This robustness addendum extends the frozen VES Stage 5B SPARC diagnostic analysis by testing whether the reported improvement over a fixed baryonic baseline is stable under galaxy-level resampling and metadata-score perturbations. The analysis uses the released metadata-merged Stage 5B table and evaluates the per-galaxy quantity ΔRMSE = RMSE (Stage5B) − RMSE (baryon), where negative values indicate improvement over the frozen baryonic baseline. Three diagnostic tests are performed: 10, 000 bootstrap resamples of the galaxy-level ΔRMSE distribution, leave-one-galaxy-out jackknife resampling, and 5, 000 Monte Carlo perturbations of the empirical Qₘeta purity-score weights by ±30 percent followed by renormalization. In the metadata-clean subset of 94 galaxies, the bootstrap mean ΔRMSE remains strongly negative, with a 95 percent confidence interval of approximately −11. 22, −8. 71. The bootstrap median interval is also fully negative. In the medium-risk subset of 78 galaxies, the mean ΔRMSE remains negative with a 95 percent confidence interval of approximately −3. 80, −2. 50. Jackknife tests show that the mean improvement is not driven by any single dominant galaxy. The Qₘeta sensitivity test shows that the clean-subset improvement fraction remains equal to 1. 0 across all tested weight perturbations, while the clean median ΔRMSE remains stably negative. The remaining failures are concentrated in systems with baryonic inner overshoot, sparse rotation-curve sampling, low SPARC quality, or elevated metadata reconstruction risk. These results strengthen the empirical status of VES Stage 5B as a statistically robust frozen diagnostic model on the SPARC clean and medium-risk subsets. They do not constitute physical validation of VES and do not replace direct comparison against RAR/MOND-like relations, NFW or Burkert halo baselines, variable mass-to-light ratio tests, or a full elliptic/radial field solver.
Mikheil Rusishvili (Sun,) studied this question.