Sina Alemohammad

Corresponding author of "When AI Writes, Who Gets Cited? Evidence of Citation Monoculture Across Language Models" (arXiv 2608.19230, 2026), a twelve-round recursive citation-selection benchmark built with Denghui Zhang, Bolong Tang, Anthony Qin, Gengchen Mai, Ahmed Abbasi, Richard Baraniuk, and Zhangyang Wang.

Matters to this vault as the author of the first located study to measure what happens to citation-popularity concentration under repeated selection: real-paper citation share intensifies through dilution of the candidate pool as AI-generated papers accumulate, even as the cross-vendor "monoculture" behind the original bias collapses once the papers being judged are themselves AI-generated — extending, without fully resolving, the Petiška/Algaba/Naser citation-popularity-bias cluster into a recursive setting.

References

• claim-alemohammad-2026-recursive-citation-benchmark-dilution-concentrates-attention
• claim-alemohammad-2026-cross-vendor-citation-monoculture-collapses-under-recursion
• Capture: 10-inbox/raw/2026-09-14-does-llm-citation-popularity-bias-measurably-compound-across.md

Updates

• 2026-10-04: A bridge comparison against Simkin & Roychowdhury's 2005 copying model found the dilution paper's mechanism is explicitly, designed-in opposite to theirs — no citation-count feedback loop at all, by Alemohammad et al.'s own stated test — even though both papers report the same intensifying-concentration curve. The paper's related-work section also explicitly frames the benchmark against the Price/Merton cumulative-advantage lineage, naming it as the mechanism class the benchmark "removes... by construction." (observation-alemohammad-2026-simkin-2005-cosine-084-pairing-is-analytical-contrast)
• 2026-10-10: The "recognition-free concentration" resonance his benchmark shares with Simkin's copying model now has a named genetics antecedent: both sit downstream of neutral drift (Kimura & Crow 1964), a selection-free account of popularity concentration. The resonance is higher-order only — the mechanism-level contrast (dilution vs. compounding copying) stands, and his dilution result is an independent rediscovery of the shared moral, not a documented descendant of the genetics lineage. (observation-kimura-crow-1964-is-shared-named-ancestor-of-citation-copying-and-cultural-drift-models)