In our article, we have analyzed these dynamics at the single paper level while multiplying the case studies, in different disciplines: biology, mathematics, economics and sociology. We developed a very original and mixed-method that had never been used before to understand the dynamics of scientific research. We started by building an egocentric co-citation network to capture the community of researchers who cite the same references as the paper under study. We then studied the structure of the network and its evolution, five years before and five years after the publication. We established that these changes can be understood according to a double dynamic. First, the increase in the density of co-citations links between cited references of the paper is an indication of the vitality of the community around it. Second, the clusters tendency (modularity) is an indication of the sub-structuring of the scientific community surrounding the article. What leads to such destinies for scientific papers?

Our results show that the increase in the vitality of the community around the paper is correlate to the powerful position and resources of its main author. The increase in the structuring of the community is linked to a more intellectually rich and varied context surrounding the paper. The combination of these two indicators (depending on whether they are positive or negative) has defined four rather exclusive destinies for the communities surrounding articles: polarization, clusterization, atomization or attrition.

Our study shows that the social, institutional and intellectual characteristics of the authors play a role in the destinies of the articles. In the polarization dynamic, the main author of the article is present and central in the network of co-citations. He is more often a man, senior and the emblematic discipline is economics. On the opposite, the attrition dynamic shows fewer central authors, more women and juniors and especially few economists and biologists. The professional and social status of some authors provides them with more ressources, which is in line with some recent results on academic capital in academia (Gonzalez-Brambila, 2014). Our results further extend the issues related to social inequalities among researchers by showing that intellectual exchanges (citations) are also part of the process.

We also interviewed researchers about their relationships with the people cited in their articles and those who cite the same references as them. We have seen that the social relations between citing and cited authors are more distant (they do not know each other or not very well), where structured sub-communities are developing. The links are stronger when there are no such dynamics. This brings us back to very classic results in network analysis (Granovetter, 1973). Weak ties between authors lead to the bringing together of different worlds and stronger ties coincide with a delimitation of the community. We have also highlighted that chains of ties are more present when co-citation networks are consolidated (when density increases). On the contrary, network transience co-occurs with islets wherein small groups of relatives are immersed in an unknown or unfamiliar environment. Here again, these results are in line with those that conclude that chains of ties are important for collective action (Tindall, 2004). Our results show that social relations are also operating in the intellectual context of science, as in any social domain. They are embedded in the citations, which are therefore key mechanisms for structuring scientific research.

We have also identified two types of community vitality around articles. These result in part from the resources mobilized for the article, including the authors’ expectations of the article. Articles presented as research capitalization (polarization dynamics) accumulate the most individual, institutional and social resources. Those that expose a research promise (clustering dynamics) need social resources (co-authors and chains of links) but also intellectual references, especially at the international and multidisciplinary level. The difference between the two dynamics underlines the importance of intellectual ties, i.e. unknown persons or people on whom researchers are just “keeping an eye” (Michaelson, 1993). This is in line with earlier findings showing the importance of purely intellectual exchanges between researchers and their interlinking with social relations (Crane, 1972). In our study, we observed that these intellectual articles are in the end the most cited, as if their promises had been fulfilled.

In conclusion, we confirm that it is not possible to predict the success of a scientific article. But we propose to better understand the conditions that support a particular destiny. These are both social and intellectual and it is important to consider the two aspects. Citations, which are key mechanisms in the multi-level structuring of research, are very helpful indicators in determining these dynamics. Our work shows that the intellectual, institutional and social conditions of research activity are not the same for all researchers and research fields. Whether it is to reducing inequalities due to academic capital or to promote more original and promising research, researchers must be given the appropriate conditions so that they read other work and be inspired by it. Therefore, providing stability and recognition to researchers that foster their intellectual curiosity would be an excellent program for research policies.

The original article upon which this translation is based can be found at: https://asistdl.onlinelibrary.wiley.com/doi/full/10.1002/asi.24732