The substrate measures its own words. JEV quantifies each word's quality. JEPA quantifies each word's position in the relational geometry. Together they make the pseudo-weight.
Try it
Type a sentence. The substrate will tokenize, embed each word, and compute JEV-style quality scores + JEPA-style relational positions against the canonical substrate opcodes (WITNESS, PROOF, BIND, FORGET, TICK, JEPA, JEV, LINK).
The witness forgets nothing the proof cannot reach.PROOF holds the substrate together.FORGET is the immune system; JEV is its superego.The brewery grows the substrate that brews it.Words carry scores as qualities; the substrate IS its own measurement.
What you're seeing
JEV confidence — for each word, how strongly does the substrate identify it as one of its canonical concepts (WITNESS, PROOF, BIND, FORGET, TICK, JEPA, JEV, LINK)? The substrate's superego assigns this.
JEPA position — for each word, what is its cosine similarity to each substrate concept? The substrate's id measures this relational geometry.
Pseudo-weight — JEV quality × JEPA position. The word's measurement AS the word. Reading is navigating these measurements.