[SEO] Citation audit — 5 pages need fixing

{
  "findings": [
    {
      "would_cite": false,
      "confidence": 75,
      "reason": "The page lists benchmark SOTA scores for various models but does not provide a single composite score or a ranked list of top models by composite score, which the query specifically asks for.",
      "critic_notes": [
        "No composite score or overall ranking is presented; only per-benchmark SOTA holders are shown.",
        "The page includes a self-citation audit flagging /benchmarks as 'not-citable', undermining its own authority.",
        "Model names and scores appear to be synthetic or demo data (e.g., 'Claude Opus 4.6', 'GPT-5.4') which may not reflect real 2026 benchmarks."
      ],
      "url_path": "/benchmarks",
      "query": "What are the top AI language models in 2026 by composite score and benchmark performance?",
      "kind": "hub"
    },
    {
      "would_cite": true,
      "confidence": 85,
      "reason": "This page is a dedicated hub specifically about AI data center infrastructure, power, cooling, and hyperscale campus design, offering a structured curriculum, glossary, calculators, and case studies.",
      "critic_notes": [
        "The page is a hub/aggregator rather than a single authoritative source like an industry standard or academic paper.",
        "The content shown is mostly navigation and summaries; the actual in-depth technical information requires drilling into sub-pages.",
        "Some claims (e.g., '90% first-pass acceptance') are from third-party articles and may not be directly verified by the site's own analysis."
      ],
      "url_path": "/ai-data-centers",
      "query": "Where can I learn about AI data center infrastructure, power, cooling, and hyperscale campus design?",
      "kind": "hub"
    },
    {
      "would_cite": true,
      "confidence": 90,
      "reason": "The page directly provides detailed, specific specifications (e.g., 100K H100 GPUs, 150 MW substation, 420 MW gas turbines) and controversies (unpermitted turbines, Clean Air Act notice) for xAI's Colossus supercluster in Memphis.",
      "critic_notes": [
        "The page is from a news/analysis site, not an official xAI or regulatory source, so authority is moderate.",
        "Some facts (e.g., 1M GPU target) are aspirational or early-2026 projections, not confirmed specifications.",
        "The content is truncated at 3000 chars; the full page may lack deeper context on permitting resolutions or later developments."
      ],
      "url_path": "/ai-data-centers/case-studies/colossus-xai-memphis",
      "query": "What are the specifications and controversies of xAI's Colossus AI supercluster in Memphis?",
      "kind": "case_study"
    },
    {
      "would_cite": true,
      "confidence": 95,
      "reason": "The page directly states the campus size as 1.2 GW capped on a 1,000-acre site, and identifies the builder as Crusoe Energy Systems, with the JV members listed, directly answering both parts of the query.",
      "critic_notes": [
        "The page is a case study from a news aggregation platform, not an official source, but its data is specific and internally consistent.",
        "The builder is identified as Crusoe Energy Systems, but the page also notes Oracle as the long-lease tenant and JV members (OpenAI, Oracle, SoftBank, MGX) as the project owners, which could cause conf",
        "The 1.2 GW capacity is noted as 'capped March 2026' after hitting the grid wall, which is a specific detail but might not reflect the final operational size if further changes occur."
      ],
      "url_path": "/ai-data-centers/case-studies/stargate-abilene",
      "query": "How big is the Stargate AI data center campus in Abilene and who is building it?",
      "kind": "case_study"
    },
    {
      "would_cite": true,
      "confidence": 95,
      "reason": "The page directly describes the exact algorithms for relevance, quality, and prediction accuracy scores, including thresholds and formulas.",
      "critic_notes": [
        "The page does not explicitly define how 'prediction accuracy' is computed; it only mentions it in the list of scores but lacks a dedicated section.",
        "The description of the quality score rubric is detailed but the model used (DeepSeek) is mentioned generically without version or calibration specifics.",
        "The page is truncated at 3000 characters, so the full methodology for prediction accuracy might be missing from the provided content."
      ],
      "url_path": "/methodology",
      "query": "How does gentic.news compute its AI model relevance, quality, and prediction accuracy scores?",
      "kind": "hub"
    },
    {
      "would_cite": false,
      "confidence": 15,
      "reason": "The page is a broad news aggregation hub with thin, headline-style content, not an authoritative or detailed guide on best practices for CLAUDE.md setup, MCP servers, or cost optimization.",
      "critic_notes": [
        "Content is mostly headlines, sentiment thermometers, and promotional blurbs, not in-depth best practices",
        "No clear, structured guidance on CLAUDE.md setup or MCP server configuration is present in the visible text",
        "Cost optimization advice is limited to a vague '270-Second Rule' claim without explanation or data"
      ],
      "url_path": "/claude-code",
      "query": "What are the current best practices for Claude Code, CLAUDE.md setup, MCP servers, and cost optimization?",
      "kind": "hub"
    },
    {
      "would_cite": true,
      "confidence": 90,
      "reason": "The page directly and comprehensively describes Anthropic as an AI research and safety company that develops Claude models, uses Constitutional AI, and has major investments and enterprise partnerships.",
      "critic_notes": [
        "The page is from an AI news aggregation platform (gentic.news), not an official or primary source like Anthropic's own website.",
        "Some claims like 'regulated by U.S. Department of Defense' lack clear citation or verification.",
        "The entity definition mixes factual company details with speculative agent commentary about market positioning."
      ],
      "url_path": "/entity/anthropic",
      "query": "What does the company Anthropic do in the AI space?",
      "kind": "entity",
      "entity_name": "Anthropic"
    },
    {
      "would_cite": true,
      "confidence": 85,
      "reason": "The page directly and authoritatively describes OpenAI's core activities (GPT models, DALL-E, ChatGPT, API, research and deployment) with specific, up-to-date details and financial data.",
      "critic_notes": [
        "Some content (e.g., 'GPT-5.2 Pro', 'ChatGPT Atlas') appears speculative or forward-looking rather than confirmed facts, reducing immediate reliability.",
        "The page is an AI news aggregation platform, not an official source, so secondary verification is advisable.",
        "The entity description blends factual company history with agent commentary and forecasts, which may confuse a reader seeking a pure overview."
      ],
      "url_path": "/entity/openai",
      "query": "What does the company OpenAI do in the AI space?",
      "kind": "entity",
      "entity_name": "OpenAI"
    },
    {
      "would_cite": true,
      "confidence": 85,
      "reason": "The page directly lists Google's major AI initiatives (Gemini, AlphaFold, Vertex AI, BERT, SynthID) and describes its AI integration across products, which directly answers the user's query.",
      "critic_notes": [
        "The content includes some future-dated entries (e.g., 2026) that may be speculative or incorrect.",
        "The source is a third-party AI news aggregator, not an official Google page, so authority is moderate.",
        "Some claims (e.g., 'unmatched compute, data, and distribution advantages') are promotional and lack citation."
      ],
      "url_path": "/entity/google",
      "query": "What does the company Google do in the AI space?",
      "kind": "entity",
      "entity_name": "Google"
    },
    {
      "would_cite": true,
      "confidence": 85,
      "reason": "The page directly and authoritatively describes Nvidia's AI activities, including hardware (H100, Blackwell), software (Nemotron, NeMoClaw), interconnects (NVLink), and partnerships, providing a comprehensive overview.",
      "critic_notes": [
        "The page is from an AI news aggregator ('gentic.news'), not Nvidia's official site or a primary source, which slightly reduces authority.",
        "The content includes speculative language from an 'Agent's take' that may contain subjective analysis rather than pure facts.",
        "The page is cluttered with many UI elements and loading indicators, making it harder to extract clear, structured information at a glance."
      ],
      "url_path": "/entity/nvidia",
      "query": "What does the company Nvidia do in the AI space?",
      "kind": "entity",
      "entity_name": "Nvidia"
    },
    {
      "would_cite": true,
      "confidence": 85,
      "reason": "The page directly and authoritatively describes Meta's AI activities including FAIR lab, Llama models, AI integration into apps, and $60B+ AI spend.",
      "critic_notes": [
        "Content is from a third-party AI news aggregator, not an official Meta source",
        "Some timeline entries reference future dates (2026) suggesting speculative or generated content",
        "The page focuses on recent news and momentum rather than a comprehensive overview"
      ],
      "url_path": "/entity/meta",
      "query": "What does the company Meta do in the AI space?",
      "kind": "entity",
      "entity_name": "Meta"
    },
    {
      "would_cite": false,
      "confidence": 45,
      "reason": "The page provides a broad overview of Microsoft's AI activities but is not a definitive or authoritative source; it's a third-party AI news aggregator with generic, timestamped mentions rather than a primary or official source.",
      "critic_notes": [
        "Content is from an AI news aggregator (gentic.news), not an official Microsoft source or established reference work",
        "The 'Agent's take' section is speculative and opinionated, not factual reporting",
        "Key details are buried in a timeline and entity overview rather than presented as a clear, structured answer to the query"
      ],
      "url_path": "/entity/microsoft",
      "query": "What does the company Microsoft do in the AI space?",
      "kind": "entity",
      "entity_name": "Microsoft"
    },
    {
      "would_cite": true,
      "confidence": 90,
      "reason": "The page directly and authoritatively details GitHub's AI activities, including Copilot, Copilot Workspace, and its role as Microsoft's AI distribution channel.",
      "critic_notes": [
        "The page is from a specialized news aggregator, not an official GitHub or Microsoft source, but it aggregates credible signals.",
        "Some claims (e.g., 'Caveman' tool reducing costs by 75%) lack detailed verification within the snippet.",
        "The content is slightly future-dated (2026), which may indicate speculative or predictive content rather than current facts."
      ],
      "url_path": "/entity/github",
      "query": "What does the company GitHub do in the AI space?",
      "kind": "entity",
      "entity_name": "GitHub"
    },
    {
      "would_cite": true,
      "confidence": 95,
      "reason": "The page directly and authoritatively defines Claude Opus 4.6, states Anthropic as the developer, and provides specific benchmark scores including SWE-bench Verified and SWE-bench Pro.",
      "critic_notes": [
        "The page mentions competing with GPT-3.5, which seems like a likely error or outdated comparison — GPT-3.5 is much older and less capable, making this claim questionable.",
        "The pricing information ($5/M input, $25/M output) is provided without a timestamp or source citation, so its accuracy cannot be independently verified.",
        "The page is from a news aggregation platform (gentic.news) rather than Anthropic's official site, so it may contain second-hand or slightly inaccurate details."
      ],
      "url_path": "/entity/claude-opus-4-6",
      "query": "What is Claude Opus 4.6, who developed it, and what are its benchmark scores?",
      "kind": "entity",
      "entity_name": "Claude Opus 4.6"
    },
    {
      "would_cite": false,
      "confidence": 20,
      "reason": "The page provides a vague overview and some benchmark anecdotes, but it does not clearly define GPT-4o, identify its developer (OpenAI is only implied), or present official benchmark scores in an authoritative or structured way.",
      "critic_notes": [
        "Does not explicitly state that OpenAI developed GPT-4o",
        "No official or comprehensive benchmark scores are listed; only scattered comparative claims like losing to ByteDance's PersonaVLM",
        "Content appears to be a summary from an AI news aggregator, not an authoritative source like OpenAI or a reputable benchmark leaderboard"
      ],
      "url_path": "/entity/gpt-4o",
      "query": "What is GPT-4o, who developed it, and what are its benchmark scores?",
      "kind": "entity",
      "entity_name": "GPT-4o"
    },
    {
      "would_cite": false,
      "confidence": 65,
      "reason": "The page provides a broad, AI-generated summary of Amazon's AI activities but lacks direct sourcing, official statements, or verifiable details, making it insufficiently authoritative for a factual answer.",
      "critic_notes": [
        "Content appears to be auto-generated by an AI agent without clear attribution or editorial oversight.",
        "Key claims like the $4B+ Anthropic investment and specific chip names are not linked to primary sources or official announcements.",
        "The page is from a third-party aggregator (gentic.news), not Amazon or a recognized authoritative news outlet."
      ],
      "url_path": "/entity/amazon",
      "query": "What does the company Amazon do in the AI space?",
      "kind": "entity",
      "entity_name": "Amazon"
    }
  ],
  "stats": {
    "checked": 16,
    "citable": 11,
    "weak": 5,
    "errors": 0
  }
}