Informing Regulatory Reasoning on Algorithmic Systems in Societal Communication with STEAM

The project, which is located between media law and computer science, aims to use a new method to visualize the functioning of news distribution on digital platforms and thus offer media regulation opportunities for new regulatory approaches.

Societal communication has changed profoundly in the wake of ongoing digitalisation. As new players are active and algorithmic systems as well as other technologies are used, new approaches to media regulation may also be required. While the ideal of maintaining media diversity has been the primary concern so far, it is now no longer scarce channels but complex systems of the most diverse actors that determine what chance a certain content has of being perceived.

Researchers from the Universität Hamburg and the Leibniz Institute for Media Research are developing a new so-called “Socio-Technical Ecosystem Architecture Method” (STEAM) using the case studies “Facebook” or “Facebook News” and the American social networking platform “Parler”.

The method being developed will provide a holistic view of news dissemination in an ecosystem such as that of news dissemination on Facebook, and will help to map these ecosystems with their relationships between actors, data flows and software components in such a way that problems can be identified and, if necessary, provide linkage opportunities for new regulatory approaches.

The project is one of seven project consortia from the social and technical sciences funded by the Volkswagen Foundation. All selected projects are scheduled to run for three to four years and will each receive funding of around 1.5 million euros. Further information on the Volkswagen Foundation’s initiative “Artificial Intelligence and the Society of the Future”.

Development of the Method

The project aims to develop a socio-technical ecosystem architecture methodology. This methodology builds on existing architecture concepts in computer science and complements them with normative concepts as well as frameworks with a legal and ethical background and horizon scanning approaches. This allows for a holistic description of an ecosystem, its actors and its action dynamics, and reveals risks to public communication as well as suitable starting points for counteracting them in a regulatory way.

Novel actors and technologies used by them, particularly related to artificial intelligence (AI), have gained relevance in the generation, aggregation, selection, curation, and prioritization of content. At the time a content item is displayed in an individual’s sphere of attention, numerous decisions are made beforehand that influence the content’s dissemination and the probability of ending up as part of the individual’s media repertoire. Specifically, various actors (e.g. content creators, digital platforms or content providers) involved in the entire process from content creation to content provision have influence on societal communication. In this relation, these actors also increasingly use algorithmic systems and, in particular, AI. Two examples illustrate this phenomenon:

1. The first illustrative example is provided by Facebook and its news-aggregating application Facebook News. The platform is constituted by a constellation of human, technical, and institutional actors and dynamics as well as interdependencies among them. Ultimately, human curators, together with recommendation and filtering algorithms, generate a selection of media contributions that are offered to the user. However, the pool of available contributions is already narrowed by various limiting factors in the broader environment. Facebook, together with external partners, selects media outlets that are admitted. Furthermore, Facebook determines publisher guidelines that providers need to adhere to and monetization schemes that publishers need to agree to. Therefore, the media repertoire of a user of Facebook News is co-determined by a range of decisions made by various actors.
2. The second example considers the American social networking platform Parler, which is often discussed for user content involving far-right posts, antisemitism, and conspiracy theories. In January 2021, three companies exerted meaningful influence on the accessibility and usage of Parler. Both major mobile platform providers – Google and Apple – removed Parler from their respective app stores on iOS and Android. In addition, Amazon excluded Parler from its cloud hosting service Amazon Web Services (AWS). Through this intervention by Amazon, Parler was apparently unable to operate its platform and had to go offline. The example emphasizes that  parties who fulfil different functions in making content accessible have an influence on what content is brought to the attention of users without prior judicial or regulatory decision to do so and without having been in the spotlight as important actors in societal communication.

Both cases demonstrate challenges to pre-existing mechanisms for safeguarding societal communication and its essential functions. The cases might hint at problems for relevant qualities of societal communication, which are not covered by the current communication law system. Therefore, we need to fundamentally rethink how we can continue to ensure that socially relevant communication takes place both openly and freely.

To achieve the research objectives, we develop the Socio-Technical Ecosystem Architecture Method (STEAM) to enable a holistic view on news dissemination in an ecosystem. This method will help in the presentation of ecosystems and their actors and relations in a way that forms the basis from a regulatory perspective to allow the assessment of the potential influence of actors. To develop STEAM, we integrate architectural thinking and normative reasoning within three iterations of a design science research (DSR) appraoch.

Photo by fabio on Unsplash