In the past half-century, the proportion of research that opens up new pathways within a field has significantly decreased.
An analysis of the disruptive impact of newly published papers on historical literature  reveals that although there has been an explosion in the quantity of scientific and technological research papers in recent decades, their “breakthrough” nature has been declining.
Data analysis of millions of papers shows that compared to research in the mid-20th century, the research conducted in the first decade of this century appears to be progressively advancing science rather than suddenly opening up new directions and displacing previous research findings. Similar trends were found in the analysis of patents from 1976 to 2010.
Russell Funk, a sociologist at the University of Minnesota and co-author of the study, stated, “Some changes can be observed from the data.” The study was published on January 4th in Nature. “The intensity of breakthroughs in current research is not as high as before.”
The authors inferred that if a study has a high level of breakthrough, subsequent research is less likely to reference the paper and more likely to directly cite it. The research team used citation data from 45 million papers and 3.9 million patents to calculate the Critical Difference (CD) index, which measures the degree of breakthrough. The CD index ranges from -1 to 1, with -1 representing the least disruptive research and 1 representing the most disruptive.
From 1945 to 2010, the average CD index for research papers declined by over 90% (see “Disruptive Research Is on the Decline”); from 1980 to 2010, the average CD index for patents declined by over 78%. The decline in breakthroughs was observed across all research fields and patent categories analyzed, even after accounting for potential differences in citation practices.
The authors also examined the most commonly used verbs in the papers and found that research in the 1950s tended to use words indicating creation or discovery, such as “produce” or “determine,” while research in the 2010s tended to use words indicating progressive advancement, such as “improve” or “enhance.”
“They provided a very rigorous explanation of this phenomenon, which is remarkable,” said Da-Shun Wang, a computational sociologist at Northwestern University. “They analyzed it from 100 different angles, and I find the overall results quite persuasive.”
Yi-An Yin, another computational sociologist at Northwestern University, stated that other studies  have also found a slowdown in scientific innovation in recent decades, but this study broke new ground by recording the changes in science in a data-driven manner.
Wang believes that groundbreaking science is not always the absolute best, and progressive science is not necessarily inferior. He stated that the direct observation of gravitational waves, for example, was not only revolutionary but also a result of progressive science.
The ideal scenario is a combination of progressive science and breakthrough research, according to John Walsh, an expert in science and technology policy at the Georgia Institute of Technology. He said, “In an era where the validity of research results is of great concern, conducting more replication and reproduction studies may not be a bad thing.”