A Year of the DSA Transparency Database: What it (Does Not) Reveal About Platform Moderation During the 2024 European Parliament Election

TL;DR

The paper analyzes 1.58B self-reported moderation actions from eight major platforms in the EU around the 2024 European Parliament elections via the Digital Services Act Transparency Database (DSA-TDB). It investigates whether platforms adapted their moderation in response to electoral integrity risks and whether the database reliably reflects such changes. The findings reveal no robust, systematic election-driven shifts in moderation practices and highlight persistent data-quality issues, including duplicates, pre-publication moderation records, and pervasive use of vague fields, with one notable election-related anomaly on LinkedIn. The work underscores limitations of self-regulatory transparency tools and argues for stronger regulatory enforcement and improved data access mechanisms (e.g., Article 40) to safeguard democratic processes.

Abstract

Social media platforms face heightened risks during major political events; yet, how platforms adapt their moderation practices in response remains unclear. The Digital Services Act Transparency Database offers an unprecedented opportunity to systematically study content moderation at scale, enabling researchers and policymakers to assess platforms' compliance and effectiveness. Herein, we analyze 1.58 billion self-reported moderation actions taken by eight large social media platforms during an extended period of eight months surrounding the 2024 European Parliament elections. Our findings reveal a lack of adaptation in moderation strategies, as platforms did not exhibit significant changes in their enforcement behaviors surrounding the elections. This raises concerns about whether platforms adapted their moderation practices at all, or if structural limitations of the database concealed possible adjustments. Moreover, we found that noted transparency and accountability issues persist nearly a year after initial concerns were raised. These results highlight the limitations of current self-regulatory approaches and underscore the need for stronger enforcement and data access mechanisms to ensure that online platforms uphold their responsibility in safeguarding democratic processes.

Figures (11)

• Figure 1: Daily time series of the number of moderation actions (cyan-colored) and their average delay (orange-colored). The blue vertical band indicates the Parliament elections days (6th--9th June), while the green vertical dashed line indicates the President election day (18th July). The red circles highlight the subset of anomalies that we analyzed in detail.
• Figure 2: Comparison between the SoRs (right-aligned, red-colored) that caused the moderation volume anomaly reported by LinkedIn on May 8, 2024 and the SoRs (left-aligned, blue-colored) by the same platform from the surrounding routine days.
• Figure 3: Comparison between the SoRs (right-aligned, red-colored) that caused the moderation volume anomaly reported by TikTok on May 10, 2024 and the SoRs (left-aligned, blue-colored) by the same platform from the surrounding routine days.
• Figure 4: Comparison between the SoRs (right-aligned, red-colored) that caused the moderation volume anomaly reported by YouTube on July 21, 2024 and the SoRs (left-aligned, blue-colored) by the same platform from the surrounding routine days.
• Figure 5: Comparison between the SoRs (right-aligned, red-colored) that caused the moderation volume anomaly reported by Pinterest on August 21, 2024 and the SoRs (left-aligned, blue-colored) by the same platform from the surrounding routine days.