Social media platforms for academics, such as Academia.edu or ResearchGate, are spaces where millions of scholars share their research and construct themselves as legitimate and productive academic workers. These services provide new metrics of scholarly impact, such as download counts, social media sharing, popularity, or network reach. They also allow the self-publication of academic content, independent of the scholarly journal system. Digital media are often presented as a democratization of scholarly publishing, which breaks out from the market dominance of incumbents such as multinational publishing companies. Yet with the emergence of new platforms that move beyond traditional open-access media, scholarly publishing has exploded and must be approached with new conceptual lenses. Academic social media intensify academic labor, and at the same time assume an ever-increasing epistemic role. Are these infrastructures of evil?

Academic social media platforms are not merely the product of technological innovation, but also represent the answer to a crisis in processes of value production and accumulation in academic labor. Two phenomena have converged to create this crisis. First, there is the neoliberal university’s obsession with numerical metrics, such as journal rankings and citation counts. Second, there is the explosion of casualized and precarious academic labor, whose output cannot be fully accommodated by scholarly journals and thus is not valorized through traditional publishing systems and metrics. Within this framework, an analysis of the political economy of academic social media highlights its effect on the academic workforce and also illuminates the ways in which knowledge is produced and validated. Indeed, these platforms are privately owned and controlled, based on value-extraction logics that can become parasitic. They tend to become bottlenecks that actively construct the boundaries of a discipline. Finally, they are based on nontransparent algorithms that process data about social interactions within the platform to determine the value of scholarly content. These platforms shape the activities of their users, who have to deal with new forms of gatekeeping and the algorithmic logics that govern them. In sum, they produce and discipline a new kind of academic.

Let me offer an example from a well-established service. One of the flagship infrastructures of the open-access movement is arXiv, a preprint repository that emerged from forms of epistolary preprint exchange in physics. Founded by particle physicists in 1991 and now run by Cornell University, arXiv has become a hegemonic space for disciplines such as physics, mathematics, and other quantitative sciences. For example, it collects more than 90 percent of all articles published in particle physics. ArXiv’s role as a central communication platform makes publication in scholarly journals almost irrelevant for some physicists. ArXiv also shapes publication and credit attribution practices for physicists, who need to abide by the temporalities and authorial practices shaped by the website. At the same time, not being based on peer review, arXiv uses algorithmic and human moderation to filter out inappropriate articles. Manuscripts may be rejected, thus excluding authors from participation in the core infrastructure of a discipline—if your work is not on arXiv, one might plausibly say, you are not a physicist. Criticism against arXiv is rare, but strikes at the core of its functioning and role. Scientists have at times accused arXiv of abuse for rejecting articles that are deemed uninteresting or fringe, or for marginalizing scholars who are not recognized as community members—for example, those who do not work in an academic lab or department.

Of course, arXiv is a nonprofit website run by members of the scientific communities it serves, and problems related to its role in patrolling the boundaries of a discipline may sound marginal. But for commodified publishing repositories, these and other issues are potentially explosive. This is the case with Academia.edu, which is privately owned, or the Social Science Research Network, which is controlled by the global academic publisher Elsevier. These corporations follow a data-capture model and capitalize on information freely provided by users through the platforms themselves. To do so, they need to datafyacademic content and interactions, thus increasing their ability to extract relevant information from them. These platforms are part of a larger trend to increase the calculability of research outputs. Their numerical metrics go beyond the impact factors and citation counts that represent the standard measures of impact in the neoliberal university, also including metrics such as download counts and network reach. For example, Academia.edu positions scholars in the top 1 percent or 0.1 percent according to opaque algorithms that calculate a scholar’s impact, while ResearchGate calculates one’s “RG score.”

As social media platforms for academic publishing broaden their reach and are increasingly used as spaces where content is shared, bonds created and maintained, and reputation constructed, a thorough analysis of their political and epistemic dark sides becomes urgent. The main result of their emergence may not be increased access to knowledge, but rather a redefinition of the relevance of knowledge in numerical terms. This exacerbates the neoliberal university’s role in producing entrepreneurial academic subjects, as it shifts performance evaluation from citations to a continuous and pervasive process of microcontrol that measures one’s ability to perform an appropriate algorithmic identity and adhere to accelerated temporalities. Not surprisingly, the new quantified academic produced by the widespread use of these platforms experiences an ever more extreme blurring of the boundaries between work and nonwork, and abides by new forms of self-discipline and self-surveillance (see Gill 2010). Yes, academic social media platforms may allow an increasingly casualized academic workforce to find routes around incumbent powers, such as editorial boards or impact factor systems. Yet they also produce an intensification of academic labor that may reinforce entrenched academic hierarchies—after all, tenured faculty do not need to worry too much about Academia.edu, at least for now. This has epistemic consequences too. These services have the ability to gather and analyze large datasets about reading, citation, and interaction patterns and thus to assign an arbitrary value to a piece of scholarship. An increase of their reach and pervasiveness will give them an unprecedented power over processes of gatekeeping and validation. What kinds of scholarship will they engender?