What Counts as Knowledge
Summary
This post provides technical definition of knowledge grounded in Information Theory and Physics of Agency framework, addressing traditional “justified true belief” problems including Gettier cases. Knowledge is pattern-encoded information within agent’s predictive structure reliably and quantifiably reducing Shannon entropy (uncertainty) regarding future events/states across quantum-branching timelines. Three components: (1) Pattern Identifier—reproducible structure (neural configuration, logical rule, algorithm, cultural convention) enabling consistent predictions; (2) Reliability—PI consistently reduces uncertainty across scenarios; (3) Entropy Reduction—quantifiable Shannon entropy decrease in bits, providing objective predictive capability metric. Definition tested against scenarios: weather forecasts (knowledge), false beliefs (fail—no reliable entropy reduction), Gettier problems (fail—no reliable predictability), quantum uncertainty (epistemic vs. ontic distinction), random guesses (fail), tacit knowledge/skills (pass), cultural knowledge (pass), conditional knowledge (context-dependent pass), QBU branching knowledge (pass), raw data without agency (fail). Integrates with Conditionalist and QBU frameworks, emphasizing agent-centricity, measurability, practical action-facilitation.
Key Concepts
- Pattern-encoded information – Knowledge as reproducible structure within agent enabling predictions, not isolated beliefs.
- Shannon entropy reduction – Quantifiable uncertainty decrease in bits as objective knowledge metric.
- Reliability criterion – Consistent entropy reduction across scenarios validates knowledge status.
- Agent-centricity – Knowledge requires agent’s predictive structure; raw data insufficient.
- Epistemic vs. ontic uncertainty – Incomplete agent knowledge versus irreducible quantum-mechanical branching.
- Gettier resolution – Stopped clock scenario fails reliability criterion despite justified true belief.
- Quantum Branching Universe integration – Knowledge enables steering choices toward favorable branching outcomes.
Evolution Notes
- Continues “What Counts as X” series with most technically sophisticated definition yet.
- Integrates multiple earlier frameworks: Physics of Agency, Conditionalism, QBU, information theory.
- Demonstrates cumulative theory-building: later work leveraging earlier conceptual foundations.
- Provides formal alternative to traditional analytic epistemology using computational/physical concepts.
- Emphasizes measurability and operationalization—knowledge assessable via entropy calculations.
- Reflects Axio’s scientific naturalism: philosophical concepts grounded in physics/information theory.
- Tacit knowledge inclusion challenges traditional propositional-only epistemology.
Tags
- epistemology
- knowledge
- information theory
- Shannon entropy
- Gettier problems
- justified true belief
- Physics of Agency
- quantum branching universe
Cross-References
Open Questions
- How do we measure Shannon entropy reduction for non-formalized knowledge (intuitions, aesthetic judgments)?
- Does the definition exclude knowledge of necessary truths (mathematical, logical) that don’t reduce contingent uncertainty?
- Can collective/distributed knowledge exist, or only individual agents possess knowledge?
- How does this framework handle self-knowledge and metacognition (knowledge about one’s own knowledge)?
- Does the agent-centricity criterion anthropomorphize knowledge unacceptably, excluding “objective” facts?
- How do we reconcile measurable entropy reduction with phenomenological aspects of knowing?
- Can the definition handle knowledge of singular past events that don’t enable future predictions?