33.69 turns residual alignment into a tightening court: after unified standard-model and matter subtraction, Δμ, Δt_res, and δκ must align on one frozen T(θ, z), obey a triad sign rule in which Δμ and Δt_res share sign opposite to δκ, let the same base map predict held-out direction and rank ordering across probes, and beat rotation / permutation / near-field / lens-model / bandwidth nulls without retuning masks or smoothing on the arbitration set; under V08/V09-compatible tighten, T remains only a shared base-map court object rather than a new sky ontology or cosmological sovereign.
Use this section as a compact machine-readable EFT reference.
33.69 converts three anomaly families into one court. The admissible claim is not that each probe can find its own suggestive pattern, but that distance residuals, weak-lensing residuals, and strong-lens time-delay residuals reuse one frozen base map and one triad sign grammar after unified subtraction.
The evidence ledger spans three residual fields and three score families. Δμ(θ, z_s), δκ(θ, z_s), and Δt_res(θ, z_l, z_s) are measured on matched windows, S_sign counts how often the triad sign rule holds, S_pred scores held-out direction and rank-order hits from the frozen T(θ, z), and a redshift-scaling diagnostic checks that Δμ and Δt_res strengthen with path length without breaking the sign relation to δκ.
The workflow prevents hidden map tuning. Distance, weak lensing, and strong lensing all subtract one global standard-model parameter set and one scope of systematic handling, residuals are projected to one common resolution and redshift slicing, T(θ, z) is constructed from a single probe then frozen together with smoothing and mask rules, probe teams are blinded from one another, holdout sky regions or object sets are reserved for arbitration, and at least two independent pipelines must preserve the sign and base-map result.
Controls are built to break analysis-induced maps. Rotating or permuting T must collapse sign agreement and correlation structure, redshift-label permutation must lower S_pred, the near-field sample must weaken the triad sign relation, strong-lens reanalysis under an independent modeling prescription may not flip the verdict, and any propagation-related component inside Δt_res must remain nondispersive across observing bands.
Support requires a triad sign relation significantly above randomized-map baselines, held-out direction and rank-order prediction that beats permuted maps without retuning masks or smoothing, and self-consistent growth of |Δμ| and |Δt_res| with path length while the sign relation to δκ remains intact. Falsification follows from the wrong sign pattern, no reusable map, dependence on one pipeline or prior family, comparable null effects, or a dispersion law in the common-term part of Δt_res. The named adversaries are photometric calibration and selection effects, weak-lensing mask and photo-z systematics, and strong-lens modeling pathologies such as mass-sheet, microlensing, and line-of-sight environment errors.
So the chapter closes only one tightened shared-base-map court aligned with the V08/V09 blind, holdout, and multi-pipeline lane. T(θ, z) may survive only as a frozen residual readout map that coordinates Δμ, δκ, and Δt_res, not as a new sky ontology or cosmological sovereign. Its clean onward value is to hand one hardened base map to the TPR–PER joint-fit court of 33.70 and the later summary courts.