Git is a file system. We need a database for the code (gist.github.com)

A prominent figure in software development has announced the need for a new version control system to replace Git, citing its increasing limitations as developers shift towards interacting with large language models (LLMs) instead of traditional coding. The speaker points out that Git’s structure poses challenges like handling monorepos, managing merges and rebases, and lacking effective code insight features, which hinder productivity. By transitioning from a file system mindset to a database paradigm for code, the goal is to implement formal and deterministic algorithms for version control, improving collaboration not just among developers but also with AI tools. The proposed system would utilize abstract syntax trees (ASTs) rather than blobs, allowing for sophisticated operations like overlay branches, a concept previously explored within the commutative replicated data type (CRDT) community. By addressing fundamental architectural issues in Git, such as data accretion and merge semantics, the new tool aims to enhance code organization and versioning. This shift is significant as it aligns with the evolving nature of software development, where understanding and collaboration with AI are becoming integral, suggesting that the tech community is ripe for a radical change in how code management and collaboration are approached.