Experiment 9 is the completed confirmation and stress-testing suite for Future Access Without Presence (FAWP), the regime in which predictive coupling remains measurably present while steering coupling has already collapsed. Building on the flagship Experiment 8 result, this paper tests whether FAWP survives meaningful changes in ablation structure, steering definition, persistence criteria, null-threshold portability, coarse regime variation, and timing alignment relative to downstream failure. The suite contains six completed core subtests: E9. 1 ablation and robustness, E9. 2 steering-definition comparison between executed action u and intervention signal xi, E9. 3 persistence-window sweep, E9. 4 null-quantile portability sweep, E9. 5 regime generalization grid, and E9. 6 lead-lag alignment of the detector relative to failure. The paper reports only the Experiment 9 empirical suite. Broader horizon terminology follows the earlier Agency Horizon framework, while later geometric extensions are outside the present E9 result block. Headline results are as follows. In E9. 1, the detector recovered a post-zero agency horizon of tauₕₚlus = 4, a functional failure cliff of tauf = 35, and a nonempty operational detection window from tau = 4 through tau = 34, with clean null behavior. In E9. 2, both steering definitions collapsed at tau = 31, the failure cliff occurred at tauf = 36, and a nonempty pre-cliff FAWP window persisted from tau = 31 through tau = 33. E9. 3 showed stability across a broad neighborhood of persistence rules. E9. 4 preserved perfect detection across beta values 0. 90, 0. 95, 0. 975, 0. 99, and 0. 995. E9. 5 found nonzero FAWP detection in 10 of 30 tested configurations for u and 19 of 30 for xi, with the strongest region centered near (a, K) = (1. 02, 0. 8). E9. 6 showed that in the flagship aggregate the detector peak coincides with the steepest failure-rate rise and precedes the cliff, while in the strong basin it precedes both rise and cliff by positive margins. Within the studied delayed unstable-control class, the completed Experiment 9 suite supports FAWP as a reproducible operational regime rather than a single-case numerical artifact.
Clayton Ralph (Tue,) studied this question.