Annabell, T. (2026). Regulating a floating signifier: Interpretations of “data” in the digital services act researcher access provision. Media, Technology and Governance. https://doi.org/10.1177/29777984261486601
Summary
This article interrogates a conceptual blind spot at the heart of the European Union’s Digital Services Act (DSA): what “data” actually means in Article 40, the provision granting vetted researchers access to platform data to study systemic risks. Despite its centrality, “data” is never formally defined in the DSA or its Delegated Act. Drawing on Laclau’s notion of the floating signifier, Annabell argues that this definitional vacuum is not neutral but strategically exploitable — platform companies narrow “data” to aggregated, anonymized, quantified, predefined datasets, a framing that sits in tension with the plural understandings of data across academic disciplines, especially qualitative traditions in the humanities and social sciences. The paper reframes data access from a technical or procedural matter into a governance struggle shaped by the political economy of Big Tech.
Key Contributions
- Identifies the concept of “data” itself — rather than data quality, access conditions, or transparency — as an under-examined site of contestation in DSA Article 40 scholarship.
- Applies Laclau’s floating signifier as an analytical lens to platform data governance, contrasted against the boundary object concept.
- Offers an empirical mapping of how VLOPs/VLOSEs articulate “data” through feedback submissions and published data catalogs during early implementation of the Delegated Act.
- Connects definitional ambiguity to concrete governance consequences — notably the risk that commercial and computational framings become the default, marginalizing qualitative research paradigms.
Methods
Fairclough-informed critical discourse analysis applied to three corpora: (1) regulatory texts, including Article 40 and Recitals 96–98 of the DSA and the draft and final Delegated Act, plus guidance from the Irish Digital Services Coordinator; (2) ten public feedback submissions (covering 15 of 25 designated VLOPs/VLOSEs) from the October–December 2024 consultation on the draft Delegated Act, coded by article and recital; and (3) publicly available DSA data catalogs published from November 2025, analyzed for format, terminology, data types, and categories. The analysis is theoretically grounded in post-foundational discourse theory and critical data studies.
Findings
- “Data” appears in none of the DSA Article 3 definitions or the Delegated Act’s Article 2 definitions; only personal data is cross-referenced via GDPR.
- Recitals tied to Article 40 gesture toward narrow numerical/aggregated interactional data, while Delegated Act Recital 11 offers a broader non-exhaustive list (profile data, engagement histories, A/B test results, moderation changelogs, internal documentation).
- The Delegated Act’s language of “data assets” and mandated data catalogs reflects and reinforces a view of data as owned, value-generating platform resources repurposed from advertising or developer contexts.
- Booking.com and Amazon argued data should mean only quantitative raw datasets; X objected to raw data access; Meta insisted providers should not be required to create new data (loggers, pipelines) and raised concerns about researcher competence.
- Data catalogs reveal divergent operationalizations: Meta and Google link to existing datasets/APIs; TikTok, Amazon, and Temu provide tables privileging quantified metrics; Zalando offered the widest range (including A/B testing and moderation docs); Shein’s vague “data on…” framing may inadvertently widen interpretive possibilities.
- Platform operationalization overall aligns with computational/data science conventions, risking the marginalization of qualitative social science and humanities understandings of data.
Connections
This paper is primarily a critical-data-studies and governance intervention rather than an empirical access study, but it speaks directly to the practical and procedural analyses of Article 40 and platform data access in Rieder2025-ju and Rieder2026-pp, and complements critiques of platform gatekeeping and the history of data access infrastructure in Freelon2018-ao and Gillespie2010-as. Its concern with how platforms institutionally define the terms of research access resonates with broader platform-power analyses such as van-Dijck2018-up and Gillespie2022-jx.
Podcast
A research-radio episode discusses this paper: 🎧 MP3 · Spotify