{ "_meta": { "built": "2026-08-28", "note": "puzzle-book content: what + pindar rationale per method" }, "canon": [ "\"The seed phrase is 11 glitches ending with a random word.\" — the clue", "\"…they have been looking for glitches all wrong.\" — Pindar, 2022-08-25", "\"It was backtested and I can get into the wallet… even knowing how to solve it, it was hard.\" — Pindar, Discord", "The puzzle is in ALL 512 pieces, not just the Act 1 Review. — Pindar via the collector", "\"use the minting order of the bitGANs\" — both solved seed puzzles", "\"Their essence is just a distraction, they hide their true BIP39 nature elsewhere\" — Solstice clue", "\"It begins with sunrise and it ends with sunset\" — Solstice (ends pinned when hinted)", "\"it doesn't matter how you find the seedphrase and there is no judge\" — byteGAN", "\"I can't give any more clues.\" — everything needed already exists", "\"There are some glitch bitGANs with intentional errors - you will find others - they are glitched\" — Pindar (Discord, 2022-01-01, re: a Rainbow skull missing its SubType)" ], "selections": [ { "id": "SEL-TWIN", "name": "Twin class", "status": "ratified", "members": [ 47, 59, 73, 121, 124, 132, 139, 157, 298, 323, 373, 379, 395 ], "plain": "Thirteen pieces that show two side-by-side creatures whose bodies never merge. Each drop draws twins its own way: skulls as two figures, robos as tiled repeats, ghosts as one wide two-eyed body, monsters as two heads. #379 (Spin Twins) is the named exception - its two figures DO merge.", "pindar": "Pindar names these pieces Twin/Twinbot/Twinvader — the name channel and the collector's eye agree completely, which is the strongest authorship signal we have. Each drop renders twins in its own dialect, meaning the class was carried deliberately across drops rather than emerging by accident. Twins read as the CLASSICAL pair in the quantum story that runs from qubitGAN to the Quantum Glitch finale." }, { "id": "SEL-ENTANGLED", "name": "Entangled class", "status": "ratified", "members": [ 217, 222, 226, 244, 260, 263, 281, 411, 436, 450, 465, 473, 481, 489, 499 ], "plain": "Fifteen pieces named 'Entangled' whose two bodies are mixed together instead of side by side. Twin = classical pair; Entangled = quantum pair. Their essences include loaded words like key, solution, book, believe.", "pindar": "Pindar's own word — fifteen pieces literally named 'Entangled', and the art shows intermingled bodies exactly where the name promises them. Entangled is quantum vocabulary: qubitGAN #130 leads in, the Entangled pieces carry it, #512 'Quantum Glitch' closes it, and the NEXT collection is about quantum computers. The essences hiding in this class (key, solution, book, believe) look like a wink." }, { "id": "SEL-NAGA", "name": "Palindrome (naga) class", "status": "ratified-complete", "members": [ 129, 141, 147, 159, 162, 174, 180, 192 ], "plain": "Eight robo pieces with palindrome names (Taco Cat, Dog God...) sitting at ALL eight possible palindrome binary positions, animated forward-then-backward. Triple-coded and mathematically complete - authored beyond doubt.", "pindar": "The most over-determined authorship in the collection: palindrome names, ALL EIGHT possible palindrome binary positions occupied, and forward-then-backward animation. Three independent channels agree, and the set is mathematically complete — nobody does that by accident. If any class teaches us how Pindar signs his work, it is this one." }, { "id": "SEL-PAIRS", "name": "Deliberate pairs", "status": "ratified-11-pairs", "members": [ 9, 14, 16, 73, 75, 100, 130, 133, 165, 170, 175, 205, 207, 209, 300, 363, 473, 479, 480, 489, 491 ], "pairs": [ [ 14, 75 ], [ 479, 480 ], [ 209, 300 ], [ 489, 491 ], [ 130, 133 ], [ 165, 207 ], [ 363, 473 ], [ 16, 100 ], [ 175, 205 ], [ 9, 73 ], [ 170, 175 ] ], "pair_reasons": { "14-75": "Byte-identical GIF files: #14 displays #75's animation. The only exact-duplicate pair in all 512.", "479-480": "The two held-frame pieces — one frame parked (delay) at f27 and f24 — and a shared one-off colour palette found nowhere else in the collection.", "209-300": "Solar skullGAN and Lunar skullGAN: day and night. The solstice drop's first and last pieces.", "489-491": "The scanline pair: both Frozen skullGANs share the scanline effect; #491 also carries the colour-speedup.", "130-133": "The same background flash — the background flips into sine waves on both. qubitGAN #130 and its echo #133.", "165-207": "The two hand-painted number carriers: #165 carries '81', #207 carries '4'. Confirmed but flagged 'suss — could be coincidence' by the collector.", "363-473": "Essences TRUE and FALSE — the only two, a matched statement. The collector's open idea: each may describe the piece's relationship to its own drop.", "16-100": "Both titles spell 'Primoridal' — Primordial with two letters transposed. The only two in all 512.", "175-205": "Both name suffixes are miscased 'Gan' instead of 'GAN' (skullGan / spookyGan). The only two in all 512; a cross-drop pair (robo + spooky).", "9-73": "Both Arweave filenames spell 'gltich' — Glitch with two letters transposed (009_s_silver_gltich.gif / 073_twin_gltich.gif). The only two in all 512. #73's Type field carries the same transposition ('Gltich').", "170-175": "Both SubTypes spell 'Raibow' — Rainbow missing its n. The only two in all 512. Note #175 also sits in pair P09 via its 'Gan' suffix." }, "plain": "Eleven two-piece sets the artist built on purpose - each pair shares ONE one-off defect that appears on exactly those two pieces in all 512 (identical files, a shared typo, matching one-off colours, day/night, TRUE/FALSE). Nine were ratified by the collector (D2); two were found 2026-09-06 by a deck-wide typo census: {9,73} share the filename typo 'gltich' and {170,175} share the SubType typo 'Raibow'. Exactly eleven pairs. #175 sits in two. Essence read of all 1,536 one-per-pair elevens swept DEAD (D43b) - the pairing is the fact, the words are still open.", "pindar": "Pindar plants the same one-off error on exactly two pieces and nowhere else — a byte-identical file, a shared typo ('Primoridal', 'Gan', 'gltich', 'Raibow'), a shared flash, day and night, TRUE and FALSE. A defect that appears on precisely two of 512 is not an accident; it is a signature. The census lands on exactly eleven such pairs — the puzzle's number — with #175 sitting in two of them. Every pair is shown side by side below with the reason it is a pair." }, { "id": "SEL-PAINTED", "name": "Painted-mark class", "status": "ratified-closed-census", "members": [ 165, 179, 207, 379 ], "unsure": [ 414 ], "values": { "165": "81", "179": "10 . 00 (maybe)", "207": "4", "379": "AI" }, "plain": "Four pieces where the artist hand-painted a small number or letters into the animation: #165 carries 81, #179 maybe 10.00, #207 carries 4, #379 carries AI. All 512 pieces have now been scanned frame by frame - this list is complete. #414 is parked as unsure.", "pindar": "Hand-painted numbers are the closest thing to a signature move: the byteGAN puzzle hid the word 'noohp' painted inside the art itself. Four carriers survived a complete, tested census of all 512 (81, 10.00, 4, AI). Small class — but under the collector's mix-and-match doctrine a 4-piece class is a component, and painted numbers beg to be read as indices." }, { "id": "SEL-CREATURE-11", "name": "Visible-creature eleven", "status": "DEAD", "members": [ 5, 13, 34, 55, 147, 159, 180, 204, 264, 268, 412 ], "plain": "Eleven pieces whose visibly-depicted creature is a BIP-39 word (whale, spirit, heart, warrior, cat, dog, wolf, clown, + the collector's eye: #264 dog, #268 bird, #412 monster). The cleanest OBJECTIVE non-essence eleven found — you read what the piece IS. DEAD: swept every creature-vs-essence combination x 7 orderings (~530 total, incl. mint), plant-validated, no hit. the collector's eye also showed the census mislabeled 3 solstice/ghost pieces (they are a tribute + two monsters, not rare creatures), so the clean 11 was partly an artifact. Naga trio (Taco Cat/Dog God/Flow Wolf) + Bird monsterGAN have creature==essence.", "pindar": "The most objective eleven the project produced: read what each piece visibly IS. Eleven pieces depict a creature whose name is a seed word (whale, spirit, heart, warrior, cat, dog, wolf, clown, bird, monster). It sits squarely inside the collector's no-guess rule — anyone can see it is a whale. It is also DEAD: every combination of creature-word vs essence-word across seven orderings (including mint, the solved precedent) was swept with no hit, and the collector's eye showed three of the eleven were mis-catalogued (a tribute and two monsters, not rare creatures). A clean idea, thoroughly closed." }, { "id": "SEL-COLOR-ESSENCE", "name": "BIP color-word essences", "status": "DEAD", "members": [ 145, 164, 166, 174, 189, 196, 262, 338, 378, 419 ], "plain": "The ten pieces whose essence is one of the ten plain color words on the BIP-39 list (black, blue, brown, gold, green, olive, orange, pink, silver, yellow). Every color word on the list is used exactly once; none in the skull drop; five sit in the robo drop where essences are hand-chosen. Seven color-ish essences (color, bright, pumpkin, avocado, coffee, rose, cinnamon) also appear once each. DEAD: swept the ten colors + each color-ish 11th word (first/last/inline) under gan/mint/alpha/spectrum/drop orders forward and reversed, plus every eleven from the 17 color-ish words: 94,106 orderings, 12.0M seeds, plant-validated, no hit (D46, 2026-09-08). [Note, 17 Sep 2026: the list also carries rose, bronze and eight other colour words; rose is #446's essence and bronze is #493's. The colour tag was widened to include them — see D67.]", "pindar": "A set defined by the wordlist, not by a guess: the BIP-39 list holds exactly ten plain color words, and every one of them is an essence exactly once — never repeated, never skipped, none in the skull drop, five in the robo drop where Pindar chose essences by hand (the same drop where the first Rainbot carries the essence 'rain'). Seven color-ish essences (color, bright, pumpkin, avocado, coffee, rose, cinnamon) also appear once each. DEAD: the ten colors with each color-ish 11th word first, last or sorted in, under GAN, mint, alphabetical, spectrum and drop orders forward and reversed, plus every eleven drawn from the seventeen color-ish words — 94,106 orderings, 12 million seeds, plant-validated, no hit. The completeness of the set is a real observation about how the essences were assigned; it is not a word source. [Note, 17 Sep 2026: the list also carries rose, bronze and eight other colour words; rose is #446's essence and bronze is #493's. The colour tag was widened to include them — see D67.]" }, { "id": "SEL-ANCHORS", "name": "Bookends: #0 universe and #512 final", "status": "ratified; shape open", "members": [ 0, 512 ], "plain": "Pindar's anchor pieces sit outside the numbered run by their very traits: #0 the Pandimensional skullGAN (essence 'universe', Dimensions 'infinite', the piece before the first) and #512 the Quantum glitchGAN (essence 'final', the only Dimension-11 piece, the last). Standing rule (D65): a hypothesis may be NINE or TEN glitch pieces with #0 and #512 as bookends, or ten with one bookend, and must say which. The Solstice puzzle pinned its ends the same way ('It begins with sunrise and it ends with sunset'). Bookend variants of the ratified visual classes were swept on the essence register only (D30, no hit); the shape itself is open. #357 (essence 'seed') was retired as an anchor on 2026-09-14 (D66): it is a mid-drop ghost with ordinary traits, an interesting observation, not a bookend.", "pindar": "The Solstice puzzle honestly pinned its ends ('begins with sunrise… ends with sunset'). Pindar built two pieces that stand outside the 512 by their very traits: #0 with Dimensions 'infinite' and essence 'universe', #512 with Dimensions 11 and essence 'final'. If the Act 1 phrase is pinned, these are the pins — which means the glitches may number nine or ten, not eleven, and every hypothesis should say which shape it assumes." }, { "id": "SEL-SPOOKY-TRIO", "name": "Spooky trio variants", "status": "ratified-variants", "members": [ 205, 206, 207 ], "plain": "The spooky drop's contribution could be any of three: #205 (miscased name), #206 (the only uncostumed spooky - a robot at a costume party), or #207 (painted 4). They cover Dimensions 3, 4 and 5, so run all three as variants.", "pindar": "The spooky drop is a costume party, and each of the three candidates breaks it differently: #205 miscased, #206 uncostumed (the only robot at the party), #207 painted. All three are burnt. the collector's ruling: don't pick — run them as variants and let the recipe decide which dimension it needs." }, { "id": "SEL-INVERTED", "name": "Inverted-frame class", "status": "confirmed-bytes", "members": [ 231, 256, 274, 479 ], "plain": "Pieces with a color-negative frame spliced in, proven at the byte level. #256 has two. #479 also carries the timing-hold glitch.", "pindar": "Byte-proven negative frames — no eye judgment needed. The trio #231/#256/#274 even shares an identical negative-frame list, and #479 adds an arrow-shaped one. Machine-verifiable authorship, which matters because Pindar said solvers 'looked for glitches all wrong' — maybe wrong CHANNEL." }, { "id": "SEL-COLORFLASH", "name": "Color-flash classes", "status": "ratified", "members": [ 130, 133, 462, 463, 464, 497 ], "plain": "the collector's color-flash finds: #130/#133 flash the background into sine waves (they are also a pair); #462-464 and #497 flash the whole frame while the background stays binary.", "pindar": "the collector's original finds, made with the naked eye before any tooling existed. #130/#133 are also a ratified pair, tying the flash channel to the pair channel. These prove the collection rewards raw watching — the same skill the artist used to make them." }, { "id": "SEL-CHAIN-CORRUPT-11", "name": "Chain-is-corrupt-side eleven (superseded)", "status": "superseded by SEL-SHEET-CHAIN-CORRUPT", "members": [ 16, 17, 44, 48, 58, 62, 66, 81, 173, 281, 309 ], "plain": "The original eleven, ratified on an incomplete chain pull. On the complete pull (D50, 13 Sep 2026) the sheet and the chain disagree on 26 pieces, with chain-side errors of the same kinds on #4, #88, #97 and #100. Kept for history; use SEL-SHEET-CHAIN-CORRUPT.", "pindar": "SUPERSEDED. PINDAR-CONFIRMED. On 2022-01-01 he replied to a collector who noticed a Rainbow skull (#58) missing its SubType: 'There are some glitch bitGANs with intentional errors - you will find others - they are glitched.' Exactly eleven pieces carry an intentional on-chain trait corruption, and they come in five flavors — transpose (#16 Primoridal), reverse (#48 Rainbow->wobniaR), delete (#58 SubType gone), swap (#62 Type<->SubType), substitute (#66 Rainbow->Flaming, #17/#44/#173/#281). The three Rainbows glitched three different ways read as a Rosetta stone for the grammar. This is the ONLY exact-eleven an artist statement directly ratifies. The catch: essence and the trait-values themselves are not the words (dead / not BIP-39), so per the Solstice rule the true word for each hides in a channel we have not yet read." }, { "id": "SEL-SHEET-CHAIN-CORRUPT", "name": "Sheet-vs-chain corruptions (26)", "status": "PINDAR-CONFIRMED-CLASS; recomputed-2026-09-13", "members": [ 4, 16, 17, 33, 39, 44, 48, 58, 62, 66, 73, 81, 88, 97, 100, 123, 124, 128, 173, 175, 281, 290, 309, 315, 383, 509 ], "plain": "Every piece where Pindar's Arweave sheet (31 Jan 2022) and the on-chain traits give different values for a name or trait, recomputed field by field on the complete chain pull (all 512 pieces have a record; the systematic Twin→Twins rename excluded). Three witnesses decide which side is odd: usage elsewhere, the Arweave filename Pindar wrote at export, and his journal. Piles: the sheet corrects a chain error (13); the substituted word describes what the art shows (7, collector's ruling); the sheet introduces the error (3); undecided number/capital differences (4).", "pindar": "PINDAR-CONFIRMED CLASS. On 2022-01-01 he replied to a collector who noticed a Rainbow skull (#58) missing its SubType: 'There are some glitch bitGANs with intentional errors - you will find others - they are glitched.' We found them by recomputing every field of Pindar's sheet against the chain: 26 pieces disagree, in nine kinds of move. The three Rainbows glitched three ways (#48 reversed, #58 deleted, #66 substituted) read as a Rosetta stone for the grammar, and #100 repeats #16's move exactly. The catch is unchanged: the essences and the trait values are not the words, so the true word for each hides in a channel not yet read." }, { "id": "SEL-CHAIN-ERROR-A", "name": "Chain-error eleven A (skull drop)", "status": "candidate-exact-11; essence, undo-and-cut and filename registers closed (D62)", "members": [ 4, 16, 48, 58, 62, 66, 81, 88, 97, 100, 124 ], "plain": "The skull pieces whose on-chain label is malformed and corrected by the sheet, excluding #73 (the chain has the correct spelling there). Exactly eleven. A rule Pindar could state: 'the mis-written skulls'. Mint order = piece order.", "pindar": "Pindar planted his intentional errors in the skull drop first (Krum found #58 there; journal 002 flagged its typos), so 'the mis-written skulls' is a rule he could have stated in a sentence. The essence, the corrected word and the filename word were all read on this eleven in mint order and none opened the wallet (D62); the picture-describes-the-label reading is the one left." }, { "id": "SEL-CHAIN-ERROR-B", "name": "Chain-error eleven B (punctuation excluded)", "status": "candidate-exact-11; essence, undo-and-cut and filename registers closed (D62)", "members": [ 4, 16, 48, 58, 62, 81, 88, 97, 100, 124, 309 ], "plain": "Every malformed chain label the sheet corrects, excluding #73 and #66's stray backtick (punctuation, not a word). Exactly eleven, one of each move: a capital, two transpositions, a reversal, a displacement, a swap, three cut suffixes, a dropped letter, an added letter; spans the skull and ghost drops. Mint order.", "pindar": "Read strictly, 'every chain label the sheet corrects' gives thirteen; drop the two that are not word errors (#73, where the chain is spelled right, and #66's backtick) and exactly eleven remain, one per kind of move, across two drops — which suits Pindar's statement that the puzzle spans the collection. Same registers closed as A (D62)." }, { "id": "SEL-PROMISE", "name": "Pindar promise pieces", "status": "lore-flagged", "members": [ 32, 119 ], "plain": "The only two pieces the artist publicly said to 'hold on to for later': #119 (odd Type E) and #32 (pointed at by the hex pair 0x20 = 32). Both are still-unspent promises. See SEL-HIDDENTYPE-MISFILE: decoding the HiddenType field shows these two are the skull-drop pair of a five-piece misfile class.", "pindar": "Pindar publicly flagged only two pieces as puzzle-relevant and said 'hold on to this info for later': #119 (the odd Type E) and — via the hex pair pointing at 32 — #32. Promises this explicit are rare; any recipe that spends them earns credibility." }, { "id": "SEL-HIDDENTYPE-MISFILE", "name": "Hidden-type misfiles (5)", "status": "candidate-decoded-2026-08-31", "members": [ 32, 119, 161, 187, 193 ], "plain": "The on-chain HiddenType field is Pindar's hidden filing system: one letter per piece recording what KIND of number its ID is - a=Fibonacci, b=prime, c=square, d=odd-composite, e=even. The exact same code runs across the skull (by piece number), robo (by binary position, plus an F override for the eight palindromes) and spooky (by octal position) drops; it stops after spooky. In all that machinery exactly FOUR pieces are filed wrong. Two are literal corruptions - the field holds the number '32' instead of a letter: #32 (its own piece number) and #161 (binary name 100000 = 32). Two are mislabels - the letter is real but wrong for that number: #119 (odd, filed under e/even - Pindar flagged this exact piece) and #187 (even, filed under c/square - essence 'mirror', the robo echo of #119). Two Pindar confessed (#32, #119), two we recovered by cracking the code (#161, #187). A 4-piece author-signed class; a stackable contributor, not the eleven. The '32' pair may point at the NUMBER 32 (the 32x32 pixel cell size) rather than a word. Word registers on it (essence/number, incl. anchored/bookended) swept DEAD 2026-08-31; awaiting the collector's ruling on the class + un-computed registers.", "pindar": "FIVE misfiles after the collector's ruling (D52) that the palindrome class F belongs to the robo drop only: #193 (spooky position 0, letter F where the scheme gives A) joins #32, #119, #161 and #187. Pindar built a hidden filing system in the on-chain HiddenType field — one letter per piece encoding what KIND of number its ID is (a=Fibonacci, b=prime, c=square, d=odd-composite, e=even), the same code across the skull, robo and spooky drops on each drop's native numbering. Then he deliberately misfiled exactly four cards: #32 and #161 hold the literal number '32' where a letter belongs (both resolve to 32, the collection's 32x32 pixel cell), and #119 and #187 wear the wrong letter for their number. Two he confessed in the hidden-types post, two we recovered by cracking the code. The exception is the message — the classic signpost move — and it's the tightest author-signed class we have, even if word registers on it are already swept dead." }, { "id": "SEL-BURNT", "name": "Burnt axis", "status": "axis", "members": [], "plain": "Not a selection by itself - a switch. 30 pieces were burnt in forges before the clue existed (verified against the on-chain burn events, D22). Every candidate set should be tried both with and without its burnt members (ruling D5).", "pindar": "All 32 burns happened BEFORE the clue existed, and only in the first three drops. the collector spotted that timing — it means the burnt pieces were real when the puzzle was authored, so burning can't disqualify them; but a solver in 2022 saw only survivors. Both readings are live, so burnt-status is an axis, never an exclusion." }, { "id": "SEL-HODL", "name": "HODL Fibonacci eleven", "status": "partially-dead", "members": [ 0, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89 ], "plain": "The artist's own Fibonacci list. Read straight it is dead; the +128-partner trick is dead; other partner-hop rules remain open.", "pindar": "Pindar posted a Fibonacci HODL list himself — it is the artist's own eleven-item sequence. The straight read is dead, but a list the artist personally published deserves partner-hop and transform attempts before burial." }, { "id": "SEL-FIB-ELEVEN", "name": "Backwards-Fibonacci eleven", "status": "exact-11 (detector caveat applies)", "members": [ 148, 152, 158, 160, 166, 170, 172, 176, 182, 190, 287 ], "plain": "Under the collector's ground-truth definition of the visual glitch — one frame re-inserting itself with the gaps between visits shrinking 5, 3, 2, 1, 1 (the Fibonacci sequence backwards) — EXACTLY ELEVEN pieces carry the complete run: ten roboGANs and one solstice piece (#287). Forty-six pieces glitch at all; only these eleven do it perfectly. An exact-eleven that emerges from the definition itself, with no trimming. HONEST CAVEAT: the project's two glitch detectors (this frame-recurrence one and the older held-frame one) disagree badly, and the theories log warns that no candidate eleven — including this one — can be called confident until that disagreement is resolved.", "pindar": "The strongest structural candidate in the catalog. The clue says eleven glitches; the collector's corrected glitch definition — validated against his eye on every hand-checked piece — produces exactly eleven perfect specimens without any human trimming. Fibonacci is Pindar's signature number-language (his own HODL list is Fibonacci), and the first drop's glitches follow a prime-and-not-Fibonacci number rule he encoded on-chain in the unpublished HiddenType field. A definition that counts to eleven by itself demands to be swept across every word register and ordering." }, { "id": "SEL-STILL-EYES", "name": "Still-eyes class", "status": "collector-eye-2026-08-28", "members": [ 129, 130, 141, 159, 161, 174, 180, 192 ], "plain": "Eight robo pieces whose EYES stay perfectly still while everything else animates. Six of the eight naga palindromes (129, 141, 159, 174, 180, 192 — the collector ruled 147 Taco Cat and 162 Star Rats OUT), plus qubitGAN #130 (the wallet-cracker) and #161 (the piece whose HiddenType is '32'). Because two nagas lack it, this is NOT naga family behavior — it is a deliberately chosen class of exactly eight.", "pindar": "Found by the collector's eye in one browsing session on the new puzzle book — the project's best instrument doing what it does. The members carry heavy freight: the qubit wallet-cracker, the '32' carrier, and five of the eight nagas including both brackets. Either naga family grammar with two meaningful outsiders, or a fresh authored class." }, { "id": "SEL-WHITE-ZERO-EYES", "name": "White-zero eyes", "status": "collector-eye-2026-08-28", "members": [ 130, 133, 141, 174 ], "plain": "the collector's sharpest mechanism observation: on these four pieces the eyes are all white inside and are literally the ZERO characters of the binary layer sitting in front of the creature — the eyes ARE digits. Both halves of the ratified pair {130,133} are here, making that pair agree in three separate channels. If eyes occupy digit positions of a binary string, those positions can be READ.", "pindar": "Eyes that ARE binary digits reframe what an 'eye' is on these pieces — and Pindar drew the robo drop with binary as its native language. If the artist chose WHICH zeros become eyes, he chose digit positions, and digit positions can encode. The triple agreement on pair {130,133} says these choices were deliberate." }, { "id": "SEL-MONTAGE-GLITCH-FRAME", "name": "Montage glitch-frame", "status": "open-multi-step", "members": [ 19 ], "plain": "A two-step idea from the collector's montage review: Pindar's official montage image shows one tile per piece — and for #19, that tile is an actual GLITCH FRAME of the animation (frames 20 and 40 are identical spliced frames, and the montage shows exactly that frame). Hypothesis: the montage silently points at each glitched piece by displaying its glitch frame. the collector also graded ~19 pieces whose montage tile does NOT match today's art, with the caveat that an OpenSea frame-shift could explain some. The simple montage-selection sweep died, but this glitch-frame version was never completed.", "pindar": "The montage is the one image Pindar personally composed containing every piece. Hiding each glitch in plain sight — by showing the glitch frame as the piece's official portrait — is exactly the kind of joke he tells ('they have been looking for glitches all wrong': everyone looked at the GIFs, not at his group photo). the collector's #19 finding is the proof of concept; his theory: 'the montage is low res so Pindar could hide the fact that he spliced in GANs sitting at the wrong frame.'" }, { "id": "SEL-POD-INVASION", "name": "PodGAN invasion", "status": "collector-curated", "members": [ 302, 310, 314, 340, 349, 355, 368, 370, 375, 377, 379, 383, 386, 398, 401, 403, 408, 412, 421, 431, 439 ], "plain": "the collector identified ghost-drop pieces being 'invaded' by the NEXT collection: monster/spaceGANs wearing the Space-Invader icon like a T-shirt, and pieces with the podGAN symbol in the background. The story writes itself: podGANs invading the bitGANs. Feeds the podGAN transplant idea — map these pieces to their podGAN counterparts and read words THERE (the answer may not even live in this collection, like the BIP/SLIP puzzle).", "pindar": "The podGANs are the NEXT collection, and the lore is that bitGANs escape outward. Pieces wearing invader icons read as the next collection reaching back — a bridge Pindar could use to move the answer OUT of this collection entirely, like the BIP/SLIP puzzle whose answer lived in the other word list. the collector curated both lists by eye." }, { "id": "SEL-GLITCH-NAME-LIARS", "name": "Glitch-name liars", "status": "collector-eye-2026-08-19", "members": [ 9, 15, 25, 35, 45, 46, 51, 56, 58, 60, 72, 75, 84, 86, 90, 102, 115 ], "plain": "Seventeen skull pieces named 'Glitch' (or Shimmer) whose animations show NO visual glitch — the name promises what the art withholds. The mirror set exists too: four Shimmer-named pieces (7, 43, 67, 71) that DO glitch, and #166 which glitches in a unique two-rhythm way without any Glitch name. Names lying about glitches, in both directions, is itself a candidate glitch class.", "pindar": "Pindar names things carefully and lies in both directions here: Glitch-named pieces that don't glitch, Shimmer-named pieces that do. A name-vs-art contradiction is visible to anyone on OpenSea — matching the collector's read that the answer should be visible without file forensics ('looking all wrong' = staring at metadata)." }, { "id": "SEL-MISLABEL-11", "name": "Mislabel eleven", "status": "DEAD", "members": [ 179, 423, 231, 511, 509, 204, 150, 490, 478, 92, 491 ], "plain": "DEAD as a selection (ruling D4): the collector graded the labels and 9 of 11 actually match the art. Only #423 and #357 are real mislabels. Note: #179 later turned out to carry a painted mark (D19) - unrelated to its label.", "pindar": "DEAD by the collector's blind grading: nine of eleven 'lying' labels actually match the art. Kept visible as a warning — the instrument that built it was confidently wrong, and the two real mislabels it surfaced (#357, #423) are near-miss homages, not label lies." }, { "id": "SEL-GLYPH-11", "name": "Letters-in-art selection", "status": "DEAD", "members": [], "plain": "DEAD as a selection (ruling D9): 15 of 17 glyph candidates were family texture. The painted-mark class (4 pieces) is what survived.", "pindar": "DEAD as a selection: fifteen of seventeen glyph candidates were family texture (robo alphabets, TEST wallpaper). What survived is the small painted class. The lesson written into the rulings: marks that survive the creature's dissolution are wallpaper; real marks live and die with the creature." } ], "registers": [ { "id": "S2-W1", "name": "essence (straight)", "kind": "atomic-extraction", "def": "Piece's Essence trait.", "examples_from_history": "~81k swept sets.", "proposed_by": "both", "in_solution_space": true, "tested": "yes — dead on every ratified eleven, and on the sheet-vs-chain error selections A and B in mint/gan/rev (D57 witness11, 151k orderings; D62). Decoy by Solstice precedent.", "pindar": "Heavily tested and dead on every principled eleven — which matches the Solstice clue calling essences 'just a distraction'. 511 distinct essences by design is a decoy signature. Kept because a MIXED recipe may still let one or two pieces speak their essence.", "plain": "Each chosen piece contributes its own Essence trait as its word. Example: #357's essence is 'seed', so #357 says 'seed'." }, { "id": "S2-W2", "name": "title/name BIP words", "kind": "atomic-extraction", "def": "BIP words in the exact-case chain name/Title (GANg suffix stripped). the collector param: if a GAN has more than one derivable title word, try them all.", "examples_from_history": "the collector: 'let's explore that title-derived third option on its own merits and as an additional step'; 'Let's also try the fibonacci solution using words derived from the titles. If GANs have more than one derivable word, let's try them all' (bf7ee7f4). name-eleven / unique-title-31 built in word_sources.", "proposed_by": "both", "in_solution_space": true, "tested": "partial (sweep provenance unverified per S2)", "pindar": "The names are the most public channel Pindar has — and his name-games (palindromes, leet, 'Primoridal') prove he edits names by hand. Sweep provenance is unverified, so this register is NOT honestly dead; audit before trusting old negatives.", "plain": "Look at the piece's NAME as shown on OpenSea. If a real seed-list word appears in it, that word is the piece's contribution. Example: 'Shy skullGAN' contains 'shy'." }, { "id": "S2-W3", "name": "SubType word", "kind": "atomic-extraction", "def": "SubType if BIP (8 distinct values ceiling).", "proposed_by": "assistant", "in_solution_space": true, "tested": "no (structurally weak; silver-11 repeat unswept)", "pindar": "SubTypes cap at 8 distinct values, so they cannot carry eleven distinct words alone — structurally weak, useful only inside mixes.", "plain": "The piece's SubType trait is the word, when it happens to be a seed-list word (only 8 SubTypes exist, so this can't carry all eleven alone)." }, { "id": "S2-W4", "name": "Type word", "kind": "atomic-extraction", "def": "Type if BIP (8 distinct ceiling).", "proposed_by": "assistant", "in_solution_space": true, "tested": "no", "pindar": "Same ceiling as SubType: 8 distinct Types. A component register at best.", "plain": "The piece's Type trait is the word (same 8-value ceiling as SubType)." }, { "id": "S2-W5", "name": "true-class word", "kind": "atomic-extraction", "def": "What the art actually IS (mislabel inversion). Blocked on the collector's namings; D4b supplies #357 candidates (yellow/blue/half/almost/spike) and #423 (recycle).", "proposed_by": "both", "in_solution_space": true, "tested": "partial (#357->yellow variants swept dead). NOW the leading open register: the collector's ruling (D59) that a substituted label describes what the art shows makes 'what the corrected label names visually' the derivation left for the chain-error elevens — adjudicated by eye, not yet applied to A/B.", "pindar": "Pindar published the mislabel idea himself ('not Hidden Typos… part of the puzzle'). The register needs the collector's true-class namings to run — the two confirmed mislabels (#357 almost-invader, #423 Eclectic-Method homage) already yield candidate words like yellow and recycle.", "plain": "Ignore the label; ask what the art actually IS. A 'Shy' skull that is really a yellow color-glitch says 'yellow'. Needs a human naming for each piece." }, { "id": "S2-W6", "name": "partner's word (pair relations)", "kind": "piece-map", "def": "First-speaks / last-speaks / other-member of a deliberate pair.", "proposed_by": "both", "in_solution_space": true, "tested": "partial (pairs+anchor 4 variants dead; pointer-11 sub-rules unswept)", "pindar": "If pairs are the eleven, someone in each pair must speak. First-speaks and last-speaks died on the essence register, but the pair-RELATION reads (what the two share, what one lacks) are untested — and the pairs themselves are ratified authorship.", "plain": "The piece doesn't speak — its PARTNER does. For each ratified pair, take the other member's word (or the first member's, or the last's — three variants)." }, { "id": "S2-W7", "name": "number -> wordlist index", "kind": "atomic-extraction", "def": "wl[GAN], wl[GAN±1], wl[mint], wl[mint±1]. the collector-directed application: reword the sentence constructs by WL[gan]/WL[gan-1] (SPX/qidx).", "examples_from_history": "the collector: 'What if we ran all of these sentence constructs... but we used the GAN numbers to ID BIP words?' (6c9a8360).", "proposed_by": "both", "in_solution_space": true, "tested": "partial (dead on historical selections; NOT on union-11/onef11)", "pindar": "The wordlist is 2048 numbered slots and Pindar numbers everything — pieces, mints, binary positions. Solved puzzles used mint order; a number-to-word lookup is the simplest mechanism that survives 'no judge'. Partially tested; NOT yet run against the live selections.", "plain": "Treat a number attached to the piece (its piece number, or its mint position) as a position in the 2048-word seed list, and read off that word. Example: piece #81 → word 81 → 'any'." }, { "id": "S2-W8", "name": "frame-index -> wordlist / -> piece", "kind": "atomic-extraction", "def": "Held/negative frame index as wl index or GAN pointer.", "proposed_by": "assistant", "in_solution_space": true, "tested": "partial", "pindar": "Frame indices are invisible on OpenSea but byte-real — the timing-hold frames (#479 f27, #480 f24) look like deliberate bookmarks. Fits 'looking for glitches all wrong' if solvers only ever looked at pictures, not at WHICH frame.", "plain": "Some pieces have one special FRAME (a held frame, a negative frame). Use that frame's number as a seed-list position, or as a pointer to another piece." }, { "id": "S2-W9", "name": "painted-number index", "kind": "atomic-extraction", "def": "Hidden painted numbers (#165 '8941', #207 '#4') as wl/piece indices.", "proposed_by": "both", "in_solution_space": true, "tested": "no (mostly untested; hunt more members first)", "pindar": "THE NEW FRONTIER: the painted class closed at four carriers with values 81, 10.00, 4, AI. byteGAN precedent says Pindar paints payloads. Reading painted numbers as wordlist positions or piece pointers is untested and now fully specified.", "plain": "Four pieces have numbers hand-painted into the art (81, 10.00, 4, AI). Read each painted number as a seed-list position or as a pointer to that piece number." }, { "id": "S2-W10", "name": "radix decode", "kind": "atomic-extraction", "def": "Binary/octal/hex titles -> numbers -> word or pointer.", "proposed_by": "assistant", "in_solution_space": true, "tested": "no (read as signpost only)", "pindar": "#465/#466 wear '0x20/0x21' where octal was demanded — a wrong-base number that revalues 16/17 into 32/33 and points at promise-piece #32. Pindar showing us a number can LIE about its base is plausibly a taught mechanism, not a joke.", "plain": "Some names write numbers in the wrong base (hex '0x20' where octal belonged). Convert the number to its 'true' value and use THAT as the position or pointer. 0x20 = 32 → piece #32." }, { "id": "S2-W11", "name": "drop-channel word", "kind": "atomic-extraction", "def": "Each drop's glitch MEDIUM names the word.", "proposed_by": "assistant", "in_solution_space": true, "tested": "no (register not fully specified)", "pindar": "Each drop glitches in its own medium (skull=splice, robo=names, spooky=costume…). If the medium itself names the word, the puzzle spans 'all 512' the way Pindar said. Not yet even fully specified — design work, high ceiling.", "plain": "Each drop glitches in its own medium (skulls splice frames, robos corrupt names…). The MEDIUM itself supplies the word — e.g. every skull glitch says its drop's word." }, { "id": "S2-W12", "name": "description words", "kind": "atomic-extraction", "def": "Contract description field per piece / per drop. the collector supplied the six per-drop description texts verbatim in chat.", "proposed_by": "both", "in_solution_space": true, "tested": "partial (ghost-description eleven closed at stateable orderings; no per-piece 11 swept)", "pindar": "Contract descriptions are hand-typed per piece and censused, never swept as a register. Low-probability but cheap to close.", "plain": "Each drop carries a written description on-chain. Harvest the seed-list words from those descriptions and assign them to pieces." }, { "id": "S2-W13", "name": "artist-tribute words", "kind": "atomic-extraction", "def": "Person-tribute pieces -> tribute-derived word. Blind-rater aptness (remember-rose, draw-cloud, elder-legend) is the calibration instrument.", "proposed_by": "both", "in_solution_space": true, "tested": "no (census not launched; rater pass done)", "pindar": "The clue page's four GIFs are the cryptoArtGAN tributes including the prize — Pindar put tribute pieces ON the clue page. A tribute-derived word register was never censused. That placement is hard to dismiss.", "plain": "Some pieces are tributes to real people (artists, collectors). The tribute target supplies the word (their name, their signature motif)." }, { "id": "S2-W14", "name": "transplant (index hop between lists)", "kind": "piece-map", "def": "Positions from list A read out in list B (podGALLERY grammar).", "proposed_by": "both", "in_solution_space": true, "tested": "partial (union->upload dead; montage/title/drop-internal transplants unswept)", "pindar": "The podGALLERY puzzle literally worked by reading positions from one list out of another. Precedent-proven mechanism; union→upload died, but montage, title-list and drop-internal transplants are unswept.", "plain": "Find the piece's POSITION in list A (say, mint order), then read out whatever sits at that same position in list B (say, upload order). The position hops between lists." }, { "id": "S2-W15", "name": "SLIP->BIP list swap (same index)", "kind": "word-transform", "def": "Same index, other wordlist, both directions.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes — dead as stated (zero adherents; survivors shown to be chance after §8 correction)", "pindar": "Tested dead as stated: zero SLIP-word adherents in the collection.", "plain": "Same position, different word list: look the piece's index up in the SLIP-39 list instead of BIP-39." }, { "id": "S2-W16", "name": "filename descriptor words", "kind": "atomic-extraction", "def": "Arweave hand-written vocabulary (ancient/elder/heart/warrior/whale...). 72 pieces carry a filename word appearing nowhere in their metadata.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes on the chain-error elevens A and B (D62 hyp61: filename word as the unpublished column, mint/rev, no hit); all 512 filenames now sourced from Arweave (D55). Elsewhere partial.", "pindar": "Arweave filenames carry hand-written vocabulary (ancient/elder/whale…) invisible on OpenSea — a private channel only a file-digger sees. Coverage is partial (339/512), which itself might be the filter.", "plain": "Pindar hand-typed descriptive words into some Arweave filenames (ancient, elder, whale…). Those private words are the contributions." }, { "id": "S2-W17", "name": "date/season decode", "kind": "atomic-extraction", "def": "Solstice-style date->season/month; re-save session dates.", "proposed_by": "assistant", "in_solution_space": true, "tested": "no (session-date reads unswept)", "pindar": "The Solstice puzzle ran on dates and seasons. Act 1 has date traits, burn dates, and eleven re-save sessions. Pindar reuses machinery between puzzles — 'calendar' is his proven register family, unswept here.", "plain": "Turn a date attached to the piece (drop date, save date, burn date) into a word — a month, a season — like the Solstice puzzle did." }, { "id": "S2-W18", "name": "letter-payload / acrostic", "kind": "atomic-extraction", "def": "Case-deviant letter+position payload; acrostics of initials/finals.", "examples_from_history": "Batch-name acrostics all dead (batch_name_decode §5); community-name internal-capital acrostic flagged but unassembled.", "proposed_by": "assistant", "in_solution_space": true, "tested": "partial (censuses done, reading never assembled)", "pindar": "The case-anomaly census found deliberate letter-level edits ('Gan', 'Primoridal'). A letter-payload assembled across pieces was verified as a mechanism (B23) but the READING was never assembled and swept.", "plain": "Collect one LETTER from each piece (a miscased letter, a swapped letter) and assemble the letters into words across the eleven." }, { "id": "S2-W19", "name": "Eyes/Dimensions decode", "kind": "atomic-extraction", "def": "Integer trait -> wl index.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes — dead by analysis (degenerate)", "pindar": "Analytically dead: Eyes/Dimensions values 0-11 collapse into the first dozen wordlist entries — degenerate.", "plain": "Use the small integer traits (Eyes 1-9, Dimensions 2-11) as seed-list positions. (Analysis shows this collapses to the list's first dozen words — a dead end.)" }, { "id": "S2-W20", "name": "event-word", "kind": "atomic-extraction", "def": "One word per glitch EVENT (which member speaks per pair).", "proposed_by": "both", "in_solution_space": true, "tested": "partial (via W6 sweeps)", "pindar": "If the clue counts EVENTS, each event needs one word. Only partially exercised through the pairs sweeps; the general event-word framing is broader than pairs.", "plain": "Count glitch EVENTS, not pieces: a pair is one event and speaks one word. Choose which member of each event does the talking." }, { "id": "S2-W21", "name": "absence-word", "kind": "atomic-extraction", "def": "Each authored absence contributes via its drop channel.", "proposed_by": "assistant", "in_solution_space": true, "tested": "no", "pindar": "Pindar authored absences: the missing robo GIF, the never-minted forge partner, 404 'Not Found'. 'Looking for glitches all wrong' fits perfectly if the glitches are the HOLES. Census incomplete; nothing swept.", "plain": "The glitch is a MISSING thing (a file that should exist, a slot never minted). The absence itself supplies a word through its drop's channel." }, { "id": "S2-W22", "name": "image-association word", "kind": "atomic-extraction", "def": "BIP word for what is VISIBLE beyond the label (dominant anomalous color, objects/symbols, background content).", "proposed_by": "both", "in_solution_space": true, "tested": "partial (#357->yellow swept dead on strong elevens)", "pindar": "Reads what the eye sees rather than what metadata says (#357→yellow is over-determined). the collector's eye is the project's best instrument, and this register is that eye formalized.", "plain": "Say what your EYE sees that the label doesn't: a wrong color (#357 → yellow), an object in the frame (recycling symbol → recycle). The seen thing is the word." }, { "id": "S2-W23", "name": "homage-target word", "kind": "atomic-extraction", "def": "Word of the ENTITY a piece homages or half-becomes (recycle for Eclectic Method, invader, Sol/Luna).", "proposed_by": "both", "in_solution_space": true, "tested": "no", "pindar": "The two confirmed mislabels are both incomplete homages — pieces trying to BECOME something else. The homage TARGET (invader, Sol/Luna, Eclectic Method) supplies the word. Matches Pindar lore about GANs 'trying to come into focus'.", "plain": "The piece almost-becomes another entity (an invader, Sol/Luna, another artist's motif). The word belongs to the thing it's imitating." }, { "id": "EC24", "name": "typo-normalization (repair the glitch, keep the word)", "kind": "word-transform", "def": "Normalize authored corruptions back to legal BIP words before seeding: err0r->error (unleet), wobniaR->Rainbow (reversal repair), TRUE/FALSE->true/false (boolean coercion repair). the collector's standing rule, used silently in many sweeps but never catalogued as a pipeline stage.", "examples_from_history": "the collector verbatim: 'Remember that for things like err0r and wobniaR, we can normalize them to error and Rainbow if we need to for the seed words' (bf7ee7f4). HET1 spec: 'essence-worded (err0r/TRUE/FALSE normalised)'. A registry keying bug from unnormalized TRUE/FALSE was later caught and fixed.", "proposed_by": "collector", "in_solution_space": false, "tested": "yes — undo-and-cut (restore the corrupted value, cut to its on-list stem, repeats allowed) run on elevens A and B in mint/rev, no hit (D62). Earlier: as a stage inside HET1/pointer sweeps.", "pindar": "Pindar's typos are surgical (err0r, Gltich, Primoridal). Repairing the typo and keeping the word treats each corruption as an arrow pointing at a clean BIP-39 word — the gentlest possible read of 'glitches ARE the words'.", "plain": "Repair the typo and keep the word: err0r → error, wobniaR → rainbow. The corruption points at a clean word." }, { "id": "EC25", "name": "spelling-transform family (reversal, atbash, Caesar, anagram, prefix4-hop)", "kind": "word-transform", "def": "Letter-level word->word maps. CLOSED PERMANENTLY by arithmetic: only 15 BIP words reverse into BIP words, 6 survive atbash, most Caesar shifts map nothing; 13,552 empirical opportunities fired zero times.", "examples_from_history": "transforms_and_mechanisms §1a/§9. Community solves use transformations but Pindar's three solved puzzles never required one.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes — closed for arithmetic reasons; do not re-test", "pindar": "Reversal is IN the collection (wobniaR, Dog God, Flow Wolf). A puzzle that demonstrates reversal on its own name plate licenses reversal as a word transform.", "plain": "Transform the word by spelling games the collection itself plays: reverse it, shift it, anagram it (Dog God, Flow Wolf prove reversal is in-bounds)." }, { "id": "EC26", "name": "wordlist-index arithmetic", "kind": "word-transform", "def": "word -> index -> ±1/±2/±3/±7/±11/±12/±128/±512/±1024, doubling, halving, complements, decimal-digit reversal, 11-bit reversal, mod 1024 -> word. All 32 are TOTAL functions — producing eleven valid words is a tautology, never evidence.", "examples_from_history": "transforms_and_mechanisms §1a calibration table; the survivors' 'success' explicitly downgraded to no-information.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (on 57 + 1,232 candidate elevens; no match)", "pindar": "Index arithmetic (±1, offsets) is the standard escape when a straight lookup misses by one — and the collection itself is ambiguous about 0- vs 1-indexing (binary positions start at 000000).", "plain": "Nudge the number: if position N misses, try N+1 or N-1 — the collection is genuinely ambiguous about counting from 0 or 1." }, { "id": "EC27", "name": "letter-arithmetic word->index (len/initial/lettersum/letterprod)", "kind": "word-transform", "def": "Turn a word into a number from its letters and read that as an index. Total but collapsing (few distinct outputs) — recorded as unusable. Also the batch-name letter-sum decode (a=1..26 sum -> GAN essence or BIP index) which produced three stateable elevens.", "examples_from_history": "transforms §1d; batch_name_decode §4 (letter-sum pointer: bs=21 -> #21 'bean', 1-based variant lands bs on 'action' — 64 sweeps negative).", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (batch-name variants swept; general family tested)", "pindar": "Letter-sums and initials are classic puzzle machinery, but nothing in Pindar's record uses them — no precedent, low prior, kept for completeness.", "plain": "Turn a word into a number by its letters (length, first letter's rank, letter sum) and use that number as a position." }, { "id": "EC28", "name": "glitch-as-instruction (the corruption names the operation)", "kind": "word-transform", "def": "A corrupted field does not supply a word but names the operation to apply: wobniaR=reverse, Gltich=transpose, Cronenber=truncate, err0r=unleet, HiddenType '32'=read-as-number. Spelling-level application fails outright (5-6 of 11 produce nothing); index-level application produces testable elevens.", "examples_from_history": "transforms_and_mechanisms §3c; resonates with the collector's 'Never assume that an error is truly an error for Pindar... That is intentional.'", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes — index level swept (D40–D42); move-as-instruction on the chain-error elevens shown infeasible: reversal / adjacent swap / dropped letter applied to each piece's restored word lands on the wordlist for 0 of 11, on the essence 1 of 11 (D62). The wordlist rewards reversal on 5 pairs, a swap on 6, a dropped letter on 198 — none of them these words.", "pindar": "The glitch names the operation: an inverted piece says 'invert something', a doubled piece says 'double something'. Self-describing corruption fits an artist who made glitches his medium.", "plain": "The glitch NAMES the operation: an inverted piece says 'invert something', a doubled piece says 'read twice'. Corruption as instruction." }, { "id": "EC29", "name": "cross-collection podGAN transplant", "kind": "piece-map", "def": "Map selected bitGANs by number/mint-position to their podGAN counterparts and use the PODGAN essences as the seed words — 'like the BIP/SLIP puzzle, the answer isn't even here.' Includes the condensed-BIP-set renumbering variant (filter podGAN essences to BIP-valid, renumber, then match).", "examples_from_history": "the collector verbatim: 'What if we tried to match them to PodGANs?... transpose the number (mint order) of the relevant bitGANs I identified to their PodGAN counterparts so we could use the PodGAN essences as our 11 seed words'; 'the condensed list is the new order' (bf7ee7f4). Data fact found en route: podGAN seed trait is a perfect SLIP-39 index map.", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (podGAN-invaded set runs; the collector audited the mapping twice)", "pindar": "Cross-collection transplant with condensed renumbering — the exact podGALLERY move, and one of the collector's four. Filter, renumber the survivors, read the new numbers. Proven in Pindar's own prior puzzle.", "plain": "Filter the collection down to a chosen subset, RENUMBER the survivors 1,2,3…, and use the NEW numbers — as seed positions here, or as piece numbers in the podGAN collection (read those podGANs' essences as our words)." }, { "id": "EC30", "name": "forge-hop word", "kind": "piece-map", "def": "Read the word of the piece across the forge relation: the SuperRare piece this artwork was burnt into ('forged' hop), or the bitGAN it was burnt WITH ('partner' hop). The forge pairing is the collection's only non-trivial cycle structure (19 two-cycles).", "examples_from_history": "multistep_chains hop inventory (partner cov 40, forged cov 40); forge-partners11 / burnt-flaming11 sets; backward run proved the forged eleven has NO preimage in Act 1.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (forge sets swept ~60 orderings; transform images tested)", "pindar": "Sixteen forges consumed 32 pieces before the clue; the forge is a MAP from burnt pieces to SuperRare outputs. Burnt-Flaming count is exactly 11 — an unread signal sitting on an exact-eleven.", "plain": "Burnt pieces map to the SuperRare pieces they forged. Hop through the forge: the burnt piece's word comes from its forged output." }, { "id": "EC31", "name": "number-word chase (self-referential cardinal hop)", "kind": "piece-map", "def": "Essences that are English number words (eight, three, twelve, zero, one, twenty, hundred, two, six, ten...) point at the artwork of that number; read the target's essence. Twelve number-word pieces = a complete mnemonic, tested directly (checksum-legal by chance, negative). 'seven' is the one BIP number word no artwork holds.", "examples_from_history": "transforms_and_mechanisms §3i.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes", "pindar": "Cardinal-number chase (a piece whose word is 'three' points three ahead). No Pindar precedent; speculative.", "plain": "If a piece's word is a number-word ('three'), hop that many pieces forward and take THAT piece's word." }, { "id": "EC32", "name": "collection-as-bit-register read", "kind": "process", "def": "One bit per artwork from a stated predicate (essence-index parity, GAN primality, glitch flag, Rainbow SubType, digit-in-title...), taken in a stated order, cut into eleven 11-bit groups, each read as a BIP index. Words produced by a process, not read off objects; 121 bits exactly matches the entropy of eleven words.", "examples_from_history": "transforms_and_mechanisms §3b (11 predicates x 2 orders x 2 slices = 44 elevens tested).", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (44 sets, negative; the 121-of-513-bits choice never forced)", "pindar": "Reading the whole 512 as a bit register is elegant but computational — Pindar said execution was hard, not derivation. Disfavored by the ≤2-sentence guardrail.", "plain": "Read the whole 512 collection as one long binary number (each piece = one bit by some rule) and decode it. (Heavy computation — out of character for Pindar.)" }, { "id": "EC33", "name": "site-page word harvest (clue page / lore pages)", "kind": "atomic-extraction", "def": "The roboGAN-precedent move: keep only the BIP-39 words contained in a published TEXT (the clue page, a journal page, a drop write-up). The clue page holds 22/35 distinct BIP words depending on scope — split, interleaved, and mapped-to-carrier-GAN variants were run.", "examples_from_history": "the collector: 'test the ghosts. i want to see an artifact mapping the 22 words to Gans with mint order and drop'; the 22-word split/every-other readings; clue-page red/BIP scan closed (35 words, not 11) in extraction_theories §closed.", "proposed_by": "both", "in_solution_space": false, "tested": "yes (page-word families swept; the roboGAN-move-on-his-own-sheet variant produced the 13 sheet hapaxes, 156 reductions tested)", "pindar": "The clue page and lore pages are Pindar's own words; door-closer work proved no per-word marking exists on the clue page. Mostly closed, kept for the record.", "plain": "Harvest words directly from the puzzle's own web page or the lore pages, in page order." }, { "id": "EC34", "name": "sheet-hapax words (roboGAN move on bitgan_properties.csv)", "kind": "atomic-extraction", "def": "Apply the roboGAN filter to Pindar's own published sheet with the Essence column dropped (Solstice exclusion): BIP words occurring exactly once outside Essence. Yields 13 (heart ancient warrior mechanic pumpkin angry solar hungry bird lunar spin fire vibrant); essence-word exclusion cuts to 6.", "examples_from_history": "transforms_and_mechanisms §2d; all C(13,2)=78 drop-two reductions tested x2 orders.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (156 orderings, negative)", "pindar": "Sheet-hapax words replicate the roboGAN move (unique words on the spreadsheet). The spreadsheet is an artist-authored artifact, so words appearing exactly once there are flagged by construction.", "plain": "Words that appear exactly ONCE in the official spreadsheet are self-highlighting — collect the hapax words as the contributions." }, { "id": "EC35", "name": "batch-label decode", "kind": "atomic-extraction", "def": "Treat the eleven Arweave batch NAMES (aa, brinkman, bs, colborn, dgans, highdef, line/lines, roses, skullall, stella, test) as ciphertext: lexicon membership, anagrams, splits, edit-distance-1, 4-letter-prefix completion, subsequence completion, leet/hex (aa=0xAA=170), base-26, letter-sum pointers, mint-index pointers, acrostics. No rule maps all eleven to words.", "examples_from_history": "batch_name_decode.md — verdict: 'the batch names are labels, not cipher-text'; the three forceable elevens swept negative (64 sweeps).", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (closed)", "pindar": "Batch labels (reprise07 missing, etc.) are producer-side vocabulary; a decode was proposed but never built. Thin.", "plain": "Decode the production batch labels (file batches like 'reprise07') into words or positions." }, { "id": "EC36", "name": "sentence-fitting word choice (aptness-constrained speaker selection)", "kind": "atomic-extraction", "def": "For multi-member batches, CHOOSE the member whose word makes the eleven read as English (instruction / read-out / ode), optionally constrained by blind-rater tribute aptness (remember-rose, draw-cloud, elder-legend). A selection-of-word rule, not a derivation — 7,168-combination space.", "examples_from_history": "the collector's canonical sentence and his 16 hand-listed variants; 'we may have to try different words in each clause since we are not always sure which words are correct'; 'other sentences... even more apt as instructions, a final read-out, an ode'. Null-model verdict: manufactured (P=0.572), but the full VA/VB spaces were closed anyway.", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (12,544-ordering closure + hand variants)", "pindar": "Aptness-constrained choice: when a piece offers several words, pick the one that makes the SENTENCE work. Dangerous (post-hoc), so it is a tiebreaker, never a driver.", "plain": "When a piece offers several possible words, pick the one that makes the eleven-word SENTENCE read most naturally. A tiebreaker, never a driver." }, { "id": "EC37", "name": "in-frame inscription read (byteGAN noohp precedent)", "kind": "atomic-extraction", "def": "Words painted faintly INTO animation frames; read by frame-stepping + enhancement (std-dev projection or neighbor-diff — never median, which erases faint marks). the collector's fleeting-letters observation keeps the channel alive; the 512-file sweep found only the TEST watermark.", "examples_from_history": "extraction_theories_A rider (A3, demoted to eyes-if-idle); word_sources §8 names it 'the single largest unexplored place a word could be hiding'; the collector challenged the 'potato/elite/book/moon' precedent claim and it did not verify.", "proposed_by": "both", "in_solution_space": false, "tested": "partial (plain-view sweep done; enhanced 512-frame read not exhaustive)", "pindar": "byteGAN hid 'noohp' painted in-frame — the strongest precedent in the whole record, and the reason the painted-mark census was worth three days of eyes. Now feeds S2-W9 with real values.", "plain": "Look INSIDE the art for written words — the byteGAN puzzle hid 'noohp' painted in-frame. This is the register the painted-mark census now feeds." }, { "id": "EC38", "name": "container-metadata read (XMP / layer names)", "kind": "atomic-extraction", "def": "byteGAN bitquence precedent: words in Photoshop layer names / GIF extension blocks. CLOSED: all 404 Arweave files carry only NETSCAPE2.0 + boilerplate XMP (DocumentID/InstanceID/DerivedFrom/CreatorTool, no layer names). XMP survives as a SELECTOR/ORDERER (final-session eleven), not a word source. Parser warning: naive sub-block parsing corrupts CreatorTool strings.", "examples_from_history": "word_sources §8; extraction_theories B §4 ('XMP as word CARRIER' killed).", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (closed as word source); XMP identifiers now decoded for all 512 files — save time, build, session (D51). Not a word register.", "pindar": "XMP re-save sessions number EXACTLY eleven — Pindar touched 77 files in eleven sittings. If the sessions are the glitches, the register reads something per session (its date, its first piece, its count). The count landing on eleven is either the answer or the best red herring in the data.", "plain": "Read the invisible file metadata (XMP saves, layer names). The 77 re-saved files cluster into exactly ELEVEN save sessions — read something per session." }, { "id": "EC39", "name": "two-channel agreement word (essence == title-derived)", "kind": "atomic-extraction", "def": "Keep only pieces where the essence and the title-derived word AGREE — the word confirmed by two independent channels is the seed word. Also the degenerate cousin: one word repeated eleven times ('skull' x11) plus the random 12th.", "examples_from_history": "the collector verbatim: 'maybe it GANs where the essence and title derived word are the same'; 'If skull is the word and it comes up in 11 candidates in a row, it could just be skull 11 times plus a totally random word' (bf7ee7f4). agree10 run ('essence == title-word') appears in the priority queue.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (agree10 ran; repeat-word readings largely unswept — silver-11 still open per S2-W3)", "pindar": "Two-channel agreement (essence == a name word) is a filter more than a register: pieces where independent channels agree are self-highlighting.", "plain": "Keep only pieces where two independent channels AGREE on the same word (essence matches a name word). Agreement is the selection." }, { "id": "EC40", "name": "TRUE/FALSE as marker-classifier", "kind": "process", "def": "TRUE (#363) and FALSE (#473) — mirrored in podGANs #941/#329 — are not words but LABELS: e.g. TRUE = puzzle clue, FALSE = glitch (or each piece's relationship to its own drop, per D2). Decode why one is true and the other false, then use that predicate to identify the other glitch pieces.", "examples_from_history": "the collector verbatim: 'something like TRUE/FALSE in the essence is not a mistake. That is intentional. Maybe TRUE = puzzle clue. or FALSE = Glitch. And if we can find why one is true and the other false, we can ID others like them' + 'I expected a more robust and creative exploration of the True/False anomaly' (bf7ee7f4); D2 note.", "proposed_by": "collector", "in_solution_space": false, "tested": "no (correlation pass ran; the classifier itself never operationalized)", "pindar": "the collector's classifier idea: TRUE (#363) and FALSE (#473) as a labeling machine over other pieces rather than words in the phrase. the collector's own note on the pair: TRUE/FALSE may be a statement about each piece's relation to its own drop. Never operationalized until now.", "plain": "Use #363 (TRUE) and #473 (FALSE) as a labeling machine: test every candidate piece against some yes/no question (does it match its drop's theme?) and keep only the TRUEs — or only the FALSEs." }, { "id": "EC41", "name": "prize-GAN encoding search", "kind": "atomic-extraction", "def": "Search bitGAN #445 (the prize itself) for ANY encoding of the words — visual, metadata, steganographic. 'You should test all kinds of encodings in the prize GAN.'", "examples_from_history": "the collector, bf7ee7f4 msg 7.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (file/metadata level covered by general censuses; no dedicated #445 stego pass on record)", "pindar": "The prize GAN itself may encode its unlock. Poetic, untested, cheap to inspect.", "plain": "Search the prize piece itself for an encoding of its own unlock." }, { "id": "EC42", "name": "montage-splice frame register", "kind": "piece-map", "def": "the collector's original never-executed theory, later executed: Pindar's published all-512 montage shows some tiles at the WRONG frame ('the montage is low res so Pindar could hide the fact that he spliced in GANs sitting at the wrong frame'); the mismatching tiles select the pieces and the displayed-frame index (or its content) supplies the word (via essence or number-index readings).", "examples_from_history": "6c9a8360 grading-artifact thread; the collector hand-graded all tiles, flagged 2/85/86/89/133/167/176/193/200/201/205/320/338/344/383/390/410/426/427/433/442; word sweeps under essence and number-index readings run per session summary.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (graded; first sweeps run; frame-content reads open)", "pindar": "Montage-splice frames identify donor pieces — the montage is an artist-built index of the collection, so splices FROM it are pointers.", "plain": "The montage's spliced tiles identify donor pieces — use which-frame-the-montage-shows as the extraction handle." }, { "id": "EC43", "name": "terminus back-derivation (the random-essence GAN as the 12th)", "kind": "process", "def": "Fix #468 (essence 'random') as the twelfth GAN — the clue read literally — and reverse-engineer the eleven predecessors by a step rule that reaches it. Only one rule reaches #468 in eleven steps (mint-position hop), yielding two chains; chain A is checksum-legal.", "examples_from_history": "the collector verbatim: 'Let's say that the GAN whose essence is random IS the 12th GAN. Can we reverse engineer the other 11 from that?' (bf7ee7f4). transforms follow-up §11a-b: extended to six termini, 224 elevens, all negative.", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (346 walks incl. terminus alignments)", "pindar": "Back-derive the 12th: if some piece's essence is literally 'random', the clue's 'ending with a random word' might be a pun naming that piece. Checks the pun before the sweep does it the hard way.", "plain": "Work backwards from the end: find the piece whose essence is literally 'random', declare it the 12th, and reverse-engineer which eleven lead to it." }, { "id": "EC44", "name": "double-mint double-speak (#390 twice / one-of-each-duplication-kind)", "kind": "atomic-extraction", "def": "The twice-minted #390 contributes TWO words (its two slots' readings in their respective positions); or the eleven is completed 'one of each' duplication kind — #390 plus one of the byte-identical pair #14/#75.", "examples_from_history": "the collector verbatim: 'What about the piece that was minted twice? What if we used it twice with its different essences in their respective slots?' and 'What if we need to do one of each. 390 plus either 14 or 75 to round out the 11?' (6c9a8360). Pointer-11's double-count gambit (B §3) is the assistant cousin, kept as untested fallback.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (D2 dbl390 essence variants swept q0827d negative; slot-specific double-speak readings not exhausted)", "pindar": "#390 was minted twice and #14/#75 are byte-identical — duplication in three different media. 'Double-speak' reads duplicated things twice. Pindar demonstrably duplicates on purpose.", "plain": "Duplicated things speak twice: the twice-minted #390 contributes at both its mint slots; byte-identical #14/#75 share one voice." }, { "id": "EC45", "name": "negative-space word (the hole as the word)", "kind": "process", "def": "The word a stated rule DEMANDS but nothing supplies is itself the payload: 'action' (index 20, displaced by #21 'bean') is the only word the consumed low band never spends; 'seven' the only number word unheld. A missing word as the eleventh/twelfth contribution.", "examples_from_history": "multistep_chains §4.1 ('the sharpest unexplained hole I found'); transforms §2e; batch_name_decode letter-sum 1-based lands bs on 'action'.", "proposed_by": "assistant", "in_solution_space": false, "tested": "no (chased as a lead, never operationalized into an eleven)", "pindar": "The hole as the word: 404→'Not Found', the missing 25th forge partner. Absence-as-payload is the darkest reading of 'looking for glitches all wrong' — and unswept.", "plain": "The HOLE is the word: 404 → 'Not Found', a missing file, a never-minted slot. Read what isn't there." }, { "id": "EC46", "name": "GAN-vocabulary / lore-term thematic words", "kind": "atomic-extraction", "def": "Mine GAN-art technical language (nearest neighbor, latent space, mode collapse...) and bitGAN lore vocabulary for thematic concepts, then locate the words visually in the collection under the no-hard-computation rule.", "examples_from_history": "the collector: 'is there anything we can do to explore the history/technical construction/lingo around GANs and GAN art... to find thematic concepts we could apply within the rules we've set'; follow-up: 'The GAN/Lore exploration is woefully underdeveloped' (6c9a8360).", "proposed_by": "collector", "in_solution_space": false, "tested": "no (brainstorm lane only; the collector twice asked for a restart)", "pindar": "Lore-term thematic words (moirai, quantum, glitch...) — mood board more than mechanism; kept as vocabulary for other registers.", "plain": "Use the collection's own lore vocabulary (moirai, quantum, glitch…) as the word pool that other rules select from." }, { "id": "EC47", "name": "visual-signature-behind-the-name census", "kind": "atomic-extraction", "def": "For each Type/SubType/GANg value, identify the VISUAL signature the name denotes (Shimmers' distinctive color-blend, reverse-timing constructions), then read pieces that carry the signature without the name — or the name without the signature — as the word carriers. Extraction cousin of the type-rule-breaker selection.", "examples_from_history": "the collector: 'Have we ever looked across all the collection at type/subtype/GANg... to ID the visual signature behind the name?' + 'Shimmers don't always tremble. They also have a very distinctive look' (6c9a8360).", "proposed_by": "collector", "in_solution_space": false, "tested": "no (feeds S1-V29 selection; never run as a word register)", "pindar": "The census behind W22: what the art LOOKS like, formalized. Feeds image-association words.", "plain": "A census of what each piece LOOKS like beneath its name — the raw material for the eye-based registers." } ], "orderings": [ { "id": "S3-O1", "name": "mint order", "def": "TokenIndex ascending — canon default; both solved seed puzzles used it verbatim.", "params": { "key": "mint_index", "dir": "asc" }, "examples_from_history": "Default first ordering on every sweep; Pindar: 'use the minting order of the bitGANs'.", "proposed_by": "both", "in_solution_space": true, "tested": "yes (on every swept set)", "pindar": "Canon default. Both solved seed puzzles said it in so many words: 'use the minting order of the bitGANs'. Every recipe runs mint first; a non-mint ordering must argue for itself.", "plain": "Line the eleven up by the order they were MINTED on-chain, earliest first. This is the canon default — both solved puzzles used it." }, { "id": "S3-O2", "name": "reverse mint order", "def": "O1 descending.", "params": { "key": "mint_index", "dir": "desc" }, "examples_from_history": "the collector: 'I presume you are also doing mint order, reverse mint order, etc.' (bf7ee7f4).", "proposed_by": "both", "in_solution_space": true, "tested": "yes", "pindar": "The free variant of canon — the palindrome class proves Pindar loves reading things backwards.", "plain": "Mint order, read backwards." }, { "id": "S3-O3", "name": "GAN number order", "def": "Piece number ascending (equals mint order 1-444; Review breaks it).", "params": { "key": "gan", "dir": "asc" }, "examples_from_history": "Standard on all sweeps.", "proposed_by": "both", "in_solution_space": true, "tested": "yes", "pindar": "Piece number equals mint order for #1-444, then Review breaks the equality — meaning the two orders DISAGREE exactly where the puzzle climaxes. That divergence is either noise or a hinge.", "plain": "Line them up by piece number (#1 … #512). Identical to mint order until the Review drop, where the two disagree." }, { "id": "S3-O4", "name": "drop-then-position", "def": "Drop sequence, then position inside drop.", "params": { "group_key": "drop", "inner_key": "drop_local_number" }, "examples_from_history": "orders6 group mechanism.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes", "pindar": "Drops are Pindar's chapters; ordering by chapter-then-page respects the collection's own table of contents.", "plain": "Sort by drop first (skull, robo, spooky…), then by position inside each drop." }, { "id": "S3-O5", "name": "Dimensions/Sudoku slotting", "def": "8 distinct-Dimension pieces pin slots 2-8 and 11; 3 wilds fill 1, 9, 10. A constraint, not a sort.", "params": { "constraint": "dimensions_value = slot", "wild_slots": [ 1, 9, 10 ] }, "examples_from_history": "the collector: 'Have we considered that Dimensions are the ordering mechanism: 1-11?' and 'the gaps are intentional... like a kind of Sudoku' (6c9a8360). D3: full pairs-x-Sudoku closed space 127,920 orderings ordered run.", "proposed_by": "collector", "in_solution_space": true, "tested": "partial (SET-1..4, Eleven-A..C; pairs-x-Sudoku space per D3)", "pindar": "Dimensions 2-11 pin eight pieces to eight slots like a sudoku; three wilds fill the rest. the collector reviewed it as viable, flagging slot 2 as the weak point. The trait exists for no OTHER known reason — a trait with no job is suspicious.", "plain": "Use the Dimensions trait as a seat number: a Dim-4 piece sits in slot 4. Eight slots get pinned by eight distinct Dimensions; three wild pieces fill the rest in mint order." }, { "id": "S3-O6", "name": "alphabetical", "def": "By title, or by word (= BIP-39 index order on words).", "params": { "key": "title_or_word", "dir": "asc/desc" }, "examples_from_history": "the collector: 'I would also try multiple orders like alphabetical' (bf7ee7f4).", "proposed_by": "both", "in_solution_space": true, "tested": "yes", "pindar": "Alphabetical is how OpenSea displays and how a casual solver browses; cheap to run, no Pindar signature.", "plain": "Alphabetical by title." }, { "id": "S3-O7", "name": "Arweave upload order", "def": "act1_uploadorder.json manifest sequence (only 68 Review pieces have ranks).", "params": { "key": "upload_rank" }, "examples_from_history": "Used as mint-order proxy early; transplant q0827c.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes (where defined)", "pindar": "The upload manifest is producer-side truth — the order Pindar's own hands touched the files. Only 68 Review pieces have ranks, limiting it to Review recipes.", "plain": "The order Pindar's own upload manifest lists the files (known only for the 68 Review pieces)." }, { "id": "S3-O8", "name": "montage display order", "def": "montage_display_seq over all 512.", "params": { "key": "montage_seq" }, "examples_from_history": "Deck artefact ordering.", "proposed_by": "assistant", "in_solution_space": true, "tested": "partial", "pindar": "The montage is an artist-built arrangement of all 512 — the only full-collection layout Pindar personally composed.", "plain": "The order pieces appear in Pindar's official montage image, read like a page." }, { "id": "S3-O9", "name": "gallery hanging order", "def": "Clue-era CryptoVoxels/bitGALLERY arrangement (was literally a password once).", "params": { "key": "gallery_position", "status": "data unrecovered; gallery mutated" }, "examples_from_history": "the collector confirmed the bitGALLERY 'changes all the time and is not the same as when the puzzle launched' (6c9a8360).", "proposed_by": "assistant", "in_solution_space": true, "tested": "no (frontier, blocked on Archive Hunter)", "pindar": "The gallery hanging order was LITERALLY a password once (B20). The data is unrecovered — archive work, not sweep work — but the precedent is exact: Pindar has used physical arrangement as a key before.", "plain": "The order the pieces hung in Pindar's virtual gallery (once literally a password; the arrangement data is not yet recovered)." }, { "id": "S3-O10", "name": "grouped-then-ordered (single-key groups)", "def": "Group by a trait, order inside by another column; group order fixed by the group key itself.", "params": { "group_key": "type/drop/trait", "inner_key": "any column" }, "examples_from_history": "the collector's seed idea: 'first grouping by name (rainbows together then flamings) and had them in order inside of those sets. That kind of two-step is very Pindar' (bf7ee7f4).", "proposed_by": "collector", "in_solution_space": true, "tested": "yes (orders6 standard)", "pindar": "Group-then-order matches how the collection is built (drops containing positions). The generic machinery behind several specific orderings.", "plain": "Group by some trait first, then sort inside each group by a second rule." }, { "id": "S3-O11", "name": "ends-pinned variants", "def": "Pin #512/'final' last; #0/Moirai first; anchor-initial variants; constraint layered on any base order.", "params": { "pin_first": [ "#0 Pandimensional", "#1 Moirai", "solstice GAN" ], "pin_last": [ "#512", "#357", "solstice GAN" ] }, "examples_from_history": "the collector: 'running those with the Pandimensional Gan or Morai in the #1 position or the Quantum Glitch in the 11th position'; 'Try the solstice GAN in the number 1 and 11 spot too'; D2: 'if seed [#357], also try phrase-INITIAL position'.", "proposed_by": "collector", "in_solution_space": true, "tested": "partial (applied ad hoc)", "pindar": "Ends-pinning is Solstice-honest: 'begins with sunrise and it ends with sunset'. When Pindar pins ends, he says so; the clue's 'ending with a random word' IS an end-pin.", "plain": "Pin the ends: force a chosen piece first (#357, the Moirai) or last (#512), and order the rest normally in between." }, { "id": "S3-O12", "name": "random-word position", "def": "Canon: random 12th LAST only. Variants placed it first / at every seam between groups / every position — a sweep-structure parameter, not a sort.", "params": { "positions_tried": [ "last", "first", "seam-between-clusters", "middle", "every position" ] }, "examples_from_history": "the collector repeatedly: 'random word in every position... starting with 1st, last and between clusters'; 'Put the random at the start, seam and end' (bf7ee7f4). Later self-corrected: 'The random word is only at the very end' — foreclosed by G5.", "proposed_by": "collector", "in_solution_space": true, "tested": "yes (cluster-union runs with seam positions), then foreclosed by canon G5", "pindar": "Foreclosed by the clue itself: the random word ends the phrase. Kept only to document the closure.", "plain": "Move the random 12th word somewhere other than last. (Foreclosed — the clue says 'ending with a random word'.)" }, { "id": "S3-O13", "name": "save/session order", "def": "XMP UUID-v1 timestamps; exact save sequence known for the final session.", "params": { "key": "xmp_save_timestamp" }, "examples_from_history": "Final-session eleven swept 'first-saved and last-saved words, in session order, GAN order, and mint order' (transcript).", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes — 39 save sessions at a 3-hour gap from the decoded XMP times (D51, save_sessions.json); as an ORDERING key, tested only on the old 86-file class. Not yet run on the chain-error elevens.", "pindar": "The XMP timestamps expose the exact save sequence — a hacker-grade channel a 2022 solver would need file forensics to see. Tier-2 by the collector's policy, alive by Pindar's 'no judge'.", "plain": "The exact order Pindar SAVED the files, recovered from hidden file timestamps." }, { "id": "S3-O14", "name": "BIP-39 index order", "def": "Words sorted by wordlist index (= alphabetical on words).", "params": { "key": "bip_index" }, "examples_from_history": "Coincides with mint order on skull selections.", "proposed_by": "assistant", "in_solution_space": true, "tested": "yes", "pindar": "Word-index order coincides with mint on skull selections; degenerate elsewhere. Documented for the coincidence.", "plain": "Sort the eleven words by their position in the official seed word list." }, { "id": "OC15", "name": "group-then-order with block permutation (2-step and 3-step stackings)", "def": "Group by key A, then PERMUTE the blocks freely (not fixed by A), sort inside each block by key B; 3-step adds a subgroup key. Extends S3-O10, whose group order is fixed by the group key — the free block arrangement is the extra degree of freedom S3 never names.", "params": { "group_key": "SubType/Type/HiddenType/GANg/Drop/Eyes/Dimensions", "block_order": "all permutations of blocks", "inner_key": "any column asc/desc", "depth": "2-3 steps" }, "examples_from_history": "the collector: 'orderings based on principles grounded in the GANs (... sub types or types grouped in order and then put together, other multi step stackings)'; 'One step, two step, three step' (bf7ee7f4). Worker built the '2-step: group by A, permute blocks, sort within by B (~2,000 orderings)' table; compound-ordering space reported tested on the strong sets.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (allorders-family runs on strongest sets only)", "pindar": "the collector's block-permutation idea: order inside blocks by mint (canon-obedient), then permute the BLOCKS (small, enumerable). Honors the mint precedent while admitting the eleven might be assembled by chapter.", "plain": "Two-step ordering: split the eleven into blocks (by drop, by class), keep mint order INSIDE each block, then arrange the blocks by a second rule. Four blocks only have 24 arrangements — all can be tried." }, { "id": "OC16", "name": "representative-then-reorder (first-by-GAN per category, reordered by mint)", "def": "Two-stage compound: select one representative per category by rule R1 (e.g. lowest GAN number), then order the representatives by a DIFFERENT key R2 (e.g. mint index). Selection key and ordering key deliberately decoupled.", "params": { "rep_rule": "first-by-GAN / last-by-GAN / first-by-mint / first-uploaded", "reorder_key": "mint / GAN / upload / batch-alpha" }, "examples_from_history": "the collector verbatim: 'For each category, take the first by GAN number, but then reorder by mint number. Try all 128 possible final words' (6c9a8360). Also the batch-speaker orderings (speaker-GAN / speaker-mint / batch-anchored mint in qspeak).", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (batch families; the specific first-byGAN-reorder-by-mint run the collector requested)", "pindar": "Representative-then-reorder: pick each category's first piece, then reorder representatives by mint. A two-step selection-ordering hybrid.", "plain": "Take one representative per category (each category's first piece), then order the representatives by mint." }, { "id": "OC17", "name": "calendar order (Solstice-precedent temporal layout)", "def": "Order words/pieces by where they fall in a yearly calendar — the mechanism the collector identifies as the actual Solstice-puzzle shape ('you could line up essence by when they appeared in a yearly calendar and get the solution'). For Act 1: order by any recoverable date semantics (seasonal words, save dates, drop dates).", "params": { "key": "calendar_position", "source": "date semantics of word or piece" }, "examples_from_history": "the collector, bf7ee7f4: 'In the solstice puzzle, the answer was a calendar.' Never generalized to an Act 1 ordering; S2-W17 covers date DECODE as a word register but S3 has no calendar ORDERING row.", "proposed_by": "collector", "in_solution_space": false, "tested": "no", "pindar": "the collector's calendar idea, and the Solstice precedent is direct: that puzzle ran spring→winter. Act 1 has drop dates, Date traits, burn chronology, and eleven re-save sessions to sort by. The artist has already used the calendar once.", "plain": "Order by CALENDAR: each piece's real date (drop date, save date, burn date), earliest to latest — or spring-to-winter like the Solstice puzzle." }, { "id": "OC18", "name": "hue / rainbow-gradient order", "def": "Arrange the selected pieces into their own color gradient (rainbow walk) and read in hue order; motivated by the ghost drop cycling one hue-step per neighbor.", "params": { "key": "dominant_hue (GAN figure, not whole frame)", "dir": "spectral" }, "examples_from_history": "the collector: 'the Ghost Gan drop cycles through colors of the rainbow... each one shaded at a one step remove from its nearest neighbors'; 'did you rearrange the out of order ones into their own color gradient to get the answer?' (6c9a8360). Worker ran the color-eleven 'in self-hue order' — negative.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (color-eleven at essence and number registers in self-hue order; other selections unswept)", "pindar": "The ghost drop is a rainbow gradient and #357 is a color-glitch IN that gradient — hue order is the one ordering where #357's anomaly is a position, not just a color.", "plain": "Order by COLOR: arrange the eleven along a rainbow/hue gradient, the way the ghost drop itself is arranged." }, { "id": "OC19", "name": "burn-chronology order", "def": "Order by the burn transactions' block/timestamp (forge history), for sets whose members have no artwork traits to sort on (forged11) or whose members were burnt.", "params": { "key": "burn_block_timestamp", "dir": "asc/desc" }, "examples_from_history": "walks_round_rules.md family F: 'burn chronology of their source burns (block/timestamp from burnt_final.json)'; a report lists 'burn order' among ~60 orderings swept on the burnt elevens.", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (forged11/burnt sets)", "pindar": "Burn chronology orders by destruction date — all pre-clue, all documented on-chain. Pairs naturally with burnt-axis selections.", "plain": "Order by the dates the burnt pieces were destroyed." }, { "id": "OC20", "name": "batch-label order (alphabetical batch names / numeric code order)", "def": "Order by the Arweave batch LABEL itself: named batches alphabetically (with/without reprise-prefix stripping, line/lines merge, dgans displacement), or numbered batches in code order 00-13.", "params": { "key": "batch_name_alpha | batch_code_numeric", "normalizations": [ "strip reprise", "merge line/lines", "512-as-container-to-end" ] }, "examples_from_history": "The whole sentence lineage: 'define cloud project collect bronze deposit rose draw legend give final' read in batch-alphabetical order; instruction_reading.md dissects the three normalizations; code-order used in the C(13,11) numbered-batch sweep ('one GAN from each code, in code order').", "proposed_by": "both", "in_solution_space": false, "tested": "yes (qspeak/qsent full spaces; VA/VB 12,544 orderings)", "pindar": "Batch labels order by production run; thin but producer-side.", "plain": "Order by production batch label (alphabetical batch names or their numeric codes)." }, { "id": "OC21", "name": "artefact text order (page/word-list order, with split and interleave readings)", "def": "Order = the sequence in which words appear in a published artefact text (clue page, drop description). Variants: first-eleven / second-eleven splits of a longer list, every-other-word interleaves, text order vs carrier-GAN order.", "params": { "key": "text_position", "variants": [ "as-written", "reversed", "first-11/second-11 split", "every-other" ] }, "examples_from_history": "Ghost-description eleven 'series art there theme this color find more about van act' in text order (walks_round_rules D — repairing a never-run negative); the 22 clue-page BIP words split/interleaved ('we take the first eleven and then the second eleven... once in the order the words appear on the page'). the collector: 'Do the ghostGAN description BIP words appear in the Essences across the entire collection? Maybe that gives the ordering.'", "proposed_by": "both", "in_solution_space": false, "tested": "yes (D family closed; page-word splits run)", "pindar": "Artefact text order reads pieces in the order some artifact lists them (clue page, lore page). The artifacts are Pindar's own publications.", "plain": "Order by where each piece (or its word) appears in some artifact text — the clue page, a journal entry — read top to bottom." }, { "id": "OC22", "name": "sentence-construction order (semantic arrangement)", "def": "Arrange the eleven words so they read as English — an instruction, read-out, or ode — including fixed semantic pair-blocks permuted as units (verb+noun clauses), clause-grouping choices, and aptness-constrained placements (payoff last, 'random final' adjacent).", "params": { "blocks": "verb+noun pairs fixed internally", "block_arrangements": "all 5,040 x word substitutions", "criteria": "payoff-last, clause coherence, blind-rater aptness" }, "examples_from_history": "the collector's canonical reading 'escape cloud source / collect reward / remember rose / draw legend / give final' and his 16 hand-written variants ('define cloud project draw legend remember rose random final collect reward', etc., 6c9a8360); qsent/qsent2 pair-block sweeps (20,160 orderings each) owner-requested; better_sentences.md repairs ('draw cloud / escape source / collect reward / remember rose / elder legend / final'). Null model verdict: manufactured (P=0.572).", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (extensively; verdict apophenia but full spaces closed)", "pindar": "Sentence-construction: arrange the eleven words so they SAY something. Pindar writes jokes into names; a phrase that parses would be very him — but it is post-hoc, so it proposes and never confirms.", "plain": "Arrange the eleven words so they form a meaningful SENTENCE. Human judgment decides the arrangement; the sentence justifies it." }, { "id": "OC23", "name": "walk-defined order (pointer-chase; selection = ordering in one act)", "def": "A step rule maps each artwork to the next (essence-index hops, mint-position hops, etc.); the eleven words are whatever the walk visits, in visit order. Includes terminating walks run backwards from a motivated terminus (#468 'random', #512 'final', #445 prize, #191, #32, #119).", "params": { "step_rules": "9-15 implemented (essence idx mod 513/512, gan+idx, bit-reversal, complement, mint-position, SLIP route)", "starts": "12 motivated anchors", "termini": "6 motivated endpoints", "alignment": "terminus as 12th word or as 11th" }, "examples_from_history": "the collector's seed: 'Let's say that the GAN whose essence is random IS the 12th GAN. Can we reverse engineer the other 11 from that?' (bf7ee7f4). transforms_and_mechanisms 3a and 11: 346 walks tested; only one rule reaches #468 in eleven steps (chain 'token mistake author... random' checksum-legal, negative). No eleven-cycle exists in the collection.", "proposed_by": "both", "in_solution_space": false, "tested": "yes (346 walks, 1,140 tests; no cycle census: zero 11-cycles)", "pindar": "Walk-defined: each piece points to the next (partner, homage, +N). Selection and ordering in one act — matches the relation-edge web the collection demonstrably has.", "plain": "Pointer-chase: start somewhere, and let each piece point to the next (its partner, its homage target, +N). Selection and ordering happen in one walk." }, { "id": "OC24", "name": "lattice walks in number space (AP / Fibonacci-gap / glitch-schedule)", "def": "Ordered selection by stepping through GAN or mint space with a gap pattern: arithmetic progressions a, a+k, ..., a+10k; Fibonacci gaps (1,1,2,3,5,8,13,21,34,55) and reversed; Pindar's measured reinsertion ramps (6,4,3,2,2)/(4,3,2,2) cycled — visit order IS the word order.", "params": { "space": "GAN 1-512 or mint 1-662", "gaps": "constant k / fib / glitch-schedule", "dir": "asc/desc" }, "examples_from_history": "walks_round_rules.md W1-W4 (stated 2026-08-23 before running); the collector's prompt behind it: 'is there a way to use the spacings to derive sets of 11?' after the Fibonacci frame-insertion discussion.", "proposed_by": "both", "in_solution_space": false, "tested": "yes (mechanised closure)", "pindar": "Number-space walks (Fibonacci gaps — the HODL list is Fibonacci!). The artist published the seed of this idea himself.", "plain": "Walk the number line by a rule — every Fibonacci number, a fixed gap — and take pieces in the order the walk visits them." }, { "id": "OC25", "name": "perturbation orderings (transpositions, rotations, move-one-to-end)", "def": "The natural order perturbed by every single transposition (55), all cyclic rotations (10), and the eleven displace-one-to-end moves — motivated by a DOCUMENTED authored failure mode (#259/#268 essences swapped in the Solstice register).", "params": { "base": "any natural order", "perturbations": [ "single transposition x2 dir", "rotation", "one-to-end" ] }, "examples_from_history": "multistep_chains.md batch 3 (919 orderings on the seven strongest sets); the collector's pair-swap eye ('309/310 feel like they should be swapped', '333/334 seem transposed') supplies the same intuition on the visual side.", "proposed_by": "both", "in_solution_space": false, "tested": "yes (on the seven strongest sets only)", "pindar": "Perturbations of mint (swap two, move one to end) formalize 'almost canon' — the shape of an artist who breaks exactly one rule at a time.", "plain": "Start from mint order, then make ONE small change: swap two neighbors, or move one piece to the end." }, { "id": "OC26", "name": "mirror / palindromic ordering", "def": "Read a set out-and-back (forward then reverse, mirror from an anchor), the ordering analogue of the nagaGAN forward-then-reverse loop convention.", "params": { "anchor": "terminus or #449", "shape": "mirror" }, "examples_from_history": "transposition_sets.json labels mirror_from_terminus, mirror_from_449; the collector: 'Calcuclac and TacoCat definitely run forwards and then in reverse... are they a pair of Glitches from Drop 2 we can take into a combined drop theory?' (6c9a8360).", "proposed_by": "both", "in_solution_space": false, "tested": "yes (per transposition_sets labels)", "pindar": "Mirror ordering is licensed by the naga class: name, position, AND motion all palindrome. If any ordering is signed into the art, it is this one.", "plain": "Make the eleven read the same forwards and backwards, or read the phrase in reverse — the palindrome class licenses this move." }, { "id": "OC27", "name": "condensed-list renumbering (filter, renumber, transplant)", "def": "Filter a list (e.g. podGAN essences that are BIP-valid), renumber the survivors consecutively — 'the condensed list is the new order' — then map the new index back to bitGAN numbers or forward to a target register. An ordering-by-reindexing move.", "params": { "filter": "e.g. BIP-valid subset", "renumber": "consecutive", "map": "new index -> bitGAN number or target list" }, "examples_from_history": "the collector verbatim: 'No the condensed list is the new order. And then you match the new number to the bitGAN number' (bf7ee7f4, podGAN transplant thread).", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (podGAN transplant runs, negative)", "pindar": "the collector's condensed renumbering as an ORDERING: filter, renumber, read positions in the new list. podGALLERY-proven.", "plain": "Filter, renumber the survivors 1,2,3…, and use the NEW numbers as the ordering (the podGALLERY move applied to order)." }, { "id": "OC28", "name": "mint-timestamp order (real timestamps, not index)", "def": "Order by actual mint timestamps rather than token index — gaps and clusters in real time could carry signal that the bare index hides.", "params": { "key": "mint_timestamp", "note": "gaps possibly meaningful; anything minted after clue date out of window (the collector)" }, "examples_from_history": "Worker queue: 'Actual mint timestamps — not just order.' the collector on gaps: 'You shouldn't look at anything minted after the contest announcement date... You'd have to run that all down' (6c9a8360).", "proposed_by": "both", "in_solution_space": false, "tested": "partial — every token's first transfer out of Pindar's wallet is now known (Blockscout, D56: first_transfers.json); for skulls it differs from mint order (#100, #97 ahead of #81, #88, #66). Not yet run as an ordering on A/B (no precedent supports it).", "pindar": "Real mint timestamps (not index) — differs from mint order only where mints batched; documents the distinction.", "plain": "Order by the real minting TIMESTAMPS (clock time), which differs from mint index where batches were minted together." }, { "id": "OC29", "name": "exhaustive permutation (11! and block-order brute force)", "def": "All 39,916,800 orderings of a fixed eleven (or all block arrangements for grouped sets). Proposed by the collector, then explicitly walked back as against the no-brute-force doctrine; one full run (person_11) was executed before the walk-back.", "params": { "space": "11! per set", "status": "doctrine-rejected except legacy runs" }, "examples_from_history": "the collector: 'Can we also just do all 11 words plus the random in every possible order?' then 'OK, kill all of these... If we've already tried the obvious orderings I don't think we should pursue a full brute force method' (bf7ee7f4). person_11 exhaustive 39.9M-ordering run completed per transcript.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (person_11 fully; family then abandoned by ruling)", "pindar": "Brute force over 11! is the anti-Pindar reading (he said statable rules, hard EXECUTION). Block-level brute force (OC15) is the sane subset.", "plain": "Try every possible arrangement (11! = 40 million). Brute force — the opposite of a statable rule; block-level version (OC15) is the sane subset." }, { "id": "OC30", "name": "two-set linking with seam placement", "def": "Organize two (or more) sets internally, then concatenate the sets in every set-order, treating the seams between sets as distinguished positions (random-word placement at seams; set A then B vs B then A).", "params": { "sets": "e.g. Monster/Cloud; test/line/aa/bs + corrupted-7", "inner_order": "each set's own orderings", "link_order": "all set permutations", "seam": "random-word candidate position" }, "examples_from_history": "the collector: 'We should consider Monster/Cloud to be sets and run variants that organize the GANs inside of those sets and link them together in different orders... random word at the beginning, in between and at the end of the two sets'; 'there are 4 groups outside of the 7 corrupted... test all combinations of those with the 7 corrupted. Put the random at the start, seam and end' (bf7ee7f4).", "proposed_by": "collector", "in_solution_space": false, "tested": "yes (cluster-union runs)", "pindar": "Two-set linking with a seam: two small classes joined, seam placement chosen. The mix-and-match doctrine turned into an ordering question.", "plain": "Join two small classes into one eleven and choose where the SEAM goes — which class leads, which follows." }, { "id": "OC31", "name": "pair-linked / additive-composite orderings", "def": "Order driven by pair relations: linked pairs kept adjacent, and additive composites per Pindar's stated '2 eyes + 3 eyes = 5 eyes' rule (order by sums/differences of paired trait values).", "params": { "key": "pair adjacency; eye-sum composites; differences" }, "examples_from_history": "Worker report: 'Swept ~60 orderings across both elevens — GAN number, burn order, mint order, alphabetical, BIP39 index, pair-linked, additive composites (Pindar's stated 2 eyes + 3 eyes = 5 eyes rule) and differences.'", "proposed_by": "assistant", "in_solution_space": false, "tested": "yes (on the two burnt/forge elevens)", "pindar": "Pair-linked orderings keep ratified pairs adjacent — if pairs are events, their members should travel together.", "plain": "Keep ratified pairs ADJACENT: order the events, and let each pair's two members travel together." }, { "id": "OC32", "name": "drop-quota assembly order", "def": "Assemble the eleven with a per-drop quota (a few words per drop, sized to the drop — 'maybe only 1 in Spooky'), ordered drop-sequence-first; the drop sequence itself carries the ordering while different glitch KINDS fill each drop's quota.", "params": { "quota": "per-drop counts proportional/1-per-small-drop", "order": "drop chronology then intra-drop key" }, "examples_from_history": "the collector: 'It may also argue for each drop having just a few of the seed words/solution. And our compiling of multiple different glitch types together might give us the latitude to fill out the possibility space' + 'visual glitches or other solutions may be different on a drop by drop basis... maybe only 1 in Spooky' (6c9a8360). D1 spooky three-variant ruling feeds this.", "proposed_by": "collector", "in_solution_space": false, "tested": "partial (Sudoku assemblies encode it; no dedicated drop-quota sweep labeled)", "pindar": "Drop-quota assembly (one from each drop, then...) mirrors how Eleven-A/B/C were built; mostly spent, documented.", "plain": "Fill a quota per drop (one from each, or a fixed count), then order the quota winners by a second rule." }, { "id": "OC33", "name": "listing-surgery corrected order", "def": "Order by OpenSea listing-creation order CORRECTED for Pindar's documented relist surgery (out-of-order relists #24/#30/#310/#311/#359/#390/#445/#446; in-slot replacements 404/420) — 'if something messed up the ordering, correcting the ordering would apply no matter what' (the collector).", "params": { "key": "listing_creation_order", "correction": "undo relists / honor replacements" }, "examples_from_history": "qsurg LS family (listing-surgery eleven, full stateable depth); the collector's challenge on whether corrected order differs from GAN order (6c9a8360).", "proposed_by": "both", "in_solution_space": false, "tested": "yes (qsurg; correction shown to equal GAN order except on relisted members)", "pindar": "Listing-surgery: the corrected order after fixing known listing errors (104→404, 120→420). Reads the collection as Pindar INTENDED it to be numbered.", "plain": "Order by the numbering as Pindar INTENDED it after fixing his known listing surgeries (104→404, 120→420)." } ], "priority_recipes": [ { "id": "PR-1", "name": "Calendar order", "uses": [ "OC17" ], "status": "never-swept", "plain": "Idea: the eleven are ordered by CALENDAR - the real dates attached to the pieces (drop dates, the Solstice date traits, burn dates, or re-save dates) - instead of mint order. How to run it: pick a selection, look up each piece's date, sort by date, read the words. Variants: which date field; earliest-first or latest-first; season order like the Solstice puzzle (spring to winter)." }, { "id": "PR-2", "name": "TRUE/FALSE as classifier", "uses": [ "EC40" ], "status": "never-swept", "plain": "Idea: #363 (essence TRUE) and #473 (essence FALSE) are not words in the phrase - they are a LABELING MACHINE. Each candidate piece gets marked true or false by some test (does its label match its art? is it burnt? does it match its drop theme?), and only the TRUE side (or only the FALSE side) joins the eleven. the collector's own note on the pair {363,473}: TRUE/FALSE may be a statement about each piece's relationship to its own drop." }, { "id": "PR-3", "name": "podGAN transplant with condensed renumbering", "uses": [ "OC27", "EC29" ], "status": "never-swept", "plain": "Idea: borrowed from the solved podGALLERY puzzle. Take a filtered list (say, just the glitched pieces), throw everything else away, and RENUMBER the survivors 1, 2, 3... Then use those new numbers - not the original piece numbers - as wordlist positions or pointers. The eleven could also be positions in the NEXT collection (podGANs) read back into this one." }, { "id": "PR-4", "name": "Block permutation", "uses": [ "OC15" ], "status": "never-swept", "plain": "Idea: order the eleven in two steps. First group them into blocks (by drop, by class, by pair), order INSIDE each block by mint, then arrange the BLOCKS themselves by a second rule (drop order, size, alphabetical, reversed). Small numbers: 4 blocks only have 24 arrangements - cheap to enumerate." } ], "grammar": { "slots": [ "SELECT", "MAP", "EXTRACT", "TRANSFORM", "ORDER" ], "plain": "A recipe fills five slots. SELECT: which pieces (or events) are the eleven glitches - one class, or a mix of small classes (the collector's doctrine: mix and match). MAP: optionally hop from each selected piece to another piece (its partner, its pointer target, its renumbered position). EXTRACT: how a piece becomes a word (a register from the 47). TRANSFORM: optional word-level fix (repair the typo, reverse it, index arithmetic). ORDER: how the eleven line up (one of the 33). The 12th word is random and sits last - we sweep all 128 checksum-valid endings, so it costs nothing.", "discipline": "A negative sweep kills exactly one full recipe (selection x register x ordering), never the class itself. Every sweep label must encode all three IDs." }, "values": { "painted_numbers": { "165": "81", "179": "10 . 00 (maybe)", "207": "4", "379": "AI" }, "radix_pair": { "465": "0x20 -> 32", "466": "0x21 -> 33" }, "non_bip_essence": { "191": "err0r" }, "boolean_essences": { "363": "TRUE", "473": "FALSE" }, "anchor_essences": { "357": "seed", "512": "final" } } }