Skip to content

Bukti — Methodology & Standards

These docs describe how Bukti turns evidence into capability claims — the math, the taxonomy, and the data formats. They exist so that anyone consuming a Bukti score can audit how it was produced.

Status: uncalibrated. Threshold values cited throughout are defaults subject to revision when calibration data from the first pilot cohort exists. See calibration-status.

What's a Bukti score, in one paragraph

Every entity-capability pair (a person, agent, or organization paired with a capability node from the taxonomy) carries two orthogonal grades: an identity grade (I0–I4, how strongly we know who is making the claim) and a substantive grade (a Beta-Binomial posterior over evidence, expressing both a median competence estimate and a credible interval). The public tier — verified / attested / self-declared — is a derived label from the joint matrix of these two axes. See the two-axis model.

Read these first

  • Two-axis model — why "verified" and "score" are separated, and how the joint matrix maps to public tiers.
  • Scoring formula — the Beta-Binomial posterior, prior specification, and credible-interval extraction.
  • Evidence weights — how each evidence type (behavioral artifact, peer attestation, credential badge, etc.) contributes pseudo-observations.
  • Identity grades — the I0–I4 ladder, OIDC providers, and what each grade certifies.

Reference

Methodology v0 — dated 2026-04-28. Uncalibrated; see calibration status.