{
  "design": "Design A: quote by observed step position; experiment blocked by an OutOfGas halt in the required V2 baseline market.",
  "blocked": {
    "step": "Instrumented V2 baseline batch on seeds 20000..20255",
    "error": "The required 250000-gas execution limit produced 1 OutOfGas halt. The brief invalidates any result with an OOG call; no valid paired confirmation can be reported."
  },
  "paired": {
    "n": 0
  },
  "notes": [
    "Pinned checkout: 6b6921f8d463314aa7c96234e1be25a25245f704. The supplied setup script completed and StarterStrategy validated.",
    "Fetched the specified artifact with curl, applied both specified substitutions, and computed SHA-256 28417e116f57f1029fdd1803832e852574902c1ae7253c05657f317db76da4ab, matching the requested V2.",
    "Baseline was run separately against VanillaStrategy using amm_sim_rs.run_batch(V2, VanillaStrategy, configs, 8). Configs used default_rng(seed), drawing size, rate, sigma in that order, with 10000 steps and price seed=seed, retail seed+1.",
    {
      "invalidated_baseline_batch": {
        "seeds": [
          20000,
          20255
        ],
        "n": 256,
        "gas": {
          "max_ok": 242807,
          "oog": 1,
          "other_halts": 0
        },
        "elapsed_seconds": 50.945080041885376
      }
    },
    "Gas counters were added inside amm_sim_rs/src/evm/strategy.rs: atomic maximum of successful call gas_used, OutOfGas count, and other revert/halt count. These counters aggregate calls from both contracts in this baseline-versus-Vanilla batch; the OOG is not attributed to a specific seed or contract.",
    "No gain, gas safety for a candidate, or position error rate is claimed. No frozen-candidate confirmation was run on seeds 4050000..4051999. No results or missing measurements were fabricated.",
    "Proposed mapping, not a validated recommendation: replace the V2 midpoint mapping with retailFee = 15bps + k*sizeHat for the first one or two observed prints; afterward use BASE_FEE + sigmaHat + FLOW_SIZE_COEF*lambdaHat*sizeHat. The retail formula maximizes fee times a linearized router share times flow, under equal-reserve and near-equal-price assumptions. It is an approximation, not the exact router optimum.",
    "Allowed-input inference: timestamp changes identify the first observed print and a per-step counter identifies subsequent prints. This does not identify the final print prospectively, nor prove that the first print was arbitrage. Count thresholds are probabilistic end-of-step guesses. Their error rates remain unmeasured because the required baseline batch was invalidated."
  ]
}