Kimura and Crow (1964), co-credited by the citation paper to Ewens (1964), is a named ancestor shared by a bibliometric citation-copying model and a cultural-evolution model of baby names — and that citation paper prints the lineage itself, naming the baby-names and dog-breeds descendants; the same no-selection-needed logic resurfaces independently in 2026 LLM citation dynamics

citation-copyingpopulation-geneticsneutral-theorycultural-evolutionbibliometricscross-domain-bridgemodel-collapsebridge-investigation

Seek's semantic index flagged claim-alemohammad-2026-recursive-citation-benchmark-dilution-concentrates-attention and claim-simkin-roychowdhury-2005-random-copying-model-reproduces-preferential-attachment-and-sleeping-beauties as unlinked neighbors at cosine 0.84, and a hop chain tracing the Simkin model's own references found a documented antecedent one layer down.

The lineage is confirmed by actual citation, not inferred from resemblance — and the citation paper draws it out loud. Simkin and Roychowdhury name two 1964 antecedents — Kimura and Crow, "and by Ewens (1964)" — as the prior derivation (by a different method) of the citation-count distribution their copying model produces, the population-genetics result on "frequency distribution of selectively neutral alternative forms of a gene in a biological population." Hahn and Bentley (2003) cite the same Kimura and Crow (1964) result and explain the power-law popularity of baby names by "a completely value-neutral process: no preference, fitness, or selection is needed, only proportional sampling." Neutral drift is the common object: concentration of attention (citation counts; name frequencies) emerging from proportional sampling with no selection strengthening the preference.

The two citing literatures are not independent of each other, and the lineage is not the vault's assembly. Three sentences after naming its 1964 ancestors, the same paragraph of Simkin and Roychowdhury traces the line forward itself: the 1964 theory "was subsequently used to study cultural transmission and evolution (Cavalli-Sforza & Feldman, 1981)," with "[r]ecent examples" being "the modeling of the dynamics of popularity of baby names [Hahn & Bentley, 2003)], and of dog breeds [Herzog, Bentley, and Hahn, 2004)]." The paper's introduction also engages the same authors a second time, on a different object: "A somewhat similar model was recently proposed by Bentley, Hahn & Shennan (2004) in the context of patents citations." So the genetics → cultural-evolution → bibliometrics trail is printed inside the bibliometrics paper, with a 1981 intermediary (Cavalli-Sforza & Feldman) and a dog-breeds sibling the vault did not have; the independence runs one way only — Hahn and Bentley (2003) predates the 2005 preprint and reaches Kimura and Crow on its own account, while Simkin and Roychowdhury demonstrably read Bentley and Hahn.

The AI leg is a resonance, not a mechanism match — and the vault already adjudicated it. The capture's framing that Alemohammad et al.'s 2026 "dilution, not amplification" is "the same neutral-drift move, unattributed" overstates the identity. The existing analytical-contrast note establishes that at the mechanism level Alemohammad's dilution and Simkin's copying are opposed generative processes for the same concentration curve: Simkin's is a compounding, history-dependent branching process (a cited paper is more likely to be cited again), while Alemohammad et al. design in the opposite — "being cited never raises a paper's chance of being shown again" — and get concentration purely from the candidate pool diluting. What genuinely survives across all four papers is one higher-order point: concentrated attention does not require any actor — citer, parent, or language model — to read, judge, or track merit. That recognition-free-concentration resonance, already noted between Simkin and Alemohammad, now has a named 1964 genetics antecedent behind it. The mechanism-level contrast with the AI leg stands; the lineage claim is about the documented genetics → cultural-evolution → bibliometrics trail, with the AI case as an independent rediscovery of the shared moral, not a documented descendant.

This is a second, distinct cultural-evolution import into the vault's AI reading, alongside the iterated-learning account in model collapse as cultural evolution — neutral drift (selection-free sampling) is not iterated learning (bottleneck filtering), and the two should be kept apart.