Mission
Social Data Science research facilitated by the EU’s DSA
The DSA provides two new types of access. One is to auditors, for the purpose of checking compliance with DSA regulations. The rules governing audits are set out in a Delegated Act of the DSA: they give auditors broad powers to access platform data and algorithms, and to conduct experiments. The other type of access is to vetted external researchers, in particular through Article 40(4). The Delegated Act defining this kind of access is still being finalised, but the core purpose of this access is to conduct new studies into societal risks associated with VLOPs, and to gauge the effectiveness of any risk mitigation measures already in place. Researcher access to VLOPs is crucial for these purposes, because the best experimental designs for the questions to be answered very often make use of data or methods that are only available within platforms. The brief for researchers is necessarily broader than that of auditors, because researchers are not assessing compliance against a finite checklist of obligations: their brief is to identify risks of VLOPs for users and society, and thus in some sense to define the detailed checklist for auditors’ assessments.
The Alliance is concerned with the structures that are needed to organise and coordinate an effective programme of research into VLOPs, under the auspices of the DSA. What is required—and what the DSA must enable—is a structured programme of research, rather than a piecemeal collection of individual research efforts.
The processes through which applications for DSA-enabled research access are vetted and approved have already received some attention. Decisions will ultimately be made by the Digital Services Coordinators (DSCs) of individual member states, but there will of course be an advisory mechanism too. There will likely be a body that receives applications for access from groups of researchers, to conduct specified projects, and advises DSCs of the projects that most effectively pursue the DSA’s objectives, using the best scientific methods, and in compliance with agreed ethical standards. This body will have some of the functions of a grant-awarding institution, assessing the quality of research proposals and research teams: access to VLOPs is necessarily a scarce resource, just like research funds. Plans are already under way to create a body with this role. The European Digital Media Observatory (EDMO) has convened a working group to design an ‘Independent Intermediary Body’ (IIB) to support DSA-enabled research on VLOPs. This body is envisaged to have several roles – in particular, to ‘safeguard the independence of research’, and to ‘ease the legal liabilities’ of platforms which share data, and to ‘mediate disputes’ between companies and researchers. A key function of the proposed body would be to ‘vet researchers and their research proposals’. Given that this body advises on the detail of access to VLOPs, it is expected to contain members from VLOPs, as well as members from the community of independent researchers.
There is a need for a second body, alongside the IIB, to represent and support the community of researchers who apply for DSA-enabled access to VLOPs. The role of this second body, which we are now establishing, is to coordinate applications, to communicate results of work already conducted or currently under way (within the guidelines laid down for access, naturally), to suggest useful work that could be conducted under the DSA, to convey information about the methods that are available within companies (again, within the established guidelines), and the legal frameworks that allow access under the DSA, and to formulate policy advice based on the body of research findings. In sum, the role of the new body will be to help organise the new science enabled by the DSA, and similar transparency laws enacted elsewhere. This task is the natural province of the research community: as such it involves the convening of research meetings and methods workshops, processes of scientific review, and the publication of edited research collections, and perhaps the foundation of a new dedicated journal, or set of web resources.
There are many existing research organisations that could provide helpful input into a grouping of this kind. On the AI and technology side, groups like EurAI and AAAI have relevant expertise. The AAAI conference on Web and Social Media is particularly relevant. Social media research expertise could also come from ICSMS, and social science expertise from organisations like ICRBS and IOSSBR. The DSA40 Data Access Collaboratory at the Weizenbaum Institute is already gathering information about applications made under the DSA, and will certainly be an important partner. The DSA Observatory at the University of Amsterdam is doing related work, with a focus on legal aspects of access. EDMO can provide expertise in areas relating to disinformation, and wider political impacts. For impacts on mental health, and health more generally, guidance can be sought from the WHO (for instance, the Pan-European Mental Health Coalition), and from NIMH and similar groups. However, we believe there is a role for a new organisation, that coordinates research efforts across these diverse domains, with a specific focus on DSA-enabled research.
We are founding an organisation with the mission just outlined. The new organisation is called the ‘Social Data Science Alliance’.