Dune/docs/TRUTHTRANCE_KNOWLEDGE.md

Truthtrance: recorded predictions and storm knowledge

Caladan occupier's auction knowledge — 29 October 2026

Prototyped / Partial. The existing auctionPeek fact also checks a Caladan occupier's successful private inspection of the original current ordinary or Advanced Black Market lot. It uses the same audience and lot identity as the auction view. The occupied inspection remains separately available when the native Atreides inspection was canceled; a different viewer, unavailable inspection or stale lot yields I don't know. The target alone sees the checked answer before replying, and bots use that same answer.

The runtime fixture uses a genuine auction and controlled Caladan occupation; it stages a conserved Truthtrance card. Eight related files pass 36 cases, including native cancellation and a recycled physical card in a later lot. This does not certify wider Homeworld interactions or deployed play.

Freeform Traitor and Face Dancer facts — 29 October 2026

Prototyped / Partial. New current-fact questions recognize an exact unique public leader name in a held-Traitor or unrevealed-Face-Dancer question, including supported single-operator combinations with other current facts. The editor previews the checked interpretation. The server uses the existing distinct predicates, and only the target sees the verified answer before replying; an unresolved Traitor choice yields I don't know. Absent or ambiguous names stay player-managed. The shared Cheap Hero / Heroine identity is supported; Duke Vidal is not an eligible named identity for this grammar.

New interpretations carry version 9; earlier versions and unstamped questions retain their existing meaning. The wording is a provisional online interpretation, not general language understanding. Runtime evidence exercises dealt Traitors and Face Dancers and stages one unresolved Traitor choice; fixtures also stage conserved Truthtrance custody. Twenty-three related freeform files pass 103 cases. Broader questions, full rules and deployed verification remain open.

Freeform No-Field and auction knowledge — 29 October 2026

Prototyped / Partial. New current-fact questions recognize bounded wording for a concealed No-Field's printed value, an inspected current auction card, the current public Richese cache card or a card in Ixians' recorded inspected auction group. Recognized questions can also combine these with other supported current facts using one AND or OR. The editor previews the interpretation and the target alone sees the checked aggregate before replying. Unavailable inspection knowledge yields I don't know; an absent or already revealed No-Field answers No under the existing fact rule.

New interpretations carry version 8; earlier versions and unstamped questions retain their existing meaning. The wording is a provisional online interpretation, not general language understanding. Runtime evidence includes an actual marker shipment and ordinary auction and Ixian group inspections, with a conserved Truthtrance card staged by the fixture. Twenty related freeform files pass 92 cases; broader questions, full rules and deployed verification remain open.

Freeform compound current facts — 29 October 2026

Prototyped / Partial. New current-fact questions can combine two to eight recognized fact clauses with a single AND or OR. This includes combining prediction or forecast knowledge with physical force counts, card facts and personal spice. For example, “Did you predict Guild and do you have at least 3 forces in your reserves?” is checked as one aggregate; the editor previews its interpretation and the target alone sees the result before responding. Unavailable knowledge retains the existing three-valued AND/OR behavior.

Newly recognized expressions carry version 7. Earlier versioned and unstamped questions retain their previous meaning. Mixed operators without explicit supported grouping, ambiguous partitions, current-turn commitments and unrecognized wording remain player-managed. This bounded grammar is a provisional online interpretation, not general language understanding. The runtime fixture stages the Truthtrance card while preserving its physical inventory; seventeen related freeform files pass 79 cases. Full rules and deployed verification remain open.

Freeform predictions, forecasts and forces — 29 October 2026

Prototyped / Partial. New current-fact questions recognize bounded wording for the recorded Bene Gesserit prediction, a submitted storm dial, Fremen's storm forecast, Atreides' known next Spice Blow, and counts of the respondent's physical forces in reserves, Tanks or on the board. For example, “Did you predict Harkonnen?” and “Do you have at least 3 forces in your reserves?” use the same checked facts as the structured editor. Unavailable knowledge yields I don't know; the preview names the interpretation and only the respondent sees the checked answer before replying. Physical counts exclude concealed markers.

Newly recognized questions carry version 6; older versions and unstamped questions retain their original behavior. The bounded wording is an online interpretation, not general natural-language understanding. The focused Spice fixture stages a conserved card to arrange the known forecast; wider grammar, current-turn commitments and deployed verification remain open. Fourteen related freeform files pass 65 cases.

Freeform compound hand-category facts — 29 October 2026

Prototyped / Partial. New fact-scoped freeform questions can combine two supported current-hand facts with one unambiguous AND or OR, including printed weapon/defense categories and supported negative card or category counts. For example, a player may ask whether the respondent holds both a poison weapon and a defense, or holds a Shield and no poison weapons. The public editor previews the interpreted aggregate; the target alone sees the verified answer before publishing it, and minimal bots follow the same authoritative fact. Category meanings use the existing printed primary-role classification.

Newly recognized compositions carry version 5. Existing versions 1–4 and unstamped saved questions keep their original behavior. Ambiguous forms, mixed operators, unsupported wording and current-turn promises remain player-managed; this is a bounded interpretation of the FAQ's general game-related yes/no permission, not general language understanding. The focused runtime fixture stages conserved physical card custody, rather than claiming the exact starting hand was earned. See parser, pure cases and runtime cases. Full rules, deployed and broader combinations remain open.

Freeform negative hand facts — 29 October 2026

Prototyped / Partial. New current-fact questions recognize bounded forms such as “Do you have no Shield?” and “Do you not have any poison weapons?” The editor shows the checked zero-card interpretation, the respondent sees the private answer, and bots use the same truthful answer. Named faces and the existing primary-role categories use their existing count predicates. Newly recognized questions carry version 4; older versions and unstamped questions retain their prior semantics. Ambiguous, unsupported compound, future and other unrecognized wording remains player-managed; the later V5 section adds bounded compound forms. The parser is a provisional online interpretation of the FAQ's game-related yes/no permission, not general language understanding. Parser, runtime cases.

Freeform hand-category facts — 29 October 2026

Prototyped / Partial. New fact-scoped questions can ask whether a player holds a weapon, defense, Worthless card, or projectile/poison weapon or defense, or compare a numeric category count using exactly, at least or at most. The editor shows the canonical printed primary-role interpretation before asking; only the target sees the checked answer before publishing it. Unsupported wording and current-turn promises keep their existing manual semantics.

Newly recognized category questions carry a version 3 interpretation. Existing version 1 and 2 parsers and unstamped questions retain their original meaning. The parser, rule cases and action journeys cover physical hands, truthful answers and minimal legal bots. This is a bounded online interpretation, not general language understanding or complete rules.

Freeform hand size and combined facts — 29 October 2026

Prototyped / Partial. New fact-scoped questions recognize exact comparisons of total cards in hand and two current-spice or named-card conditions joined by one AND or OR. For example, “Do you have at least 3 cards in your hand?” and “Do you have at least 5 spice and a Shield?” receive a displayed interpretation and a verified aggregate answer. The table does not receive separate condition results; all minimal bot profiles use that same aggregate.

Version 2 retains the earlier exact forms for newly asked questions. Saved version 1 questions are validated against the unchanged version 1 parser, and unstamped older questions keep their original behavior. Mixed operators, ambiguous names, unsupported negation, future promises and other grammar remain player-managed with unknown bot answers. This bounded grammar is an online interpretation of the FAQ's game-related yes/no permission, not general natural language understanding. See the parser and runtime evidence.

Recognized freeform current facts — 29 October 2026

Prototyped / Partial. New freeform questions explicitly scoped as current facts recognize the exact forms “Do you have at least/exactly/at most N spice?” and “Do you have a/an [printed card name]?” with case-insensitive wording and optional question mark. The former counts the respondent's own current spice; the latter checks their physical hand. The public editor and pending question show the canonical interpretation. The respondent alone sees the verified answer until publishing it; the server rejects an inconsistent answer. All bot profiles use the same answer path.

Only anchored supported phrases are interpreted. Ambiguous, unsupported, compound, future or otherwise unsupported wording keeps the existing human-managed behavior, and bots answer unknown. Current-turn promise text is never parsed. The server stamps a versioned interpretation on each newly asked recognized question and validates it against its text; older unstamped questions keep their original behavior. This does not implement general language understanding or enforcement of arbitrary promises. The FAQ allows game-related yes/no questions; this limited parser is an online interpretation, not a printed restriction. See the parser and its rule and runtime tests.

Public Advanced Richese cache lot — 29 October 2026

Prototyped / Partial. A structured cache-lot question can also name the currently offered public Richese cache card in Advanced play. Any seated respondent can answer while that actual lot is active. A missing, settled, non-cache or inaccessible lot yields unknown; the private Black Market hand card is never exposed by this question. The existing Yes/No answer and minimal bot paths are unchanged.

The Advanced journey starts an Advanced game and exercises the real cache and Black Market actions, human answers and all four bot profiles. Full Truthtrance, wider combinations and deployed acceptance remain open. The selector expresses the FAQ's general yes/no question permission; it is a provisional online interpretation.

Public Basic Richese cache lot — 28 October 2026

Prototyped / Partial. A separate {kind:'cacheLot', name} fact asks whether the currently offered public Basic Richese cache card has a named printed face. Any seated respondent can give a definite answer while that actual lot is active. At that checkpoint missing, completed, Advanced and Black Market lots returned unknown; Advanced public cache support is described above. Only the offered card is read, not the other cached cards; the existing Atreides inspection question keeps its original meaning. Single and compound editors use the public name registry and minimal bots use the same answer path.

The GF9 FAQ p. 8 permits game-related yes/no questions. This selector and its active-lot boundary are a provisional online interpretation; broader special auctions and full Truthtrance acceptance remain open.

Already answered battle Prescience — 28 October 2026

Prototyped / Partial. A current-fact question can compare the leader, weapon, defense or exact dial already revealed through native battle Prescience. The field must match the answered inspection and the respondent must be entitled to the existing battle insight. Pending or unavailable inspection, a different field and an expired battle return unknown. The normal inspection category conversion is reused, including an Ecaz battle special represented as no element. The question exposes no inspected value to the asker beyond the target's ordinary Yes/No response.

The GF9 FAQ p. 8 permits game-related yes/no and compound questions. This bounded selector expresses that general permission and is not a new printed restriction. Other forms of private knowledge, general freeform interpretation and complete Truthtrance coverage remain unfinished.

Eleven focused files pass 58 cases, including genuine Basic Prescience answers, Truthtrance questions, all four minimal bot responses and battle continuation for all four fields. The native fixture orders unopened cards without staging hands or positions. Controlled component callbacks exercise the selectors; npm run check:quick and the production build also passed. Browser and deployed acceptance remain open.

Advanced Ixian auction knowledge — 29 October 2026

Prototyped / Partial. Native Advanced Ixians can answer the existing structured auction-group question using the private group they actually inspected in the current Bidding phase and turn. Advanced Technology updates the recorded group when a substitution succeeds; a canceled exchange keeps the previous record. An absent, stale, canceled or unauthorized inspection yields unknown. This is the known group, including later sold cards, not a prediction of the next lot or another player's hand. The respondent sees the verified answer and may answer normally; the group remains private.

The Advanced journey starts a native Advanced game, exercises actual inspection and a substitution, and checks the human and minimal legal bot paths. The selector is a software interpretation of the FAQ's game-related yes/no question permission, not a printed restriction. Borrowed Homeworld knowledge, wider interactions, full games and deployed acceptance remain open.

Basic Ixian auction knowledge — 28 October 2026

Prototyped / Partial. The structured fact {kind:'ixPoolPeek', name} asks whether a named printed Treachery Card is in the Basic Ixian's recorded inspected auction group. The group records cards retained after Ixians returned one to the deck; it is not the current unsold lots, and retains cards sold afterward. The current Bidding phase and turn, native Ixian seat and matching recorded controller are required. A canceled, absent, stale or unauthorized inspection yields unknown. Other factions do not receive the group; the target alone sees the verified scalar, and the normal answer publishes the response.

The GF9 FAQ p. 8 permits game-related yes/no and compound questions. This selector and its boundary are a software interpretation, not a printed restriction. At that checkpoint Advanced Technology and borrowed Homeworld inspection were separate. Eight related unit files pass 27 cases, including a genuine Basic inspection, private questions and all-profile responses. Missing and aged observations are explicitly staged; broader combinations and deployed acceptance remain open.

Advanced Black Market inspection — 29 October 2026

Prototyped / Partial. The existing auctionPeek fact now also reports the native Atreides seat's successful inspection of an active Advanced Richese Black Market lot, regardless of its normal, Once Around or Silent method. It uses the offered card bound to that lot, or the reserved physical seller card in older frames. Missing, canceled, expired, completed or other-faction observations return unknown. This does not disclose the card to the questioner. A real Advanced setup exercises the offer, inspection, private Yes/No question and four bot responses for Silent and Once Around; a conserved Truthtrance is transferred between hands by the fixture. Semuta and broader auction interactions remain open. The later Caladan section above adds its occupied inspection.

Atreides ordinary auction knowledge — 27 October 2026

Prototyped / Partial. The structured fact {kind:'auctionPeek', name} asks whether Atreides' currently inspected ordinary auction card has the named printed face. The answer uses only the current lot's successful native inspection. Other factions, canceled or expired inspections and missing lots return unknown. The Black Market extension is described above, while public cache lots use the separate cacheLot fact. The later Caladan section above adds the occupied inspection. The card identity is not added to another seat's view; the selected answer is public through the normal Truthtrance flow. Single and compound selectors use the public printed name registry, and existing bots answer from their own projected scalar.

The GF9 FAQ p. 8 permits game-related yes/no and compound questions. This selector and its software boundary are a bounded interpretation, not a printed restriction. Native Basic/Advanced auction progression exercises an actual card inspection, question, answer and all four minimal AI profiles; the unavailable-inspection case is staged. Eight related unit files pass 32 cases. Broader question language and special-auction interactions remain open.

Atreides Spice Blow knowledge — 27 October 2026

Prototyped / Partial. The structured fact {kind:'spicePeek', value} asks whether Atreides' currently known next Spice Card is a sandworm, Sandtrout, or a blow naming one of the printed territories. A Discovery card at the named territory also matches that territory question. Only the native Atreides seat with an active, uncanceled Shipment and Movement peek receives a definite answer; an unavailable peek, another faction or an empty deck yields unknown. This reports existing knowledge rather than granting a forecast, looking at a future replacement card or promising a future outcome. Both single and compound controls use public face choices; the owner-only scalar, physical Truthtrance handling and four legal bot answer paths remain shared.

The existing general GF9 FAQ p. 8 permits game-related yes/no and combined questions. These particular selectors and entitlement boundaries are software interpretations, not a new printed restriction. Five focused files pass 24 cases; two runtime cases include real Basic/Advanced progression to the successful peek, actual asking and answering, all four bot profiles, a staged canceled-peek state and physical card custody. Full Truthtrance language, broader interactions and deployed acceptance remain open.

14 September 2026. Prototyped, with Partial coverage. This extends the existing question flow with four typed predicates; freeform interpretation, general future commitments and complete mode acceptance remain unfinished.

Authority and semantics

The GF9 November 2020 FAQ, p. 8 permits game-related Yes/No questions and AND/OR combinations. Its unknown-answer flow allows another question or retaining the card. Publisher-indexed text was retrieved again for this batch. These precise predicate definitions implement part of that broad permission; they are not additional printed restrictions.

• Prediction faction: compare the respondent's already-recorded native Bene Gesserit prediction with a named faction.
• Prediction turn: compare the turn in that stored prediction using exactly, at least or at most. This asks what was selected, not whether that future win will happen or whether the player promises to make it happen.
• Submitted storm dial: compare this respondent's locked value while the current Storm phase and its submission remain active. A not-yet-submitted or expired dial is unknown; zero is a real first-turn value, not missing data. The separate combined turn-action question can bind an eligible Basic player’s future exact dial; it does not turn an unsubmitted dial into a recorded fact.
• Known forecast: compare the current storm card only for native Advanced Fremen when the existing knowledge flag is true. Canceled peeks, unknown cards and other factions receive unknown; no new deck lookup or borrowed knowledge is granted. The question concerns the known card, not a promise about eventual storm movement after other effects.

The shared parser accepts canonical faction identities, prediction turns 1–10, dials 0–20 and forecasts 1–6, and rejects malformed comparisons and values. Stored dial entitlement also checks the turn's actual legal dialing range. Missing knowledge yields unknown rather than inventing a choice or an answer.

Connected behavior and privacy

truthKnowledgeOf constructs only the target's eligible facts. Both server answer validation and target-only answer projection use that same snapshot. AND/OR uses the existing three-valued evaluation: a false AND term or true OR term can settle the aggregate despite an unknown sibling. No matching-clause list is published.

The public question and response controls are independent of private values. The target sees the verified scalar, and its answer becomes public through the normal history flow. Neither prediction fields nor an unrevealed dial/forecast are added to rivals' views. Equality answers can naturally reveal the queried value, as the chosen question intends. A definite answer discards the actual Truthtrance once; unknown retains its retry/save continuation. The original storm, auction, battle or other interaction remains in place while the question is open.

The four UI choices share faction/value fields and numeric comparison controls. Both compound clauses support them, with invalid-input and busy submission guards. All four AI profiles answer the existing owner-only scalar. They do not receive opponent knowledge, and this does not claim new strategic question selection or freeform language understanding.

Verification boundary

Focused tests use real Basic/Advanced setup, a real locked prediction and ordinary storm/forecast progression. They stage only the asker's physical Truthtrance acquisition by relocation, without duplicating cards. They cover private Yes/No and unknown answers, changed/expired entitlement, invalid queries, dishonest and duplicate answers, compound indistinguishability, all profiles and JSON views. The isolated component tests check all numeric comparisons and ranges.

Thirteen new cases pass (six rules, two controls and five production SQLite recovery cases); the focused union with existing Truthtrance/spice coverage passes 44 cases. Authenticated saved-seat tests restore pending questions and the original storm-peek response, preserve all private views and physical custody, reject stale, dishonest and duplicate answers, and allow only one competing version write. Malformed saved predicates fail without a database write. Independent source and privacy review found missing numeric comparison controls; integration review also fixed the parent selector mapping and invalid-input submission guard.

Two genuine-setup, four-profile games finish without rejected actions: Basic in 192 actions with five periodic JSON rounds, Advanced in 518 with fourteen. Exact card/force custody holds and all eight final seat views restore. Explicitly scripted legal questions use naturally held cards: the Basic sample asks about a prediction faction and a submitted storm dial; Advanced asks about a prediction turn. Neither sample exercises a forecast question. These are bounded continuation checks, not evidence of autonomous AI question strategy or strength calibration.

The separate browser exercise uses real Advanced setup, a recorded prediction and ordinary forecast progression, relocating only two physical Truthtrances to the asker. It verifies invalid dial input, an unknown answer and rephrasing, a compound faction/turn question, a forecast comparison and refresh after submission and resolution. The final history is unknown/Yes/Yes, both physical cards are discarded, and unrelated resources and private projections remain unchanged. Pending-answer restoration is established by SQLite tests, not a paused browser observation.

All 767 opening games and original seat/recovery records match the backup after the reported outage. Integrity and foreign keys pass; the healthy server is reused. The new browser room brings the preservation baseline to 768 before broad checks. Required broad results, source fingerprint, final preservation totals and Git delivery are recorded in the checkpoint commit and private report. These results do not certify all Truthtrance questions, complete expansions or publication readiness.