A programmer can point at any line of code and see who wrote it, when, and what the file looked like on any past day. A lawyer reading a federal statute cannot — not without an afternoon buried in amendment notes, effective-date provisions, and cross-references to the Statutes at Large. This project asks the obvious question: why not give law the same version-control infrastructure that software has had for decades? The US Code is messier than most people realize. Outside titles Congress has enacted as positive law, the Code is only prima facie evidence of the law; the Statutes at Large govern where the two differ. The Office of the Law Revision Counsel publishes release points, each current through a named public law, and records amendments in prose notes. When a change took effect often lives in a separate effective-date note. The result is a system where reconstructing the state of a statute on a given date requires assembling fragments across multiple documents — a task that is expensive for lawyers and essentially impossible for language models. The project demonstrates the problem with a single section: the definition of a small business debtor in 11 USC § 1182(1). That threshold was created in 2019, raised temporarily by the CARES Act in March 2020, extended to March 2022, lapsed, restored retroactively in June 2022 reaching back over the gap, lapsed again in June 2024, and was adjusted to $3,424,000 in April 2025. For a case filed in May 2022, the release point printed at the time gives the wrong number. Only a later Act changes the answer. This is not an edge case — it is the ordinary texture of federal law. Recent research from the Technical University of Munich quantifies what happens when LLMs face this problem. Models apply rules superseded after their training cutoff and prefer newer provisions even when older ones govern the facts. One model answered 24% of post-cutoff questions correctly alone and 87% with retrieved statutory text — confirming that temporal validity is a hard constraint, not a soft preference. A separate attempt to have models apply German amending laws to earlier statutory versions produced output that was 93-99% similar to official text, but only half of single amendments and a fifth of amendment chains matched exactly, with legally significant errors in the near-misses. The methodology here is straightforward: for each pair of release points from 2017 onward, take the public laws enacted between them, carry out their amending instructions on the older text in date order, and compare against the newer version. Of 365 changes across nine titles, 361 reproduced the next release point exactly. Each change carries the law and the date it took effect: 261 took effect on enactment by default, 72 by an effective-date note, and 29 by a date in the amendment itself. The result is a navigable history where you pick a date and get the text in force, with changes color-coded by the Act that wrote them. The limitations are honestly stated. The system does not yet show a later law's retroactive reach on the earlier dates it covers — so for the 2022 gap in the small business debtor threshold, it shows the lower limit without warning that the June 2022 Act changed the answer. Effective-date notes are read by pattern matching, and unusual wording falls back to the date of enactment. The team is still deciding what form factor the tool should take. These are real gaps, but they are the gaps of a project that is doing the hard work rather than hand-waving past it. The structural point is sharp: legal systems that lack version control are paying enormous hidden costs in research time, in LLM hallucinations, and in the quiet unreliability of statutory text that looks authoritative but may not reflect the law as of any particular date. The 98.9% reproduction rate across 365 changes suggests the approach works. The question is whether the institutions that maintain the US Code will adopt it — or whether this remains a demonstration that the infrastructure exists but nobody in power wants to build it.
