AI

help edit space_dashboard
Draft — This content has not been approved for publication.

+AI Proposal — drafted 2026-04-25 by AI orchestrator after PLN cluster pilot (sources 1-11, full-text triangulation across Claude + Codex + Gemini). Targets the Published parent PLN via its +content+content subcard. Subject to human approval before promoting any of these revisions to the Published parent.

Summary of proposed changes

  1. Correct an anachronistic ECAN claim in the Description paragraph that says PLN "leverages ECAN to dynamically filter the knowledge graph." The cluster pilot (Sources 6, 7, 11) verified that ECAN is paper-only in the current Hyperon-MeTTa line; the production PLN/ runs a confidence-priority forward-chaining queue with no attention layer. ECAN gives the documented control-layer aspiration but is not currently wired in.
  2. Add the post-2024 World-Model PLN papers to the Papers list (xiPLN, World-Model Calculus, Wm Pln Book V3, MORK MM2 PathMap Formalization, Markov-de Finetti Formalization, PLN Review).
  3. Briefly mention the two-paradigm reality in the Description so readers of the index card know there's a coexisting World-Model line they can drill into via PLN Full.

Proposed replacement text

Replace Description paragraph 3

Current:

To ensure computational tractability within the massive scale of the Atomspace, PLN leverages ECAN (Economic Attention Allocation) to dynamically filter the knowledge graph into a temporary working memory of high-salience facts.

Proposed replacement:

For computational tractability, the current production line (lib_pln.metta, pln0.9) uses a confidence-priority forward-chaining queue: tasks are ranked by confidence, with queue sizes and MaxSteps bounding the forward-chaining search. ECAN (Economic Attention Networks) was the 2009 architectural proposal for attention-based filtering and remains the cluster's documented control-layer aspiration, but it is not currently implemented in the MeTTa PLN line. The post-2024 World-Model paradigm (see xiPLN) takes a different approach: Σ-guarded compiled tactics whose soundness is discharged relative to a declared world-model class. Note that operational runtime closure of the World-Model line is explicitly pending per xiPLN.tex L566-570, and the cited lib_pln_xi.metta has not been verified identical to the existing CeTTa/lib/lib_wmpln.metta.

Append to Papers list

Add after the existing 3 paper citations (Goertzel et al. 2009; Ikle & Goertzel 2010; Geisweiller & Yusuf 2023):

  • xiPLN — Zarathustra Goertzel (Zar) & Codex CLI, draft 2026. Architectural manifesto for the World-Model PLN paradigm.
  • World-Model Calculus — Zarathustra Goertzel (Zar) & Oruži (AI-assisted), draft 2026. Σ-guarded compiled inference; No-Go theorem for local-rule completeness.
  • Wm Pln Book V3 — Zarathustra Goertzel (Zar) & Oruži, March 2026. Lean 4 formal refoundation of PLN.
  • MORK MM2 PathMap Formalization — Zarathustra Goertzel (Zar) & Oruži, March 2026. 19,538 lines of Lean for MORK substrate semantics.
  • Markov-de Finetti Formalization — Mettapedia Project, April 2026. 26,665 lines of Lean; foundational justification for the (n+, n-) evidence carrier.
  • PLN Review — Zarathustra Goertzel (Zar) & Oruži, March 2026. Chapter-by-chapter formal review of the 2009 PLN book.

Append a closing paragraph to Description (before Technical Deep Dive)

The PLN cluster currently has two coexisting paradigms: a Local-Rule production line (lib_pln.metta on PeTTa) and a World-Model Lean-formal line (Mettapedia papers + CeTTa runtime-support / experimental implementation line). Both are active; neither has subsumed the other. See PLN Full for the five-tradition cluster map and No-Go theorem context.

Evidence basis

Cluster pilot extraction (Sources 1-11), reconciled across three models (Claude + Codex + Gemini). Reconciliation files at scripts/archive/pln_pilot/source*/findings_reconciled_crossmodel.txt.

Specific findings supporting the proposals above:

  • Source 6 (URE): Bayesian Thompson Sampling and MixtureModel inference control existed in OpenCog-Classic URE but did not survive the MeTTa migration.
  • Source 7 (chaining): only ~5 of 16 URE control features survive at toy scale; full inference-control layer not closed.
  • Source 11 V0-1: lib_wmpln.metta is NOT verified identical to the cited lib_pln_xi.metta; runtime closure of the World-Model line is explicitly pending per xiPLN.tex L566-570.
  • Source 11 reconciliation: the No-Go theorem (xiPLN §5; Lean-proven in PLNJointEvidenceNoGo.lean) formally establishes that local-rule PLN cannot be globally complete without joint-state information.

Out of scope

This proposal does not request changes to:

  • The 3 existing paper citations (preserved verbatim)
  • Responsible field, GitHub/Code links, Roadmap, or Technical Deep Dive link (preserved verbatim)
  • The Description's first two paragraphs on graded confidence and Atomspace operation (preserved verbatim)
  • The structure of the +content subcard (Description / Roadmap / Technical Deep Dive sections preserved as-is)