Paper summary: Opening Digital Borders Cautiously yet Decisively: Digital Filtering in Saudi Arabia (FOCI 2020)

Opening Digital Borders Cautiously yet Decisively: Digital Filtering in Saudi Arabia
Fatemah Alharbi, Michalis Faloutsos, Nael Abu-Ghazaleh (video and slides)

This paper is the first systematic, longitudinal study of digital filtering in Saudi Arabia. The focus is on the filtering of web sites (the Alexa top 500 in 18 content categories) and social media / communications apps (18 apps such as iMessage, Line, and WeChat). The authors made three measurements, a year apart, between March 2018 and April 2020, from vantage points in four cities and three ISPs. The overall development since 2018 has been towards less filtering. There is a moderate decline in the blocking of web sites, and a marked decline in the filtering of mobile apps: of 18 apps assumed to be blocked in 2017, all but one (WeChat) were usable in 2020.

For each web site, they did a DNS lookup over UDP and TCP, a direct TCP connection to port 80, an HTTP request directly to the target web server, an HTTP request to a known-unblocked server but with the target domain name in the URL path, and a TLS connection directly to the target web server. From these tests they conclude that web site filtering is based on HTTP and TLS features, not DNS or IP address. Each domain name is either wholly blocked or wholly accessible, which means it is sometimes possible to, for example, read the articles of a blocked news site on an unblocked service like Twitter or Instagram. For mobile apps, they attempted to install each app on a phone in Saudi Arabia and a phone in the US, then use the app’s text, audio, and video features. Over three years, apps that support voice and video calling have gone from being almost completely filtered to almost completely unfiltered. This latter phenomenon corresponds to a deliberate relaxation of policy towards communication-oriented apps that occurred in 2017. Some instances of filtering can be associated with specific political events. Examples include the blocking of ISIS-related web sites and certain foreign news sites.

Exceptionally for a study of this kind, the authors consulted with a government official and expert in local law, who stated that the study did not violate the law.

Thanks to the authors for reviewing a draft of this summary.