Source Tiers T1–T5
How evidence sources are classified and how tier determines scoring weight.
Screenshot / Personal Account
Unverified first-person accounts, screenshots, social media posts, forum threads, and user-submitted reports without independent corroboration.
Examples
- ·Reddit post describing a suspicious job listing
- ·Screenshot of a profile flagged as fake
- ·LinkedIn comment claiming ghost jobs are common at a company
T1 evidence is accepted because pattern-of-account data has signal value at scale, even when individual items are unverified. T1 entries require a human reviewer to assess plausibility before approval.
Journalism / Industry Survey
Published journalism from identified news organizations, industry surveys, third-party analyst reports, and investigative pieces with named sources.
Examples
- ·Wall Street Journal investigation into job board ghost listings
- ·LinkedIn Workforce Report showing application rates
- ·Resume-writing industry survey on application-to-response rates
T2 sources are externally published and carry reputational accountability. The journalist or publisher may be wrong, but they have a public record attached to the claim.
Academic Research
Peer-reviewed research papers, preprints with institutional affiliation, university research reports, and studies with documented methodology.
Examples
- ·University study on bot-to-human follower ratios on professional networks
- ·ACM paper on coordinated inauthentic behavior detection
- ·Preprint analyzing scraping patterns across major job boards
T3 sources involve systematic investigation and documented methodology. The peer-review process provides a structural check, though it does not guarantee correctness.
Platform Transparency Report
Official disclosures by the platform itself — transparency reports, investor filings, press releases citing specific figures, and formal statements to regulators.
Examples
- ·LinkedIn Transparency Report citing fake account takedown counts
- ·Platform SEC 10-K acknowledging a significant percentage of accounts are inauthentic
- ·Official blog post disclosing the number of ghost jobs removed in a period
T4 evidence carries high evidentiary weight because the platform itself has admitted or disclosed the fact. However, platforms can selectively disclose, and T4 figures should be read as minimum bounds. Admissions made to avoid greater regulatory exposure may understate scope.
Regulatory / Legal
Government enforcement actions, court orders, FTC consent decrees, CFPB findings, state AG settlements, and similar legally binding documents involving the platform.
Examples
- ·FTC consent decree against a platform for deceptive job posting practices
- ·Court filing citing evidence of coordinated fake profile networks
- ·State AG settlement over misleading claims about job application outcomes
T5 evidence represents the highest evidentiary weight — an external, adversarial process with formal legal standards has produced a documented finding. These are the most difficult to dispute and receive the highest coefficient.
Tier assignment
Tier is assigned by the evidence ingestion operator when a claim is extracted from a source artifact. Tier upgrades (e.g., a T2 claim later corroborated by a T5 regulatory action) are recorded as separate evidence entries — the original entry retains its original tier. Both entries contribute independently to the score.