{"id":"2056437462423990593","url":"https://x.com/cdriclion/status/2056437462423990593","text":"Introducing Open Collider: an open-source engine that mechanically improves LLM creativity.\n\nIt generates non-trivial, high-quality ideas at scale, for any ideation problem.\n\nLLMs collapse on the same ideas. Sample the same brief 100 times → most outputs land in the same place. Researchers call it the Artificial Hivemind (Jiang et al., 2025).\n\n\"Be more creative\" moves the LLM's output by ~0.04 in embedding space.\nForcing structurally distant domain collisions moves it by ~0.28.\n\n7× more. Same model, same brief.\n\nSo I built Open Collider: a pipeline based on the theory of bisociations (Koestler 1964), the same model that drives human creativity.\n\n📊 Across 12 real-world ideation problems:\n•⁠ ⁠12/12 sign-test wins on embedding distance (p = .0002)\n•⁠ ⁠60%+ originality wins on 4,320 blind LLM-judge verdicts\n•⁠ ⁠4–13× further from the default cloud than \"be original\" prompts or longer context\n•⁠ ⁠Idea relevance holds (win rate >50% on overall quality)\n\n💻 Engine: first reply 👇\n📝 Launch study: pinned tweet\n\nTry it, Break it, Tell me what you find!","author":{"name":"Cédric Lion","username":"cdriclion","avatarUrl":"https://pbs.twimg.com/profile_images/2056469035487723520/lh9oWp2O_200x200.jpg"},"createdAt":"Mon May 18 18:11:12 +0000 2026","engagement":{"replies":26,"retweets":81,"likes":691,"views":233112},"media":{"photos":[{"url":"https://pbs.twimg.com/media/HInuczzXoAEzZz9.jpg?name=orig","width":966,"height":295},{"url":"https://pbs.twimg.com/media/HInuhHjXkAA4Rlb.jpg?name=orig","width":1654,"height":1218}],"videos":[]},"quoteTweet":{"id":"2056419767603298607","url":"https://x.com/cdriclion/status/2056419767603298607","text":"","author":{"name":"Cédric Lion","username":"cdriclion","avatarUrl":"https://pbs.twimg.com/profile_images/2056469035487723520/lh9oWp2O_200x200.jpg"},"createdAt":"Mon May 18 17:00:53 +0000 2026"}}