🔬 physics

Normalized fractional Hirsch Index and Modified Pareto Law

This paper proposes a normalized fractional Hirsch index to demonstrate that for highly successful scientists, approximately 14% of their top-cited papers generate about 86% of their total citations, a distribution that aligns with Pareto's 80-20 law and self-organized critical sandpile models.

Original authors: Asim Ghosh, Bikas K. Chakrabarti

Published 2026-09-29✓ Author reviewed ⓘ

📖 4 min read☕ Coffee break read

CC0 1.0

Original authors: Asim Ghosh, Bikas K. Chakrabarti

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ This is an AI-generated explanation of the paper below. It is not written by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of science, measuring the impact of a researcher's work is a constant challenge. For decades, the standard tool for this has been a simple count: how many times has a scientist's work been cited by others? In 2005, a specific metric known as the Hirsch index, or h-index, became the most popular way to summarize this. It attempts to find a balance between the number of papers a scientist has written and how often those papers are read and referenced. If a researcher has an h-index of 20, it means they have published 20 papers, each of which has been cited at least 20 times. While this number is widely used by universities and funding agencies to judge reputation, it has a significant flaw. The index is not a fixed scale; it grows as a scientist publishes more work. A highly productive researcher will naturally have a higher number than a less prolific one, even if their work is equally influential. This makes it difficult to compare scientists across different career stages or fields, much like trying to compare the wealth of a small town to a giant nation using only a raw dollar count without adjusting for population.

To solve this, researchers have long looked to economics, where experts measure inequality using normalized indices. These tools, such as the Gini index, strip away the total size of a population or economy to reveal the underlying distribution of wealth. If a small fraction of people hold a massive share of the wealth, the index reflects that concentration regardless of whether the total wealth is a million dollars or a trillion. Recently, scientists have applied similar logic to scientific citations, discovering that the distribution of citations often follows a pattern where a small number of papers receive the vast majority of attention. This phenomenon is reminiscent of the Pareto principle, often called the 80-20 rule, which suggests that roughly 80 percent of effects come from 20 percent of causes. However, when looking at the most successful scientists, the numbers appear to shift, suggesting an even more extreme concentration of impact.

A team of researchers, Asim Ghosh and Bikas K. Chakrabarti, set out to bridge the gap between the unbounded Hirsch index and these normalized economic measures. They analyzed the publication records of 126,067 scientists, a group that included 95 Nobel Laureates, to see if they could create a new, standardized version of the Hirsch index. Their goal was to determine if the most successful scientists shared a common pattern in how their citations were distributed, similar to how wealth is distributed in highly unequal societies. They found that while the raw Hirsch index varied wildly among the Nobel Prize winners—ranging from a low of 25 to a high of 260—this variation was simply a reflection of how many papers each person had written, not necessarily a difference in the structure of their impact.

When the researchers adjusted the data to account for the total number of papers and total citations, a striking pattern emerged. They calculated a new value, a normalized fractional index, which represents the fraction of total citations earned by the fraction of the most successful papers. For the 95 Nobel Laureates, this new metric converged on a very specific number. They found that approximately 14 percent of a scientist's most cited papers were responsible for generating about 86 percent of their total citations. This result is remarkably close to the findings in economic studies of extreme inequality, where the top 14 percent of the population often holds 86 percent of the wealth. It is also a more precise fit than the traditional 80-20 rule, which would suggest that 20 percent of papers generate 80 percent of citations.

The researchers noted that this 86-14 split is not unique to human achievement. It mirrors the behavior of self-organized critical systems, such as piles of sand. In computer simulations of sandpiles, as grains are added, the system reaches a critical point where avalanches occur. In these models, it is observed that roughly 14 percent of the largest avalanches dissipate about 86 percent of the total mass. The fact that the citation patterns of the world's most distinguished scientists align so closely with these physical models suggests that scientific impact may be governed by similar underlying dynamics of competition and accumulation. The study suggests that while the raw Hirsch index is useful for counting output, it fails to capture the true nature of scientific reputation. By normalizing the data, the researchers revealed that the most successful scientists, regardless of their total publication count, operate within a remarkably consistent framework of inequality, where a tiny fraction of their work carries the overwhelming weight of their influence.

Technical Summary: Normalized Fractional Hirsch Index and Modified Pareto Law

Problem Statement
The Hirsch index ( $h$ ), introduced in 2005, is a widely adopted scientometric measure for assessing the productivity and citation impact of scientific authors. It is defined as the maximum integer $h$ such that an author has published $h$ papers, each cited at least $h$ times. Despite its popularity in databases like Google Scholar and Web of Science, the $h$ -index possesses significant limitations as a comparative metric. It is an unbounded integer that statistically scales with the square root of the total number of publications ( $N_t$ ) or total citations ( $C_t$ ). Consequently, the $h$ -index lacks a normalized upper bound, making it difficult to compare the "reputation" or inequality structures across scientists with vastly different publication volumes. For instance, among 95 Nobel Laureates, $h$ values scatter significantly (ranging from 25 to 260), failing to converge on a characteristic value that might indicate a universal level of scientific impact.

In contrast, socio-economic inequality measures, such as the Gini index ( $g$ ) and the Kolkata index ( $k$ ), are normalized fractions bounded between 0 and 1. These indices are independent of total wealth or population size. Recent studies have observed that for highly successful groups (e.g., Nobel Laureates) and in self-organized critical (SOC) sandpile models, these indices converge to a value of approximately $g = k \simeq 0.86$ . This suggests a "modified Pareto law" where roughly 86% of the wealth (or citations) is held by the top 14% of the population (or papers), deviating from the classical Pareto 80-20 law ( $k=0.80$ ). The central problem addressed by this paper is the lack of a normalized, bounded metric for citation inequality that can be directly compared to these socio-economic and physical models, specifically to determine if the $h$ -index can be transformed to reveal similar universal scaling behaviors.

Methodology
The authors propose a normalization of the Hirsch index to create a "normalized fractional Hirsch index," denoted as $\bar{h}^{(f)}_c$ . The methodology involves the following steps:

Data Collection: The study analyzes Google Scholar data for a sample of 126,067 scientists, each with at least 100 publications to ensure statistical robustness. This sample includes 95 Nobel Laureates across Physics, Chemistry, Medicine, and Economics.
Definition of Fractional Components:
- $h^{(f)}_n = h / N_t$ : The fraction of the author's total papers that constitute the successful $h$ papers.
- $h^{(f)}_c = (\sum_{h} n_c) / C_t$ : The fraction of the author's total citations ( $C_t$ ) contributed by those $h$ successful papers.
- The authors note that since $h$ scales as $\sqrt{N_t}$ , both $h^{(f)}_n$ and $h^{(f)}_c$ scale as $1/\sqrt{N_t}$ , meaning they remain unnormalized and dependent on the total volume of work.
Normalization: To eliminate the dependence on total publication volume, the authors define the normalized fractional Hirsch index for citations as:
$\bar{h}^{(f)}_c = \frac{h^{(f)}_c}{h^{(f)}_c + h^{(f)}_n}$
The complementary normalized fraction is defined as $\bar{h}^{(f)}_n = \frac{h^{(f)}_n}{h^{(f)}_c + h^{(f)}_n}$ . By construction, $\bar{h}^{(f)}_c + \bar{h}^{(f)}_n = 1$ , and both values are bounded between 0 and 1.
Comparative Analysis: The authors plot the distributions of these new indices against the Gini index ( $g$ ) and the Kolkata index ( $k$ ) for the entire dataset and specifically for the Nobel Laureates. They compare the convergence of $\bar{h}^{(f)}_c$ with the known convergence of $g$ and $k$ in high-inequality regimes.

Key Contributions

Proposal of a Normalized Metric: The paper introduces $\bar{h}^{(f)}_c$ , a bounded, normalized version of the Hirsch index that accounts for both the fraction of successful papers and the fraction of total citations they generate.
Unification of Scientometric and Socio-Economic Inequality: The study bridges the gap between author-level scientometrics and socio-economic inequality theory by demonstrating that citation distributions follow similar statistical laws to wealth distributions and physical SOC models.
Identification of a Universal Scaling Value: The analysis reveals that while the raw $h$ -index varies wildly among top scientists, the normalized index $\bar{h}^{(f)}_c$ converges to a specific value for highly successful researchers.

Results
The analysis of the 126,067 scientists yields the following observations:

Convergence of Indices: For the general population of scientists, there is scatter in the values of $g$ , $k$ , and $\bar{h}^{(f)}_c$ . However, for the subset of 95 Nobel Laureates, these indices converge significantly.
The "Modified Pareto" Value: For the Nobel Laureates, the Gini index ( $g$ ) and Kolkata index ( $k$ ) converge to approximately $0.86$. Crucially, the normalized fractional Hirsch index $\bar{h}^{(f)}_c$ also converges to this same value ( $\bar{h}^{(f)}_c \simeq 0.86$ ).
Interpretation of the Value: This convergence implies that for the most successful scientists, approximately 86% of their total citations are earned by the top 14% of their papers (since $\bar{h}^{(f)}_n \simeq 1 - 0.86 = 0.14$ ).
Comparison with Raw $h$ : The raw $h$ -index for these Nobel Laureates ranges from 25 to 260, showing no characteristic value. In contrast, $\bar{h}^{(f)}_c$ is restricted to a narrow range ( $0.68 \leq \bar{h}^{(f)}_c \leq 0.98$ ), with a strong clustering around 0.86.
Correlation with SOC Models: The observed value of $\simeq 0.86$ aligns with results from self-organized critical sandpile models (e.g., Bak-Tang-Wiesenfeld and Manna models), where 86% of avalanche mass is dissipated through 14% of the largest avalanches near the critical point.

Significance and Claims
The paper claims that the normalized fractional Hirsch index ( $\bar{h}^{(f)}_c$ ) provides a more characteristic measure of scientific reputation and inequality than the traditional $h$ -index. By normalizing the index, the authors demonstrate that the citation inequality among top-tier scientists (Nobel Laureates) follows a "modified Pareto law" ( $k \simeq 0.86$ ) rather than the classical 80-20 law ( $k=0.80$ ).

The authors suggest that this convergence indicates a lack of correlation between the raw $h$ -index and the specific "reputation" or inequality structure of distinguished scientists, whereas the normalized index reveals a universal critical value. This value ( $\simeq 0.86$ ) appears to be a signature of highly competitive, self-organized systems, similar to those observed in economic markets and physical sandpile models approaching criticality. The study posits that $\bar{h}^{(f)}_c$ can serve as a robust, bounded metric for comparing scientific impact across authors with different publication volumes, aligning scientometric measures with established principles of socio-economic and physical inequality.

Technical Summary: Normalized Fractional Hirsch Index and Modified Pareto Law

More like this