Quad-lens flux ratios, clean at radio and mid-infrared wavelengths, defy smooth models that fit the image positions perfectly, demanding small-scale perturbers whose required abundance, from CLASS onward, runs several times above the CDM subhalo supply at the radii where quads form, where the same simulations deplete subhalos by tidal destruction. JWST's expanding clean samples keep the demand at or above the CDM ceiling: the galaxy-scale sibling of the cluster GGSL excess.
The model assigns every anomaly to dark subhalos, so the anomaly rate measures the subhalo mass function: a rate above the simulated supply leaves ΛCDM owing perturbers it tidally destroyed. The test built to confirm substructure instead audits its shortage.
SCT supplies the perturbation strength rather than extra perturbers: the luminous satellites, dwarfs, and compact groups that ΛCDM bookkeeping treats as negligible carry coherence-amplified effective masses, with substructure-scale amplification A_sub of 3.2 to 4.0 and lensing cross-sections scaling as its square, a factor of 10 to 16 (the same external-superposition structure that produces the cluster GGSL excess). Perturbations strong and frequent enough to anomalize flux ratios then come from the visible small-scale inventory plus the mesh's granularity (P50, P51, P52, P53): the anomaly rate tracks real, observable structure rather than a dark population the simulations cannot keep alive at quad radii.
The discriminating statistics follow: anomalies should correlate with detectable luminous substructure and local environment richness more strongly than CDM-subhalo models predict (where the perturbers are dark and uncorrelated with light), and deep imaging of anomalous quads should keep finding the responsible visible structures, as JWST has begun to. The tidal-destruction paradox dissolves because nothing dark needs to survive: the perturbers are the satellites that demonstrably exist, weighed correctly.
This is the same A_sub coherence behind the Meneghetti GGSL excess, scaled to galaxy lenses. There is no need to owe the lens models a dark population the model's own dynamics removes.
The registered kill carries over: if raising CDM subhalo concentrations within simulations fully reproduces the anomaly statistics at all radii, the amplification is unnecessary and the excess is a concentration artifact. The correlation test cuts directly: anomaly rates statistically independent of detectable luminous substructure, with deep JWST imaging consistently finding no visible counterpart where anomalies demand perturbers, would favor the dark-subhalo reading and break the amplified-visible account.