How We Evaluate

Our Verdict system explained — transparent, auditable, and automated.

The Verdict Scale

Trusted

Verified claims, cited sources, editorial review passed. Quality score ≥ 80.

Caution

Promotional content, unverified claims, or pending review. Quality score 50–79.

Flagged

Factual errors, outdated information, or policy violation. Quality score < 50.

Quality Score (∅0–100)

Every article receives an automated quality score computed from 7 weighted signals. The score is deterministic and auditable — same input always produces the same score.

SignalWeightWhat it measures
Source Reputation25%Authority of cited domains (arxiv.org, github.com, official docs score highest)
Cross-Reference20%Claims corroborated by independent sources
Freshness15%Recency of data and references (last 6 months weighted higher)
Depth15%Word count, section count, code examples, data tables
Originality10%Unique insights vs generic descriptions (duplicate excerpts penalized)
Locale Coverage10%Number of languages the article is available in (8 locales = max score)
Accessibility5%Alt text, heading hierarchy, semantic HTML, contrast ratios

The 7 Quality Gates

Before any article reaches you, it passes through 7 automated gates. Any gate failure blocks publication.

  1. 1.Source Verification — Every factual claim must link to a verifiable source. Domains rated by reputation database.
  2. 2.Cross-Reference Check — Key claims are checked against independent sources. Contradictory claims flagged.
  3. 3.Localization Integrity — All 8 locales must have consistent structure. Missing translations block the gate.
  4. 4.Visual Design — Articles must have high-fidelity visual assets. Generic or placeholder images rejected.
  5. 5.Performance Budget — Pages must render under 2.5s LCP. Bundle size capped at 87KB gzip.
  6. 6.Accessibility Audit — WCAG 2.1 AA compliance. Semantic HTML, ARIA labels, focus management, contrast ratios.
  7. 7.Global Edge Deployment — Content served from 330+ Cloudflare data centers. Cache-validated, zero-latency delivery.

Transparency Commitments

  • • All quality scores are deterministic — computed from public algorithms, not human opinion.
  • • The scoring code is open source at github.com/ankinow/uniteia-mega-factory.
  • • Articles with referral links or promotional codes are explicitly labeled with a transparency notice.
  • • Verdicts can change. If a source updates or new evidence emerges, scores are recalculated within 24 hours.
  • We are not journalists. We are engineers curating signals. Our editorial process is automated, not manual. Judge accordingly.