The splashback radius is ΛCDM's cleanest structural prediction, set by collisionless dynamics near 1 to 1.5 R200m, and the measurements find it about 20 percent smaller around SDSS redMaPPer clusters, echoed in DES, with SZ-selected samples closer to prediction but noisier. The optical-selection defense must convert percent-level weights into a 20 percent dynamical displacement, and the feature's galaxy-color dependence adds structure the artifact reading does not predict.
The edge's location is pure collisionless CDM: particles orbit, pile at apocenter, and the profile steepens where the mass function and accretion history dictate. There is no dial; a displaced edge indicts either the cluster-finding or the collisionless dynamics itself.
SCT relocates the physics of the edge: a cluster's effective mass profile is its baryonic distribution amplified by A(r), and the amplification has its own boundary, the coherence length at which A_obs falls away from the virialized fixed point, registered to coincide with the virial radius within 30 percent (P50, P52, P62). The steepening feature in stacked lensing and galaxy profiles marks where coherent amplification releases, an edge set by coherence geometry rather than by first-orbit apocenters, and there is no reason it should land where collisionless splashback bookkeeping puts it: modestly inside the CDM expectation is the natural placement for a boundary tied to the virialized region itself. The 20 percent displacement reads as the difference between two edge definitions, only one of which the sky uses.
The galaxy-population dependence follows naturally: red and blue populations sample different orbit families and different coherence states within the cluster frame (the C-function's velocity dependence), so their apparent edges differ, dynamical structure the selection-artifact account cannot supply. The registered test is the scaling: across 50 or more clusters spanning a decade of mass, the measured edge must track the virial radius within 30 percent, with stacked Euclid lensing on SZ-selected samples the clean dataset.
This is the same A(r) profile structure behind lensing-is-low and the two-mass SZ resolution, read at the cluster boundary. There is no need to stretch selection weights into dynamical displacements.
The registered kill: no correlation between the inferred coherence-decay scale and the virial radius at more than 1σ across the stacked samples breaks the boundary identification. The cleaner converse also bites: Euclid and Rubin lensing-only splashback measurements landing exactly at the collisionless CDM prediction, with the optical displacement fully traced to selection, would restore the standard edge and leave the coherence boundary without its observable.