Talking at Cross Purposes? A Computational Analysis of the Debate on Informational Duties in the Digital Services and the Digital Markets Acts

Abstract

In the latest Commission proposals, the Digital Markets Act (DMA) and Digital Services Act (DSA), ex ante informational obligations for online intermediaries, platforms, and ‘gatekeepers’ figure prominently. Some are new, others are already state-of-the-art for many operators. Because the efficacy of these duties is widely questioned, one wonders how they are implemented in the normative proposals. The question is largely uncovered in the literature. To fill this void, the paper investigates whether there was any agreement among the stakeholders who participated in the consultation over the DSA and DMA proposals. We do so by using NLP techniques to analyze whether key terms of transparency are used in the same way by different stakeholders.
We find significant differences in the employment of terms like ‘simple’ or ‘meaningful’ in the position papers that informed the drafting of the two proposals. These findings are informative for both rule-makers and legal scholars, and may explain why informational duties fail so often to reach their goal.

Keywords: Digital Services Act, Digital Markets Act, Big Platforms, Computational Analysis, Machine Learning, Transparency Duties, Law and Technology, Natural Language Processing, Word Embedding Model; Regulation

JEL Classification: K3, K4, O3