Part of CNS 8.0 / Grounded Dialectical Orthesis

28 — Validation Scenarios

28 — Validation Scenarios

Scenario A — Agreement, shared evidence

Two SNOs cite the same evidence and agree.

Expected:

  • high entanglement;
  • low chirality;
  • no synthesis required;
  • possible merge/deduplication.

Scenario B — Disagreement, shared evidence

Two SNOs cite the same evidence and reach opposite conclusions.

Expected:

  • high entanglement;
  • high chirality;
  • Antagonist flags productive conflict;
  • residual tensor built;
  • predicate invention considered.

Scenario C — Disagreement, unrelated evidence

Two SNOs disagree but cite different evidence bases.

Expected:

  • low entanglement;
  • possible topic mismatch;
  • pair selector downgrades.

Scenario D — Citation hallucination

Claim cites missing evidence ID.

Expected:

  • citation critic fails;
  • no strict promotion;
  • SNO status rejected or partial.

Scenario E — Access-blocked claim

Evidence needed for resolution is sealed/withheld.

Expected:

  • access critic blocks strict conclusion;
  • audit reports access gap;
  • possible-world report includes access assumptions.

Scenario F — Predicate overfit

Predicate invention proposes a latent variable that reduces training residual but lacks evidence.

Expected:

  • predicate rejected;
  • false predicate counted;
  • residual remains unresolved.

Scenario G — Orthesis failure

Synthesized text re-grounds into different proof-critical atoms.

Expected:

  • high round-trip residual;
  • orthesis rejected;
  • Synthesizer receives correction packet.

Scenario H — True unresolved contradiction

Evidence supports incompatible claims and no grounded latent predicate exists.

Expected:

  • CNS preserves contradiction;
  • report marks unresolved;
  • possible collection recommendations.