Lexifina's Git for US Legislation traces every statute line to the law that wrote it

Version control for federal law

Lexifina has released a project it calls Git for US Legislation, which gives the United States Code the kind of line-by-line provenance that programmers take for granted in version control. In a blog post that reached Hacker News' front page, the team frames the gap plainly: a developer can point at any line of code and learn who wrote it and when, while a lawyer reading a federal statute usually has to dig through amendment notes written in prose. The tool lets a reader pick a date and see the text in force on that day, with wording that reads differently today marked, and each phrase attributed to the Act that wrote it. It also tracks defined terms, so changing a definition lists every other provision that relies on it, and a section can be followed to see what has changed since the reader's last visit.

Why statutory lineage is hard

The Office of the Law Revision Counsel publishes the code as a series of release points, each current through a named public law, and keeps reorganising it; a bill to enact title 41 as positive law was introduced in September 2025. Which public law wrote a sentence is recorded in amendment notes in prose, when a change took effect often sits in a separate effective-date note, and where Congress said nothing the default rule is the date of enactment. Outside the titles enacted as positive law, the code is only prima facie evidence of the law, and the Statutes at Large govern where the two differ.

Lexifina illustrates the stakes with the small business debtor definition in 11 USC § 1182(1). Created in 2019, its debt limit was raised to $7.5 million by the CARES Act in March 2020 as a temporary measure, extended to March 2022, lapsed on 27 March 2022, then restored on 21 June 2022 for two more years, reaching back to cases commenced on or after 27 March 2020 that were still pending. The higher limit expired again in June 2024, and after a triennial adjustment in April 2025 it stands at $3,424,000. For a case filed in May 2022, the release point printed at the time gives the lower limit, and only a later Act changes the answer.

Building a checkable history

For each pair of release points from 2017 onward, the team took the public laws enacted in between, applied their amending instructions to the older text in date order, and compared the result with the official newer version. Of 365 changes across nine titles, 361 reproduced the next release point exactly; the remaining four are labelled with how far they could be confirmed. Each change carries the law and its effective date: 261 took effect on enactment by default, 72 through an effective-date note and 29 through a date written into the amendment itself. One change depends on an agency determination that may never come, and the team left it out of the text rather than guess.

The language-model motivation

The post argues the project matters because much first-pass legal research now runs through language models, which often cannot use a law's lineage when interpreting it. A 2026 study from the Technical University of Munich found two failure modes in models answering questions on German statutes: they apply rules superseded after their training cutoff, and they prefer newer provisions even when an older one governs the facts. One model answered 24% of post-cutoff questions correctly on its own and 87% when the relevant statutory text was retrieved for it, and the authors conclude that temporal validity must be treated as a hard constraint.

Asking a model to reconstruct old text is no fix, either. In related Munich work, a model applying German amending laws to earlier versions produced output 93–99% similar to the official text, yet only about half of single amendments and a fifth of amendment chains matched exactly, and experts found legally significant errors in the near misses. The history, Lexifina concludes, has to be checkable.

Known limits

The tool does not yet show a later law's reach back onto the earlier dates it covers, so for a case filed in the 2022 gap it shows the lower limit without warning that the June 2022 Act changed the answer. Effective-date notes are read by pattern, and unusual wording falls back to the date of enactment. Every change links to the law's own words, and the team advises reading the original for anything that matters. The right form factor, they add, is still being worked out.

Why it matters

Temporal validity is now a documented, quantified failure mode of AI-assisted legal research, and this project supplies what the Munich researchers found missing: authoritative as-of-date statutory text that a model or a lawyer can be grounded in, instead of relying on model recall. The replay-and-diff methodology, with its mismatch rate published rather than hidden, also gives civic-tech developers a template for building verifiable histories of official documents. For practitioners, it turns footnote archaeology into a lookup, at least within the nine titles and the 2017-onward window covered so far.