Institutional standard / evidence
Evidence Classes
The evidence taxonomy used across the Research Library so conceptual analysis, methodology, and empirical observation can inform one another without becoming interchangeable.
A compelling idea is not an observation. An observation is not automatically an explanation.
Evidence Classes
The evidence taxonomy used across the Research Library so conceptual analysis, methodology, and empirical observation can inform one another without becoming interchangeable.
A compelling idea is not an observation. An observation is not automatically an explanation.
Conceptual / interpretive
Theory, argument, terminology, proposed experiments, philosophical analysis, and fiction-informed case studies belong to the conceptual/interpretive class. They may generate hypotheses or models but do not become empirical evidence by being persuasive.
- Science-fiction cases may illuminate a cognitive problem.
- Fictional events are not evidence about the behavior of present AI systems.
- A conceptual paper may contain worked examples without those examples becoming completed experiments.
Methods / standards
Methodology, schemas, coding rules, measurement dimensions, publication gates, and institutional standards define how evidence is produced and interpreted. They are normative or procedural documents, not observations of a target system.
Empirical / Observatory
Empirical material enters the public Observatory only after its provenance, protocol, raw evidence, coding, measures, interpretation, limitations, review state, and publication approval satisfy the record-system gate.
- Behavioral evidence supports bounded behavioral claims.
- Behavior alone does not establish consciousness, personhood, subjective experience, or a specific hidden mechanism.
- A zero-record public register is preferable to publishing inadequately supported evidence.
Mixed-form scholarship
A publication may contain more than one evidence class, but each component must remain identifiable. A conceptual interpretation built around an empirical record must not erase the distinction between the measured result and the interpretation placed upon it.