2.  Property Differentiation: The Sole Primitive

Before there is space, or time, or matter, there has to be at least one difference. This chapter says what a difference is when you are not allowed to use anything else to explain it. Two things differ simply because they are not the same thing. Nothing more is needed, and nothing less would do.

2.1.  Differentiation by non-identity alone

A property is differentiated from another not by virtue of occupying different positions in space or occurring at different times (positions and times do not yet exist), but simply by being other. Two properties are differentiated if and only if they are not identical. Nothing further is required: no metric distance, no causal connection, no ordering relation, no extensional structure of any kind. Property differentiation is the minimal condition under which the concept of a distinction is meaningful, and it is the sole primitive of the precausal substrate.

The concept of otherness that this definition invokes is purely logical. It is the primitive of non-identity: $a$ is other than $b$ if and only if it is not the case that $a=b$. This does not require the two properties to be distinguishable by any observer, separable in any physical sense, or related by any causal or spatial connection. It requires only that identity fails. Two properties that are distinct in this sense need share nothing beyond their mutual non-identity; they carry no further structure, no mass, no charge, no location, no duration. The precausal substrate is therefore a domain in which the only thing that is true is that some properties are not the same as others. All of physics, in the PST programme, must be derivable from that single fact.

This choice of primitive is not arbitrary, and it is not merely the result of a preference for minimality. It is the result of asking what must be true of any domain that is to give rise to structure at all, and then identifying the unique weakest condition that satisfies the requirement. The requirement is this: a domain from which structure can emerge must be capable of nontrivial differentiation. If every element of the domain were identical to every other, the domain would be featureless and nothing could arise from it. Property differentiation is precisely the weakest condition under which nontrivial differentiation is possible. It is weaker than any spatial, causal, or metric structure, and it is not derivable from anything weaker than itself: any domain that contains less structure than property differentiation contains no nontrivial distinctions at all and is, in effect, the null domain from which nothing can arise. Property differentiation is therefore not merely minimal; it is the unique minimal structure with the required generative capacity. The substrate is genuinely precausal: it exists in a mode of being in which causality has no purchase, because causality requires a temporal ordering of events, and temporal ordering requires positions and durations, neither of which are available at this level.

2.2.  The substrate triple $(D,\delta,v)$

The substrate is the triple $(D,\delta,v)$:

• $D$ is a set of primitive properties.

• $\delta\colon D\times D\to\{0,1\}$ is the distinction relation, symmetric and irreflexive:

  $$\delta(a,b) \;\implies\;\delta(b,a)$$ (4)

  $$\neg\,\delta(a,a)$$ (5)

• $v\colon D\to V_{7}$ is the direction assignment: each property $a\in D$ has inherent directional content $v(a)\in V_{7}$ in the 7-dimensional direction algebra.

• $\delta$ is faithful to $v$: a difference in directional content is recorded as a distinction,

  $$v(a)\neq v(b) \;\implies\;\delta(a,b)$$ (6)

One condition beyond (4) and (5) is imposed on $\delta$, and it is a condition of coherence with $v$ rather than of structure. Faithfulness (6) adds nothing to $\delta$ that is not already in the triple: it requires only that the relation whose job is to record distinctions record the ones the triple carries. Beyond it nothing is imposed. There is no transitivity requirement, no totality requirement, no metric on $D$, no topology, and no ordering.

Faithfulness is a premise and is carried as one (Appendix C). It is not a theorem of (4) and (5): the two-property substrate with $\delta$ empty and $v(a)\neq v(b)$ satisfies both and violates it (Computation 184). Its motivation is the one just given, that without it $\delta$ is not the distinction relation this chapter names; its consequence for Born agreement was found afterwards, in Computation 184, and is recorded in that order deliberately. What it buys structurally is set out there and is worth stating here, because a reader will ask: contraposed, (6) says that two properties carrying no distinction between them share their directional content, so $\delta$ contains the complete multipartite graph on the partition of $D$ by $v$-value and the only freedom left lies inside a $v$-block. Where $v$ is injective, $\delta$ is complete. The direction assignment $v$ replaces a labelled-set structure with a $V_{7}$-vector content; without it, distinctions would carry magnitude only (“this property is distinct”) with no record of what kind of distinction. With $v$, each distinction is a vector in the maximal Hurwitz direction algebra, structured by the octonion vector product.

2.3.  The direction algebra $V_{7}$

$V_{7}\cong\mathbb{R}^{7}$ as a vector space, equipped with:

• a positive-definite inner product $\langle\cdot,\cdot\rangle$,

• a bilinear vector product $\times\colon V_{7}\times V_{7}\to V_{7}$ satisfying $\langle u\times v,u\rangle=\langle u\times v,v\rangle=0$ and $|u\times v|^{2}=|u|^{2}|v|^{2}-\langle u,v\rangle^{2}$.

This is the unique 7-dimensional algebra satisfying these properties (Hurwitz); equivalently, it is the vector part of the octonions $\mathbb{O}$ under octonion multiplication. The automorphism group of $V_{7}$ (preserving both the inner product and the vector product) is the exceptional Lie group $G_{2}$, of real dimension $14$. The choice of dimension $7$ for the direction algebra is forced: no other dimension supports a maximal normed-product structure over $\mathbb{R}$.

2.4.  Configurations and configuration space

A configuration $C\subseteq D$ is a collection of properties considered jointly. The space of configurations is $\mathcal{C}=\mathcal{P}(D)$. The total direction content of a configuration is the sum

$$v(C)\;=\;\sum_{a\in C}v(a)\;\in\;V_{7}.$$ (7)

This is the natural $V_{7}$-valued aggregate of the directional content of the properties in $C$. The asymmetric tension functional of P2 is structurally related to (but distinct from) $v(C)$: $T(C)$ encodes the magnitude and directional bias of the configuration’s tension, with the magnitude governing modal sublimation and the direction inheriting from $v$ via the asymmetric tension construction of Chapter 3.

It is within the configuration space $\mathcal{C}$ that the modal structure of PST is expressed. These are not temporal dynamics; they concern what is possible, what is necessary, and what cannot remain uninstantiated.

2.5.  The canonical measure on configurations

Before assembling the functional (Chapter 4), the measure $\mu$ on $\mathcal{C}$ must be fixed. The variational formulation demands integration over all configurations, and the choice of measure is not arbitrary: it must follow from the structure of $(D,\delta)$ alone, without invoking any geometric, probabilistic, or metric primitive that would introduce a new assumption at the level of the substrate.

The configuration space $\mathcal{C}=\mathcal{P}(D)$ is the power set of $D$: the collection of all subsets of the property domain. The structure $(D,\delta)$ carries a natural symmetry group, its automorphism group $\mathrm{Aut}(D,\delta)$, consisting of all bijections $f:D\rightarrow D$ that preserve the distinction relation: $\delta(a,b)\iff\delta(f(a),f(b))$ for all $a,b\in D$. Every such automorphism lifts canonically to a bijection on $\mathcal{C}$ by acting elementwise: $f\cdot C=\{f(a):a\in C\}$. Since the only primitive is $(D,\delta)$, and since the definition of a configuration is simply a subset of $D$, the measure $\mu$ must be invariant under this lifted action: no configuration should receive more weight than any automorphically equivalent configuration.

The key observation is that in the precausal substrate, all properties in $D$ are structurally equivalent: the distinction relation $\delta$ is irreflexive and symmetric but otherwise unconstrained, so there is no intrinsic difference between any one property and any other beyond their bare identity. No property is marked, distinguished, or preferred. This means that the full symmetric group on $D$ is contained in $\mathrm{Aut}(D,\delta)$ as a subgroup: any permutation of $D$ is an automorphism, since $\delta$ treats all pairs equally. A measure invariant under all permutations of $D$ is invariant under all relabellings of the properties, which is the minimal requirement for a measure that introduces no additional structure beyond $(D,\delta)$.

For a finite domain $D$ of cardinality $n$, the canonical measure on $\mathcal{P}(D)$ is the uniform measure, assigning equal weight to all $2^{n}$ subsets:

$$\mu(C)=2^{-n}\qquad\text{for all }C\in\mathcal{C}$$ (8)

Permutation invariance alone does not force this: any weighting that depends only on the cardinality $|C|$ is permutation-invariant. What forces it is the per-property form of complementation invariance. The primitive $(D,\delta)$ draws no distinction between the presence and the absence of each individual property, so for every $a\in D$ the single-property flip $\sigma_{a}\colon C\mapsto C\mathbin{\triangle}\{a\}$ is a symmetry of the substrate. The group generated by the flips acts transitively on $\mathcal{P}(D)$, so the unique flip-invariant probability measure is the uniform one; permutation invariance then holds automatically.

For an infinite domain $D$, the natural extension is the Bernoulli product measure $\mu=\bigotimes_{a\in D}\mathrm{Bern}(1/2)$, the unique probability measure on $\{0,1\}^{D}$ invariant under all permutations of the index set $D$ and under every single-property flip $\sigma_{a}$. The flips act transitively on every finite cylinder, forcing the uniform marginal on each finite window, and Kolmogorov extension then gives the product measure:

$$\mu\bigl{(}\{C\in\mathcal{C}:C_{0}\subseteq C\}\bigr{)}=2^{-|C_{0}|}\qquad\text{for all finite }C_{0}\subseteq D$$ (9)

This is the projective limit of the uniform measures on all finite sub-domains of $D$.

The value $1/2$ for the Bernoulli parameter is not arbitrary: it is the unique value for which $\mu$ is invariant under complementation, $C\mapsto\bar{C}=D\setminus C$. Complementation invariance is required by the structure of property differentiation: the distinction relation $\delta(a,b)$ is symmetric, and no configuration is intrinsically more natural than its complement. A configuration $C$ and its complement $\bar{C}$ represent exactly the same collection of pairwise distinctions, viewed from two opposite orientations that are not distinguished by anything in $(D,\delta)$. Assigning $\mu(C)\neq\mu(\bar{C})$ would introduce a polarity into the substrate that has no basis in the primitive. The Bernoulli parameter $1/2$ is therefore forced by complementation invariance alone, independently of the permutation argument.

The derivation may be summarised as a theorem.

2.5.1.  Theorem (Canonical Measure)

Let $(D,\delta)$ be a precausal property domain with $\delta$ symmetric and irreflexive. The unique probability measure $\mu$ on $\mathcal{C}=\mathcal{P}(D)$ that is (i) invariant under all automorphisms of $(D,\delta)$ and (ii) invariant under every single-property complementation $\sigma_{a}\colon C\mapsto C\mathbin{\triangle}\{a\}$, $a\in D$ (the global complementation $C\mapsto\bar{C}$ is their composition), is the Bernoulli product measure $\mu=\bigotimes_{a\in D}\mathrm{Bern}(1/2)$.

The proof is immediate from (ii) alone: the group generated by the single-property flips acts transitively on $\mathcal{P}(D)$ for finite $D$, and on every finite cylinder for infinite $D$, so an invariant measure weights all configurations (all cylinders over a given window) equally, which is the Bernoulli product measure; (i) then holds automatically. Condition (ii) is the exact measure-theoretic expression of P1’s symmetry between the presence and the absence of each distinction. The per-property form is essential: global complementation together with permutation invariance would not suffice (a mixture supported on $\{\emptyset,D\}$ satisfies both), which is why the flip invariance is the honest hypothesis.

The measure $\mu$ is not a free parameter or an auxiliary assumption: it is uniquely determined by the symmetry structure of $(D,\delta)$, without invoking any concept from outside the precausal domain.

2.6.  Summary

This chapter develops P1, the property differentiation primitive, and assumes only P1. No space, no time, no energy, no forces, and no laws of nature are assumed here. PST’s sole ontological primitive is the capacity for distinctions to exist; from this alone, the substrate’s internal structure (the set $D$ of primitive properties and the Bernoulli measure $\mu$) is derived. The Bernoulli measure $\mu=\bigotimes_{a}\mathrm{Bern}(\tfrac{1}{2})$ is derived as the unique measure on $\mathcal{P}(D)$ invariant under relabelling and per-property complementation; adopting the direction algebra $V_{7}=\mathrm{Im}(\mathbb{O})$ as the maximal directional structure (§1.5) was carried as a modelling choice until Computation 168, and is now a consequence of P1: a proper subalgebra such as the cross product $V_{3}$ on $\mathbb{R}^{3}$ costs a selector P1 does not supply, and $V_{7}$ costs none. What remains upstream is the norm-multiplicativity that places the substrate in the Hurwitz classification at all (§1.7, §2.3). No conjectures enter at the postulate level. The two further postulates, P2 (asymmetric vector tension, Chapter 3) and P3 (modal sublimation, Chapter 4), supply the substrate’s asymmetry and its threshold-crossing transition respectively.