{
 "schema_version": "1.3",
 "table_size": 5,
 "id": "tptk-pot-control-vs-tight-coldcaller",
 "format": "multi_option",
 "free": true,
 "rebuild": "v2",
 "provenance_class": "dealt_hand",
 "updated_at": "2026-08-09T00:00:00Z",
 "source": {
  "session_id": "MoreHandReviewwithCurrentHands",
  "verdict_audio": "quiz/audio/tptk-pot-control-vs-tight-coldcaller.m4a",
  "verdict_beats": "RE-CUT 2026-08-08 (this rebuild), 16.21 s, four word-trimmed beats registered in tools/cut_quiz_audio.py, all spk=tyler + spk_confident: w6603-6628 (s296 head, t=2637.44-2642.46) + w6630-6636 (s296 tail, 2644.44-2646.22) + w6714-6739 (s304 head, 2689.04-2695.22) + w6761-6766 (s304 mid, 2703.48-2704.28). The OLD splice (s296 + s305-306) aired Tyler's plan for the whole hand — 'call flop, call turn, hold river' — naming two streets this fixture's decision node never renders (validator I4 bars that on v2 rebuilds; same defect/fix as akk-3way). It also carried ~9 s of dead air (Tyler's 2 s holds inside 'like' and 'call') and a stray fragment that re-ASR'd as 'Money in.' The dropped plan is preserved verbatim in hand_record.spoken_facts and disclosed in unverifiable_claims. New beat 3 adds the MECHANISM and stops before w6749-6750 ('25 13', this villain's HUD line — never aired); beat 4 starts after w6760 'bet' so Tyler's parallel clause about hero's own 1.80 stab cannot read as a fifth option on a raise/call/fold quiz (verified in isolation: re-ASR of the final 1.8 s returns 'I think we've got a call.' with no 'bet'). Echo-checked 2026-08-08, faster-whisper small.en — see aired_transcript. Old clip backed up to scratchpad dh3/tptk-pot-control-vs-tight-coldcaller/old_audio_backup.m4a.",
  "lesson_slug": null,
  "master_slug": "MoreHandReviewwithCurrentHands_master",
  "master_t": 2142.3,
  "src_t": 2637.4,
  "ser_sentences": [
   194,
   195,
   196,
   240,
   243,
   247,
   249,
   250,
   296,
   302,
   303,
   304,
   307,
   308
  ],
  "hand_record": {
   "lane": "FRAME_READ",
   "played": "A real hand of Eric's, reviewed in-session. From t≈1716 to t≈1800 Eric screen-shares his PokerTracker hand replayer over Skype (title bar 'Playback for <room> Hand #… - $1 NL (6 max) - Holdem', typed action-log pane, FLOP/TURN/RIVER transport bar) with the replay PAUSED on this flop decision; Tyler then talks over that frozen node for ~17 minutes (Flopzilla range work on his own screen) before they hit play at t≈2767 and watch the hand out. Frames are READ-ONLY provenance — never aired, never rendered (the title bar carries poker-room branding).",
   "hand_id": "Illegible at this capture resolution. 18x temporal-averaged crops of the title bar (t=1745, 200-frame tmix) resolve only the shape of a ~10-digit id after the '#' (a leading '4' and a '424' run); not asserted.",
   "stake": {
    "source": "Replayer title bar, t=1745: 'Playback for <room> Hand #… - $1 NL (6 max) - Holdem'. The stake glyph sits at the legibility floor of this 1324x720 capture — 8x/14x/18x temporal-averaged crops give '$1 NL (6 max)' in shape but cannot separate '$' from a digit, so the DOLLAR stake is NOT asserted anywhere in this fixture and the room name is withheld per the no-operator-branding policy (recorded only in the audit trail).",
    "note": "bb-NATIVE fixture: the replayer is set to BB display and every amount in this fixture is read in BB directly off the felt — the blind lines in the action log ('posts the small blind 0.50 BB' / 'posts the big blind 1 BB'), the chip plates ('7 BB', '1.80 BB', '6 BB'), the stack plates ('95.70 BB', '93.63 BB') and the pot box ('Pot: 7.80 BB' -> 'Pot: 14.80 BB'). NO $-to-bb conversion was performed anywhere. Weak independent cross-check only: the felt banner reads 'Avg Stack: $88' against a 5-stack hand-start average of 90.07 BB, consistent with bb=$1 and with no other stake; the banner is a table average, not a per-hand figure, so it is corroboration and nothing more."
   },
   "frames": [
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1730 — decision node, villain plate labelled 'RAISE': 'Pot: 14.80 BB', 'Odds: 2.85:1 - 26%', villain chip '7 BB', hero chip '1.80 BB', centre disc '6 BB', hero plate 'Hero - 86% / 95.70 BB'",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1745 — one street-step earlier in the same paused replay: 'Pot: 7.80 BB', 'Odds: 4.33:1 - 19%', hero chip '1.80 BB' out, centre disc '6 BB', villain plate 'P5-d18cedc94… - 15% / 100.63 BB' (7x/8x zoom crops)",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1745 — action-log pane, 4x zoom + 100-frame temporal average (see log_verbatim)",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1775 — plate/chip zooms at the decision: hero '95.70 BB', villain '93.63 BB' (100.63 − 7), villain chip '7 BB'; 4-colour card zooms: hero A on the grey/black spade colour with a legible ♠ pip and Q on the club green with a legible ♣ pip; board J club-green, Q heart-red, 6 diamond-blue, all three pips legible at 5x",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1730 — 12x zoom on the disc in front of hero's seat: PokerTracker's dealer button graphic. It sits at a DIFFERENT seat in the neighbouring hand on the same felt (t=1700, disc at x≈710 rather than x≈619), which is what makes it a button and not decoration; the log line 'Seat 4 (Hero) is the dealer' is the primary source.",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=1745 — HUD line on the villain's seat, 6x zoom: '▯/VP 25/PR 13/3B 0/CB 100' over 'P5-d1/103 BB/FCB 100/(16)'. READ-ONLY: no HUD figure is displayed by this fixture or aired (Tyler speaks '25 13' at w6749-6750 and that span is deliberately outside the verdict cut).",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=2637 — first verdict word (s296). Tyler's screen, not the felt: Flopzilla ('Registered version') with a hand-marked 13x13 range plus a Notepad of max-value-bet arithmetic. NONE of that arithmetic is displayed or aired by this fixture — see gated_numbers.",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=2620 — the villain's OWN marked range, flop-filtered: Flopzilla v1.8.6 grid with '167 combos in preflop range' (15.4%) and the stats panel reading set 1.48 / two pair 1.97 / overpair 0 / top pair 6.90 / middle pair 8.87 / weak pair 32.5 / ace high 8.87 / no made hand 39.4 / OESD 3.94 / gutshot 1.48; Notepad beside it: 'SB - 250 Combos in range?', 'We lose to QJo at 4.5 combos and 66 at 3 combos', 'AQ at 250 combos at 95.8%'. READ-ONLY: gated, never displayed or aired. This is the frame the 2026-08-09 coherence fix rests on.",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=2684 — the CONTRAST frame, taken while Tyler speaks s303: the same Flopzilla grid now holds only the pocket-pair diagonal TT/99/88/77/66/55/44/33/22, '51 combos in preflop range' (4.72%), stats set 5.88 / weak pair 94.1 / two pair 0 / top pair 0 / OESD 0 / gutshot 0; Notepad's last line 'Against set miner capped at 16 times pot (closer to 8 times pot)'. Proves 'a guy who has deuces through tens' is the hypothetical set-miner, not this opponent.",
    "cache/MoreHandReviewwithCurrentHands.mp4 t=2783 / t=2795 — the replay resumed (result provenance only, past this quiz's node): turn 6♠, villain bets 5.50 into 20, 'Pot: 25.50 BB'; river A♦, 'Pot: 31 BB', hero '85 BB', villain '88.13 BB'."
   ],
   "log_verbatim": [
    "Action-log pane, read at 4x with a 100-frame temporal average (t=1745) and re-read at the river (t=2795/t=2815). Every amount below is cross-anchored to a crisp felt element (chip text, plate, pot box) before being asserted. Seat IDs are the room's anonymized handles.",
    "Seat 4 (Hero) is the dealer                      [felt: dealer disc in front of hero, t=1730]",
    "P5-d18cedc94… posts the small blind 0.50 BB      [half-blind; no ante anywhere in the log]",
    "P6-d10a84429… posts the big blind 1 BB           [the BB scale this fixture is denominated in]",
    "P1-581ce51c1… folds",
    "P3-28ce00b6c… folds",
    "Hero raises to 2.50 BB                           [felt: hero 100 BB start -> 95.70 behind after 2.50 + 1.80]",
    "P5-d18cedc94… calls 2 BB                         [felt: villain 103.13 start -> 100.63 at t=1745]",
    "P6-d10a84429… folds                              [1 BB dead -> centre disc '6 BB' = 2.50 + 2.50 + 1]",
    "Flop: [Jc Qh 6d]                                 [felt: 4-colour board, pips legible, t=1775]",
    "P5-d18cedc94… checks",
    "Hero bets 1.80 BB                                [felt: '1.80 BB' chip, 'Pot: 7.80 BB', t=1745]",
    "P5-d18cedc94… raises to 7 BB                     [felt: '7 BB' chip + 'RAISE' plate, 'Pot: 14.80 BB', stack 100.63 -> 93.63, t=1730/1775]  <-- THE DECISION NODE",
    "Hero calls 5.20 BB                               [past the node; result provenance only]",
    "Turn: [6s] / P5 bets 5.50 BB / Hero calls 5.50 BB   ['Pot: 25.50 BB' then 31 BB, t=2783/2795]",
    "River: [Ad] / P5-d18cedc94… checks               [log at t=2815; the share moves on before the log scrolls further]"
   ],
   "seat_derivation": "The title bar says '(6 max)' and the felt shows six seat ovals, but only FIVE are dealt: five name plates (P6, P1, P3, Hero, P5) and one vacant dark oval on the right of the ring. Clockwise from hero (bottom-left) the occupied seats run Hero -> P5 (left) -> P6 (top-left) -> P1 (top-right) -> [vacant, right] -> P3 (bottom-right), which is exactly the log's action order (P5 posts SB, P6 posts BB, P1 acts, P3 acts, Hero acts). With 'Seat 4 (Hero) is the dealer' that fixes: hero = BTN, P5 = SB, P6 = BB, P1 = UTG, P3 = CO, and the vacant seat is the HJ. Rendered exactly 5-handed. The v1 fixture rendered a full 6-max ring with hero in the CO and the villain on the BTN — both wrong, and the position error inverted who was in position.",
   "villain_read": "Tyler's own words about THIS opponent, all spk=tyler + spk_confident: s240 'let's look at the small blind now … he looks rational … we don't know a ton about this guy but he definitely looks rational'; s243 'the tight passive side, if I was just going to mark what sort of hands I thought the guy had I would come up with something like this' (the diarizer splits this utterance mid-phrase at w5778 and gives its first half to Eric — the second half is unambiguously Tyler, who is the one marking the range in Flopzilla; only the tyler-confident half is relied on); s194-195 'what do we know about small [blind] cold calling ranges — pretty tight, right' (SER renders the word as 'small line'; 'small blind' is the ASR-corrected reading, fixed by s240's 'let's look at the small blind now' over the same seat — the load-bearing word, 'tight', is verbatim); s304 'how nut pedal your opponents are and how tight they are pre-flop'. The rendered seat label is therefore 'TIGHT'. 'Tight' here means tight PREFLOP, which is what Tyler measures ('how tight they are pre-flop', s304); it is not a claim that his flop check-raising range is nutted — see coherence_fix_2026-08-09. The v1 fixture's 'NIT' is retired: Tyler never applies 'nit' to any opponent on this tape — a whole-tape scan (848 sentences, 7472 s) returns the stem exactly three times, s651-653 at t=5722-5750, inside a DIFFERENT hand, and there he applies it to HIMSELF ('the nit in me says I should just wait around until I have jack nine or a flush'; 'this is a board texture where I'm going to be exceedingly nitty'). The quiz id and the concept KEY keep the word only as the delivery key the site, the manifest and the registry already link to; renaming a key is a cross-repo change, so it is escalated rather than taken here. CORRECTED 2026-08-09: this field previously ended 'nothing a player reads says nit', and that was FALSE of this quiz's own delivered surface — quiz/concepts.json['top-pair-pot-control-vs-tight-coldcaller'].label read 'Don't stack off top pair vs a nit', and the site RENDERS that label on the reveal card (../wayoflono.xyz/js/quiz.js:415, via conceptLabel() at 529-532; again in the study loop at js/the-way.js:1085-1086). A player did read 'nit'. The label now reads 'Don't stack off top pair vs a tight cold-caller' — the same read the seat label and the context line already carry, sourced to s194-195 / s240 / s243 / s304. See concept_entry_fix_2026-08-09.",
   "stacks": "Hand-start, read off the plates and reconciled: hero 100.00 BB (95.70 behind at the decision + 2.50 + 1.80; the HUD line on hero's seat independently reads 'Hero / 100 BB'), villain 103.13 BB (93.63 behind + 2.50 + 7.00; HUD rounds it to '103 BB'), BB 29.99, UTG 101.50, CO 115.71 (folded plates, t=1745). Effective depth at the decision is therefore 100 BB (hero is the shorter stack) — recorded here rather than in the context line because the villain's true 103.13 BB start would fail validator I7's single-depth conservation check.",
   "result": "Past this quiz's node, and NOT narrated by the fixture: hero called 5.20, the turn came 6♠ pairing the board, the villain bet 5.50 and hero called, the river came A♦ and the villain checked ('Pot: 31 BB', hero 85 BB, t=2795). The showdown is not visible before the share moves on, and Eric's later 'the guy checked called you with king ten' (s317) is his recollection of a different, previously-covered hand — it does not match this hand's log, so it is not used.",
   "spoken_facts": [
    "s296 (aired): 'so you can see if you got somebody who just nut pedaled you truthfully you can't put that much money in at all with that hand like you never want to get stacks in'",
    "s304 (partly aired): 'so you do have to pay a lot of attention here basically of how nut pedal your opponents are and how tight they are pre-flop and so when I'm looking at that hand with 25 13 I'm really I mean I think we've got a bet I think we've got a call but I mean in general here I'm never trying to put the money in'",
    "s305-306 (NOT aired, preserved here): 'so I mean if you want the line here with that hand is basically call that call flop / call turn hold river and' — Tyler's plan for the whole hand. The flop leg of it IS this quiz's key; the turn/river legs name streets the rendered felt never reaches, so they are documented rather than aired.",
    "s307-308 (NOT aired): 'you can corollary this to don't bet if checked to / otherwise you put in more money on the check node than you do on the bet' — the corollary the option_feedback for CALL passes on. Eric arrives at the same place himself at s316 over the replay: 'I probably made the mistake of putting money in on the check node.'",
    "s302-303 (NOT aired, and NOT USED — see coherence_fix_2026-08-09): 'on these type of boards I mean really like what's this dude even gonna bluff with / I mean there's no obvious bluffing combo from a guy who has deuces through tens here'. Its referent is NOT this villain. At t=2684, while Tyler speaks the line, his Flopzilla grid holds only the pocket-pair diagonal TT-22 ('51 combos in preflop range', 4.72%) with the stats panel reading set 5.88 / weak pair 94.1 / two pair 0 / top pair 0 / OESD 0 / gutshot 0 — literally 'deuces through tens', the hypothetical set-miner tag he builds at s292-295 ('if you look at kind of a typical tag range here … they might just legitimately just have like pocket pairs'), and the Notepad line beside it reads 'Against set miner capped at 16 times pot'. The v2 draft imported that read onto the real opponent; it is a property of another model, not of this node.",
    "s249-250 (NOT aired): 'we lose to queen jack 4.5 combos and sixes at three combos' — Tyler's own count of the combos in the range he marked that beat A♠Q♣ on J♣Q♥6♦. The FOLD feedback names QJ and sixes as the hands that beat hero; it does not display either count."
   ],
   "gated_numbers": "Everything Tyler computes on his own screen between s247 and s289 is deliberately absent from this fixture's displayed text and audio: '250 combos in this range', 'our hands get 95 percent of the time' / Notepad 'AQ at 250 combos at 95.8%', '.977/.023', '22 times pot', '42 times pot', 'value bet closer to 11 times pot', 'around 40 big blinds', 'capped at 16 times pot (closer to 8 times pot)'. Reasons, in order: (1) most of them are maximum-money-in figures for OTHER configurations he types out on the Notepad ('BB vs BT' and 'SB vs BT' on a hypothetical QJ622 runout), i.e. other nodes/sizings, which the honest-numbers rule bars even relabelled; (2) the ones that are about this spot are derived from a range Tyler hand-marks in Flopzilla, an assigned-range model, not a measurement at this node; (3) the v1 fixture's principle line displayed the rounded '96%' as if it were a fact about the hand, which is exactly the class of leak the corpus audit was called on. Added 2026-08-09: the Flopzilla CATEGORY readout for this villain's marked range (t=2620 — set 1.48 / two pair 1.97 / overpair 0 / top pair 6.90 / middle pair 8.87 / OESD 3.94 / gutshot 1.48) is gated on the same grounds, even though the coherence fix rests on it; the displayed text names the categories qualitatively and no figure from the panel is shown or aired. The replayer's own hero-equity plate is also NOT displayed: it prints 'Hero - 86%' at t=1730 and 85% at t=1775, and with the villain's cards face-down it is equity versus a random hand, which is not the number that decides this spot.",
   "coherence_fix_2026-08-09": "BLOCKING DEFECT, found by the batch-3 coherence gate and confirmed here from fresh frames. The v2 draft's FOLD feedback asserted 'he is not raising with air: Tyler's read on this exact player is that there is no obvious bluffing combo in his range on this board', and its RAISE feedback said the money 'arrives with QJ and sets'. Together those two strings left the check-raising range as QJ + sets on a rainbow board — a range with nothing A♠Q♣ beats — so the quiz argued FOLD while grading CALL: vs QJ (6 combos after removal — 2 queens x 3 jacks) AQ has THREE clean outs twice, the aces, for ~12% (corrected 2026-08-09 from 'about four clean outs twice (~16%)': a third queen fills QJ up, so queens are not outs), and vs 66 (3 combos) it is drawing to runners (~4%), a blend near 11% against the 26% breakeven this price demands (5.20 to win 14.80; the replayer prints 'Odds: 2.85:1 - 26%'). CAUSE (pixels): the 'no obvious bluffing combo' sentence is Tyler's s303, spoken at t=2684 over a Flopzilla grid holding ONLY TT-22 ('51 combos', OESD 0%, gutshot 0%, two pair 0%, top pair 0%) — the hypothetical set-miner tag he builds at s292-295, with 'Against set miner capped at 16 times pot' typed beside it. THIS opponent's range, marked by Tyler at s243-247 and still filtered on the same flop at t=2620, reads overpair 0 / set 1.48 / two pair 1.97 against top pair 6.90 / middle pair 8.87 / OESD 3.94 / gutshot 1.48: the check-raise cannot be nut-only, and the ~19% of worse combos the price needs is comfortably available. A read true of another model was relabelled onto this node — the same defect class as a figure borrowed from another street. FIX: the false absolute is deleted (retracted in unverifiable_claims), 'sets' is corrected to 'sixes', and both the CALL and FOLD strings now name the worse queens, jacks and open-enders that are actually in the range. REASON CORRECTED 2026-08-09 (the conclusion was right, the stated reason was not): this note used to justify 'sixes, not sets' with 'overpair 0% means no QQ/JJ in the marked range'. That inference is wrong — on J♣Q♥6♦ a pocket QQ or JJ is a SET, not an overpair, so the panel's 'overpair 0%' rules out only AA/KK. The reason that does hold is the grid itself plus Tyler's own count: at t=2620 the QQ and JJ cells sit at exactly the pale, unselected shade AA and KK do (8x zoom, side-by-side against the t=2690 frame where the same four cells are pale and TT/99 are marked), while 66 is marked in dark; and the only sets Tyler counts out loud against A♠Q♣ are 'sixes at three combos' (s250). Sets in this range are therefore sixes. The principle line no longer argues the fold ('the few combos that beat you are exactly the ones putting money in' is gone). NO audio change: the aired clip asserts none of this — it is Tyler's conditional principle plus his verdict, and s303 was never inside the cut. Correct answer stays CALL, which is Tyler's own ruling on this hand and this villain ('when I'm looking at that hand with 25 13 … I think we've got a call', s304, HUD line VP 25 / PR 13 confirmed on the felt at t=1730).",
   "concept_entry_fix_2026-08-09": "BLOCKING HONESTY DEFECT (stage-5 fidelity gate, verdict STOP), confirmed here from my own frames, my own whole-tape scan and the site's own JS, then fixed. The gate found NO disagreement on structure — hand, board order, action, every size, every stack, the graded answer and the aired clip all reproduce independently — and I reproduced the load-bearing ones again at fresh timestamps of my own (t=1766: 'Pot: 14.80 BB', 'Odds: 2.85:1 - 26%', villain chip '7 BB' + 93.63 BB, hero chip '1.80 BB' + 95.70 BB, centre disc '6 BB', board J-club-green / Q-heart-red / 6-diamond-blue, hero A-spade / Q-club, villain HUD 'VP 25 / PR 13'; t=2620 stats panel verbatim set 1.48 / two pair 1.97 / overpair 0 / top pair 6.90 / middle pair 8.87 / weak pair 32.5 / ace high 8.87 / no made hand 39.4 / OESD 3.94 / gutshot 1.48). What failed was viewer-facing HONESTY, one file over. (1) The CONCEPT ENTRY this quiz points at is a delivered, rendered surface — ../wayoflono.xyz/js/quiz.js:415 draws concepts[quiz.concept].label on the REVEAL CARD, and js/the-way.js:1085-1086 draws it in the study loop — and it read 'Don't stack off top pair vs a nit'. That is a villain read Tyler never gives any opponent on this tape (whole-tape scan: the stem occurs only at s651-653, another hand, about himself), i.e. false biography on the one line of this quiz a player is guaranteed to read, while this fixture's own villain_read certified that 'nothing a player reads says nit'. (2) The same entry's description shipped 'Being 96% ahead isn't a stack-off' — the rounding of Tyler's Flopzilla 'AQ at 250 combos at 95.8%' that gated_numbers names as exactly the leak class the corpus audit was called on — beside a pct_provenance_note claiming this quiz displays no percentage. (3) The same description still argued 'the tiny slice that beats you is exactly what puts money in', the fold-arguing absolute coherence_fix_2026-08-09 RETRACTED, and ended 'call flop, call turn, hold river', the multi-street plan the 2026-08-08 audio re-cut removed for naming streets the felt never renders. FIX: label -> 'Don't stack off top pair vs a tight cold-caller' (Tyler's own read, s194-195 / s240 / s243 / s304, and the same word the seat plate and context line already show); description -> 'Top pair, top kicker is far ahead of a tight cold-caller's range, and his check-raise doesn't narrow him to the hands that beat you — so it buys a call, not a raise. Keep the pot cheap and never play one pair for stacks.' — no percentage, no retracted absolute, no street the render never reaches, rebuilt from this fixture's own principle. The concept KEY is unchanged (the manifest, the registry and the site's stored progress all link to it); renaming it is a cross-repo change and is escalated, not taken here. The frozen exporter's CONCEPTS dict was patched in the same pass (tools/export_quiz_json.py) because concepts.json is NOT covered by the fixture-is-truth rule and a --force re-export would otherwise resurrect the leak. NOTHING ELSE CHANGED: no displayed string in this fixture, no audio, no beat registry, no structural number, correct answer still CALL. Frames were read-only. Also fixed in this pass, all provenance-only: the 'TIGHT' plate moved out of cosmetic_invention (a sourced read never belongs there); the 'overpair 0% means no QQ/JJ' inference replaced with the reason that actually holds (QQ/JJ are SETS on this board — the pixels that rule them out are the pale unselected QQ/JJ cells at t=2620 plus Tyler's 'sixes at three combos'); the 'four clean outs (~16%)' estimate corrected to three outs (~12%, a queen fills QJ up); the 'raising 7 to at least ~21' arithmetic corrected to the real minimum legal raise, to 12.20; and the Flopzilla combo/percentage non-reconciliation disclosed in claims[3]. STILL OPEN, escalated rather than acted on: (a) schema_version stays '1.3' — the gate asked for a bump to 1.4 because this file carries aired_transcript/claims/size_claims/unverifiable_claims/pct_provenance, but ALL 50 fixtures in quiz/ are '1.3' and 40+ of them carry that same field set, so a one-file bump would assert a distinction the corpus does not have, and a corpus-wide bump is a schema change that CLAUDE.md rule 6 says must be announced to the site director first; (b) quiz/gates/tptk-pot-control-vs-tight-coldcaller.json does not exist and the I8 ledger interlock will block bundle entry until it does — writing it is the ledger owner's step, not something to fabricate here; (c) the LIVE site still serves the pre-rebuild fixture (../wayoflono.xyz/data/quiz/tptk-pot-control-vs-tight-coldcaller.json, principle: 'Being 95.8% ahead isn't a stack-off…') and the old concept entry — both leaks clear only when this bundle is delivered, and delivery is Eric's gate."
  }
 },
 "context": "ONLINE 6-MAX TABLE, 5 DEALT · BTN vs A TIGHT SB COLD-CALLER",
 "hero": {
  "pos": "BTN",
  "cards": [
   "As",
   "Qc"
  ],
  "stack_bb": 95.7,
  "avatar": "lono_ghoul",
  "button": true
 },
 "seats": [
  {
   "pos": "SB",
   "avatar": "ghoul_637",
   "name": "TIGHT",
   "stack_bb": 93.63,
   "villain": true
  },
  {
   "pos": "BB",
   "avatar": "ghoul_5129",
   "folded": true
  },
  {
   "pos": "UTG",
   "avatar": "ghoul_2851",
   "folded": true
  },
  {
   "pos": "CO",
   "avatar": "ghoul_4509",
   "folded": true
  }
 ],
 "streets": [
  {
   "name": "preflop",
   "actions": [
    {
     "pos": "UTG",
     "act": "fold"
    },
    {
     "pos": "CO",
     "act": "fold"
    },
    {
     "pos": "hero",
     "act": "open",
     "to_bb": 2.5
    },
    {
     "pos": "SB",
     "act": "call"
    },
    {
     "pos": "BB",
     "act": "fold"
    }
   ],
   "pot_bb": 6
  },
  {
   "name": "flop",
   "board": [
    "Jc",
    "Qh",
    "6d"
   ],
   "actions": [
    {
     "pos": "SB",
     "act": "check"
    },
    {
     "pos": "hero",
     "act": "bet",
     "to_bb": 1.8
    },
    {
     "pos": "SB",
     "act": "raise",
     "to_bb": 7
    }
   ]
  }
 ],
 "prompt": "A♠Q♣ · TOP PAIR, TOP KICKER · YOUR FLOP BET GETS CHECK-RAISED TO 7 — RAISE, CALL, OR FOLD?",
 "pot_bb": 14.8,
 "options": [
  {
   "id": "raise",
   "label": "RAISE"
  },
  {
   "id": "call",
   "label": "CALL"
  },
  {
   "id": "fold",
   "label": "FOLD"
  }
 ],
 "correct": "call",
 "option_feedback": {
  "raise": "Raising is the one thing Tyler rules out. This player cold-called from the SB — a tight range Tyler marks out by hand on the tape — and the part of it that wants a bigger pot on this board is queen-jack and sixes. Raise and you build exactly the pot those hands want built. Tyler's rule here is a calibration, not a label: the more nut-peddling your opponent is, the less money you can put in with one pair — and you never want to get stacks in with it.",
  "call": "Correct — A♠Q♣ is ahead of nearly all of the range he cold-calls the SB with, and a check-raise does not narrow him to the hands that beat you: the range Tyler marks out on the tape also holds worse queens, jacks and open-ended draws. So paying 5.20 into 14.80 with top pair, top kicker is clear, and folding it is far too weak. What you must not do is escalate — he is the one raising, and against a player this tight you never want to get stacks in. Call, keep the pot cheap, and take it to showdown.",
  "fold": "Too weak. Tyler marks this guy's cold-calling range out combo by combo on the tape, and A♠Q♣ beats nearly all of it — what actually has you beaten is queen-jack and sixes, a countable handful, while the same range is stocked with worse queens, jacks and open-ended draws that have every reason to raise your bet. At 5.20 into 14.80 you do not need him to be bluffing often for this to be a clear call. The mistake Tyler is warning about here is the opposite one — stacking off — and calling is exactly what avoids it."
 },
 "verdict_text": "If you got somebody who just nut-peddled you, truthfully you can't put that much money in at all with that hand — you never want to get stacks in. You do have to pay a lot of attention to how nut-peddling your opponents are and how tight they are preflop. I think we've got a call.",
 "principle": "Top pair, top kicker is one pair. A tight player's check-raise doesn't narrow him to the hands that beat you, so it buys a call rather than a fold — but the more nut-peddling he is, the less you can afford to escalate. Pay the raise, keep the pot cheap, and never play one pair for stacks.",
 "concept": "top-pair-pot-control-vs-tight-coldcaller",
 "concepts_secondary": [],
 "topic": "pot-control",
 "difficulty": 5,
 "aired_transcript": "And so you can see if you got somebody who just nut-pedaled you, truthfully, you can't put that much money in at all with that hand. Like, you never want to get stacks in. So you do have to pay a lot of attention here, basically, of how nut-pedaled your opponents are and how tight they are pre-flop. And I think we've got a call...",
 "aired_transcript_note": "Verbatim faster-whisper small.en re-ASR of quiz/audio/tptk-pot-control-vs-tight-coldcaller.m4a (16.21 s, re-cut + re-run 2026-08-08). Beats w6603-6628 + w6630-6636 + w6714-6739 + w6761-6766. Three decoder artifacts, checked against the SER words: the opening 'And' and the closing 'And' are the decoder's connectives at splice heads (the isolated final beat re-ASRs as 'I think we've got a call.' with no preceding word, and w6760 'bet' ends 0.06 s before the beat's head handle); 'Like,' at 5.7 s is the audible tail of w6629 'like', whose SER span holds a 2 s pause. No cards, positions, sizes, streets past the flop, HUD figures or population magnitudes are aired.",
 "pct_provenance": {},
 "pct_provenance_note": "Intentionally empty: this fixture displays no percentage anywhere. The figures that were available — Tyler's Flopzilla '95 percent' / Notepad '95.8%' (an assigned-range model, and the v1 principle displayed it rounded to '96%'), the replayer's 'Hero - 85%' plate (equity versus a random hand, the villain's cards being face-down), the villain's 'VP 25 / PR 13' HUD line and the replayer's own 'Odds: 2.85:1 - 26%' — are all documented in source.hand_record and claims[] instead of being shown. The 26% is recomputed in claims[] as the price this fixture's own numbers imply. SCOPE CORRECTED 2026-08-09: 'this fixture displays no percentage anywhere' was true of this JSON file and FALSE of the surface this quiz delivers — the concept entry it points at, quiz/concepts.json['top-pair-pot-control-vs-tight-coldcaller'], carried the gated 95.8% rounded to '96%' in its description until today. The concept entry ships in the same bundle as the fixture and the site renders its label, so it is inside the rule-7 boundary, not outside it. It is now percentage-free (see concept_entry_fix_2026-08-09) and the claim above holds of both files.",
 "size_claims": {
  "_note": "No BET-size adjective (small / tiny / half-pot / pot-sized / overbet / huge …) appears in any displayed string, in any option label or in the aired clip; 'SB' is a position, not a size word. Two POT-size words do appear and are deliberate, not sizing claims: 'bigger pot' in the RAISE feedback (corrected 2026-08-09 — this read 'raising 7 to at least ~21', which was wrong: the raise increment is 5.20, so the MINIMUM legal raise is to 12.20, and even that smallest possible raise makes a 30.40 BB pot if called — 6.00 + 12.20 + 12.20 — against the 20.00 BB pot a call makes; any realistic raise sizing is larger still, so 'bigger pot' holds at every legal size) and 'cheap' in the CALL feedback and principle, which describes the line as a whole — calling adds 5.20 to make a 20 BB pot and leaves 90.50 behind, against the alternative of playing a 100 BB stack for it. These entries document the real bets read off the replayer felt so the numbers stay checkable.",
  "open to 2.5": {
   "street": "preflop",
   "actor": "hero (BTN)",
   "to_bb": 2.5,
   "source": "action log 'Hero raises to 2.50 BB' (t=1745); felt: hero 100 BB start -> 95.70 behind after 2.50 + 1.80"
  },
  "flop bet 1.80": {
   "street": "flop",
   "actor": "hero (BTN)",
   "to_bb": 1.8,
   "pct_of_pot": 30.0,
   "source": "felt chip '1.80 BB' into the settled 6 BB, 'Pot: 7.80 BB' (t=1745); log 'Hero bets 1.80 BB'"
  },
  "flop check-raise to 7": {
   "street": "flop",
   "actor": "SB",
   "to_bb": 7,
   "pct_of_pot": 89.7,
   "source": "felt chip '7 BB' + 'RAISE' plate, 'Pot: 14.80 BB', villain stack 100.63 -> 93.63 (t=1730/1775); log 'P5… raises to 7 BB'. Raise-size framing: 7 into a 7.80 BB pot, i.e. hero must add 5.20 to a pot that would be 20."
  }
 },
 "claims": [
  {
   "text": "the decision price: the SB check-raises to 7 over hero's 1.80; a call is 5.20 more to win 14.80",
   "lhs": "hero's price at the node",
   "op": "=",
   "rhs": "5.20 / (14.80 + 5.20) = 26.0% breakeven; the replayer itself prints 'Odds: 2.85:1 - 26%' (14.80 / 5.20 = 2.846)",
   "source": "pixels t=1730/t=1775 (chip plates, pot box, odds line); arithmetic recomputed"
  },
  {
   "text": "pot chain: 6 preflop -> 7.80 after hero's bet -> 14.80 facing the raise",
   "lhs": "recomputed street pots",
   "op": "=",
   "rhs": "2.50 + 2.50 + 1 dead BB = 6.00; 6.00 + 1.80 = 7.80; 7.80 + 7.00 = 14.80",
   "source": "every on-screen pot element matches: centre disc '6 BB', 'Pot: 7.80 BB' (t=1745), 'Pot: 14.80 BB' (t=1730/1775)"
  },
  {
   "text": "hero is the BUTTON and the villain is the SB cold-caller — hero is IN POSITION for the whole hand",
   "lhs": "seat/position derivation",
   "op": "=",
   "rhs": "log 'Seat 4 (Hero) is the dealer' + 'P5 posts the small blind 0.50 BB' + 'P6 posts the big blind 1 BB'; dealer disc in front of hero's seat (t=1730) and at a different seat one hand earlier (t=1700); five dealt seats, one vacant oval",
   "source": "pixels t=1730/t=1745; Tyler's framing agrees ('let's look at the small blind now', s240; 'what do we know about small blind cold calling ranges', s194)"
  },
  {
   "text": "the villain's marked range is NOT nut-only on this flop — it contains a large majority of holdings A♠Q♣ beats, including worse queens, jacks, open-enders and gutshots",
   "lhs": "Flopzilla category readout for Tyler's hand-marked SB range on J♣Q♥6♦ (t=2620, '167 combos in preflop range' = the 250-combo range minus As/Qc/Jc/Qh/6d)",
   "op": "=",
   "rhs": "overpair 0% and set 1.48% and two pair 1.97% (the part that beats AQ, matching his own 'we lose to QJo at 4.5 combos and 66 at 3 combos') against top pair 6.90%, middle pair 8.87%, weak pair 32.5%, OESD 3.94%, gutshot 1.48% (the part AQ beats)",
   "source": "pixels t=2620 (Flopzilla stats panel + 13x13 grid + Notepad); CONTRAST t=2684, where the grid holds only TT-22, '51 combos', OESD 0% / gutshot 0% — the hypothetical set-miner the 'no obvious bluffing combo' line (s303) actually describes. No figure from either readout is displayed or aired; the claim recorded here is qualitative and is what makes CALL coherent at a 26% price. RECONCILIATION NOTE (2026-08-09): the panel's percentages do NOT divide out to whole combos against the combo counter on the same screen (set 1.48% of '167 combos' = 2.5, while Tyler counts 'sixes at three combos'), and his own QJo count — '4.5 combos', s249-250 — is itself fractional. Two causes, both visible in my own frames: the range was being EDITED live while it was read (I sampled '245 combos in preflop range' at t=2612, '203' at t=2618, '167' at t=2620, '144' at t=2624, '70' at t=2630 as Tyler clicked cells out, and the stats panel moves with it — set 1.38 / two pair 1.83 / gutshot 5.05 at t=2618 versus 1.48 / 1.97 / 1.48 two seconds later), and Flopzilla's percentile slider admits partially-weighted cells, which is what produces half-combos. So no combo count anywhere in this fixture is derived from a percentage: the range claim rests on WHICH CELLS ARE MARKED (qualitative, re-read at 8x) plus Tyler's spoken counts, and neither the cell counts nor the panel figures are displayed or aired."
  },
  {
   "text": "effective depth at hand start is 100 BB (both players are still that deep at the decision, minus what is already in front of them)",
   "lhs": "min(hero start, villain start)",
   "op": "=",
   "rhs": "min(95.70 + 2.50 + 1.80, 93.63 + 2.50 + 7.00) = min(100.00, 103.13) = 100.00",
   "source": "plate zooms t=1775 + the hero HUD's own '100 BB'. Kept out of the context line only because validator I7 conserves a single declared depth across every seat and the villain genuinely started 3.13 BB deeper."
  },
  {
   "text": "the hand class is top pair, top kicker with the queen that matches the jack",
   "lhs": "A♠Q♣ on J♣ Q♥ 6♦",
   "op": "=",
   "rhs": "hero's Q pairs the highest board card, the A is the best possible kicker; the club queen is the blocker Tyler asks for out loud ('we really want to block jack queen', s227; 'I guess we've got the right club, I mean the queen of clubs is our… we've got the highest value queen', s315)",
   "source": "5x/6x 4-colour card zooms t=1775 with pips legible; log 'Flop: [Jc Qh 6d]'; Tyler reads both back on tape at s253-255"
  },
  {
   "text": "no ante, no straddle, standard half small blind",
   "lhs": "preflop dead money",
   "op": "=",
   "rhs": "0.50 + 1 posted, hero opens 2.50, SB calls to 2.50, BB folds 1 dead -> the centre disc reads exactly '6 BB'",
   "source": "log blind lines + felt disc, t=1745"
  }
 ],
 "unverifiable_claims": [
  "The DOLLAR stake. The title bar's limit field resolves in shape only ('$1 NL (6 max)') at 1324x720 even under 18x temporal-averaged crops, so no $ figure is asserted; the fixture is bb-native and needs no conversion. The felt banner 'Avg Stack: $88' against a 90.07 BB hand-start average is corroboration for bb=$1 and nothing stronger.",
  "The hand id. Only a ~10-digit shape after the '#' survives averaging; not asserted.",
  "The action-log pane is shape-level at this resolution. Every amount in log_verbatim is asserted only where a crisp felt element (chip text, plate, pot box) confirms it — the anchors are listed line by line — and the two lines with no felt anchor (the two 'folds' before hero's open) affect nothing but the seat derivation, which the dealer disc confirms independently.",
  "Tyler's 'call flop, call turn, hold river' (s305-306) is NOT aired by this quiz's clip and its later legs are not displayed: the rendered felt stops at the flop decision, and a verdict that prescribes streets the viewer never sees is the defect validator I4 exists to catch. The flop leg is the graded answer; the rest is preserved in hand_record.spoken_facts.",
  "The seat label 'TIGHT' is Tyler's stated read on this real opponent (s243 'the tight passive side'; s194-195 'small blind cold calling ranges — pretty tight'; s304 'how tight they are pre-flop'), not a verifiable fact about the account and not population data. Note the SER diarizer splits the s242/s243 utterance mid-phrase; only the tyler-confident half is relied on.",
  "RETRACTED 2026-08-09. The v2 draft's FOLD feedback carried 'he is not raising with air: Tyler's read on this exact player is that there is no obvious bluffing combo in his range on this board.' That is false of this opponent and is removed: s302-303 is spoken over a 51-combo pocket-pairs-only set-miner range (t=2684 frame), not over the 250-combo range Tyler marked for this SB. See coherence_fix_2026-08-09.",
  "The composition now named in the CALL/FOLD feedback — 'worse queens, jacks and open-ended draws' — is read off the flop-filtered stats panel of the range TYLER HAND-MARKED for this opponent (t=2620: top pair 6.90 / middle pair 8.87 / OESD 3.94 / gutshot 1.48 / overpair 0 / set 1.48 / two pair 1.97, over '167 combos in preflop range' = the 250-combo range minus the five known cards). It is an assigned-range judgment, not a measurement, and no figure from it is displayed or aired. Which of those combos actually chose to check-raise is unknowable — the fixture claims only that they exist in the range, which is what the 'no bluffs' line wrongly denied.",
  "Hero's in-game action IS known and is deliberately not narrated by the quiz: he called, called a 5.50 turn bet, and saw the villain check the A♦ river (t=2783/2795). Nothing in the fixture claims or implies a result."
 ],
 "cosmetic_invention": [
  "Ghoul avatars (lono_ghoul hero, ghoul_637 SB, ghoul_5129 BB, ghoul_2851 UTG, ghoul_4509 CO) — art only; the room shows anonymized handles (P5-d18cedc94…, P6-d10a84429…, P1-581ce51c1…, P3-28ce00b6c…).",
  "That is the entire list. Nothing else on this felt is invented — see sourced_not_invented. Removed 2026-08-09: the villain's display name 'TIGHT' used to be listed here as well as in sourced_not_invented. It is a SOURCED read (s194-195 / s240 / s243 / s304), and the DP1 R2 ruling puts sourced details in sourced_not_invented and NEVER in cosmetic_invention — filing a real read as cosmetic is how an unsourced nameplate would get waved through. Its single entry is now sourced_not_invented item 6."
 ],
 "sourced_not_invented": [
  "Hero's cards A♠Q♣ and the board J♣ Q♥ 6♦, SUITS INCLUDED — read off the 4-colour deck with the pips legible at 5x (t=1775), confirmed by the action log ('Flop: [Jc Qh 6d]') and by Tyler reading both back on tape (s253-255 'jack queen six club heart diamond … ace of spades queen of clubs'). The club queen is load-bearing, not cosmetic: Tyler spends s213-232 counting which queen blocks the villain's QJ combos and returns to it at s315 ('the queen of clubs is our… the highest value queen').",
  "Positions BTN / SB / BB / UTG / CO — derived from the log's 'Seat 4 (Hero) is the dealer' plus the blind posts and the clockwise seat ring, corroborated by the dealer disc's position (t=1730 vs t=1700).",
  "The 5-handed structure at a 6-max table, with the HJ seat vacant — five name plates and one empty oval on every frame of the paused replay.",
  "Every amount: the 2.50 open, the 1.80 flop bet, the 7 check-raise, the 6 BB settled pot, the 14.80 facing pot, hero's 95.70 and the villain's 93.63 — all read in BB off the felt, no conversion.",
  "The flop action shape — villain CHECKS, hero BETS, villain CHECK-RAISES. The v1 fixture had hero checking and facing a 4 BB bet, which inverted who applied the pressure.",
  "'nit' was Tyler's word for HIMSELF (s651-653), never for this opponent, and as of 2026-08-09 it appears nowhere in this fixture at all — the quiz id and concept key were renamed to ...-vs-tight-coldcaller (quiz/renames.json, 301 site-side) rather than asserting an unverifiable 'nowhere a player can read'. The sourced read is 'tight' (s194-195, s240, s243, s304), which is what the seat plate shows.",
  "The range composition the CALL and FOLD feedback lean on — 'worse queens, jacks and open-ended draws' — read off the flop-filtered stats panel of Tyler's own marked range at t=2620 (top pair, middle pair, OESD and gutshot all non-zero; overpair zero). Not invented, and not the same range as the pocket-pairs-only set-miner at t=2684.",
  "The graded answer CALL — Tyler at this node: 'I think we've got a call' (s304, aired), 'in general here I'm never trying to put the money in' (s304), 'the line here with that hand is basically call … flop' (s305)."
 ],
 "decision": {
  "pot_settled_bb": 6.0,
  "bets": {
   "hero": 1.8,
   "SB": 7
  }
 },
 "coherence_fix_2026-08-09b": {
  "defect": "principle + option_feedback.raise asserted the villain IS a nut-peddler. That is the antecedent of Tyler's CONDITIONAL ('if you got somebody who just nut-peddled you') stated as fact about this seat, and claims[3] denies it: his marked range is not nut-only. Applied literally the premise grades FOLD (blend ~11% equity vs a QJ+66 raising range, price demands 26.0%), contradicting the graded CALL.",
  "fix": "both strings re-framed to Tyler's actual calibration — 'the more nut-peddling he is, the less you can afford to escalate' — which supports CALL and matches s-tape. Also removed the off-node 'when he checks to you, don't bet at all' from the CALL feedback: it advised against an action the render shows hero taking, and the 2026-08-08 audio re-cut had already excised the parallel spoken clause for that exact reason.",
  "unchanged": "hand, board, sizes, stacks, correct answer, verdict_text, aired_transcript, audio"
 }
}