Data access rules in the DSA: SNV's response to the European Commission's call for evidence

The EU's Digital Services Act (DSA) contains an important new rule requiring online platforms to provide data to vetted researchers in certain circumstances. The details of this rule will be fleshed out in a "delegated act", for which the European Commission requested feedback. SNV experts Dr. Anna-Katharina Meßmer, Dr. Martin Degeling and Dr. Julian Jaursch submitted a response to the Commission's call for evidence, which can be found at the download link.

The summary of the key points can be found below (and also at the beginning of the document submitted to the Commission).

SNV strongly welcomes the DSA’s intended push for improvements in platform-to-researcher data access. A privacy-friendly, accessible EU-wide regime is necessary to replace the current system of arbitrary and piecemeal access points provided by some platforms on a voluntary basis. This system has created unequal access for researchers, offers incomplete data and lacks key technical features, as SNV has experienced first-hand by conducting platform research and using several platforms’ APIs.

While the provisions in Article 40 DSA are a good start, they do need to be spelled out in greater detail. That is why we applaud the Commission for focusing on this issue and starting work on a delegated act as provided for in Article 40(13). The delegated act can ideally lead to major improvements in how researchers can access platform data. To achieve this goal, we offer several suggestions in this response to the Commission’s call for evidence.

Data access needs
- Current research APIs only provide a limited set of variables and less than is publicly available, for example, via scraping. While the current trend is to limit APIs and the available data, we suggest the opposite: All data that can already be accessed by scraping should be available through APIs.
- Research often struggles with lack of relevant meta data. That is why platforms need to properly document data schemas and content removal.
- It is necessary to clarify that scraping of public data is in accordance with Article 40(12) to remove legal uncertainties about this vital research practice.

Data formats and technical specifications
- Platforms should offer regular APIs as well as stream APIs.
- Experimentation with automated user accounts should be possible on the platform or in sandboxes to study algorithms like recommender systems.
- Data donations and exploratory research should be allowed.

Vetting process
- It should be reiterated in the delegated act that researchers both from academic and civil society organizations can apply for data access, including those from outside the EU.
- In addition to applications regarding specific research projects, it should be allowed for researchers and institutions as such to apply, which would further support exploratory and inductive research.

Governance structure
- We strongly suggest setting up an independent advisory mechanism as mentioned in Article 40(13). This body, made up of academic and civil society researchers, could facilitate the development of common vetting standards, support DSCs in the actual vetting process and serve as a forum for exchange.
- The Commission, the European Centre for Algorithmic Transparency, the DSCs as well as researchers themselves should immediately engage in capacity-building efforts to make the best use of Article 40. Member states should equip their DSCs with strong research and data science units to conduct data analyses themselves and support researchers.