Public 2024 orbit samples predict the hour-scale TOI-178 d delay in three 2026 fits with fixed transit shapes

We test whether public 2024 orbit samples predict the large 2026 transit delay of TOI-178 d from an earlier constant-period clock. We propagate all 300 CDS samples with five interacting planets under stated Jacobi and stellar-mass conventions. We do not fit or reweight the samples using 2026 data. The model predicts a delay of about 63 minutes. Three main TESS fits with fixed transit shapes differ from the predicted medians by +4.18, -1.08, and -7.80 minutes. Their joint covariance score is 4.614. A predictive simulation gives a tail fraction of 0.2014. The corresponding fraction for all 11 accepted later times is 0.1955. The model timing errors are strongly correlated. A transit template with a 10% shorter duration gives a tail fraction of 0.000445 for the same three d events. Thus, the known dynamics accounts for the hour-scale shift, but the smaller residual test depends on the assumed transit shape. Time-step checks change times by at most 0.045 seconds in three tested orbit rows. An independent IAS15 calculation agrees within 0.0086 seconds at 48 checked transits. The separately archived earlier CDS timing table is not exactly reproduced: the largest median difference is 123 seconds. Its cause is unresolved. This is a retrospective test of the public sample and stated measurement errors. It does not identify another planet or validate unique planet masses. We preserve input provenance, the signed parent data package, all 300 orbit calculations, joint predictions, code, figures, and reproduction instructions.

TOI-178 · TESS · transit timing variations · dynamical prediction · public orbit samples · correlated uncertainty · reproducible astronomy