{
 "schema": "nagi-arena-v2-update-v1",
 "meta": {
  "title": "Nagi Burst V2, Nagi Burst V1 and Jev on the frozen Arena eval seeds",
  "status": "frozen eval (M4)",
  "split": "eval",
  "tick_ms": 250,
  "v2_reply_profile": "nominal 1-tick (decision for boundary t applies at t+1)",
  "model_sha256": "a24e1158b00a59ad1fc103001c301981aa94da033e59fcee7b645ae2deef5624",
  "generated_by": "site/arena/build/make_v2.py",
  "sources_sha256": {
   "arena_v2.json": "edfb183fb15543e7deb9371a61b01b59291c54e8ec04e7c6f4a466462bfed15b",
   "CANDIDATE.json": "28fe5de62675e93e03ba0b82d2e3e6f1f8a52d98423ec0415759f8c6145c65d3",
   "jev_pair_check.json": "13cc6835c0304b58ff178b5ba8174e085bd41b7200499e5cf764282b3d2ef12e",
   "jev1s_snake.json": "e7e0ce965d655ee111903a763a2df496f161709b95240b61fcfb40666f812ad2",
   "jev1s_rr.json": "5cd53676b8db2b82551685ec179e15e058bc1515202c3f056849d95e46352be5",
   "jev1s_booster.json": "eb9c6ad3b1803229148e4b14d2dc166d985553732239c11b937d87c3859992bc",
   "arena_v2_videos.json": "ca9abec18f1e89eb06b9f138d823e9bbcb7a05f7d792bc8adcdff0aa62bfe7ab"
  },
  "replay_pack": {
   "exact": 72,
   "replayed": 72
  },
  "jev_pair_check_image": {
   "tag": "bv2n-rr:py31112-np246",
   "id": "sha256:bf9b6dae746787f7b3eacbce8f62446b93ea90991b0aad9ebca5ee468d61bb84",
   "platform": "linux/amd64 under Rosetta 2 on Apple M2 (SSE4.2; no AVX/FMA)"
  },
  "jev_engine_identity": "src/arena, src/bv2n/contract, src/bv2n/data, scripts/nc_r_arena_rt.py and scripts/fair_public_jev.py are byte-identical (empty git diff) between the Jev re-play harness commit c938276ae9a37af3fd25166ef4c7fa60e14e2d9d and the M4 commit de5335e36cfd62a27158e37dd547cf45326fdec2"
 },
 "players": {
  "v2": {
   "label": "Nagi Burst V2",
   "sub": "release policies: Snake Rush = SDK search pack (no learned core); Rotorwash-Ramp and Booster Gauntlet = the core's game head (no lookahead)"
  },
  "v1": {
   "label": "Nagi Burst V1",
   "sub": "Nagi-ENORMOUS-Burst (27B), pinned eval records (PREREG_V2G §3)"
  },
  "jev": {
   "label": "Jev",
   "sub": "jev-1.13.0, vendor API, single-stream re-play (PREREG_V2G §3, descriptive)"
  }
 },
 "claim_rule": {
  "v2_vs_v1": "Gate A (PREREG_V2G §5.3): PASS iff the t lower 97.5% bound of W > 1/2 and the exact sign test p <= 0.025, per game.",
  "jev": "Descriptive only (PREREG_V2G §3, §5.6, R3.15): Jev contrasts were preregistered as descriptive, on identical seeds, with no test and no claim words. Only counts are shown."
 },
 "games": [
  {
   "game": "snake",
   "label": "Snake Rush",
   "metric": "survival time (s, capped at 60 s)",
   "unit": "s",
   "decimals": 2,
   "rules_id": "snake_rush:rush-2:base:announced:L0",
   "n": 24,
   "seeds": "10–33",
   "means": {
    "v2": 50.65625,
    "v1": 12.572916666666666,
    "jev": 10.354166666666666
   },
   "ci95": {
    "v2": [
     45.37256690715566,
     55.93993309284434
    ],
    "v1": [
     9.815090818336474,
     15.330742514996858
    ],
    "jev": [
     7.838645926560162,
     12.86968740677317
    ]
   },
   "pairing_jev": "PAIRED",
   "pairing_detail": {
    "same_index_and_seed": 24,
    "rules_id_equal": true,
    "record_sha256_pinned": 24,
    "replay_exact_local_image": 24,
    "score_recomputed_equal": 24,
    "prompts_regenerated": 966,
    "prompt_mismatch": 0,
    "replay_exact_in_venue": true
   },
   "jev_profile": {
    "model": "jev-1.13.0",
    "client_streams": 1,
    "one_tick_share": 0.9958592132505176,
    "latency_p50_ms": 171.991,
    "latency_p95_ms": 210.502,
    "nonok_replies": 0
   },
   "pairs": [
    {
     "n": 24,
     "W": 1.0,
     "W_ci95": [
      1.0,
      1.0
     ],
     "sd": 0.0,
     "t_lb": 1.0,
     "p": 5.960464477539063e-08,
     "wins": 24,
     "ties": 0,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "v1",
     "a_vs": "v1",
     "claim": "gate A",
     "result": "PASS"
    },
    {
     "n": 24,
     "W": 1.0,
     "W_ci95": [
      1.0,
      1.0
     ],
     "sd": 0.0,
     "t_lb": 1.0,
     "p": 5.960464477539063e-08,
     "wins": 24,
     "ties": 0,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    },
    {
     "n": 24,
     "W": 0.6875,
     "W_ci95": [
      0.49247337130562663,
      0.8825266286943734
     ],
     "sd": 0.4618606038528339,
     "t_lb": 0.49247337130562663,
     "p": 0.04656982421875,
     "wins": 16,
     "ties": 1,
     "losses": 7,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": false,
     "a": "v1",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    }
   ],
   "per_seed": [
    {
     "index": 10,
     "seed": 1729125481,
     "v2": 60.0,
     "v1": 19.5,
     "jev": 6.5
    },
    {
     "index": 11,
     "seed": 532168729,
     "v2": 36.5,
     "v1": 9.5,
     "jev": 8.5
    },
    {
     "index": 12,
     "seed": 957618623,
     "v2": 44.75,
     "v1": 5.5,
     "jev": 19.0
    },
    {
     "index": 13,
     "seed": 206514868,
     "v2": 60.0,
     "v1": 3.0,
     "jev": 10.0
    },
    {
     "index": 14,
     "seed": 626850533,
     "v2": 59.75,
     "v1": 4.0,
     "jev": 3.0
    },
    {
     "index": 15,
     "seed": 1150166326,
     "v2": 59.25,
     "v1": 21.25,
     "jev": 18.0
    },
    {
     "index": 16,
     "seed": 928849423,
     "v2": 20.0,
     "v1": 6.0,
     "jev": 10.0
    },
    {
     "index": 17,
     "seed": 1115084172,
     "v2": 54.5,
     "v1": 13.0,
     "jev": 11.5
    },
    {
     "index": 18,
     "seed": 1954269682,
     "v2": 60.0,
     "v1": 25.75,
     "jev": 4.5
    },
    {
     "index": 19,
     "seed": 220883336,
     "v2": 34.5,
     "v1": 6.5,
     "jev": 4.0
    },
    {
     "index": 20,
     "seed": 134963249,
     "v2": 18.75,
     "v1": 12.0,
     "jev": 16.0
    },
    {
     "index": 21,
     "seed": 898002520,
     "v2": 58.0,
     "v1": 11.5,
     "jev": 19.5
    },
    {
     "index": 22,
     "seed": 385912263,
     "v2": 60.0,
     "v1": 7.0,
     "jev": 6.5
    },
    {
     "index": 23,
     "seed": 81718780,
     "v2": 60.0,
     "v1": 12.0,
     "jev": 10.5
    },
    {
     "index": 24,
     "seed": 2125293014,
     "v2": 52.5,
     "v1": 18.5,
     "jev": 7.0
    },
    {
     "index": 25,
     "seed": 791674996,
     "v2": 51.5,
     "v1": 9.5,
     "jev": 6.5
    },
    {
     "index": 26,
     "seed": 2000442638,
     "v2": 60.0,
     "v1": 20.5,
     "jev": 4.5
    },
    {
     "index": 27,
     "seed": 1638123596,
     "v2": 46.0,
     "v1": 6.0,
     "jev": 12.5
    },
    {
     "index": 28,
     "seed": 1475788145,
     "v2": 58.25,
     "v1": 18.0,
     "jev": 5.0
    },
    {
     "index": 29,
     "seed": 1851417534,
     "v2": 60.0,
     "v1": 11.5,
     "jev": 25.0
    },
    {
     "index": 30,
     "seed": 140764203,
     "v2": 39.5,
     "v1": 25.25,
     "jev": 17.0
    },
    {
     "index": 31,
     "seed": 780576269,
     "v2": 60.0,
     "v1": 11.5,
     "jev": 11.5
    },
    {
     "index": 32,
     "seed": 1805007521,
     "v2": 48.25,
     "v1": 11.0,
     "jev": 3.5
    },
    {
     "index": 33,
     "seed": 1783304786,
     "v2": 53.75,
     "v1": 13.5,
     "jev": 8.5
    }
   ],
   "replay": "https://github.com/nagisanzenin/nagi/tree/main/bench/burst_v2/replays/snake",
   "video": "media/v2-SnakeRush-V2-V1-Jev-seed32.mp4",
   "poster": "media/v2-SnakeRush-V2-V1-Jev-seed32.jpg",
   "video_sha256": "5896c66f5a34fbf8d88736b3597c1afe0aa4196df537c6b443c3729f10944442",
   "video_seed_index": 32,
   "video_seed_rule": "per game, margin_i = V2_i - Jev_i (V2 from docs/burst_v2_general/receipts/m4/<cell>/v2g_<game>.csv, Jev from docs/burst_v2_next/p3/jev1s_<game>.json per_seed[].score), rounded 1e-9"
  },
  {
   "game": "rotorwash_ramp",
   "label": "Rotorwash-Ramp",
   "metric": "K3-RR2.1 score Y",
   "unit": "",
   "decimals": 3,
   "rules_id": "rotorwash_ramp_v31:ramp2.1:base:announced:L1",
   "n": 24,
   "seeds": "34–57",
   "means": {
    "v2": 1.0479166666666668,
    "v1": 0.3509151777424368,
    "jev": 0.24375
   },
   "ci95": {
    "v2": [
     1.0297873821602508,
     1.0660459511730829
    ],
    "v1": [
     0.2976156750425404,
     0.4042146804423332
    ],
    "jev": [
     0.19154400848416375,
     0.29595599151583624
    ]
   },
   "pairing_jev": "PAIRED",
   "pairing_detail": {
    "same_index_and_seed": 24,
    "rules_id_equal": true,
    "record_sha256_pinned": 24,
    "replay_exact_local_image": 8,
    "score_recomputed_equal": 24,
    "prompts_regenerated": 1179,
    "prompt_mismatch": 0,
    "replay_exact_in_venue": true
   },
   "jev_profile": {
    "model": "jev-1.13.0",
    "client_streams": 1,
    "one_tick_share": 0.49363867684478374,
    "latency_p50_ms": 248.887,
    "latency_p95_ms": 300.012,
    "nonok_replies": 2
   },
   "pairs": [
    {
     "n": 24,
     "W": 1.0,
     "W_ci95": [
      1.0,
      1.0
     ],
     "sd": 0.0,
     "t_lb": 1.0,
     "p": 5.960464477539063e-08,
     "wins": 24,
     "ties": 0,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "v1",
     "a_vs": "v1",
     "claim": "gate A",
     "result": "PASS"
    },
    {
     "n": 24,
     "W": 1.0,
     "W_ci95": [
      1.0,
      1.0
     ],
     "sd": 0.0,
     "t_lb": 1.0,
     "p": 5.960464477539063e-08,
     "wins": 24,
     "ties": 0,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    },
    {
     "n": 24,
     "W": 0.6666666666666666,
     "W_ci95": [
      0.46332871687006694,
      0.8700046164632663
     ],
     "sd": 0.48154341234307685,
     "t_lb": 0.46332871687006694,
     "p": 0.07579481601715088,
     "wins": 16,
     "ties": 0,
     "losses": 8,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": false,
     "a": "v1",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    }
   ],
   "per_seed": [
    {
     "index": 34,
     "seed": 1573483930,
     "v2": 1.05,
     "v1": 0.1,
     "jev": 0.3
    },
    {
     "index": 35,
     "seed": 1276506956,
     "v2": 1.15,
     "v1": 0.5,
     "jev": 0.1
    },
    {
     "index": 36,
     "seed": 688035671,
     "v2": 1.05,
     "v1": 0.45,
     "jev": 0.1
    },
    {
     "index": 37,
     "seed": 1829884556,
     "v2": 1.0,
     "v1": 0.5,
     "jev": 0.35
    },
    {
     "index": 38,
     "seed": 134911655,
     "v2": 1.0,
     "v1": 0.45,
     "jev": 0.2
    },
    {
     "index": 39,
     "seed": 1886341571,
     "v2": 1.05,
     "v1": 0.4,
     "jev": 0.15
    },
    {
     "index": 40,
     "seed": 812869557,
     "v2": 1.0,
     "v1": 0.45,
     "jev": 0.1
    },
    {
     "index": 41,
     "seed": 947752742,
     "v2": 1.05,
     "v1": 0.25,
     "jev": 0.3
    },
    {
     "index": 42,
     "seed": 1812224925,
     "v2": 1.1,
     "v1": 0.25,
     "jev": 0.4
    },
    {
     "index": 43,
     "seed": 1687470739,
     "v2": 1.0,
     "v1": 0.25,
     "jev": 0.5
    },
    {
     "index": 44,
     "seed": 1181143952,
     "v2": 1.05,
     "v1": 0.3,
     "jev": 0.25
    },
    {
     "index": 45,
     "seed": 1205738295,
     "v2": 1.1,
     "v1": 0.4,
     "jev": 0.1
    },
    {
     "index": 46,
     "seed": 543573612,
     "v2": 1.0,
     "v1": 0.3,
     "jev": 0.2
    },
    {
     "index": 47,
     "seed": 704075304,
     "v2": 1.05,
     "v1": 0.2,
     "jev": 0.3
    },
    {
     "index": 48,
     "seed": 2026972281,
     "v2": 1.05,
     "v1": 0.3,
     "jev": 0.15
    },
    {
     "index": 49,
     "seed": 433784576,
     "v2": 1.1,
     "v1": 0.5,
     "jev": 0.3
    },
    {
     "index": 50,
     "seed": 1269055079,
     "v2": 1.05,
     "v1": 0.2,
     "jev": 0.25
    },
    {
     "index": 51,
     "seed": 1513302480,
     "v2": 1.05,
     "v1": 0.25,
     "jev": 0.5
    },
    {
     "index": 52,
     "seed": 932654653,
     "v2": 1.1,
     "v1": 0.5,
     "jev": 0.15
    },
    {
     "index": 53,
     "seed": 193623682,
     "v2": 1.05,
     "v1": 0.6219642658184833,
     "jev": 0.3
    },
    {
     "index": 54,
     "seed": 1025847886,
     "v2": 1.0,
     "v1": 0.35,
     "jev": 0.1
    },
    {
     "index": 55,
     "seed": 206109221,
     "v2": 1.0,
     "v1": 0.25,
     "jev": 0.4
    },
    {
     "index": 56,
     "seed": 1929627173,
     "v2": 1.1,
     "v1": 0.35,
     "jev": 0.15
    },
    {
     "index": 57,
     "seed": 1257136617,
     "v2": 1.0,
     "v1": 0.3,
     "jev": 0.2
    }
   ],
   "replay": "https://github.com/nagisanzenin/nagi/tree/main/bench/burst_v2/replays/rotorwash_ramp",
   "video": "media/v2-RotorwashRamp-V2-V1-Jev-seed44.mp4",
   "poster": "media/v2-RotorwashRamp-V2-V1-Jev-seed44.jpg",
   "video_sha256": "f8d3e6a75e22176c8441646d2984df15f6299682545d488d1c90b91b8129ebff",
   "video_seed_index": 44,
   "video_seed_rule": "per game, margin_i = V2_i - Jev_i (V2 from docs/burst_v2_general/receipts/m4/<cell>/v2g_<game>.csv, Jev from docs/burst_v2_next/p3/jev1s_<game>.json per_seed[].score), rounded 1e-9"
  },
  {
   "game": "booster_gauntlet",
   "label": "Booster Gauntlet",
   "metric": "K3 landing score Y (first hop)",
   "unit": "",
   "decimals": 3,
   "rules_id": "booster_gauntlet:g1.2:base:announced:L0:250ms",
   "n": 24,
   "seeds": "34–57",
   "means": {
    "v2": 0.9583333333333334,
    "v1": 0.0,
    "jev": 0.0
   },
   "ci95": {
    "v2": [
     0.8721392662325397,
     1.044527400434127
    ],
    "v1": [
     0.0,
     0.0
    ],
    "jev": [
     0.0,
     0.0
    ]
   },
   "pairing_jev": "PAIRED",
   "pairing_detail": {
    "same_index_and_seed": 24,
    "rules_id_equal": true,
    "record_sha256_pinned": 24,
    "replay_exact_local_image": 24,
    "score_recomputed_equal": 24,
    "prompts_regenerated": 1013,
    "prompt_mismatch": 0,
    "replay_exact_in_venue": true
   },
   "jev_profile": {
    "model": "jev-1.13.0",
    "client_streams": 1,
    "one_tick_share": 0.9486673247778875,
    "latency_p50_ms": 204.163,
    "latency_p95_ms": 249.202,
    "nonok_replies": 0
   },
   "pairs": [
    {
     "n": 24,
     "W": 0.9791666666666666,
     "W_ci95": [
      0.9360696331162698,
      1.0
     ],
     "sd": 0.10206207261596575,
     "t_lb": 0.9360696331162698,
     "p": 1.1920928955078125e-07,
     "wins": 23,
     "ties": 1,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "v1",
     "a_vs": "v1",
     "claim": "gate A",
     "result": "PASS"
    },
    {
     "n": 24,
     "W": 0.9791666666666666,
     "W_ci95": [
      0.9360696331162698,
      1.0
     ],
     "sd": 0.10206207261596575,
     "t_lb": 0.9360696331162698,
     "p": 1.1920928955078125e-07,
     "wins": 23,
     "ties": 1,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": true,
     "a": "v2",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    },
    {
     "n": 24,
     "W": 0.5,
     "W_ci95": [
      0.5,
      0.5
     ],
     "sd": 0.0,
     "t_lb": 0.5,
     "p": 1.0,
     "wins": 0,
     "ties": 24,
     "losses": 0,
     "dropped_per_shard": [
      0,
      0
     ],
     "gate_A_form": false,
     "a": "v1",
     "b": "jev",
     "a_vs": "jev",
     "claim": "descriptive",
     "result": null
    }
   ],
   "per_seed": [
    {
     "index": 34,
     "seed": 1636330441,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 35,
     "seed": 366125730,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 36,
     "seed": 586983780,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 37,
     "seed": 2116701015,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 38,
     "seed": 1591650288,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 39,
     "seed": 1648056398,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 40,
     "seed": 933113555,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 41,
     "seed": 359876553,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 42,
     "seed": 963747841,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 43,
     "seed": 1382794440,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 44,
     "seed": 1800540133,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 45,
     "seed": 755044608,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 46,
     "seed": 1194435068,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 47,
     "seed": 452817344,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 48,
     "seed": 929521102,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 49,
     "seed": 1003369363,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 50,
     "seed": 5246151,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 51,
     "seed": 1524652023,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 52,
     "seed": 596673862,
     "v2": 0.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 53,
     "seed": 1592964193,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 54,
     "seed": 1483049570,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 55,
     "seed": 1758577711,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 56,
     "seed": 928955108,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    },
    {
     "index": 57,
     "seed": 387129241,
     "v2": 1.0,
     "v1": 0.0,
     "jev": 0.0
    }
   ],
   "replay": "https://github.com/nagisanzenin/nagi/tree/main/bench/burst_v2/replays/booster_gauntlet",
   "video": "media/v2-BoosterGauntlet-V2-V1-Jev-seed34.mp4",
   "poster": "media/v2-BoosterGauntlet-V2-V1-Jev-seed34.jpg",
   "landings": {
    "v2": 23,
    "v1": 0,
    "jev": 0
   },
   "video_sha256": "ded887ad4e927a33ce95e3f783f30b001a67503fec1a16314b9155a66a1d35b4",
   "video_seed_index": 34,
   "video_seed_rule": "per game, margin_i = V2_i - Jev_i (V2 from docs/burst_v2_general/receipts/m4/<cell>/v2g_<game>.csv, Jev from docs/burst_v2_next/p3/jev1s_<game>.json per_seed[].score), rounded 1e-9"
  }
 ],
 "overall": {
  "v2_vs_v1": {
   "theta": 0.9930555555555555,
   "se": 0.006944444444444444,
   "df_ws": 23.0,
   "lb97_5": 0.9786898777054232,
   "wins": 71,
   "ties": 1,
   "losses": 0,
   "n": 72
  },
  "v2_vs_jev": {
   "theta": 0.9930555555555555,
   "se": 0.006944444444444444,
   "df_ws": 23.0,
   "lb97_5": 0.9786898777054232,
   "wins": 71,
   "ties": 1,
   "losses": 0,
   "n": 72
  },
  "v1_vs_jev": {
   "theta": 0.6180555555555555,
   "se": 0.045399424625286074,
   "df_ws": 45.92011529282934,
   "lb97_5": 0.5266669632776854,
   "wins": 32,
   "ties": 25,
   "losses": 15,
   "n": 72
  }
 },
 "notes": [
  "Seen tasks: the V2 core was trained on these three Arena games. The eval seeds were sealed and played once.",
  "Timing: V2 was played with the nominal 1-tick reply (the decision for tick t applies at t+1), justified by its measured latency gate L. The V1 and Jev numbers come from live realtime records, where a slow reply lands later.",
  "Snake Rush: V2 plays the SDK Snake search pack (20,000 simulations, horizon 12), which does not read the learned core. A predeclared rule chose it from dev seeds only, before the sealed look.",
  "Booster Gauntlet: V2 landed the first hop on 23/24 seeds; V1 on 0/24 and Jev on 0/24 (K3 = 0), so V1 vs Jev is all ties there. Game Arena v3's Board 1 documents V1's Booster pattern: flying under control until the first hop's landing deadline, without landing.",
  "Rotorwash-Ramp is a float64 engine. Re-checked here under Rosetta (no AVX/FMA), the Jev records recompute the recorded score on 24/24 seeds but replay state-hash-exactly on only 8/24. The recording job's own replay check on its x86-64 host passed on every seed. The release's V1 Rotorwash-Ramp check needed a real x86-64 CPU for the same reason.",
  "Not comparable with the Board 1 leaderboard below: different seeds (the release's sealed eval seeds here, Board 1's screen seeds there), Rotorwash-Ramp and Booster Gauntlet are scored by their K3 scores here and by survival time there, and Jev here is the single-stream re-play."
 ]
}
