From Content-Addressed Storage to Governed Semantic Project Lineage

This paper examines whether the proposed Semantic Project Lineage System represents a genuinely new computing architecture or primarily recombines established mechanisms. The proposed system is a local-first, content-addressed, transtemporal substrate for storing, reconstructing, governing, and navigating evolving digital projects. It represents projects through exact content objects, persistent file-state lineages, version-scoped path maps, provenance and authority relationships, and reconstructible external projections such as ZIP archives and working directories. The system is designed in particular for human--AI project workflows. An AI agent may repeatedly emit a complete project artifact under the same visible filename. Before that artifact replaces the prior visible version, the substrate independently verifies continuity, identifies every predecessor-file disposition, refuses unexplained losses, commits newly admitted content, and preserves exact reconstruction of prior states. The review finds substantial prior art for nearly every primitive involved: content-addressed immutable storage; global chunk deduplication; snapshot-based version control; mutable objects formed from immutable claims; semantic filesystem indexing; data provenance; reproducible derivation; versioned data repositories; staged agent filesystem mutation; Git-like agent memory and rollback. However, the reviewed systems do not appear to combine these mechanisms into the same authority model and operating workflow. The strongest differentiation is not a new hashing, chunking, graph, or delta algorithm. It is an architectural unification: The ordinary filesystem, project archive, and agent workspace become projections of a governed semantic-temporal graph whose authoritative state includes exact content, file ancestry, project membership, provenance, authority, causality, and reconstructible external representations. The paper therefore concludes that the proposed system is best characterized as a plausibly novel systems synthesis built from established primitives. Its strongest research contribution would be a governed correspondence model connecting exact storage, semantic project history, safe agent publication, and ordinary filesystem usability. The novelty claim remains review-bounded: it describes the comparison set represented in this draft, not universal priority, patentability, or the absence of other systems. The paper also supplies a formal object model and proof obligations so that the synthesis can be evaluated as more than a feature list.