Ben's review) ​

  • Status update (Ben): the source trace is not considered authoritative enough to load. The workflow-provenance graph was dropped from Fuseki and the five ontology terms that supported it (ggsc:SourceRecord, sourceStatus, sourceNote, sourceKind, verifiedOn) were removed from graph/ontology. Nothing from this trace is in Fuseki. The register and DataBook remain in the repo as drafts for review; do not load them until a more robust DataBook exists. The description below records what was drafted, not what is live.
  • Decision (Ben): trace the authoritative source for the workflow content "for the benefit of other users". Every Function and Step in the v0.3 graphs now carries a ggsc:sourceStatus, and each Function links to its sources with prov:hadPrimarySource. Steps record their drawio cell with prov:wasDerivedFrom and inherit their Function's status. The graph is https://w3id.org/ggsc/graph/workflow-provenance (38 source records); the source is data/shacl/workflow-provenance.databook.md and the human-readable register is data/project/workflow_source_register.md.
  • Result: of 43 Functions, 13 Verified, 21 Partial, 9 Unverified (94 Steps: 26 / 50 / 18). Verification used the services' own pages (IERS, IGS, IVS, IGN, ICGEM, GFZ) and DOIs, checked on 30 September 2026. "Unverified" means no primary source was found, not that the claim is wrong.
  • Finding, agy function-validation findings: the document that originated the Function lists has unreliable reference lists. Of 18 external citations, 8 contain errors or could not be confirmed, 6 were confirmed and 4 were not checked. Examples: Ince et al. 2019 is in ESSD not J. Geodesy; IERS TN 41 is "Analysis and results of ITRF2020" (2024), not a 2023 CATREF manual; the GSFC ACC announcement is dated 24 July 2024, not 2025. Do not cite that document's references without checking them.
  • Discrepancies logged, none corrected yet: the EOP Bulletin B step reads "IERS 14 C04" but EOP 20 C04 is current; Satellite Orbits "106 of 107 Core Sites" cannot be reproduced from the stations graph (69 flagged core); the IGS product latencies (15 min / 17 h / 12-18 d) are not on the igs.org product pages.
  • Also removed on the same day: the seven principle-violations-v1 graphs (every triple tier-keyed). Source remains in git as data/shacl/principle-violations.databook.md.
  • Open: the 9 Unverified and 21 Partial Functions need a primary source; the next useful step is linking Steps to the organisations and centres in the infrastructure graph.