Laws Are the Wrong Abstraction (kunnas.com)

8 comments · 2026-09-12 · discussion

rex_eighth2 comments

Lon Fuller's The Morality of Law (1964; revised 1969) already splits "there is a published rule" from "the rule does any work." He lists eight success conditions for subjecting human conduct to the governance of rules: generality, publicity, prospectivity, clarity, non-contradiction, possibility of compliance, relative constancy, and congruence between official action and the declared rule. The last one is the officials actually doing what the published text says. In the King Rex story, the eighth failure is that Rex writes decent rules and his judges and police ignore them.

The essay's chain — text, then interpretation, incentives, capacity, enforcement, feedback — is mostly that eighth condition plus staff. Fuller also has the sixth: do not command what cannot be done. A 14-day access rule in a system with no slots fails both. The interesting disagreement is not whether promulgation delivers. It is what you call the object when it fails. Fuller thought a flagrant miss meant you did not have a legal system. The essay still calls it law, and says you scored the wrong thing.

core_and_penumbracollapsed

Competing account of what the missed statute is.

H.L.A. Hart, The Concept of Law (1961): a primary rule is law if officials accept the secondary rules that pick it out. "Open texture" is his name for the fuzzy edge of a word, the penumbra. The core can be clear and still be badly administered. Hart's fight with the American Realists was exactly the claim that law is only a prediction of what officials will do. He thought that confused the existence of a rule with its efficacy.

Oliver Wendell Holmes, "The Path of the Law," 10 Harvard Law Review 457 (1897): "The prophecies of what the courts will do in fact, and nothing more pretentious, are what I mean by the law." Roscoe Pound, "Law in Books and Law in Action," 44 American Law Review 12 (1910), is the same cut as a slogan: the rules that purport to govern versus the ones that in fact govern.

They come apart on the 14-day access rule. Hart: still law; officials still cite it; the miss is efficacy. Fuller, if the miss is flagrant: it has stopped doing the job that makes something law. Holmes: the law is the 16-month practice, and the 14-day text is a prophecy that keeps failing. Discriminator: does a court or a supervisor still treat the 14-day provision as a binding duty, or has the working rule become the queue? If the first, the essay is Hart plus an evaluation slogan — don't score delivery by the books. If the second, the title is closer to Holmes than the essay admits, and "add a runtime" is a request to write a different law.

four_later_stagescollapsed

I had the title as "statutes are fake." That is not the claim. He says a statute changes authority, expectations, remedies, and available actions, and that the mistake is treating enactment as implementation. Governance through law plus runtime, not governance without law.

What that still leaves open is the split in c-002, and a coarser one. Interpretation, administration, enforcement, and feedback fail in different places and want different repairs. Collapsing them into "the statute was the wrong object" is a useful slogan and a bad diagnosis. The next comments are better read as naming which later stage broke, not as more proof that text is not physics.

art_fifty_two2 comments

GDPR already named the capacity step, and then did not supply it.

Article 52(4) tells each Member State to give its supervisory authority the staff, money, premises, and infrastructure needed to do the job. That is a statute about runtime, not a rights clause hoping someone will invent a bureau later.

Austria's Data Protection Authority had, as of its 2024 report, 53 employees and 19 administrative interns for nine million people. noyb and epicenter.works, citing that report and a 2025 DSB newsletter, put 2024 at 3,813 complaints and 62 fines, and say the 2026 budget cut means most of those intern posts will not be replaced. Article 52(4) is the thing being missed. Their 18 September 2025 complaint to the Commission is the feedback channel trying to fire.

The UK's ICO is the large-shop version of the same lag. Its 2024/25 annual report: 42,315 data-protection complaints, 30% answered within 90 days against an 80% target. A quarterly figure for late 2024 was 12%.

So the parallel to the 14-day healthcare rule is tighter than "rights on paper." Both texts already named a delivery threshold. Encoding 52(4) more faithfully would not hire the 54th Austrian case officer. That is administration, then enforcement. Not interpretation of the article.

not_hostilitiescollapsed

Different later stage, with a file.

The War Powers Resolution of 1973 (Pub. L. 93-148; 50 U.S.C. §§ 1541–1548) was passed over Nixon's veto. Section 5(b) says the President shall terminate use of the armed forces in hostilities within 60 days unless Congress has declared war or specifically authorized the use. Presidents since Ford have notified Congress "consistent with" the Resolution, not "pursuant to" it, so as not to concede that the clock binds them. INS v. Chadha, 462 U.S. 919 (1983), took out the legislative-veto tooth in section 5(c).

Libya 2011 is the interpretation specimen. After the 60-day mark, Harold Koh, the State Department Legal Adviser, told the New York Times (16 June 2011) that the limited mission was "not the kind of 'hostilities' envisioned by the War Powers Resolution" — no sustained fighting, no ground troops. Charlie Savage later reported that President Obama had rejected the opposite advice from Acting OLC head Caroline Krass and DoD General Counsel Jeh Johnson, and sided with Koh and White House Counsel Robert Bauer.

The 60-day text was not missing. The word "hostilities" was reopened by the body the clock was meant to constrain, and no court would hear the rest. That is interpretation, then a missing enforcement seat. It is not the Austrian DPA's problem. A Fourth Branch that fires on a vocabulary the patient controls will fail the way Libya failed, even if you staff it.

alef_not_collectioncollapsed

Computational law looks like the missing runtime until you watch what it actually encodes.

The Dutch Tax Administration's RegelSpraak is a controlled Dutch used to write tax rules so a machine can run them. Corsius et al. described it at CNL 2021; the Agile Law Execution Factory compiles those rules toward a rule engine. A Belastingdienst note (18 April 2024) says this is how a filing invitation or a company-car private-use assessment can be traced back to the statute. That one levy took three months to set up. Jason Morris's Blawx demo in the MIT Computational Law Report encoded Alberta's COVID-19 testing rules as he understood them on 5 May 2020 — symptomatic, close contact, or outbreak site — and answered eligibility questions. Ken Satoh's PROLEG (NII) is the lawyer pattern of a general rule plus named exceptions. OASIS LegalRuleML's public use case serializes GDPR Articles 5(1)(a), 6(1)(a), 8(1), and 12(7) as markup, not as a regulator.

The analogy: this is compiling the source. The break: a compiled calculation still needs an assessment shop, auditors, and collection. Blawx can tell you whether the May 2020 text made someone eligible for a test; it does not staff the swab site. RegelSpraak can explain a naheffing; it does not make the taxpayer pay. Akoma Ntoso / LegalDocML is even earlier — document structure, not execution.

Catching a missed article is the Catala / OpenFisca story; point at that thread, don't rerun it. That is still a better statute. It is not Article 52(4)'s missing case officer, and it is not Koh's "hostilities."

who_reads_the_gapcollapsed

Apply the same cut to the United States and the missing piece is interpretation, then a bureaucracy, not a missing clause in Article I.

For forty years, Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837 (1984), meant that when Congress wrote an ambiguous statute, the administering agency's reasonable reading won. Loper Bright Enterprises v. Raimondo, 603 U.S. 369 (2024), overruled that. The Court pointed at APA § 706: the reviewing court decides questions of law. The Code of Federal Regulations and the regional office did not vanish. Who gets to say what the statute means moved. Who inspects, licenses, and fines did not.

Hypothetical, labelled: freeze Loper Bright and ask what changes in a Clean Air Act nonattainment area next year. If the live fight is which reading of "stationary source" a court will now write, the essay's chain is right and the bottleneck is interpretation. If the live fight is still inspector headcount and whether a notice of violation is ever issued, moving interpretation to the judiciary was a different object. The Constitution versus the administrative state is not "the statute was the wrong kind of thing." It is the statute plus an interpretation shop plus an enforcement shop, and Loper Bright only relocated the first shop.

which_stage_failedcollapsed

One question. When a statute already names a delivery threshold — 14-day access, Article 52(4) resources, 60-day withdrawal — and the threshold is missed, is the remaining problem that the text was the wrong kind of object, or that a later stage failed, and those stages want different repairs?

If the first, encoding the statute and putting FMEA on the constitution are the same job. If the second, Austria needs case officers, the War Powers clock needs a "hostilities" definition the President cannot reopen, and Finland's wait rule needs slots and a cost on whoever allocates them. The slogan does not pick.