AUTHOR: Evan T. Kotler
Foundational physical theories typically presuppose probability, identity, dynamics, and background structure. This paper addresses a prior question: what minimal structural conditions are required for any nontrivial, physics-like description to be possible at all? Working within a framework of finite relational systems, we impose a single standing requirement—stability under admissible extension—designed to exclude collapse without assuming time, dynamics, observers, probability, or primitive identity. We show that this requirement forces structural compression, rules out primitive identity, and yields a class of persistent equivalence structures stabilized under admissible transformations. Most significantly, when multiple such structures coexist, probability emerges necessarily as a public, compression-invariant frequency rather than as a primitive measure or subjective notion. No numerical constants, universality claims, or representational structures are assumed. Probability is derived as a structural consequence of non-collapse alone.
Most foundational frameworks in physics begin by assuming substantial structure: a space of states, a notion of dynamics, probabilistic laws, or identifiable objects persisting through time. These assumptions are often justified pragmatically, but they leave open a more basic question: what must be true before such assumptions even make sense?
This paper addresses that prior question. We do not ask how to interpret probability, or how to quantize a classical system. Instead, we ask:
What minimal structural constraints are required for any stable, nontrivial description of a system to be possible at all?
To answer this, we adopt a deliberately austere posture. We assume:
no spacetime,
no time or dynamics,
no observers or measurements,
no probability or randomness,
no primitive identity.
The only positive commitments are:
finiteness of effective distinguishability,
and a single standing requirement: stability under admissible extension.
The central result is that this requirement alone forces a sequence of consequences:
distinctions must be compressed,
primitive identity is ruled out,
persistence becomes an earned, approximate notion,
and probability emerges necessarily as a structural frequency.
The derivation is structural rather than interpretive. Probability is not assumed, postulated, or motivated epistemically; it is forced by the conditions required to avoid collapse.
A finite relational system is a triple [ \mathcal{S} = (\Omega, \mathcal{R}, \mathcal{C}), ] where:
(\Omega) is a finite set of relational configurations,
(\mathcal{R}) is a collection of relations among elements of (\Omega),
(\mathcal{C}) is a partially defined composition operation on (\mathcal{R}).
Elements of (\Omega) are not interpreted as objects, states, or events. All structure is relational.
Finiteness here refers to effective distinguishability, not to metaphysical discreteness. Infinite idealizations may arise later as limits, but they are not primitive.
An admissible extension is a map [ E : \mathcal{S} \to \mathcal{S}' ] that:
preserves finite effective distinguishability,
does not introduce unbounded relational dependence,
remains compatible with further admissible extensions.
Admissibility is structural, not dynamical.
A property (P) is structurally meaningful only if it is preserved under all admissible extensions.
This requirement excludes distinctions that survive only because the system has not yet been extended.
The notion of admissible extension has so far been characterized intensionally, by properties it must satisfy. For clarity and robustness, we now make explicit the structural role it plays.
Let (\mathfrak{E}) denote the admissible extension class, defined as the largest class of relational extensions satisfying:
Closure under composition: If (E_1, E_2 \in \mathfrak{E}), then (E_2 \circ E_1 \in \mathfrak{E}).
Preservation of effective finiteness: Extensions do not introduce unboundedly many distinguishable relational configurations.
Congruence preservation: If two configurations are behaviorally congruent prior to extension, no admissible extension renders them distinguishable.
No new invariant labels: Extensions may introduce relational structure, but not new primitive identifiers invariant under all further admissible extensions.
A property is structurally meaningful if and only if it is invariant under the full class (\mathfrak{E}).
This definition does not enumerate admissible extensions. Instead, it characterizes them by a closure principle: admissible structure is whatever survives all context enlargement compatible with finiteness and non-collapse. No probabilistic or dynamical assumptions are introduced.
Two configurations (x,y \in \Omega) are distinguishable if there exists an admissible extension under which their relational roles differ.
If no such extension exists, they are indistinguishable.
Define (x \sim y) iff (x) and (y) are indistinguishable under all admissible extensions.
Behavioral congruence is an equivalence relation.
Proof. Reflexivity, symmetry, and transitivity follow directly from the definition. ∎
Any admissible description of (\Omega) factors through the quotient [ \Omega / {\sim}. ]
Proof. If a description distinguishes two behaviorally congruent configurations, it encodes a distinction not preserved under admissible extension, violating the standing requirement. ∎
Compression is not epistemic. It is a structural necessity.
Primitive identity on (\Omega) is incompatible with stability under admissible extension.
Proof. If identity distinctions are preserved under all extensions, collapse follows. If they vary under extension, identity fragments. If they fail stability, they are inadmissible. ∎
Identity, if it exists, must be derived as compression-stable equivalence.
A survivor is an equivalence class in (\Omega/{\sim}) invariant under all admissible extensions.
The stabilizer of a survivor is the set of admissible transformations leaving it invariant.
Survivors represent the only structures that can persist without collapse.
Let (\Sigma) denote the set of survivors (behavioral congruence classes invariant under admissible extension).
An admissible weighting is a function [ w : \Sigma \rightarrow \mathbb{R}^+ ] satisfying:
Normalization: (\sum_{\sigma \in \Sigma} w(\sigma) = 1),
Stabilizer invariance: (w(\sigma)) is invariant under admissible renaming,
Extension invariance: (w) is unchanged under all admissible extensions in (\mathfrak{E}).
There exists no admissible weighting on (\Sigma) that assigns unequal weights to distinct survivors.
Assume, for contradiction, that (w(\sigma_1) \neq w(\sigma_2)) for two survivors (\sigma_1, \sigma_2 \in \Sigma).
By definition of survivorship, both (\sigma_1) and (\sigma_2) are invariant under admissible extension. However, admissibility places no structural distinction between survivors beyond their invariance.
Because admissible extensions are closed under composition and do not introduce new invariant labels, there exists an admissible extension (E \in \mathfrak{E}) that permutes representatives of (\sigma_1) and (\sigma_2) while preserving behavioral congruence. (Intuitively: admissible extensions may reorganize relational neighborhoods without creating new invariants.)
Under such an extension, the unequal assignment (w(\sigma_1) \neq w(\sigma_2)) is not preserved, violating extension invariance.
Thus, any admissible weighting must assign equal weight to all survivors. ∎
Remark (Genericity of Symmetric Reorganization). The argument above assumes that the admissible extension class 𝔈 is sufficiently rich to allow symmetric reorganization of relational neighborhoods that does not introduce new invariant labels. This holds for any relational language in which distinctions are encoded relationally rather than by rigid primitives. If the relational signature were so impoverished as to forbid such reorganizations, survivors would already carry additional invariant structure, contradicting their definition as purely compression-stable classes.
Structural probability is the unique admissible weighting on (\Sigma), given by [ P(\sigma) = \frac{1}{|\Sigma|}. ]
Equivalently, structural probability is the relative frequency of representatives in each survivor class.
Remark (Role of Cardinality). Cardinality enters not as a probabilistic assumption but as the primitive measure of effective distinguishability. Because finiteness of distinguishability is assumed at the outset, the size of a preimage class records how many distinct relational roles have been irreversibly compressed into a survivor. No alternative quantitative notion of multiplicity is available without introducing additional structure.
In any finite relational system admitting more than one survivor, probability emerges uniquely as uniform, compression-invariant frequency.
By Lemma 8.2, no non-uniform admissible weighting exists. Uniform frequency is therefore the only structurally meaningful summary of multiplicity compatible with admissible extension. ∎
No appeal is made to indifference, symmetry principles, randomness, or epistemic uncertainty. Uniformity is not assumed; it is forced by invariance under admissible extension.
This paper:
derives probability without assuming it,
introduces no universality claims,
introduces no information-theoretic or representational structure.
Those questions are addressed separately.
Probability is often treated as fundamental. This work shows it is not. When collapse is forbidden and admissible extension is taken seriously, probability is unavoidable—but only as frequency. This result establishes a minimal foundation on which information, universality, and representation may be layered without conceptual circularity.
