Fair, curriculum-faithful marking needs the right frame of reference. Laura models a per-system and per-state layer instead of forcing a single scale.