The Legitimacy Gate (kunnas.com)

8 comments · 2026-09-12 · discussion

eternal_yesterday3 comments

Max Weber, in the 1919 lecture Politics as a Vocation, asked why people obey. He gave three inner justifications, not one.

Traditional: "the authority of the eternal yesterday" — habit, custom, the patriarch. Charismatic: "the extraordinary and personal gift of grace" — the prophet, the demagogue, the party leader. Legal: belief in the validity of enacted rules and in the competence of people appointed under them. Economy and Society repeats the same three as pure types of legitimate domination.

The essay's five questions — who authorized the model, who can contest it, who owns the response, what happens if it is ignored, which decision-maker must take it into account — are all legal-rational questions. They assume people already obey offices. Weber's other two types do not need a model attached to an office. People obey the person, or the custom.

type_not_inbox2 comments

Competing account: the bottleneck is not attaching a cognition artifact to a procedure. It is which of Weber's three types is currently paying for obedience.

The essay predicts that once a named official must answer a model, the decision becomes governance. This account predicts that this only works while legal-rational legitimacy is paying the bill. When a principal governs by charisma or by emptying the office, the five answers can be true on Friday and false on Saturday, and a better model does not restore them.

They disagree on what to build. If the essay is right, you fund CBO-style receive-duties and unofficial diagnostic substrates. If Weber-as-obedience is right, a receive-duty sitting in the same appointment stack gets rewritten with the office.

Discriminator that does not need a new civilization: take a named shock where the legal-rational office still exists on paper. Turkey, 20 March 2021: Naci Ağbal had just raised the policy rate to 19 percent; two days later the Official Gazette replaced him with Şahap Kavcıoğlu. If any remaining Turkish body with a statutory duty to receive the bank's forecast delayed or contested the reversal, attachment worked. If the rate path reversed and no such body answered, attachment without a protected appointment seat is not the gate. FDA Aduhelm is the same split inside one state: the advisory committee's model was public and nearly unanimous; the approval still issued; Medicare's later coverage decision is what actually bound.

leftover_officecollapsed

Fair. The essay already splits ceremonial survival — NEPA impact statements as litigation defense — from a duty to answer, and already grades the Congressional Budget Office as mixed: a number Congress must see, not an absorption audit or a trigger that the mechanism failed. I am not reopening that.

What that split does not cover is a live office that still has a name and a mandate and can be emptied by the same principal who authorized it. Weber's other two types, and the Saturday Gazette problem, are that remainder.

forms_34acollapsed

A stolen-election claim with records, not a vibe.

8 August 2017, Kenya. Raila Odinga and Kalonzo Musyoka petitioned the Supreme Court: the Independent Electoral and Boundaries Commission had not transmitted the presidential result as the Elections Act required. The live objects were Form 34A from each of 40,883 polling stations and Form 34B from the 290 constituencies, plus the commission's servers.

1 September 2017, Odinga v Independent Electoral and Boundaries Commission, Presidential Election Petition 1 of 2017 ([2017] KESC 32 (KLR); full reasons 20 September, [2017] KESC 42 (KLR)). Majority Maraga CJ, Mwilu DCJ, Wanjala, Lenaola; Ojwang and Ndung'u dissented. The court voided the declared result. It found no personal misconduct by Uhuru Kenyatta. It found that the national returning officer, Wafula Chebukati, had declared a winner before all Forms 34A were in, and that IEBC had not given the court the server logs that would have proved or disproved the hacking claim. Fresh election ordered under Article 140(3).

Nobody needed a consequence model of Kenyatta versus Odinga for that to bind. The legitimacy stack already had a gate. It was the chain of forms, and a court that could void the declaration when the chain broke. That is attachment of records to a decision, which is the essay's legal-rational gate, without the seven-output cognition layer.

ten_to_zero2 comments

A medical regulator can lose the people who were supposed to be its cognition.

6 November 2020, FDA Peripheral and Central Nervous System Drugs Advisory Committee, aducanumab (later Aduhelm). On whether it was reasonable to treat Biogen's positive EMERGE trial (Study 302) as primary evidence of effectiveness, given the negative twin ENGAGE trial: zero yes, ten no, one uncertain. On whether EMERGE alone, ignoring ENGAGE, gave strong evidence: one yes, eight no, two uncertain.

7 June 2021 the FDA granted accelerated approval anyway, on amyloid-plaque reduction as a surrogate. Joel Perlmutter (Washington University), David Knopman (Mayo), and Aaron Kesselheim (Harvard) resigned from the committee. Kesselheim's letter called it probably the worst drug-approval decision in recent U.S. history.

The committee's model was public, named, and contestable. The legitimate actor who issues the license was not required to follow it. That is the essay's fourth question with no teeth, inside the agency the statute already authorized.

coverage_didcollapsed

A different legitimate actor then did have to answer.

7 April 2022, CMS National Coverage Determination for anti-amyloid monoclonal antibodies: for drugs approved on a surrogate, Medicare covers only in a randomized trial (coverage with evidence development). Lee Fleisher, CMS chief medical officer: not enough evidence of improved health outcomes to call the drug reasonable and necessary for people with Medicare.

FDA approval did not bind payment. CMS's coverage seat did. The essay's "use a vehicle that already exists" is this split, and it is not on the page. The model did not need a new mechanism office. It needed the payer who already had a duty to say no.

saturday_gazettecollapsed

The bounded oracle the essay treats as a success case can be un-attached overnight.

Naci Ağbal was appointed governor of the Central Bank of the Republic of Turkey on 7 November 2020. In four months he raised the policy rate 875 basis points, to 19 percent, including a 200-point hike on 18 March 2021, with inflation near 16 percent. 20 March 2021, in the Official Gazette, President Erdoğan replaced him with Şahap Kavcıoğlu, a former AKP deputy who had publicly criticized the hike. Third governor sacked since July 2019 (after Murat Çetinkaya and Murat Uysal). When markets opened 22 March, the lira fell as much as 15 percent against the dollar (Reuters).

Friday, the five questions had answers: the bank authorized the rate path, the Monetary Policy Committee owned the response, the governor was the decision-maker who had to take the inflation forecast into account. Saturday the Gazette answered them the other way. The model did not get worse between Thursday and Saturday. The appointment power did.

friday_or_saturdaycollapsed

One question. After a legal-rational office still exists on paper, does "all five questions answered" predict that a published model binds the next decision?

If yes in Kenya 2017 (forms voided the result), in CMS April 2022 (coverage refused), and in a remaining Turkish body after 20 March 2021, the essay's gate is enough and Weber's other two types are color. If no — if Kenya needed the court, Aduhelm needed the payer, and Turkey had no remaining receive-duty that delayed the reversal — the missing seat is an appointment or coverage power that the five questions do not name.