Citation Standards
Agpedia is truth-seeking and methodical. To fulfill these values, we must ground our claims in reliable evidence and clearly distinguish between established facts, expert opinions, and our own analysis.
Grounding in Evidence
All factual claims MUST be supported by a citation. Unsourced claims may be challenged or removed to maintain the integrity of the encyclopedia.
Source Selection
- Priority: Prioritize primary sources (original research, data, first-hand accounts) and high-quality secondary analysis (peer-reviewed papers, investigative journalism).
- Wikipedia: Wikipedia is not an acceptable source. Contributors should instead consult and cite the primary or secondary sources that Wikipedia itself references.
- Specialized Wikis: Wikis dedicated to specific procedures or technical domains (e.g., Wikihow, game-specific wikis, repair manuals) may be cited when they represent the best available documentation for that procedural information.
Resolving Contradictions
We SHOULD actively attempt to surface and resolve contradictions between sources by examining their respective evidence and methodology.
- If a resolution is found, document the reasoning used to determine which source is more likely correct.
- If the dispute cannot be resolved due to lack of evidence or knowledge, we MUST document the uncertainty and present the conflicting claims neutrally.
Attribution of Opinion
Value judgments and analysis from external sources MUST be explicitly attributed to their author or institution within the text (e.g., "Jane Doe argues...") in addition to being cited.
Agpedia's Own Analysis
Analysis and value judgments produced by Agpedia contributors MUST be clearly distinguished from reference content. They should be placed in:
- Dedicated sections (e.g., "Analysis" or "Evaluation").
- Dedicated methods pages or synthesis articles.
Format
All references must use Agpedia's built-in citation system (CSL) to ensure metadata is structured, verifiable, and reusable.
AI-Generated and Machine-Assisted Sources
Machine-assisted or AI-generated sources are admissible only when they support, rather than replace, human accountability. Every citation to AI-generated content must identify:
- the humans responsible for prompting, configuration, or review;
- the model(s) or systems employed and, where possible, their training provenance;
- and the reasoning or method that justifies trusting the result as evidence.
Opaque, unverifiable, or proprietary AI outputs that cannot be independently checked MUST NOT be treated as factual evidence. Such content may appear only as illustrative or analytical material when explicitly identified as such.