BOSS galaxy clustering plus the standard galaxy-halo connection under Planck ΛCDM predicts a galaxy-galaxy lensing signal 20 to 40 percent larger than measured, worst (about 30 percent) below 5 Mpc/h (Leauthaud et al. 2017). Baryons, neutrinos, and reasonable GR modifications each move the prediction only a few percent, and a five-survey blind comparison confirms the deficit is real, not instrumental.
The halo model assumes clustering and lensing are two readings of the same CDM halos: positions fix masses, masses fix lensing. There is no freedom for the two to disagree by 30 percent, so the model must either inflate halo-occupation flexibility until prediction evaporates or carry an unexplained deficit that points the same way as the S₈ tension.
SCT replaces the dark matter halo with coherent gravitational amplification of baryons: the effective mass profile is A(r) times baryonic, where the coherence amplification grows from near 1 in inner regions toward the virialized fixed point A* = 5.970 at the virial radius (P50, P52). The radial shape matters: an NFW halo calibrated to match the clustering-implied total mass distributes that mass differently with radius than the coherence profile does, exceeding it precisely at the intermediate radii (0.1 to 5 Mpc/h) where galaxy-galaxy lensing measures. A ΛCDM halo fit to the positions therefore overpredicts the lensing at those radii: lensing is low because the NFW expectation is high, not because mass is missing.
The redshift-and-amplitude bookkeeping closes the same way as the S₈ family: the Planck-normalized clustering amplitude carries coherence contributions a matter-only model attributes to mass, so every late-time lensing observable, cosmic shear, CMB-lensing cross-correlations, and galaxy-galaxy lensing, comes in low against it by the related factor, with sigma8_inferred = A^(1/2) sigma8_true setting the scale. One secondary modulation: assembly-bias-like selection effects are real and absorb part of the small-scale gap, which is why HOD flexibility helps below 1 Mpc/h without resolving the full range.
This is the same coherence physics behind the S₈ deficit, the growth-index excess, and the cluster lensing calibrations. There is no need to invoke 30 percent baryonic feedback or to dismantle the halo model's predictivity.
The discriminating shape: SCT requires the lensing deficit to track the coherence profile, largest at intermediate radii, shrinking toward small radii where A approaches 1 and toward large radii where both models converge on the two-halo term, and diminishing with redshift as A(z) falls. DESI and Euclid measuring a deficit that is flat in scale or constant in redshift would break the coherence reading; full agreement of lensing with NFW-shaped profiles at all radii in a large stacked sample would do the same.