Veranstaltungsprogramm

State-backed information operations leveraging social media manipulation and coordinated inauthentic behaviour first arrived in the centre of widespread public attention after the 2016 US presidential election, when a large number of inauthentic accounts operated from a now infamous troll farm in St Petersburg sought to influence the election in favour of Donald Trump. While the question whether the campaign had a significant impact on actually swaying public opinion towards the Republican candidate might remain an eternal unknown, the coordinated inauthentic operations backed by the Russian government did succeed in reaching millions of US voters both through organic growth, paid advertisements, and in part also automated accounts.

With more and more public attention attributed to the threat of foreign interference, platforms, researchers, and law enforcement agencies have since uncovered dozens of other cases in which governments, most of the time authoritarian ones, have resorted to social media manipulation to conduct information operations abroad. While the 2016 IRA troll factory represented a relatively straightforward case where the modus operandum of the perpetrators was both revealed and attributed, many of the subsequently discovered operations featured much blurrier boundaries. For example, Twitter and Facebook have suspended and disclosed inauthentic networks which in some cases were directly to CCP-backed actors, while other cases saw much blurrier boundaries where elements of central coordination seem to have mingled with decentralised forms of (semi-)organic content creation such as coordinated firewall hopping. A similar trend seems to have emerged during Russia’s war on Ukraine in 2022, where research disclosed by the UK Foreign Office showed that Russian actors relied on a mix of paid troll workers and voluntary content creators loosely coordinated via Telegram.

Using a number of sources and methods, including primary testimony from an interview with a former employee of the IRA’s St Petersburg troll factory and investigative research experience from numerous information operations previously discovered by the author, this presentation will sketch the evolution of modus operandi of state-backed information operations, describe the main epistemological challenges, and propose methodological solutions and innovation for future research.

Researchers still try to understand politics, power and security in the Cyber Space by adapting modern pre-cyber theories of International Relations to the issue. Yet, what if those attempts are inadequate because of the incompatible nature of those theories towards the matter they try to investigate? Traditional understandings of space, actors and activities seem to be very problematic in the eye of the fluidity of cyberspace. Borders between those categories are blurring and it becomes very difficult to understand and describe the phenomenon without having a regress in timeworn pre-cyber theories.

I argue, that it is worth to return to the starting point of understanding politics and international relations, to re-assess the phenomenon of cyberspace at hand. The circumstances might have changed, but the philosophical foundations of political analysis and understanding have not. Instead of trying to adapt offline concepts into the online world, we should try to rethink our perception.

Phenomenology provides ways of perception and analysis, that are capable to develop new and maybe even more accurate methods to investigate Cyber Space. There are several important aspects of Cyber Affairs that need re-definition, to which the school of phenomenology has in my opinion the right approach. This does not mean that we have to abandon all insights of already existing theories, but we can learn from the difficulties they have with the subject and the fallacies they occasionally made.

Space is one of the subjects that lost its certainty and form in Cyber Space and contemporary theories seem to have their difficulties with the fluidity of it. Taxonomy and Morphology of the phenomenon Space in Cyber Space will be re-asserted in this paper, this includes as well human factors and agglomerations of power in the net and their inherent fluidity.

With the phenomenologic demand for the definition of Chorology and Chronology the interlink between contemporary online and offline world will be sought as well as its genesis and constant reproduction.

It is an ongoing debate in academia whether the use of social media increases political polarisation and the fragmentation of the public. Many empirical studies on these phenomena make use of network analysis to demonstrate the emergence of homophilic clusters on Twitter. These approaches often rely on hashtag- or keyword-based samples and reconstruct networks according to information such as follows, mentions or retweets. Thereby, researchers leave out an arguably fundamental aspect of communication via social media: the possibility of two or more users to interact directly with each other via replies. The discourse dynamics unfolding in direct replies may be responsible for the emergence of conflictual discussions and their de/escalation. Such conflictual discussions may contribute to polarisation, which, again, cannot become available to analysis when methodologically focusing on isolated tweets. Accordingly, we present a Twitter dataset that shifts from keyword- or hashtag-based collections of isolated tweets to the reconstruction of reply-networks in a tree structure where branches of the tree represent chains of direct replies by users to each other. The dataset is based on Twitter activity during the first presidential debate of the US election 2020. It enables us to analyse both discourse and network dynamics by combining qualitative, quantitative and computational methods. The interactions in the conversation tree can be studied using qualitative methods such as conversation analysis, and our initial observation is that while conversations were only a small proportion of the Twitter activity during the first debate, many of those conversations were conflictual. Our future work involves using qualitative coding of discourse dynamics to improve approaches of machine learning and natural language processing by taking sequentiality into account, and we will use quantitative and network analysis to test hypotheses about structural patterns and types of discourse dynamics derived from qualitative analysis. An overarching aim is to assess to what degree conflictual discussions should be understood as an expression of a preexisting polarisation between political opponents and to what degree we should understand conflictual discussions as a phenomenon that reflects de/polarisation as an ongoing process.

Social media have become central arenas of social polarization. It is not only ‘hatespeech’ between individual users and the ‘brutalization’ of communication culture that contributes to this. Moreover, strategic actors are intentionally exploiting social media logics in order to increase polarization in their own interests. Especially on Twitter the existing possibilities for automating communication are heavily used for this purpose. In particular, the use of so-called socialbots has gained a dubious fame in this regard. As previous research has shown, these automated actors are mainly used to disseminate polarizing content (Bastos and Mercea 2017), which often promotes political positions from the right-wing political spectrum (Bessi and Ferrara 2016; Kollanyi et al. 2016; Hagen et al. 2020). Up to now, however, there is no consent among researchers on how socialbots can best be detected and how exactly they contribute to polarization and influence communication.

In our presentation, we present an interdisciplinary and integrative approach to identify automated activities on Twitter and investigate their influence on processes of public opinion formation. The basis for this approach is an empirical case study about the emergence of the hashtag #laschetlacht on twitter during the German federal election campaign 2021. In this study, we detected automated activities and investigated their influence on public opinion formation. The results indicate that (partially) automated accounts significantly contributed to the spread of the hashtag. Moreover, the dissemination of the hashtag drew the attention of the mass media to a faux pas of the candidate for chancellor Armin Laschet, which subsequently led to significant negative coverage and poor poll ratings. That is, in the investigated case study automated activities contributed to the development of public attention dynamics that may have influenced the outcome of the federal election.

For several decades now journalism (research) has been characterized by its constant readjustment as it seeks to capture the transformation processes triggered by the digitalization of media production, distribution and use. In the course of this development, journalism has also become an increasingly interesting attractive application field for computer science: Journalistic content, including accompanying user comments, can be inexhaustibly reviewed as “data material”; almost all social media, algorithms and AI-based forms of automation play a role in each phase of news production, distribution, and usage; and journalistic “products”, are increasingly “built” using methods from software development (Loosen/Solbach 2020; Hepp et al. 2021a).

This is the background against which I will first take a disciplinary look at journalism in order to outline the development of technically supported news production, as well as interrogating communicative phenomena that are “suspected” of being journalistic or that participate in the manufacturing of the public sphere in a significant way – such as algorithms, communicative AI (Guzman/Lewis 2020) as well as utterances of audiences. This will be followed by an interdisciplinary exploration along the intersection of journalism (research) and computer science, for which I use examples from own research projects that address the coming together of the two disciplines (Reimer 2022). In this way, it becomes clear that the development of (media) technology is just as important for the scientific problems and the questions faced by journalism research as it is for its connection to other disciplines and the development of methods, which, with the increasing importance of computational approaches (Haase et al. 2022; Hepp et al. 2021), is rendering the field increasingly inter- and transdisciplinary (Loosen et al. 2022). In sum, I consider journalism in the context of inter- or “de-disciplinary” research as a kind of case study for media change, changes to the public sphere, and societal change, and the reflexive relations that exist between them. In this way, I want to show that this kind of de-disciplinary research is about oscillating between different points of view and that interdisciplinarity and transdisciplinarity primarily stand for the “mindsets” of mutual understanding and do not exist on purely operational terms.