We report TDE v16.9, a methodology‑first release designed to expose (and audit) a discrete spectral structure while strictly avoiding per‑particle lookup tables or post‑hoc tuning. The release cleanly separates strict forward‑only mass prediction from a clearly labeled diagnostic lattice‑projection layer. In the strict layer, masses are computed with frozen constants and deterministic rules on a discrete octant ladder (A,B,C), using the locked form Mpred=A⋅E0+B⋅E1+C⋅E2. Crucially, PDG masses are never used to assign quantum labels, rank levels, resolve multiplets, or fit parameters; they serve solely as evaluation metrics. On the PDG‑based benchmark (N=527), the strict forward‑only system achieves a coverage of 318/527 (60.34%). Within this strict subset, we obtain MAE = 1.94 MeV, RMSE = 3.37 MeV, and a mean APE = 0.108%, with a max |error| = 15.75 MeV. The remaining 209/527 (39.66%) states are unhandled strictly due to missing or non‑unique spectroscopic assignments in PDG summaries (dominated by 162 baryons and 47 mesons). For these, v16.9 introduces Octant Ladder Tagging with explicit provenance: states are marked as PDG‑explicit, PDG‑decoded, TDE‑inferred, or UNRESOLVED, allowing readers to distinguish first‑principles coverage from diagnostic completion at a glance. Finally, v16.9 produces a reproducible list of candidate missing resonances identified as vacant octant levels, specified by (quark content, JPC/JP, radial n, orbital L, A,B,C, predicted mass). The current forward generator yields 34 candidate states. For collider energies s=X, states below the kinematic ceiling are accessible; TDE flags these unoccupied octant levels as natural targets for resonance searches, without claiming guaranteed production absent a dedicated QCD production model.
Michał Karol Surowiecki (Sun,) studied this question.