Neel Nanda

{
  "id": "neel-nanda",
  "wikiId": "E214",
  "path": "/knowledge-base/people/neel-nanda/",
  "filePath": "knowledge-base/people/neel-nanda.mdx",
  "title": "Neel Nanda",
  "quality": 26,
  "readerImportance": 84.5,
  "researchImportance": 40.5,
  "tacticalValue": 72,
  "contentFormat": "article",
  "causalLevel": null,
  "lastUpdated": "2026-03-16",
  "dateCreated": "2026-02-15",
  "summary": "Comprehensive biographical profile of Neel Nanda covering his role as DeepMind's mechanistic interpretability team lead, key contributions (TransformerLens, Gemma Scope, grokking paper), and his evolving views on interpretability's limitations and strategic pivot toward applied safety problems. Notably includes his 2025 admission that some high-profile mech interp results were 'wrong or significantly weaker than originally claimed' and that the 'most ambitious vision of the field is probably dead.'",
  "description": "Senior Research Scientist and Mechanistic Interpretability Team Lead at Google DeepMind; creator of TransformerLens; co-author of 'A Mathematical Framework for Transformer Circuits'; researcher on grokking, induction heads, and sparse autoencoders",
  "ratings": {
    "novelty": 2,
    "rigor": 3,
    "completeness": 4.5,
    "actionability": 2.5
  },
  "category": "people",
  "subcategory": "safety-researchers",
  "clusters": [
    "ai-safety"
  ],
  "metrics": {
    "wordCount": 2632,
    "tableCount": 2,
    "diagramCount": 0,
    "internalLinks": 31,
    "externalLinks": 8,
    "footnoteCount": 30,
    "bulletRatio": 0.23,
    "sectionCount": 15,
    "hasOverview": true,
    "structuralScore": 14
  },
  "suggestedQuality": 93,
  "updateFrequency": null,
  "evergreen": true,
  "wordCount": 2632,
  "unconvertedLinks": [
    {
      "text": "neelnanda.io",
      "url": "https://www.neelnanda.io",
      "resourceId": "028435b427f72e06",
      "resourceTitle": "Mechanistic interpretability work"
    },
    {
      "text": "neelnanda.io",
      "url": "https://www.neelnanda.io",
      "resourceId": "028435b427f72e06",
      "resourceTitle": "Mechanistic interpretability work"
    }
  ],
  "unconvertedLinkCount": 2,
  "convertedLinkCount": 0,
  "backlinkCount": 16,
  "citationHealth": {
    "total": 4,
    "withQuotes": 4,
    "verified": 4,
    "accuracyChecked": 4,
    "accurate": 2,
    "inaccurate": 0,
    "avgScore": 0.9874999970197678
  },
  "hallucinationRisk": {
    "level": "medium",
    "score": 60,
    "factors": [
      "biographical-claims",
      "low-rigor-score",
      "low-quality-score",
      "well-cited"
    ]
  },
  "entityType": "person",
  "redundancy": {
    "maxSimilarity": 16,
    "similarPages": [
      {
        "id": "chris-olah",
        "title": "Chris Olah",
        "path": "/knowledge-base/people/chris-olah/",
        "similarity": 16
      },
      {
        "id": "connor-leahy",
        "title": "Connor Leahy",
        "path": "/knowledge-base/people/connor-leahy/",
        "similarity": 14
      },
      {
        "id": "anthropic-core-views",
        "title": "Anthropic Core Views",
        "path": "/knowledge-base/responses/anthropic-core-views/",
        "similarity": 14
      },
      {
        "id": "interpretability",
        "title": "Mechanistic Interpretability",
        "path": "/knowledge-base/responses/interpretability/",
        "similarity": 14
      },
      {
        "id": "technical-research",
        "title": "Technical AI Safety Research",
        "path": "/knowledge-base/responses/technical-research/",
        "similarity": 14
      }
    ]
  },
  "changeHistory": [
    {
      "date": "2026-02-19",
      "branch": "claude/citation-pipeline-iteration-KvR2n",
      "title": "Citation pipeline improvements and footnote normalization",
      "summary": "Fixed citation extraction to handle all footnote formats (text+bare URL), created a\nfootnote normalization script that auto-converted 58 non-standard footnotes to\nmarkdown-link format, switched dashboard export from JSON/.cache to YAML/data/ for\nproduction compatibility, ran the citation accuracy pipeline on 5 pages\n(rethink-priorities, cea, compute-governance, hewlett-foundation,\ncenter-for-applied-rationality) producing 232 citation checks with 57% accurate, 16%\nflagged, re-verified colorado-ai-act archive outside sandbox (18/19 verified), and\nimproved difficulty distribution to use structured categories (easy/medium/hard) with\nnormalization fallback.",
      "model": "claude-opus-4-6",
      "duration": "~1h"
    },
    {
      "date": "2026-02-18",
      "branch": "claude/fix-issue-240-N5irU",
      "title": "Surface tacticalValue in /wiki table and score 53 pages",
      "summary": "Added `tacticalValue` to `ExploreItem` interface, `getExploreItems()` mappings, the `/wiki` explore table (new sortable \"Tact.\" column), and the card view sort dropdown. Scored 49 new pages with tactical values (4 were already scored), bringing total to 53.",
      "model": "sonnet-4",
      "duration": "~30min"
    },
    {
      "date": "2026-02-17",
      "branch": "claude/review-wiki-editing-scCul",
      "title": "Wiki editing system refactoring",
      "summary": "Six refactors to the wiki editing pipeline: (1) extracted shared regex patterns to `crux/lib/patterns.ts`, (2) refactored validation in page-improver to use in-process engine calls instead of subprocess spawning, (3) split the 694-line `phases.ts` into 7 individual phase modules under `phases/`, (4) created shared LLM abstraction `crux/lib/llm.ts` unifying duplicated streaming/retry/tool-loop code, (5) added Zod schemas for LLM JSON response validation, (6) decomposed 820-line mermaid validation into `crux/lib/mermaid-checks.ts` (604 lines) + slim orchestrator (281 lines). Follow-up review integrated patterns.ts across 19+ files, fixed dead imports, corrected ToolHandler type, wired mdx-utils.ts to use shared patterns, replaced hardcoded model strings with MODELS constants, replaced `new Anthropic()` with `createLlmClient()`, replaced inline `extractText` implementations with shared `extractText()` from llm.ts, integrated `MARKDOWN_LINK_RE` into link validators, added `objectivityIssues` to the `AnalysisResult` type (removing an unsafe cast in utils.ts), fixed CI failure from eager client creation, and tested the full pipeline by improving 3 wiki pages. After manual review of 3 improved pages, fixed 8 systematic pipeline issues: (1) added content preservation instructions to prevent polish-tier content loss, (2) made auto-grading default after --apply, (3) added polish-tier citation suppression to prevent fabricated citations, (4) added Quick Assessment table requirement for person pages, (5) added required Overview section enforcement, (6) added section deduplication and content repetition checks to review phase, (7) added bare URL→markdown link conversion instruction, (8) extended biographical claim checker to catch publication/co-authorship and citation count claims.\n\nSubsequent iterative testing and prompt refinement: ran pipeline on jan-leike, chris-olah, far-ai pages. Discovered and fixed: (a) `<!-- NEEDS CITATION -->` HTML comments break MDX compilation (changed to `{/* NEEDS CITATION */}`), (b) excessive citation markers at polish tier — added instruction to only mark NEW claims (max 3-5 per page), (c) editorial meta-comments cluttering output — added no-meta-comments instruction, (d) thin padding sections — added anti-padding instruction, (e) section deduplication needed stronger emphasis — added merge instruction with common patterns. Final test results: jan-leike 1254→1997 words, chris-olah 1187→1687 words, far-ai 1519→2783 words, miri-era 2678→4338 words; all MDX compile, zero critical issues.",
      "pr": 184
    }
  ],
  "coverage": {
    "passing": 8,
    "total": 13,
    "targets": {
      "tables": 11,
      "diagrams": 1,
      "internalLinks": 21,
      "externalLinks": 13,
      "footnotes": 8,
      "references": 8
    },
    "actuals": {
      "tables": 2,
      "diagrams": 0,
      "internalLinks": 31,
      "externalLinks": 8,
      "footnotes": 30,
      "references": 1,
      "quotesWithQuotes": 4,
      "quotesTotal": 4,
      "accuracyChecked": 4,
      "accuracyTotal": 4
    },
    "items": {
      "summary": "green",
      "schedule": "red",
      "entity": "green",
      "editHistory": "green",
      "overview": "green",
      "tables": "amber",
      "diagrams": "red",
      "internalLinks": "green",
      "externalLinks": "amber",
      "footnotes": "green",
      "references": "amber",
      "quotes": "green",
      "accuracy": "green"
    },
    "editHistoryCount": 3,
    "ratingsString": "N:2 R:3 A:2.5 C:4.5"
  },
  "readerRank": 58,
  "researchRank": 340,
  "recommendedScore": 113.07
}

No external links