{
  "acceptance_version": "0.1",
  "object_path": "examples/the-primap-hist-national-historical-emissions-time-series-17",
  "manifest": "agent/tools.json",
  "published_results_path": "results.json",
  "notes": "All but two calls stay within the main source variant: the variants are 42 to 98MB CSVs, so the suite is built on the 44MB main one. The two exceptions, rounding-no-rounding-delta-published and rounding-no-rounding-count-published, fetch the 98MB no_rounding variant because claim 4 is computed from it and nothing in the main variant reaches it; that is one extra download per check. no_extrap and no_extrap_no_rounding remain reachable through the variant parameter but are not exercised.",
  "calls": [
    {
      "call_id": "confirmatory-2019-published",
      "tool": "divergence_test",
      "args": {},
      "expect": {
        "kind": "published_pointer",
        "result_path": "/confirmatory/test1_mann_whitney"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "fisher-2019-published",
      "tool": "exact_agreement_test",
      "args": {},
      "expect": {
        "kind": "published_pointer",
        "result_path": "/confirmatory/test2_fisher_exact"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "fisher-n2o-off-reference",
      "tool": "exact_agreement_test",
      "args": {
        "entity": "N2O"
      },
      "expect": {
        "kind": "literal",
        "value": {
          "contingency": {
            "annex1_notequal": 39,
            "annex1_equal": 1,
            "nonannex1_notequal": 78,
            "nonannex1_equal": 87
          },
          "frac_equal_annex1": 0.025,
          "frac_equal_non_annex1": 0.5272727272727272,
          "odds_ratio": 43.5,
          "p_value": 3.7404479923548464e-10
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The reporting-existence asymmetry holds for N2O as well, and slightly harder than for the aggregate basket (odds ratio 43.5 against 38.1). 39 of 40 Annex I countries have differing scenarios; 87 of 165 non-Annex I countries have identical ones. The proxy is not an artefact of the aggregate Kyoto basket."
    },
    {
      "call_id": "temporal-1990-published",
      "tool": "divergence_test",
      "args": {
        "year": "1990"
      },
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/temporal/1990"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "temporal-2000-published",
      "tool": "divergence_test",
      "args": {
        "year": "2000"
      },
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/temporal/2000"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "temporal-2010-published",
      "tool": "divergence_test",
      "args": {
        "year": "2010"
      },
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/temporal/2010"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "gas-co2-published",
      "tool": "divergence_test",
      "args": {
        "entity": "CO2"
      },
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/by_gas/CO2"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "gas-ch4-published",
      "tool": "divergence_test",
      "args": {
        "entity": "CH4"
      },
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/by_gas/CH4"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "gas-n2o-untested-off-reference",
      "tool": "divergence_test",
      "args": {
        "entity": "N2O"
      },
      "expect": {
        "kind": "literal",
        "value": {
          "n_annex1": 40,
          "n_non_annex1": 165,
          "median_abs_rel_diff_annex1": 0.2290229885057471,
          "median_abs_rel_diff_non_annex1": 0.0,
          "mann_whitney_u": 4753.5,
          "p_value": 6.851744856393609e-06,
          "cliffs_delta": 0.44045454545454543,
          "cliffs_delta_ci95": [
            0.29939015151515147,
            0.5727310606060606
          ]
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "N2O is a Kyoto-basket gas the published run never tests, and it carries the LARGEST effect of any entity examined: delta 0.440, against CH4 at 0.416 and the headline aggregate at 0.383. The median Annex I divergence is 22.9%, with a non-Annex I median of exactly zero. This strengthens claim 3 rather than qualifying it: the gas robustness sampled two of three main gases and happened to omit the strongest."
    },
    {
      "call_id": "temporal-1970-off-reference",
      "tool": "divergence_test",
      "args": {
        "year": "1970"
      },
      "select": "/cliffs_delta",
      "expect": {
        "kind": "literal",
        "value": 0.2427710843373494
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The series runs from 1750 and the published temporal robustness starts at 1990. At 1970 the effect is present but weaker (delta 0.243, p = 0.014), consistent with the monotone rise the reported years show (0.214 in 1990 to 0.383 in 2019). The divergence grows as national inventory reporting matures."
    },
    {
      "call_id": "bootstrap-stability-off-reference",
      "tool": "divergence_test",
      "args": {
        "n_boot": 40000,
        "seed": 7
      },
      "select": "/cliffs_delta_ci95",
      "expect": {
        "kind": "literal",
        "value": [
          0.23764307228915663,
          0.5225903614457832
        ]
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Four times the resamples on a different seed reproduces the published CI [0.2346, 0.5230] to within 0.004 at each end. The interval reflects the data, not the 10,000-resample default."
    },
    {
      "call_id": "exact-agreement-count-published",
      "tool": "exact_agreement_count",
      "args": {},
      "select": "/n_exact_equal_total",
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/rounding/main/n_exact_equal_total"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "exact-agreement-count-1970-off-reference",
      "tool": "exact_agreement_count",
      "args": {
        "year": "1970"
      },
      "select": "/n_exact_equal_total",
      "expect": {
        "kind": "literal",
        "value": 83
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The same count as in 2019, 83, but not the same countries: Honduras and Japan agree exactly in 1970 and not in 2019, China and Luxembourg the reverse. Nor is the count independent of the year: the same tool gives 81 for 1980. What the two years do share is the asymmetry the Fisher test measures: in each, exactly one of the 40 Annex I countries is in the exact-agreement group (Japan in 1970, Luxembourg in 2019)."
    },
    {
      "call_id": "rounding-no-rounding-delta-published",
      "tool": "divergence_test",
      "args": {
        "variant": "no_rounding"
      },
      "select": "/cliffs_delta",
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/rounding/no_rounding/cliffs_delta"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Claim 4 through the tool surface: the 2019 primary test recomputed from the no_rounding variant, delta 0.395 against 0.383 from the rounded main file."
    },
    {
      "call_id": "rounding-no-rounding-count-published",
      "tool": "exact_agreement_count",
      "args": {
        "variant": "no_rounding"
      },
      "select": "/n_exact_equal_total",
      "expect": {
        "kind": "published_pointer",
        "result_path": "/robustness/rounding/no_rounding/n_exact_equal_total"
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Claim 4's second number: without rounding, exact agreement falls from 83 countries to 80. Rounding creates the three extra matches (China, Luxembourg and the Holy See) and removes none. Shares the no_rounding download with the call above."
    }
  ]
}