Mali: Social media disruptions amid 2018 presidential election?

Two days ago, on 29th July 2018, Malians went to the polls to vote in presidential elections and reported that social media was inaccessible.

This is not the first time that social media is reportedly blocked in Mali. Facebook and Twitter were blocked two years ago amidst violent protests against the detention of a popular radio host. Last month, Internet Sans Frontieres reported that social media was disrupted in Mali, following a violent police crackdown on banned opposition protests calling for transparency and accountability in the 2018 presidential elections.

In this post, OONI and Internet Sans Frontieres share OONI network measurement data, corroborating local reports on the blocking of WhatsApp and Twitter during Mali’s 2018 presidential election.

Measuring internet censorship in Mali

Daily OONI network measurements have been collected from Mali for more than a year, thanks to our local partners. These measurements, collected mainly from Orange Mali (AS30985), show that most of the (internationally relevant and local) tested websites have been accessible, along with popular instant messaging apps, like WhatsApp and Facebook Messenger.

Access to WhatsApp and Twitter though was suddenly disrupted last Sunday on election day, as suggested by OONI data. We also observe a censorship policy change in relation to the accessibility of circumvention tool sites guardianproject.info and anonymizer.ru.

WhatsApp

OONI’s WhatsApp test is designed to measure the blocking of WhatsApp’s app and web version. The test attempts to establish TCP connections to WhatsApp’s registration server and endpoints, and performs DNS lookups and HTTP requests to web.whatsapp.com. Over the last year, this test has been run on a daily basis from Orange Mali (AS30985), with most (if not all) measurements confirming the app’s accessibility in the country.

WhatsApp though may have been blocked on 29th July 2018, as suggested by OONI data. The table below shows results (from AS30985) pertaining to the testing of WhatsApp’s registration server, endpoints and web version in the evening of Mali’s latest elections.

Our findings suggest that WhatsApp’s web version was likely blocked, since TCP connections to web.whatsapp.com were unsuccessful and rendered errors. Access though appears to have been restored by 19:26 local Mali time. Quite similarly, access to WhatsApp’s app might have also been blocked, since TCP connections to WhatsApp’s registration server failed.

It’s worth highlighting though that WhatsApp wasn’t blocked by all ISPs. Many OONI measurements, for example, collected from SOTELMA (AS21271) show that both WhatsApp’s app and web version were accessible between 12:55 to 20:53 local time on 29th July 2018.

Twitter and circumvention tool sites

Access to Twitter appears to have been blocked in the evening of Mali’s elections, along with two circumvention tool sites (guardianproject.info and anonymizer.ru).

The following table shows that although these sites were accessible (from AS30985) up until (at least) the early hours of 29th July 2018, their testing presented failures in the evening, suggesting that access to them was blocked.

All of the URLs included in the above table were accessible over the last months based on daily OONI Probe testing. The fact that they suddenly presented network anomalies on the same evening of the same date (29th July 2018) suggests a change in censorship policy and that Orange Mali (AS30985) may have (temporarily) blocked access to them. Similarly to the testing of WhatsApp, measurements collected from SOTELMA (AS21271) on 29th July 2018 show that Twitter was accessible.

Circumvention tools are usually blocked around the world in an attempt to make censorship circumvention harder. The blocking of the Guardian Project (which develops various censorship circumvention tools) and anonymizer.ru could potentially be an attempt to reinforce the blocking of social media in Mali.

The blocking of certain social media comes at a highly tense political period. The west African country has been in turmoil since 2012, when ethnic and Islamist insurgents seized the northern region of Mali. According to UN figures, more than 300 civilians have died in ethnic clashes this year alone.

Leading up to the latest presidential election, armed groups issued threats of violence, warning that they would try to disrupt the elections, particularly in the country’s northern region. While no major incidents were reported in the south and the voting “went smoothly” in the capital, several violent incidents were reported in northern and central Mali, limiting voter turnout.

In addition, a controversy on the existence of a parallel voter list emerged a few weeks before the election, casting doubts in the opposition on whether the election would be free and fair. If no candidate gains more than 50% of the vote, a second round of elections will take place on 12th August 2018.

As part of this study, we used free and open source software, open methodologies, and open data. Our research methodologies can therefore be reproduced and this study can be expanded upon as political events unfold.