Epistemic Boundaries

Protocol Charter & Methodology

DONK AI preserves human testimony, client-side encryption, and blind independent corroboration without confusing personal memory with settled historical fact.

Core Epistemic Equations
Popularity ≠ Corroboration ≠ Evidence Integrity ≠ Historical Support
Stake ≠ Truth • LPS-1 Conformance ≠ Historical Infallibility

1. Explicit Boundaries: What DONK AI Proves

✓ What Is Cryptographically Proven
  • Exact un-tampered record payload at time $T$.
  • Author account/key authorized the commitment.
  • Evidence CIDs match SHA-256 Merkle tree leaves.
  • Corroborator submitted before seeing the target record.
✕ What Is NOT Claimed
  • Does not prove the event physically happened.
  • Does not verify real-world human identity.
  • Does not claim majority consensus equals truth.
  • Does not accept generative AI as source evidence.

2. Bounded Historical Review Rubric

Reviews avoid binary "true/false" ratings and produce one of five transparent states:

Contextualized
Partially Supported
Contested
Insufficient Evidence
Unresolved

3. Immutable Protocol Metadata Quartet

{
  "protocolSpecificationVersion": "2.0",
  "commitmentAlgorithm": "lps1-merkle-v1",
  "fixtureCorpusVersion": "lps1-v1",
  "eip712DomainVersion": "1"
}