The EU election observation mission has issued its preliminary statement on Saturday’s Presidential election in Sri Lanka. Overall the mission found the election to be well run, but says that there was an uneven playing field for the candidates resulting from the absence of a finance law and the bias of both state and privately owned media. But there are highly damaging findings on the use of online and social media. I’ve re-produced the relevant section and footnotes below:

“A coordinated distortion of the information environment online undermined voters to form opinions free from manipulative interference.

Thirty-four per cent of Sri Lankans have access to the internet and use smartphones to send and receive information. The digital literacy rate is low, leaving the online discourse prone to manipulation. (65) Constitutionally guaranteed freedom of expression is not explicitly extended to online content (66), and in the absence of all-encompassing privacy and data protection legislation, parties do not declare their use of voters’ personal data, which is collected by mobile applications or campaign staff. (67) Such practice is at odds with international standards. (68)

Among the social media platforms, Facebook is the prime contributor to the crafting of political narratives in the public space and to setting the electoral agenda. (69) The EC had only an informal understanding with Facebook on the removal of hate speech and disinformation. (70) Citizen observers also reported harmful content online to the EC and Facebook. (71) However, Facebook’s reluctance to take action, coupled to high levels of anonymous, sponsored content, enabled a mushrooming of hateful commentary and trumped-up stories that capitalised on long-standing ethnic, religious and sectarian tensions. (72) It continued also during the campaign silence, when Facebook removed only a small proportion of such paid content. (73) This was detrimental to the election and at odds with inter- national standards. (74)

Coordinated dissemination of outright false and/or demeaning information presented in various for- mats and across digital platforms dwarfed credible news threads. Suppression of credible news en- tailed the use of sponsored content on Facebook and coordinated sharing of political memes that sow discord and political gossip, both of which served as a source for multiple posts on political support group pages. In the majority of cases, the SLPP campaign benefitted from this. One such campaign undermined the integrity of postal voting; (75) four narratives capitalised on underlying fears and/or recycled previously debunked information. (76)

The use of algorithms and human curation to mislead the debate on Twitter was observed. (77) It in- cluded a high number of recently registered accounts amplifying certain political messages that also appeared on political support group pages. Ten days before the election, the SLPP further skewed the online discourse by announcing a “sharing” contest for 50,000 subscribers of the SLPP app VCAN. (78) A few professional fact-checking organisations exposed deceptive and false stories, but their staff levels and reach are far smaller than those of political actors, and they were not supported by broadcast or print outlets. (79) Overall, a damaging online environment distorted public debate and curbed voters’ access to factual information on political choices, an important element for making a fully informed choice.”

Footnotes:

65 – Computer and digital literacy are 27.5 and 40.3 per cent respectively (2018). Department of Census and Statistics Sri Lanka; joint declaration on freedom of expression and “fake news”, disinformation and propaganda by recognised international bodies, section 3, says: “States should take measures to promote media and digital literacy”.

66 – UNHRC, Promotion, Protection and Enjoyment of Human Rights on the Internet, 4 July 2018:“[…] the same rights that people have offline must also be protected online, in particular freedom of expression…”.

67 – The VCAN app promoted by the SLPP (Google Play) demands registration with a National Identity Card (NIC); if permitted, it has access to a phone’s geolocation, can read the content of USB storage and view network connections. The privacy and developers’ pages are empty. The use of the VCAN app and enclosed private information by campaigners was confirmed to EU observers in Anuradhapura, Batticaloa, Colombo, Gampaha, Kandy, Kalutara, Kurunegala and Trincomalee.

68 – ICCPR art. 17, “No one shall be subjected to arbitrary […] interference with his privacy”. ICCPR, HRC GC 16 par. 10: “The gathering and holding of personal information […] on computers, data banks and other devices, whether by public authorities or private individuals or bodies, must be regulated by law.”

69 – As of October 2019, there are 6.5 million Facebook users, 1.1 million Instagram, 261,700 YouTube, and 182,500 Twitter users in Sri Lanka. Several million people use WhatsApp. Eight million communicate on Viber.

70 – Facebook has not published data about requests by, or its response to, government authorities in 2019. It notes that without a written agreement with the EC, the timeline for decisions remains at its discretion. The number of removed posts is not public. FB did not consider the EC’s media guidelines applicable to it.

71 – By 6 November the EC complaints centre has registered 102 complaints about campaign violations and hate speech online. 

72 – The EU EOM analysed 340 Facebook pages supporting one of the two leading candidates, including 167 prominent in one district. Twenty-six of national-level and 10 per cent of local-level pages featured sharply negative content. On 25 of the most popular meme pages, the EU EOM identified 47 political memes with a menace, including sectarian undertones. The EU EOM assessed public videos by ten of the most subscribed to YouTube influencers (more than 20,000 followers each) and identified at least 10 cases of sharply divisive rhetoric, including two with a racist message.

73 – On 14 and 15 November the EU EOM identified at least 300 sponsored posts/adverts with campaign content. CSOs identified 700 such sponsored posts, FB removed less than half of it.

74 – ICCPR, HRC GC 25, par. 19: “Voters should be able to form opinions independently, free of violence or threat of violence, compulsion, inducement or manipulative interference of any kind.” See also the joint declaration on freedom of expression and “fake news”, disinformation and propaganda, sec. 4 ‘Intermediaries’.

75 – On 31 October an anonymous fan page, Iraj Production, posted “postal voting” results featuring the SLPP’s victory with 95 per cent of votes cast. The post cited the EC. It was shared and re-shared 30,000 times. The post was also shared by a FB page promoting the SLPP and serving as “a mother page” for sharply negative and manipulative content against the UNP. FB closed the fan page on 6 November. The “landslide victory of the SLPP in postal voting” was repeated in rallies.

76 – These include the UNP candidate pictured with a Muslim doctor falsely accused of sterilising 4,000 Buddhist women (de- bunked in July 2019), a claim that the SLPP candidate is supported by a Muslim politician, who, in turn, is falsely associated with “masterminds of the Easter bombings” (debunked in July 2019). The posts and sponsored content also capitalised on anti-foreign sentiments by falsely stating that the government was giving away 18 per cent of Sri Lanka’s land by signing an agreement with US.

77 – From 4 October to 4 November the EU EOM downloaded all tweets trending the neutral electoral hashtags #PresPollSL and #prespolls2019. Out of 2,000 accounts 500 were randomly selected for assessment. Nine per cent were established less than four months before the election, 4.62 per cent were deleted by Twitter, 28 per cent shared only negative content, and 35 per cent were only re-tweeted.

78 – By following the application, the EU EOM observed automated and concerted activities, including multiple sharing of the same post at odd hours (3:30am, 4:22am) but from different profiles which could indicate the activity of bogus accounts. 

79 – Three fact-checking projects employ a robust news/photo verification methodology. They have debunked 74 false sto- ries/statements/images by and about political figures. These included a claim of a politically motivated fight that followed a Premadasa rally, the arson of houses belonging to Rajapaksa supporters, and a letter from a cardinal opposing a big govern- ment agreement with US (the Millennium Challenge Corporation, MCC). The debunking of false postal voting results was shared just 17 times as opposed to the 30,000 shares of the false news post itself. The EU EOM also identified an organisation that was impersonating a fact-checking organisation. It was featured in the mainstream media as a guardian of fairness in online campaigning, while in practice it was striving to discredit non-partisan fact-checkers.