What types of novelty are most disruptive?

Abstract

This work builds on recent advances in the study of novelty and impact to investigate whether novel scientific papers tend to disrupt the scientific literature, and whether some types of novelty are more disruptive than others. We develop a computational measure to capture the ways in which scientific articles are novel (e.g., developing a new method or presenting a new result). We also utilize Funk and Owen-Smith’s (2017) CD index to capture the different ways an article can influence a stream of literature: by consolidating (C) or disrupting (D) the status quo. Drawing upon classic sociology of science scholarship on the distinction between cumulative normal science research and non-cumulative paradigm shifts (Kuhn 1962; 1977), we develop hypotheses about which type of novelty is more likely to disrupt knowledge streams and which is more likely to consolidate them. By integrating data from the Web of Science (to measure the nature of scientific influence) with reflective essays written by authors of Citation Classics (to access novelty claims) and by joining computational text analysis with statistical analyses, we demonstrate clear and robust patterns between type of novelty and the nature of scientific influence. As we expected, articles that develop and present a new method tend to be more disruptive, likely because they tend to cross disciplinary divides, which boost creative insight and alters citation patterns (Abbott 2004; Koppman & Leahey 2019). To our surprise, new theory (especially the most abstract kind of theory) is associated with a more consolidating type of influence. This may be attributable to the nature of theory and theory-building, which tends to involve synthesis and bricolage.