Identical Ancestor Point Axiom

• The Identical Ancestor Point Axiom is a structural rule stating that for any organism in a set, almost all members are either its descendants or non-descendants.
• It complements a convexity condition to form specieslike clusters in a directed graph model of an infinite biosphere, where vertices represent organisms.
• The framework reduces subjectivity in species classification by using generator sets, common ancestry, and reflection properties to control internal genealogical splits.

Searching arXiv for the primary paper and closely related genealogy/common-ancestor work to ground the article in current preprints. The Identical Ancestor Point Axiom (IAP axiom) is a structural condition on subsets of an organismal genealogy, introduced in Alexander’s "Specieslike clusters based on identical ancestor points" (Alexander, 5 Feb 2026). In that framework, the ambient object is the directed graph of all organisms—past, present, and future—where edges represent biological parenthood, and a candidate species is represented by a subset of vertices. The axiom states that for every organism in such a subset, either the subset contains only finitely many descendants of that organism, or only finitely many non-descendants of that organism. Together with a convexity condition, the axiom is used to define specieslike clusters, and the paper argues that this pair of axioms reduces the subjectivity of species in a technical sense (Alexander, 5 Feb 2026).

1. Formal setting in the infinite biosphere

The IAP axiom is formulated inside an infinite biosphere, which is a directed graph $G$ equipped with a birthdate function $t:G\to\mathbb{R}$ (Alexander, 5 Feb 2026). Each vertex is an organism, and an edge $v\to w$ means that $v$ is a biological parent of $w$. The biosphere satisfies four background conditions: parents are older than children; only finitely many organisms exist before any given time; each organism has finitely many children; and the graph is infinite, reflecting the simplifying assumption that life does not go extinct (Alexander, 5 Feb 2026).

An organism $v$ is an ancestor of $w$ if there is a directed path

$v=v_1\to v_2\to \dots \to v_n=w,\qquad n>1,$

and $w$ is then a descendant of $v$ (Alexander, 5 Feb 2026). A subset $t:G\to\mathbb{R}$0 is treated both as a set of vertices and as the induced subgraph. Connectivity is taken in the underlying undirected sense: $t:G\to\mathbb{R}$1 is connected if any two vertices in $t:G\to\mathbb{R}$2 can be joined by a path staying in $t:G\to\mathbb{R}$3 when edge directions are ignored (Alexander, 5 Feb 2026).

Within this abstraction, a putative species is simply a subset $t:G\to\mathbb{R}$4. The central project is therefore not taxonomic description by morphology or ecology, but the search for axioms that make a purely genealogical subset biologically reasonable as a species (Alexander, 5 Feb 2026). The IAP axiom is one of the two main axioms in that program.

2. Definition of the Identical Ancestor Point Axiom

For a subset $t:G\to\mathbb{R}$5, the Identical Ancestor Point Axiom is:

For every $t:G\to\mathbb{R}$6, at least one of the following holds: - all but finitely many members of $t:G\to\mathbb{R}$7 are descendants of $t:G\to\mathbb{R}$8, or - all but finitely many members of $t:G\to\mathbb{R}$9 are non-descendants of $v\to w$0 (Alexander, 5 Feb 2026).

Equivalently,

$v\to w$1

or, in positive form,

$v\to w$2

(Alexander, 5 Feb 2026).

The intuitive content is that each organism in $v\to w$3 must be genealogically either eventually negligible or eventually ubiquitous within $v\to w$4. What is excluded is a permanently mixed position in which one organism has infinitely many descendants in $v\to w$5 and simultaneously coexists with infinitely many members of $v\to w$6 that never descend from it (Alexander, 5 Feb 2026).

Alexander’s terminology is motivated by classical identical ancestor point results in population genetics, as discussed in the paper through Chang 1999 and Rohde et al. 2004. In that literature, an identical ancestor point is a time in the past by which every individual alive then is either a common ancestor of all people alive today or a common non-ancestor of all of them. Alexander’s axiom does not define a single time slice; instead, it imposes a structural property on an arbitrary subset $v\to w$7 of the entire biosphere (Alexander, 5 Feb 2026). The paper accordingly presents the axiom as a strong formalization of lack of permanent splits.

3. Convexity and the definition of specieslike clusters

The second main axiom is the convexity axiom: $v\to w$8 (Alexander, 5 Feb 2026).

This says that if some ancestor of $v\to w$9 is in $v$0 and some descendant of $v$1 is in $v$2, then $v$3 itself must lie in $v$4. In genealogical terms, a path through ancestry cannot leave the set and later return to it; the paper interprets this as forbidding “out-and-back” behavior and relates the intuition to Dollo’s law (Alexander, 5 Feb 2026).

A specieslike cluster is then defined as a subset $v$5 that is connected, satisfies the IAP axiom, and satisfies the convexity axiom (Alexander, 5 Feb 2026). These are proposed as genealogically natural candidates for species, abstracting away from non-genealogical criteria.

The framework admits trivial examples: every singleton $v$6 is a specieslike cluster, since it is connected and vacuously satisfies both IAP and convexity (Alexander, 5 Feb 2026). The paper therefore emphasizes maximal specieslike clusters, meaning specieslike clusters not properly contained in larger ones. However, IAP together with convexity and connectedness alone does not guarantee that every organism belongs to a maximal specieslike cluster; the paper gives a counterexample showing that some organisms can lie only in endlessly extendable clusters (Alexander, 5 Feb 2026).

This construction matters because it shows that the IAP axiom is not intended as an isolated combinatorial property. Its role is specifically as one half of a joint structural criterion, with convexity providing genealogical closure and IAP preventing enduring internal bifurcation.

4. Generators and the reduction of subjectivity

For $v$7, a vertex $v$8 is a generator of $v$9 if $w$0 contains at most finitely many non-descendants of $w$1; equivalently, almost all members of $w$2 are descendants of $w$3. The generator set is written $w$4 (Alexander, 5 Feb 2026).

The IAP axiom yields an immediate simplification: if $w$5 satisfies IAP and $w$6 has infinitely many descendants in $w$7, then $w$8 (Alexander, 5 Feb 2026). In other words, inside an IAP set, any organism with infinite genealogical reach must in fact generate almost all of the set.

Using generators, the paper defines an equivalence relation on infinite subsets: for infinite $w$9,

$v$0

(Alexander, 5 Feb 2026). Two sets are therefore equivalent if they have almost the same generators.

The principal result is the Objective Species Theorem: if $v$1 and $v$2 are infinite specieslike clusters and $v$3 is infinite, then $v$4 (Alexander, 5 Feb 2026). The proof uses both IAP and convexity. The technical significance is that two infinite specieslike clusters cannot share infinitely many central generating organisms while remaining genuinely distinct, except for finitely many anomalies.

This is the sense in which the paper says that IAP together with convexity reduces the subjectivity of species (Alexander, 5 Feb 2026). The reduction is not a claim of uniqueness of one global species partition. Rather, it is a constraint on overlap: once one restricts attention to infinite specieslike clusters, heavy overlap in generators forces equivalence. A plausible implication is that the framework replaces unrestricted taxonomic discretion with a tightly controlled equivalence structure.

5. Maximal clusters, common ancestor, and reflection

To obtain maximal clusters containing every organism, the paper adds two further constraints beyond IAP and convexity (Alexander, 5 Feb 2026).

The Common Ancestor Property (CA) requires that there is a vertex $v$5 such that every other $v$6 is a descendant of $v$7; the set thus has a unique common ancestor of all its members (Alexander, 5 Feb 2026). The Reflection Property (REF) requires that if $v$8 has infinitely many descendants in the ambient biosphere $v$9, then $w$0 also has infinitely many descendants in $w$1 (Alexander, 5 Feb 2026). REF is interpreted in the paper as ruling out non-representative subsets that truncate an organism’s genealogical future.

The paper then studies nonempty sets satisfying all four conditions: $w$2 (Alexander, 5 Feb 2026). Such a set is connected by CA, specieslike by IAP and convexity, rooted by a unique common ancestor, and reflective of infinite descendant lines.

The main existence theorem states that for every $w$3, there exists an $w$4-maximal set containing $w$5 (Alexander, 5 Feb 2026). Thus, under this four-axiom regime, every organism belongs to some maximal genealogically admissible cluster.

The paper also proves a minimality result. If IAP is dropped, the maximal sets become too trivial; and if IAP is retained but any one of convexity, common ancestor, or reflection is removed, there are infinite biospheres in which some organism belongs to no maximal set satisfying the remaining axioms (Alexander, 5 Feb 2026). Within the framework of the paper, the quartet $w$6 is therefore minimal for the coverage theorem.

6. Biological motivation, examples, and limits

The paper gives five informal arguments for why real biological species might satisfy the IAP axiom: identical ancestor point phenomena in roughly panmictic populations; isolation scenarios such as colonies or islands that eventually speciate; Hennig’s no-split intuition; a probabilistic argument based on biased mixing between descendants and non-descendants; and a combinatorial $w$7-ancestors argument for genealogical saturation (Alexander, 5 Feb 2026). These are explicitly presented as heuristics rather than proofs.

A central interpretive point is that IAP concerns genealogical ancestry, not genetic ancestry. The paper stresses that an organism can be an ancestor of almost everyone without contributing genetic material to many of them, invoking the contrast between genealogical and genetic ancestry through the Gravel–Steel observation mentioned there (Alexander, 5 Feb 2026).

The stylized examples in the paper illustrate the partitioning behavior of the axioms. In a single two-way split, there are two maximal clusters sharing the entire pre-split history, but only finitely many individuals are shared, so the objectivity theorem is not violated (Alexander, 5 Feb 2026). In a one-way split, one branch continually contributes individuals to another, and the interaction of convexity and reflection with IAP prevents the receiving branch from simply absorbing the entire pre-split lineage (Alexander, 5 Feb 2026). These examples show that the axiom is designed to forbid eternal internal splits while still allowing nontrivial branching histories.

The paper also states clear caveats. The infinite-future assumption is a simplifying device; purely asexual lineages are a challenge because the heuristics for IAP rely largely on sexual or occasionally sexual reproduction; and the author does not claim that $w$8 is the only reasonable axiom set (Alexander, 5 Feb 2026). One open direction explicitly identified is to obtain an existence theorem without requiring the common ancestor property.

7. Relation to other common-ancestor frameworks

Alexander’s IAP axiom is a property of subsets of a genealogical graph, not a general theorem about all directed acyclic graphs. This distinguishes it from information-theoretic work such as "Information-theoretic inference of common ancestors" (Steudel et al., 2010), where strong redundancy in mutual information implies that any compatible DAG must contain a common ancestor of at least $w$9 observed variables. That result is existential and observational: it infers the necessity of ancestral overlap from dependence patterns. By contrast, Alexander’s axiom stipulates a structural dichotomy inside a candidate species subset [(Alexander, 5 Feb 2026); (Steudel et al., 2010)].

The axiom also differs from branching-process descriptions of common ancestry. In "The coalescent point process of branching trees" (Lambert et al., 2011), coalescent times $v=v_1\to v_2\to \dots \to v_n=w,\qquad n>1,$0 encode the genealogy of neighboring individuals in Bienaymé–Galton–Watson and continuous-state branching settings, and MRCA depths are extracted from maxima of coalescence times. In "The Feller diffusion conditioned on a single ancestral founder" (Burden et al., 28 Apr 2025), a single-founder conditioning produces an exact ancestral concentration in which all current individuals descend from one founder. These frameworks analyze coalescence, MRCA times, and ancestor counts, but they do not formulate species as connected convex subsets satisfying a descendant/non-descendant dichotomy [(Lambert et al., 2011); (Burden et al., 28 Apr 2025)].

A further contrast appears in "Common ancestor type distribution: a Moran model and its deterministic limit" (Cordero, 2015). There, the core fact is that at each time in a finite Moran model there exists almost surely a unique individual whose descendants eventually occupy the entire population, and the paper studies the type distribution of that common ancestor (Cordero, 2015). This again concerns forward fixation and ancestral type bias, rather than the combinatorial condition that no member of a species subset may have both infinitely many descendants and infinitely many non-descendants within that subset.

Taken together, these comparisons suggest that the Identical Ancestor Point Axiom occupies a specific niche. It is neither a generic causal-DAG inference principle nor a coalescent-time statistic nor a fixation theorem. It is a genealogical axiom for defining specieslike subsets of an infinite biosphere, with the distinctive consequence that permanent internal lineage splits are ruled out inside any admissible cluster (Alexander, 5 Feb 2026).