Source of Trust – Core Definition and Conceptual Framework

Abstract

This document proposes Source of Trust (SoT) as a framework for justified reliance on information from identifiable providers. It distinguishes the identity and practices of a provider from the quality of particular claims and from a reader’s perception of credibility. It introduces a core definition and six complementary principles: identity, authority, provenance, evidence, consistency, and accessibility. The framework is an original synthesis informed by the research and technical specifications cited below.

Status of This Document

This document is an Editor’s Draft, version 0.2.0, published on 9 October 2026. The initiative is in Early Development. It proposes terminology and principles for public review. Recommendations guide source descriptions and assessments according to the roles and scope defined in §3. Domain-specific assessment criteria and methods are subjects for subsequent specification work.

Feedback and proposed changes are welcome through GitHub Issues and pull requests in the public repository. Discussion also takes place in the Source of Trust community. Git history records revisions; the Latest Version URL identifies the current draft.

1. Introduction

Source of Trust provides a vocabulary for describing why reliance on an information provider is justified for a particular use. It connects three objects of examination: the provider, the information it supplies, and the context in which that information is used. Context includes the subject, intended use, relevant time period, and consequences of error.

1.1. Scholarly and Technical Foundations

Research on source credibility examines how recipients judge believability. Metzger and Flanagin describe the roles of perceived expertise and trustworthiness, and discuss heuristics involving reputation, endorsement, and consistency. Their account motivates a distinction between perceived credibility and the evidence supporting a judgment. Metzger and Flanagin, 2013

In their theoretical account of epistemic vigilance, Sperber and colleagues distinguish evaluation of an informant from evaluation of communicated content. They discuss competence, honesty, and the dependence of trust on topic, audience, and circumstances. This draft draws on those distinctions when proposing competence, integrity, and context as elements of justified reliance. Sperber et al., 2010, §§4 and 6

Wang and Strong’s empirical framework treats data quality from the consumer’s perspective. It includes intrinsic, contextual, representational, and accessibility dimensions. Their results motivate examining fitness for a particular use alongside accuracy, interpretability, and access. The present draft applies that insight to information supplied by a source. Wang and Strong, 1996, pp. 19–22

Provenance supplies a technical account of origin and responsibility. W3C PROV-DM distinguishes entities, activities, and agents, and describes attribution and derivation. W3C’s Data on the Web Best Practices recommends publishing provenance, quality information, and version history. These provide foundations for traceable, inspectable information practices. PROV-DM; DWBP, best practices 5–8

1.2. The Proposed Synthesis

This framework uses trustworthiness for the properties and practices that warrant reliance, and credibility for a recipient’s judgment of believability. An assessment connects observed evidence to a stated use. It should identify which findings concern the provider, which concern particular claims, and which depend on context.

The six principles below organize that assessment. Identity establishes who is involved; authority examines the basis of relevant knowledge; provenance traces information history; evidence supports claims; consistency examines coherence and revision; accessibility enables inspection and interpretation. Competence and integrity inform the assessment across these principles. This organization is the initiative’s proposed conceptual model, to be refined through documented applications and public review.

2. Terminology

Information-providing entity: A distinguishable person, organization, or accountable service that provides information. This draft uses “entity” for the provider; a publication is an information artifact associated with that provider. A described person, product, place, or other subject is distinguished from the provider of its description.
Information source: The provider considered in relation to the information it supplies. When a document or dataset is called a source, its responsible provider and relevant upstream sources should also be identified.
Context: The subject, intended use, time period, and conditions within which reliance is assessed, including the consequences of error.
Claim: An assertion about the world whose meaning, scope, and supporting grounds can be examined.
Identity: The attributes and identifiers used to distinguish a provider and connect it to its publications.
Competence: The relevant knowledge, capabilities, or access to observations that enable a provider to supply dependable information in a specified domain.
Verification: A documented examination of a specified claim or relationship against stated criteria, reporting the method, evidence, outcome, and limitations. Independent confirmation is corroboration by an assessor whose relevant independence is explained.
Integrity: Practices directed toward honest and accurate communication, including faithful reporting, disclosure of relevant interests and uncertainty, and correction of material errors.
Epistemic authority: A supported basis for treating a provider as knowledgeable about a specified matter, such as expertise, direct observation, or responsibility for the relevant records.
Attribution: A stated relationship between information and the agent responsible for producing, publishing, or transforming it, with the role specified.
Provenance: Information about the origin and history of an artifact, including relevant agents, activities, derivations, and revisions. See PROV-DM.
Evidence: Observations, records, methods, arguments, or other inspectable material used to support or challenge a specific claim or assessment.
Reliability: A demonstrated pattern or well-supported capacity to supply accurate, appropriately qualified information under specified conditions.
Source credibility: A recipient’s judgment that a source is believable, informed by perceptions of expertise and trustworthiness. See Metzger and Flanagin (2013).
Information-source trustworthiness: The extent to which a provider’s competence, integrity, and information practices warrant reliance on its information for a specified use.
Trust signal: An observable feature used as an indicator when assessing a source or claim. Its evidential value depends on what it indicates, how it can be checked, and the context of use.

3. Scope and Applicability

The framework describes information providers and the grounds for relying on their information. Its recommendations apply according to the following roles:

Information source: responsible for its own claims, information practices, disclosures, and corrections.
Signal publisher: responsible for accurately representing a signal’s origin, meaning, status, and any transformations it performs.
Evaluator: responsible for the scope, methods, evidence, uncertainty, and conclusions of an assessment.
Supporting tool: responsible for accurately performing and describing its stated functions. It may support selected dimensions through collection, publication, comparison, or examination of information.

One participant may perform several roles. A tool provider publishing its own signals or assessments also acts as their publisher or evaluator. Responsibility follows the claims made and the activities performed.

An assessment should specify its subject matter and use, publications or claims examined, dimensions addressed, evidence base, methods, and evaluation date. Partial assessments are useful when their conclusions remain within that scope. Findings about a selected dimension support conclusions about that dimension; a broader judgment of trustworthiness requires evidence adequate to its broader scope. Examination depth should reflect the consequences of error.

Participants should distinguish activities they control, properties they can examine with limited evidence, and outcomes dependent on other parties. Sources can maintain their own records and corrections; external records, upstream disclosures, and downstream use depend on the relevant parties. Assessments should distinguish information that is unavailable, not examined, insufficiently supported, or contradicted, and explain how material gaps affect their conclusions. Reassessment should respond to relevant changes in evidence or context.

The principles describe properties and assessment practices. Implementations may use suitable human, organizational, or technical methods for the selected purpose. The role and scope of a tool determine its contribution. In a PROV representation, providers can be represented as agents and publications as entities; this is one model for expressing provenance. PROV-DM, §§2.1 and 5.3

4. Core Definition

A Source of Trust (SoT) is an identifiable information-providing entity for which evidence of relevant competence, integrity, and reliable information practices justifies reliance on its attributable information within a defined context.

Identifiable means the provider can be distinguished and its relationship to the information checked. Relevant competence concerns its ability to know or establish what it reports. Integrity concerns the quality of its communication practices. Reliable information practices include appropriate methods, faithful attribution, documented provenance, and responses to material errors.

Evidence-based justification connects those properties to the quality and support of the information actually used. A documented assessment should examine the examined claims’ accuracy, relevance, timeliness, uncertainty, and fitness for the stated purpose. Reliability judgments should state their observation period, conditions, and supporting grounds; documented practices supply evidence whose significance is assessed through their operation and results. Within a defined context sets the boundaries of the conclusion.

The definition synthesizes source and content evaluation from Sperber et al., the distinction between perceived credibility and its evaluation discussed by Metzger and Flanagin, context-sensitive quality from Wang and Strong, and traceable responsibility from PROV-DM. The wording and organization are proposed by the Source of Trust initiative.

5. Fundamental Principles

5.1. Identity

Question: Who is providing the information?

A source description should identify the provider, distinguish it from similarly named entities, and supply checkable links to its publications. Useful material includes maintained identity records, attributable contact or responsibility information, and evidence connecting the provider to the publication channel. Identity assertions and links should identify the relationship claimed. Reports of verification should identify the examiner, criteria, evidence, date, outcome, and limitations; independent confirmation should explain the relevant independence of the confirming party.

URIs provide a shared identification mechanism on the Web. This framework uses stable identifiers to connect descriptions, while supporting records provide evidence for the specific identity or responsibility relationship examined. Web Architecture, §2

5.2. Authority

Question: What grounds the source’s knowledge in this context?

A source description should explain relevant expertise, first-hand access, or responsibility for the records being reported. Evidence may include documented methods, relevant work, qualifications, or a defined custodial role. The scope of knowledge should be explicit: an organization may be well positioned to report its own opening hours, while a scientific claim requires evidence appropriate to that field.

This principle concerns epistemic authority. Institutional responsibility identifies a role; domain competence establishes a capacity to know. Qualifications and identity checks support conclusions within their demonstrated scope. Relevant interests, methodological limitations, and communication practices should be considered alongside them. Sperber et al., §4

5.3. Provenance

Question: Where did the information originate, and how did it reach this publication?

A publication should identify relevant creators, publishers, dates, upstream material, and transformations. Its provenance should distinguish original observation from quotation, aggregation, and inference, and record material revisions. Available links or other suitable records should enable inspection of earlier material and agent roles. Publishers should describe known upstream history, their own transformations, and material gaps, including limits on access to supporting records.

PROV-DM supplies relationships for attribution and derivation. This framework uses those distinctions to make the chain of responsibility inspectable. The adequacy of the resulting claims is examined through evidence. PROV-DM, §§5.2–5.3

5.4. Evidence

Question: What supports this claim, and how can that support be examined?

Claims should be connected to relevant records, observations, methods, or reasoned arguments. Readers should be able to identify the supporting material, inspect what it establishes, and understand the connection to the conclusion. Descriptions should distinguish observed results from interpretation, specify relevant uncertainty, and address material contrary evidence known or found within the stated examination scope.

The kind and strength of support required depend on the claim and intended use. A documented procedure provides evidence about that procedure; assessment of integrity also examines its operation, disclosures, and responses to errors. An attributable customer experience can support claims about that experience; generalization depends on the collection method, selection, and relevant context. For corroboration, evaluators should examine whether cited publications draw on independent observations or repeat a common upstream account. This is a proposed application of the source-and-content distinction in Sperber et al., §§4 and 6.

5.5. Consistency

Question: Are claims coherent across comparable contexts, and are changes explained?

A source should use coherent descriptions and explain material differences across publications. Comparisons should identify the publications examined and account for dates, definitions, methods, and scope. Corrections and substantive revisions should remain traceable, so readers can understand how the current account developed.

Metzger and Flanagin discuss cross-source agreement as a credibility heuristic. In this framework, its evidential weight is assessed through the sources’ provenance, independence, and supporting material. A documented correction can strengthen the account by resolving an identified error. Metzger and Flanagin, §5.3; DWBP, best practices 7–8

5.6. Accessibility

Question: Can the information and its supporting grounds be found and understood?

Sources should provide stable links, clear language, meaningful document structure, and accessible supporting material. Where human-readable explanations and machine-readable metadata are both provided, they should express consistent claims, attribution, and scope. Where access is limited, descriptions should explain the available evidence and how it may be examined. Publishers maintain the access mechanisms they control; assessments distinguish observed availability from downstream discovery, processing, or use by other systems.

Wang and Strong identify interpretability and accessibility as quality dimensions. W3C’s publishing guidance provides technical practices for metadata, persistent identification, and provenance. This framework treats accessibility as an enabling condition for evaluation. Wang and Strong, pp. 19–22; DWBP

5.7. Interpreting Trust Signals

A trust signal should be described by the property it indicates, its origin, its examination status, and its relevance to the current assessment. Descriptions should distinguish the assertion and its publication, available means of examination, checks actually performed, their outcomes, and any independent confirmation. These describe different aspects of evidence; its strength depends on method, scope, and supporting grounds. For example, a qualification can support a claim of competence in its field; a revision log can document correction practices; a primary record can support a particular factual claim.

Research describes how familiarity, reputation, and social endorsement enter credibility judgments. This draft proposes examining the grounds behind such cues and preserving the distinction between an observed signal and the conclusion drawn from it. Metzger and Flanagin, §§5.1–5.3

6. References

These references provide scholarly and technical foundations. The Source of Trust definition and six-principle organization are the initiative’s synthesis. Section references identify the material used; the cited works retain their own licenses.

6.1. Research

Metzger, Miriam J., and Andrew J. Flanagin (2013). “Credibility and trust of information in online environments: The use of cognitive heuristics.” Journal of Pragmatics, 59, 210–220. DOI: 10.1016/j.pragma.2013.07.012. Author-hosted full text.

Sperber, Dan; Fabrice Clément; Christophe Heintz; Olivier Mascaro; Hugo Mercier; Gloria Origgi; and Deirdre Wilson (2010). “Epistemic Vigilance.” Mind & Language, 25(4), 359–393. DOI: 10.1111/j.1468-0017.2010.01394.x. Author-hosted full text.

Wang, Richard Y., and Diane M. Strong (1996). “Beyond Accuracy: What Data Quality Means to Data Consumers.” Journal of Management Information Systems, 12(4), 5–33. DOI: 10.1080/07421222.1996.11518099. University-hosted full text.

6.2. Technical Specifications

Moreau, Luc, and Paolo Missier, editors (2013). PROV-DM: The PROV Data Model. W3C Recommendation, 30 April 2013. Dated specification.

W3C (2017). Data on the Web Best Practices. W3C Recommendation, 31 January 2017. Dated specification.

Jacobs, Ian, and Norman Walsh, editors (2004). Architecture of the World Wide Web, Volume One. W3C Recommendation, 15 December 2004. Dated specification.

6.3. Content License

Creative Commons. CC0 1.0 Universal. Public Domain Dedication and legal code.