In January of 2023, a study was published in Nature which has generated a significant amount of discussion. Against the background of perceived exponential progress in innovation and technological progress, the authors analyzed 45 million papers over six decades of research across multiple fields of study and quantified their “disruptiveness”.
The authors of the study, using a previously established metric called the CD index and showing declines in the CD over time across numerous fields, concluded that the disruptiveness of published research has declined over time. The authors also demonstrate that this decline appears to be observable with alternative indicators, and show that use of language in scientific publications indicating departures from the status quo has been decreasing over time.
Their analysis, including that of citation practices, suggests that scientists are increasingly building their work on top of a narrower, and older, set of past knowledge. Reasons are discussed, including the career benefits that may accrue though the use of such strategies and the difficulty scientists may have in keeping up with the pace of novel developments.
It is suggested that depending on the factors that have produced this observed phenomenon, it may be amenable to change through adjustments in policy or other means of modifying incentives. The concerns that this new study brings to the forefront are, a least in this writer’s opinion, relevant to efforts to reform academic publishing, some of which we have previously discussed in this newsletter, as well as the increasing awareness of the need for increased accountability and integrity in research (see this article by the International Science Council for a brief discussion). Whatever the case may be, in order to preserve progress in our increasingly technologically dependent society and economy, we must pay careful attention to any significant obstacle to future innovation.
Click here for the Japanese version.