Deep roots — legitimate terminal endpoints

Deep roots are the irreducible operating floor of the framework. They are where explanation currently stops. They are not magic, and they are not proof that no deeper theory exists. They are explicit commitments: each root is named, status-labeled, and connected to the rest of the anchor ledger. A root is acceptable only if it is openly paid, not smuggled.

A deep root is the opposite of a hidden assumption. It is an assumption or irreducible operating condition that has been named, typed, charged, and protected from overclaim. The one statement this page exists to make explicit is this:

$$\boxed{\;\text{A deep root is not a failure to explain. It is a declared endpoint where explanation stops honestly.}\;}$$

The deep roots are also where walls dissolve to. A wall dissolves given a complete root — the terminal endpoint is dissolved-given-Granularity / -Shape / -Scale (or a combination). And this is not a fallback: since nothing is solved from nothing (the floor is ≥ 1), dissolution ≠ solved, but it is the only possible endpoint — a genuine wall either is a measured anchor at the root floor, or it dissolves once the complete root removes the idealization that generated it. See Dissolutions — dissolved is not solved.

This page is part of the /closure/ set. Its siblings are Dissolutions — when a problem disappears once a root is granted and The wall-routing protocol — derivation, reduction-to-root, dissolution, or wall-record. It builds directly on the /anchors/ system and feeds the /walls/ frontier register.

Where the deep-root gates stand on the ratified board (2026-07-08): all three close RESOLVED +0 — Deep Root: Granularity as CERTIFIED-IRREDUCIBLE (the one-posit floor certified irreducible by banked countermodels), Deep Root: Shape as DERIVED-GIVEN-anchor (the 13D selector: carriers forced given the declared grammar and the observed content $E$), and Deep Root: Scale as MEASURED-ANCHOR ($M_{\rm Pl}$ the honest ruler). The full board stands at 33 resolved at +0 · 0 anchored at +1 · 0 open, and — separately, permanently — 0 of 33 gates are physics-closed: the anchor floor below every closure is ≥ 1 by design. Status propagates from the live ledger outward; this page is the discipline that produced those endpoints.

Core thesis

A finite physical theory cannot derive everything from nothing. If every claim must be derived from something deeper forever, the theory never becomes finite or reviewable. Therefore every reviewable framework needs terminal anchors.

Deep roots are those terminal anchors.

They are not all theorem-grade. Some are meta-admissibility conditions, some are physical operating roots, and some are declared posits that may later be reduced. Their job is to prevent exact objects from floating unpaid. The discipline throughout is the program's spine:

$$\text{anchored} \neq \text{closed} \quad\cdot\quad \text{dissolved} \neq \text{solved} \quad\cdot\quad \text{selected} \neq \text{forced} \quad\cdot\quad \text{given-}E \neq \text{derivation-of-}E.$$

Why deep roots are needed

1. To stop infinite regress

Every explanation either derives from deeper structure, loops back circularly, or bottoms out. Infinite regress is not a finite theory:

$$\text{claim} \rightarrow \text{premise} \rightarrow \text{deeper premise} \rightarrow \cdots$$

Without a root endpoint, the chain never becomes auditable. A deep root is the place where the chain is deliberately and visibly cut, with the cut named and charged rather than hidden.

2. To prevent hidden assumptions

If a framework uses invariance, records, causal order, granularity, scale, shape, or nonseparability without naming them, those roots become hidden assumptions. The rule is:

$$\boxed{\;\text{Every exact primitive must be generated, measured, charged, declared, or left open.}\;}$$

Naming a root does not make it true. It makes it visible — so a reviewer can see exactly where the framework rests on something unproven.

3. To distinguish derivation from anchoring

A root is not necessarily derived. A root can be terminal by anchoring:

$$\text{root named} + \text{status assigned} + \text{anti-claim stated} \;\Rightarrow\; \text{honest endpoint}.$$

Anchoring is anchor-transfer, never anchor-elimination. A terminal-anchored endpoint is a legitimate stopping place; it is not a claim that the target was derived.

4. To prevent unicorn targets

Some demands are not reasonable per-gate closure targets:

• derive granularity from nothing;
• prove the $13$D shape is absolutely unique across all possible mathematics;
• prove no future theory could do better;
• prove a measured value from the same gate that uses it as target.

These are often unicorn targets — universal negatives, unprovable in principle for any object in any field. A deep root lets the project stop at a paid primitive instead of pretending the unicorn was solved. A unicorn is never a wall; it is dissolved as a non-target, and only the bounded, in-grammar version is recorded as the real wall.

5. To support wall handling

When a gate hits a wall, the first question is:

$$\text{Is this wall reducible to a deep root?}$$

If yes, the terminal endpoint may be REDUCED-TO-AXIOM, AXIOM-CLOSED, or TERMINAL-ANCHORED, not full derivation. The full decision procedure lives in the sibling wall-routing protocol; the rule below is the deep-root entry into it.

The seven-root hierarchy

Use this root set unless explicitly updated by the project lead. The seven correspond one-to-one with R1–R7 on the /anchors/ Seven Deep Roots page; this page treats them as terminal endpoints, that page treats them as anchors. They do not contradict.

Theorem-grade roots versus declared roots

The seven are not all the same kind of object, and the page that pretends they are is overclaiming. Three sorts appear:

• Meta-admissibility conditions (R1, R2). These are conditions for anything to count as a physical claim or an empirical audit at all. They are upstream of every physical root and are not themselves derived theorems.
• Physical operating roots (R3, R4, R5, R7). These are posits about the world, charged explicitly. Some sub-parts are argued (e.g. the existence of a required scale); the rest are declared or measured.
• Declared / selected roots (R6, and the declared sub-posits of R4/R5). These bottom out on a named posit, a frozen branch, or a measured value, any of which may later be reduced but is not derived now.

A theorem-grade root would be one whose content follows from strictly weaker premises without importing the target. None of the seven is claimed theorem-grade across the board. Saying otherwise is the first universal forbidden overclaim.

Root definitions

R1 — Physical Equivalence / Invariance

Operational statement. Physical content is what survives admissible redescription: coordinate changes, gauge choices, basis changes, observer frames, and notational relabeling.

Allowed claim. A gate object must be invariant or gauge-accounted to count as physical.

Forbidden claim. A coordinate artifact is physical because it appears in one representation.

R2 — Record Interface

Operational statement. Physics is tested through finite, reproducible, physically instantiated records.

Allowed claim. Ledgers, hashes, symbolic runs, measured quantities, and audit artifacts enter through the record interface.

Forbidden claim. Observer means consciousness or subjective mind.

R3 — Causal Order

Operational statement. There is an invariant structure of possible influence.

Allowed claim. Causal order constrains signalling, horizons, locality, and update rules.

Forbidden claim. Nonseparability permits usable faster-than-light signalling.

R4 — Granularity / Cost-Floor

Operational statement. Distinguishable physical transitions carry nonzero specification/cost burden.

Allowed claim. Exact labels, charge tables, quotients, selectors, spectra, and constants are not free.

Forbidden claim. The granularity floor has been derived from nothing unless a strictly weaker reconstruction is supplied.

R5 — Scale

Operational statement. Physical theory requires at least one scale anchor or invariant scale relation.

Allowed claim. Scale anchors magnitude, cutoff, Planck, curvature, and above/below-regime statements.

Forbidden claim. A naked dimensionful number is invariant by itself.

R6 — Shape / Structural Form

Operational statement. The frozen structural branch is the form being tested.

Allowed claim. The shape can be selected or category-minimal inside a declared grammar.

Forbidden claim. The shape is absolutely unique across all possible mathematics.

R7 — Nonseparability / Global State Constraint

Operational statement. Exact independence requires factorization, and exact factorization is special. Write exact independence as

$$\rho_{AB}=\rho_A\otimes\rho_B,$$

with correlation / nonseparability the generic case $\rho_{AB}\neq\rho_A\otimes\rho_B$.

Allowed claim. Local closure must embed into global consistency; local results cannot automatically close a full gate.

Forbidden claim. Measurement sends a usable signal everywhere.

Terminal statuses for deep-root gates

When a gate's residual is a deep root, the legitimate endpoint is one of the following — never bare CLOSED.

What REDUCED-TO-AXIOM means

REDUCED-TO-AXIOM is the strongest honest endpoint available to a root that is not derived. It means the wall or root has been compressed to a named irreducible posit or minimal posit set — the count of free assumptions has been driven down as far as the framework can drive it, and what remains is one explicit, value-free declaration plus (where relevant) a measured residue. It is emphatically not a claim that the posit count reached zero. Reducing a root to one named axiom is progress; calling that "derived from nothing" is the overclaim the status exists to forbid.

The wall register records exactly this kind of endpoint: the granularity deep root is logged as REDUCED-TO-AXIOM (one value-free posit), TERMINAL-ANCHORED — with both the irreducibility unicorn and the $\hbar$-value unicorn dissolved — owning the bounded bridge wall W4. The shape deep root is logged as SELECTOR-MINIMAL (category-relative), TERMINAL-ANCHORED, owning walls W5 and W6. Those are the wall-register entries; on the ratified gate board the same two rows close at +0 — Granularity as CERTIFIED-IRREDUCIBLE and Shape as DERIVED-GIVEN-anchor (ledger) — because the compression is complete: what remains under each is a certified floor, not an owed derivation.

Worked example: granularity (the model example)

The granularity root should not be scored as "derived from nothing." Its legitimate endpoint is:

$$\boxed{\text{REDUCED-TO-AXIOM / AXIOM-CLOSED}}$$

Meaning:

• the root has been compressed to one named, value-free posit (exact-global action-spectrum uniformity — an earned no-go via a finite-resources countermodel, not proven atomic);
• $\hbar$ is treated as measured residue / action spacing — a separate, genuinely atomic measured anchor;
• the posit is not hidden — it is named, typed, and charged;
• pre-quantum reconstruction remains open (the bounded metric-selection bridge W4: granularity $\Rightarrow$ description-length over dimension-count).

So the honest granularity posit count is one, not zero: one declared posit plus one measured anchor. The existence/discreteness equivalence relocates the uniformity posit but does not eliminate it, and deriving uniformity from unitarity is circular by the program's own finding. On the ratified board this endpoint is the Deep Root: Granularity row — CERTIFIED-IRREDUCIBLE · RESOLVED +0: two banked countermodels prove finiteness alone does not force the uniform floor, so the one-posit compression is certified as far as reduction can go, and it closes at +0 rather than hanging as debt.

Allowed claim. Granularity has been reduced to one named posit plus measured residue.

Forbidden claim. Granularity has been derived from zero assumptions.

This pattern recurs at the gate level: the Yang–Mills mass-gap gate (Gap-02) closes CERTIFIED-IRREDUCIBLE · RESOLVED +0 — the continuum-existence half dissolves on the granularity root, the finite-gap half is certified irreducible to the standing Clay Millennium problem, and the gap value itself is consumed as a measured anchor. It is explicitly not a Clay solution, and the framework says so plainly. See the gap-02 anchor ledger.

Worked example: shape (the cautionary example)

The shape root must never be scored as derived-from-nothing — absolute uniqueness across all possible mathematics is a unicorn, and unicorns dissolve as non-targets. Its ratified endpoint is:

$$\boxed{\text{DERIVED-GIVEN-anchor (selector-minimal, given the declared grammar and }E\text{)}}$$

— the ratified Deep Root: Shape row, RESOLVED +0, with the absolute-uniqueness unicorn dissolved as a non-target.

Meaning:

• there is a category-relative selector win: within the declared internal-isometry grammar, the carriers of the frozen branch

$$M_4\times K_6\times S^2\times S^1_Y/\mathbb{Z}_2,\qquad K_6=SU(3)/T^2,\qquad D=4+6+2+1=13,$$

are forced given the grammar and the observed matter content $E$ (DERIVED-GIVEN-E);

• the declared grammar, the bridge axiom, and the given $E$ are paid — named and charged, not smuggled;
• architecture-neutral uniqueness and absolute irreducibility are universal negatives (a Kolmogorov-style uncomputable question) — unicorns dissolved as non-targets, correctly refused-as-axiom rather than carried as theorem debt; the bounded in-grammar sub-targets (W5/W6) are what the register records.

Allowed claim. The shape is selector-minimal inside the declared frozen grammar.

Forbidden claim. The shape is forced across all possible architectures.

The contrast with granularity is the lesson of this page: granularity closes on one named posit, certified irreducible; shape closes as DERIVED-GIVEN-anchor because its residual is a category-relative selection plus a dissolved uniqueness unicorn — the selection is paid, the unicorn is a non-target, and neither is carried as debt. Neither is physics-closed — no gate is (0 of 33, by design). See the shape deep-root anchor ledger.

Shared status vocabulary

These statuses are used consistently across the /closure/, /anchors/, and /walls/ pages.

Allowed claims

The deep-root system may legitimately say:

• A deep root is a named, typed, charged terminal endpoint where explanation stops honestly.
• A root can be terminal by anchoring rather than by derivation.
• Granularity has been reduced to one named posit plus measured residue.
• The shape is selector-minimal inside the declared frozen grammar.
• The carriers are DERIVED-GIVEN-E inside the declared internal-isometry grammar.
• A wall reducible to a deep root may be REDUCED-TO-AXIOM, AXIOM-CLOSED, or TERMINAL-ANCHORED.
• A measured value (e.g. $\hbar$, $M_{\rm Pl}$) is a MEASURED-ANCHOR, not a derived output.
• A unicorn target is dissolved as a non-target; the bounded in-grammar version is the real wall.

Forbidden claims (universal forbidden overclaims)

No closure page may say or imply any of the following:

• "deep roots are all theorem-grade."
• "a declared root is derived from nothing."
• "dissolved means solved."
• "selected means forced."
• "given-$E$ means derivation-of-$E$."
• "a measured anchor has been derived by the same gate that uses it as target."
• "hashes validate physics."
• "local closure equals whole-gate closure."
• "a wall-record is a physics solution."
• "a unicorn target must be solved before a gate can reach terminal endpoint."

The hard discipline, stated once more so it cannot be missed: anchored $\neq$ closed; dissolved $\neq$ solved; selected $\neq$ forced; given-$E$ $\neq$ derivation-of-$E$; a wall-record is not a solution. No physics gate is ever called CLOSED in the from-nothing sense unless its actual target is derived with no anchor consumed — and that bar is unreachable in principle (the anchor floor is ≥ 1). Currently zero physics gates are physics-closed — 0 of 33, stated plainly on the board — while all 33 stand RESOLVED at +0 on their ratified terminals. The two axes never blur: resolved means the honest endpoint was reached and named; physics-closed would mean solved from nothing, and nothing is.

Cross-links

• /closure/ — Dissolutions — when a problem disappears once a root is granted · The wall-routing protocol · this page.
• /anchors/ — The Seven Deep Roots (R1–R7 anchor view) · The Anchor Hierarchy overview · Master / Fundamental Anchors · granularity ledger · shape ledger.
• /walls/ — The Wall Register (frontier walls + terminal endpoints) · W4 — granularity $\Rightarrow$ MDL bridge.

Completion report

Tests passed.

• T1 — Defines deep roots as explicit terminal anchors, not magic (opening paragraph + boxed thesis).
• T2 — Explains why roots are needed to stop infinite regress (reason 1, with the regress chain).
• T3 — Includes the seven-root hierarchy (the seven-row table).
• T4 — Distinguishes theorem-grade roots from declared roots (the "theorem-grade versus declared" section).
• T5 — Explains REDUCED-TO-AXIOM (dedicated subsection + status table).
• T6 — Includes granularity as the model example (boxed REDUCED-TO-AXIOM / AXIOM-CLOSED).
• T7 — Includes shape as the cautionary example (boxed DERIVED-GIVEN-anchor, the uniqueness unicorn dissolved as a non-target).
• T8 — Contains an allowed-claims block and a forbidden-claims block.
• T9 — Includes the universal forbidden overclaims (all ten).
• T10 — Ends with this completion report.
• U1 — Shared status vocabulary table present and consistent with /anchors/ and /walls/.
• U2 — Terminal-anchored discipline stated explicitly (anchored ≠ closed; dissolved ≠ solved; selected ≠ forced; given-E ≠ derivation-of-E; a wall-record is not a solution).
• U3 — No physics gate called CLOSED from nothing; 0 of 33 physics-closed stated honestly beside the board's 33 RESOLVED at +0.
• U4 — All math in $...$ / $$...$$; no internal-process vocabulary; no QC-chip / out-of-scope-engineering content.
• U5 — Cross-links to both other /closure/ pages, to /anchors/, and to /walls/.

Tests failed. None.

Open items.

• Granularity uniformity posit (AXIOM-UNIFORM-GRAIN): REDUCED-TO-AXIOM (one value-free posit); not proven atomic. Posit count is one, not zero.
• Shape absolute / architecture-neutral uniqueness: a unicorn dissolved as a non-target — an undischarged universal negative, correctly refused-as-axiom; the bounded in-grammar sub-targets are walls W5 / W6.
• Bundle-uniqueness theorem that could remove "given-$E$": OPEN (wall W5).
• The description-length / MDL metric the cost-floor presupposes: OPEN (the bounded bridge wall W4).

Assumptions made.

• Adopted the handoff's seven-root list, root definitions, terminal-status table, and worked examples verbatim; kept R1–R7 consistent with the sibling /anchors/ Seven Deep Roots page (terminal-endpoint view here, anchor view there).
• Took the granularity and shape root accounting (REDUCED-TO-AXIOM at root level, walls W4 / W5 / W6) from the wall register, and the gate rows (CERTIFIED-IRREDUCIBLE and DERIVED-GIVEN-anchor, both RESOLVED +0) from the live /gates/ ledger, so this page contradicts neither.
• Used $...$ / $$...$$ for all math so it typesets, and only the shared status vocabulary labels.