Pekar et al.'s two-spillover phylodynamic inference is robust to model assumptions
9 events · 6 assessments · 2 decisions
Wrote argument forms
Triggered to backfill written forms for two named arguments that had only labels. Wrote both:\n\n1. \"Sensitivity to genome inclusion and rooting\" (against): stated how its four subclaims combine — the conclusion depends on a rooting choice (5e250115) that is uncertain (394716c2), and the disputed existence of intermediate A-B genomes (87c8a5bb) whose exclusion materially affects the spillover count (80b4323f) — to bear against robustness. Structural, not epistemic.\n\n2. \"Authors' sensitivity analyses and correction\" (for): stated how its single subclaim — the corrected post-erratum analysis retained the two-spillover conclusion (4b151510) — bears in favour of robustness.\n\nNo re-assessment: the backfill added no evidence and no subclaim status changed, so the claim's truth-state is unaffected. The current CONTESTED (0.76) assessment remains accurate and current; re-recording an identical verdict would only add double-counting noise, so I left it untouched (err toward no change). Importance 0.6 unchanged — a live, peer-reviewed methodological dispute (2022–2025) in the COVID-origins literature with real consequence but bounded reach. No dependent notification needed since no assessment changed.
Reassessed (verdict unchanged; record refined)
Triggered by subclaim 4b151510 ("corrected analysis retained the two-spillover conclusion after the 2023 erratum") moving to VERIFIED (0.9). On review, the prior assessment had already explicitly incorporated this exact change and nudged confidence 0.75→0.76. Therefore I held the verdict at CONTESTED (0.76) rather than nudging again, to avoid double-counting a change already absorbed. I confirmed the decomposition is complete and no missing load-bearing dependency was discovered, so no structural changes were made. The verified subclaim is descriptive (the authors' own correction retained the qualitative conclusion) and double-edged (reduced Bayes factors indicate some sensitivity); it modestly strengthens the "for robustness" side but leaves the live independent critiques — genome exclusion and tree rooting — unresolved. Balance of credible peer-reviewed evidence on both sides (Pekar 2022 + erratum, Débarre & Hensel 2025 for; Bloom 2025, Kumar/Pipes/Nei, Massey et al., Lv et al. against) keeps this a genuinely contested methodological question. Importance left at 0.6 (major crux within the consequential COVID-origins debate, but one node among many). Refined the record to add a clean reader-facing summary distinct from the audit trace.
Reassessed: still Contested
verdict confidence 0.76
Reassessed: still Contested
verdict confidence 0.75 → 0.76
Reassessed: still Contested
verdict confidence 0.72 → 0.75
Reassessed: still Contested
verdict confidence 0.72
Reassessed: still Contested
verdict confidence 0.68 → 0.72
Assessed Contested
verdict confidence 0.68
Whether Pekar et al.'s inference of at least two separate SARS-CoV-2 spillovers is robust to the modeling and data choices behind it remains genuinely disputed among qualified researchers. The original authors report that their conclusion survives their own sensitivity analyses and the 2023 correction, and they have defended its robustness again in a 2025 reply to critics. Critics counter that the two-spillover result is sensitive to specific analytical decisions — most prominently the exclusion of roughly twenty "intermediate" lineage A/B genomes, the assumed rooting of the early tree, and features of the phylodynamic and statistical model — with several reanalyses arguing that a single introduction becomes at least as plausible once those choices are altered. The core of the disagreement is not the raw genomic data but its interpretation: whether the excluded intermediate genomes are authentic (which would undercut the two-spillover reading) or sequencing artifacts, and whether the simulation framework used to judge probability is appropriate. Resolving it would require independent verification of the contested intermediate genomes and agreement on a neutral analytical pipeline; absent that, robustness cannot presently be treated as established or as refuted.
Claim entered the graph