Chapter 1: The Subtractive Foundations of the Unit Carrier

Scope: current implemented directed-cover comparison candidate. It is not the EDCM target carrier unless it integrates the required Möbius causal chain, superpositioned Structural Null, and new-gonol initiation twist. See UCNS_OPTION_DECISIONS.md.

The construction of UCNS did not begin by adding axioms. It began by removing illusions.

Early attempts to build a payload-bearing geometric number system on a directed unit carrier imported metaphors from physics and from familiar topological models: zero-point energy vaults accumulating at the origin, continuous frame inversions producing destructive interference, arbitrary bounds on causal memory, signed fibers reintroducing negation, and antipodal placements that forced the second representative across the carrier rather than along the same directed path. These imports did not clarify the object; they concealed the distinctions the carrier actually possesses and erased distinctions it retains.

The required reset therefore removed every coordinate, sign, seam, inversion, and physical interpretation that was not intrinsic to the structure itself. What remained is stricter:

• distinction determines radius;
• a directed 720° carrier projects onto a visible 360° geometry;
• carrier pairing and typed payload interaction are separate operations;
• algebraic zero is not Structural Null;
• memory is ordinary structure and contributes exactly the breadth it retains.

This chapter formalizes the current directed twofold branched angular cover candidate with a 720° lifted period and a 360° visible projection. It does not settle the entire UCNS carrier option space. For EDCM, the Möbius-origin/hidden-zero construction now has decided-constraint standing: the twist interprets space as singular and superpositioned and supplies a causal chain needed by the future hyper-dimensional lattice relating letters and words. Its exact integration with this directed candidate remains open.

1.1 The Radial Map and the Unique Null

Candidate-only null semantics. EDCM defines Structural Null as the singular superpositioned space and places the twist at the initiation of a new gonol. The coordinate-free center below documents the current executable comparison model; it is superseded as the EDCM target semantics and cannot decide the origin question.

Let $\mathcal{O}$ be the class of UCNS carrier objects. Each object $A \in \mathcal{O}$ carries a nonnegative scalar called its faithful breadth $B(A) \ge 0$, which measures the complete structural distinction retained by the object.

Carrier radius is obtained by the monotonic map

Consequently $0 \le a(A) < 1$ for every finite object. The outer rim is approached only in the limit:

The center is occupied by exactly one object, the Structural Null $\mathbf{N}$, defined by the equivalence

Null is not a container, latent field, suppressed payload, or erased receipt. It is the complete absence of distinction and therefore contains none of the following: carrier coordinate, payload, type or shape, metadata, provenance, causal receipt, recursive content, or retained relation.

Any surviving distinction requires $B(A) > 0$ and therefore $a(A) > 0$. Breadth grows outward; null alone occupies the center.

1.2 The Directed Branch Law

For every non-null radius $a > 0$, the lifted carrier has directed angular period $4\pi$ (720°), so the angular coordinate belongs to $\phi \in \mathbb{R}/4\pi\mathbb{Z}$. The visible geometry has period $2\pi$ (360°). The projection is

Every visible point away from the center therefore possesses exactly two lifted representatives:

These representatives occupy the same continuous directed path and share the same heading; the second is simply one visible lap ahead of the first. The half-period deck translation is

and $J^2$ completes the lifted return.

A 360° displacement produces none of the following automatically: negation, reflection, parity reversal, chirality reversal, frame inversion, or destructive interference. Those effects arise only when an explicit payload algebra or interaction driver defines them. Topology supplies only directed deck translation.

The branch law is therefore

At the center there is no angular circumference, hence no distinction between “here” and “one lap ahead.” The otherwise doubled carrier is intrinsically single at null.

A precise model is the lifted carrier

and the visible disk

with $p(\mathbf{N}) = 0$ and the natural projection for $a > 0$.

1.3 The Separation of Zeros

A formal system becomes unstable when it conflates distinct kinds of absence, neutrality, and cancellation. UCNS therefore maintains four non-interchangeable concepts:

A multiplicative unit may satisfy $M(A) = 1$ while remaining structurally present ($W(A) > 0$, $B(A) > 0$, $a(A) > 0$). An algebraic result may satisfy $P' = 0_{\mathcal{V}}$ while retaining type, shape, coordinate, provenance, or receipt; such an object is still non-null. A tensor may contract every index and still leave a structural receipt. Algebraic zero does not imply Structural Null.

1.4 The Parallel Valuation Triad

Within this directed-cover candidate, UCNS describes three scalar valuations $W$, $M$, and $B$. For active EDCM research, unit support is the declared $W$ baseline, while all three registered $M$ candidates and all three registered $B$ candidates must be displayed together with no selected winner. Over the manifest domain $\mathcal{O}^* = \mathcal{O} \setminus \{\mathbf{N}\}$ these valuations are canonically decoupled; they share no functional dependence or conversion law. Their only required coincidence is the common zero at Structural Null:

For every non-null object, all three are strictly positive.

Why three, and not one — the no-go

A single scalar cannot simultaneously be additive under carrier pairing and faithful to Structural Null.

Proposition (No faithful additive real breadth). Let $e$ be the $\boxtimes$-unit ($e \boxtimes A = A$) with $e \ne \mathbf{N}$. There is no map $\beta : \mathcal{O} \to \mathbb{R}_{\ge 0}$ that is both

1. additive: $\beta(A \boxtimes C) = \beta(A) + \beta(C)$, and
2. faithful: $\beta(A) = 0 \iff A = \mathbf{N}$.

Proof. The unit is idempotent: $e \boxtimes e = e$. Additivity yields $\beta(e) = \beta(e \boxtimes e) = 2\beta(e)$, hence $\beta(e) = 0$. But $e \ne \mathbf{N}$, contradicting faithfulness. The obstruction extends to any non-null idempotent or torsion element. $\blacksquare$

The proposition forces division of labor: $B$ is faithful but carries no additive law; $M$ (and its logarithm where defined) is multiplicative/additive but not faithful; $W$ aggregates support. Each escapes the no-go by declining exactly one premise.

Support weight $W$

Each potential cell $c$ has nonnegative support weight $\mu(c) \ge 0$. A cell is absent exactly when $\mu(c) = 0$. For an object with indexed cells $\{c_i : i \in I\}$,

Support is not payload value; a cell holding algebraic zero may retain positive support through coordinate, type, shape, state, provenance, or relation. Under pairing, $W(A \boxtimes C) = W(A)W(C)$.

Product character $M$

$M : \mathcal{O} \to \mathbb{R}_{\ge 0}$ satisfies $M(\mathbf{N}) = 0$, $A \ne \mathbf{N} \implies M(A) > 0$, and

Its neutral value is $M = 1$, a product-theoretic baseline, not structural absence. On the manifest domain the logarithmic character $\lambda(A) = \log M(A)$ is additive:

This additivity is partial and derived; by the no-go proposition it cannot also be faithful.

Special elements lie at the neutral baseline: any non-null idempotent $T$ satisfies $M(T) = 1$; any finite-order element likewise satisfies $M = 1$. Nilpotence under pairing forces null: if $X^{\boxtimes k} = \mathbf{N}$ then $M(X) = 0$, hence $X = \mathbf{N}$. Structural Null is absorbing: $\mathbf{N} \boxtimes A = A \boxtimes \mathbf{N} = \mathbf{N}$.

Faithful breadth $B$

$B$ measures complete surviving distinction, is null-faithful ($B(A) = 0 \iff A = \mathbf{N}$), and determines radius via $a(A) = 1 - e^{-B(A)}$. No universal additive or multiplicative law is imposed. Typed interaction may reduce, preserve, or increase distinction; $B$ reports whatever structure the complete object actually retains (payloads, carrier state, recursion, metadata, provenance, receipts, relations).

Thus the valuations hold separate jurisdictions and are welded only at their common floor.

1.5 The Rectangular-Zero Lemma

Carrier pairing $\boxtimes$ constructs the Cartesian product of supported cells. By definition the paired-cell support weight is strictly multiplicative:

A paired cell exists if and only if both parent cells exist.

Let $Z_A$ and $Z_C$ be the absent-cell index sets of $A$ and $C$. The absent cells of the paired carrier form the Cartesian union of complete rows and columns:

Pruning zero-support cells therefore commutes with pairing:

Aggregate support is multiplicative:

Deleting zero-support cells leaves this sum unchanged. The lemma rests solely on the multiplicative cell law for $\mu$; no claim about the independent product character $M$ is required.

The lemma authorizes pruning of structurally absent cells only. It does not authorize deletion merely because a cell’s payload is algebraic zero, numerically zero, neutral, nilpotent, or idempotent. Any retained structural distinction keeps the cell’s support positive.

1.6 Carrier Pairing and Typed Dispatch

For the active EDCM configuration, processing stops at carrier pairing. Typed payload dispatch remains outside the configuration pending more information. The two-stage description below is retained as broader candidate machinery, not an active EDCM rule.

UCNS interaction proceeds in two ordered stages.

Carrier pairing $A \boxtimes C$ constructs the complete Cartesian cross-pairing of supported cells. It determines which structural positions meet; it does not decide the algebraic meaning of the encounter.

Typed payload dispatch executes the interaction via a modality-sensitive driver:

UCNS exposes one public interaction surface but imposes no universal payload algebra. Different modalities may declare field operations, superposition, phase laws, syntax composition, tree construction, contraction, wedge products, geometric products, etc. Unsupported modality pairs fail closed. Topology determines where structures meet; typed algebra determines what the meeting means.

Typed dispatch may alter faithful breadth in either direction. Let $B_{\mathrm{paired}}$ be the breadth of the complete paired state before reduction and $B_{\mathrm{out}}$ the breadth of the surviving output. Then $\Delta B = B_{\mathrm{paired}} - B_{\mathrm{out}}$ carries no presupposed sign. Radius is always computed from the post-dispatch breadth: $a_{\mathrm{out}} = 1 - e^{-B_{\mathrm{out}}}$.

1.7 Memory as Geometry

A retained causal receipt $R$ is ordinary structure. If retained, it possesses positive support $W(R) > 0$ and contributes faithful breadth $B(R) > 0$, therefore contributing radius. An object retaining ten distinct receipts genuinely contains more distinction than one retaining one; its larger breadth and radius are faithful reporting, not inflation.

Receipt semantics must be declared by each interaction driver: whether a receipt is emitted, what it contains, whether duplicates or ordering or causal edges are preserved, and what erasure rules apply. The chosen representation (set, multiset, sequence, graph, tree, …) determines which distinctions are actually retained and therefore where the object sits on the carrier. An implementation that assigns identical breadth to objects with differently retained receipt structures must justify the equality by explicit canonical equivalence; otherwise the measurement is not faithful.

1.8 The Complete Collapse Rule

An interaction driver returns a complete raw output object that may contain resulting payload, causal receipt, carrier state, type and shape, recursive payloads, relations, and metadata. Collapse cannot be decided by inspecting payload value alone.

Let $\operatorname{erase}_\tau$ be the canonical erasure declared by the relevant typed driver. The canonical post-dispatch state is obtained by pruning after erasure. The output collapses to Structural Null exactly when no structural support survives:

Because all three valuations share the common null floor, this is equivalent to $W = M = B = 0$ (and therefore $a = 0$) for the canonical output.

The familiar two-part test (payload algebra declares semantic erasure and no receipt is retained) is sufficient only when the driver contract also guarantees that no coordinate, type, shape, state, relation, metadata, or recursive structure remains. Thus $P' = 0_{\mathcal{V}}$ does not imply null, nor does $R = \varnothing$. Only complete structural absence implies null. Any surviving structure—even the bare memory that an interaction occurred—keeps the object off-center and alive on the carrier.

1.9 Implementation Boundary

Within the current directed-cover comparison candidate, Chapter 1 fixes the architecture, jurisdictions, and laws of that candidate only. The EDCM target additionally requires the Möbius causal carrier, superpositioned Structural Null, a twist whenever a new gonol initiates, the exact public 157-position Unicode code-point carrier, words as the smallest gonols, and every source code point in the profile-pinned Unicode White_Space set assigned to the U+0020 origin as the superpositioned interface through which all larger gonols compose. That carrier assignment retains the exact source value, code point, and offset and performs no normalization, rewriting, trimming, or collapse. EDCM also requires ordered concatenation, one unit of support per complete speaker turn, full-corpus execution, exact-evidence baseline, carrier pairing only, plural M and B displays, and an EDCM-specific profile. It determines that faithful breadth must satisfy $B(A) \ge 0$, $B(A) = 0 \iff A = \mathbf{N}$, and $a(A) = 1 - e^{-B(A)}$, and that $B$ must report every distinction retained by payloads, carrier state, recursion, metadata, provenance, and receipts. It fixes the zero-test and jurisdiction of cell support $\mu$, the aggregate support law of $W$, the paired-support axiom of $\boxtimes$, the multiplicative codomain and laws of $M$, the directed branch topology, the separation of pairing and dispatch, and the complete collapse rule.

The separate v0.18 research boundary can apply the surviving exact signed-local circle candidate to an explicitly initiated word when independent exact rational coordinates are supplied. That bounded evidence neither selects the Chapter 1 breadth laws nor derives a coordinate from source identity. The v0.19 separately derives one exact trace-local midpoint candidate from ordered source address. Candidate selection, cross-scope and higher-gonol composition, total Structural Null topology, higher geometry, and completion remain outside this chapter.

Canonical evaluators or primitive assignments for $\mu$, $M$, and $B$ remain formal implementation obligations. A conforming implementation must:

1. assign support weights $\mu$ and demonstrate the zero-test;
2. construct a nontrivial product character $M$ that is multiplicative, has unique zero at null, and is provably not identical to $W$;
3. build the canonical faithful-breadth evaluator $B$ that reports all retained distinction and passes invariance tests;
4. define canonical structural equivalence so that encoding artifacts do not manufacture breadth;
5. verify the null, pairing, pruning, dispatch, receipt, and radius laws.

Separation of $W$ and $M$. Both valuations are multiplicative under $\boxtimes$ and therefore share the same formal shape. Different names alone do not prove they measure distinct properties. A conforming implementation must supply concrete witnesses: non-null objects $A, C$ with $W(A) = W(C)$ yet $M(A) \ne M(C)$, and non-null objects $D, E$ with $M(D) = M(E)$ yet $W(D) \ne W(E)$. Until such witnesses and invariance tests exist, the triad is three-by-specification but only provably two.

The EDCM observation profile now makes source-preserved carrier assignment, word, SPACE-origin boundary, speaker-turn support, and full-corpus execution testable. It records but does not geometrically implement the initiating Möbius twist. The next work is the formal Möbius carrier, the higher-gonol composition law, construction of the measuring instruments, and comparison of complete configurations against complete EDCM corpora.

Offstage Excavation

This geometry reads cleanly because nine turns of subtraction occurred offstage. A reader encountering it for the first time may think, “Obviously a 360° displacement on a directed path does not invert the frame—why would it?” or “Obviously a retained receipt is ordinary structure—where else would it go?”

That apparent obviousness is the result of active removal. The completed surface does not display the gravitational pull of the metaphors that had to be resisted: the seam that tried to localize the twist, the signed fiber that reintroduced negation, the antipode that placed the second representative across rather than ahead, the frame inversion that forced cancellation, the zero-point vault that hid structure at null, the inverted radius that placed infinity at the center, the bounded receipt that made breadth lie about memory, and the additive breadth that had to be broken into three valuations because faithfulness and additivity cannot inhabit one real scalar.

Those models were not removed for lack of imaginative power. They were removed because they imposed distinctions the carrier does not possess or erased distinctions it retains. The subtractive process is invisible in the final axiomatic structure; it is nevertheless the reason the structure can now carry weight. This candidate omits the twist that supplies EDCM’s origin and causal chain. That omission is disqualifying for the target EDCM configuration unless formal integration shows that the directed cover preserves the required Möbius and superposition semantics.

hmmm — this chapter now survives as comparison evidence, not as the EDCM answer. Selection and cross-scope composition beyond the v0.19 trace-local coordinate candidate, total superposition topology, graph contribution laws, projection scale, higher composition, completion, and real-system failure corpora are exactly the places where honest traffic should make the current construction look incomplete.