Physical realization of minimal loop evolution (PE-A): multiple abiotic architectures satisfying a pre-frozen operational criterion

Abstract

Version 0.3 supersedes v0.2 after a correction. Following publication, the corresponding author of one primary study clarified that systems A and B of the competitive-exclusion work contain different replicator sets. The A/B contrast is therefore not a matched intervention/control pair, and architecture A1 has been downgraded from PASS to QUALIFIED. Independent clean passes: 2, which still meets the registered threshold of at least two. None of the five conditions is refuted; the requirement of a complete 5/5 within one compatible system is not met for A1. Full account in section 12 of the manuscript. Nature of the work. This is a retrospective comparative reanalysis of already published experiments, not a new laboratory study. All experimental effects were discovered and reported by the cited original authors. The contribution is (i) an operational criterion frozen in advance, (ii) the classification of independently published systems against it, and (iii) a reproducible recomputation from open primary data. Evolution is usually recognized through organisms, reproduction, inheritance and genomes. This work asks a question that is at once more general and narrower: does the physical world contain a minimal class of historical continuation in which distinguishable alternatives share a finite constraint and, through their own physical consequences, change their subsequent representation without any externally designated winner? This class is denoted PE-A. Origin of the criterion. PE-A was not abstracted from the systems it classifies. It was produced by a prior constructive digital programme: a finite simulated medium in which a distributed material state written by the system's own history proved causally necessary for maintaining its boundary (doi:10.5281/zenodo.21414141), and an integer-exact digital crucible in which alternative continuations under one shared material budget changed their representation through their own consequences, with each retained layer a causally necessary basis for the next. Independent blind recomputation of the crucible returned no failed gate, with one boundary carried forward here: under a consequence-equalized control the network topology was identical to live, differing only in edge weights. Consequences changed how much material each continuation carried, not which operators came into existence, which is precisely the PE-A / PE-B distinction retained in this paper. A constructive proof in a designed physics establishes possibility, not physical realization; that is why the criterion was frozen and taken to independently published experiments rather than elaborated further in simulation. A five-condition criterion (alternatives; shared finite constraint; change of representation; dependence on the class's own consequences; no externally designated winner) together with additional evidential gates was frozen before the final decision on four pre-declared architectures. After recomputation of open data, verification against the primary publications, an independent adversarial audit, and a subsequent clarification by the corresponding author of one primary study, two materially independent abiotic architectures obtained a clean PASS: deracemizing NaClO3 crystals (Viedma, PRL 2005) and dissipative anhydride droplets (Tena-Solsona et al., Nature Communications 2018). Macrocyclic fibre replicators (Eleveld et al., Nature Chemistry 2025) and polymeric PISA protocells were left QUALIFIED and are excluded from the threshold count. The result establishes E2 MULTIPLE-PHYSICAL-PE-A-CONFIRMED under the stated criterion: neither an organism nor DNA/RNA is necessary for this minimal class. Both surviving clean passes are non-templated, so the conclusion that template copying and a distinguished parent-offspring pair are not required now rests on 2 of 2; NaClO3 provides a fully inorganic crystalline branch. Four negative controls, including a genuine measured response to artificial selection that nevertheless fails the criterion, and an explicit comparison with Lewontin's conditions, Godfrey-Smith's Darwinian populations, competitive exclusion and niche construction support the status N2 DISTINCTIVE-FRAMEWORK. Delimitation. What distinguishes PE-A from theoretical accounts of history-dependent dynamics, including constructive-dynamics frameworks in which the effective operator depends on the population state, is not the idea that past consequences shape subsequent transitions, but the apparatus around it: a decision rule frozen before the systems were examined, an explicit condition excluding externally designated winners, documented negative controls that reject systems the criterion is not meant to admit, and retrospective application to independently published physical experiments. Membership in PE-A is therefore a verifiable property of a material system rather than a characterization of a class of models. Limits. This is the status of a discriminating operationalization, not a claim to a new universal law of nature. The work does not establish the universality or prevalence of PE-A, open-ended evolution, a rich non-genomic heritable alphabet, the origin of life, a physical instance of the template-replicator branch, or a prospective discriminating prediction. The literature search was broad but not a systematic PRISMA review; the result therefore concerns the existence of the class, not its prevalence. All classifications of published systems are retrospective.