THE MINIMAL-SHAPE SUITE (AI EDITION)

The Rigorous, Honest Case That the Frozen 13D Geometry Is Selector-Minimal

Document class: Selector-minimality case file / no-overclaim instrument / read-only synthesis. Date: 2026-06-24. (Note: several cited anchors — REVIEW_SHAPE_SUITE_2026-06-23.md, THREE_ROOT_CLOSURE_PATHS_2026-06-23.md — are dated 2026-06-23; the one-day lead is harmless and simply reflects a synthesis built the day after its sources.) Scope: The Local Structural Form root (R2 / "Shape") — the frozen 13D active branch $\mathcal{B}_{\rm active} = [M_4 \times K_6 \times S^2 \times S^1_Y/\mathbb{Z}_2]_{(\times)} \oplus [F^+ \oplus C_{\rm admiss}]_{(\oplus)} \otimes [E_{\rm matter}\oplus E_{\rm gauge}\oplus E_{\rm Higgs}\oplus E_{\rm proton}]_{(\otimes)}.$ No status was ever upgraded. This document derives nothing, closes no gate, certifies no new number, and mutates no frozen object. It argues a status, and it argues it conservatively. Frozen branch dcc66f1b2685 / a5b1e6f9d951 — READ-ONLY, confirmed unmutated.

0. The one sentence this document refuses to write

The crank version of this document would say:

"We proved the geometry is THE unique minimum — the geometry is forced."

This document will not say that, anywhere, in any disguise, because it is false. The honest claim is strictly weaker, and the gap between the two is the entire intellectual content of the suite:

Inside a declared, frozen search category, under a pre-declared Occam funnel, the frozen 13D active branch is the lexicographically-minimal complete survivor; no Tier-1 preferred competitor currently beats it; and any architecture that meets the physical burden must carry at least three functional roles. None of this derives the geometry. It selects it. Selection is not derivation. And the selection itself bottoms out on $E$ — the Standard-Model chiral content — which no known principle forces.

That is the thesis. The rest is the careful defense of each clause, the honest fencing of what is EARNED, and the plain admission of what is NOT EARNED. The two-branch fork at the end (Shape stands OR Shape folds) is left genuinely open, because closing it falsely would be the exact overclaim this document exists to prevent.

This document is written to survive blind adversarial AI review — i.e., a reviewer who does not trust the authors, who will grep for every "QED" and ask what it discharges, who will hunt for the tuning to the known answer smuggle, and who will reject any universal negative dressed up as a theorem. Where the suite that this document synthesizes has already been adversarially reviewed (REVIEW_SHAPE_SUITE_2026-06-23.md), the confirmed defects are reproduced here verbatim and folded into the NOT-EARNED ledger, not hidden.

1. What "minimal shape" honestly means here

1.1 Two completely different claims that share a word

The word "minimal" is doing dangerous double duty in physics-foundations writing. There are (at least) two claims that both get called "the geometry is minimal," and they are not the same claim, not even close:

The entire discipline of this document is to keep these two apart, at every step, including in places where collapsing them would be rhetorically convenient.

1.2 Selection is not derivation — stated as a hard rule, not a hedge

A selection ranks options that you already wrote down. A derivation produces an option you did not have to write down, from premises you had independent reason to accept. The corpus encodes the difference as a rejected move:

Funnel-winner-forced = RELABEL = REJECTED. ("selection ≠ derivation," on-disk at …/TOE/AXIOM_LEDGER.md line 126; "Funnel-winner-forced = RELABEL = REJECTED" at …/TOE/AXIOM_LEDGER.md line 186; cross-referenced in THREE_ROOT_CLOSURE_PATHS_2026-06-23.md §3.3 item (iii).)

The reason this matters operationally: an Occam funnel that returns a unique winner is psychologically indistinguishable from a derivation. You run the machine, exactly one branch survives, and it is tempting to write "therefore the branch is forced." But uniqueness of the survivor is uniqueness relative to the category and the ranking you supplied. Change the category, and the survivor can change. A derivation has no such dependency: it survives any enlargement of the universe of discourse, because it is an implication, not a search outcome.

So the operative rule for this whole document is:

A unique funnel survivor is reported as "selector-minimal," never as "derived" or "forced." The moment a sentence reads "therefore the geometry must be ...," it is wrong unless the "must" is explicitly relativized to the declared category.

The selector theorem (T_SHAPE_SELECTOR_MINIMALITY_CERTIFICATE_THEOREM.md §0) opens by refusing the strong claim:

"We should not try to prove $\mathrm{Scale}+\mathrm{Granularity} \Rightarrow \mathrm{Shape}$. That is too strong and is not currently supported."

This document inherits that refusal as a load-bearing commitment, not as throat-clearing.

1.3 "Category-relative" is the honest ceiling, and it is a real ceiling

A category-relative result is a real result. It is not a non-result, and it is not vibes. It is a precise, falsifiable statement of the form:

Given (i) the declared search category $\mathfrak{B}_{\rm search}$, (ii) the constraint set $\mathcal{C}_{\rm GUT}$, (iii) the lex-min Occam order $\mathfrak{R}_{\rm Occam}$ with completeness binding, and (iv) the freeze/no-smuggling discipline, the frozen branch is $\operatorname*{argmin}_{B \in \mathrm{Adm}(\mathfrak{B}_{\rm search})} \mathfrak{R}_{\rm Occam}(B)$.

The honesty consists in never silently dropping the "Given (i)–(iv)." The GUT manuscript states the ceiling as its own binding sentence (Appendix B2, quoted in GUT.md line 6198):

"This theorem is category-relative. It does not prove that no other mathematical architecture could exist outside the declared search category. Reviewers who reject the search category should attack Section 2.7 and Appendix B1 (the selector / Occam definition), not this theorem."

That sentence is the model for the entire posture: it tells the hostile reviewer exactly where to attack, which is the opposite of a smuggle. A document that names its own load-bearing assumption and invites attack on it is behaving like a theorem, not like an advertisement.

1.4 The difference between selecting a winner and deriving it — a worked contrast

Consider two statements about $K_6 = SU(3)/T^2$:

• (Selection) "Among the admissible internal geometries we enumerated, $K_6$ is the cheapest one whose index returns exactly three families as a topological integer; the cheaper $\mathbb{CP}^2$ fails completeness because its family count is tunable." — This is true and EARNED. It is a statement about a search outcome over a declared menu.
• (Derivation) "$K_6$ is forced: any consistent theory of the observed world must compactify on $SU(3)/T^2$." — This is false and NOT claimed. It quantifies over all consistent theories, a class the suite explicitly does not control (spectral triples, string compactifications, finite-state models, etc., are live competitors per §5).

The two sentences differ by exactly one thing: the scope of "all." The first scopes "all" to the declared menu; the second scopes it to all of physics. The first is a certificate; the second is an overclaim. Keeping that distinction visible at every node is what "honest minimality" means in this document.

2. The search category and the Occam funnel rule

2.1 Why a constraint-first search even makes minimality meaningful

The geometry is not produced by scanning all compact manifolds and admiring one. It is produced by a constraint-first funnel (GUT.md §3.2, line 648):

"Instead of scanning all possible compact geometries, we start from the structural facts of the Standard Model and ask which candidate branches survive every required closure gate. Branches are filtered by gauge recovery, charge consistency, chirality, anomaly cancellation, stabilization, threshold unification, Higgs protection, proton safety, and flavor closure. Surviving branches are then minimized by Occam's razor."

This ordering is essential to the honesty of the claim. Minimality is applied last, and only to branches that already passed every gate. That is what stops "minimal" from degenerating into "cheapest thing that does almost nothing." The empty geometry (bare $\mathcal{M}_{3,1}$) is the cheapest object in the universe and it passes the anomaly ledger vacuously (GUT.md line 2800) — and it is eliminated, because it fails gauge recovery. Cheapness without completeness is disqualifying, by rule.

2.2 The declared search category — stated, not assumed

The search category $\mathfrak{B}_{\rm search}$ is the declared outer boundary of every statement (GUT.md Appendix B1, line 5858):

"The declared search category is the outer boundary of every statement in this appendix: every constraint, selector step, Occam ordering, and freeze entry is evaluated inside this category. Scope-reduction during selection (silently shrinking the category to make a candidate look minimal) is forbidden by B.8."

Two features of this are load-bearing for adversarial review:

1. It is declared in advance. A category drawn after seeing which branch you wanted is a fit. A category declared before the comparison is a hypothesis. The freeze rule (§4.7 of GUT.md) is what converts "we drew the category honestly" from an assertion into a checkable property: the category, the selector, the Occam order, and the eliminated-branch ledger are all frozen objects with hashes.
2. Shrinking it mid-selection is explicitly forbidden (B.8). The classic smuggle — "minimize over everything, then quietly delete the competitor that beat you" — is named and banned. This is exactly the move a hostile reviewer looks for, and the corpus pre-empts it with a named rule.

The admissible set is then: $$\mathrm{Adm}(\mathfrak{B}_{\rm search}) = \{ B \in \mathfrak{B}_{\rm search} : B \models \mathcal{C}_{\rm GUT},\ B \text{ obeys freeze-before-compare},\ B \text{ obeys no-smuggling} \}.$$

2.3 The constraint vector $\mathcal{C}_{\rm GUT}$ (Gates 1–10)

The ten constraints (GUT.md §2.2 of the selector theorem; GUT.md gate cards) are:

1. geometry specification;
2. Standard Model gauge recovery;
3. hypercharge and electric charge recovery;
4. chirality / no mirrors / three families;
5. anomaly cancellation;
6. stabilization of used compact moduli;
7. threshold unification;
8. Higgs protection;
9. flavor closure;
10. proton safety.

Gate 11 (claim-boundary discipline) is treated as a governance guard, not a shape-output constraint. This is an honesty point worth flagging in advance: the selector theorem keeps Gate 11 out of the shape-output vector, which is correct — but the architecture-neutral constraint set $\mathcal{C}_{\rm phys}$ used later (§4) re-imports governance items (C11/C12), and that re-import is one of the two confirmed MAJOR defects of the suite (see §4.4 and §6). The two constraint vectors must not be conflated.

2.4 The Occam funnel rule: completeness is BINDING; minimality only ranks COMPLETE branches

This is the heart of §2 and the single most important rule for keeping the claim honest. The Occam razor $\mathcal{R}$ (GUT.md §4.6, lines 1485–1505) is:

Plain. "Occam's razor here is a demolition rule with one absolute restriction: you may remove a wall only if the building still stands. The machine does not select the cheapest theory; it selects the cheapest complete theory — and 'complete' is decided by the gates, not by taste."

The precise priority order:

1. Do not remove any structure required for a gate. Completeness is binding.
2. Prefer lower-dimensional / lower-complexity geometry whenever the gates remain closed.
3. Prefer fewer declared free inputs.
4. Prefer fewer independent operator choices.
5. Prefer one mechanism per closure role; reject double-closures.
6. Remove factors whose only role is decorative.
7. Reject any extension whose extra structure does not improve gate closure.

In compressed form: $$\boxed{\text{completeness} \;>\; \text{minimality}, \qquad \text{minimality applies only among complete branches}.}$$

The funnel rule does two opposite jobs with one priority, and the suite shows both directions, which is what makes it a genuine rule rather than a one-off rationalization:

• Direction A (retain $F^+$). The flavor chamber $F^+$ is simpler to drop than to keep. Rule 2 alone would drop it. But dropping it reopens the flavor gate (Gate 9), so rule 1 forbids the drop. $F^+$ is retained by the razor's own rule, not over the razor's objection (GUT.md line 1505; line 5936). The cheaper theory is incomplete, so it loses.
• Direction B (retain $K_6$ over $\mathbb{CP}^2$). Run the same priority the other way: rule 2 prefers the cheaper 4-D $\mathbb{CP}^2$ over 6-D $K_6$. But $\mathbb{CP}^2$'s family count is a tunable bundle choice, which is incomplete under the chirality gate (Gate 4) once the anti-fitting rule converts "tunable" to "fail." So rule 1 forbids the cheaper option, and $K_6$ is retained.

The fact that the same binding rule explains both a retention of extra structure ($F^+$) and a payment of extra dimensions ($K_6$) is the strongest available evidence that completeness-over-minimality is a real prior constraint and not a criterion inserted to reach a target. We work Direction B in full in §3, because it is the case a hostile reviewer will single out as the smuggling surface.

2.5 Why the funnel rule is the anti-smuggle, not the smuggle

A blind reviewer's first suspicion: "completeness > minimality is just a license to keep whatever you want by declaring it 'required for a gate.'" The defense is that "required for a gate" is not a taste judgment — it is a binary certificate outcome. A structure is required iff removing it causes a named gate's certificate to flip to fail, and that flip is reproducible by a frozen pipeline (GUT.md §4.8, certificate fields include an explicit Failure condition and a negative control). So:

• "$F^+$ is required" is checkable: delete $F^+$, re-run, watch Gate 9 fail.
• "$K_6$ is required over $\mathbb{CP}^2$" is checkable: the family count on $\mathbb{CP}^2$ is a continuous modulus, so Gate 4 has no forced integer to certify — the certificate cannot pass without a hand-chosen value, which the anti-fitting rule (no post-hoc tuning) bars.

The smuggle would be a taste-based "required." The corpus uses a certificate-based "required." That is the difference between a razor and a rationalization, and it is the difference a blind reviewer is paid to find. Here it survives.

3. The $\mathbb{CP}^2$ elimination, worked in full

This section is the crux of the adversarial defense, because it is the one place where the selector pays more (two extra dimensions) rather than less, and a hostile reviewer will read that as "they reverse-engineered a 'forcedness' bonus to reach the three-generation target $E$." We work it completely and show it is the legitimate, uniform application of completeness > minimality, identical in kind to the $F^+$ retention — not a target-aligned special pleading.

3.1 The candidates, side by side

From GUT.md §3.5 (line 1201) and Appendix GS (line 2727):

Both carry $SU(3)$ exactly. Both clear gauge recovery (Gate 2). They diverge at Gate 4 (chirality / three families), and only there.

3.2 The decisive criterion: "tunable counts as fail" is the anti-fitting rule, declared independently

The verdict (GUT.md line 2727):

"$\mathbb{CP}^2$ ... Spin-$\mathbb{C}$ only (not spin) — admissible; but the three-family count becomes a continuous bundle-moduli choice rather than a forced integer. Eliminated — family count not topologically forced (the anti-fitting rule converts 'tunable' into 'fail')."

The load-bearing move is: a quantity that can be dialed to the observed value is not a pass; it is a fit. The anti-fitting / freeze-before-compare discipline (GUT.md §4.7) is declared as a general rule of the whole programme — it is the same rule that makes every Yukawa, phase, and threshold a frozen declared input rather than a post-hoc adjustment. It is not a rule invented at the $\mathbb{CP}^2$ node. So when $\mathbb{CP}^2$ offers the right family count only by choosing a bundle modulus, it is offering a fit, and the pre-existing anti-fitting rule rejects fits.

3.3 Why this is completeness > minimality, not a "forcedness bonus"

Here is the precise logical structure, stated so a reviewer can check there is no extra criterion:

1. Rule 2 (minimality) strictly prefers $\mathbb{CP}^2$ (fewer dimensions). No dispute.
2. Rule 1 (completeness) asks: does $\mathbb{CP}^2$ close Gate 4 under the frozen, no-fitting pipeline? Answer: no — because the only way to get three families out of $\mathbb{CP}^2$ is to choose a continuous modulus after the fact, which the anti-fitting rule forbids. So $\mathbb{CP}^2 \notin \mathrm{Adm}(\mathfrak{B}_{\rm search})$: it is incomplete, not merely "less forced."
3. Therefore minimality (rule 2) never gets to rank $\mathbb{CP}^2$ at all, because rule 1 already removed it from the complete set. $K_6$ is the cheapest complete survivor.

The phrase "the selector pays two extra dimensions for forcedness" (GUT.md line 1201) is a plain-language gloss on this — and it is exactly the phrase a hostile reviewer flags. The honest reading, which the manuscript itself now states (GUT.md §4.13, line 1625), is:

"The §3.5 retention of 6-D $K_6$ over the cheaper 4-D $\mathbb{CP}^2 = SU(3)/U(2)$ is not a special 'forcedness' preference reverse-engineered to pick $K_6$; it is a clean instance of that pre-declared rule — $\mathbb{CP}^2$'s three-family count is a tunable bundle choice, barred by the anti-fitting discipline, hence incomplete under the chirality gate — the same completeness-binding rule that retains the $F^+$ chamber."

So "forcedness" is not a new criterion being maximized; it is a renaming of "passes Gate 4 without a fit." The selector is not rewarding $K_6$ for being forced; it is disqualifying $\mathbb{CP}^2$ for being incomplete (it cannot pass Gate 4 without a banned fit). These are the same operation as the $F^+$ case: the cheaper option is incomplete, so the dearer complete option wins.

3.4 The adversarial objection stated at full strength, and the honest answer

Objection (blind reviewer). "The anti-fitting rule is doing all the work, and you only invoke it where it happens to eliminate the competitor that would have given you a different number of families. The target is 3 generations ($E$). 'Tunable = fail' is a criterion you reach for precisely when a competitor's freedom would let it miss the target. That is tuning to the known answer."

This is the strongest form of the objection and it must be answered without flinching. The answer has three parts, and the reviewer should weigh whether they fully discharge it:

1. Uniformity test (passes). The anti-fitting rule is applied everywhere, not just at $\mathbb{CP}^2$. Every continuous post-hoc adjustment in the entire programme is barred by the same rule (the freeze manifest R1, hash a5b1e6f9d951, locks Yukawas, phases, thresholds — GUT.md line 1523). If "tunable = fail" were a $\mathbb{CP}^2$-only criterion, that would be a smuggle. It is not; it is the programme's universal anti-fit discipline. This part of the objection is answered as to smuggling (it does not show the criterion is $\mathbb{CP}^2$-special; the residual in part (3) remains).
2. Direction test (passes). The same rule retains extra structure ($F^+$) and pays extra dimensions ($K_6$). A target-fitted criterion would only ever cut toward the target; this one cuts in both directions (it keeps an expensive chamber the razor would drop, and it pays for an expensive manifold the razor would skip), governed by a single priority. A criterion that sometimes costs you simplicity to preserve completeness is behaving like a constraint, not like a fit. This part of the objection is answered as to smuggling (the criterion is not a target-only cut; the residual in part (3) remains).
3. Residual honesty (the part that does NOT fully defeat the objection). The objection has a true residue that no amount of uniformity-checking removes: the criterion does bottom on $E$. "Three families, forced" is only a requirement because we are matching the observed three families — i.e., because $E$ (the SM chiral content) is a fixed input. The anti-fitting rule is applied uniformly, yes — but the reason Gate 4 demands a forced integer rather than, say, a forced two, is that the world has three generations, and that fact is an input ($E$), not a derived output. So the $\mathbb{CP}^2$ elimination is a clean, uniform, completeness-over-minimality call given $E$ — and it is not a derivation of why three. The honest verdict is: the elimination is legitimate as a selection step; it is not, and is not claimed to be, a derivation of the family number. (This is exactly the $E$-residual that §7 makes the centerpiece.)

So the $\mathbb{CP}^2$ elimination is EARNED as a selection move and explicitly NOT a derivation of $E$. That dual statement is the honest disposition, and it is the disposition the manuscript itself now carries (§4.13).

3.5 A second uniformity witness: $S^3$ vs $S^2$, $T^2/\mathbb{Z}_n$ vs $S^1/\mathbb{Z}_2$

To show the funnel is a machine and not a story told once, note the other eliminations run by the same razor in the same appendix (GUT.md lines 2703, 2716):

• $S^3 \cong SU(2)$ eliminated in favor of $S^2$: "one extra dimension, one extra $SU(2)$, no gate served" — rule 2 + rule 6.
• $T^2/\mathbb{Z}_n$ eliminated in favor of $S^1/\mathbb{Z}_2$: "everything it offers is already supplied at lower dimension and moduli cost" — rule 5 (redundant mechanism).

These are minimality cutting toward cheapness — the ordinary direction. The $\mathbb{CP}^2 \to K_6$ case is the extraordinary direction (paying more), and it is governed by the same priority list. A reviewer who accepts the $S^3 \to S^2$ cut as legitimate Occam must, for consistency, accept that the $\mathbb{CP}^2 \to K_6$ cut is the same machine with rule 1 active. The only thing one may still contest is the input $E$ that makes Gate 4 demand a forced integer — and contesting that is contesting the residual, which the document concedes is unforced (§7).

4. The functional-role FLOOR ($k_{\rm role} \ge 3$)

4.1 What the floor is, and what kind of claim it is

The selector-minimality result of §2–3 is category-relative: it lives inside $\mathfrak{B}_{\rm search}$. The functional-role floor attempts something architecture-neutral: a statement about any admissible physics, not just product-manifold geometries. The claim (T_SHAPE_FUNCTIONAL_ROLE_NECESSITY_THEOREM.md):

$$\mathcal{C}_{\rm phys} \Rightarrow \mathrm{Stage} + \mathrm{Rulebook} + \mathrm{Actors}.$$

In words: any architecture $B$ that meets the architecture-neutral physical burden $\mathcal{C}_{\rm phys}$ must contain functional equivalents of three roles —

The competitor need not write $\times/\oplus/\otimes$; it must carry the functions those symbols name. A finite-state model, a spectral triple, a string compactification — each must, the theorem argues, contain a stage-like, a rulebook-like, and an actor-like component, on pain of failing some clause of $\mathcal{C}_{\rm phys}$.

4.2 What this earns: a NECESSARY floor, NOT a sufficient bridge

This is the cleanly EARNED part: $k_{\rm role} \ge 3$ is necessary for any admissible architecture. The proof (theorem §6) is a three-fold contradiction:

• assume $\mathrm{Stage}(B) = \varnothing$ ⇒ no carrier for the 4D sector / gauge / index ⇒ $B \not\models \mathcal{C}_{\rm phys}$;
• assume $\mathrm{Rulebook}(B) = \varnothing$ ⇒ no admissibility / anomaly-selection / freeze ⇒ $B \not\models \mathcal{C}_{\rm phys}$;
• assume $\mathrm{Actors}(B) = \varnothing$ ⇒ no matter/gauge/Higgs/proton DoF ⇒ $B \not\models \mathcal{C}_{\rm phys}$.

The submitted branch has $k_{\rm role}(\mathcal{B}_{\rm active}) = 3$, hitting the floor exactly.

Crucially, this is necessary, not sufficient. It says every admissible competitor has at least three roles; it does not say the submitted realization of those roles is the cheapest (that is the open realization-minimality theorem, §6). The "Unfold" device (theorem §4) blocks the cheap smuggle "I removed the rulebook" by extracting the role even when it is fused into an action, a measure, or a selection functor. So the floor is robust against the "my theory has fewer pieces" trick: fewer pieces is not fewer roles.

4.3 The honest caveat #1: role-necessity is near-circular if roles restate the physics requirements

This is a CONFIRMED MAJOR defect (review §3 defect 1) and it must be stated plainly, not buried. The role definitions (theorem §3.3 Stage, §3.4 Rulebook, §3.5 Actors) are written as restatements of clauses of $\mathcal{C}_{\rm phys}$ (§3.2). The proof then "derives" each role's non-emptiness by assuming it empty ⇒ a $\mathcal{C}_{\rm phys}$ clause fails ⇒ contradiction. But that contradiction holds only because the role was defined to be those clauses. The theorem is, in skeleton, $P \Rightarrow P$.

Concretely: "Stage" is defined to be "the thing that supports the observed 4D sector, gauge carriers, index data, low-energy comparison" — which is just a list of $\mathcal{C}_{\rm phys}$ clauses. So "$\mathcal{C}_{\rm phys}$ ⇒ Stage $\ne \varnothing$" is "these clauses ⇒ a name for these clauses." That is true, but it is near-tautological.

What survives the caveat: the floor count $k_{\rm role} \ge 3$ is still meaningful, because the partition into three (rather than two, or one) carries a small amount of non-tautological content: it asserts that the burden cannot be discharged by fewer than three functionally distinct clusters (you cannot fuse "where things live," "what is allowed," and "what exists" into two without one secretly carrying two roles — the Unfold argument). That partition claim is weak but not empty. What does NOT survive: any use of role-necessity as "the gateway to absolute irreducibility" (the fork theorem's phrase). A near-tautology cannot be a gateway to a substantive universal. So role-necessity earns the floor and nothing past it.

4.4 The honest caveat #2: $\mathcal{C}_{\rm phys}$ is NOT truly architecture-neutral (C11/C12 are governance, not physics)

This is the second CONFIRMED MAJOR defect (review §3 defect 2). The architecture-neutral constraint set $\mathcal{C}_{\rm phys}$ (theorem §3.2) lists twelve items. Items C11 (freeze-before-compare / no tuning to the known answer) and C12 (reproducibility / certificate-status discipline) are methodological commitments of this programme, not physical constraints any admissible theory must satisfy.

A pure Standard Model plus measured constants (audit-matrix row 16) is a physically admissible description of the world. It does not carry "freeze-before-compare" as an internal property — freeze-before-compare is a property of how we argue about a theory, not of the theory itself. By baking the programme's own audit discipline into the supposedly neutral burden, the constraint vector guarantees that only corpus-shaped architectures pass. A genuinely architecture-neutral $\mathcal{C}_{\rm phys}$ would contain only C1–C10 (the physics) and would route C11–C12 to the meta-level (how any of these theories must be presented to be auditable), not the object level.

Disposition: the functional-role floor should be read as earned over $\mathcal{C}_{\rm phys}^{\rm physics} = \{C1, \dots, C10\}$ only. Over the full twelve-item set it is partly an artifact of including governance. The EARNED claim is therefore: necessary role-floor $k_{\rm role} \ge 3$ for any architecture meeting the physical (C1–C10) burden. The "architecture-neutral" label is downgraded to "architecture-neutral over the physics clauses; the governance clauses C11/C12 are programme methodology and are excluded from the neutral burden."

4.5 Net on the floor

The floor is real but modest: it tells you the geometry is not over-built in role count (three roles, three present), and that no competitor can undercut it by having fewer than three roles. It tells you nothing about whether the fillings of those roles are minimal. That is the honest extent.

5. The competitor audit (current-record, not nonexistence)

5.1 The structure of the claim, and its strict limit

The strongest negative a selection can carry is "no preferred competitor currently survives." This is failure-to-find over an audited list — emphatically not a nonexistence proof. The audit matrix (SHAPE_COMPETITOR_AUDIT_MATRIX.md) enumerates 18 competitor classes; the Tier-1 theorem (T_SHAPE_TIER1_COMPETITOR_ELIMINATION_THEOREM.md) hardens the five most dangerous. The result is labeled, correctly and repeatedly, "QED-current-record" and "current-record elimination, not nonexistence."

The beat condition a competitor $B'$ must satisfy to defeat the shape (Tier-1 theorem §2): (BC1) $B' \models \mathcal{C}_{\rm phys}$; (BC2) supplies a frozen certificate package (declared inputs, anti-fitting firewall, no hidden knobs, pass/fail, reproducibility); (BC3) is strictly simpler after unfolding into Stage/Rulebook/Actors; (BC4) no smuggling. A competitor that is "a known broad framework" but supplies no frozen, reproducible, lower-burden certificate does not meet the bar — not because it is impossible, but because no one has supplied the package.

5.2 The five Tier-1 rival classes and why none currently survives

5.3 The honest weakness of the audit (CONFIRMED instrument gap)

A blind reviewer must be told the audit's real limitation (review §3 defects 6–7), and this document states it without softening:

• The "simpler after unfolding?" column is mostly "Unknown." For the five Tier-1 classes, the decisive cell — is it actually simpler after unfolding? — reads "Unknown / Usually not / Likely no." The most dangerous classes (NCG, KK, finite-state) are precisely the un-adjudicated ones. So the elimination is "no one has handed us a winning package," not "we checked and they lose." That is a weaker statement, and it is the true one.
• Neither $\models \mathcal{C}_{\rm phys}$ nor $\prec_{\rm abs}$ is operationalized as a finite test. The complexity vector $\mathfrak{K}$ (the realization-minimality theorem §3) has no units and no aggregation rule, so the lex order $\preceq_{\rm abs}$ is not yet a well-defined total order. "Simpler after unfolding" is therefore not yet mechanically decidable. The audit is a structured TODO with priorities, not a finished falsifiable instrument.
• Unfold is conceptual, not algorithmic. There is no procedure mapping an arbitrary architecture to its $(\mathrm{Stage}, \mathrm{Rulebook}, \mathrm{Actors}, \mathrm{Residue})$ tuple. Combined with falsifier F2 ("a competitor that appears role-free is secretly carrying the role"), role-necessity becomes hard to falsify in practice — any apparent counterexample can be reclassified as "secretly carrying" a hidden role. A reviewer is right to note this asymmetry, and it caps how much the audit can claim.

5.4 What the audit nonetheless EARNS

Despite the gaps, the audit earns a real, bounded result:

EARNED: No Tier-1 preferred competitor currently survives. The word "currently" is never upgraded to "ever." The reopen rule is live and explicit (each card carries a reopen condition; the matrix is a standing program).

And it earns something procedurally valuable even where it does not earn a verdict: it defines exactly what a competitor must do to win (BC1–BC4 + a filled scorecard). That converts a vague "no one has beaten us" into a precise falsification target. A hostile reviewer who wants to refute R2 now knows the precise package to supply — which is the correct posture for a falsifiable claim, and is itself a (modest) point in the suite's favor.

5.5 The reopen rule, stated as a live commitment

The reopen condition (selector theorem F2; Tier-1 theorem §8; GUT.md §4.13):

Any $\mathcal{B}'$ in the category that (i) satisfies the physical burden $\mathcal{C}_{\rm phys}$, and (ii) is strictly simpler after unfolding into Stage / Rulebook / Actors, with (iii) a frozen, reproducible, tuning-free certificate, immediately reopens R2 realization-minimality and may fold the shape.

This is not decorative. It is the operational meaning of "current-record." A claim that can be reopened by a named, suppliable object is a falsifiable claim. A claim that cannot is dogma. R2's negative is the former.

6. What is NOT EARNED

This section is the spine of the document's honesty. Two things are not earned, and a printed "QED" appears in the suite over one of them; both are stated plainly.

6.1 Realization-minimality is NOT EARNED (5 open sub-lemmas; the "QED" is conditional only)

The realization-minimality theorem (T_SHAPE_REALIZATION_MINIMALITY_THEOREM.md) aims to upgrade the role floor (§4) to realization minimality: not just "three roles are necessary," but "the submitted fillings of those three roles are the cheapest possible." The conditional theorem: $$\forall B \in \mathfrak{B}_{\rm abs},\ B \models \mathcal{C}_{\rm phys} \Rightarrow \mathrm{Unfold}(\mathcal{B}_{\rm active}) \preceq_{\rm abs} \mathrm{Unfold}(B).$$

This is NOT EARNED. All five load-bearing sub-lemmas are OPEN (theorem §7):

The "QED" problem, stated for the reviewer. Theorem §6 prints "QED" — but the proof body only chains the five lemmas: it shows $(\mathrm{L1} \wedge \mathrm{L2} \wedge \mathrm{L3} \wedge \mathrm{L4} \wedge \mathrm{L5}) \Rightarrow \text{argmin}$. Since all five are open, the "QED" discharges only the logically empty implication $(A_1 \wedge \dots \wedge A_5) \Rightarrow (\text{conclusion that is just } A_1 \wedge \dots \wedge A_5 \text{ re-assembled})$. It adds zero new deductive content. The conditionality is disclosed (the header reads "NOT FULLY PROVEN"; the corpus review forced the line to read "QED (conditional on Lemmas 1–5 of §7, all currently OPEN)"), so this is not fabrication — but a reader who lifts §6 in isolation would see a clean "QED" over an open result. The honest label: realization-minimality is a valid conditional skeleton, not a result. Any "QED" here is conditional-only.

Compounding defect: the complexity vector $\mathfrak{K}$ has no units or aggregation rule, so $\preceq_{\rm abs}$ (lexicographic over an unordered tuple) is not a well-defined total order. Argmin uniqueness under an ill-defined order is asserted, not established. Until $\mathfrak{K}$ and $\preceq_{\rm abs}$ are operationalized, even the statement of realization-minimality is not fully well-posed, let alone proven.

6.2 Absolute irreducibility is NOT EARNED — the genuine two-branch fork

The fork theorem (T_SHAPE_ABSOLUTE_IRREDUCIBILITY_FORK_THEOREM.md) is a dedicated no-overclaim instrument. Its main result is, deliberately, a limitation:

Theorem (Absolute Shape Irreducibility Is Not Yet Claimable). Under the current record, R2 may be claimed as category-relative selector-minimal but may not be claimed as absolutely irreducible.

The proof is honest and short: every result in the stack is scoped to a declared category (B2 is category-relative by its own binding sentence; the axiom ledger records Shape as "irreducible-under-known-reductions," not "provably irreducible"). Absolute irreducibility would require a universal competitor class $\mathfrak{B}_{\rm abs}$, an architecture-neutral preorder $\preceq_{\rm abs}$, a no-smuggling theorem, and a proof that no simpler admissible competitor exists — a universal negative over an open-ended quantifier, which (per THREE_ROOT_CLOSURE_PATHS_2026-06-23.md §3.2) is never closable. Worse, the one certificate form that would be absolute ("the shape is the unique solution of constraint set $C$") is not a certification at all — it is a reduction, and a successful absolute certificate self-defeats into the very derivation the document disclaims.

So the document keeps the fork genuinely open, with both branches live:

• Fork A — Shape stands. If, eventually, no simpler admissible competitor ever appears (and ideally a generative no-alternative theorem is proven), R2 becomes a genuine third brute fact alongside Scale and Granularity.
• Fork B — Shape folds. If a simpler admissible competitor is exhibited (the finite-state / discrete route is flagged as the most likely folder, possibly collapsing Granularity + Shape), then Shape is not absolutely irreducible. This is not a failure. The fork theorem says so explicitly: "This is not a failure. It is the correct compression outcome." Folding the shape into a deeper principle would be a win for the programme's stated goal (a smaller, sharper posit set), not a defeat.

A document that names the scenario in which its own central object disappears, and calls that scenario a success, is not engaged in promotion. That is the strongest single signal that the suite is disciplined rather than self-serving — and it is the posture this document adopts wholesale.

6.3 A provenance flag a blind reviewer should know

The fork theorem §2 cites an "axiom ledger" framing ("irreducible under known reductions," "smaller set genuinely implies the larger," "reductions count only if...") that a grep of GUT/GUT.md does not match (review §3 defect 8). The other fork anchors (the category-relative binding sentence; Appendix B1/B2) do verify. The axiom-ledger citation resolves to the on-disk TOE authority …/TOE/AXIOM_LEDGER.md (verified to exist and to carry the load-bearing "selection ≠ derivation" commitment at line 126 and the "Funnel-winner-forced = RELABEL = REJECTED" entry at line 186; this is the file the §1.2 hard rule is pinned to). The exact fork-theorem phrasings ("irreducible under known reductions," "smaller set genuinely implies the larger," "reductions count only if...") are not byte-matched in AXIOM_LEDGER.md as quoted and remain pending owner supply for verbatim wording; the cited file, not an unverifiable GUT.md phrase, is the correct on-disk anchor. Flagged, not hidden.

7. The irreducible residual is $E$

7.1 The decisive structural fact

Everything banked under R2 — every genuine sub-reduction — is computed from $E$ and presupposes $E$. $E$ is the Standard-Model chiral content: which representations, and three generations. The banked intra-geometry reductions are (GUT.md §4.13, line 1634; THREE_ROOT_CLOSURE_PATHS_2026-06-23.md §3.3):

Read the dependency: $\mathbb{Z}_6$ lands on $E$ + faithfulness; the spin form is read on the carriers chosen to host $E$; the Euler/index character $\chi(K_6, E) = -3$ has $E$ inside the index. None of these forces $E$; each consumes $E$ and returns a label. So the reducible part of the shape genuinely compresses — but it compresses onto $E$, and stops there.

7.2 Forcing $E$ is the decisive wall — no surviving forcing theorem

The question that would exit R2 (turn selection into derivation) is: what forces $E$? The candidate forcing principles and their fates (THREE_ROOT_CLOSURE_PATHS_2026-06-23.md §3.3; GUT.md §3.4–3.5):

There are, additionally, external results that make forcing $E$ look genuinely hard, not merely unattempted: there are infinitely many anomaly-free $U(1)$ extensions of the SM (Allanach et al., arXiv:2111.04148), and the generation number is not fixed by anomaly cancellation. So "no forcing theorem yet" is not a temporary gap that obviously closes; it is a wall with known structural reasons for standing.

7.3 Why this makes the suite's honest residue $E$, not "Shape"

The competitor suite (selector/role/realization/audit theorems) bottoms on the label "Shape" and folds $E$ into an input constraint (C4: chirality / no mirrors / three families). The honest framing — which GUT.md §4.13 now carries and the competitor suite omits — is that $E$ is the irreducible CORE, and "Shape" is the compressible wrapper around $E$. The distinction matters for a blind reviewer because:

• treating $E$ as an input constraint (C4) makes the suite's minimality result read as "given the SM spectrum, the cheapest complete carrier" — which is correct and honest as a selection;
• but it omits the load-bearing admission that the SM spectrum itself is the unforced thing. A reviewer scanning for "what is actually being assumed" should land on $E$, and the competitor suite does not point there. This document points there explicitly.

So the corrected residue statement is:

Accept $E$ (the actor content: which representations, three generations) and the three-layer $\times/\oplus/\otimes$ architecture. Everything else about the shape is selected-minimal-given-$E$, or derived-from-$E$. Forcing $E$ is the single decisive obstruction, and no forcing theorem survives the relabel falsifiers.

7.4 The relabel falsifiers that keep this honest

Every candidate forcing/reduction is run through three falsifiers before banking (THREE_ROOT_CLOSURE_PATHS_2026-06-23.md §3.3):

• F1 — pre-existence: is the principle independently motivated, not invented to hit the shape?
• F2 — derivation: does it derive (unique, no free family), not assert?
• F3 — non-circularity: is it free of any shape-equivalent fed in (e.g., $\chi = -3$, or $\theta_F$ "fixed by $|V_{us}|$" then used to derive $|V_{us}|$)?

Any failure = relabel = REJECTED. The $\mathbb{Z}_6$ reduction passes all three (it is a genuine computation, not a relabel) — but it lands on $E$ and so relocates within R2 rather than exiting it. Banking it as "Shape reduced" without naming the $E$-residual would be a soft overclaim, and this document does not.

8. Honest verdict, and exactly what would upgrade it

8.1 The status, stated at the correct strength

R2 — Local Structural Form ("Shape") — status 2026-06-24:
  IRREDUCIBLE BUT COMPRESSIBLE; accepted as an explicit axiom.

  EARNED:
    (1) category-relative SELECTOR-MINIMAL
        — the lex-min complete survivor of the declared scoped-GUT search
          category under the §4.6 Occam order (completeness > minimality binding);
        — the CP^2 -> K_6 elimination is a clean, uniform instance of that rule
          (CP^2 incomplete under Gate 4: tunable family count = fail by anti-fitting),
          NOT a reverse-engineered "forcedness" smuggle.
    (2) architecture-neutral functional-role FLOOR  k_role >= 3
        — NECESSARY, not sufficient; over the physics clauses C1-C10 only.
    (3) NO TIER-1 PREFERRED COMPETITOR CURRENTLY SURVIVES
        — current-record over 5 audited classes; "currently" never upgraded
          to "ever"; reopen rule live.

  NOT EARNED:
    (a) REALIZATION-MINIMALITY — 5 sub-lemmas OPEN; any printed "QED" is
        conditional-only; preorder \preceq_abs not yet well-defined.
    (b) ABSOLUTE IRREDUCIBILITY — OPEN, genuine two-branch fork:
        Shape stands, OR a simpler admissible competitor appears and Shape folds
        (the correct compression outcome, not a failure).

  RESIDUAL CORE: E (the SM chiral spectrum / 3 generations), which no known
    principle forces. Every banked sub-reduction (Z6, spin-form, chi=-3) is
    computed FROM E and presupposes it. "Shape" is the compressible wrapper around E.

  SELECTION != DERIVATION throughout. Funnel-winner-forced = RELABEL = REJECTED.
  No status was ever upgraded. Frozen branch dcc66f1b2685 / a5b1e6f9d951 READ-ONLY.

8.2 Comparison to the sibling roots (why Shape is the swing, and the weak link)

The asymmetry is the honest punchline: SCALE has an impossibility theorem; SHAPE has a selection plus an audit. They are not the same epistemic object, and this document does not pretend they are. SHAPE is the root whose status can still move — it can compress toward "$E$ + layering," or fold entirely (Fork B) — which is exactly why it is the swing root and why the strongest available claim about it is a selection, not a derivation.

8.3 Exactly what would upgrade each NOT-EARNED line

Stated as concrete, suppliable objects so the claim is falsifiable and the upgrade path is mechanical:

To upgrade realization-minimality (from skeleton to result): 1. Operationalize $\mathfrak{K}$ and $\preceq_{\rm abs}$ — give the complexity vector units and an aggregation/tie-break rule so the lex order is a well-defined total order. Without this the theorem statement is not even well-posed. 2. Discharge L3 (Actor minimality) first — it is the narrowest, the $\mathbb{Z}_6$ computation already lands there, and if it fails realization-minimality fails fast. Target: $E_{\rm matter}\oplus E_{\rm gauge}\oplus E_{\rm Higgs}\oplus E_{\rm proton}$ is the minimal actor realization (a no-alternative proof, not a best-found). 3. Discharge L2 (Rulebook minimality) — the known weakest link ($F^+$); needs a formal lower-burden-or-bust argument for the flavor chamber. 4. Discharge L1 (Stage minimality) and L4 (no cross-role compression) with a formal Unfold algorithm (not the current conceptual sketch) and a no-smuggling metric.

To upgrade the competitor audit (from current-record to a real bounded negative): 5. Fill the "simpler after unfolding?" cells for the five Tier-1 classes — convert "Unknown" to an adjudicated verdict by actually unfolding each and comparing burdens under the now-well-defined $\preceq_{\rm abs}$. Convert the matrix from a structured TODO into a finite falsifiable instrument.

To upgrade absolute irreducibility (the fork): 6. Prove a generative no-alternative theorem of the form "$B \models \mathcal{C}_{\rm phys} \Rightarrow B$ contains $\ge$ the functional complexity of $\mathcal{B}_{\rm active}$" — but note this requires (5) and a non-tautological role partition (fixing review defect 1), which is genuinely hard and may be unreachable in principle (universal negative). Honest odds: LOW for an absolute certificate; MODERATE for a bounded certificate of the CORE ($E$ + layering).

To touch the residual $E$ (the only thing that turns selection into derivation): 7. Supply a forcing theorem for $E$ that passes F1/F2/F3 — pre-existing, deriving (unique, no free family), non-circular. None exists; external results (infinitely many anomaly-free $U(1)$ extensions; unfixed generation number) suggest this is a wall, not a gap. Honest odds: LOW. The realistic terminal state is "$E$ + layering" as a sharper, smaller, still-accepted posit — never zero.

8.4 The bottom line

The frozen 13D geometry is selector-minimal, in the precise, category-relative, honestly-fenced sense developed here: it is the lex-min complete survivor of a declared, frozen search under a completeness-binding Occam funnel; it sits exactly on the necessary three-role floor; and no Tier-1 competitor currently beats it, with the reopen rule live. That is a real certificate, and it is stronger than "selected."

It is not a derivation, not realization-minimal (skeleton only; conditional "QED"), and not absolutely irreducible (a genuine fork in which the shape may legitimately fold). And the whole edifice bottoms on $E$ — the SM chiral content — which no known principle forces.

The honest one-line summary, which this document is willing to defend against any blind adversarial reviewer:

Shape is not derived. Inside the declared category it is the selector-minimal survivor; necessarily it carries at least three roles; currently nothing beats it. But selection is not derivation, the realization-minimality "QED" is conditional only, absolute irreducibility is an open fork, and the irreducible core is $E$, which is accepted, not forced.

No status was ever upgraded. This document states an acceptance and its honest defense, not a derivation. It closes no gate, derives no number, certifies or removes no axiom, and mutates no frozen object. Grounded in GUT/GUT.md (§3.5, §4.6, §4.13, Appendix B1/B2), …/TOE/SHAPE_R2_CERTIFICATE_SUITE/ (selector / functional-role / realization / competitor-audit / Tier-1 / fork theorems), …/TOE/REVIEW_SHAPE_SUITE_2026-06-23.md, and …/TOE/THREE_ROOT_CLOSURE_PATHS_2026-06-23.md (SHAPE section). Frozen branch dcc66f1b2685 / a5b1e6f9d951 READ-ONLY and confirmed unmutated.