First Interdisciplinary, Longitudinal Study of Internet Shutdowns and Outages

Government-ordered shutdowns impact more than 1 billion of the world's Internet users globally. Since 2018, the IODA team has been documenting outages observed on the IODA platform. Through collaboration with political scientists and merging of IODA's outage database with Access Now's #KeepItOn STOP shutdown database, we conducted the first longitudinal analysis of government-ordered Internet shutdowns and spontaneous outages (i.e., disruptions not ordered by the government). Download and read the full paper using this link, checkout the recorded presentation, and read our overview below.

Recorded Presentation

Lead author, Dr. Zachary Bischof presents a 10-minute overview of the paper.


Our primary goal in conducting this research is to provide insights on the identifying features of government-ordered, national-scale Internet shutdowns and how they differ from spontaneous outages caused by power outages, severe weather, cable cut, misconfiguration, cyberattacks, etc.

Highlights from our insights:

  • Countries with shutdowns are the most autocratic scoring countries in the dataset. However, countries that experience spontaneous outages also score more autocratic than countries that experience neither.
  • Internet shutdowns are significantly more likely to occur on days of political mobilization. Shutdowns are 9 times more likely to co-occur with protests, 16 times more likely to co-occur with elections, and 286 times more likely to co-occur during coups. Outages are not significantly more likely to occur on days of political mobilization.
  • Spontaneous outages tend to have shorter durations compared to shutdowns.
  • Shutdowns are likely to occur precisely 1, 2, 3, or 4 days after a previous shutdown
  • Shutdowns are significantly more likely to start on the hour compared to spontaneous outages.

Please read the paper for more findings and in depth discussion. We hope these insights into distinguishing characteristics of Internet shutdowns can help allow for more effective and efficient monitoring for Internet shutdowns. In the future we also plan to create a "Shutdown Indicator" based on this data in IODA.


Please reach out with any questions or inquiries:

Authors: Zachary S. Bischof (Georgia Tech), Kennedy Pitcher (UC San Diego), Esteban Carisimo (Northwestern University), Amanda Meng (Georgia Tech), Rafael Bezerra Nunes (Yale University), Ramakrishna Padmanabhan (Amazon Web Services), Margaret E. Roberts (UC San Diego), Alex C. Snoeren (UC San Diego), Alberto Dainotti (Georgia Tech)