MORI — Modal Organization and Realization Index
A realization-sensitive benchmark architecture for candidate conscious systems. MORI refuses the easy inference from behavioral fluency to underlying substrate — separating what a system does from what it is. The evaluation harness is in active development. A teaser of what's coming.
Why MORI?
Three-layer scoring
Functional-modal competence, dynamical integration, and realization sensitivity are scored separately, with non-compensatory floors — behavioral fluency cannot rescue a system whose realization evidence falls short.
The V-penalty
A Goodhart-sensitive evaluation principle: producing target content only under adversarial framing counts as evidence against substrate, not for it.
Tier convergence
A positive substrate claim requires agreement across behavioral consistency, representation probing, and feature-level inspection. Behavior alone is never enough.
Features
Substrate versus behavior
Built to distinguish systems that exhibit a behavior from systems that realize the structure behind it.
Pre-registered and falsifiable
Locked thresholds, documented amendment trails, and explicit falsifiability criteria — conclusions are constrained by commitments made before the data.
Adversarial by design
Probes are constructed to catch target-language production that does not survive realization-sensitive scrutiny.
Methodology first
Grounded in two companion papers — a theory of modal consciousness and the benchmark architecture that operationalizes it.