The writer asserts that Anthropic’s Scion is a deliberate, close copy of Mononoke rather than an independent or accidental resemblance. After initially offering a charitable read, private confirmations and overlapping personnel convinced them the similarity is intentional. The likeness goes far beyond the headline idea of “derived data” and extends to the entire data model and surrounding structures: both systems use a write-ahead-log style Merkle DAG where a changeset lists files changed relative to its parent and points to content-addressed blobs, with changesets created server-side and instantly available. The same kinds of derived indexes (e.g., blame) and a similar server-side query/mutation API appear, and even the names bear a suggestive relation (Scion/Bonsai).
The piece emphasizes that Mononoke’s Bonsai model was contrarian and technically novel when built - requiring new algorithms like server-side rebases - and has proven durable in production over many years. The complaint is not about reuse per se but about failing to credit the original work: vague nods to “Sapling” or “hard-won lessons” are inadequate given the deep duplication. That degrades Anthropic’s public claim that an LLM built and productionized the system in months, though reproducing it quickly is still technically notable. The writer calls for clearer attribution and recognition of the team who developed and maintained the original system.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.