About LawVM
LawVM compiles legal state from amendment acts. It tells you what each provision says at any date, which act changed it, and where its answer is uncertain.
It does not prove: that the source corpus is complete, that the source faithfully reflects legislator intent, that any oracle is authoritative, or that broader doctrinal interpretation is correct. It proves text-state derivation from the sources it has, with explicit evidence for what was observed, inferred, recovered, and left unresolved.
What LawVM optimizes for
- Auditability over magic
- Typed structure over ad hoc string patches
- Evidence over hand-wavy correctness claims
- Bounded investigation over endless architecture expansion
Non-goals
- Pretending all jurisdiction frontends are equally mature
- Hiding source defects behind silent cleanup
- Glossy claims without evidence
- Presenting editorial consolidations as the only truth
Author
Elias Kunnas — kunnas.com
Open source
Get involved
Researchers: See Finland for empirical evidence and the verification ledger. Papers in preparation — email for collaboration or preprints.
Government agencies: See Finland for proof of concept. The architecture supports multiple jurisdictions with different authority models. Email to discuss a jurisdiction pilot.
Contributors: Clone the repo, run uv sync, try uv run lawvm --help. The architecture documentation is in notes/. Start with bounded tasks: one phase, one test, one adjudication classifier.
Contact: elias at kunnas.com
Legal note
LawVM is research and engineering infrastructure. It is not legal advice. The legal authority of consolidated texts varies by jurisdiction. Official legal effect depends on each jurisdiction's source regime.