# Lens 5 — ORDER (swarm of 21 September 2026)

Ideas only. Nothing here was run against a wallet, no sweep tool was used, no network was touched, and no existing file was changed. Every number below was computed from primary_facts.py, bip39.txt, pairs.json, tags.sql, the same-names data, fib_return_frames.json, the sweeps/*_tasks.json files (read only, to see which orders were already run), theories/_registry.json and theories/DECISIONS.md.

## What the order lens found before any idea

I first asked the question the brief set: on the elevens we already hold, does any untried rule give one clear order with no ties? Here is what the data says.

**Eyes and Dimensions as slot numbers: no held eleven passes.** I checked all fifteen elevens from the big systematic run (broken essences, old chain list, chain lists A and B, "not a substitution", creature, backwards-Fibonacci, mislabel, the three Named elevens, number versus time, letter typos, title glitches, the drop rule), plus the out-of-sequence eleven, the twelve "so" words and the collector's nine. For a trait to work as slot numbers, all eleven pieces must carry different values. The best any of them does on Eyes is 7 different values (broken essences); most have 3 to 6. On Dimensions the best is 5. Eyes plus Dimensions, and Eyes times Dimensions, also leave ties on every one. So on the selections we hold, neither trait gives an order. This is a clean negative.

**Remainders do not work either.** Piece number, mint slot and wordlist place, each taken as a remainder after dividing by 11 or by 12, never give eleven different slots on any held eleven.

**"A displaced word says where it goes" adds nothing new.** For the eleven out-of-sequence words there are two honest readings. Order by the seat each word sits in is piece-number order. Order by the hole each word left behind is wordlist order, because the holes run in the same direction as the wordlist. Both were run (D88, all twelve plain orders, both shapes). The sizes of the jumps (bean +136, mistake +1078 and so on) are not small numbers and do not number slots.

**"A pair marks from and to" also collapses.** Ordering pairs by their earlier piece or by their later piece gives the same line-up, except that the TRUE/FALSE pair (#363, #473) and the key/solution pair (#411, #450) overlap and trade places. Both line-ups were inside the one-of-each-pair runs. The only pair ordering never run is ordering by the gap between the two pieces (idea 4).

**What this means.** About twelve plain orders have been run on every held selection with seven word rules. If one of those selections and one of those word rules were right, a plain order would very likely have hit, because Pindar's two solved puzzles used plain mint order. So the order is the unread part only for a set that carries its own built-in order. The ideas below look for such sets, or give a way to test for one without a wallet.

---

## Idea 1 (best) — The number words: a set that states its own order

**Side:** all DATA (the essence words on Pindar's sheet and the wordlist).

- **Selection, as Pindar could say it:** "The glitches are the pieces whose essence is a counting number." The wordlist holds twelve counting words: zero, one, two, three, six, seven, eight, ten, twelve, twenty, hundred, million. Ten of them are essences: eight #137, three #176, twelve #177, zero #192, one #316, twenty #392, hundred #400, two #424, six #441, ten #442. Only 'seven' and 'million' are missing.
- **Word, as Pindar could say it:** "Each number tells you which piece to read: take the essence of the piece with that number." That gives zero → #0 universe, one → #1 abandon, two → #2 ability, three → #3 able, six → #6 absent, eight → #8 abstract, ten → #10 abuse, twelve → #12 accident, twenty → #20 act, hundred → #100 arrest. (Plain alternative, same selection: use the number word itself as the seed word.)
- **Order, as Pindar could say it:** "Count upward." The words sort themselves: 0, 1, 2, 3, 6, 8, 10, 12, 20, 100. There are no ties and no judgement.
- **Shape:** the pointer reading produces the bookend shape without anyone adding it by hand: universe first, then nine words, then 'final', then the random twelfth: *universe abandon ability able absent abstract abuse accident act arrest final + random word*. With the number words themselves the shapes are universe + ten + random, or ten + final + random.

**What in the data teaches it.**
- Ten of the twelve counting words are essences. All ten sit in drops where Pindar chose words freely (four robo, six ghost); none is forced by the alphabetical run of the skull and solstice drops. With about 276 freely chosen essences taken from about 1,800 words, the chance of holding ten or more of the twelve counting words is about 1 in 3 million. He placed them on purpose.
- Two of them tell the truth about their own piece: 'eight' sits on the robo whose binary counter is 001000, which is eight, and 'hundred' sits on ghost piece 100/144. So Pindar looked at the counter when he placed number words. The other eight are numbers that do not match their piece: a label that lies, which is the collector's untried reading.
- In piece order (and in mint order) the ten read 8, 3, 12, 0, 1, 20, 100, 2, 6, 10. That is a count shown out of sequence, the collector's theme one level up. 'sort' is the essence of #494.
- 'number' is the essence of #470, the piece with the most eyes in the whole collection (10, shared only with #401), and it sits in working group reprise11, one of the groups that is out of order. 'random' is the essence of #468, two pieces earlier.
- The missing counting word is 'seven', and the missing working group is reprise07.

**Already tested?** Only partly, and not in these forms. In August the twelve number-like essences (the ten plus 'double' and 'half') were tried as a complete twelve in piece-number order, and the pointer words were computed, but the note says "neither the twelve nor the ten is eleven" and it stopped there (theories/transforms_and_mechanisms.md section 3i; extraction catalog entry "number-word chase"). That was before the bookend shape became a standing option (D65). The registry holds no set with even six of the ten number words, and none of these keys: the ten plus final, the ten plus universe, the ten plus number, the pointer ten plus final. Order by value was never run. DECISIONS.md has no entry for it.

**What it predicts that we can check without a wallet.** (1) The collector can look at the ten pieces side by side and say whether they share anything by eye. (2) If the pointer reading is right, the pointed-at pieces should matter: they are #0, #1 The Moirai, #2 The Artist, #3 The Geometer, #6, #8, #10, #12, #20 and #100, and the collector already calls #100 a Moirai echo (#12 carries the same note only as a soft call). (3) The zero-based version of the pointer (the wordlist counted from 0) turns 'twenty' into 'action', the one word from 1 to 224 that no piece holds. That is a second, separate reason to try the zero-based count.

**Phrases needed:** about 20 (three word readings, up and down, two or three shapes each), each with the 128 legal twelfth words. This is one of the cheapest tests on the board.

**Why Pindar would do it this way.** His solved puzzles used a word you can read off the piece and an order nobody can argue about. Counting words are the only seed words that come with their own order. It also explains why he would call the last word "random": the first ten-plus-one are counted, the twelfth is not.

**Weakest point.** These ten pieces are not glitches in any way we have measured; only #176 'three' carries the frame glitch. The colour words were a family that Pindar also filled completely (ten of ten), and that family was swept dead (D46), so a filled word family can be a theme and nothing more.

---

## Idea 2 — A LESSON: Eyes is the only trait built to number eleven slots, so use it as a free test

**Side:** all DATA (the Eyes column on the sheet). This is a lesson and a filter, not a selection.

- **Selection:** "The eleven glitches all have a different number of eyes."
- **Word:** whatever word rule the selection already uses (the essence is the default).
- **Order:** "Put them in eye order, fewest eyes first," from 0 eyes to 10 eyes.
- **Shape:** eleven words plus the random twelfth.

**What in the data teaches it.** Across the 512 pieces the Eyes column takes exactly eleven values: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10. Dimensions, which the collector once proposed as the slot numbers, takes only eight values (2 to 8, and 11 on #512), which is why that idea needed three wild slots. The end values are rare, so they pin slots hard: 10 eyes is only #401 'dolphin' and #470 'number'; 9 eyes is only #100, #310, #431; 0 eyes is only #131 'blind', #137 'eight', #145, #195; 8 eyes is seven pieces; 1 eye is six pieces, among them #39, the cyclops the collector marked "on purpose", and #259. Several of these rare pieces are already glitch carriers on the data side: #131 'blind' has zero eyes and is a displaced essence, #100 is an error piece, #310 was minted out of turn, #470 is in an out-of-order working group.

**What I computed.** No eleven we hold has eleven different Eyes values (best 7). In 200,000 random draws of eleven pieces, none had more than 9 different values, and only 7 draws reached 9. So an eleven with all eleven eye counts cannot happen by accident. If one ever turns up from an honest rule, that is strong proof before any wallet test, and it arrives with its order already fixed and free of ties.

**Already tested?** Eyes was used once as a sort key, on the twelve 'so' words only (D80). Eyes as slot numbers over 0 to 10 appears nowhere in DECISIONS.md or the ordering catalog.

**What it predicts without a wallet.** Every candidate eleven from the other lenses can be checked in seconds: are the eleven eye counts all different? The nearest natural pool today is the "out of sequence, from the data" tag (31 pieces). It covers ten of the eleven eye counts and has exactly one piece at 0 eyes (#131), 1 eye (#259), 6 eyes (#73), 7 eyes (#472) and 10 eyes (#470). It has no piece with 8 eyes, so it fails as it stands.

**Phrases needed:** none for the filter. If a pool ever passes, one or two phrases per passing eleven.

**Why Pindar would do it this way.** The glitchGANs are drawn around their eyes, his lore counts eyes ("2 eyes + 3 eyes = 5 eyes"), and an eye count is something you can read off the piece.

**Weakest point.** No natural pool passes today. Forcing one would mean picking among many pieces with 2, 3 or 4 eyes, which is brute force.

---

## Idea 3 — A LESSON: most position words are forced, and the three end markers say the order is already right

**Side:** all DATA (essences and the wordlist). This is a lesson.

**The lesson in three sentences.** (1) Skull and solstice essences follow the wordlist from word 1 to word 224, so every position or instruction word in that range is there because the alphabet put it there: above #5, across, ahead, among, apart, around, away, before #240, behind, below #245, between #252, beyond, bottom #287, border, base, axis, and also add #26, again #38, arrange #99, begin #241, adapt, adjust, alter, advance, avoid, bind. Pindar did not choose those words, so they cannot place their neighbours. (2) Only eight position words were freely chosen: under #179, edge #311, over #346, off #355, position #395, index #408, upper #426, south #508. The freely chosen re-ordering words are flip #142, jump #152, mirror #187, cross #347, swap #402, move #417, sort #494. (3) Wherever Pindar did place an "end" word, it sits on the last piece of its set by piece number: 'bottom' #287 is the last of the eleven backwards-Fibonacci pieces (the word was forced, but the choice to put the full glitch on that piece was free), 'south' #508 is the last of the twelve 'so' words by piece number and by wordlist, and 'final' #512 is the last piece of all. The first of the backwards-Fibonacci eleven, #148, carries 'hint', and the first 'so' word, #450, is 'solution'.

**Three slots for the set this lesson points at.** Selection: "the eleven pieces where one frame comes back at gaps 5, 3, 2, 1, 1" (all VISUAL; the collector's own definition). Word: not known; this is the open part. Order: "read them upward by piece number; 'hint' starts and 'bottom' ends." Shape: eleven plus the random twelfth; or, if 'bottom' is only a marker, the ten robo pieces plus 'final' plus the random twelfth.

**What I computed.** I looked at the first and last essence of every held eleven in piece order. Only three sets end on an end word: the backwards-Fibonacci eleven ('bottom'), the invented-name Named eleven ('final') and the 'so' block ('south'). The others end on frost, author, lucky, negative, reward, calm, zero, skull, network.

**Already tested?** The orders were: piece order on these sets has been run with all seven word rules (D77 to D80). That is the point of the lesson. On the one set that is "eleven glitches" by the collector's own eye, the order is very likely already correct, so the unread part is the WORD. The order lens says: stop re-ordering that set and look for a new word rule on it.

**What it predicts without a wallet.** Any new eleven should be checked for an end word on its last piece. The eight free position words are a short list to check against any new selection: #426 'upper' is the clean same-name partner of error piece #315, #311 'edge' is one of the eight pieces minted out of turn, and #179 'under' is in the mislabel eleven.

**Phrases needed:** none.

**Why Pindar would do it this way.** He signs the ends of things: #0 'universe', #512 'final', Solar and Lunar at the two ends of the solstice drop.

**Weakest point.** Three end markers is a small sample, and 'bottom' lies inside the forced run.

---

## Idea 4 — The nine standing data pairs, in the bookend shape, ordered by the gap between the two pieces

**Side:** all DATA. The nine pairs known from files, names and essences that still stand after the collector's review: identical files #14/#75, held frames #479/#480, Solar and Lunar #209/#300, TRUE and FALSE #363/#473, 'Primoridal' #16/#100, 'Gan' #175/#205, 'gltich' filenames #9/#73, 'Raibow' #170/#175, key and solution #411/#450. The scanline pair, the flash pair and the painted-number pair are known only by eye and stay out.

- **Selection:** "Each glitch is a piece that comes back: two pieces that share something no other piece has."
- **Word:** "Take the essence of the piece that comes back" (the later piece of each pair); the mirror rule is the earlier piece.
- **Order:** "Widest gap first, as the glitch frame returns at 5, then 3, 2, 1, 1." The nine gaps are 110, 91, 84, 64, 61, 39, 30, 5, 1. They are all different, so the order has no ties. Later-piece words in that order: false, brief, arrest, animal, announce, solution, grace, skull, fossil.
- **Shape:** universe + nine + final + random twelfth. Nine is exactly what that shape needs.

**What in the data teaches it.** The standard glitch is one frame that returns after shrinking gaps. A pair is the same thing one level up: a piece that returns later in the collection. The gap is then the natural measure, and it is the only pair ordering never run. A side observation that I flag as MIXED and so not usable as a selection: among the twelve standing pairs, exactly four have a short gap, and those gaps are 5 (#170/#175), 3 (#130/#133), 2 (#489/#491) and 1 (#479/#480), the glitch's own numbers, each once. #473 FALSE and #474 'truth' would supply the second 1.

**Already tested?** Not in this form. The one-of-each-pair run (D88, D89) used ten data pairs, including #256/#479, which the collector withdrew the next day (D91), and eight plain orders: piece number, wordlist place and date made, each both ways. Nine pairs in the universe + nine + final shape were never run, and gap order was never run. D91 says "No runs" after his pairs review.

**What it predicts without a wallet.** Nothing strong. The collector is still "not sure" on five of the nine.

**Phrases needed:** 8 with the fixed choosers (later or earlier piece, widest or narrowest gap first, two shapes), or 512 × 2 if every one-of-each choice is allowed (#175 sits in two pairs, which the rules allow).

**Why Pindar would do it this way.** It reuses the one mechanism he shows again and again, and nine plus his two bookends is a tidy eleven.

**Weakest point.** Five of the nine pairs rest on spelling slips, which the collector's taste says look random. His five "sure" pairs mix data and by-eye evidence, so they cannot be used together.

---

## Idea 5 — A new plain order: the counter printed in the title, read in its own base

**Side:** all DATA. This is an order to add to the standard twelve; it applies to any selection.

- **Selection:** any held eleven.
- **Word:** any of the seven standard word rules.
- **Order:** "Read the counter I printed on each piece." Skulls count n/128, robos count in binary (000000 to 111111), spookies in octal, ghosts n/144, solstice and Review show the whole-collection number. Fourteen Review pieces carry binary or hex counters that are not their piece number (#458 shows 1000001, which is 65; #466 shows 0x21, which is 33).
- **Shape:** both.

**What I computed.** This order is free of ties on 13 of the 15 systematic elevens, and it differs from piece-number order on 11 of them. Example, broken essences: #129 (0), #131 (2), #141 (12), #21, #57, #189 (60), #191 (62), #363 (63), #259, #268, #473. It leaves a tie on the drop-rule eleven, the old chain list and the collector's nine, because #44 and #173 both show 44 (#173's counter is 101100). Among the 29 sheet-versus-blockchain pieces there is one more clash: #73 and #281 are both the 73rd piece of their drop. About one clash is expected by chance, so I draw nothing from it.

**Already tested?** No. "Place in its own drop" was used as a WORD rule in the systematic run (D79) but never as an ORDER. The twelve orders there were piece number, mint slot, date made, first transfer, wordlist place and title A to Z, each both ways.

**What it predicts without a wallet.** Nothing on its own.

**Phrases needed:** 2 directions × about 15 elevens × 7 word rules × 2 shapes, roughly 400 to 2,000 after the leave-one-out versions.

**Why Pindar would do it this way.** Changing the number base from drop to drop is one of the loudest features of the collection, and the counter is printed on the piece where anyone can read it.

**Weakest point.** For any selection inside a single drop it is the same as piece order. Both solved puzzles used plain mint order.

---

## Idea 6 — The scrambled count as a ten-place key

**Side:** all DATA.

- **Selection:** any ten-piece glitch set in the bookend shape (for example the collector's nine plus one "not sure" piece, or the ten robo pieces of the backwards-Fibonacci eleven).
- **Word:** the set's own rule (the essence is the default).
- **Order:** "The number pieces tell you where each word goes." In piece order the ten number words read 8, 3, 12, 0, 1, 20, 100, 2, 6, 10. Ranked by value that is the key 6, 4, 8, 1, 2, 9, 10, 3, 5, 7: the first glitch goes to place 6, the second to place 4, and so on (or the reverse reading of the key).
- **Shape:** universe + ten + random, or ten + final + random.

**What in the data teaches it.** The same things as idea 1. The key is a full rearrangement of ten places with no ties, and ten is the size the bookend shape needs.

**Already tested?** No. The matching idea for the 'so' block (its 12-place and 11-place keys) was run on 17 elevens with no match (D80). That is why this sits last.

**What it predicts without a wallet.** Nothing on its own; it stands or falls with idea 1.

**Phrases needed:** 2 readings of the key × 2 shapes × the ten-piece sets we hold, a few hundred at most.

**Why Pindar would do it this way.** "11 glitches ending with a random word" needs an order, and a count is the plainest way to give one.

**Weakest point.** A transposition key is a cipher-style step, and Pindar's solved puzzles had none.

---

## Rules that give one clear order on real elevens (summary)

| Rule | Gives an order with no ties? | Already run? |
|---|---|---|
| Eyes as slot numbers | No held eleven (best 7 different values) | Only as a sort key on the 'so' words |
| Dimensions as slot numbers | No held eleven (best 5; the column has only 8 values) | Partly (the August slot work) |
| Eyes + Dimensions, Eyes × Dimensions | No | No |
| Piece, mint or wordlist number, remainder by 11 or 12 | No | No |
| Counter printed in the title | Yes on 13 of 15; differs from piece order on 11 | No |
| Number words by their own value | Yes, always | No |
| Pairs by the gap between the two pieces | Yes on the nine and on the twelve standing pairs | No |
| Displaced word by its seat, or by its hole | Yes, but these are piece order and wordlist order | Yes (D88) |
| Pairs by earlier or by later piece | Yes, but already inside the plain orders | Yes (D88, D89) |
| Pindar's journal number classes, then number | Only with the "then number" tie-break | On the 'so' words and the reprise originals only |