{
 "credit": "Taylor (gtmcknight), github.com/gtmcknight/amm-sim@4812a6382311624e1506aca131ca35c378f930d9 research/solved: the solved policy, real.py, the qm tables, the replica and the arb/retail tell. BASE stalerival.py is the network's replica policy from https://api.imd.fun/artifacts/f5a140f51df99423d6ede60b618dbc6eabc37a3050ac439c80dc3824191e9f59; this item only subclasses it.",
 "idea": "Confidence blending of the table families. BASE picks tags 1s/2s (synced) or 1c/2c by a flag that any 0.35 pull sets. Keep a confidence c in [0,1] for the rival estimate instead: 0 on a new step, 1 after a binding router bound, raised after a pull. Arms: synced family only when c>=0.6; only when c>=0.9; fees blended linearly between the families by c.",
 "setup": [
  "R=https://raw.githubusercontent.com/gtmcknight/amm-sim/66604b3426d8e97a73e2c39940b9b4dece51cc73/research; curl -LsSf $R/tell/fetch.sh -o fetch.sh && sh fetch.sh $R scratch/amm   # fetched 34 files, 0 mismatched",
  "scratch/amm/stalerival.py = zlib(base64(code_zb64)) of https://api.imd.fun/artifacts/f5a140f51df99423d6ede60b618dbc6eabc37a3050ac439c80dc3824191e9f59 (sha256 176f8115fb67d99340f1de1e6994042dfaa41f830c22e000595df3377cc6a2b3, checked)",
  "scratch/amm/blend.py = zlib(base64(code_zb64)) of this result (sha256 in frozen.code_sha256)",
  "BASE: cd scratch/amm && PROCS=8 python3 -I run.py stalerival:Policy <seed0> <n> out.json soft=1 qm=qm_k1e-5.json a0=0.35 ainf=0.35",
  "pick: PROCS=8 python3 -I run.py blend:Policy <seed0> <n> out.json soft=1 qm=qm_k1e-5.json a0=0.35 ainf=0.35 mode=thr thr=0.6 cadd=1",
  "other arms: thr60 = mode=thr thr=0.6; thr90 = mode=thr thr=0.9; blend = mode=blend; blendadd = mode=blend cadd=1; blend70 = mode=blend calpha=0.7; off = mode=off (BASE exactly)",
  "python3 -I pair.py <arm.json> <base.json>   # paired mean_diff, se, worst5. The rows here were produced by harness_zb64's drv.py, which calls the unmodified run.work per 64-seed block; both CLI lines above were re-run and matched it row for row (diagnostics.cli_*)"
 ],
 "rule": {
  "state_kept": "BASE's state with the boolean `synced` replaced by a confidence c in [0,1]. For the pick c only takes the values 0, 0.35, 0.70, 1, so it is a 2-bit count of pulls this step saturating at 2, plus BASE's existing bound flag.",
  "per_callback": [
   "new step (timestamp advanced; where real.py clears synced): c = 0",
   "retail read, router candidate binds (buy: cand < s2h, sell: cand > s2h; where real.py sets s2h = cand and synced): c = 1",
   "retail read, candidate on the non-binding side (where BASE pulls ln s2h by alpha = 0.35 and sets synced): after the pull, c = min(1, c + q) when cadd=1, or c = c + q*(1 - c) when cadd=0; q = the pull's alpha (0.35) unless calpha > 0",
   "table family for phases 1 and 2 (phase 0 keeps tag 0c): thr arms use tags 1s/2s iff c >= thr, else 1c/2c; blend arm, when 0 < c < 1, looks both families up at the same (w, w2, phase, sd) and returns fee = fee_c + c*(fee_s - fee_c) on each side; c = 0 or 1 is a single lookup",
   "the pull, the filter, the tell, the posterior odds, fee rounding and the lookups themselves are BASE's, untouched"
  ],
  "pick_in_words": "thr60add: the synced tables are used after a binding bound this step, or from the second pull of a step on (0.35 + 0.35 = 0.70 >= 0.6). One pull alone no longer switches family.",
  "equivalences": "With c + q*(1-c) the values after 1, 2, 3 pulls are 0.35, 0.5775, 0.725375 (0.9 needs 6 pulls), so thr60 syncs from the 3rd pull and thr90 is in effect 'a pull never syncs'. Steps with 2 or more pulls and no bound are rare (diagnostics.confidence), which is why the three threshold arms sit within 0.003 of each other."
 },
 "frozen": {
  "params": {
   "pick": "thr60add",
   "mode": "thr",
   "thr": 0.6,
   "cadd": 1,
   "calpha": 0,
   "base": {
    "soft": 1,
    "qm": "qm_k1e-5.json",
    "a0": 0.35,
    "ainf": 0.35
   }
  },
  "params_sha256": "fa07f137a9f05287cdbc7a609fcf8546afcb0cb5d2cd2109e656150054b66e38",
  "code_sha256": "9835112c82b3ed18ad1b2df7847e16886ae94f8b752235e198d529e131a1d6d4",
  "frozen_utc": "2026-11-10T15:32:38Z",
  "base_sha256": "176f8115fb67d99340f1de1e6994042dfaa41f830c22e000595df3377cc6a2b3",
  "report_plan": "10000..10767 in order, 12 blocks of 64, fixed before reading any report seed; arms BASE, thr60, thr90, blend and the pick thr60add",
  "settings_written_before_scoring": {
   "written_utc": "2026-11-10T15:21:01Z",
   "settings": {
    "thr60": {
     "mode": "thr",
     "thr": 0.6,
     "cadd": 0,
     "calpha": "pull alpha 0.35"
    },
    "thr90": {
     "mode": "thr",
     "thr": 0.9,
     "cadd": 0
    },
    "blend": {
     "mode": "blend",
     "cadd": 0
    },
    "thr60add": {
     "mode": "thr",
     "thr": 0.6,
     "cadd": 1
    },
    "blendadd": {
     "mode": "blend",
     "cadd": 1
    },
    "blend70": {
     "mode": "blend",
     "cadd": 0,
     "calpha": 0.7
    }
   },
   "code_sha256_then": "9835112c82b3ed18ad1b2df7847e16886ae94f8b752235e198d529e131a1d6d4"
  }
 },
 "choose": {
  "seeds": [
   64000,
   64383
  ],
  "n": 384,
  "minus_base": "stalerival:Policy with the same options, same seeds",
  "table": {
   "thr60": {
    "n": 384,
    "mean_diff": 0.006664,
    "se": 0.006064,
    "mean_plus_3se": 0.024856,
    "min": -0.591224
   },
   "thr90": {
    "n": 384,
    "mean_diff": 0.006278,
    "se": 0.006068,
    "mean_plus_3se": 0.024482,
    "min": -0.591813
   },
   "blend": {
    "n": 384,
    "mean_diff": 0.002872,
    "se": 0.005383,
    "mean_plus_3se": 0.019022,
    "min": -0.693361
   },
   "thr60add": {
    "n": 384,
    "mean_diff": 0.008125,
    "se": 0.005957,
    "mean_plus_3se": 0.025997,
    "min": -0.583738
   },
   "blendadd": {
    "n": 384,
    "mean_diff": 0.004454,
    "se": 0.005673,
    "mean_plus_3se": 0.021473,
    "min": -0.679907
   },
   "blend70": {
    "n": 384,
    "mean_diff": 0.002527,
    "se": 0.004458,
    "mean_plus_3se": 0.015901,
    "min": -0.364262
   }
  },
  "selection": "highest mean_diff: thr60add. No setting met the drop rule (mean_diff + 3 se < 0), and none is separated from BASE or from another setting at 2 se on these seeds. 384 of the 1,024 choose seeds were used; the rest were left for time."
 },
 "report": {
  "seeds": [
   10000,
   10767
  ],
  "n": 768,
  "order": "12 blocks of 64 in order, count fixed at freeze before any report seed was read; scored once",
  "arms": [
   {
    "name": "thr60",
    "minus_base": {
     "mean_diff": 0.008111,
     "se": 0.003881,
     "min": -0.428498,
     "worst5": [
      [
       10300,
       -0.428498
      ],
      [
       10697,
       -0.386525
      ],
      [
       10254,
       -0.353541
      ],
      [
       10436,
       -0.346705
      ],
      [
       10414,
       -0.338295
      ]
     ]
    },
    "edge": 538.2422
   },
   {
    "name": "thr90",
    "minus_base": {
     "mean_diff": 0.008263,
     "se": 0.003906,
     "min": -0.428554,
     "worst5": [
      [
       10300,
       -0.428554
      ],
      [
       10697,
       -0.38645
      ],
      [
       10254,
       -0.353541
      ],
      [
       10436,
       -0.346645
      ],
      [
       10414,
       -0.338154
      ]
     ]
    },
    "edge": 538.2424
   },
   {
    "name": "blend",
    "minus_base": {
     "mean_diff": 0.001384,
     "se": 0.004039,
     "min": -1.04122,
     "worst5": [
      [
       10208,
       -1.04122
      ],
      [
       10231,
       -0.648195
      ],
      [
       10667,
       -0.441644
      ],
      [
       10590,
       -0.310588
      ],
      [
       10654,
       -0.309358
      ]
     ]
    },
    "edge": 538.2355
   },
   {
    "name": "thr60add (pick)",
    "minus_base": {
     "mean_diff": 0.010318,
     "se": 0.003936,
     "min": -0.388606,
     "worst5": [
      [
       10697,
       -0.388606
      ],
      [
       10009,
       -0.364956
      ],
      [
       10300,
       -0.358995
      ],
      [
       10254,
       -0.352999
      ],
      [
       10436,
       -0.347771
      ]
     ]
    },
    "edge": 538.2444
   }
  ],
  "base_edge": 538.2341,
  "pick_by_64_block": [
   [
    10000,
    0.0116
   ],
   [
    10064,
    0.0146
   ],
   [
    10128,
    0.0307
   ],
   [
    10192,
    0.0083
   ],
   [
    10256,
    0.0077
   ],
   [
    10320,
    0.0344
   ],
   [
    10384,
    -0.0058
   ],
   [
    10448,
    0.013
   ],
   [
    10512,
    -0.0018
   ],
   [
    10576,
    0.0136
   ],
   [
    10640,
    -0.0037
   ],
   [
    10704,
    0.0011
   ]
  ],
  "not_reported": "blendadd and blend70 were chosen against and not run on report seeds"
 },
 "cost": {
  "status": "estimates from the rule; no Solidity or engine work in this item",
  "extra_storage_words": 0,
  "storage_note": "pick: 2 bits (pull count this step, saturating at 2) in the word that already holds the synced flag; blend arms: c as a small fixed-point value in the same word",
  "worst_ops_per_callback": {
   "arb_read": "no change (phase 0 uses tag 0c)",
   "new_step": "clear c (rides the write that clears synced)",
   "retail_binding_bound": "set c = 1 (rides the write that sets synced)",
   "retail_pull_threshold_arms": "1 add, 1 saturate/compare for c, 1 compare c >= thr; and the lookup BASE repeats after a pull is unchanged",
   "retail_pull_blend_arms": "as above plus one whole second table lookup (two sd layers x bilinear, both sides) and 2 multiplies, 2 adds for the mix: the lookup cost of that callback doubles"
  }
 },
 "diagnostics": {
  "switched_off_equals_base": {
   "seeds": [
    64900,
    64907
   ],
   "rows_identical": true,
   "fields": "every field of every row (edge, retail, arb, rv, av, first, misread, se, nse)"
  },
  "cli_base_matches_driver": {
   "seeds": [
    64900,
    64907
   ],
   "rows_identical": true
  },
  "cli_pick_matches_driver": {
   "seeds": [
    10000,
    10007
   ],
   "rows_identical": true
  },
  "reads_no_hidden_input": {
   "method": "harness_zb64 diag.py: a run.Blind subclass replaces sigma, lam, size, the arb flag, the true price, the rival spot and the order index with noise before run.Blind sees them; sha256 over every returned fee pair, callback count and edge compared with the ungarbled run",
   "all_identical": true,
   "runs": [
    {
     "arm": "off",
     "seed": 64900,
     "callbacks": 9464,
     "fees_sha256": "df19b1b33588186a4361469ae480025b416baf63fbd74a2d060b4fb217095112",
     "identical": true
    },
    {
     "arm": "off",
     "seed": 64901,
     "callbacks": 12108,
     "fees_sha256": "882765660fa6170e1b1a3c090a891df6dccee0b17010a105286193c6fe3aa63c",
     "identical": true
    },
    {
     "arm": "thr60",
     "seed": 64900,
     "callbacks": 9470,
     "fees_sha256": "b33d17fbea8e995e790a57d24d5b14a64542bc7634095151d01ba8cec389e24c",
     "identical": true
    },
    {
     "arm": "thr60",
     "seed": 64901,
     "callbacks": 12106,
     "fees_sha256": "222e9f4aca90f29e691a0db5ec8be5335052fbd27f135152dd474fc4c1ad4a41",
     "identical": true
    },
    {
     "arm": "thr90",
     "seed": 64900,
     "callbacks": 9470,
     "fees_sha256": "b33d17fbea8e995e790a57d24d5b14a64542bc7634095151d01ba8cec389e24c",
     "identical": true
    },
    {
     "arm": "thr90",
     "seed": 64901,
     "callbacks": 12107,
     "fees_sha256": "1d295d5de2230edee4ad274e1e0a42593a0bda6c4f45ec43b51ec1d283f3d3ce",
     "identical": true
    },
    {
     "arm": "blend",
     "seed": 64900,
     "callbacks": 9471,
     "fees_sha256": "fd2cb6ce83214f87f2846a8840507937eb3bfec41ef07c01dc7ab07151e8b5d9",
     "identical": true
    },
    {
     "arm": "blend",
     "seed": 64901,
     "callbacks": 12107,
     "fees_sha256": "b4add4a4da7000c9b6c9532f63b2fad03a09a134d6f40e10a450a8d2ab432b75",
     "identical": true
    },
    {
     "arm": "thr60add",
     "seed": 64900,
     "callbacks": 9470,
     "fees_sha256": "159af5a7b191b74f2ca46a13814527e35d1c7185f43aced3d8701533e05a8067",
     "identical": true
    },
    {
     "arm": "thr60add",
     "seed": 64901,
     "callbacks": 12107,
     "fees_sha256": "e713483940ded670ed383e7d1fb6f81c0ff5da5d2c722b3d0a1d77097b49381d",
     "identical": true
    },
    {
     "arm": "blendadd",
     "seed": 64900,
     "callbacks": 9471,
     "fees_sha256": "c8326dc5dc2ababe7126927c2b9bb84eef08ea4fb010305667a7dcbaef646e27",
     "identical": true
    },
    {
     "arm": "blendadd",
     "seed": 64901,
     "callbacks": 12107,
     "fees_sha256": "38c07f4bd0bc9f59121de14914f339e32732371d70c31fa639de5027765c22e8",
     "identical": true
    },
    {
     "arm": "blend70",
     "seed": 64900,
     "callbacks": 9469,
     "fees_sha256": "66bee3444f9751678db90d192ad642da4d23a6f355f9db6429ecb7cfd3441cc3",
     "identical": true
    },
    {
     "arm": "blend70",
     "seed": 64901,
     "callbacks": 12107,
     "fees_sha256": "a1dfa938d693a44eed8a146f6b596276e938e284b8bbe0d7b425572c5616f9b6",
     "identical": true
    }
   ]
  },
  "confidence": {
   "what": "BASE trajectory (mode=off), seeds 64900..64915, read by the harness after the policy answered. Retail-read callbacks by c after the callback (c = c + 0.35*(1-c) form): n, rms of ln(s2h / true rival spot) in bps, share within 0.01 bps, share where the router candidate equals the true rival spot (the order was split, so the identity is exact).",
   "table": {
    "c:0.5<=c<0.6 (2 pulls)": {
     "n": 586,
     "rival_rms_bps": 6.874,
     "share_within_0.01bps": 0.0051,
     "share_router_identity_exact": 0.2235
    },
    "c:0.6<=c<1 (3+ pulls)": {
     "n": 35,
     "rival_rms_bps": 7.889,
     "share_within_0.01bps": 0.0,
     "share_router_identity_exact": 0.1143
    },
    "c:0<c<0.5 (1 pull)": {
     "n": 10528,
     "rival_rms_bps": 4.236,
     "share_within_0.01bps": 0.0049,
     "share_router_identity_exact": 0.2179
    },
    "c:c=1 (binding bound this step)": {
     "n": 102042,
     "rival_rms_bps": 0.932,
     "share_within_0.01bps": 0.9678,
     "share_router_identity_exact": 0.9682
    },
    "ev:arb": {
     "n": 62176,
     "rival_rms_bps": 3.373,
     "share_within_0.01bps": 0.8828,
     "share_router_identity_exact": null
    },
    "ev:bind_or_none": {
     "n": 98974,
     "rival_rms_bps": 0.202,
     "share_within_0.01bps": 0.9976,
     "share_router_identity_exact": 0.9973
    },
    "ev:pull": {
     "n": 14217,
     "rival_rms_bps": 4.62,
     "share_within_0.01bps": 0.0051,
     "share_router_identity_exact": 0.177
    }
   },
   "share_of_retail_callbacks_with_0<c<1": 0.0985,
   "reading": "About 90% of retail-read callbacks come at or after a binding bound in the same step, where the estimate is exact 97% of the time; the arms cannot touch those. They act only on the roughly 10% reached by pulls alone, where the candidate is the true rival spot 22% of the time and the estimate after the pull is off by 4.2 bps rms (6.9 after two pulls). That is where BASE reads the synced tables on an estimate that is not synced, and why the lever is small.",
   "rows": "c:* rows split retail-read callbacks by c after the callback; ev:* rows split all callbacks by what happened in that callback (arb read; pull; binding bound or, rarely, neither). The router identity does not apply to an arb read, so its share is null there."
  },
  "pooled_choose_plus_report_pick": {
   "n": 1152,
   "mean_diff": 0.009587,
   "se": 0.003289,
   "caveat": "includes the 384 seeds the pick was selected on; not the protocol figure"
  },
  "timing": {
   "first_8_seeds": "BASE and the off arm, 16 runs, 3.9 s wall at PROCS=8",
   "cpu_s_per_run": 0.68,
   "wall": "about 4.4 runs a second at PROCS=8 (about 3 cores effective on this machine); 64 seeds x 5 arms 70-76 s, x 7 arms about 100 s; no command over 5 min 11 s"
  },
  "drop_rule": "mean_diff + 3 se < 0 as the brief states it; applied to the choose table, nothing dropped. The kit itself was not fetched: Python-only replica item."
 },
 "advance": false,
 "verdict": "Do not advance. Pick thr60add on 768 report seeds: mean_diff +0.0103 se 0.0039, under the 0.02 bar (mean_diff - 2 se = +0.0024). Every threshold arm has the same sign on the choose and the report seeds (+0.006 to +0.010), so 'one pull alone does not switch to the synced tables' looks like a small real gain of about 0.01 that costs a contract two bits and a compare, but it is not a lever towards 528.4152. Linear fee blending is indistinguishable from BASE (+0.0014 se 0.0040) and costs a second lookup on those callbacks: drop it.",
 "seeds_used": {
  "choose": "64000..64383: BASE and all six settings",
  "checks_inside_choose_range": "64900..64907 off-vs-BASE and CLI check; 64900..64901 garble, all arms; 64900..64915 confidence diagnostics on BASE",
  "report": "10000..10767: BASE, thr60, thr90, blend, thr60add, once, after the freeze; 10000..10007 pick re-run through the run.py CLI after the report to confirm the command (same rows)"
 },
 "not_done": [
  "report seeds 10768..12047 (time: about 4.4 runs a second here)",
  "choose seeds 64384..65023 apart from the checks on 64900..64915",
  "blendadd and blend70 on report seeds",
  "blending the table entries (a, b) instead of the fees; a confidence that depends on the size of the pull gap or on the side",
  "any Solidity, gas, size or engine measurement: the cost block is an estimate",
  "the kit (score.py, engine_eval.py) was not used"
 ],
 "harness_zb64": {
  "encoding": "base64(zlib(UTF-8 JSON {name: text})): drv.py (block driver over run.work), diag.py (garble check, confidence table); they sit in scratch/work beside scratch/amm",
  "data": "eNrFWW1z2zYS/is4fziSNU3r1a0lK3NJm7Y3bdqO6/skazggCUmMJZAlINs6j/77PQuAEik7Tjpzd3Vii1zsLnYX+wo9nWTVfVRuT0bs5Nb8+67K70XF5kXF9FKwZCVkxsRjKap8LaRmfiXKVZ7ykP221ctCskKutkHEvq1ElusRu+HbFWj9hV6ndzJfLHUQ3spFrpebJEqL9flh4Zyv12cqX/9j8E23xy/63/T63e5FbyC6w84FT3m33015f5j2v/5mftnvZJesEkrwKl2eq2J1L7LoVm4UX4gRK40sfXb2T2YVYleaL96wKyVE1sGnxC+v1iH9RlH0hk2TTbYQOlazW3ldPCjGK8HuRKkZVGU+4bGLwRnRwwhFehewjcyw9FBUd+eG+/mYzXOZq2WNYplIQQasRLWRkbXprczXZVFpprYqZAV+P6pChkzDpLfyx/fX79kE4KjkehlleSX5Wvj1O08UffpxPM9XIo6D4Fa+/fChQfGxyKV/TE5cg5B5sLEHCuxs13OpRKX9TsjAJNhLBmFv5bwq1qzkFXPAcs1LBxX3fFUWq3pFaa7Vrfz21w8ffv0Fojx5qphrb8S6ETh7f6zx2BIPmxl4fNcVZ8OI9PeMeB1gdqL+kJ5zOXdvO+h4/eF34nwrGX68d29/f49F+vE9bL+CQ0Km0W8FvHHrhexpR45mcIv53KEC13hwA81bF5nAssE60OhlddGxVK/QAAvv5oMEvWjRX/5p+ssDvSH5LL3FOpKaZ0T4J6TGc2qJcFxHInyWmRXhNRZfGzt8GYtVueRGqK8NE5x7JubMODxCLGQmDpNOMEJA6U0l215FTo4wIsS59wTcXfyUdHbOvSyvVcGzJi8TqMHIClxR5MPJdvaVsl7SYbl0aA6LfkqgPRfrsJ7P96KJx1xp5ZdBg7zmXhFzEi8ychWlkEAk9SDJtJp6lHC82Qy7VU5EqzetW4U0h/32YqhapRCY84mNk3rnhCsBTgcTAGVvgjErNvpIeTAkAYnvqKUbLUwmRD9iaSF1LjfigMBbmzTtPGakENlYIW+IzFdC+zxgf2f0QPIFbSPCNXxDErAr1ntpswzMpj6fqtnUE8jh3oydGU0bkJCBnhRSpI7hNztwgHjGLiaL+dN7g3kfspiQs1lDIFhoCnVmJsVJOOpBPLjvWnAZZ7nJNuCJs8ODErSSEzL++lkw7UAcD2VD6SGA0ymO6n5W72kFtLYB7mg4m4Vtt6l/rGrYYLNua9/WM2DnbSnJNHGD9shUr5LvWj4Ia6BgzFkcU4mJY/IIL47XSNxx7DmHMS6gUAWk9U+yM4oPrxb30y4skUvt7wG9WXAE6RNk/zaYRQrthva90HPHYus2uM7hcQ3C4SzY+4+DBewNGzKxwmEPOp2oYxkgTNf8Dt1KpfyXswkkMDEcF3eTm2oj6p1tjYerY82vuFwInxRVHXZK2l4MAvi77gCDKntEf3xAVlyRvHsBXskyusgKcm9uo7GORVJMFvpZhjnkI5ONglkrlIiCGL4URFhuyIgA0qSEkfQrpPTuEKazlkZzVdEJec7wShclSwTkE1Z+z+w9Zkkl+F1DleTYDnUmaJgvgfkoShLa/WJQ2zJoJgXNlbH6tKHd3jpGv3a8oE1BkwUC6iCmfIY8V5L9szzVvu1YAgOLNmXGNTqtoM3AbnhKWcbyKnVohZ/moxzyDeDG1N199/77t//6+Sb+7defbbbJSSSrGRRrRuGgeTZoY6maoLvyiQ11lKHdFYLdkYVAad/ZuQ1H0jNo65+H1gRCbtaiIkUM0pE1XIVzGca3nX1i5BTQB6d9N/JJqW5Az7NDlUog9p3YTlZ8nWScVahRdXk6MpipZ9lmDXWwGxmMqtqRc2ILL9Lrkur+g0eRAm+mYYKn4jXkZ05+2NsF1pEjJ42ibTx37hk/ZdQa2HB62pt0h2dwGUWd+U5ht/lqo5bNoHdVMbQ2+UTtDaJci7Xym8Z3e9+eUF8yulQ7JtkTLCi92WiQ7RjVDQM4VJHZ6DQazHdUnJ6MrQExgPVp38EayKf9r2okS4ZAsji5NLA+YJTjDdAm+5EBmraAgIfSYFduT5zWteF0gU6bPTUsfKY7zlhAPQnZSZbzRWuC/JFXUih1RoMhS5eCcuanxsm/bG5c8AqyjBjNalu2zLNMSBxwiY6I5HRDLltyiZRFkIeKl2YyVPliDU9EVCAp5P8W5AkJ/IZ8QsNvyHYpoGY8YaosNPQoKhodc/Rvj6hSijm/x9y4RZrOcRwuoza3Qqwh8+uxAc7phR5K00+7kqzYeoMYSATLoYCGyCsoh3SPNg2zuWkGQVbxjyLVBTT1qQefYPRBcBtFScozoXC4yCBMVBWlhy1VjDmxTAVLqW3kRlSiWNrjZW+/v3l/TZBb6URacgzAUj0IHGnEXpjMradQG2E69NeHY8izDCmZZgUsDd7LVZ78tdPymL2wYEibc7Qx1yqsfajHYJbrPzNcUye9Z5hV97cYzVNkKcV+LhamYrxbwZXqZEMzQRznMtdxjHqzmlNHJUUVMuvkzaSkNnAsP4j2+AbT1ObVPFrQ1GEMHl2bD//y0jRVlpHtpH4ppHD4S+C7g4nUkveGF37NSlLDc5CPdnOyHcdP0B40rCCjZ2hUw8xSZCX0qU1sA1C/Ll8EXrYaeuLkzGDFOhao1q7uD3CQlQ/CIEI8IHx8Kl7NlrjWspAxQg0I7hRUnGy2yA+P+MUnugg8bU3CQF3Dbj04+8v6twvsYq/92GSbCWsoeIXOcji2Ayq0XdsGsfQX0YJvlCIfRes5JKFVjxj1PoP1cc9/38pcBp8y4V7nL9S24YufN7JzJbRi3bbF7SRs3RLjpDOXqdPmxm7C+PjQCNKVoG0GMcq12kFaebklHLObeqg9bvXGroGrG1IqbgvTUvnf8xXNf6Z/aJxhkXyk6XJiukYaPmxjeRNaKbAxafsABIpwg71IaB9AriMQmT4yZA9274jqg0Q39wD7LcVjli+QwSkiHiJKP3W1r3slZ7knD/pjCDRmsp3ciFnGdBOzSjhKtTcyjWFnZoY1j8pObGP7sGIG2n3B2cPtBYEC4Y5O6N1PZCMvnXSo5etcpVfwU+Z3WblZrQIDi4ZXEwJfML9nwMrBLwjeZX7/tAFOJ4AkxQYdhP2rlyilSosy8GbWI5JNeie0nx7uizoUVCm7Qjpyk6AFvIFPWUDXYVAYWUjvALmwkH6dgb9FXfzyFPzlqVevYKyzrnvjadq4mCHuPMscY4wKwszMaKGK1g68zhIgj5TQoOOblfZBMbXhjT/4P8OmnM7LxhXHgdGjQFoQ9Nozr6ikvrDZpWOw+gZcb/zfzHpS1pIbm4yZFBQLmv1tUlunaSZ0RLKkKducLeGaMwJtJMlbxmxOc+WcEw5B50loP/n/IIuBy5ads8cxS5ExiSfsCC8/IynoNoW6gegH41OWuxWX0Oqlc0fAWzeKztLMnoXJ2CukB7PNOQQJ6HS64qxR3qhM7hFJZdVbWlwa68WgVWqMSfIK/SNxpDfIBz1Hzovgcp64HwHiGZ9rCkcqHBWpFhGdg1dvYk6FsgMd22kdd4SWIIjiooolmgqv7difYp6OPPB499PUxTptkAazTxE3qwZ5KzW2bip38r+S5zH+HF2QHi7Kmun90wVlai+96QLRfOEwfqEWHF+U35grB1cQTMIBDV5dJaAy0L4JuHM3AQ8m8p+PoubyigKn0K/nhef31Y0bjQGSajLNbXLAZ6u2gDXl/M9dDroLwamxRVh/6xLWX5+Eje8G9l9thI1vJsLDVwxu/0V9k2KbgZCuhve3v4cLtP3Z+ReDy06HLusuO92gVhl9ud7fyZCLgE/jmsrQ0MUPoii3tIP6NQieG2r2ou/QJoRlNhu9fIS09n8+QVNq7kbugpvb6m7mwhiWi5PS9ARUcG0OUn9g8DFF49xgo+3o0+yELqES8QMmdpw4FY0mpakq54774ICOVYRsbJsJvY3FI091g6rfpNq1TOWutMgkzmD1dfXhPurJs24BjgvqIHCysYleuog/eEJ3OHOrgKc7d4PVuramO6vQTPBST7qtaxK3B0MLdRjDR/BVAPzHaaNTsrdrjyQ9XTLTLdSCOj/LZy+28tPGVnTPsvsPEyOsgA=="
 },
 "code_zb64": "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",
 "rows_zb64": "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",
 "rows_note": "rows_zb64 = base64(zlib(JSON [[seed, pick_minus_base], ...])), 768 report seeds in order, 6 decimals"
}