# Source Register — The Enterprise Has No Memory

Retrieved: 2026-10-07

Status: external evidence hardening for Issue #1319. These sources corroborate components of the thesis. None is authority for Tony's synthesis or for GhostMesh architecture.

## S1 — NASA APPEL Knowledge Management

Publisher: NASA
Title: Knowledge Management
Canonical URL: https://www.nasa.gov/learning-resources/for-professionals/appel-knowledge-management/
Page updated: June 12, 2026
Source class: authoritative first-party operational practice

Supports:
- NASA explicitly treats knowledge continuity across programs, projects, centers, and employee transitions as an operational concern.
- NASA describes critical knowledge as experience-derived, not easily replaceable, and important to well-informed decisions.
- NASA provides knowledge-capture and transfer resources for employee retirement and transition.

Does not establish:
- that enterprises generally have "no memory";
- that document repositories are categorically inadequate;
- that a cognitive graph or GhostMesh architecture is the necessary solution;
- that all conversational context should be retained.

Use in article:
Corroboration for the employee-transition and knowledge-continuity problem.

## S2 — NASA Lessons Learned

Publisher: NASA Office of the Chief Engineer
Title: NASA Lessons Learned
Canonical URL: https://www.nasa.gov/nasa-lessons-learned/
Source class: authoritative first-party operational system

Supports:
- NASA maintains an official reviewed lessons-learned system.
- Lessons include the original driving event and recommendations.
- Lessons feed training, best practices, policies, and procedures.

Does not establish:
- full decision-lineage reconstruction;
- the proposed cognitive-graph architecture;
- the proposition that current enterprise systems erase causality.

Use in article:
Strong example that useful institutional learning preserves the relationship between an event and a recommendation, rather than storing an isolated artifact.

## S3 — NASA Knowledge Loss / Experienced Workforce Lesson

Publisher: NASA Lessons Learned Information System
Lesson: Reduction in NASA Knowledge Base Due to Loss of Experienced Employees and Hiring of Inexperienced Employees
Lesson number: 1171
Canonical URL: https://llis.nasa.gov/lesson/1171
Lesson date: 2001-02-01
Source class: authoritative historical first-party lesson

Supports:
- NASA documented loss of experienced personnel as a knowledge-base risk.
- The lesson states there was no systematic effort at that time to capture experienced personnel knowledge before departure.

Does not establish:
- current NASA practice;
- a general quantitative rate of enterprise knowledge loss;
- the complete thesis.

Use in article:
Historical evidence that workforce transitions can create material knowledge continuity problems.

## S4 — Stein, Organization Memory review

Publisher: International Journal of Information Management / Elsevier
Author: E.W. Stein
Title: Organization memory: Review of concepts and recommendations for management
Publication: 1995, Volume 15, Issue 1, pp. 17–32
DOI: 10.1016/0268-4012(94)00003-C
Canonical URL: https://www.sciencedirect.com/science/article/pii/026840129400003C
Source class: established prior academic work

Supports:
- organizational memory is longstanding prior art;
- acquisition, retention, maintenance, and retrieval of organizational memory have been studied across multiple disciplines.

Does not establish:
- Tony's specific continuity-of-cognition architecture;
- modern LLM/RAG limitations;
- that existing organizational-memory models solve current enterprise AI governance.

Use in article:
Prior-art boundary. The article must not imply that organizational memory itself is a new concept.

## S5 — Alvarado, Bañares-Alcántara, Trujillo

Publisher: Journal of Petroleum Science and Engineering / Elsevier
Title: Improving the Organisational Memory by recording decision making, rationale and team configuration
Publication: 2005, Volume 47, Issues 1–2, pp. 71–88
Canonical URL: https://www.sciencedirect.com/science/article/pii/S0920410505000173
Source class: established prior academic architecture

Supports:
- explicit representation of decision-making rationale has prior art in organizational-memory architecture;
- organizational memory may include workflow, participants, rationale, experience, and dynamic process evolution.

Does not establish:
- GhostMesh novelty;
- policy-mediated joins, model-independent context fabric, or the full current thesis;
- any claim that Tony invented decision-rationale capture.

Use in article:
Important anti-overclaim boundary and corroboration that rationale and decision process have long been recognized as valuable memory components.

## S6 — NASA knowledge-management history

Publisher: NASA APPEL Knowledge Services
Title: History of APPEL Knowledge Services
Canonical URL: https://www.nasa.gov/learning-resources/for-professionals/appel-history-and-timeline/
Source class: authoritative first-party historical context

Supports:
- NASA has invested for decades in training, lessons learned, on-the-job learning, and knowledge sharing as part of technical workforce development.

Does not establish:
- causality between a specific knowledge-management practice and mission outcomes;
- the article's broader enterprise-AI thesis.

## Attribution rule

The article should say that the problem of organizational memory and rationale capture has substantial prior art.

Tony's contribution here is the synthesis connecting:
- organizational memory;
- modern model/RAG systems;
- temporal and provenance-aware context;
- cognitive graph projections;
- deterministic authority;
- model/provider portability;
- recovery and continuity of cognition.

That synthesis may be original to this body of work, but no "first" claim is authorized without a broader prior-art search.

## S7 — Anthropic context engineering

Publisher: Anthropic
Title: Effective context engineering for AI agents
Canonical URL: https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents
Published: 2025-09-29
Retrieved: 2026-10-07
Source class: first-party current agent-engineering guidance

Supports:
- modern agent engineering requires management of context beyond prompt wording;
- context can include system instructions, tools, external data, and message history;
- long-running/multi-turn agents require active curation of evolving context state.

Does not establish:
- organizational memory as defined in this Work;
- that all enterprise context should be retained;
- temporal decision lineage, authority semantics, or GhostMesh architecture;
- that Anthropic endorses this thesis.

Use in article:
Current corroboration that the engineering focus has expanded from prompts toward managed context state.

## S8 — NIST AI Risk Management Framework / Generative AI Profile

Publisher: National Institute of Standards and Technology
Titles:
- AI Risk Management Framework
- Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1)
Canonical URLs:
- https://www.nist.gov/itl/ai-risk-management-framework
- https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence
Published: AI RMF 1.0 in 2023; GAI Profile 2024; current framework revision and resources rechecked 2026-10-07
Retrieved: 2026-10-07
Source class: authoritative government risk-management guidance

Supports:
- AI risks should be managed across design, development, use, and evaluation;
- trustworthy AI requires explicit risk-management practices rather than treating model output as self-validating authority;
- testing, evaluation, verification, and validation remain relevant operating practices.

Does not establish:
- the specific GhostMesh authorization model;
- the article's institutional-memory thesis;
- a requirement for cognitive graphs;
- any claim that NIST endorses SharePlane or GhostMesh.

Use in article:
Governance corroboration for the distinction between machine reasoning and institutional authority.
