The series to either side of this paper refuses association. Paper 1 derived the Grounded/Carried grading; Paper 2 assigned suits to addresses and was honest enough to call its own table “associations, not rulings”; Papers 6–13 will establish Progenitors by the fixed method of Derive, Coin, Test. This paper is the irrational supplement between the integers: it is about association — what it is, why the system inherited it, what it costs, and what replaces it. It does not repeat, it does not terminate cleanly into either neighbor, and it holds the material that reduces to neither.
It also performs one act of repair. Earlier in this series’ planning record, the names Vast Mountain, Dark Star, and Desolate Mountain were rejected as Progenitor candidates — correctly, at the time, because they entered through a miscategorized source sketch and were reused inconsistently across tiers. This paper is the answer to that rejection. Two of those names re-enter here, not resurrected but re-derived: they arrive at ace positions because the mathematics entails a position and the Coin step attaches them to it, with failure conditions stated. A name rejected for arriving by association may be re-earned by arriving through derivation. Dark Star does not re-enter; its position remains open.
The suit system of record is a lookup table: four suits, two poles each, eight suit-pole combinations associated one-to-one with the eight lattice addresses. The file says of itself that these are associations — editable metadata, revisable by editing the file. That honesty is the file’s virtue and its indictment. Nothing in the table could not have been otherwise. Re-associate heart with a different complement pair and every record re-addresses without contradiction, because there is nothing for a contradiction to break against. An association has no failure condition, and a claim with no failure condition is not a claim.
The papers exist to overcome this. The overcoming is not a better table. It is the elimination of the table as a foundation — the addresses must be entailed, so that a wrong assignment is not a stylistic disagreement but a checkable error.
The alphabet is Σ = {yin, yang}. Nothing else is primitive.
A word is a finite ordered sequence of letters from Σ, including the empty word ∅. The sole operation is extension: appending one letter to a word. The set of all words under extension is the free monoid Σ*.
Two measures attach to a word, and they are not the same measure. The depth of a word is its length; ∅ has depth 0. The mark-count of a word is the number of yang letters it contains; ∅ and every all-yin word have mark-count 0. Depth locates a word's stratum in the census; mark-count measures its deviation from the all-yin word of its stratum. Conflating them is the same class of error Proposition 2 forbids.
That is the entire foundation. It is unownable in the strongest sense: anyone, anywhere, can re-derive everything below from these three definitions, and the definitions themselves are the common property of mathematics. No suit meaning, no agreement label, no grade, no classical glyph, and no house vocabulary appears at this tier — the quarantine of classical vocabulary is satisfied not by discipline but by construction, because there is nothing at this tier a quarantined vocabulary could attach to.
The number of words at depth n is 2n:
Two words with the same letter-counts but different orderings are distinct. yin·yang·yang and yang·yang·yin are different words. There is no operation in the system that identifies them. Any reasoning that treats trigrams as multisets of marks rather than ordered words is invalid at the foundation, and everything built on it inherits the invalidity.
The empty word is the Absent root. The first extension writes the first letter, and there are exactly two: this is the first deviation from Absent to present, and it is not a fourfold or an eightfold — it is the binary itself.
The two depth-1 words are the two stances. Presence enters a structure in exactly two ways: as ground beneath the observer, read from the bottom up (Grounded — what do I stand on?), or as arrival upon the observer, read from the top down (Carried — what reaches me?). These are not properties a word has; they are the two positions from which any word can be read, and they arise together at depth 1 or not at all. Two observers facing one another across the same structure each read the other’s yang as their own yin. The stance relation is the reading direction, and the reading direction is therefore derived, not decreed.
This resolves what Paper 1 grades and what the pip-tree diagrams draw as “first deviation from root, open/closed”: the first stratum below the root is Grounded/Carried, and every deeper word carries its reading stance as an inheritance from its first letter.
The eight depth-3 words are the eight trigrams, and the eight trigrams are the eight aces of the two decks. Writing yin as a broken line and yang as a solid line, and fixing once that string position runs top-to-bottom in the Carried direction, each word is drawn here beside its address. Addresses on this page are therefore carried-first tokens; the system of record writes grounded-first, and string reversal is the dictionary between the two serializations — cross-page comparison goes through the words themselves, never through raw tokens. The figures are house-rendered line marks; no classical glyph appears on this page or any live page of the compendium.
| Figure | Word | Address | Mark-count |
|---|---|---|---|
| yin·yin·yin | 000 | 0 marks | |
| yang·yin·yin | 100 | 1 | |
| yin·yang·yin | 010 | 1 | |
| yin·yin·yang | 001 | 1 | |
| yin·yang·yang | 011 | 2 | |
| yang·yin·yang | 101 | 2 | |
| yang·yang·yin | 110 | 2 | |
| yang·yang·yang | 111 | 3 marks |
Call two depth-3 words dual when they are bitwise complements. Dual pairs are exactly those whose mark-counts sum to 3 and whose marked positions are disjoint. Duality is not built into the notation by a complement operator; it is a derivable property of pairs of words. There are exactly four dual pairs:
The word 000 contains no marks. There is no presence in the word for an occupant to stand in. Its dual, 111, contains only presence. Every address, 000 included, is fully nameable — the address string is itself a complete description of its position, and the lattice labels name every position without exception. What is asymmetric is occupancy: one cannot coin an occupant for a position whose word asserts that nothing occupies it. The pure-yin ace has a name and no occupant; the pure-yang ace admits both. The prior Absent ruling is hereby reclassified from Decree to theorem of occupancy, which is a promotion in robustness — a Decree can be reversed; this cannot, without abandoning the foundation.
Among the four dual pairs, one pair (000, 111) has zero internal mixture: the pure pair. Each remaining pair is characterized by the unique position at which its minority letter sits. Reading positions in the Carried direction — top, then middle, then bottom — the pairs order themselves: pure pair first, then top-position (100, 011), then middle-position (010, 101), then bottom-position (001, 110). Associating the four suits to the four pairs, this yields heart, hourglass, star, mountain. Reorder the suits and the deviation positions no longer descend — a visible, checkable failure.
The ordering is a four-frame descent: a single mark entering at the top, passing through the middle, coming to rest at the bottom — presence arriving (Carried) until it becomes ground (Grounded).
What Paper 2 left as four open Questions of assignment, this section closes as far as ordering is concerned; which suit-name attaches to which pair remains the one residual association, addressed in §7.
Mark-count counts marks; it cannot tell the three one-mark words apart. But the marks sit on lines, and the lines have positions. Assign each line a value by where it sits in the drawing — top = 1, middle = 2, ground = 3 — and let a word’s line value be the sum of the values of its marked lines. Now the one-mark words separate: 100 marks only the top (v = 1), 010 only the middle (v = 2), 001 only the ground (v = 3). The value is counting lines, not marks.
The same table makes Grounded and Carried legible at a glance. A word is Grounded exactly when its ground line is marked — when the 3 appears in its sum. A word is Carried exactly when its top or middle line is marked — when a 1 or a 2 appears. The two stances of §3 are not new machinery; they are two questions asked of the same drawing: is there a mark where I stand? and is there a mark above me?
| Word | Address | Marked lines | Line value v | Grounded | Carried |
|---|---|---|---|---|---|
| yin·yin·yin | 000 | — | 0 | no | no |
| yang·yin·yin | 100 | top (1) | 1 | no | yes |
| yin·yang·yin | 010 | middle (2) | 2 | no | yes |
| yin·yin·yang | 001 | ground (3) | 3 | yes | no |
| yin·yang·yang | 011 | middle + ground (2 + 3) | 5 | yes | yes |
| yang·yin·yang | 101 | top + ground (1 + 3) | 4 | yes | yes |
| yang·yang·yin | 110 | top + middle (1 + 2) | 3 | no | yes |
| yang·yang·yang | 111 | top + middle + ground (1 + 2 + 3) | 6 | yes | yes |
One claim is left standing here for further research, and argument on it is welcome. The value is a sum, and a sum forgets which lines produced it: 001 and 110 share v = 3, yet one is Grounded and not Carried, the other Carried and not Grounded. The line value separates what mark-count conflates, but Grounded and Carried read the positions, never the total — Proposition 2’s lesson arriving a third time. What that collision opens, rather than closes, is the arithmetic question: addition is the operation that produced it, and addition is order-blind by definition. How the Grounded and Carried stances behave under operations that are not order-blind — multiplication and division of line values, the order of operations itself as a stance-declaration, whether a product or quotient can preserve the positional information a sum discards — is untreated here and is exactly the kind of question a reasoning system that consumes these strings will eventually force. The authors regard it as a live seam between this foundation and the larger conversation about machine reasoning, and invite the reader who sees further into it to write.
A pair of playing-card decks presents eight suit-instances but only two colors. The system is the same shape: the eight aces are four suits under two colors, and the colors are the alphabet. Deck 1/2 is the yin deck — the empty poles, the Grounded readings, addresses 000, 100, 010, 001. Deck 2/2 is the yang deck — the full poles, the Carried readings, addresses 011, 101, 110, 111. Every card’s dual is its same-suit counterpart in the other deck, ace to ace, by Proposition 3.
The decks are not symmetric distances from the root, and this must be stated plainly because the notation can be abused to hide it: the yin aces sit at mark-counts 0 and 1; the yang aces at mark-counts 2 and 3. Presence accumulates. It does not arrive by complement-flip. A construction that writes the yang aces as complements of shallow words conceals two marks behind an operator and is rejected here as associative — the honest string is the word itself.
Two decks hold 104 cards: eight suit-lines of thirteen ranks. The census supplies 8 words at depth 3, 32 at depth 5, and 64 at depth 6, and 8 + 32 + 64 = 104. Under each ace, its depth-5 continuations number four and its depth-6 continuations number eight; with the ace itself, 1 + 4 + 8 = 13 — a full suit-line, exactly. The two decks are therefore precisely the trigrams, the five-line words, and the hexagrams.
The depth-4 words — sixteen of them, two beneath each ace — appear in no deck. Their absence is systematic, not accidental: the rank arithmetic closes at 13 only because depth 4 is skipped in every line. The authors record, with a straight face, that this is a marketing plan for an expansion game, and record beneath the joke that a real derivation of the skip is pending. Until it exists, the sixteen are reserved space: no card, no coinage, no association permitted to squat there.
Everything above §5 is tier 1: mathematics. It is unownable, and its unownability is a feature — the papers can invite any reader to re-derive the census, the dual pairs, the ordering, the 104, and the standing challenge of the series depends on that invitation being real.
Tier 2 is the authored string: the specific expression, in the house’s designed signature, entered into the system of record. A theorem belongs to no one; a statement of it — this notation, this signature, this record — is a work. Pearl does not own causality; he authored the do-calculus [1]. The house does not own the free monoid on two letters; it authors the strings that occupy the entailed positions. This is the IP-bearing tier, and it is deliberately the stable one: the string survives every prose revision, every corpus audit, every re-rendering of the compendium, because it is pinned to tier 1 by derivation.
Tier 3 is the adaptation: the per-trope prose, the compendium pages, the card text in Troped. It is retirable by corpus audit, by design. It churns, and it is supposed to churn.
An earlier draft framing of this series rested the house’s claim on tier 3 — the one layer the papers themselves declare revisable. That was backwards: it placed ownership on the most perishable stratum. The correction, adopted here as architecture, moves authorship to tier 2. The consequence is stated as the paper’s central practical claim: the prose can be burned and reprinted from the positions, but the strings as authored are the work. The residual association noted in Proposition 5 — which suit-name attaches to which dual pair — lives at tier 2 as well: it is an authored choice, revisable only by the author of record, and the paper marks it as the system’s single surviving association, now confined, named, and signed rather than diffused through a table.
Four ace positions carry authored occupants as of this draft; four are open. Stated with their derivational addresses so that the coinage is checkable against the entailment:
Vast Mountain and Desolate Mountain are the re-derived names of §0: rejected when they arrived through a miscategorized sketch, re-earned here because 110 and 001 are the bottom-position dual pair — the pair where the mark has come to rest as ground — and the coinages’ meanings are entailed by the words rather than chosen for vibe. They remain subject to the Test step against actual corpus phase-traversals, as all Progenitors are. Dark Star, rejected in the earlier ruling, holds no reservation on either star position.
A reader who accepts §2 takes on a set of commitments, and the paper asks to be held to them. Trigrams are ordered words, not mark-counts (Proposition 2). The yin-root Absence stands as a theorem rather than a reversible editorial choice (Proposition 4). The suit ordering carries a visible failure condition, so any reordering must answer to it (Proposition 5). Depth 4 stays reserved — no card, no coinage — until its skip is derived (§5). The house’s claim includes both the authored strings, and the mapping of semantic identities, e.g., the prose (§6). And lattice labels such as Shattered Foundation remain tier-1 descriptors, distinct from Progenitors, which are tier-2 works — the tiers are now marked well enough that confusing them is a correctable error rather than an easy slide.
What remains open, honestly listed: the derivation of the depth-4 skip; the four pending coinages; the residual suit-name association, confined to tier 2 and the author’s signature; and the grade-vocabulary collision flagged elsewhere in the series, which this paper inherits but does not solve.
[1] Pearl, J. (1995). Causal diagrams for empirical research. Biometrika 82(4), 669–688; and Pearl, J. (2009). Causality: Models, Reasoning, and Inference, 2nd ed. Cambridge University Press. Cited for the do-calculus as the exemplar of authored expression over unownable mathematics.
[2] Adaptor House, AH-PAPER-1: derivability of the Grounded/Carried grading. adaptorhouse.com/works/papers.
[3] Adaptor House, AH-PAPER-2: the suit-assignment paper, whose open Questions this paper narrows. adaptorhouse.com/works/papers.
[4] Adaptor House lattice core (AH-CORE-2), the eight-address agreement lattice. System of record.
Citation of AI contribution: portions of this paper were derived in working session between the Architect and Fable (Claude, Anthropic), per the house's AI-integrated citation standard (AMOS).