Skip to content

Q002 — Assessment

Query

The W3C Credibility Coalition and Credible Web Community Group developed content credibility indicators (CCIV) and a credibility signals specification. What is the current status of this work? Is it actively maintained, adopted, or abandoned? Does it include any of: hierarchical evidence quality scale, calibrated confidence language, structured bias assessment, or source reliability tiering?

Hypotheses Tested

  • H1 (Active/Comprehensive): The work is actively maintained and includes evidence quality scales, confidence language, bias assessment, and source tiering.
  • H2 (Abandoned/Limited): The work is abandoned or dormant and never included those features.
  • H3 (Nuanced): The work produced useful taxonomies but stalled at the indicator level without reaching epistemological depth, and is now inactive.

Finding

H3 is best supported by the evidence.

The W3C Credible Web Community Group and Credibility Coalition occupy a middle state between active and abandoned. Specification development has effectively stopped (last GitHub commit circa 2019, last dated specification February 2020), but the community group continues to hold quarterly meetings in a networking/maintenance mode. The specifications produced are valuable taxonomies of observable credibility signals but do NOT include any of the four features asked about.

Confidence: HIGH

Status Assessment

REPORTED: The W3C Credible Web Community Group held quarterly check-in meetings through 2024 (February, May, August, September/TPAC). The August 2024 meeting featured Japan's Originator Profile initiative (SRC-Q2-10).

FACT: The most recent dated specification output is "Reviewed Credibility Signals" dated 2020-02-24 (SRC-Q2-03). The CCIV document is explicitly marked as archival (SRC-Q2-04). The GitHub repository shows no commits more recent than 2019 (SRC-Q2-11).

JUDGMENT: The initiative is best described as "dormant with periodic check-ins" rather than either "actively maintained" or "abandoned." The specification development is complete or stalled, while the community maintains a minimal meeting cadence. This is a common pattern in W3C Community Groups that have completed their initial output but lack resources or mandate for further development.

Feature Assessment

Requested Feature Present? Evidence
Hierarchical evidence quality scale No Signals are organized hierarchically by TOPIC (grouping), not by QUALITY. The hierarchy identifies subject types (claim, article, source, media), not evidence strength levels (SRC-Q2-12).
Calibrated confidence language No The specification defines signals as observable features. It does not include any calibrated language for expressing confidence or uncertainty in assessments (SRC-Q2-02, SRC-Q2-03).
Structured bias assessment No No signals specifically address cognitive or analytical bias in the assessment process. Some signals relate to organizational bias indicators (e.g., funding transparency), but these are about the SUBJECT being assessed, not about the ASSESSOR's reasoning (SRC-Q2-03).
Source reliability tiering Partial Some signals function as source reliability indicators (journalism awards, site longevity, corrections policies — SRC-Q2-03). However, these are BINARY observational signals (present/absent), not a TIERED reliability scale. There is no mechanism to combine signals into a hierarchical source reliability rating.

What the Specifications Actually Provide

The Credibility Signals specification (SRC-Q2-02, SRC-Q2-12) is a VOCABULARY/TAXONOMY, not a METHODOLOGY. It answers the question: "What observable features of content might be relevant to credibility assessment?" It does NOT answer: "How should these features be weighted, combined, or interpreted to produce a quality judgment?"

Key characteristics:

  1. Descriptive, not prescriptive: Signals describe what CAN be observed, not what SHOULD be concluded (SRC-Q2-05).
  2. User-empowering, not authoritative: The design philosophy is to help users make their OWN credibility decisions, not to produce an authoritative quality rating (SRC-Q2-05, SRC-Q2-13).
  3. Feature-level, not judgment-level: Signals are atomic units of information (individual observations), not composite assessments. The specification does not define how to aggregate signals into overall credibility judgments (SRC-Q2-13).

Adoption

JUDGMENT: Adoption is limited and indirect.

  • ClaimReview (Schema.org) was developed in parallel but separately from the CredWeb work, and has achieved much broader adoption.
  • The Credibility Signals vocabulary has influenced academic research on automated credibility assessment (SRC-Q2-14) but has not been widely adopted as a standard.
  • The CredCatalog (SRC-Q2-09) catalogs credibility initiatives but does not demonstrate broad adoption of the signals specification itself.

Gaps and Limitations

  1. Meeting minutes from 2023-2024 were not accessible via search; the characterization of meetings as "networking/maintenance mode" is inferred from the meeting titles (quarterly check-ins, guest presentations) rather than from reading the minutes.
  2. The W3C mailing list archives were not deeply searched; additional context about group decisions may exist there.
  3. Individual participant organizations may be using the signals vocabulary in ways not documented publicly.
  4. The assessment of GitHub activity is based on web search results, not direct repository inspection; there may be recent activity not indexed by search.