Today, a fourth perspective may be emerging in academic communities. Computer scientists and information technology professionals have acknowledged that digital technologies can have negative implications on society. They seek to solve these social implications and often propose ways to solve them through using more advanced technologies. They are using the term computational social science to reflect the digitalised nature of society (i.e. it is computational) but seek to propose that their work has a dimension of social science as well. For example, the 2019 European Symposium on Societal Challenges in Computational Social Science1.2 states that they seek to
address one of the most pressing challenges in todayâs digital society: understanding the role that digital technologies, the Web, and the algorithms used therein play in the mediation and creation of inequalities, discrimination and polarisation.'
An example of this type of approach is Garimella et al. (2017). They demonstrate how controversy in social media could be decreased. The main contribution in their work is to develop a computational approach that works in a digital environment. The gist of their approach is a content recommendation considering the social network and the positions different user accounts have in the controversy. They validate their method by measuring how it functions in several Twitter data sets and a variety of methods developed by others and conclude that `our methods outperform other baselines' (Garimella et al., 2017, 88). This choice of terminology already suggests that their research contribution is not to describe society (real or artificial) but rather a computational approach to solve a social issue - and doing it more efficiently (with some metric) than previous work in this area. These types of factors are critical in the computer science domain where these algorithms are developed.
A similar example of this type of work is work around algorithmic discrimination. This phrase refers to potential discriminatory practices caused by machine learning systems, such as different rental recommendations in AirBnB depending on the neighbourhood's racial composition. refs The common approach in this work is to first quantify discrimination in some manner. Following this, it measures existing approaches to decrease algorithmic discrimination and propose some novel approaches that may be better than the existing approaches.
With this formulation of the focus, the meaning of computational social science seems to bear similarity to internet research and new media scholarship. These are established areas within traditional social sciences that study the interplay of a digital technology society. A potential differentiator between this form of computational social science and more traditional social science acknowledging technology (e.g. internet research) is the field of study. Based on the symposium series, most people have their main discipline related to computers of information technology. This is different from traditional social science perspectives, where the main discipline is in social science or humanities. The difference may manifest itself in the traditions, established practices or choice of theory and methodology for the research. As we have illustrated in the two cases above, the main contribution, for example, appears to be a computational approach that â on some measurements â decreases a negative social phenomenon or increases a positive social phenomenon. The success of these are evaluated against established standards set up in computer sciences, including, for example, computational efficiency (see e.g. Garimella et al., 2017, 88)