Clarivate Analytics, managers of the Web of Science database, asserts that human capital (talent, intelligence, creativity, etc.) is the most important factor in the race for knowledge. This contradicts previous research indicating the influence of other factors, such as gender, language, and funding, on scientific output. This study investigates whether human capital alone is the primary driver of scientific advancement by examining a random sample of highly cited researchers from the 2018 Clarivate Analytics database. The hypothesis is that if human capacity and talent were the sole determinants of success, the profile of prominent researchers would be evenly distributed across gender, country development level, funding access, and languages. The study aims to explore the relative importance of human capital against these other factors.

Existing literature highlights gender as a significant factor, with studies showing negative impacts on female scholars due to stereotypes, family commitments, implicit bias, and perceived lower productivity. Language is another barrier, with English's dominance creating disadvantages for non-English speakers. Access to funding is also crucial, its influence varying based on funding amount. The algorithm used by automated search engines, originating from Eugene Garfield's work in 1955, plays a substantial role, initially intended for organizing scholarly literature but now impacting journal purchasing, author assessment, productivity measurement, and research funding. Bradford's law, a core assumption of the algorithm, focuses on high-impact journals, often in English, potentially overlooking high-quality research in other languages or less impactful journals. This algorithm also prioritizes publication numbers in high-impact journals, potentially disadvantaging scholars from less developed countries with heavier teaching loads. The algorithm's reliance on citation counts ignores negative citations and may reinforce established schools of thought over innovative research, particularly problematic when considering the high number of authors involved in some works. Finally, the algorithm's assumption that factors such as race, nationality, and gender are irrelevant in science, while aspirational, ignores the realities of researcher bias and the influence of personal characteristics.

This research uses a random sample of 198 scholars from the 2018 Clarivate Analytics Highly Cited Researchers database. Scholars were selected using a random number generator to choose three letters of the alphabet, then selecting the first three scholars in each of the 22 knowledge areas whose last names begin with those letters. Scholars whose gender couldn’t be determined were replaced. The study analyzes the influence of gender, funding (sponsorship), journal of publication, number of co-authors, language, and country development level on the researchers' success, factors identified in previous research as relevant. Country of primary affiliation data, obtained from the Clarivate Analytics dataset, was analyzed using the Human Development Report 2019 to determine the level of human development. English fluency was assessed based on the percentage of English speakers in each country (at least 30%). Gender was determined using Google Photos, while data on the percentage of male and female scholars per country came from UNESCO's Institute for Statistics. The researchers’ latest highly cited publication was analyzed to collect data on citations, co-authors, sponsors, and journal of publication. Descriptive statistical methods (average, minimum, maximum, standard deviation, median, and mode) were used to analyze the data. The study specifically considered the distribution across 'very-highly developed', 'high', 'medium' and 'low' levels of human development.

The analysis reveals a highly concentrated profile of highly cited researchers. Approximately half (49.5%) are affiliated with US institutions, with the remaining scholars distributed across 29 other countries, leaving 83% of UN-recognized countries unrepresented. A similar percentage (83%) of the sample are male. UNESCO data for 21 of the 30 countries representing the scholars shows an average of 67% male scholars, while the sample itself has 80% male scholars among those from these countries. This illustrates that women are underrepresented in science and experience more difficulty reaching top positions. The vast majority (98%) of scholars are affiliated with 'very-highly developed' countries, with only one from a 'medium' human development country and none from 'low' human development countries. The distribution by country shows a significant representation of scholars from China and other 'very-highly developed' countries, suggesting the level of country development is crucial. Similarly, 93% of the scholars are from countries where English is spoken by at least 30% of the population. This concentration is even more pronounced among women (N=33), all of whom are from 'very-highly developed' English-speaking countries. No female scholars were from Latin America, Africa, or Oceania, and only two were from Asia (South Korea). The number of authors per publication varied greatly (1 to 2834, average 56.3, median 11). Excluding outliers, the average drops to 36.5, still indicating a high tendency for large author groups, facilitating increased publications and reinforcing cross-citation bias. 90% of the authors reported sponsorship, with only one non-sponsored scholar not affiliated with a 'very-highly developed' country. The number of sponsors per publication also varied significantly (0 to 221). Excluding outliers, the average was 5.08 sponsors per publication. The 198 papers were published in 130 journals, all in English, illustrating a significant concentration of publications in a small number of journals (e.g., Nature, Science). The highly-cited researchers showed a bias toward publishing in a small number of high-impact journals, suggesting that the algorithm of science disproportionately rewards publications in these select journals, neglecting other important research in other journals or languages.

The findings strongly challenge Clarivate Analytics' assertion that talent is the primary driver of scientific advancement. The absence of representation from various geographic regions, particularly women from developing countries, and the overwhelming dominance of English and high-impact journals indicate significant biases. These biases stem from factors such as language barriers, funding disparities, and the inherent limitations of the algorithm itself. The data suggests that gender, language, funding, and the level of development of the country are far more influential than human capital alone. The concentration of publications in a limited number of journals further compounds the issue, suggesting that the current system disproportionately favors scholars from specific backgrounds. The overrepresentation of scholars from specific countries and the lack of diversity undermine the validity of scientific knowledge by limiting perspectives and potentially excluding crucial research from underrepresented communities.

This study demonstrates that the claim that human capital is the primary driver of scientific advancement is misleading. The significant biases identified in the distribution of highly cited researchers across gender, language, funding, and country development highlight the crucial role of these external factors in scientific success. Search engines like Web of Science, which wield significant influence over scientific visibility, should adjust their algorithms to address these biases, improving the inclusivity and validity of scientific knowledge. This adjustment is crucial for fostering a more equitable and representative scientific community, particularly in light of the growing distrust in science.

The study's reliance on a single database (Clarivate Analytics) might limit the generalizability of the findings. The methodology of random selection, while aiming for objectivity, might not perfectly capture all biases in the entire researcher population. Information on factors such as religion and ethnicity was unavailable, which could have further enriched the analysis. The study's focus on 'highly cited researchers' could potentially exclude talented researchers whose work is not yet widely recognized.