This table exists to stop one type of evidence being used to substantiate a different type of claim.
| Claim type | Appropriate proof |
|---|---|
| Code exists | Repository plus deterministic test |
| Verifier portability | Offline external reconstruction |
| SDK usability | External developer integration, unassisted |
| Reuse | Second and third materially different integration |
| Scientific sufficiency | Domain review plus empirical evaluation |
| Buyer value | External recipient actually uses the artifact |
| Willingness to pay | Money |
| Recurring value | Renewal or recurring contract |
| Defensibility | Behaviour that persists despite available alternatives |
| Portfolio leverage | Measured reduction in time or cost, or a new product |
| Market standardisation | Unsolicited compatible adoption or requests |
A repository proves code exists. It does not prove reuse. A pilot proves capability. It does not prove willingness to pay. Only money proves willingness to pay.
used as proof of
external reliance
used as proof of
repeatability
used as proof of
willingness to pay
The evidence needed depends on the claim.
Code can prove code exists. It cannot prove developer adoption. An integration can prove the SDK works for one source. It cannot prove reuse. A buyer interview can support a problem thesis. It cannot prove willingness to pay. A paid pilot proves a transaction once. It does not prove recurring margin.
The review should state the burden before reading the evidence.
The native reading. Claim status, evidence burden, proof gates, falsifiers, contradictions, capital-to-proof, governance, downside, and explicit reasons to stop or narrow the thesis.