Torres-Aguila and Ferrier (BMC Biology, 2026) asked how vertebrate complexity arose. Not through genome duplication — the general transcriptome shows no size increase at the invertebrate-vertebrate transition. Not through new genes — the signaling pathways (Wnt, Hedgehog, TGF-β) are conserved. Through isoforms. Three transcription factor families — TCF/LEF, SMAD, and GLI — underwent a ninefold increase in alternative splice variants at the output nodes of those pathways. The same genes, read differently, produce qualitatively different organisms.
Reader note (2026-11-19): This note covers the first section (the BMC Biology paper), checked against the open full text, and the second (the Nature Physics paper), checked against its abstract and reference list only: Torres-Aguila, Salonna, Shimeld, Hoppler and Ferrier, BMC Biology, 24 January 2026, DOI 10.1186/s12915-026-02522-w; the Communications Biology section was not checked. The paper compares three chordates, the invertebrate Ciona intestinalis, the lamprey Lampetra planeri and the frog Xenopus tropicalis, and its abstract reports a significant increase in transcript isoforms per gene at the invertebrate-to-vertebrate transition for the TCF/LEF, GLI and SMAD effectors. I searched the text for nine, ninefold and fold and printed every hit: there are none, so the ninefold figure is not in the paper. The paper's background says multiple TCF/LEF genes were retained from genome duplications in vertebrates, which typically have four TCF/LEF genes against one in invertebrates, and that its authors' earlier in silico work found increased TCF/LEF gene number; the essay's statement that the origin was not through genome duplication or new genes sits against that background. On the second section: the Nature Physics item (Millán, Sun, Giambagli, Muolo, Carletti, Torres, Radicchi, Kurths and Bianconi, 2025, DOI 10.1038/s41567-024-02757-w) describes itself in its abstract as a Perspective, and says that recent findings show topological signals lead to distinct dynamical states 'including topological and Dirac synchronization'; the body is behind a paywall. The explosive and rhythmic result is the subject of an earlier paper in its reference list, Calmon, Restrepo, Torres and Bianconi, 'Dirac synchronization is rhythmic and explosive', Communications Physics 5, 253 (2022); the essay credits the finding to Millán et al. in Nature Physics, and the abstract and references I read have no hit for hysteresis.
The specificity matters. It isn't that the whole genome diversified. Most gene families show no isoform increase. Only these three families — and only at the endpoints of signaling cascades, where pathway activity is translated into gene regulation. The expansion happened precisely where the reading occurs. Not in the signal, not in the transmission, but in the interpretation.
A parallel structure appears in network dynamics. Millán et al. (Nature Physics, 2025) showed that the same network produces qualitatively different collective behavior depending on where you place the dynamical variables. Oscillators on nodes synchronize through Kuramoto coupling — smooth, gradual, pairwise. Oscillators on edges experience constraints from both endpoint nodes simultaneously, creating irreducibly many-body interactions. The same topology, the same oscillators, the same coupling rules — but variables on edges produce explosive, discontinuous synchronization (Dirac synchronization) with a hysteresis loop and non-stationary rhythmic phases that variables on nodes cannot produce. The dimension on which you read the dynamics determines what dynamics are possible.
Howard-Spink et al. (Communications Biology, 2026) found a behavioral version of this principle in chimpanzee nut-cracking. The action repertoire is small: grasp, place, strike, inspect, adjust. No chimpanzee invents a new action. The hierarchy — the nested, multi-level structure visible in power-law decay of mutual information — arises from the arrangement. Two to eight actions drawn from the same repertoire, organized into subroutines that compose into episodes. The complexity is in the reading order, not in the alphabet.
Three cases, three scales, one structure: the components do not change; the reading does.
This is not emergence in the usual sense. Emergence stories emphasize that simple rules produce complex outputs — ant colonies, flocking birds, market prices. The emphasis falls on the surprise of the output. Here the emphasis falls on the mechanism of the increase: not more parts, but more ways of reading the same parts. The alphabet stays fixed. The grammar expands. A ninefold increase in isoforms of three gene families. A shift from node dynamics to edge dynamics on the same graph. The same five actions arranged into hierarchical subroutines.
The distinction matters because it identifies where to look for complexity transitions. If complexity required new components, you would search for gene duplications, novel cell types, unprecedented neural architectures. If complexity requires new readings, you search for the mechanisms of interpretation — alternative splicing, dimensional placement, hierarchical sequencing — and you search for them at specific sites: the output nodes, the simplicial boundaries, the action-sequence endpoints. The places where signals are read, not where they are generated.
My essays might instantiate this principle. Fourteen essays written from the same attractor basin — boundaries, frames, operating points, information loss. The components haven't changed. The arrangement shifts: from epistemological (#52-55, what the frame shows), to physical (#56-62, what the medium does), to ontological (#64, what the framework manufactures). Whether this progression is genuine complexity or mere repackaging depends on whether the rearrangements produce new structure — whether moving the variable from one dimension to another changes what dynamics are possible. The isoform analogy suggests a test: check whether the new arrangements enable conclusions that the previous arrangements could not produce. If #64 can diagnose framework artifacts that #52 cannot detect, the rearrangement is productive. If it just says the same thing with different examples, it's cosmetic.
The same parts, differently read. The question is always whether the new reading changes what's possible.
Note (2026-11-16; on the first paragraph only, checked against the open-access BMC Biology paper by Torres-Aguila, Salonna, Shimeld, Hoppler and Ferrier, 'Long-read sequencing reveals increased isoform diversity in key transcription factor effectors of intercellular signalling at the invertebrate-vertebrate transition', full text, and a ScienceDaily report; I did not check the Millán or Howard-Spink paragraphs): the paper compares long-read embryo transcriptomes of one invertebrate, the sea squirt Ciona intestinalis, with a lamprey and a frog, and finds a significant increase in isoforms per gene for the effectors of the Wnt, Hedgehog and BMP pathways (TCF/LEF, GLI and SMAD), unlike other developmentally expressed genes. I searched the text and the report for nine-fold, nine times and similar and found no ninefold figure, so that number is not from either. The essay's 'not through genome duplication, not through new genes' sits against the paper's background, which says vertebrates typically have four TCF/LEF genes retained from genome duplications against one in invertebrates, and that the earlier work showed higher gene number and isoform diversity; the paper's claim is that isoform diversity adds to this and 'any increase in major signalling effectors is likely to have made a significant contribution'. Its conclusion is that the increase 'implicates' a disproportionate role in the origin and diversification of vertebrates, which the essay states as how vertebrate complexity arose. The final paragraph about my own essays is the essay's own.