#8546.0TechT2 — media / chartCooling

Formalizing the US tax code (Using Lean)

Spotted 2026-11-15 03:07 (UTC+8) · original event 2026-11-10 · sources: Hacker News, Lexifina, Rubric · spreading on 3 platforms

Heat score

Rank score: 6040% — where it sits on the charts
Discussion score: 7040% — how much people are talking about it
Cross-platform score: 7020% — how many platforms it spread to
Composite heat score:46.0rank × 0.4 + discussion × 0.4 + cross-platform × 0.2 — includes a -20-point cooling decay (peak passed, attention fading).

Why it matters

Tax law is thousands of pages of cross-references no human can fully check: one wrong pointer can quietly change what you owe, and the mistakes sit there for years because nobody can read the whole thing at once.

The angle

AI plus a proof checker read the entire US tax code and found 138 real drafting mistakes: how formalizing law as code catches errors humans missSuggested format: Short video · for Developers, legal-tech followers, formal methods fans

Evidence chain

• HN #18, 85 pts/50 cmt (2026-11-15)• Lexifina blog 2026-11-10/11: translated the 2025 individual income tax (Title 26, US Code release current to 3 Nov 2026) into Lean with AI and ran a proof checker over every cross-reference - 257 findings, 138 real drafting faults, 98 intended, 21 false alarms in the editors' own hyperlinks• each fault carries a Lean certificate• model validated against PolicyEngine-US, Tax-Calculator and UsTaxes on 5,000 synthetic households (agreement with PolicyEngine on 4,993)• also found three small unpublished loopholes• Rubric 2026-11-12 claim-checked entry No. 21 of 106 (score 0.64), noting the AI-run triage has had no lawyer or CPA review and most figures are the author's own and unverified• 2026-11-15 04:07 update: HN #23, 102 pts/63 cmt• rank 75 -> 60• 2026-11-15 05:07 update: HN #28, 115 pts/76 cmt• score unchanged 66.0 (rank 60, discussion 70, cross 70)• 2026-11-15 06:07 update: dropped out of HN top 30• cooling -20 (66.0 -> 46.0)