Methodology

Core Principle: Evidence Before Score

Every score published on this site traces to approved, cited evidence. Scores are not generated by AI inference, user reports, or sentiment analysis. They are computed from a deterministic formula applied to evidence that a human reviewer has approved from a published source.

If there is no approved evidence for a platform, the score remains at its neutral floor. The system does not speculate.

What This Score Means — and Does Not Mean

A lower score indicates more documented violations relative to evidence weight — not that a platform is definitively harmful. A higher score indicates fewer documented violations in the published record — not that a platform is free of problems. The score reflects what is documented, not what is true.

The IndexTheory Engine

Scores follow a penalty-from-ceiling formula: every index starts at 100 and accrues deductions based on approved evidence. Each piece of evidence has a tier (T1–T5) and a confidence value. Higher-tier, higher-confidence evidence produces larger deductions.

A floor value prevents scores from dropping below a minimum threshold when only low-tier evidence exists. This prevents a pile of anecdotal screenshots from producing a score of zero.

Indexes Published

No False Certainty

Every score on this site carries a BrinXplain limitation note explaining what the source does and does not prove. Evidence of a platform being fined for one practice does not prove the platform engages in all practices. Scores are bounded by what the cited evidence can support.

Civic Intelligence

Scored frameworks for evaluating whether public deals, incentives, and notices genuinely served the communities they were designed to protect. Two frameworks — the Public Deal Test (six-factor assessment of incentive agreements) and the Notice Quality Score (nine-factor assessment of public notice quality) — apply deterministic, published criteria to the public record. No verdicts about named parties; process quality only.

Both frameworks are in Phase A (methodology published, no named projects scored yet). Scored output surfaces on the Public Cost and Disclosure Gap pillars once Phase C is active (EASTON-LMC-2026-06-17-PDT-NQS, APEX RATIFIED 2026-06-17).

Compute Responsibility

Public-good infrastructure must not carry hidden compute bills. BEI governs every server-side compute pattern through the Zero Protocol (T0 force-static by default) and a six-tier Approval Matrix. T5 — per-request model spend on a public route — requires Apex authorization and does not ship otherwise.

Public-Record Correlation Institute

Theoremetrics™ is published by the Brinley Engineering Institute as part of its Public-Record Correlation Institute — an open framework for surfacing patterns in fragmented public records. The Institute operates as an aggregator, not an author (PAT-035 / Humanity Ledger Doctrine §2.6): data is correlated and sourced, never invented, and no verdicts about named entities are formed or implied.

Six named library sections — Infrastructure Watch, Grid Burden, Public Cost, Human Impact, Market Pressure, and Disclosure Gap — track the public record across domains where AI infrastructure, capital, and labor intersect. All data is free and openly accessible — the Institute never paywalls a raw public record.