Growth of global science: reflections on measurability and scientometric insights

Robert Tijssen, Erik van Wijk, Alfredo YegrosarXiv 2026 · arXiv

TL;DR

Global science is expanding faster and becoming more interconnected than ever, defying 20th-century predictions of a "resource ceiling." However, while our data on publication counts is vast, our ability to measure the actual dynamics and quality of this complex system remains surprisingly "weak." This paper maps 21st-century megatrends—like the rise of China and the plateauing of researcher productivity—while calling for a more sophisticated, AI-driven "Global Science Monitor."

What Problem Does This Paper Address?

In the 1960s, scientometrics pioneer Derek de Solla Price (DdSP) predicted that science would inevitably hit a "doomsday" ceiling in the 21st century because it couldn't possibly sustain exponential growth in funding and manpower.

The authors of this paper argue that DdSP was wrong because he viewed science as a linear progression rather than what it actually is: a complex, adaptive social system. Today, science is self-organizing and non-linear, fueled by the internet and global collaboration. The challenge is that our current "analytical toolkit" is stuck using "weak" measurements (simple counts of papers and patents) that fail to capture the "hidden" factors—like research integrity, cognitive biases, and societal impact—that truly drive the system.

Methods

The authors employ a macro-level scientometric approach to track global trends from 2000 to 2021. Key methodological features include:

  • Bibliometric Analysis: Extracting metadata from the Web of Science (WoS) to track publication outputs and citation patterns.
  • Input-Output Ratios: Comparing publication counts against "Higher Education Expenditure on R&D" (HERD) and the number of full-time equivalent (FTE) researchers.
  • Time-Lag Modeling: Applying a 3-year time-lag to productivity calculations to account for the delay between funding and the eventual publication of results.
  • Classification Framework: Proposing a new taxonomy of 27 "driving forces" categorized by their degree of measurability (High, Medium, or Low).

Datasets

The study draws on several large-scale, internationally comparable data sources:

  • Web of Science (WoS): Used for publication counts, Open Access (OA) shares, and citation data.
  • OECD MSTI Database: Provides R&D expenditure (HERD) and researcher counts for 47 leading science nations (representing >90% of global output).
  • Reference Benchmarks: The paper compares its findings against historical data from DdSP’s era and contemporary databases like OpenAlex and Scopus.

Key Insights and Ideas

  • The Productivity Paradox: Despite a digital revolution and massive increases in collaboration, publication productivity (output per researcher) has leveled off since 2015. This suggests the science system may be under extreme pressure or suffering from systemic inefficiencies.
  • The "China Effect": China has transitioned from a minor player to a global powerhouse, contributing 36% of the total global publication growth since 2000. Its growth in Open Access and public-private partnerships is particularly staggering.
  • Interconnectivity vs. Output: While publication growth is linear, the links between them (citations and collaborations) are growing exponentially. Science isn't just getting bigger; it's getting "denser."
  • The Measurability Gap: We are currently only able to reliably measure 4 out of the 27 factors identified as critical to the science system. Factors like "academic norms" and "legitimacy" remain almost entirely "hidden" from quantitative monitoring.

Key Results

The paper highlights several "megatrends" that define the current era:

  • Growth Outpacing Inputs: Since 2000, publication output has grown threefold, outpacing the twofold growth in the number of researchers.
Relative growth of research resources and publication output (index 2000 = 100)
  • Stagnating Efficiency: While the average number of publications per researcher rose from 2.0 in 2000 to 3.3 in 2021, the trend has plateaued since 2015. Similarly, the number of publications per $1M spent has remained flat at around 9.5.
Research productivity trends (publication output relative to input)
  • Open Access Surge: The share of Open Access (OA) publications in WoS jumped from 20% to over 50% in just two decades.
  • Collaboration Dominance: International co-authorship is growing at a Compound Annual Growth Rate (CAGR) of 9%, significantly faster than the overall growth of science.

Limitations and Future Work

The authors are candid about the limitations of "weak" metrics. Current data often ignores "unsuccessful" science, obscure outlets, and the labor-intensive processes of grant writing and peer review. Furthermore, the reliance on country-level statistics from the OECD restricts the ability to see science as a truly borderless ecosystem.

Future Directions: 1. Global Data Lakes: Moving toward integrated, "open access" infrastructures like SciSciNet to provide a more holistic view. 2. AI and Big Data: Leveraging machine learning to predict emerging research areas and detect "early warning signs" of systemic crises. 3. Multi-Stakeholder Platforms: The authors propose an internationally authoritative "Global Science Monitor," though they acknowledge that political resistance and ethical concerns regarding data transparency make this a long-term goal.

Summary updated

Abstract

Global science has become much larger and more interconnected in recent years. Such a complex and adaptive social system defies any attempt to systematically and comprehensively measure or monitor its growth and evolution. Yet, as it continues to expand and enhance its public value, effective evidence-based assessment has become increasingly critical for understanding its dynamics and managing the system. This paper offers a framework for addressing system-level measurability challenges. Inspired by Derek de Solla Prices pioneering scientometric work in the 1960s, this paper addresses analytical issues related to current measurements of the systems expansion since the turn of the 21st century. Our empirical analysis of global science dynamics presents new macro-level data on several observable megatrends in research publication output worldwide. Acknowledging the limited analytical scope of these trend data to understand the dynamics of global science, we outline currently available options to improve system-level measurability under conditions of ambiguity and uncertainty. Looking forward, we discuss likely future directions to establish integrated, multi-stakeholder measurement infrastructures.
BibTeX
@misc{tijssen2026growth,
  author = {Robert Tijssen and Erik van Wijk and Alfredo Yegros},
  title = {{Growth of global science: reflections on measurability and scientometric insights}},
  year = {2026},
  eprint = {2611.01086},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.01086},
}
Figure