Layer 1 · verbatim base + tierguard · claim-vs-quote driftLayer 2 · shared structureLayer 3 · your overlays
Layer 3 — your overlays → the conclusion
—
The visible result is computed from editable Update Nodes. Source admission, priors, prior strength, source shifts, admissibility multipliers, dependence weights, and child projections can be inspected and changed in Layer 2.
Base sources and typed model structure are shared across readers; overlays are per-reader edits to admissibility, priors, shifts, weights, and child projections.