The small-world network of global protests

Abstract

Protest diffusion is a cascade process that can spread over different regions of the planet. The way and the extension that this phenomenon can occur is still not properly understood. Here, we empirically investigate this question using protest data from GDELT and ICEWS, two of the most extensive and longest-running data sets freely available. We divide the globe into grid cells and construct a temporal network for each data set where nodes represent cells and links are established between nodes if their protest events co-occur. We show that the temporal networks are small-world, indicating that the cells are directly linked or separated by a few steps on average. Furthermore, the average path lengths are decreasing through the years, which suggests that the world is becoming “smaller”. The persistent temporal hubs present in both data sets indicate that protests can spread faster through the hubs. This topological feature is consistent with the hypothesis that protests can quickly diffuse from one region to any other part of the globe.

Bots influence opinion dynamics without direct human-bot interaction: the mediating role of recommender systems

“…A recent study found that human accounts are significantly more visible during political events than unverified accounts (González-Bailón and De Domenico 2021). This finding casts doubt on the centrality and impact of bot activity on political mobilizations' coverage (Ferreira et al 2021). Overall, these findings show that, notwithstanding the well-documented spread of bots and troll factories on social media, their effect on influencing opinions may be limited.…”

An agenda for addressing bias in conflict data

“…If taken at face value, the size of these collections may give the impression that a vast number of conflict events are missed by researcher-led datasets due to the comparatively limited capacity of subject-matter experts. Even when those presumptions are proven false 2 and the large tranche of false positives are uncovered 3 , 4 , many analysts believe that automated data collection provides a relatively accurate ‘big picture’ of patterns, while getting some smaller details wrong 5 , 6 .…”

Does Privacy Regulation Harm Content Providers? A Longitudinal Analysis of the Impact of the GDPR

“…Our empirical strategy focuses on capturing metrics correlated with websites' ability to provide new, quality content. The metrics include variables used in the previous literature, such as the amount of new content URLs generated by online publishers over time, the volume of traffic they receive, and the degree of social media engagement with new content (Shiller et al, 2018;Gallea and Rohner, 2021;Ferreira et al, 2021). In addition, we capture information such as content providers' exit from the market, and changes in revenue-and data-generating strategies, which may also signal websites' distress with the impact of the GDPR.…”

“…GDELT gathers and provides metadata for articles from news and media websites going back to 2015 from both domestic (US) and international sources. The database provides metadata including the URL, publication date, and publisher website for each article, and has been used in studies that examined global events (Gallea and Rohner, 2021;Ferreira et al, 2021). We use GDELT data to count the number of new URLs of content (GDELT URLs) published by each website in the sample in the week surrounding each wave of data collection (three days before and after each OpenWPM observation).…”