Universality without Totality: Diagonal Constraints on Information, Coherence, and Holography

Universality without Totality: Diagonal Constraints on Information, Coherence, and Holography Andrei T. Patrascu FAST Foundation, Destin FL, 32541, USA email: andrei.patr[email protected] We argue that the black–hole information paradox is not a paradox of information loss, but a manifestation of a more general structural limitation: in sufficiently expressive, self–referential quantum systems, information about the system cannot be totally internalized as a single, globally coherent, and uniformly decodable code. By “information about the system” we mean semantic distinctions or predicates concerning the system’s states, observables, or histories, as opposed to the mere existence of physical correlations. While such information may exist and be preserved in correlations, diagonal arguments of Gödel–, Turing–, and Lawvere–type imply that no internal procedure can encode and decode all such distinctions uniformly within the same system once self–reference is present. We formulate this limitation in the language of categorical holography by viewing bulk–to– boundary maps as encoding functors and show that any encoding sufficiently rich to internalize its own reconstruction cannot serve as a total classifier of bulk predicates. Consequently, holographic reconstruction can at best be universal in a relative sense, i.e. restricted to code subspaces, sectors, or contexts, but never total. We demonstrate that modern mechanisms such as quantum error– correcting code subspaces, entanglement wedge reconstruction, and the island prescription realize this non–totality physically through state– and context–dependent reconstruction rather than through information loss. Finally, we connect this semantic non–totality to a physical mechanism, Higher Categorical Coherence Breakdown (HCCB), in which global linear and unitary descriptions fail to glue coherently across contexts despite local consistency. In this view, black holes represent maximally self–referential systems where the limits of total internal encoding become unavoidable and geometric, revealing a universal constraint on information in quantum theory rather than a pathology specific to gravity. I. INTRODUCTION A. From “information loss” to limits of decoding The black–hole information paradox is often narrated as a tension between (i) unitary quantum evolution, (ii) semiclassical locality, and (iii) Hawking’s semiclassical computation of thermal radiation and ever–growing entanglement entropy. Hawking’s original conclusion was that gravitational collapse may lead to a breakdown of predictability and an apparent loss of information [ 3 ]. Subsequent work reframed the question in operational terms: if a black hole evaporates unitarily, what is the time–dependence of the information accessible in the Hawking radiation, and what is the mechanism by which it becomes accessible? A central benchmark in this discussion is the Page curve [ 4 ], which predicts that the fine– grained entropy of the radiation increases until roughly the Page time and then decreases back to zero in a globally pure evaporation process. The modern landscape has sharpened the paradox in two complementary directions. On the one hand, the firewall argument [ 5 ] showed that demanding a single, globally consistent semiclassical description that simultaneously realizes unitarity, effective field theory at the horizon, and standard entanglement monogamy leads to an apparent inconsistency. On the other hand, developments in holographic entanglement and bulk reconstruction provided concrete tools for tracking quantum information in gravitational systems: the Ryu–Takayanagi prescription [ 6 ] and its covariant extension [ 7 ] relate boundary entanglement entropy to extremal surfaces in the bulk; quantum corrections [ 8 ] and the quantum extremal surface (QES) framework [ 9 ] incorporate bulk entanglement effects; and the operator–algebraic/quantum–error– correcting viewpoint of AdS/CFT [ 10 ] together with modular/relative entropy control [ 11 ] clarified why bulk reconstruction is naturally subsystem–relative rather than globally absolute. A decisive recent step was the emergence of the island prescription and replica wormholes [ 14 – 16 ], which recover a Page–curve–consistent fine–grained entropy from semiclassical geometry by allowing the entanglement wedge of the radiation to include interior “islands”. In parallel, entanglement wedge reconstruction was used to articulate precisely when (and in what sense) interior operators can be reconstructed from the radiation [ 13 ], building on the older Hayden–Preskill insight that, after the Page time, rapid scrambling can make newly infalling quantum information recoverable from the radiation given sufficient control [12]. 2 These developments strongly suggest that the “paradox” is not simply a conflict between unitarity and gravity, but a conflict between certain totality assumptions about decoding and the structure of self–referential quantum systems. The guiding thesis of this paper is that the black–hole information paradox is best understood as exposing a universal limitation: Information about the system may exist and be preserved in correlations, yet cannot always be totally internalized as a single, globally coherent, uniformly decodable code within a sufficiently expressive, self–referential system. The remainder of this introduction makes this statement precise enough to be assessed logically, and motivates why it naturally leads to (i) diagonal/undecidability constraints (Gödel/Turing/Lawvere) [ 17 – 19 ], (ii) “universality without totality” in categorical holography, and (iii) a physical realization in terms of Higher Categorical Coherence Breakdown (HCCB), in which global gluing of a linear/unitary description fails despite local consistency. B. The hidden assumption: total internal decoding We begin by separating three distinct notions that are frequently conflated: (A) Existence/preservation of correlations: quantum states determine correlation functions and entanglement measures, and unitary dynamics preserves global purity. (B) Information about the system: semantic distinctions or predicates concerning the system’s states, observables, or histories (e.g. “which microstate?”, “which interior configuration?”, “which sector?”). (C) Internal encodability/decodability: the existence of an internal representation and a uniform decoding procedure which, from within the same system/theory, recovers the answers to all predicates of interest. Diagonal arguments constrain (C), not (A). In particular, the claim that “information is not totally encodable” is not a claim of information destruction; it is a claim that no single internal coding/decoding structure can uniformly represent all semantic distinctions about a sufficiently expressive, self–referential system. To formalize the hidden assumption, let Sys denote the “system” (e.g. the full quantum gravitational system describing an evaporating black hole plus radiation) and let Q denote a class of questions about Sys . Concretely, Q may be taken as a space of predicates on states, a family of bulk observables, or a family of operationally meaningful propositions that one would like to decide. We write an answer set as a set (or space) A. An encoding is any map (or functorial assignment) E:Sys −→ Code,(1) which associates to the system an internal code object Code (e.g. the radiation state, a boundary algebra, a family of reduced density matrices, a set of accessible correlators). A decoder is a procedure D:Code × Q −→ A,(2) intended to output the answer to question q∈ Q using only the internally available code. Total internal decoding (informal definition). We call (E,D)atotal internal decoder if: (i) Totality: D is defined for all q∈ Q and all relevant system states (i.e. it halts/returns an answer uniformly). (ii) Correctness: for each q∈ Q , the value D ( E ( Sys ) , q )agrees with the “ground truth” semantics of q for Sys (e.g. correct expectation values, correct membership in a sector, correct interior predicate, etc.). (iii) Internality and uniformity: E and D are definable within the same theoretical framework describing Sys and do not rely on external oracles (no meta–theory privileged viewpoint). 3 Many informal arguments in the black–hole literature implicitly assume such a total decoder exists, at least in principle, for the Hawking radiation. The Hayden–Preskill protocol, for instance, assumes “unlimited control” over the radiation in order to analyze retrieval times [ 12 ]; and many versions of the paradox tacitly treat decoding as an abstract operation unconstrained by self–reference. However, the diagonal phenomena behind Gödel incompleteness [ 17 ] and the halting problem [ 18 ] teach that total internal decoders are not generically available once the system is expressive enough to encode statements about its own encoding. C. Diagonal logic: why total internal decoding fails The underlying logical structure is classical: whenever a system can represent (enough of) its own semantics or operational behavior internally, universal “deciders” become impossible by diagonalization. In arithmetic, Gödel’s diagonal lemma produces a sentence asserting its own unprovability, implying that no consistent sufficiently expressive system is complete [ 17 ]. In computation, Turing’s diagonal construction shows that no program can decide halting for all programs [ 18 ]. In category theory, Lawvere gave a unifying formulation of these diagonal arguments in cartesian closed categories (CCC), showing that a suitable “evaluation + weak surjectivity” hypothesis forces fixed points for all endomorphisms—and hence yields contradictions in the presence of negation–like maps [ 19 ]. We will later give a detailed formulation adapted to holographic encoding functors, but the meta–lesson can already be stated: If a decoding scheme is rich enough to internalize its own action on encodings, then demanding totality (a uniform correct decoder for all predicates) forces a diagonal predicate/operator which the decoder cannot handle consistently. This is the sense in which self–reference imposes a limit of total encodability: the relevant “information” is information about the system (predicates, properties, and semantic distinctions), not the raw existence of correlations. D. Universality without totality in holography and islands The AdS/CFT correspondence motivates the idea that boundary data encodes bulk physics. However, modern developments already indicate that reconstruction is intrinsically relative: bulk locality behaves like a quantum error–correcting code, and reconstruction is controlled on a code subspace rather than across the full Hilbert space [ 10 ]. Entanglement wedge reconstruction, guided by RT/HRT and its quantum refinements [ 6 – 9 ], makes reconstruction explicitly dependent on the boundary region and the state. The island rule and replica wormholes [ 14 – 16 ] go further: the reconstructible region itself can change discontinuously (a QES phase transition), yielding a Page curve consistent with unitarity while avoiding naive “total decoder” expectations. This picture is sharpened by analyses that tie the Page transition directly to entanglement wedge reconstruction of the radiation [13]. Our proposal is to interpret these facts through a single structural lens: holography supports universality of encoding but forbids totality of decoding. In other words, there can be universal cores—minimal structures through which all successful reconstructions factor—without the possibility of a single globally defined decoder for all bulk predicates. The island prescription then appears not as an ad hoc fix, but as a geometrized implementation of contextual (state–dependent) decoding consistent with diagonal constraints. E. Preview: Higher Categorical Coherence Breakdown (HCCB) The diagonal discussion above is semantic: it constrains what can be internally decided or uniformly represented in a self–referential system. A separate but complementary question is physical: how does the theory realize these constraints dynamically and operationally? The guiding idea of Higher Categorical Coherence Breakdown (HCCB) is that the minimal consistency requirement in quantum theory is not “global symmetry” or even “global unitarity” as an absolute structure, but rather coherence closure—the ability to glue local/sectorial descriptions consistently. When higher coherence conditions fail, one expects the global description to cease to be a single linear/unitary object while retaining local consistency; physically this appears as sectorization, history dependence, and effective completely positive (CP) 4 dynamics on accessible subalgebras. These ideas are developed in detail in our HCCB framework [ 1 , 2 ]. In this paper, we use the black–hole setting as a maximally self–referential regime where these issues are geometrically unavoidable, and we propose a unified narrative: diagonal non–totality provides the logical reason total decoders fail, while HCCB provides a physical mechanism by which the world implements coherence without totality. Roadmap. Section I has isolated the hidden totality assumption. In subsequent sections we: (i) develop the diagonal constraint in a form suited to holographic encoding functors, using the Lawvere–style categorical diagonal mechanism [ 19 ] and its logical ancestors [ 17 , 18 ]; (ii) reinterpret entanglement wedge reconstruction and islands as concrete realizations of “universality without totality” [ 6 – 9 , 13 , 14 , 16 ]; and (iii) connect these semantic constraints to HCCB as a unifying physical mechanism. II. WHAT DO WE MEAN BY “INFORMATION”, “ENCODING”, AND “DECODING”? A. Why definitions matter: avoiding a category error A large fraction of the confusion surrounding “information loss” in gravitational settings comes from conflating distinct layers of description: (i) physical correlations carried by quantum states, (ii) semantic distinctions (predicates) about the system, and (iii) the existence of a uniform internal procedure that can encode and decode all such distinctions from within the system. Diagonal arguments (Gödel/Turing/Lawvere) constrain the third layer, not the first [17–19]. Accordingly, we fix terminology precisely. Throughout, “information” is understood in the operational/information–theoretic sense introduced by Shannon [ 20 ] and generalized to quantum systems by von Neumann’s entropy and its refinements [ 21 , 22 ]. We distinguish this from “information about the system”, meaning semantic distinctions or predicates concerning states, observables, and histories. This section formalizes (i) what it means for such information to exist and be preserved, (ii) what it means to be encodable and decodable, and (iii) what “total” means in each case. B. Operational information: correlations as constraints States and observables. Let H be a (separable) Hilbert space and let B ( H )denote the bounded operators. A state is a density operator ρ≥ 0with Trρ = 1. Operational predictions are expectation values hOiρ = Tr ( ρ O )for observables O in an admissible algebra (often a von Neumann algebra in QFT contexts). Entropy and mutual information. The von Neumann entropy S(ρ) = −Tr(ρlog ρ)(3) quantifies uncertainty of ρ and reduces to Shannon entropy on classical distributions [ 20 , 21 ]. For a bipartite state ρAB with marginals ρA= TrBρAB,ρB= TrAρAB, the mutual information I(A:B)ρ=S(ρA) + S(ρB)−S(ρAB)(4) measures total correlations. It can be written as a relative entropy: I(A:B)ρ=D(ρAB kρA⊗ρB),(5) where D ( ρkσ )is the (Umegaki) quantum relative entropy [ 23 ]. (Equation (5) follows directly from the definition D ( ρkσ ) = Tr ( ρ ( log ρ−log σ )) and basic logarithm rules when σ = ρA⊗ρB .) Relative entropy is the operational measure that controls distinguishability and obeys monotonicity under physical channels, a property crucial for any discussion of accessible information [21, 22]. Monotonicity and coarse–graining. Aquantum channel (CPTP map) Φis a completely positive, trace–preserving linear map on density matrices. The monotonicity (data processing inequality) D(ρkσ)≥D(Φ(ρ)kΦ(σ)) (6) expresses the fact that distinguishability cannot increase under physical processing [ 22 , 23 ]. This is the precise mathematical form of “forgetting” or coarse–graining: channels destroy or hide information by reducing distinguishability. We emphasize that (6) is a statement about accessible information and correlation structure; it does not by itself impose any limitation on the existence of semantic truths about the system. 5 C. Encoding and decoding as internal channels We now formalize encoding and decoding in a manner suitable both for quantum information theory and for holographic/bulk–boundary settings. Encoding map. An encoding is modeled as a channel (or functorial assignment) E:S(Hbulk)−→ S(Hcode),(7) where S ( H )denotes the state space on H . In holographic language, Hcode can be the boundary/radiation subsystem (or an algebra thereof), and E is the physical map that produces the accessible data (e.g. restriction to a subalgebra, partial trace, or boundary encoding). For generality, we keep the channel formulation; it is standard that every channel admits a Stinespring dilation, i.e. can be realized as a unitary on a larger space followed by discarding an environment [21, 24]: E(ρ) = TrenvU(ρ⊗ |0ih0|)U†.(8) This is important conceptually: “encoding” is not mysterious; it is just a physical interaction plus discarding degrees of freedom. Decoding map. Adecoder is a channel D:S(Hcode)−→ S(Hbulk),(9) intended to reconstruct bulk information from the code. In quantum error correction, one demands that D ◦ E act as the identity on a specified code subspace (or on a set of states), not globally [ 21 , 22 ]. This distinction (relative vs global decoding) will be central when we interpret holographic reconstruction and islands as contextual decoding mechanisms rather than total ones. Semantic information vs operational reconstruction. To connect decoding to information about the system, introduce a family of predicates or queries Q. Operationally, a query can be: •an expectation functional qO(ρ) = Tr(ρO)for some observable O, •a sector label (superselection index), •a statement about membership in a code subspace, •or any decision problem definable from the available algebra of observables. A decoder is said to answer a query if the reconstructed state reproduces the same operational statistics for the relevant family of observables/predicates. This makes explicit that “information about the system” is not merely S ( ρ )but a structured family of semantic distinctions one would like to recover from the internal code. D. Total encoding/decoding: what “total” means We now define the notion that will be ruled out by diagonal logic in the regimes of interest. Total decoding (definition). Fix a class of queries Qand an operational semantics mapping J·K:Q×S(Hbulk)→ A,(10) where A is an answer space (numbers, distributions, labels, etc.). An internal decoding scheme for ( E,Q ) is a decoder D such that the composite D ◦ E reproduces the semantics for all queries and all states in a specified domain Ω⊆ S(Hbulk): Jq, ρK=Jq, (D ◦ E)(ρ)K,∀q∈ Q,∀ρ∈Ω.(11) We call the decoder total (relative to Q and Ω) if it is uniformly defined on E (Ω) and satisfies (11) for all queries in Q. Total internal decoding (strengthened). The qualifier “internal” means that the scheme ( E,D )is implementable within the same physical/theoretical framework that defines Sys and its semantics, without an external oracle. In particular, when Q includes questions about the encoding/decoding process itself (self–reference), the demand that a single uniformly defined D answer all such questions becomes the analogue of a “total halting decider” or “complete truth predicate” [ 17 – 19 ]. This is precisely the demand that will be shown to be untenable in sufficiently expressive self–referential regimes. 6 Relative (non-total) decoding. Importantly, nothing in physics requires totality in the above sense. In quantum error correction and in holographic reconstruction, one typically has: (D ◦ E)(ρ)≈ρonly for ρ∈Ωcode,(12) a restricted set of states (a code subspace, a sector, a semiclassical regime). This is universality without totality: reconstruction is robust within a controlled context but not globally uniform. E. Existence vs encodability vs decodability We now explicitly separate the three notions that will recur throughout the paper. (i) Existence/preservation. Information in the operational sense exists whenever the state has nontrivial correlations (e.g. nonzero mutual information (4) ) and is preserved under unitary evolution on a closed system. The statement “information is preserved” is therefore, at base, a statement about the existence and evolution of correlations in ρ(t). (ii) Encodability. Encodability is the existence of an internal representation that makes a chosen class of semantic distinctions accessible from within a code. Formally, given Q and Ω, encodability asks whether there exists a code space and an encoding E such that E (Ω) retains enough distinguishability to answer Q. (iii) Decodability. Decodability asks whether there exists a decoder D satisfying (11) on the relevant domain. Total decodability is the strongest form: one decoder, defined uniformly, answers all queries on all states in the domain. Diagonal arguments constrain (iii), not (i), and often only constrain (iii) once self–reference makes Q sufficiently rich. F. What is not claimed Because the paper’s claims can be misread, we state explicitly what is not asserted. No claim of information destruction. We do not claim that information in the sense of correlations is destroyed. To the contrary, the modern semiclassical Page curve obtained via islands is consistent with global unitarity of the evaporation process (as reviewed in Section I). No claim of fundamental nonunitarity per se.We do not claim that quantum dynamics must violate unitarity as a fundamental law. Our claim is more structural and strictly weaker: even if global unitarity holds, the demand for a total internal decoding of all semantic distinctions about a sufficiently expressive self–referential system is too strong and is obstructed by diagonal logic [ 17 – 19 ]. In later sections we will connect this semantic obstruction to HCCB as a physical mechanism that replaces global linear/unitary descriptions by coherence closure on accessible algebras [1, 2]. Summary. Information (correlations) may exist and be preserved, while information about the system may fail to be totally internalizable as a single globally coherent code admitting a uniform decoder. This is the precise sense in which “universality” need not imply “totality”, and it is the conceptual foundation for our subsequent diagonal and holographic analysis. III. SELF-REFERENCE AND THE LIMITS OF TOTAL ENCODING A. Goal and logical stance The purpose of this section is to introduce, in a unified and logically transparent way, the diagonal mechanism that underlies the strongest impossibility results about “total internal decoding.” The main theme is simple but profound: Once a system is expressive enough to represent descriptions of its own behavior (or its own encodings) internally, any attempt to provide a total internal classifier/decoder for all semantic distinctions about the system generates, by diagonalization, a predicate that escapes classification. This is the shared logical skeleton behind: 7 •Cantor’s theorem (no set encodes all its subsets) [25], •Gödel’s first incompleteness theorem (truth 6=provability) [17, 26], •Turing’s halting theorem (behavior 6=decidability) [18, 27], •Lawvere’s fixed–point theorem (categorical diagonalization) [19]. As emphasized in Section II, the obstruction concerns totality of internal encodability/decodability of information about the system (semantic predicates), not the existence of correlations or the possibility of restricted (relative) reconstruction. B. Self-reference as internal representability of descriptions We now formalize the notion of self-reference at the level needed for diagonal arguments. Self-reference (operational form). A system (or formal theory) is self-referential (for our purposes) if: (SR1) it admits an internal representation of descriptions of its own objects (e.g. encodings of predicates, programs, proofs, decoders), (SR2) it admits an internal mechanism to apply a description to an input (evaluation, execution, interpretation), (SR3) the class of admissible queries Q (Section II) includes queries about this very representability/evaluation (i.e. the encoding is itself a valid subject of inquiry). In logic this is realized by Gödel numbering (syntax becomes arithmetic) [ 17 ]; in computation by the existence of universal machines and program self-application [ 18 ]; and categorically by the availability of evaluation morphisms in a cartesian closed category [19]. Why self-reference matters for “total decoding”. Recall the notion of a total internal decoder from Section II D: a single, uniformly defined procedure that decides all semantic predicates of interest from an internal code. If (SR1)–(SR3) hold, the system can encode the statement “the decoder outputs 1on the encoding of this predicate” and can form predicates that talk about (and flip) their own decoded value. This is the diagonal move. C. The diagonal argument as a general schema We state an abstract schema that will be instantiated in Cantor/Gödel/Turing/Lawvere. Diagonal schema. Suppose we have: •a domain Xof “names” or “codes”, •a class of predicates on some space Y(often Y=X), •a map (an “enumerator” / “classifier” / “decoder”) Φ : X−→ P(Y),(13) which claims to associate to each code x∈Xa predicate Φ(x)⊆Y, and • a means of self-application allowing us to compare Φ( x )with the value of Φ( x )at x (this is the crucial self-reference step). Define the diagonal predicate D:= {y∈Y:y /∈Φ(y)}.(14) Then D differs from every Φ( x )at least at the point x , hence D6 = Φ( x )for all x . Therefore Φis not surjective: the claimed classification misses at least one predicate. This is the entire mechanism. The only nontrivial issue in each application is whether (SR1)–(SR3) allow the construction (14) internally and whether the meaning of /∈ (negation) is available in the relevant setting. 8 D. Cantor: no total powerset encoding We begin with the cleanest diagonal theorem, which requires no computation or proof theory. Theorem III.1 (Cantor, diagonal form) . For any set X , there is no surjection f : X→ P ( X )onto its power set. Equivalently, |X|<|P(X)|. Proof. Assume for contradiction that f:X→ P(X)is surjective. Define the diagonal subset D:= {x∈X:x /∈f(x)}∈P(X).(15) By surjectivity, there exists a∈Xsuch that f(a) = D. Now ask whether a∈D. By definition (15), a∈D⇐⇒ a /∈f(a). But since f(a) = D, this becomes a∈D⇐⇒ a /∈D, a contradiction. Therefore no such surjection exists. Interpretation for “encoding”. Cantor’s theorem says: there is no total encoding of all predicates on X (i.e. all subsets of X ) by elements of X itself. This is the archetype of “no total internal classifier.” In later sections, the role of “subsets” will be played by predicates about states/observables, and the role of X by a space of codes/representations available internally. Cantor’s theorem thus provides the simplest intuition for why “totality” is dangerous in self-referential settings [25]. E. Gödel: truth is not identical to provability We next isolate the diagonal core of Gödel’s first incompleteness theorem. Setup. Let T be a recursively axiomatized theory extending a weak fragment of arithmetic sufficient to represent primitive recursive functions (e.g. Robinson arithmetic Q suffices for the diagonal lemma; stronger hypotheses are used for standard incompleteness statements). Gödel’s key move is to arithmetize syntax: formulas and proofs are coded by natural numbers (Gödel numbers). This supplies (SR1). Provability predicate. Because T is recursively axiomatized, there exists an arithmetic formula ProvT ( x ) expressing “ x is the Gödel number of a T -provable sentence”. Crucially, ProvT is internal to arithmetic and therefore to T(SR1–SR2). Diagonal lemma (self-reference engine). The diagonal lemma states that for any one-free-variable formula F(x)there exists a sentence Gsuch that Tproves G↔F(pGq),(16) where pGq is the numeral coding the sentence G . This is the formal realization of self-application (SR2–SR3). A detailed presentation can be found in [26]. Gödel sentence. Choose F(x) := ¬ProvT(x). Then (16) yields a sentence Gwith T`G↔ ¬ProvT(pGq).(17) Thus Gasserts “Gis not provable in T”. Theorem III.2 (Gödel, first incompleteness (diagonal core)) . If T is consistent, then T0G . Moreover, under standard mild additional hypotheses (e.g. ω -consistency, or suitable soundness assumptions), T0¬G as well; hence Tis incomplete. Proof of T0Gunder consistency. Assume for contradiction that T`G . Then T also proves ProvT ( pGq ), since a proof of G is a witness to its provability. But (17) implies T` ¬ProvT ( pGq ). Thus T proves a contradiction. This contradicts the consistency of T. Hence T0G. Interpretation. The theorem shows that the semantic notion “truth in the standard model” cannot be totally internalized as “provability in T ” once T is expressive enough to talk about its own proofs. In our language: “provability” is an attempted internal decoder for semantic truth about arithmetic; diagonalization produces a predicate ( G ) that escapes total decoding. This is the direct analogue of Cantor’s Din the space of sentences rather than subsets. 9 F. Turing: behavior is not identical to decidability We now state the computational incarnation of the same diagonal mechanism. Setup. Fix a standard model of computation (Turing machines). Let H ( M, x )denote the predicate “machine Mhalts on input x”. Theorem III.3 (Turing, halting problem) . There is no total computable function (algorithm) Halt ( M, x ) ∈ {0,1}that returns 1iff Mhalts on input xand returns 0otherwise, for all (M, x). Diagonal proof. Assume for contradiction that such a total decider Halt exists. Define a new machine D that, on input M, does the following: 1. computes Halt(M, M); 2. if Halt(M, M)=1, then Dloops forever; 3. if Halt(M, M)=0, then Dhalts immediately. This definition is effective if Halt is effective. Now evaluate D on its own code: consider D ( D ). If Halt ( D, D ) = 1, then by definition D ( D )loops forever, contradicting halting. If Halt ( D, D ) = 0, then by definition D(D)halts, contradicting non-halting. Thus Halt cannot exist. Interpretation. Here the “information about the system” is the semantic predicate “halts/does not halt.” The “internal decoder” is the hypothetical halting decider. Self-reference arises because programs can take programs as input, and because the decider is itself a program [ 18 , 27 ]. The diagonal machine D plays the same role as Cantor’s Dand Gödel’s G. G. Lawvere fixed points: diagonal logic in categorical form Cantor/Gödel/Turing are often presented as separate theorems. Lawvere’s contribution was to exhibit a common categorical spine in cartesian closed categories (CCC) [ 19 ]. We present the conceptual statement and its proof in a form that makes the diagonal mechanism transparent and that will be adaptable later to categorical holography. Cartesian closed background. Let C be a CCC. For objects A, B ∈ C , there exists an exponential object BAand an evaluation morphism ev : BA×A→B, (18) which satisfies the usual adjunction C(X×A, B)∼ =C(X, BA). Weak point-surjectivity. A morphism φ : A→BA is called weakly point-surjective (wps) if for every morphism g:A→Bthere exists an element a: 1 →A(a “global point”) such that g= ev ◦(φ◦a×idA).(19) Informally: every predicate/behavior g is realized by evaluating a code φ ( a )at its input. This is precisely the categorical form of “the system can internally represent all predicates.” Theorem III.4 (Lawvere fixed-point theorem (diagonal form)) . Let C be a CCC, let A, B ∈ C , and assume there exists a wps morphism φ : A→BA . Then every endomorphism h : B→B has a fixed point in the following sense: there exists a global element b: 1 →Bsuch that h◦b=b. (20) Proof. Let h:B→Bbe given. Consider the composite A∆ −→ A×Aφ×idA −−−−→ BA×Aev −→ Bh −→ B, (21) where ∆is the diagonal map. Denote this composite by g:A→B: g:= h◦ev ◦(φ×idA)◦∆.(22) 16 Self-reference requirement (holographic form). The diagonal mechanism requires that the system be able to express predicates that reference the encoding/decoding operation. In categorical terms, this means (schematically): • the probe category P supports a sufficiently rich predicate logic (e.g. subobject classifier in a topos, or evaluation in a CCC, or an equivalent representability structure) [19, 36, 37]; • the encoding functor U is rich enough that the image U ( X )carries internal representations of predicate claims about Uitself (closure of Q). Under these conditions, demanding total classification (29) forces a diagonal predicate, exactly as in (14) . The essential idea. If Θ X exists for all X and the predicate logic includes negation/complement in the relevant sense, then for each Xone can define a “liar-style” predicate DX:= ¬Θ−1 X(“DXholds of the code that names DX”),(30) whose content is: “this predicate fails exactly when the boundary classifier says it holds of itself.” By construction, DX cannot be consistently classified by Θ X . This is the same structure as Cantor’s D (Theorem III.1), Gödel’s G(Theorem III.2), and Turing’s diagonal machine (Theorem III.3). The statement (30) is schematic because its fully formal expression depends on the precise internal logic adopted for predicates in P ; the unifying categorical engine behind all such diagonal constructions is Lawvere’s fixed point theorem [19]. F. Key Proposition 2: No-Total-Holography Proposition V.1 (No-Total-Holography).Let U:B → P be an encoding functor. Assume: (H1) (Predicate expressivity) The probe category P supports a sufficiently rich internal predicate logic to represent predicates about objects in P , including predicates about the action of U and its would-be reconstruction (e.g. via a topos/subobject classifier or via CCC evaluation as in Lawvere’s framework) [19, 36, 37]. (H2) (Self-reference) The class of admissible queries Q is closed under reflection on the encoding itself: predicates about the output of the would-be decoder applied to encodings are themselves admissible. (H3) (Total classification demand) There exists a natural family of maps Θ X as in (29) that purports to classify all bulk predicates from boundary predicates for all Xin the intended domain. Then the demand (H3) cannot hold simultaneously with (H1)–(H2): there is no encoding functor U that is a total classifier of bulk predicates in a self-referential regime. Equivalently, holography can be universal only relatively (restricted to contexts such as code subspaces/sectors/probe families), but not total. Proof sketch (diagonal reduction). Assume (H1)–(H3). By (H1) the probe logic can represent predicates about U(X)and, crucially, about the act of decoding/classifying itself. By (H2) such “meta-predicates” are admissible queries and are therefore subject to the classification map Θ X . Now form the diagonal predicate DX whose truth value is defined to disagree with the classifier’s own prediction about DX (as in the general diagonal schema (14) and the schematic liar predicate (30) ). By construction, DX cannot be consistently classified by Θ X without contradiction. Hence (H3) fails. The only way to avoid contradiction is to weaken totality: restrict the domain (code subspace), restrict the query class, or accept contextual/state-dependent classification. Interpretation and physics reading. Proposition V.1 is not an “anti-holography” statement. It says that, in self-referential regimes, boundary encoding cannot be promoted to a single globally defined decoder that decides all bulk predicates uniformly. Instead, one expects precisely the structures seen in modern holography: •reconstruction valid on code subspaces (relative universality) [10], •entanglement-wedge/state/region dependence [6, 9], •algebra/center/edge-mode extensions in gauge/gravity localization [31], •and, in evaporating black holes, island-induced reconstruction transitions [13, 14, 16]. All of these are consistent realizations of “universality without totality.” 17 G. Why “universality” survives: abduction and factorization The diagonal obstruction targets only totality, not universality. In the abductive (universal-property) sense discussed earlier, universality means: there exists a minimal core structure through which all successful reconstructions factor. Categorically, such a core is characterized by factorization/universal properties, not by total classification of all predicates. This distinction is the backbone of our narrative: holography can remain a powerful universal principle precisely because it does not promise total internal decoding. VI. ENTANGLEMENT WEDGE RECONSTRUCTION AS RELATIVE UNIVERSALITY A. Goal and relation to No-Total-Holography Section V established that “total holography”—a single globally defined decoder/classifier for all bulk predicates—is too strong in self-referential regimes (Proposition V.1). The purpose of this section is to show that modern holographic reconstruction already implements the correct logic: it provides relative universality rather than totality. Concretely, bulk reconstruction is best understood as a quantum error correction (QEC) phenomenon: bulk degrees of freedom are redundantly encoded in boundary degrees of freedom, and the existence of recovery maps depends on the code subspace, the boundary region, and (in gravitational settings) sometimes the state through which the semiclassical geometry is defined. This is not a defect; it is precisely the form of universality compatible with diagonal constraints. B. Quantum error correction perspective: encoding is physical, decoding is relative The QEC viewpoint on AdS/CFT was articulated clearly in [ 10 ] and has become standard: a bulk effective Hilbert space (or algebra) is encoded into the boundary Hilbert space in a way analogous to a quantum code. Code subspace. Let HCFT denote the boundary Hilbert space. A code subspace Hcode ⊆ HCFT models the subspace of states admitting a semiclassical bulk interpretation (low-energy excitations on a fixed background, or a controlled family of backgrounds). The inclusion isometry V:Hcode ,→ HCFT (31) plays the role of an encoding (in the sense of (7) ): for a code state ρcode , the corresponding physical boundary state is ρCFT =V ρcode V†.(32) The crucial point is that V is not claimed to encode all of HCFT into a bulk; it encodes a restricted domain. This restriction is the simplest, most concrete manifestation of “universality without totality.” Subsystem recovery. Let the boundary degrees of freedom be partitioned (at least conceptually) into a region R and its complement ¯ R with HCFT ∼ =HR⊗ H ¯ R . A central QEC question is: given access only to R , can one recover certain code observables? This is precisely the operational meaning of “bulk reconstruction on a boundary region.” C. Operator algebra quantum error correction and recovery criteria The most robust formulation uses operator-algebra quantum error correction (OAQEC), which naturally matches holography’s emphasis on reconstructing algebras of operators rather than arbitrary states. A convenient entry point is the work relating boundary and bulk relative entropies [ 11 ] and the subsequent development of operator algebra reconstruction in the entanglement wedge framework [38]. Algebraic reconstruction statement (informal). Let Abulk ( W )be the algebra of (low-energy) bulk operators supported in a bulk region W (typically an entanglement wedge), and let AR be the boundary 18 operator algebra on R . Entanglement wedge reconstruction asserts that, under appropriate conditions, for each O∈ Abulk(EW (R)) there exists an operator e O∈ ARsuch that V†e O V =Oon Hcode.(33) Equation (33) is the precise meaning of “Ois reconstructible from R” within the code subspace. Relative entropy control. A powerful characterization of when reconstruction holds is via relative entropy. One formulation (in a suitable regime) is that relative entropy of reduced states on R equals relative entropy of corresponding bulk states in the entanglement wedge [ 11 ]. This “entropy equality” is a strong compatibility condition indicating that Rcontains exactly the information needed to distinguish code states as far as bulk observables in EW ( R )are concerned. Because relative entropy is monotone under channels (6) , equalities of this type are highly constraining and naturally express “recoverability”. OAQEC language (sketch). In OAQEC one corrects an algebra A rather than a full code space. The correctability condition can be phrased in terms of commutants and the action of the noise channel on the code (see e.g. [ 39 ]). In holography, the “noise” is the restriction (partial trace) to the accessible region R , and the algebra to be corrected is the bulk algebra associated to EW ( R ). The key point is: the existence of recovery depends on the algebra and the code. D. Why stateand region-dependence are not flaws The dependence of reconstruction on region and state is often presented as puzzling. In our framework it is expected and conceptually necessary. Region-dependence. In QEC, which subsystem can correct which information is a structural feature of the code. Different regions R have different error-correcting capabilities. Thus reconstruction being region-dependent is not a failure of holography; it is the precise content of holography-as-QEC [10, 38]. State-dependence. The code subspace itself is commonly defined relative to a reference semiclassical geometry. In gravitational settings, the map between bulk effective degrees of freedom and boundary operators can depend on the background state. This is not an “ad hoc” feature; it is a manifestation of working in a restricted domain where bulk effective field theory is valid. When one tries to enlarge the domain toward “all states,” diagonal obstructions appear (Section III) and totality must fail. Debates about state-dependent interior operators can be read as concrete instantiations of this general non-totality pressure. Relative universality. The correct conceptual status of reconstruction is therefore: Universality holds within a context (code subspace, sector, region), not as a global, stateindependent decoder for all bulk predicates. This is exactly what Proposition V.1 predicts should happen once self-reference is present. E. Non-reconstructible operators as diagonal remainders We now connect the QEC viewpoint to diagonal logic. Diagonal arguments guarantee that, beyond a certain expressive threshold, any attempt at total classification produces predicates/operators that escape. In holography, the analogue is: There exist bulk operators/predicates that are not reconstructible from a given boundary region (or not reconstructible in a single state-independent manner) because doing so would constitute a total internal decoder. Two precise senses of “non-reconstructible.” (i) Outside the entanglement wedge: if an operator is supported outside EW ( R ), it is not expected to be representable on R without enlarging the accessible algebra. This is geometric non-reconstructibility. (ii) Outside the code domain: even if an operator is reconstructible on a small code subspace, there is no guarantee the same reconstruction exists on a much larger state space. Attempting to define a single reconstruction for all states is the totality demand obstructed by diagonal logic. 19 Diagonal remainder as a structural prediction. From our standpoint, “non-reconstructible operators” are not mysteries; they are the necessary remainder left when one imposes consistency constraints strong enough to avoid total internal decoding. This remainder may manifest as: •superselection/center data (edge modes), •contextual dependence of the reconstruction map, •or intrinsic limitations on jointly measurable interior observables. All of these appear across gauge theory, AQFT, and holography (Section IV). F. Takeaway Entanglement wedge reconstruction, understood through quantum error correction, is an explicit realization of “universality without totality”: • Bulk operators are reconstructible on a boundary region R relative to a code subspace and relative to an algebra of interest. • Regionand state-dependence are not pathologies; they are structural necessities once self-reference rules out total decoding. • Non-reconstructible operators are expected diagonal remainders—the precise place where the system refuses to close into a single global decoder. This prepares the ground for Section VII, where islands will be interpreted as a geometric mechanism that adjusts the reconstructible region itself to maintain consistency in an evaporating black hole. VII. ISLANDS AS A PHYSICAL REGULARIZATION OF NON-TOTALITY A. Goal and logical stance Sections III–VI developed a unified message: self-referential regimes obstruct total internal decoding (Proposition III.5), and modern holography already realizes this as relative universality via code subspaces and entanglement wedge reconstruction (Section VI) [ 10 , 13 , 38 ]. The purpose of this section is to show that the island mechanism is not merely a technical trick to reproduce the Page curve, but can be understood as a physical regularization of non-totality: islands implement, in semiclassical gravity, the only consistent alternative to “total decoding” by making reconstruction contextual and regime-dependent. This will be stated precisely as Proposition VII.1 below. B. The QES/island prescription: statement and meaning We first review the quantum extremal surface (QES) framework and the island formula at the level needed for our logical narrative. From RT/HRT to quantum extremal surfaces. For holographic CFT states with a semiclassical bulk dual, the Ryu–Takayanagi (RT) prescription [ 6 ] and its covariant Hubeny–Rangamani–Takayanagi (HRT) generalization [ 7 ] propose that the von Neumann entropy of a boundary region R is given, at leading order in 1/GN, by the area of an extremal surface γRhomologous to R: S(R)≈Area(γR) 4GN .(34) Quantum corrections were incorporated by Faulkner–Lewkowycz–Maldacena [ 8 ], leading to the generalized entropy functional Sgen(R;γ) := Area(γ) 4GN +SbulkΣγ,(35) 20 where Σ γ is a bulk region bounded by R∪γ and Sbulk is the bulk entanglement entropy of quantum fields across γin the semiclassical state. Engelhardt and Wall refined this into the QES prescription [ 9 ]: the relevant surface is a quantum extremal surface, i.e. a surface γ that extremizes Sgen , and the entropy is given by evaluating (35) on the minimizing quantum extremal surface. Formally: S(R) = min γ∈Ext Area(γ) 4GN +Sbulk(Σγ),(36) where Ext denotes the set of candidate (quantum) extremal surfaces. Islands. In evaporating black hole settings, the region R of interest is typically a region of radiation (outside the black hole), and one allows the bulk region Σ γ to include an “island” region I inside (or near) the black hole. The island formula is a specialization of (36): S(R) = min Iext Area(∂I) 4GN +Sbulk(R∪I),(37) where I ranges over candidate island regions, ∂I is their boundary (a QES), and Sbulk ( R∪I )is the bulk entropy of quantum fields on R∪I . In the modern derivations, (37) arises from gravitational replica computations and replica wormholes [14–16]. A key conceptual point. The optimization in (37) means that the entropy, and therefore the entanglement wedge of the radiation, is determined by a variational principle. Hence the reconstruction region itself becomes a stateand regime-dependent output rather than a fixed input. This is precisely the structural feature we need: non-totality is not patched; it is regulated by allowing the effective decoding region to adapt. C. Page curve from islands: the mechanism in one line We briefly recall how islands reproduce Page behavior [4]. Two competing saddles. In an evaporating black hole, there are typically two relevant saddles for (37) : (i) No-island saddle: I=∅, giving Sno-island(R) = Sbulk(R),(38) which grows with time as Hawking quanta accumulate. (ii) Island saddle: I6=∅, giving Sisland(R) = Area(∂I) 4GN +Sbulk(R∪I),(39) which can remain bounded and eventually decrease, because the bulk entropy of R∪I does not grow indefinitely when Icaptures the interior partners. The min/ext prescription (37) selects the smaller of these, producing a Page-like rise-and-fall. Logical reading. From our perspective, this competition is the physical manifestation of non-totality: the naive (no-island) decoding picture treats the radiation as a fixed subsystem from which one attempts total recovery; the island saddle corrects the notion of “what is encoded in the radiation” by making the reconstructible region state-dependent. Thus the Page transition is a context switch in reconstruction, not an information-loss event. D. Reconstruction region as a state-dependent output Entanglement wedge reconstruction (Section VI) tells us that bulk operators in EW ( R )are reconstructible from R(relative to a code subspace) [38]. The island prescription changes EW(R)itself. Entanglement wedge with islands. In the presence of an island I , the entanglement wedge of the radiation includes I , so operators supported in I become reconstructible from the radiation algebra (again, relative to a suitable code domain). Penington emphasized this perspective early in the island era [13]. 21 Why this is not “cheating”. The key is that the map “boundary region → reconstructible bulk region” is not fixed by kinematics alone; it is selected by the variational principle (37) , which depends on the state (via Sbulk ) and on the semiclassical geometry (via the area term). Thus, reconstruction is not a single global functor valid for all regimes; it is a family of reconstruction maps indexed by state/sector/context. This is exactly the pattern predicted by Proposition V.1. E. Monogamy, AMPS, and the pressure toward totality The AMPS firewall argument [5] can be read as diagnosing an inconsistent simultaneous demand: •purity/unitarity of the radiation after the Page time, •semiclassical EFT entanglement across the horizon, • and a single global factorization/decoding picture in which the same interior mode is simultaneously purified by both early radiation and near-horizon partners. This demand is a form of totality: it implicitly assumes that the decoding map from radiation to interior can be treated as globally valid and simultaneously compatible with the semiclassical horizon description. From the viewpoint of Sections III–V, this is exactly the regime where diagonal/self-reference constraints are expected to bite: the system is attempting to close on a single globally coherent decoder that works for all questions and all observers at once. Islands provide a concrete resolution by changing the effective reconstruction region so that one does not demand incompatible simultaneity of encodings. In other words, islands implement the abandonment of totality required for consistency. F. Key Proposition 3: Islands as a Diagonal Resolution Proposition VII.1 (Islands as a diagonal resolution of non-totality) . In semiclassical evaporating black hole settings where: (I1) the Hawking radiation R is treated as an internal code subsystem whose entropy is computed by a generalized entropy extremization principle (37), and (I2) entanglement wedge reconstruction is valid on an appropriate code domain (so that bulk operators in EW (R)can be represented on R) [38], the island mechanism provides a physical realization of the semantic non-totality constraints of Section III and Proposition V.1: Island formation implements a consistent alternative to total internal decoding by making the reconstructible bulk region a stateand regime-dependent output. Consequently, information about the system may be preserved and encoded in correlations without admitting a single global, uniform decoder valid across all regimes. Argument (structural, not a new replica derivation). We do not re-derive the replica wormhole calculation; those derivations are in [ 14 – 16 ]. Instead we show the logical role of islands. Assume one insists on a fixed notion of reconstruction from the radiation (a fixed decoding region and a globally uniform decoder). In the post-Page regime, this fixed-decoder demand conflicts with the simultaneous semiclassical entanglement structure at the horizon (AMPS-type pressure) [ 5 ]. By Section III, such a demand is precisely the kind of “total internal decoder” assumption that becomes inconsistent in maximally self-referential settings. The island formula (37) replaces the fixed-decoder demand by a variationally selected entanglement wedge EW ( R )that can include an interior island I . Thus the effective decoding map is not global but contextually selected (state/regime dependent). This is exactly the weakening of totality prescribed by Proposition V.1. Therefore island formation constitutes a physical regularization of non-totality: it preserves unitarity-compatible Page behavior [ 4 ] without requiring a single global internal decoder of all interior predicates from the radiation alone. 22 G. Takeaway Islands make precise what our diagonal/categorical narrative predicts: • The radiation can encode information about the system and yield a Page curve consistent with unitarity, but this does not imply the existence of a single globally defined decoder. • The reconstructible region is selected by a state-dependent extremization principle; hence reconstruction is inherently contextual and regime-dependent. • This contextuality is not an ad hoc patch; it is the physically realized alternative to the impossible demand of total internal decoding in a maximally self-referential system. In the next section we connect this to Higher Categorical Coherence Breakdown (HCCB) as a general physical mechanism by which coherence closure replaces global totality. VIII. HIGHER CATEGORICAL COHERENCE BREAKDOWN (HCCB) A. Goal: from semantic non-totality to a physical mechanism Sections III–VII established a semantic constraint: in sufficiently expressive self-referential regimes, one cannot demand total internal decoding of all semantic predicates about the system (Proposition III.5), and holography/islands implement a consistent alternative by making reconstruction contextual and regime-dependent (Propositions V.1 and VII.1). The remaining question is physical: What dynamical/structural mechanism produces (and stabilizes) this non-totality in the actual physics, while preserving local consistency and operational predictability? Higher Categorical Coherence Breakdown (HCCB) is our proposed answer [ 1 , 2 ]. In this section we (i) isolate the notion of coherence closure as the minimal consistency requirement, (ii) explain why demanding global symmetry/unity/totality is stronger than coherence and is generically obstructed, (iii) show how breakdown of higher coherence naturally yields sectorization and history dependence, and (iv) show why the operational effective dynamics becomes completely positive (CP) and typically semigroup-like on accessible algebras. B. Coherence vs symmetry vs totality We first separate three notions that are often conflated: Coherence (minimal consistency). In categorical language, coherence is the existence of consistent gluing/composition data: commuting diagrams and higher coherence conditions ensure that local identifications compose consistently [ 34 ]. Coherence is fundamentally about closure under composition: different ways of assembling a composite process yield canonically equivalent results. Symmetry (a sufficient but not necessary condition). Symmetry is a stronger property: invariance under a group (or groupoid, or higher group) action. Symmetry typically implies coherence (there are canonical identifications), but coherence can exist without a global symmetry principle. This is precisely the guiding insight we have emphasized in our broader program: symmetry is sufficient but not necessary for coherent gluing; coherence/closure is the minimal requirement. Totality (global closure under all queries). Totality is stronger still: it demands that a single global structure internalizes and decides all semantic predicates about the system (Sections II and III). Diagonal logic shows that in self-referential regimes, totality is generically impossible. Thus, the correct physical demand is not totality but coherence closure. C. Higher coherence and gluing: the mathematical core We now state the structural idea of HCCB. 23 Why “higher” coherence? Ordinary categorical coherence (1-categorical) controls associativity/unitality up to unique isomorphism (Mac Lane coherence) [ 34 ]. However, many physical constructions involve families of identifications and transformations between transformations: gauge choices, renormalization prescriptions, observer-dependent coarse-grainings, subsystem embeddings, and code-subspace reconstructions. These are naturally organized not in a mere category but in a 2-category or higher category (bicategory / (∞,1)-category), where: •objects are contexts (sectors, algebras, code subspaces, observers, scales), • 1-morphisms are admissible transitions between contexts (coarse-grainings, embeddings, encodings, reconstructions), • 2-morphisms (and higher) are coherence data (natural transformations between 1-morphisms, higher homotopies between transformations). Standard references for coherence in higher categories include [40]. Coherence closure vs coherence breakdown. Coherence closure means that all higher associativity/compatibility constraints can be satisfied (possibly up to controlled equivalence) so that any composite of transitions is well-defined independent of parenthesization/path. HCCB posits that, in sufficiently rich physical settings, these higher coherence conditions can fail: there exist closed loops in context space whose composite coherence data does not reduce to an identity (or does so only up to a nontrivial defect). In the physics language, the defect is a coherence curvature or holonomy gap in the space of descriptions. Local consistency, global obstruction. A key point is that coherence breakdown can occur even when each local piece is perfectly well-defined: each local patch of description is unitary/linear, but gluing them globally requires higher coherence that may be obstructed. This is the structural analogue of: •gauge/gribov-type global obstructions, •sheaf/descent obstructions (no global section), •diagonal obstructions (no total classifier). In other words, HCCB is the physical avatar of “universality without totality.” D. From coherence defects to loss of global unitary/linear semantics We now explain why coherence breakdown leads to an effective breakdown of global unitarity and linearity as a single total semantics, without claiming that local unitarity must fail in each patch. Patchwise unitarity. Suppose each context c carries a Hilbert-space (or algebraic) description with a unitary evolution Uc ( t ). If contexts are related by intertwiners (change-of-description maps) Wc→c0 , then global consistency would require higher coherence conditions such as Wc→c00 ≃Wc0→c00 ◦Wc→c0,(40) and analogous higher relations for the intertwiners themselves. Coherence breakdown means that around some loop c→c0→c00 →cone obtains a nontrivial defect: Wc→c:= Wc00 →c◦Wc0→c00 ◦Wc→c06≃ idc.(41) This defect obstructs the existence of a single globally defined unitary dynamics on a single Hilbert space that simultaneously represents all contexts. Why linearity breaks globally. Linearity is a property of a chosen state space (vector space/Hilbert space) and a chosen evolution map acting linearly on it. If there is no globally valid identification between contexts (due to higher coherence defects), then there is no single globally valid linear state space on which all dynamics can be represented. Instead one has: •a family of linear spaces (one per context) plus nontrivial gluing, •and defects that appear as effective nonlinearities when pushed into a single chart. Thus “breakdown of linearity” is not posited ad hoc; it is an emergent descriptor of attempting to force a global linear chart on a space that only admits local charts with nontrivial holonomy. 24 E. Emergence of CP flows, sectorization, and history dependence We now connect coherence defects to the operational appearance of CP dynamics and memory. Accessible algebras and coarse-graining. In practice, observers access a subalgebra Aacc of the full algebra (or a reduced state on a subsystem). The act of restricting to accessible degrees of freedom is a channel (Section II): ρ7→ ρacc = Φ(ρ),ΦCPTP.(42) If coherence defects prevent a globally consistent unitary identification across contexts, then from the viewpoint of Aacc the effective dynamics is generically non-unitary. The correct structural class for such reduced dynamics is completely positive maps, because complete positivity is precisely what guarantees consistency under extension by ancillas and avoids negative probabilities (Section II). Kraus/Stinespring representation. Any CPTP map admits a Kraus representation Φ(ρ) = X k Kkρ K† k,X k K† kKk=1,(43) equivalently a Stinespring dilation (8) [ 24 , 41 ]. This representation is not a model assumption; it is a theorem: CP evolution is exactly the class of physically admissible reduced evolutions. Semigroups and Lindblad form. When the effective dynamics is Markovian (memoryless) on the accessible algebra, one expects a one-parameter semigroup of CPTP maps { Φ t}t≥0 with Φ t+s = Φ t◦ Φ s and continuity at t = 0. The general structure theorem (Gorini–Kossakowski–Sudarshan–Lindblad) states that the generator L of such a uniformly continuous quantum dynamical semigroup has the Lindblad form dρ dt =L(ρ) = −i[H, ρ] + X αLαρL† α−1 2{L† αLα, ρ},(44) for some effective Hamiltonian H and noise (jump) operators Lα [ 42 , 43 ]. We emphasize: HCCB does not assume Markovianity; it predicts that coherence defects produce effective CP dynamics on accessible algebras, which may be history-dependent and non-Markovian. The semigroup/Lindblad case is a clean limit. Sectorization and superselection-like structure. Coherence defects generically imply that not all contexts can be globally identified; the system decomposes into sectors labeled by the holonomy/obstruction class. Operationally this appears as superselection-like structure: some observables distinguish sectors, but dynamics within a sector is well-defined while transitions between sectors are forbidden or contextdependent. History dependence. If the effective map on Aacc depends on the path taken in context space (e.g. which coarse-graining sequence, which reconstruction choice, which observer chain), then the reduced dynamics is non-Markovian. In HCCB language, memory is the operational trace of higher coherence holonomy. This matches the “directionality/history dependence” logic we previously emphasized in your work on operational geometry and related contexts. F. Key Proposition 4: Physical realization of diagonal non-totality Proposition VIII.1 (Physical realization via HCCB) . In sufficiently expressive self-referential quantum systems, diagonal logic forbids a total internal decoder of all semantic predicates about the system (Proposition III.5). HCCB provides a physical realization of this non-totality as follows: (P1) Local descriptions remain consistent and can be unitary/linear within contexts (patchwise semantics). (P2) Higher coherence conditions required to glue these contexts into a single global unitary/linear semantics fail, producing nontrivial coherence defects/holonomy. (P3) When restricted to accessible algebras or operational probes, these coherence defects manifest as intrinsically CP reduced dynamics, typically with sectorization and history dependence, rather than as a single global decoder. Thus diagonal non-totality is realized dynamically as higher categorical coherence breakdown, replacing “global unitarity as total semantics” by coherence closure as the minimal consistency requirement [1, 2]. 25 Proof sketch (structure-to-dynamics). Proposition III.5 forbids total internal decoding once self-reference is present. Attempting to enforce totality requires globally consistent identifications of all contexts and reconstructions. In HCCB, these identifications are encoded as higher morphisms and coherence data. If higher coherence fails (nontrivial holonomy (41) ), then no single global unitary/linear semantics exists that can serve as a universal decoder. Operational observers access reduced descriptions; reduction is necessarily CPTP (Stinespring/Kraus) [ 24 , 41 ], and therefore coherence-defect-induced context dependence yields CP effective dynamics on accessible algebras. In regimes where the reduction is approximately memoryless, the GKLS theorem gives the Lindblad generator (44) [ 42 , 43 ]. The sectorization and history dependence follow from the existence of distinct holonomy classes and path dependence of gluing. This is exactly the dynamical/operational imprint of non-totality. G. Takeaway HCCB supplies the missing physical bridge: •diagonal logic explains why total internal decoding is ill-posed in self-referential regimes; •categorical holography and islands show how gravity realizes relative decoding; • HCCB explains how the world maintains local consistency while abandoning global totality: global unitarity/linearity may fail as a single total semantics because higher coherence cannot be glued, and the operational remnant is CP dynamics with sectorization and memory. In Section IX we return to black holes as maximally self-referential systems where these mechanisms become unavoidable and geometric. IX. BLACK HOLES AS MAXIMALLY SELF-REFERENTIAL SYSTEMS A. Goal and synthesis statement We now synthesize the logical and physical threads developed so far. • Section III established the diagonal obstruction to total internal decoding of information about the system (Proposition III.5) using the shared Cantor/Gödel/Turing/Lawvere mechanism [ 17 – 19 , 25 ]. • Sections V–VI showed how holography naturally implements universality without totality via encoding functors, code subspaces, and entanglement wedge reconstruction [10, 11, 38]. • Section VII interpreted islands as a physical regularization of non-totality, making the reconstructible region a state-dependent output of a variational principle [13–16]. • Section VIII introduced HCCB as the physical mechanism realizing semantic non-totality through higher coherence breakdown and CP/sectorial operational dynamics [1, 2]. The remaining step is to explain why black holes are the regime in which these issues are maximally forced: the horizon converts self-reference from a merely semantic option into an unavoidable geometric/operational feature. B. Horizon as irreversible observer separation Causal separation and observer classes. An event horizon partitions spacetime into regions that are causally inaccessible to one another for certain observers. An exterior observer cannot access interior degrees of freedom directly. This is not a matter of experimental limitation but of spacetime causal structure. 32 3. Links to complexity and computation Diagonal logic is historically tied to computation (Turing) and incompleteness (Gödel) [ 17 , 18 ]. In the black-hole context, decoding is often argued to be computationally complex (e.g. the Hayden–Preskill decoding problem) [12]. A promising direction is to relate: •computational hardness of decoding (complexity barriers), •semantic non-totality (diagonal barriers), •and coherence defects (HCCB barriers), as complementary manifestations of the same underlying limitation: the system cannot fully internalize and execute a total decoder of its own semantic content. A rigorous bridge here would significantly strengthen the conceptual unification proposed in this paper. G. Closing perspective The main message is a shift in what we ask from fundamental theory. Rather than demanding global totality (a single internal decoder for all predicates), we should demand coherence closure (consistent gluing across contexts) and accept sectorization and contextual reconstruction as intrinsic where self-reference is unavoidable. Black holes are the place where this becomes geometrically sharp, but the phenomenon is universal across quantum theory. XII. CONCLUSION A. Summary of results We conclude by summarizing the logical structure of the paper and the main results, with emphasis on the distinction between (i) the existence/preservation of correlations and (ii) total internal decodability of semantic predicates about the system. (1) Definitions: what is (and is not) being constrained. In Section II we separated: • operational information as correlations and distinguishability (entropy, mutual information, relative entropy) [20, 22], • information about the system as semantic distinctions/predicates concerning states, observables, or histories, • encodability/decodability as the existence of a uniform internal representation and decoder for a specified query class. Diagonal constraints target the last notion (total internal decoding), not the first (existence of correlations). (2) Diagonal logic: the semantic no-go. In Section III we presented the shared diagonal mechanism behind Cantor, Gödel, and Turing and its categorical unification via Lawvere fixed points [ 17 – 19 , 25 ]. We formulated the resulting semantic obstruction as Proposition III.5: sufficiently expressive self-referential systems do not admit a single total internal decoder/classifier of all semantic predicates. (3) Categorical holography: universality without totality. In Section V we translated diagonal constraints into holographic language by modeling bulk-to-boundary encoding as a functor U : B → P and showing that “total holography” (a total classifier of all bulk predicates) is too strong in self-referential regimes. This was crystallized in Proposition V.1. (4) Modern holography already implements the correct logic. In Section VI we showed that entanglement wedge reconstruction, understood through the quantum error correction viewpoint, is intrinsically relative: it holds on code subspaces and depends on region and sometimes state [ 10 , 11 , 38 ]. Non-reconstructible operators are therefore not anomalies but expected “diagonal remainders.” (5) Islands as physical regularization of non-totality. In Section VII we interpreted islands and the QES prescription as a geometric implementation of contextual, regime-dependent reconstruction that avoids the over-demand of total decoding and reproduces Page-curve behavior in semiclassical gravity [9, 13, 14, 16]. This was formalized as Proposition VII.1. 33 (6) HCCB: a physical mechanism. In Section VIII we proposed Higher Categorical Coherence Breakdown (HCCB) as a dynamical/structural mechanism by which semantic non-totality is realized physically: global unitary/linear semantics may fail to glue coherently across contexts even when local descriptions remain consistent, yielding sectorization, history dependence, and CP effective dynamics on accessible algebras [1, 2]. This was summarized in Proposition VIII.1. (7) Black holes as maximally self-referential systems. In Section IX we argued that black holes sit at the extreme of the self-reference hierarchy: horizons enforce observer separation, Hawking radiation is an internal encoding, and decoding is an internal physical process. As a result, total decoding demands generate paradox, while relative decoding resolves it without information loss. Finally, in Section X we showed that the same “universality without totality” pattern appears beyond gravity in AQFT, gauge theory, measurement chains, and open quantum systems. B. Final perspective The conceptual shift advocated here is that what fails in the black-hole setting is not information itself, but the totality demand that information about the system be internalizable and uniformly decodable from within the same self-referential system. Once one replaces this demand by the weaker and more robust requirement of coherence closure, the modern picture of holography, entanglement wedge reconstruction, and islands fits naturally into a single logical narrative. Final sentence. Black holes do not destroy information; they expose a universal limit of information itself—namely, that in sufficiently self-referential quantum systems, information about the system can exist and be preserved in correlations without ever being totally decodable by a single internal, globally coherent, and uniformly valid decoding scheme. [1] A. T. Patrascu, “Higher-order categorical coherence breakdown: a geometric framework for nonlinear quantum mechanics and its applications to strongly correlated electron systems,” Eur. Phys. J. B 98 (2025) 210. doi:10.1140/epjb/s10051-025-01062-6. [2] A. T. Patrascu, “Higher Categorical Coherence Breakdown as a Quantum Process: Measurement without Classicality and Its Dual Role in Quantum Computing Limitations and Algorithmic Advantages,” Zenodo preprint (v1, published Aug. 2025). doi:10.13140/RG.2.2.11749.72168. [3] S. W. Hawking, “Breakdown of Predictability in Gravitational Collapse,” Phys. Rev. D 14 (1976) 2460–2473. doi:10.1103/PhysRevD.14.2460. [4] D. N. Page, “Average Entropy of a Subsystem,” Phys. Rev. Lett. 71 (1993) 1291–1294. doi:10.1103/PhysRevLett.71.1291. arXiv:gr-qc/9305007. [5] A. Almheiri, D. Marolf, J. Polchinski and J. Sully, “Black Holes: Complementarity or Firewalls?,” JHEP 02 (2013) 062. doi:10.1007/JHEP02(2013)062. arXiv:1207.3123 [hep-th]. [6] S. Ryu and T. Takayanagi, “Holographic Derivation of Entanglement Entropy from AdS/CFT,” Phys. Rev. Lett. 96 (2006) 181602. doi:10.1103/PhysRevLett.96.181602. arXiv:hep-th/0603001. [7] V. E. Hubeny, M. Rangamani and T. Takayanagi, “A Covariant Holographic Entanglement Entropy Proposal,” JHEP 07 (2007) 062. doi:10.1088/1126-6708/2007/07/062. arXiv:0705.0016 [hep-th]. [8] T. Faulkner, A. Lewkowycz and J. Maldacena, “Quantum corrections to holographic entanglement entropy,” JHEP 11 (2013) 074. doi:10.1007/JHEP11(2013)074. arXiv:1307.2892 [hep-th]. [9] N. Engelhardt and A. C. Wall, “Quantum Extremal Surfaces: Holographic Entanglement Entropy beyond the Classical Regime,” JHEP 01 (2015) 073. doi:10.1007/JHEP01(2015)073. arXiv:1408.3203 [hep-th]. [10] A. Almheiri, X. Dong and D. Harlow, “Bulk Locality and Quantum Error Correction in AdS/CFT,” JHEP 04 (2015) 163. doi:10.1007/JHEP04(2015)163. arXiv:1411.7041 [hep-th]. [11] D. L. Jafferis, A. Lewkowycz, J. Maldacena and S. J. Suh, “Relative entropy equals bulk relative entropy,” JHEP 06 (2016) 004. doi:10.1007/JHEP06(2016)004. arXiv:1512.06431 [hep-th]. [12] P. Hayden and J. Preskill, “Black holes as mirrors: quantum information in random subsystems,” JHEP 09 (2007) 120. doi:10.1088/1126-6708/2007/09/120. arXiv:0708.4025 [hep-th]. [13] G. Penington, “Entanglement Wedge Reconstruction and the Information Paradox,” arXiv:1905.08255 [hep-th]. [14] A. Almheiri, R. Mahajan, J. Maldacena and Y. Zhao, “The Page curve of Hawking radiation from semiclassical geometry,” arXiv:1908.10996 [hep-th]. [15] A. Almheiri, N. Engelhardt, D. Marolf and H. Maxfield, “The entropy of bulk quantum fields and the entanglement wedge of an evaporating black hole,” JHEP 12 (2019) 063. doi:10.1007/JHEP12(2019)063. arXiv:1905.08762 [hep-th]. [16] A. Almheiri, T. Hartman, J. Maldacena, E. Shaghoulian and A. Tajdini, “Replica Wormholes and the Entropy of Hawking Radiation,” JHEP 05 (2020) 013. doi:10.1007/JHEP05(2020)013. arXiv:1911.12333 [hep-th]. 34 [17] K. Gödel, “Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I,” Monatshefte für Mathematik und Physik 38 (1931) 173–198. [18] A. M. Turing, “On Computable Numbers, with an Application to the Entscheidungsproblem,” Proc. Lond. Math. Soc. 2 42 (1936) 230–265; erratum 2 43 (1937) 544–546. doi:10.1112/plms/s2-42.1.230. [19] F. W. Lawvere, “Diagonal arguments and cartesian closed categories,” Theory Appl. Categ. 15 (2006) 1–13; reprint of Lecture Notes in Mathematics 92 (1969) 134–145. Available at https://www.tac.mta.ca/tac/reprints/articles/15/tr15.pdf. [20] C. E. Shannon, “A Mathematical Theory of Communication,” Bell System Technical Journal 27 (1948) 379–423, 623–656. [21] M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press (2000). [22] M. M. Wilde, Quantum Information Theory, 2nd ed., Cambridge University Press (2017). [23] H. Umegaki, “Conditional expectation in an operator algebra. IV. Entropy and information,” Kodai Mathematical Seminar Reports 14 (1962) 59–85. [24] W. F. Stinespring, “Positive Functions on C∗-Algebras,” Proc. Amer. Math. Soc. 6(1955) 211–216. [25] G. Cantor, “Über eine elementare Frage der Mannigfaltigkeitslehre,” Jahresbericht der Deutschen Mathematiker-Vereinigung 1(1891) 75–78. [26] R. M. Smullyan, Gödel’s Incompleteness Theorems, Oxford University Press (1992). [27] M. Sipser, Introduction to the Theory of Computation, 3rd ed., Cengage Learning (2012). [28] E. P. Wigner, “Remarks on the Mind-Body Question,” in I. J. Good (ed.), The Scientist Speculates, Heinemann, London (1961). [29] D. Frauchiger and R. Renner, “Quantum theory cannot consistently describe the use of itself,” Nature Communications 9(2018) 3711. doi:10.1038/s41467-018-05739-8. arXiv:1604.07422 [quant-ph]. [30] H. Casini, M. Huerta and J. A. Rosabal, “Remarks on entanglement entropy for gauge fields,” Phys. Rev. D 89 (2014) 085012. doi:10.1103/PhysRevD.89.085012. arXiv:1312.1183 [hep-th]. [31] W. Donnelly and L. Freidel, “Local subsystems in gauge theory and gravity,” JHEP 09 (2016) 102. doi:10.1007/JHEP09(2016)102. arXiv:1601.04744 [hep-th]. [32] R. Haag, Local Quantum Physics: Fields, Particles, Algebras, 2nd rev. and enl. ed., Springer, Berlin (1996). [33] L. Susskind, L. Thorlacius and J. Uglum, “The stretched horizon and black hole complementarity,” Phys. Rev. D 48 (1993) 3743–3761. doi:10.1103/PhysRevD.48.3743. arXiv:hep-th/9306069. [34] S. Mac Lane, Categories for the Working Mathematician, 2nd ed., Springer, New York (1998). [35] S. Awodey, Category Theory, 2nd ed., Oxford University Press (2010). [36] P. T. Johnstone, Sketches of an Elephant: A Topos Theory Compendium, Oxford University Press (2002). [37] B. Jacobs, Categorical Logic and Type Theory, Elsevier (1999). [38] X. Dong, D. Harlow and A. C. Wall, “Reconstruction of Bulk Operators within the Entanglement Wedge in Gauge-Gravity Duality,” Phys. Rev. Lett. 117 (2016) 021601. doi:10.1103/PhysRevLett.117.021601. arXiv:1601.05416 [hep-th]. [39] C. Bény, A. Kempf and D. W. Kribs, “Generalization of the Knill–Laflamme conditions for quantum error correction,” Phys. Rev. Lett. 98 (2007) 100502. doi:10.1103/PhysRevLett.98.100502. arXiv:quant-ph/0608071. [40] T. Leinster, Higher Operads, Higher Categories, Cambridge University Press (2004). [41] K. Kraus, States, Effects, and Operations: Fundamental Notions of Quantum Theory, Lecture Notes in Physics 190, Springer (1983). [42] G. Lindblad, “On the generators of quantum dynamical semigroups,” Commun. Math. Phys. 48 (1976) 119–130. doi:10.1007/BF01608499. [43] V. Gorini, A. Kossakowski and E. C. G. Sudarshan, “Completely positive dynamical semigroups of N -level systems,” J. Math. Phys. 17 (1976) 821–825. doi:10.1063/1.522979.