Proof Layer
The verification rubric, asymmetry profile, and verification cost analysis built BEFORE the capability. Every AI system needs one. No exceptions.
Why It Exists
Generation costs are collapsing and the bottleneck has moved to verification - knowing whether the output is correct. Build the rubric first. If you can't describe what "good" looks like before building, you won't recognize it after.
Rosetta Stone
Four circles, four readings of the same object. Each role reads the artifact through its own lens.
Pre-investment due diligence, hardened. Define what "good" means before you deploy capital; otherwise you have no way to price success.
The rubric you build before turning the agent on. If you cannot describe what good looks like in advance, you won't recognize it when it ships.
Test-driven development for operations. Write the assertions before the implementation. Pass the assertions, ship the work.
A pre-registered evaluation protocol. Write the null hypothesis, the test, and the decision rule before observing data. Bayesian-valid posteriors require the prior be fixed in advance.
Related Terms
Gold Standard - A set of human-verified, ground-truth examples used to calibrate and evaluate AI output.
Templeton Ratio - T = time_to_do / time_to_check.