Sri Lanka’s March 2018
Social Media Block, analyzed with data

by Yudhanjaya Wijeratne | March 19, 2018

Image: an article by Tisaranee Gunasekara in the Sunday Observer. Photo by Sanjana Hattotuwa.

Quantitative analysis: the impact to online discussion

A couple of days ago, Ray Serrato, Senior Programme Officer at DRI, had a look at the impact of the Sri Lankan social media ban. He looked at a Facebook group called තුලාව (justice), a known discussion space that mixes Sinhala-Buddhist racism with high support for the Rajapakse regime.

Anecdotally, we knew that the block wasn’t working, but Ray’s single group analysis put a number on the effectiveness.

But because one group cannot be taken as representative of all of Sri Lankan Facebook, I extended his analysis a little: I looked at a spectrum of groups, ranging from newsgroups to known racist discussion forums to tourism groups to ecommerce groups.

My analysis looked at 63,842 posts collected from 15 groups (names listed below) boasting a combined total of 781,349 users. Of course, these groups may share the same users — there is no way to establish a clear margin without violating people’s privacy. These posts were generated by 10,715 individual accounts — 1.37% of the accounts have the ear of the rest of them.

The data was collected by looking at the number of posts each group had over the past 30 days, then doubling that figure. The y-axis is the number of posts, and the x-axis is the date. The chart is highly reliable from February 2018 onwards to now (mid-February looks like an exceptionally busy month), and loses accuracy the further back in time we go.

Let’s examine the period around the block:

My data confirms Ray’s pattern perfectly. The social media block took place on the 6th of March. The riots and arrests went on until the 8th. None of this took people offline — it dropped the network to about 60% activity — the same level seen in non-block times in February 2018 and in January 2018.

This correlates perfectly with the rise in searches for “VPN”: the moment the block happened, people simply found ways around it. People were online all through the riots.

The dip in March 10–11 is interesting: people seem to have been offline. Newspapers from the date show the Prime Minister of Sri Lanka visiting houses in Kandy and pledging to get the situation under control. Bangladesh chased down Sri Lanka’s 214 in the Nidahas Trophy T20 in Colombo. Rather than blocked, people seem to have been merely…distracted.

What were people talking about? For that, we turn to Twitter. An archive of #lka, the common Sri Lanka discussion thread, gave 30,574 conversations from the 7th to the 15th of March. All top topics — anything with over 500 occurrences — revolved around the riots.

It’s a testament to how small Twitter in in Sri Lanka that this entire discussion spanned only 4676 users.

Impact to businesses

However, this VPNing around may have caused severe damage to businesses that rely on Facebook for advertising.

Warnings put out by embassies regarding Sri Lanka will of course impact tourism for a while, but digital marketing was interrupted when Sri Lankan users suddenly appeared from different countries via VPNs.

Consider this image from the start of last week:

The figure above shows the user engagement figures for 5 diverse businesses with large (over 100,000) likepools. Notice the extremely low engagement figures. The figure below, in contrast, shows these same pages after the block was lifted. GFlock, a clothing company that relies heavily on Facebook ads, shows four times as much engagement. Glitteray shows an improvement of 10x.

There is not enough data here to make a broad statement, but discussions among the startup community yielded figures anywhere of between 30 and 70% drops in revenue. These figures varied widely depending on the type of business and the use case, of course. There is a case to be made here that the one-week ban, so callously imposed, may have caused severe issues for smaller companies that rely on social media advertising for their revenue.

Key findings:

  1. Contrary to government expectation, the ban on social media merely brought Facebook activity to January-February levels (on average, a roughly 50% drop from the three days before the riots)
  2. Twitter was not blocked at all, and conversation continued unimpeded
  3. The government, if anything, shot itself in the foot by slowing the flow of news, impacting businesses and enraging the user population

Qualitative analysis: everything else

Note: I will be making liberal use of tweets from Nalaka Gunawardene for this part of the document, as in my archive of 30,000 Nalaka is a reliable and consistent chronicler these events.

The common narrative, especially that favored by the President’s Office, seems to be that blocking Facebook promptly made the great Valley giant sit up, promise to take hate speech off their platform, and bow down to the government’s benevolence.

Reality is not so simple. Anecdotes filtering through from people involved note that when the Digana and Teldeniya narratives started making the rounds on social media, Internet citizens promptly began a campaign of reporting racist pages and people. The Mahason Balakaya’s Facebook presence, for example, has been wiped out, and in fact a Facebook video of the (apparent) head of the Balakaya discussing riot logistics may have served to convince many of the need to urge the government to take action.

It should be noted that Facebook has, in this country, been used very effectively to coordinate citizen response (see Said response could be for good or ill, but given that the block only took place after the riots were in full swing, it seems to be a wasted opportunity for citizen response.

Social media was not the only thing impacted. Random websites were blocked, including my own (ironically, the website I used to criticise the governments for the last ten years runs perfectly: they had blocked the website I maintain for my science fiction writing).

Faced with a potentially block-happy government and a lack of information, many expressed fears over what the President and Facebook were discussing, especially when faced with statements like this:

Meanwhile, of course, those in power were busy violating their own orders:

In light of this, especially to understand whether some measure of political censoring may be implemented, Nalaka Gunawardene, Sanjana Hattotuwa and I had a call with the Facebook team, including those who flew down here to Sri Lanka. On their end, they stated that they have pledged only to work on racial violence and hate speech as per their published Community Standards (, and have not engaged with the government on anything else that does not meet their Community Standards criteria.

This allowed us to put some fears to rest ( with regard to political surveillance or further blocking done by the government.

Facebook’s own reporting system, however, is severely flawed by a lack of understanding. Case in point: Jeevanee Kariyawasam, probably the most visible activist reporting pages and people during this incident, found her own content taken down, while her own statuses were filling up with comments threatening her life.

The ban is widely regarded as a colossal mistake or a case of chronic misinformation. The clearest articulation of this in English is by Rohan Samarajiva for the New York Times. ( Variants of this message are found scattered through liberal media in Sri Lanka, with state-sponsored newspapers taking the opposite tack:


  1. The government responses were uncoordinated: various Ministers spoke of the ban being lifted at different times, and none went through.

2. By being uncoordinated and by deploying draconian tactics of the sort that one would have expected from the Rajapakse regime, the government has lost a great deal of credibility.

3. There now appears to be an atmosphere of deeper distrust towards the government.

Final conclusion

This block was about as useful as a one-legged man in an ass-kicking competition, and seems to have had a similar effect.

Legal implications

I am not a lawyer, so I contacted separate sources regarding this segment. According to Gehan Gunatillake, a lawyer and doctoral student at the Oxford Law Faculty,

Section 66 of the Telecommunications Act gives the Minister in charge of the subject of telecommunications the power to issue directions to the regulatory authority — which in 1996 became the TRC. So only the Minister can issue directions, not the President (whose portfolios are confined to Defence, Mahaweli, Environment and National Integration & Reconciliation). The TRC cannot operate on the instructions of the president. The Emergency Regulations do not confer such powers to the President either. So the social media block is not in compliance with any law.

The only exception in the Telecommunications Act to the Minister’s Authority is section 5(f) — where the GoSL can give instructions to the TRC in the interest of national security and public order. A direction from the GoSL would require a cabinet decision, and cannot be a direction from the president following the 19A. In any event, we have to go back to the emergency regulations of 6 March 2018 — which do not confer powers to the President on regulating telecommunications.”

However, the official story indicates that there is some channel of power that the President can use here. Consider this statement from the state-sponsored Sunday Observer:

….the intelligence agency of the government, the State Intelligence Service, requested the Defence Ministry to consider a blockade on social media to control hate speech and rumours that was fast spreading. Thereafter, the Secretary to the Ministry of Defence, Kapila Waidyaratne recommended to the Director General of Telecommunications Regulatory Commission of Sri Lanka (Austin Fernando) to consider such a blockade….

Austin Fernando is the current Secretary to the President of Sri Lanka. The Defense Minister is one of the President’s advisors. Obviously there is some structure here, whether fully legal or in a gray area, that allows the President to channel such powers.

Dr Deepika Udagama, the Chairperson of the Human Rights Commission of Sri Lanka, in a letter sent to the Chairman of the Telecommunication Regulatory Commission (TRCSL) Chairman Austin Fernando, stated that the HRCSL had received several complaints from the public regarding “the continued restrictions on social media even after the violence in the Kandy District was under control”.

“The Commission recognizes the critical necessity to protect freedom of expression and the right to information as guaranteed by the Constitution of Sri Lanka and Sri Lanka’s international human rights obligations,” the letter said.

“ln doing so, we recognize the need to strike the necessary balance between those rights and maintenance of public order and the protection of the rights of all,” it said.

According to the Colombo Gazette, “the HRCSL had reiterated the urgent need to take legal action against those who were using social media to propagate communal hatred and incite sectarian violence, under applicable laws, in particular under the ICCPR Act №56 of 2007”.

Anyone wishing to understand the legalities of this issue should probably start at the act quoted by the HRCSL and work up from there. Many jury-rigged spheres of influence seem to be at play here.



Data scientist, public policy and tech, @LIRNEasia. Nebula Award nominated author. Numbercaste (2017) / the Inhuman Race (2018). @yudhanjaya on Twitter.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Yudhanjaya Wijeratne

Data scientist, public policy and tech, @LIRNEasia. Nebula Award nominated author. Numbercaste (2017) / the Inhuman Race (2018). @yudhanjaya on Twitter.