Evidence behind the competence floor

Provisions that name the people who must recover, coded on seven questions

The competence floor is the current human capacity needed to complete non-deferrable fallback work within a recovery tolerance. This corpus is the evidence behind it: recovery and continuity regimes, AI-governance regimes, and a contrast panel of regimes known to requalify their people. Each provision is coded on whether it names human competence, defines it, states a level, tests it, puts a clock on it, ties it to a measurable, and treats it as something lost with disuse. Rows appear as coded, with the verification note beside them. Counts follow the codebook: rows graded watch and rows not yet coded are left out of every denominator. A pattern here is a description of the coded corpus, not a population estimate.

Seven questions, three panels

Two clocks

The codebook reads two things at instrument level. A machine clock is an explicit numeric system-recovery time in the instrument. A person clock is an explicit personnel cadence: annual, quarterly, n-year. An instrument that only tells the entity to set its own recovery objective has no machine clock, and a cadence written as regular or periodic is no person clock. Each dot is one instrument.

Stated recovery tolerances reducible to hours

The workbook reduces a tolerance to hours only where the wording allows it. Its rule: next business day is entered as 24 for P1-15 and P1-16; within the business day takes the instrument's own intraday figure, so P1-18 is 2; where an instrument states both a next-business-day and a shorter figure, the shorter one is held, so P1-14 is 2. The stated wording is always beside the number.

Year the operative text was adopted

The provisions

Select a row for its verbatim trigger text, qualifiers, coding rationale, source and verification note. Every column of the workbook is there.

Candidates not yet coded

Codebook