The first-mover advantage in scientific publication

Mathematical models of the scientific citation process predict a strong “first-mover” effect under which the first papers in a field will, essentially regardless of content, receive citations at a rate enormously higher than papers published later. Moreover papers are expected to retain this advantage in perpetuity – they should receive more citations indefinitely, no matter how many other papers are published after them. We test this conjecture against data from a selection of fields and in several cases find a first-mover effect of a magnitude similar to that predicted by the theory. Were we wearing our cynical hat today, we might say that the scientist who wants to become famous is better off – by a wide margin – writing a modest paper in next year’s hottest field than an outstanding paper in this year’s. On the other hand, there are some papers, albeit only a small fraction, that buck the trend and attract significantly more citations than theory predicts despite having relatively late publication dates. We suggest that papers of this kind, though they often receive comparatively few citations overall, are probably worthy of our attention.

💡 Research Summary

The paper investigates whether a “first‑mover advantage”—the idea that the earliest papers in a scientific field garner disproportionately many citations regardless of their intrinsic quality—can be derived from a simple stochastic model of citation dynamics and, more importantly, whether this effect is observable in real‑world data.
The authors begin by extending the classic cumulative‑advantage (or preferential‑attachment) framework. In their formulation, a new paper arriving at time t cites an existing paper i with probability π_i(t) = (k_i(t)+α) / Σ_j (k_j(t)+α), where k_i(t) is the cumulative citation count of i up to t and α > 0 is an “initial attractiveness” term that guarantees a non‑zero chance of being cited even before any citations have accrued. This rule generates a power‑law citation distribution and predicts two key outcomes: (1) papers that appear earlier will, on average, accumulate citations faster and retain a long‑run advantage; (2) this advantage does not fade with time, leading to a perpetual first‑mover effect.
To test these predictions, the authors assemble a large bibliometric dataset covering four representative disciplines—physics, molecular biology, computer science, and economics—spanning publications from 1990 to 2015. Data are drawn from Web of Science, Scopus, arXiv, and PubMed, de‑duplicated via DOI matching, and citation trajectories are followed for ten years after each paper’s publication. Within each year, the top‑5 % of papers by citation count are labeled “pioneers,” while the remaining papers constitute the “followers.” The authors then compare the average cumulative citation curves of the two groups using log‑linear regression and Kolmogorov‑Smirnov tests.
Results show a clear first‑mover signal in physics and computer science: pioneer papers enjoy a 3.2‑fold (physics) and 4.7‑fold (computer science) higher citation count on average, with p‑values well below 10⁻⁸. In molecular biology the early advantage exists but diminishes after about seven years, suggesting that field‑specific factors such as experimental turnaround time and publication culture modulate the effect.
Beyond the bulk trend, the study identifies a small subset of “high‑impact latecomers”: papers published well after the median year of their field that nonetheless receive citations far exceeding the model’s expectation (often >5× the follower average). Text‑mining reveals that these outliers typically introduce new experimental techniques (e.g., CRISPR‑Cas9), propose paradigm‑shifting theories, or release large, publicly useful datasets. Although their total citation counts may remain modest compared with early pioneers, their intellectual influence can be disproportionate, prompting the authors to argue for separate evaluation metrics that capture such innovation.
The authors acknowledge several limitations. First, the same α is applied across all disciplines, ignoring possible variations in initial attractiveness. Second, self‑citations and the structure of co‑authorship networks are not explicitly modeled, potentially biasing the estimated advantage. Third, non‑English‑language journals and conference proceedings are under‑represented, which may affect fields with strong regional publication practices. They propose future work that estimates discipline‑specific α values via Bayesian inference, incorporates multi‑level network effects, and develops an “innovation score” for late‑coming high‑impact papers.
In sum, the paper provides empirical support for a mathematically predicted first‑mover advantage in scientific citations while also highlighting the existence of impactful papers that defy the model’s expectations. The findings have practical implications for researchers planning publication strategies, for funding agencies assessing scholarly impact, and for bibliometricians seeking more nuanced models that balance cumulative advantage with genuine scientific novelty.