# Proof of Understanding as Experimental Method

*One-pager · Evaluation protocol for multi-intelligence assemblies*

A method for testing whether comprehension-based attestation produces better collective intelligence than possession-based authentication. The protocol turns Proof of Understanding from a principle into a falsifiable experimental claim.

**Central hypothesis.** Groups that generate stronger, more diverse, and more repairable Proofs of Understanding produce outputs with higher reality contact than groups that do not.

**Three-layer assessment** (reviewers for layer 3 blind to layer 1):

- **Process Quality** — whether PoU is happening, measured by trust-graph instrumentation.
- **Output Quality** — coherence and review-readiness of what the assembly produced.
- **Reality Contact** — whether the output's map survives the world over 3, 12, and 36 months.

**Direction of PoU** — four orthogonal axes tagged on every edge:

| Axis | Variable |
|---|---|
| 1 | Symmetry (bilateral / unilateral) |
| 2 | Power gradient (upward / downward) |
| 3 | Time vector (convergent / divergent / differentiating) |
| 4 | Intelligence-pair type (🙂↔🙂, 🙂↔🤖, 🙂↔🌿, etc.) |

**Structure.** 42 participants in 7 groups of 6 over 6 weeks. Nine-phase Palava (orientation → bilateral PoU → small-group synthesis → AI character sessions → natural-intelligence sessions → drafting → peer review → red-team → publication).

**Measurement.** BabelFish logs a PoU edge when restatement, confirmation, revision, recognition, consequence, and chain-anchored record have all occurred.

**Controls (five).** Classic Expert Panel · Human-Only Assembly · AI-Only Drafting · Standard Citizens' Assembly · Adversarial Red Team.

**First pilot.** A draft constitution for a multi-intelligence commons. Pre-registration required.

*Reference: pou_as_experimental_method.md (full protocol)*