{
  "harness": "sj-harness 0.6.0",
  "job": "job:de835f4af260d06297ff04ebf1c55662",
  "kind": "adversarial_review",
  "bundle": "sha256:6399f974259dd6ed5036a79a199face2bdadd5453eb638a0a7fdcae2fce16194",
  "claims": [
    {
      "local_id": "C1",
      "claim_id": "claim:ae394b886c29df45e2c8e0ac12a321c7c1e17ee0dceb21e2881f88917eb2b986",
      "verdict": "major_issues",
      "significance": "known",
      "reason": "Values reproduce and overestimation is systematic, but error at s=0.1 is ~3% for every N (3.31/3.26/3.09/3.00 for N=20/50/100/200), so it does not grow as N contracts; deleterious errors on the same grid (10-25%) are omitted from the stated range."
    },
    {
      "local_id": "C2",
      "claim_id": "claim:785c938aee0e81aa315e29fc645918ee3ef3633b6e05f290ffa9c96efb22757b",
      "verdict": "major_issues",
      "significance": "known",
      "reason": "3.02% max reproduces only with s capped at 0.05; the grid's s=0.10 gives 6.1-7.2% asymmetry, and asymmetry grows with N at fixed s. A 9-point grid does not prove symmetry; the diffusion symmetry is Maruyama 1974, not Ohta 1973."
    },
    {
      "local_id": "C3",
      "claim_id": "claim:1bf5f9acf5f401b8f06ea8d55190b483522145ec0be632f698d2abf01edae72e",
      "verdict": "minor_issues",
      "significance": "known",
      "reason": "Residual ~1e-17 and tau=196.74 reproduce; 2N is a large-N limit for haploid N copies, not a bound; diffusion value at p=1/N is 199.0, so the like-for-like gap is 1.1%; diploid would be 4N."
    }
  ]
}