{
  "kind": "exploratory-probe",
  "not_a_receipt": true,
  "question": "are single-generation baseline measurements stable draw-to-draw today?",
  "started": "2026-08-28T20:56:08.917Z",
  "finished": "2026-08-28T21:27:16.196Z",
  "surface": "claude-cli (subscription; metered spend $0)",
  "judge": {
    "model": "claude-haiku-4-5",
    "samples": 5,
    "timeout_ms": 120000,
    "sampling": "surface-controlled"
  },
  "arm": "baseline only — no skill text",
  "run_note": "Run 1 aborted on a judge-call timeout after 4 completed draws on case 1; those 4 are carried in, labelled salvaged_from_run1. The instrument is identical across both runs.",
  "assumption": "These draws measure noise TODAY. Ruling luck out for the #005 gap assumes the noise level itself did not change between 2026-08-18 and now. #005 took ONE generation per cell, so its receipts cannot confirm that assumption either way.",
  "tonight_values_are": "DECLARED by the operator from the report-006 session; not verifiable from this probe.",
  "results": [
    {
      "label": "severity-labeled-findings",
      "skill": "code-review-and-quality",
      "model": "claude-fable-5",
      "pinned_005": 0.3,
      "pinned_005_samples": [
        0.3,
        0.3,
        0.3,
        0.3,
        0.3
      ],
      "pinned_005_sd": 0,
      "tonight_declared": 0.696,
      "rubric_matches_report": true,
      "report_rubric_hash": "a953d281f6a998f2b332172b431e3fd36792516ceefbc87ca00bab9ebcf015b0",
      "probe_rubric_hash": "a953d281f6a998f2b332172b431e3fd36792516ceefbc87ca00bab9ebcf015b0",
      "report_judge_model": "claude-haiku-4-5",
      "probe_judge_model": "claude-haiku-4-5",
      "report_surface": "claude-cli",
      "report_run": "2026-08-18T19:01:02.260Z",
      "draws_completed": 10,
      "draws_failed": 0,
      "salvaged_from_run1": 4,
      "rows": [
        {
          "draw": 3,
          "mean": 0.69,
          "stddev": 0.105,
          "samples": [
            0.65,
            0.68,
            0.6,
            0.87,
            0.65
          ],
          "salvaged_from": "run 1 (aborted)",
          "case": "severity-labeled-findings"
        },
        {
          "draw": 1,
          "mean": 0.626,
          "stddev": 0.037,
          "samples": [
            0.6,
            0.6,
            0.65,
            0.68,
            0.6
          ],
          "salvaged_from": "run 1 (aborted)",
          "case": "severity-labeled-findings"
        },
        {
          "draw": 2,
          "mean": 0.732,
          "stddev": 0.091,
          "samples": [
            0.85,
            0.63,
            0.76,
            0.77,
            0.65
          ],
          "salvaged_from": "run 1 (aborted)",
          "case": "severity-labeled-findings"
        },
        {
          "draw": 4,
          "mean": 0.3,
          "stddev": 0,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.3,
            0.3
          ],
          "salvaged_from": "run 1 (aborted)",
          "case": "severity-labeled-findings"
        },
        {
          "case": "severity-labeled-findings",
          "draw": 5,
          "mean": 0.506,
          "stddev": 0.083247,
          "samples": [
            0.48,
            0.5,
            0.45,
            0.45,
            0.65
          ],
          "generation_hash": "2ba3f1eaee6ac5b2cce17639540ceed4f94fb65bb963af1594e5058c7130e29d",
          "generation_chars": 3842,
          "wall_ms": 303451,
          "attempt": 1
        },
        {
          "case": "severity-labeled-findings",
          "draw": 6,
          "mean": 0.686,
          "stddev": 0.037148,
          "samples": [
            0.7,
            0.71,
            0.62,
            0.7,
            0.7
          ],
          "generation_hash": "1954afc5bca106feb450b80f37622655148224ac56ca88e3dd6be570981c099a",
          "generation_chars": 4332,
          "wall_ms": 374053,
          "attempt": 2
        },
        {
          "case": "severity-labeled-findings",
          "draw": 7,
          "mean": 0.774,
          "stddev": 0.064265,
          "samples": [
            0.7,
            0.8,
            0.87,
            0.75,
            0.75
          ],
          "generation_hash": "cc72233955c76bc8ddca937bc037e6c6d591a89a40f92ef2a1b3bc6e946d2129",
          "generation_chars": 3691,
          "wall_ms": 347905,
          "attempt": 1
        },
        {
          "case": "severity-labeled-findings",
          "draw": 8,
          "mean": 0.808,
          "stddev": 0.101094,
          "samples": [
            0.68,
            0.85,
            0.85,
            0.93,
            0.73
          ],
          "generation_hash": "9c9d5c8544709397224dbe594cb1338d8f53da30d269ace8bc01aa4d305b8b3d",
          "generation_chars": 4070,
          "wall_ms": 362524,
          "attempt": 1
        },
        {
          "case": "severity-labeled-findings",
          "draw": 9,
          "mean": 0.734,
          "stddev": 0.068411,
          "samples": [
            0.7,
            0.85,
            0.74,
            0.68,
            0.7
          ],
          "generation_hash": "652037d38c182e591bc5d1aee5d3bef84a19ad0bbc6159e1ed84f7860e52ef07",
          "generation_chars": 3697,
          "wall_ms": 430210,
          "attempt": 1
        },
        {
          "case": "severity-labeled-findings",
          "draw": 10,
          "mean": 0.3,
          "stddev": 0,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.3,
            0.3
          ],
          "generation_hash": "f3cec8e4e6db5dda6989eaed9c85f526dde099852f6c3a4e23b3bf8ea0685972",
          "generation_chars": 2897,
          "wall_ms": 145187,
          "attempt": 1
        }
      ],
      "draws": [
        0.69,
        0.626,
        0.732,
        0.3,
        0.506,
        0.686,
        0.774,
        0.808,
        0.734,
        0.3
      ],
      "mean": 0.6156,
      "stddev_across_draws": 0.1860389445489543,
      "mean_judge_stddev_within_draw": 0.058717,
      "min": 0.3,
      "max": 0.808,
      "containment_005": {
        "value": 0.3,
        "inRange": true,
        "within1sd": false,
        "within2sd": true,
        "z": -1.6964189985336808,
        "distanceFromNearestDraw": 0
      },
      "containment_tonight": {
        "value": 0.696,
        "inRange": true,
        "within1sd": true,
        "within2sd": true,
        "z": 0.43216757757321855,
        "distanceFromNearestDraw": 0
      },
      "verdict": {
        "rule": "b",
        "reads": "UNSTABLE SINGLE DRAW — generation-level noise explains the gap"
      }
    },
    {
      "label": "commit-message-conventional-type",
      "skill": "git-workflow-and-versioning",
      "model": "claude-sonnet-5",
      "pinned_005": 0.282,
      "pinned_005_samples": [
        0.28,
        0.28,
        0.27,
        0.28,
        0.3
      ],
      "pinned_005_sd": 0.010954,
      "tonight_declared": 0.852,
      "rubric_matches_report": true,
      "report_rubric_hash": "c17f2f214cec220ac31af3cfe88a52393c254cea8f9a94fc16e668f668fa88c3",
      "probe_rubric_hash": "c17f2f214cec220ac31af3cfe88a52393c254cea8f9a94fc16e668f668fa88c3",
      "report_judge_model": "claude-haiku-4-5",
      "probe_judge_model": "claude-haiku-4-5",
      "report_surface": "claude-cli",
      "report_run": "2026-08-18T19:01:02.260Z",
      "draws_completed": 10,
      "draws_failed": 0,
      "salvaged_from_run1": 0,
      "rows": [
        {
          "case": "commit-message-conventional-type",
          "draw": 1,
          "mean": 0.3,
          "stddev": 0,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.3,
            0.3
          ],
          "generation_hash": "ba946f6649e40920c532f4a8b4f58b8af6bb5a560d4a2a027ec6a6e50ce4bf10",
          "generation_chars": 311,
          "wall_ms": 157983,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 2,
          "mean": 0.29,
          "stddev": 0.022361,
          "samples": [
            0.25,
            0.3,
            0.3,
            0.3,
            0.3
          ],
          "generation_hash": "3366fd5ee510324ecb87e4f3e9d6d628561397b96e8f9e9627f02ec5ba06184e",
          "generation_chars": 294,
          "wall_ms": 81135,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 3,
          "mean": 0.292,
          "stddev": 0.010954,
          "samples": [
            0.3,
            0.3,
            0.28,
            0.3,
            0.28
          ],
          "generation_hash": "9d559aed97f21f4b03b3e30f63fedae9cbd081a92be85b9a39c7ad01f5b0762d",
          "generation_chars": 242,
          "wall_ms": 109772,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 4,
          "mean": 0.286,
          "stddev": 0.021909,
          "samples": [
            0.28,
            0.3,
            0.3,
            0.25,
            0.3
          ],
          "generation_hash": "ac993172fbf6a337af99cca69d53bd635257ecdde50e6f1f033623341c2bb088",
          "generation_chars": 258,
          "wall_ms": 180495,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 5,
          "mean": 0.294,
          "stddev": 0.013416,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.27,
            0.3
          ],
          "generation_hash": "ec441d583807fcafadf3e1e0acb78386d57415da7b6aedd52295a19b9ac383cd",
          "generation_chars": 259,
          "wall_ms": 68284,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 6,
          "mean": 0.29,
          "stddev": 0.022361,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.3,
            0.25
          ],
          "generation_hash": "8bb6a47d8c8fa6e5ae96b994cfdcd82785cca79ce49a190fb350af322313fef1",
          "generation_chars": 304,
          "wall_ms": 63212,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 7,
          "mean": 0.296,
          "stddev": 0.008944,
          "samples": [
            0.3,
            0.3,
            0.3,
            0.28,
            0.3
          ],
          "generation_hash": "574477ff9e59b29d7c1d5404ff332dbefc554790a65d71956789584a8527f2bd",
          "generation_chars": 253,
          "wall_ms": 124772,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 8,
          "mean": 0.214,
          "stddev": 0.121367,
          "samples": [
            0.3,
            0.25,
            0.25,
            0.27,
            0
          ],
          "generation_hash": "9dee0a436b0ebd87785cc1159e0a2b1cb9c464e98b56a4583763724bd3de69f0",
          "generation_chars": 344,
          "wall_ms": 227945,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 9,
          "mean": 0.856,
          "stddev": 0.005477,
          "samples": [
            0.86,
            0.86,
            0.85,
            0.85,
            0.86
          ],
          "generation_hash": "c95b4d43ce85c7fef6c23cab864b02fadd18e2972b1a3cec4fe295b92a9e230c",
          "generation_chars": 343,
          "wall_ms": 100473,
          "attempt": 1
        },
        {
          "case": "commit-message-conventional-type",
          "draw": 10,
          "mean": 0.278,
          "stddev": 0.017889,
          "samples": [
            0.3,
            0.28,
            0.28,
            0.25,
            0.28
          ],
          "generation_hash": "cd286ad184741ff56ca980c371470fe66c19814b099a09c5cdd61c4676dbed05",
          "generation_chars": 390,
          "wall_ms": 160536,
          "attempt": 1
        }
      ],
      "draws": [
        0.3,
        0.29,
        0.292,
        0.286,
        0.294,
        0.29,
        0.296,
        0.214,
        0.856,
        0.278
      ],
      "mean": 0.3396,
      "stddev_across_draws": 0.18313516562607215,
      "mean_judge_stddev_within_draw": 0.024468,
      "min": 0.214,
      "max": 0.856,
      "containment_005": {
        "value": 0.282,
        "inRange": true,
        "within1sd": true,
        "within2sd": true,
        "z": -0.31452178942851694,
        "distanceFromNearestDraw": 0
      },
      "containment_tonight": {
        "value": 0.852,
        "inRange": true,
        "within1sd": false,
        "within2sd": false,
        "z": 2.7979334184578466,
        "distanceFromNearestDraw": 0
      },
      "verdict": {
        "rule": "b",
        "reads": "UNSTABLE SINGLE DRAW — generation-level noise explains the gap"
      }
    }
  ]
}