Exploring YouTube's Political Communication Networks during the 2024 French Elections

TL;DR

This study analyzes YouTube-based political communication in France during the 2024 European and legislative elections, comparing news-media channels (NM) with politicians' channels (PP). It builds a large dataset of 43.5k videos and 7.4M comments across NM and PP channels, using bipartite and augmented graph representations to map commenter networks and cross-channel interactions. The results show that left-wing and far-right actors drive higher activity and engagement, while PP channels form tighter, more cohesive communities; Shorts have a conditional, audience-dependent impact. Politicians’ appearances on NM channels increase audience overlap with PP channels, revealing an intricate mix of echo-chamber tendencies and cross-partisan engagement with potential political influence beyond traditional media boundaries.

Abstract

In 2024, France was shaken by the far-right National Rally's victory in the European elections. In response to this unprecedented result, French President Emmanuel Macron dissolved the National Assembly, triggering legislative elections just two weeks later. A whirlwind campaign followed, partly on social media, as is now the norm, and concluded with the victory of a left-wing coalition. This article examines the YouTube activity of two key actors during this period, news media and politicians, and the commenting behavior they generated. We built a dataset of 35 news media channels, 28 politicians and parties channels, 43.5k videos posted from three months before the European elections to one week after the second round of the legislative elections, and 7.4M associated comments. We examined upload activity and engagement across political orientations and used network analysis methods to uncover the structure of their commenting communities. We also identified politicians' appearances on news media channels and assessed their impact on commenting user bases. Our findings show that, among politicians and parties channels, far-right and left-wing ones were significantly more active and received substantially higher engagement (views, likes, and comments) than other groups, with denser and more clustered commenting communities. About 7% of commenters commented across political orientations and were much more active than in-group commenters. News media channels tended to favor politically aligned guests, while centrist politicians were over-represented. Finally, politicians' presence in the videos of a specific news media channel increased the share of commenters who were active on this channel and political channels, regardless of their orientation.

Figures (8)

• Figure 1: Example of an AVCG, using the videos and commenters from the channel "Les Républicains". A shows the bipartite network with videos and commenters, and an edge when a commenter commented on a video. B shows the same network with added edges between commenters if they co-commented on at least two videos from the channel.
• Figure 2: Empirical CCDFs of uploads and view counts, grouped by orientation. For each NM and PP orientation, we aggregated all videos from channels of that orientation and computed the distribution of uploads and view counts.
• Figure 3: Network measures aggregated over channels' orientations. Each measure was computed for each channel and then averaged over channels of the same orientation.
• Figure 4: Percentage of commenters who commented on videos from channels in the y-axis group who also commented on videos from the x-axis group. Commenters are counted as part of a group if they commented at least once on any channel in that group.
• Figure 5: ChPWG of all PP channels in our dataset, with weighted edge between the 25 pairs with highest Jaccard similarities between channels' commenters.