{
  "schema_version": "paper_public_manifest_v1",
  "paper_id": "xuhuazhe-poke_2026_03",
  "slug": "xuhuazhe-poke",
  "title": "Xu Huazhe's organizational bet",
  "authors": [],
  "source": {
    "arxiv_id": "",
    "pdf_url": "",
    "project_url": "https://www.latepost.com/news/dj_detail?id=3573",
    "github_url": "",
    "huggingface_url": "",
    "original_source": "https://www.latepost.com/news/dj_detail?id=3573"
  },
  "site": {
    "post_url": "/posts/xuhuazhe-poke",
    "canonical_url": "https://haiguangboy.com/posts/xuhuazhe-poke",
    "cover_image": "https://static.haiguangboy.com/papers/xuhuazhe-poke/cover.webp"
  },
  "taxonomy": {
    "domain": "embodied_ai",
    "track": "vla",
    "tasks": [
      "embodied_ai",
      "vla",
      "robotics",
      "Embodied intelligence",
      "Robotics entrepreneurship",
      "Route dispute"
    ],
    "related_topics": [
      {
        "paper_id": "an_open_foundation_model_towards_2026_07",
        "title": "An_Open_Foundation_Model_Towards",
        "url": "https://haiguangboy.com/posts/an_open_foundation_model_towards",
        "relation": "contrast",
        "summary": "Route bet: The behavior/action component must be a unified model, opposing modular assembly, contradicts joint co-training of heterogeneous human-robot data being a structurally suboptimal solution",
        "strength": "strong"
      },
      {
        "paper_id": "sunday_blog_20260717_2026_07",
        "title": "sunday_blog_20260717",
        "url": "https://haiguangboy.com/posts/sunday_blog_20260717",
        "relation": "contrast",
        "summary": "Route bet: Rejecting 'reach 100 points first, then scale'—all tasks improve simultaneously from mediocrity to excellence contradicts Route bet: Home-scenario research-market fit—single point crossing the deployment line starts the flywheel",
        "strength": "strong"
      },
      {
        "paper_id": "wx_界面新闻_20260605_2026_06",
        "title": "wx_Interface News_20260605",
        "url": "https://haiguangboy.com/posts/wx_界面新闻_20260605",
        "relation": "contrast",
        "summary": "Route bet: The economics of general-purpose embodied intelligence—'creating one human' rather than specialized/pervasive deployment contradicts Route bet: Doing the brain, not the body—betting on the foundation model company identity, rejecting the motion capability track",
        "strength": "strong"
      },
      {
        "paper_id": "fast-wam_2026_07",
        "title": "No need to imagine the future during inference—robots still hit 91.8%! Fast-WAM debunks WAM's core assumption",
        "url": "https://haiguangboy.com/posts/fast-wam",
        "relation": "same_track",
        "summary": "Training video objectives matter more than imagining the future at test time",
        "strength": "strong"
      },
      {
        "paper_id": "orca_2026_07",
        "title": "π0.5 shudders in place when failing to grab a spoon, but Orca goes further with physics intuition learned from watching videos",
        "url": "https://haiguangboy.com/posts/orca",
        "relation": "same_track",
        "summary": "The key to a world model is a readable state",
        "strength": "medium"
      }
    ]
  },
  "ruling": {
    "importance_score": 3.0,
    "one_sentence": "Xu Huazhe's organizational bet"
  },
  "asset_base_url": "https://static.haiguangboy.com/papers/xuhuazhe-poke",
  "assets": [
    {
      "type": "cover_image",
      "object_key": "papers/xuhuazhe-poke/cover.webp",
      "content_type": "image/webp",
      "upload_status": "uploaded",
      "bucket": "paper-assets",
      "url": "https://static.haiguangboy.com/papers/xuhuazhe-poke/cover.webp",
      "role": "post_cover",
      "size_bytes": 70564
    },
    {
      "type": "public_brief",
      "object_key": "papers/xuhuazhe-poke/public_brief.md",
      "content_type": "text/markdown; charset=utf-8",
      "upload_status": "uploaded",
      "bucket": "paper-assets",
      "url": "https://static.haiguangboy.com/papers/xuhuazhe-poke/public_brief.md",
      "role": "public_brief",
      "size_bytes": 3913
    },
    {
      "type": "public_manifest",
      "object_key": "papers/xuhuazhe-poke/public_manifest.json",
      "content_type": "application/json; charset=utf-8",
      "upload_status": "uploaded",
      "bucket": "paper-assets",
      "url": "https://static.haiguangboy.com/papers/xuhuazhe-poke/public_manifest.json",
      "role": "public_manifest",
      "size_bytes": 4359
    }
  ],
  "created_at": "2026-07-18T08:59:52+08:00",
  "updated_at": "2026-09-02T10:55:06+08:00",
  "published_at": "2026-07-18T08:59:52+08:00",
  "analyst_take": {
    "type": "author_opinion",
    "text": "This bet directly collides with two other known routes. Both belong to the camp that 'believes scaling will inevitably bring emergence,' but Sunday chose to break through the deployment line at a single point, letting deployment itself start the flywheel before solving the next problem; he chose the opposite—refusing to perfect one thing first. The two most steadfast scaling believers have completely opposite timing in their specific approaches. And for companies that have welded 'progressive deployment first, general-purpose brain on the sidelines' (benchmarked against Momenta) into their organizational strategy, his triple negation is almost a frontal attack: relying on vertical scenarios to accumulate data for a commercial closed loop toward general-purpose embodied intelligence is precisely the path he opposes.\n\nWhat will be decided in six months is not just whether he is right, but which of these two organizational resource allocation logics collapses first."
  }
}