Falsification Packs

Pack Structure

Each falsification pack should include:

• target claim or result
• context
• inputs and reference links
• expected structure
• evaluation criteria
• failure condition
• pack status (draft / current / canonical / pending)
• related Results, Corpus, and Verify links

Pack Status Vocabulary

• draft — the pack is being assembled; structure may change.
• current — the pack is the active production surface for its target.
• canonical — the pack has been frozen as the reference structure for its claim family.
• pending — the pack target is identified but the bundle is not yet assembled.

Current State

The first production surface is the 30-item physics falsification pack (status: current). This Verify page defines the pack grammar so future numeric, structural, and domain-specific packs can be added consistently.

Packs may be executed manually, computationally, or through LLM-assisted review where appropriate. Packs are operational bundles for inspection, not claims of completed falsification.

Related Verify Surfaces

• Assessment Protocols — manual and LLM-assisted review routes.
• Measurement Bridges — the construction step where empirical accountability begins.
• LLM-Assisted Protocols — for assisted pack review.