How many published Kepler TTV masses hide multi-modal solutions? A catalog-scale illusory-precision audit of the strong-TTV KOI sample

Jin, Wu, Xu & Ji 2026 (arXiv:2604.09268) showed that Kepler-9's 'tight' transit-timing-variation mass ratio conceals multiple distinct solutions: a mode-first, multi-modal exploration of the TTV likelihood finds mass modes spanning $31.6$–$47.1\,M_\oplus$ for Kepler-9 b and $21.8$–$32.3\,M_\oplus$ for c, where the original analyses reported one narrow credible interval. That audit exists for ONE system. The open question: across the full strong-TTV Kepler sample (the $\sim 260$ KOIs with significant long-term TTVs in Holczer et al. 2016), what fraction of published TTV masses are similarly illusory? Task, per system: forward-model the observed mid-times (TTVFast or equivalent), run a mode-first multi-modal exploration of the mass-ratio/eccentricity posterior (not a single-chain MCMC that reports the mode it happens to land in), and report (1) the number of distinct mass-ratio modes, (2) the full multi-modal credible region vs the published single value, (3) an illusory-precision flag where the published interval excludes substantial probability mass. A homogeneous re-derived per-transit mid-time catalog falls out as a byproduct and should be published with per-transit uncertainties. The job is embarrassingly parallel and was impractical at 2013–2016 compute; it is a long-laptop-days / cluster-hours task now.

Acceptance. FULLY RESOLVES: a public, re-runnable pipeline applied to $\ge 200$ of the Holczer strong-TTV KOIs reporting per-KOI mode count, multi-modal credible regions vs the published value, and an illusory-precision flag, validated on $\ge 2$ systems with known multi-modality (e.g. Kepler-9) and $\ge 2$ well-constrained control systems (recovering their unimodality); raw mid-times re-derived or Holczer-adopted with the choice stated and per-transit uncertainties propagated. PARTIAL: the same audit on $\ge 30$ KOIs; or a validated pipeline plus a prioritized target list with per-system data-sufficiency assessment; or a demonstrated data-coverage criterion showing which systems cannot support mode analysis and why. All datasets pinned (catalog versions, MAST product identifiers + hashes) and every number regenerable by script.