Burmese Online Information Landscape Report No. 1

A Pilot Study

October - December 2022 

Executive Summary

As part of an ongoing project aiming to promote tolerance and social cohesion in Myanmar, Digital Public Square has conducted a pilot study on the online information landscape in the country. Through a mixed-methods approach, this first report tracks and analyzes prevalent pro-regime mis and disinformation narratives and provides a high level overview of important topics and events in the online space, during the observed reporting period (1 October 2022 - 31 December 2022). This is the first report produced as a result of this pilot.

Table of Contents

About this Report 

Narratives Catalog

Methodology

Findings & Analysis

Codified Content of Pro-Regime Telegram Channels

Methodology

Findings & Analysis

Targeted Stakeholder Analysis

Methodology

Findings & Analysis

Key Takeaways

About this Report

As in almost every society today, social media plays an important role in Myanmar’s political culture. Even prior to the military coup in 2021, the use of printed media in Myanmar had already been on decline, as more and more people adopted online media to consume information and news over the traditional media outlets. Unsurprisingly, this trend accelerated in the aftermath of the coup, as printed editions of independent media - which were already struggling to survive - were halted by the military coup. Other forms of physical space for public discourse and free expression, such as public meetings, panel discussions, demonstrations, and even political discussions at tea shops, were sabotaged by brutal repression campaigns by the military, including arbitrary arrests and the confiscation of property.

With independent media under threat and in-person gatherings becoming increasingly dangerous, online spaces have become, for many, the primary place to transact information. The exponential uptake of social media in parallel to these trends has led to increasingly decentralized information production, with citizen journalism playing an important role in documenting what’s happening in the country. However, in this new media environment, it can be challenging to tell fact from fiction, placing an even heavier burden on independent media and civil society to verify facts and counter malicious information online. Following the coup, media channels on Telegram became particularly popular. As public engagement in these spaces grew, they quickly became  an epicenter of the military's misinformation campaigns.

To better understand the contours of the Burmese information landscape, DPS launched a pilot study, the initial findings of which are included in this inaugural report. The pilot study will aim to provide a longitudinal review of the shifting information landscape in the country through a series of reports. The report’s mixed-methods approach consists of three research tracks: i) cataloging refuted mis/disinformation narratives online, ii) codifying of content in pro-military Telegram channels, and iii) targeted analysis of key stakeholder accounts via CrowdTangle. The combination of these methodologies offers diverse insight into important trends and issues related to Myanmar’s online information landscape.

Who we are

Digital Public Square is a Canadian not-for-profit organization that aims to serve communities-in-need with good technology. Our work has grown to reach millions of people worldwide across the Americas, Europe, the Middle East, Africa, and Asia. Digital Public Square has been engaged in Myanmar since November 2019, on a program that seeks to increase tolerance and inclusion for religious and ethnic minorities. We work closely with multiple local partners and experts in the country, adapting to the dramatically dynamic environment.

Narratives Catalog 

The first research track explores the narratives underlying the military’s mis- and dis-information campaigns, primarily on Telegram. The focus of this analysis on Telegram channels is a reflection of recent events and trends in the online information space. While Facebook continues to be the most popular social media platform in the country, many people have adopted Telegram as both a social networking platform and instant messaging services following the military coup. This rapid adoption was a product of the military’s decision to censor Facebook and the growing perception that Messenger was no longer secure in the face of greater internet surveillance.

Other functions beyond instant messaging service were explored and utilized quickly. Channels were first used to create open source scout platforms to inform those who are concerned about the security force’s movements and sizes. Virtual neighborhoods were created only on the basis of townships and shared security related updates in the groups and channels. Subsequently, the military also started using it to compensate for its online communities being banned on Facebook as part of its anti-hate speech and misinformation efforts.

Now, three years into the military’s control in Myanmar, the military’s or its affiliated channels on Telegram enjoy substantial follower counts ranging from several thousands to forty thousands. Rather than tracking, mapping channels, groups and accounts, this stream of research deals with exploring generalizable trends and outliers in the contents of the mis- and dis-information that we captured and cataloged. The inquiry, consequently, does not include the extent to which the mis- and dis-information efforts from different sources are being coordinated.

Methodology

This research track uses quantitative data to assess the nature and frequency of false narratives being circulated online. To do so, we compiled a list of 10 online accounts (see Appendix) that consistently identify and counter false information in the Burmese online environment. We then reviewed, cataloged and categorized the instances of false information that have been debunked via these selected research accounts. This approach relies on the reliable research conducted by fact-checker sites and organizations already operating in this space, and aims to provide a quantitative lens to these efforts. In this period, we cataloged 94 instances of false information circulated in the October-December 2022 period.

Findings & Analysis

Targets of Disinformation

Findings from the observed period demonstrate the efforts of the military and pro-military actors to undermine key individuals, entities and movements. The National Unity Government, People’s Defence Force, Committee Representing Pyidaungsu Hluttaw, and National Unity Consultative Council were the groups most targeted by the cataloged mis- and disinformation instances in the observed period. In our analysis, we decided to group these bodies together since they are all in the political leadership of the mainstream resistance movement. Approximately 56 per cent of the cataloged narratives targeted the NUG/PDF/CRPH/NUCC, either alone or in combination with other groups such as ethnic resistance organizations, Rohingya and the civil disobedience movement.

The second most frequent target was ethnic resistance organizations (EROs) with 31 instances targeted against them. In this study, the term “ethnic resistance organization” is an umbrella term encompassing multiple groups which are entirely independent from each other, though sometimes coordinated. As shown in the table, the NLD is the target of only 3 instances of disinformation collected through this methodology. This reflects the military’s perceived relevance of the NLD in the ongoing political crisis in Myanmar. It is important to note that the Rohingya were specifically targeted twice. Mis and disinformation targeting minorities is important to monitor, as the military may seek to exploit existing religious tensions to divide the public.

Actors and Actions 

Our methodology also aims to assess the intent and impact of mis and disinformation in Myanmar’s online space by cataloging the actors and the actions featured in the observed instances. Figure 2 provides a breakdown of the frequency of certain disinformation narratives observed in this study. Figure 2 only illustrates narratives for which there were two or more supporting instances cataloged, but the outlier section further below summarizes some of the interesting findings from the one-off narratives. When coding actors and actions, we coded the NUG, CRPH, PDF and NUCC as separate entities, while EROs were grouped together.

Significant Trends

According to registered frequencies, narratives most commonly portray certain entities as terrorists or public enemies. People’s Defense Forces and EROs are most frequently portrayed as actively perpetrating attacks against the Burmese population. In fact, over 44 per cent of cataloged instances claimed that the general Burmese population was being persecuted and attacked by various non-military actors, including the NUG, the PDF, EROs and the Rohingya. This, combined with the designation of the NUG and PDF as terrorists, suggests that the military is attempting to sow division by claiming that the NUG and the PDF are not on the side of the Burmese population. Another small but observable trend is the association between PDF and various EROs. The small proportion (6 instances out of 94) can be explained by the fact that the association of PDFs and EROs is already recognized by the general public. The portrayed association seems only to point out that these entities are “partners in crime” that are conspiring against the Burmese people. In our dataset,the association almost always comes together with claims that the PDF is perpetrating attacks against the public. The proliferation of this type of defaming disinformation is observed in context with the surge of the news reports on clashes between military’s forces and PDF and EROs in the last year (see the Targeted Stakeholders Analysis section below for more information).

Outliers 

While the most prominent theme among observed narratives is the discrediting of the PDF and EROs, our dataset also includes a number of noteworthy outliers. .  First, our dataset points to a recurring  narrative that the military is winning against the EROs and NUG, which accounts for approximately 9 per cent of the cataloged instances. These instances often refer to victories on the battlefield or with respect to gaining international support, and aim to convey that the military enjoys an upper hand over the entities that oppose it. This is reinforced through the less frequent but similar narrative that suggests that EROs are in trouble and struggling in the battle against the military forces.

Another noteworthy narrative is that PDFs are specifically attacking and targeting Buddhism. This narrative has been depicted through instances of disinformation which claim that PDFs assassinate influential monks and shell Buddhist sites with heavy artillery. The exploitation of religion - and of longstanding religious tensions within the country - to undermine the legitimacy of opposition groups has long been a signature tactic of the military. We shall continue to monitor these types of narratives in the upcoming reporting periods.  Lastly, narratives that seek to sow division are also common in the dataset, i.e. PDF perpetrating attacks against EROs and the NLD. These instances of disinformation seek to exploit the historical fragmentation of the military’s opposition along various faultlines (such as political, religious, and ethnic) and aim to sow division between the NUG, the PDFs, and the EROs to ultimately weaken the military’s main opposition.

Tracking these patterns in the cataloged mis/disinformation narratives provides valuable insight into the prominence of false narratives being circulated online, notably  that the PDFs, NUG and EROs are leading attacks against Burmese citizens and endangering the lives and livelihoods of the people. It is also interesting to note that some of the relatively lower frequency narratives analyzed in the outlier section above are similar to disinformation campaigns used in Myanmar’s recent history, including the Rohingya genocide and earlier communal violences dating back to 2012. It appears that some of these narratives have been recycled in the aftermath of the coup in 2021 and adapted to the new political landscape.

Figure 4: Frequency of narratives cataloged from research accounts | October - December 2022

Codified Content of Pro-Regime Telegram Channels

Methodology

In parallel to the narrative catalog methodology described above, we also sampled, categorized and analyzed content from public pro-military Telegram channels. This research offers a qualitative analysis of the type of hate speech, misinformation and propaganda narratives being circulated on the pro-regime channels. For this pilot study, our team focused on six public pro-military Telegram channels, each of which had over 10,000 followers.

The study codified a random sample of 500 posts per channel, each of which were published between 1 October 2022 and 31 December 2022. Descriptive characteristics of each channel are presented below.

Figure 5: Number of followers, by Telegram channel

Figure 6: Average number of posts published every day, by Telegram channel

Figure 7: Proportion of cataloged posts containing false information, by Telegram channel

Figure 8: Proportion of cataloged posts containing negative sentiment, by Telegram channel

Findings & Analysis

Of the 3,000 total posts cataloged across the 6 Telegram channels, 17 percent of posts perpetuated mis or disinformation. Three prevalent types of mis and disinformation were identified in this study:

1. Information that can be proven as false
2. Information on real and confirmed incidents mixed with unverified or unsourced accusations
3. Information on unverified incidents

Under the first classification, the analysis revealed a pattern of misappropriation and manipulation of media content (photos & videos) to support false pro-military narratives. These tactics were used to perpetuate narratives that the NUG, the PDF and EROs are perpetrating attacks against the Burmese people and that the military is waging and winning a just war against the opposition. An example of this practice is highlighted in Figures 9 In Figure 9, the administrator of Channel C shared images of PDF walking through fields in Kalay, with a caption that suggests the PDF have seized the crops from the locals through the use of force.

However, the original source of the images is actually from a PDF Kalay Battalion, and the images themselves were taken while the troops were helping the locals with the harvest.

Figure 10 is another sample from our dataset that shows a post made by the administrator of Channel F, claiming that the featured photos were evidence that the NUG was trafficking young women and sending them abroad. These images were actually from an article published by Radio Free Asia in December about a person who was trafficked in Dubai.

While the NUG and the PDF were often the targets of the false information circulated in the Telegram channels, the study found that mainstream media such as BBC, RFA, Mizzia, Irrawaddy, and Myanmar Now were targeted the most. A recurring observed pattern in the sample was channel administrators sharing screenshots of news articles from these media outlets with sarcastic or mocking commentary, with the overall goal of decreasing the trust in the accuracy and reliability in mainstream media. An example of this tactic is featured in Figure 11. This is also an example of the second type of mis and disinformation observed in the study, in which the news or incident being discussed can be factually verified, but the tone and added commentary implies additional details and/or accusations that cannot be verified.

The last type of content identified in the study – information that is difficult or impossible to verify – often references murders and robberies allegedly committed by members of the opposition, as well as military seizures of illegal firearms. These types of claims are incredibly difficult to verify, since these incidents are only reported in military press releases and pro-military media channels.  

Of the 3,000 total posts cataloged across the 6 Telegram channels, 64 percent of posts expressed negative sentiment towards specific groups and actors. Of the 1,929 posts that expressed negative sentiment towards specific groups and actors, 27 percent were targeted towards the People’s Defence Force, 10 percent towards the NUG, and 16 percent towards EROs.

People’s Defense Force. In the studied pro-military Telegram channels, the PDF are regularly characterized as terrorists, alongside the NUG and EROs such as the Kachin Independence Army. The PDF are often accused of carrying out terrorist and illicit activities, including bombings, burning down villages and extorting bribes from the public, and that they are negatively impacting the daily lives and security of the Burmese population. Furthermore, the PDF are often mocked for and characterized as incompetent, illiterate, corrupt, and thieves.

National Unity Government. In the studied pro-military Telegram channels, the NUG are accused of supporting and affiliating with terrorist groups. The legitimacy of the NUG is challenged in these channels and if considered to be an “illegal government.” NUG and CRPH leaders were often referred to as fugitives, and NUG members are accused of being thieves, which further underscores the narrative that they are part of illegitimate and illegal political activities. Furthermore, there were some narratives that suggested the NUG was planning on eliminating important leaders in the PDF, such as General Bo Naga from the Pale Township based in Sagaing. These types of narratives suggest that there is an attempt to sow division and confusion between the NUG and the PDF, likely for the purpose of weakening both parties.

Ethnic Resistance Organizations. In the studied pro-military Telegram channels, the Arakan Army (AA) the Kachin Independence Army (KIA) and the Karen National Union (KNU) receive the most defamation and are most often accused of being terrorist groups. In Sagaing region, the KIA are accused of providing military training and firearms to the PDF, and are ultimately blamed for the fires and destruction in the region. These narratives aim to characterize EROs as being against and ultimately waging war on the Burmese population, particularly the Bamar majority.

Civil Disobedience Movement. In the studied pro-military Telegram channels, doxxing was a common practice used against CDM supporters. The names, social media accounts and addresses of pro-democratic activists were often shared in these channels, and the administrators urged the channel followers to harass and/or arrest them. Of particular note was that posts promoting doxxing largely targeted women in the pro-democratic movement. Furthermore, well-known pro-democracy women who donate money to the revolution and lead fundraising initiatives have been accused of supporting terrorism by providing resources to the PDF.

Targeted Stakeholder Analysis

Targeted Stakeholder Analysis is a monthly review of content posted by key media stakeholders. In contrast to the Narrative Catalog and Codified Content sections above that focus on pro-regime content, this part of the study focuses on information that is published by a wide range of actors including media outlets, ethnic media, key political groups, and civil society organizations (CSOs). The purpose of this activity is to provide a view into prevailing trends, topics, and issues in the Burmese online landscape - with a particular focus on discourse related to the coup and resistance movements - using Facebook.

Methodology

The Facebook pages of relevant stakeholders have been categorized into 3 groups: Myanmar Media (30 pages), Ethnic Media (33 pages), and Key Political Groups and CSOs (28 pages). The full list of pages can be found in the Appendix.

These lists were carefully curated in CrowdTangle to capture diverse opinions and priorities from different stakeholders in Myanmar, and to ultimately explore the most popular posts from the different stakeholder groups. For the last quarter of 2022, we cataloged 195 posts from three different stakeholder groups in total. For November and December, we cataloged the top 20 overperforming posts, as indicated by CrowdTangle, for each of the three stakeholder groups, for that particular month.^[1] This analysis provides quantitative insights into online discourse on Facebook in Myanmar, with the goal of identifying changes and trends in Myanmar’s digital information landscape. Among a number of data points available on Crowdtangle, we chose to use ‘Overperforming’ as our criteria to rank the posts. According to Crowdtangle’s definition, overperformance is calculated by the multiplier of the difference between the performance of a new post and the mean performance of the last 100 posts with the same post type i.e. a photo post or a video post. Unlike ‘Total Interaction’ which ranks posts according to their absolute number of interactions, thereby disproportionately representing the most popular pages, ‘Overperforming’ ensures the inclusion of less popular pages by numerous stakeholders. In this study, 195 overperforming posts were cataloged and reviewed.

Findings & Analysis

In this stream of research, the content topic categories are designed to ensure the coverage of relevant themes and political developments since the coup. Figure 13 below illustrates the top three overperforming topics for each stakeholder group across the months of October, November and December 2022.

Figure 13: Top Content Categories from Targeted Stakeholder Analysis | October - December 2022

Generally, content related to the ongoing violence in Myanmar received the most engagement  in the last quarter across all three stakeholder groups. The junta violence and atrocities, the armed struggle by the resistance forces and their encounters are different categories of  violence-related content. Across the three groups and in all three months, content relating to the armed struggle made the most appearances in our study (36 posts), followed by junta atrocities related contents (34 posts). References to the Civil Disobedience Movement were found in only 5 posts. Only one post which is related to the federal constitution drafting process has been cataloged out of 195 posts in total showing relatively low public interest and reporting on the matter. Three posts related to women and gender based violence (GBV), two of which being efforts by Myanmar CSOs to raise awareness and fight against GBV. Only one cataloged post related to the Rohingya, reporting the deaths of 13 Rohingya people in the Yangon region.

We also included miscellaneous content under the designations of ‘other’ and ‘other coup related’ categories. As the names suggest the former category, ‘other’ includes only content which is not relevant to the coup. This content includes, among others, religious ceremonies, general hardships of people, and feature video stories about a particular topic. We observed a significant increase in this type of  content from October to December of 2022. In October, there were only 7 examples of ‘other’ content in our sample across all three stakeholder groups. In November and December, it increased to 21 posts and 19 posts respectively, suggesting that the popularity of certain topics began to diversify. This is likely due to various external factors and events, such as the World Cup for example. While the time span covered by this report is not long enough to make generalizations, this trend of information diversification should be carefully tracked in the following quarters.

When comparing the top topics across the stakeholder groups within the same months, we note that Key Political Groups received high levels of engagement on posts related to protests and posts with specific calls to action. While this is likely not surprising given the political nature of the groups within this stakeholder cluster, it will be interesting to track whether these particular categories remain top performers in the reporting periods to come. Furthermore, at this stage, there seems to be high levels of similarity between the top topics from Burmese media and Ethnic Media groups. It will be interesting to see whether this changes over the course of the upcoming reporting periods and if any significant trends or divergences appear between these two stakeholder groups.

In addition to categorizing the posts according to their thematic content, we also categorized ‘relevant actors’ to designate which key political actors were mentioned explicitly or had strong relevance to the post. (Figure 14)

Figure 14: Frequency of key actors mentioned during Targeted Stakeholder Analysis, October - December 2022

When it comes to related actors, the junta appears to be the largest newsmaker as nearly one-third (61) of posts had direct or indirect reference to this actor. This is followed by various protest groups, which were featured in 26 posts. International actors, such as the United Nations, the United States, the European Union, ASEAN, China and Russia, were relevant to 25 of the cataloged overperforming posts, representing nearly 13 per cent of the sample. All but 2 posts featuring international actors were related to the topic of the coup.

Approximately 7 per cent (15) of posts related to Burmese Civil Society Organizations (CSOs) for the 15 posts as a related actor. These posts often included calls for action and featured reports relating to CSO’s lines of work. Approximately 14 of these posts were related to the coup.  

Key Takeaways

This is the first report in a pilot study that will aim to track trends and patterns in the Burmese online information landscape. The mixed-methods approach employed during this pilot study provides some preliminary insights into narratives and tactics being employed in Myanmar’s digital environment. Looking across the data collected as part of this pilot study, we have identified three key takeaways that organizations engaged in the Burmese information environment or civil society should consider as they try to engage communities around reliable information.

1. First, the themes highlighted in the Narrative Catalog shed light on the military’s tactics for weaponizing information with the goal of destabilizing and dividing Burmese society. In particular, this activity analyzes how specific instances of mis and disinformation work together to exploit pre-existing grievances or historical cleavages between groups. One of the most important findings from our Narratives Catalog is the identification of mis and disinformation that is currently being circulated with the aim of sowing division among and/or weakening the democratic opposition and public support for democratic institutions. Understanding these overarching narratives, and how they change over time, can provide independent media and civil society with a view of how to build resilience against future mis and disinformation. These resilience-building efforts must go beyond fact-checking and myth-busting to include proactive messaging and digital media literacy campaigns that seek to mitigate against the harmful effects of mis and disinformation narratives before they are produced and disseminated by malign actors.  
2. Second, codifying pro-regime content that circulates on Telegram provides more specific and immediate insights to inform counter-mis and disinformation efforts that are currently being led by civil society and independent media. This particular type of research is becoming increasingly important due to the size of the following that pro-regime channels have amassed, and the increasing efficacy with which the military and its supporters are weaponizing the platform against the pro-democracy movement.
3. Finally, our Stakeholder Analysis aims to provide a high-level overview of which coup-related topics are currently receiving the most attention. The findings presented in this pilot study illustrate interesting differences across the three key stakeholder groups and the topics that generate engagement for each group. In this initial study, the sample size of data collected is relatively small, however tracking trends in the type of information produced by a variety of key stakeholders in the Burmese online environment over time is important. Understanding the volume of engagement with information about coup-related events and political themes in the Burmese environment can shed light on public sentiment. For example, it will be interesting to track the proportion of posts cataloged in each stakeholder's top 20 overperforming list that do not pertain to coup-related news (and therefore classified as “other”). While some actors naturally post less frequently on coup-related topics than others, this trend may shed light on how the supply of information in the country changes over time, either in response to changes in consumer demand or as a result of constraints on freedom of speech.

DPS will continue to employ and refine this approach to track trends in the Burmese online information environment, including instances of mis and disinformation. The findings from our ongoing analysis will be presented in upcoming reports with the goal of identifying trends over time.

Appendix

Research Information Accounts

Stakeholder Groups