Gap Scan — 2026-05-25

Live run. The 6 new pages and 4 expansions below were researched and committed directly to their real Wiki/ folders (status: active; new pages confidence: medium, except entities/herbert-simon at high for an uncontested historical record). No foundational source was queued this run — see "Authenticity-verification failures" and "Deferred backlog."

Scanned

Recent window (last 48h): 6 New Developments Log/ files (2026-05-23-1524 through 2026-05-25-0511); ~93 wiki pages with last_updated inside the window. The two genuinely new dev-logs — 2026-05-24-1510 and 2026-05-25-0511 — were both already folded by the nightly developments-log / ingest-reflect cycles (the May 24 lint report confirms the 05-24 file; the cluster of companies/ and legislation/ edits at 05:12–05:26 on May 25 confirms the 05-25 file). No recent-thread item was left unhandled by the nightly run. Rotation slice: 4 of 14 — entities/ A–G (~120 pages). Broken-link analysis: 24,186 wikilinks across 1,247 pages → 556 unique broken targets. After alias resolution, doc-artifact filtering (briefings/, policy-briefs/, log.md, wikilinks, etc.), and dropping items already actioned in the 2026-05-22 / -23 / -23-run2 / -24 gap-scan reports, items with a queued INGEST- task, and items with an existing Raw Sources/ file, ~30 genuine gaps remained. The recent window and the rotation slice both converged on one dominant cluster: the Knight Columbia 2026 symposium foundational sources, now ingested, whose author-entities and microfoundation-concepts have no pages. The 13 highest-scored were actioned (10 substantive + 3 alias-fix clusters).

Gaps actioned (13 of ~30 found)

New pages created (live)

• Classification Institutions — gap type 4: both The Democratic Matrix and Normative Competence explicitly name [[concepts/classification-institutions]] as "the Hadfield-Weingast microfoundation" with no page behind it (5 inbound). The institution that produces a society's shared classification scheme of punishable/permitted behaviors; built from the wiki's own framing plus the established Hadfield-Weingast (2012, 2014) theory. Core safety/governance area; twice-deferred (2026-05-24).
• Rakshit Trivedi — gap type 3: co-author of Building AI for the Democratic Matrix; 7 inbound from The Democratic Matrix, Gillian K. Hadfield, Dylan Hadfield-Menell, and the source page. Verified: postdoctoral associate, Algorithmic Alignment Group, MIT CSAIL (with Hadfield-Menell). Twice-deferred.
• Herbert A. Simon — gap type 3: foundational AI-history figure (Turing Award 1975, Nobel 1978; bounded rationality; Logic Theorist) referenced from AI as Social Technology, Henry Farrell, and the Democratic Matrix source. Created at confidence: high — uncontested historical record.
• Amy X. Zhang — gap type 3: co-author of Levels of Autonomy for AI Agents; 6 inbound from Levels of Autonomy for AI Agents (Feng-McDonald-Zhang framework), Kevin Feng, and the source page. Verified: Associate Professor, UW Allen School; head of the Social Futures Lab.
• David W. McDonald — gap type 3: co-author of Levels of Autonomy for AI Agents; 6 inbound (same cluster). Verified: Professor & Chair, UW Department of Human Centered Design & Engineering.
• Nik Marda — gap type 3: co-author of A Conceptual Model to Guide AI Risk Governance Strategies; 5 inbound from Sociotechnical AI Risk Governance, Deirdre K. Mulligan, and the source page. Verified: Technical Lead for AI Governance at Mozilla (through 2025), formerly White House OSTP. One affiliation discrepancy noted on-page and below.

Pages expanded (live)

• Center for Security and Emerging Technology (CSET) — thin anchor: in-degree ~30, 275 words, with three malformed relationship links ([[ai-national-security]], [[cset-csis-us-china-natsec]], [[export-controls-ai]], [[ai-race-dynamics]] — all missing their folder prefix). Expanded to ~620 words: founding (Jan 2019, Georgetown Walsh School), founding director Jason Matheny, funding history (Open Philanthropy ~$38.9M founding grant; $42M follow-on; >$100M cumulative), the Emerging Technology Observatory data tooling. 4 malformed links repaired to typed wikilinks; 2 supporting (Source: URL) cites folded. Held at confidence: medium.
• Cosma Rohilla Shalizi — thin anchor: in-degree 16, 198 words; live Farrell-Shalizi thread. Expanded to ~470 words: CMU career detail (associate professor since 2014, at CMU since 2005), the Three-Toed Sloth notebook, the Advanced Data Analysis textbook, and a fuller account of his statistical-modeling / coarse-graining contribution and the 2010 "Singularity began with the Industrial Revolution" essay. 2 supporting cites folded; held at confidence: medium.
• Christoph Winter — thin anchor: in-degree 10, 74 words. Expanded to ~390 words and corrected a factual error — the page said he is "affiliated with" the Institute for Law & AI; he is its Director. Added Cambridge appointment, EU GPAI Code of Practice advisory role, and a full publication record. 1 supporting cite; held at confidence: medium.
• Charlie Bullock — thin anchor: in-degree 10, ~60 words at confidence: low. Expanded to ~310 words: role precision (Senior Research Fellow at the Institute for Law & AI), publication record. 1 supporting cite folded; confidence raised low → medium.

Link aliases fixed

• [[entities/electronic-frontier-foundation]] → [[entities/eff]] — 2 instances in Privacy Without Monopoly: Data Protection and Interoperability — Bennett Cyphers and Cory Doctorow (EFF, 2021). The real page is entities/eff (it exists); stale (planned) parentheticals dropped.
• [[concepts/interpretability]] → [[concepts/mechanistic-interpretability]] — 2 instances in AI as Social Technology. Alias-resolution confirmed the wiki's canonical slug is mechanistic-interpretability; no separate concepts/interpretability page is warranted.
• [[companies/nvidia]] → [[companies/nvidia-tsmc]] — 1 instance in Data Centers in Space: A Pipe Dream, or AI's Next Big Thing? — Micah Maidenberg, Merrill Sherman (WSJ, May 12 2026). The canonical company page is nvidia-tsmc.

Queued — foundational sources

None this run. The candidate pool was dominated by missing author-entities and concept anchors of already-ingested Knight Columbia sources — gap types 3 and 4, which the AUTO lane handles live. No new dangling foundational source surfaced that was not already queued or deferred. The four standing INGEST- tasks (Robin, Law-Following AI, AI RMF, Secret Loyalties) remain queued for the next ingest cycle.

Authenticity-verification failures

None — no raw source was pulled or saved this run. One forward-looking candidate was deliberately not pulled: the Pope Leo XIV AI encyclical, unveiled today (2026-05-25) with Anthropic's Chris Olah present (per the 05-25 dev-log). It is a primary text and a foundational candidate (broken slug anthropic-pope-leo-encyclical-may-2026, 2 inbound), but a same-day-of-publication pull risks an unstable or pre-final text. Per the protocol ("when in doubt, do not save"), it is deferred for verification against vatican.va once the canonical text has settled. See Deferred backlog.

Notes for the next lint / ingest pass

• Affiliation discrepancy — Nik Marda. The Knight Columbia Risk Governance source page lists Marda with a UC Berkeley affiliation (alongside Mulligan, UC Berkeley iSchool). His extensively-documented role is Technical Lead for AI Governance at Mozilla (through 2025; he left Mozilla in mid-2025, before the March 2026 paper). The new Nik Marda page records both rather than over-specifying; a future ingest of the source could confirm his affiliation as printed on the paper.
• concepts/ai-coarse-grainings is still referenced by Cosma Rohilla Shalizi (kept as a related: link, annotated "concept stub — to be created") and the Farrell-Shalizi source. It is a genuine gap type 4; deferred this run (score 2).
• entities/victor-zhenyi-wang — the third co-author of the Mulligan-Marda-Wang paper — was deliberately not created. Unlike his two co-authors he is sparsely documented in public sources; per the quality gate, a confident, accurate page could not be written this run. Deferred for a targeted pull.

Deferred backlog (over the daily cap — re-surfaces next run)

• entities/victor-zhenyi-wang — co-author, Mulligan-Marda-Wang; 5 inbound. Quality-gate deferral (sparse public record). Score 3.
• concepts/ai-and-children — child-safety topical hole; 3 inbound (Tumbler Ridge Families v. OpenAI + the Blackburn "one rulebook" source). Score 3.
• entities/john-moolenaar — Congressman, chair of the House Select Committee on the CCP; namesake of SCALE Act — Rep. John Moolenaar (April 2026); 3 inbound. Score 3.
• concepts/seeing-like-a-state — James C. Scott's analytic lens, used by Farrell-Shalizi; 3 inbound. Score 3.
• concepts/ai-political-economy — Farrell-Shalizi cluster; 3 inbound from AI as Social Technology, The Long Industrial Revolution, and the source. Score 2.
• entities/cnas + entities/nscai — Center for a New American Security; National Security Commission on AI. 2 inbound each, from the Frazier-Rozenshtein "dominance by understanding" cluster; core national-security area. Score 2 each.
• entities/pim-de-witte — General Intuition / Medal founder; the entities/pim-dewitte ↔ entities/pim-de-witte slug split (5 combined inbound) needs the page created before the alias can be fixed. Score 2.
• concepts/ai-coarse-grainings — gap type 4, Farrell-Shalizi statistical-modeling concept. Score 2.
• concepts/autonomy-certificates — still quality-gate-ambiguous (stub vs. merge into Levels of Autonomy for AI Agents (Feng-McDonald-Zhang framework)); needs a human keep/merge call. Score 2.
• concepts/ai-risk-management — deferred again: INGEST-ai-risk-management-framework-2026-05-14 is queued and NIST AI Risk Management Framework 1.0 exists; a generic concept page risks rework. Score 2.
• Pope Leo XIV AI encyclical — foundational primary text; pull and verify against vatican.va once the canonical text settles. Score 2.
• concepts/chinese-ai-policy — possible scope overlap with China Generative AI Registration (da moxing bei'an / 大模型备案); needs scoping before creation. Score 1.
• concepts/fifth-amendment-and-ai — niche; 1 live content inbound (Farahany neurorights source). Score 1.
• entities/jack-dorsey, entities/brian-armstrong — tech CEOs (Block, Coinbase) referenced from Intelligence Replaces Hierarchy; peripheral to AI policy. Score 1 each.
• Mulligan-cluster concept stubs — concepts/regulatory-managerialism, concepts/safetywashing, concepts/handoffs (2 inbound each, single-source). Score 1.
• Carried lint cleanups (not gap-identifier pulls): sources/anthropic-pwc-expanded-partnership (7 broken — likely supporting, folds into Anthropic); sources/trustworthy-agents-in-practice (6+3 — log.md records the page as both created and deleted). Flagged for lint.
• Carried entity pages: david-mcdonald ✅ actioned this run; remaining from prior backlogs — nathan-delisle, herbert-simon ✅ actioned, amy-zhang ✅ actioned, rakshit-trivedi ✅ actioned, nik-marda ✅ actioned.

One-line summary

Ten substantive gaps actioned plus three alias clusters: six new pages created live — the Hadfield-Weingast microfoundation Classification Institutions, AI-history figure Herbert A. Simon, and four Knight Columbia paper co-authors (Rakshit Trivedi, Amy X. Zhang, David W. McDonald, Nik Marda) — and four thin entity anchors expanded (Center for Security and Emerging Technology (CSET), Cosma Rohilla Shalizi, Christoph Winter, Charlie Bullock), with a factual correction to Christoph Winter's ILAI role and a Nik Marda affiliation discrepancy flagged for the next ingest.