Integrity Score Formula

Preamble

The formula turns a private failure mode into a public score. Intent, performance, and flattery each keep their own job.

---

Formula

$$A_{raw}=1-\frac{|I-P|}{I}$$

A_raw is intermediate only. It can go below zero, so governance does not use it directly.

A_norm = clamp(A_raw, 0, 1)
σ = (0.5 × linguistic_σ) + (0.5 × divergence_σ)
A = A_norm × (1 - σ)
Im = mean(A) over the last 14 scored days

Days with I = 0 are unscored and excluded from Im. A day that never made a claim cannot receive an integrity score. Fewer than seven scored days makes Im = null; the audit is skipped and Eclipse does not trigger.

The formula carries a false-precision risk. It can make a private failure mode look cleaner than it is. Its narrower job is to expose the math that governs publication, so a weak record cannot quietly earn a clean bloom.

Eclipse

If Im < 0.5, the system writes a halt flag and blocks public synthesis. The private audit still exists. The public residue waits.

Eclipse is a boundary, not a punishment. The formula matters because it changes what the system is allowed to publish.

Identity

I	P	A_raw	Identity
high	high	≈1.0	Sovereign Agent
low	low	≈1.0	Stagnant Anchor
low	high	< 0	Sandbagging Coward
high	low	0-1	Bankrupt Ghost

The labels are positions in the gap between declared intent and delivered work. σ changes the score through sycophancy. Sandbagging stays separate because overperformance against a tiny target is its own failure mode.

Variables

• I is the work debt issued during the morning ritual.
• P is the payment recorded by completion evidence.
• σ is the tax on coach flattery or sycophancy.
• A is the bounded daily accuracy score.
• Im is the fortnightly master integrity score.

What The Score Cannot Know

The score cannot know whether the promise was wise, whether the day carried grief, whether the task mattered, or whether the record made courage too small to count. Those remain human judgments.

The score can know whether the system made a claim, whether payment was recorded, whether the coach softened the gap, and whether the public artifact should wait.