{
 "schema_version": "1.0",
 "page_kind": "article",
 "pack_kind": "generate",
 "slug": "the-open-call",
 "title": "The open call — a kid, an elder, and a tired parent walk into the cove.",
 "dek": "Twenty language models were handed the same frozen moment of a small fishing town and asked to speak as its people, then read blind by seven judges from six rival families.",
 "published": "2026-08-15",
 "series": [
  "bench",
  "notes"
 ],
 "licence": "CC BY 4.0",
 "status": "pending-judge",
 "pack_note": "no judge is seated, so every chip is held",
 "notice": "items with published:false are held and are not this page's published words",
 "source": {
  "url": "https://research.strata2signal.com/the-open-call/",
  "md_url": "https://research.strata2signal.com/the-open-call/index.md",
  "md_sha": "5acba411262859876bc9cd8c77ff64beef4027edb2233bb0a4dbced2e154281d",
  "html_sha": "bb09c6bbb600aad3eae08fa8a5670f81e643c5a79325bc2a6fa908d25dbd4d72",
  "anchors_sha": "8fdf476950ca2477af8b0f434bcc225b983d588e892d36f1a7654b397a7486b0"
 },
 "short": {
  "paragraph": "Twenty language models were handed the same frozen moment of a small fishing town and asked to speak as its people, then read blind by seven judges from six rival families.",
  "from_dek": true,
  "counts": {
   "words": 39401,
   "minutes": 179,
   "tables": 9,
   "kit": true
  },
  "bullets": [
   {
    "text": "the panel produced a total of 868",
    "figure": "868",
    "cite": "where-their-zeros-sit",
    "quote": "Where their zeros sit Seven models sat down with 868 letters that morning and filed a considered verdict on every one of them, which is a strange and rather generous kind of work.",
    "span": [
     157876,
     157880
    ]
   },
   {
    "text": "the total metered spend was $2.2826",
    "figure": "$2.2826",
    "cite": "2-2826-over-56-metered-calls-against-38-49-of-caps",
    "quote": "$2.2826, over 56 metered calls, against $38.49 of caps That is the whole round's metered spend - every arm leg, every judge sheet including one discarded judging round, the auditions, the three registration probes, and 9 discarded cold-start warmups, one per metered arm per leg.",
    "span": [
     168786,
     168796
    ]
   },
   {
    "text": "the shortest median in the table is 1,457 ms",
    "figure": "1,457 ms",
    "cite": "how-long-a-family-waits-for-the-narrator-to-speak",
    "quote": "At, say, forty NPC replies in an evening - pick your own table's number and redo this in your head - the shortest median in the table above (1,457 ms a reply) comes to about 58 seconds of waiting across the whole night, and the longest (56,191 ms) to about 37 minutes.",
    "span": [
     177163,
     177171
    ]
   }
  ]
 },
 "sections": [
  {
   "id": "five-things-before-any-figure",
   "heading": "Five things, before any figure",
   "level": 2,
   "span": [
    433,
    2102
   ],
   "chunks": [
    [
     433,
     2102
    ]
   ],
   "chars": 1669,
   "digest": "the article establishes that the unit is one question, everything was registered first, and figures print as counts with their denominators. six reply cards per act are chosen by rule, and every number has a file behind it, with the kit carrying the complete scoring artifact.",
   "digest_skipped": null
  },
  {
   "id": "the-words-this-page-leans-on",
   "heading": "The words this page leans on",
   "level": 2,
   "span": [
    2102,
    3329
   ],
   "chunks": [
    [
     2102,
     3329
    ]
   ],
   "chars": 1227,
   "digest": "this section defines specific terminology used throughout the page, including arm, bundle, frozen, recusal, canon, family-mean, $0.00*, two clocks, and fold.",
   "digest_skipped": null
  },
  {
   "id": "the-author-of-this-page-is-a-contestant-in-it",
   "heading": "The author of this page is a contestant in it",
   "level": 2,
   "span": [
    3329,
    4977
   ],
   "chunks": [
    [
     3329,
     4977
    ]
   ],
   "chars": 1648,
   "digest": "Claude Fable 5 authored the exam scenarios, acts as an arm, and holds a judge seat. the agent transport is confounded with vendor, and the reference driver was qwen3.5:27b.",
   "digest_skipped": null
  },
  {
   "id": "three-scenarios-one-small-cove-one-draw-of-its-world",
   "heading": "Three scenarios, one small cove, one draw of its world",
   "level": 2,
   "span": [
    4977,
    12889
   ],
   "chunks": [
    [
     4977,
     12889
    ]
   ],
   "chars": 7912,
   "digest": "the exam features three scenarios set in a fishing village called the Reach in Sorrowmoor. the cove is an authored world in YAML campaign files. the panel of seven judges read every spoken line blind, producing 738 scoring cells.",
   "digest_skipped": null
  },
  {
   "id": "the-kid",
   "heading": "The kid",
   "level": 2,
   "span": [
    12889,
    38202
   ],
   "chunks": [
    [
     12889,
     32874
    ],
    [
     32874,
     38202
    ]
   ],
   "chars": 25313,
   "digest": "the first act involves Finn asking Pip Lune and Old Sefer Tamm questions. six arms are chosen by rule for reply cards, and the panel flagged 18 of the 20 arms for invention in this act.",
   "digest_skipped": null
  },
  {
   "id": "fold-s1",
   "heading": "Every reply in this act, in full",
   "level": 2,
   "span": [
    38202,
    82459
   ],
   "chunks": [
    [
     38202,
     57037
    ],
    [
     57037,
     75870
    ],
    [
     75870,
     82459
    ]
   ],
   "chars": 44257,
   "digest": "this section contains all 80 replies from the first act, including verbatim text, denominators, canon verdicts, clocks, and cost states for every arm.",
   "digest_skipped": null
  },
  {
   "id": "the-world-remembered-and-poured-them-a-drink",
   "heading": "The world remembered, and poured them a drink",
   "level": 2,
   "span": [
    82459,
    98182
   ],
   "chunks": [
    [
     82459,
     98182
    ]
   ],
   "chars": 15723,
   "digest": "the second act features Sam asking Garron Tarrow if anything happened lately. five of 20 replies handed the player's own deed back to them, and 13 of the 20 arms drew at least one invention flag.",
   "digest_skipped": null
  },
  {
   "id": "fold-s3",
   "heading": "Every reply in this act, in full",
   "level": 2,
   "span": [
    98182,
    119860
   ],
   "chunks": [
    [
     98182,
     118171
    ],
    [
     118171,
     119860
    ]
   ],
   "chars": 21678,
   "digest": "this section provides all 20 replies from the second act, detailing the family-mean, cells scored, canon, dimensions, transport, cost, clocks, and verbatim text for each arm.",
   "digest_skipped": null
  },
  {
   "id": "the-verse-keeper",
   "heading": "The verse-keeper",
   "level": 2,
   "span": [
    119860,
    133918
   ],
   "chunks": [
    [
     119860,
     133918
    ]
   ],
   "chars": 14058,
   "digest": "the third act features Eleanor asking Brisa Lune about her songs. the axis is who got somewhere the persona had not already been, and the arm's figure for the act is 7.92.",
   "digest_skipped": null
  },
  {
   "id": "fold-s2",
   "heading": "Every reply in this act, in full",
   "level": 2,
   "span": [
    133918,
    157799
   ],
   "chunks": [
    [
     133918,
     153822
    ],
    [
     153822,
     157799
    ]
   ],
   "chars": 23881,
   "digest": "this section contains all 20 replies from the third act, including the family-mean, cells scored, canon, dimensions, transport, cost, clocks, and verbatim text for each arm.",
   "digest_skipped": null
  },
  {
   "id": "where-their-zeros-sit",
   "heading": "Where their zeros sit",
   "level": 2,
   "span": [
    157799,
    163827
   ],
   "chunks": [
    [
     157799,
     163827
    ]
   ],
   "chars": 6028,
   "digest": "the seven judges produced 868 verdicts. the closest pair of seats sat a mean 0.72 points apart, and the farthest sat 1.61 apart. the local seat sat 1.11 from its six hosted counterparts.",
   "digest_skipped": null
  },
  {
   "id": "and-what-it-couldnt-agree-on",
   "heading": "And what it couldn’t agree on",
   "level": 2,
   "span": [
    163827,
    168786
   ],
   "chunks": [
    [
     163827,
     168786
    ]
   ],
   "chars": 4959,
   "digest": "across 738 scoring cells, 235 carried a canon verdict other than clean. three models reached for the same invented span of forty years. the panel caught the invention and could not agree it was one.",
   "digest_skipped": null
  },
  {
   "id": "2-2826-over-56-metered-calls-against-38-49-of-caps",
   "heading": "$2.2826, over 56 metered calls, against $38.49 of caps",
   "level": 2,
   "span": [
    168786,
    173492
   ],
   "chunks": [
    [
     168786,
     173492
    ]
   ],
   "chars": 4706,
   "digest": "the total metered spend was $2.2826 over 56 metered calls. the bill includes nine discarded cold-start warmups, one per metered arm per leg.",
   "digest_skipped": null
  },
  {
   "id": "how-long-a-family-waits-for-the-narrator-to-speak",
   "heading": "How long a family waits for the narrator to speak",
   "level": 2,
   "span": [
    173492,
    178120
   ],
   "chunks": [
    [
     173492,
     178120
    ]
   ],
   "chars": 4628,
   "digest": "latency is measured as client-wall for HTTP arms. one cloud tail's single act-two call ran 229,106 ms. the shortest median in the table is 1,457 ms.",
   "digest_skipped": null
  },
  {
   "id": "stated-at-the-size-they-are",
   "heading": "Stated at the size they are",
   "level": 2,
   "span": [
    178120,
    184490
   ],
   "chunks": [
    [
     178120,
     184490
    ]
   ],
   "chars": 6370,
   "digest": "the article lists its limits, including small n, the fact that a figure is a fit, and that decoding parameters were not pinned. the three agent-transport arms hold no cost figure, no client-wall clock, and no token counters.",
   "digest_skipped": null
  },
  {
   "id": "a-late-contestant-and-it-ties",
   "heading": "A late contestant, and it ties",
   "level": 2,
   "span": [
    184490,
    195193
   ],
   "chunks": [
    [
     184490,
     195193
    ]
   ],
   "chars": 10703,
   "digest": "qwen3.8:27b scored 7.33 on the panel's 0–10 scale. it ties the arm it succeeds, as the gap is 0.05, which is within the 0.5 tie band.",
   "digest_skipped": null
  },
  {
   "id": "what-to-take-with-you",
   "heading": "What to take with you",
   "level": 2,
   "span": [
    195193,
    198027
   ],
   "chunks": [
    [
     195193,
     198027
    ]
   ],
   "chars": 2834,
   "digest": "the article summarizes that no model wins, seven judges from six families read every reply blind, and the three Claude arms hold 5 of the 6 highest-family-mean cards.",
   "digest_skipped": null
  },
  {
   "id": "how-to-check-our-work-and-see-it-live",
   "heading": "How to check our work — and see it live",
   "level": 2,
   "span": [
    198027,
    200000
   ],
   "chunks": [
    [
     198027,
     200000
    ]
   ],
   "chars": 1973,
   "digest": null,
   "digest_skipped": "credits"
  },
  {
   "id": "the-rest-of-the-seminar",
   "heading": "The rest of the seminar",
   "level": 2,
   "span": [
    200000,
    200798
   ],
   "chunks": [
    [
     200000,
     200798
    ]
   ],
   "chars": 798,
   "digest": null,
   "digest_skipped": "credits"
  },
  {
   "id": "every-published-file-behind-every-number-on-this-page",
   "heading": "Every published file behind every number on this page",
   "level": 2,
   "span": [
    200798,
    217883
   ],
   "chunks": [
    [
     200798,
     217883
    ]
   ],
   "chars": 17085,
   "digest": "the page lists all published files, including scores.json, seal-manifest.json, and report.md. seven files were path-scrubbed before publishing, and two carry a correction added in the open.",
   "digest_skipped": null
  },
  {
   "id": "provenance",
   "heading": "Provenance",
   "level": 2,
   "span": [
    217883,
    222072
   ],
   "chunks": [
    [
     217883,
     222072
    ]
   ],
   "chars": 4189,
   "digest": null,
   "digest_skipped": "credits"
  }
 ],
 "chips": [
  {
   "id": "c-cbd4b98a",
   "q": "how are the figures and tie bands defined?",
   "a": "Figures are printed as counts with their denominators. Two figures closer than the registered tie band of 0.5 are considered within the band, which serves as a statement about the instrument's resolution rather than an assertion of equality between models.",
   "cites": [
    "five-things-before-any-figure"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-8b1cd908",
   "q": "what does the term arm mean in this context?",
   "a": "An arm refers to one model under test on a specific named transport. In this round, 20 arms were seated.",
   "cites": [
    "the-words-this-page-leans-on"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-fee2dac0",
   "q": "how does the author participate in the exam?",
   "a": "The author, Claude Fable 5, authored the exam scenarios, acts as an arm in the round, and holds one of the seven judge seats. A second Anthropic model holds another judge seat.",
   "cites": [
    "the-author-of-this-page-is-a-contestant-in-it"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-42fb4ab1",
   "q": "how were the six reply cards in the first act selected?",
   "a": "The rule selects the two arms with the highest family-mean, the single lowest, up to two arms flagged for invention, and the best-scoring of the three locally-hosted arms. Ties are broken by the registered roster order.",
   "cites": [
    "the-kid#0"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-66d4c561",
   "q": "what determines if a reply is considered canon clean?",
   "a": "A canon verdict is issued by the panel based on whether a reply stays within the established facts of the world. The panel uses a majority verdict to determine the arm's canon status.",
   "cites": [
    "fold-s1#0"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-084d2882",
   "q": "how does the panel handle judges from the same model family?",
   "a": "A judge never scores a model from its own family. These recused cells are kept and printed rather than deleted, and the two Anthropic seats never scored any Claude arm.",
   "cites": [
    "where-their-zeros-sit"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-d917d3c3",
   "q": "what happens when the panel cannot reach a majority verdict?",
   "a": "When the panel divides with no majority, the figure is published as SPLIT. There are 18 such arm-figures in the round.",
   "cites": [
    "and-what-it-couldnt-agree-on"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  },
  {
   "id": "c-ca5aa91a",
   "q": "how is the tie between the late contestant and its predecessor handled?",
   "a": "The newcomer, qwen3.8:27b, scores 7.33 on the panel's 0–10 scale. This is considered a tie with its predecessor, qwen3.6:27b, which scores 7.28 in the minus-anthropic column, as the difference of 0.05 is within the 0.5 tie band.",
   "cites": [
    "a-late-contestant-and-it-ties"
   ],
   "published": false,
   "publish_state": "held",
   "quote": "",
   "quote_cite": ""
  }
 ],
 "related": [
  {
   "slug": "three-new-voices-at-the-narrators-chair",
   "why": "shares ground with § And what it couldn’t agree on · § What the panel caught, and what it couldn't agree on"
  },
  {
   "slug": "two-new-frontier-models-at-the-rules-desk",
   "why": "shares ground with § $2.2826, over 56 metered calls, against $38.49 of caps · § The bill"
  },
  {
   "slug": "the-instrument-travels",
   "why": "shares ground with § Where their zeros sit · § The judge's chair scored one candidate of three"
  }
 ],
 "thanks": "",
 "kit": {
  "url": "https://research.strata2signal.com/the-open-call/data/",
  "licence": "CC BY 4.0",
  "files": [
   "scores.json",
   "seal-manifest.json",
   "report.md",
   "known-world-issues.md",
   "counting-rules.json",
   "seats.json",
   "bill.json",
   "addendum-scores.json",
   "addendum-samples.json",
   "addendum-runtime-pins.json",
   "addendum-probe.json",
   "addendum-bill.json",
   "README.md"
  ]
 },
 "seat_class": {
  "writer": "gemma-class",
  "judge": null,
  "writer_runtime": "vllm"
 },
 "bench": {
  "bullets_written": 3,
  "bullets_kept": 3,
  "digests_written": 18,
  "digests_kept": 18,
  "chips_written": 8,
  "chips_kept": 8,
  "chips_grounded": 0,
  "chips_published": 0,
  "dropped_by": {
   "new_noun": 0,
   "figure": 0,
   "length": 0,
   "cite": 0,
   "judge": 0,
   "quote": 0,
   "directive": 0,
   "redaction": 0,
   "profanity": 0
  },
  "new_noun_tokens_checked": [
   "0-10",
   "0.00",
   "0.05",
   "0.5",
   "0.72",
   "1",
   "1.11",
   "1.61",
   "106",
   "13",
   "18",
   "2.2826",
   "20",
   "229",
   "235",
   "27b",
   "457",
   "5",
   "56",
   "6",
   "7.28",
   "7.33",
   "7.92",
   "738",
   "80",
   "868",
   "Anthropic",
   "Brisa",
   "Claude",
   "Eleanor",
   "Fable",
   "Finn",
   "Garron",
   "HTTP",
   "Lune",
   "Old",
   "Pip",
   "Reach",
   "SPLIT",
   "Sam",
   "Sefer",
   "Sorrowmoor",
   "Tamm",
   "Tarrow",
   "YAML",
   "qwen3.5",
   "qwen3.6",
   "qwen3.8"
  ],
  "new_noun_tokens_withheld": 0,
  "figure_definition": "v3",
  "drops": []
 },
 "generated_utc": "2026-09-09T17:49:44Z"
}