Meta in Myanmar, Part I: The Setup

“Technology is like a bomb in Myanmar.” —Kyaw Kyaw, frontman of Burmese punk band Rebel Riot ¹

Back in early July, I started working on a quick series of posts about online structures of refuge and exposure. In a draft of what I meant to be the second post in the series, I tried to write a tight two or three paragraphs about the role Meta played in the genocide of the Rohingya people in Myanmar, and why it made me dubious about Threads. Over time, those two or three paragraphs turned into a long summary, then a detailed timeline, then an unholy hybrid of blog post and research paper.

What I learned in the process was so starkly awful that I finally set the whole series aside for a while until I could do a lot more reading and write something more substantial. Nearly three months later, I’m ready to share my notes.

Here’s a necessary personal disclosure: I’ve never trusted Facebook, mostly because I’ve been around tech for a long time and everything I’ve ever learned about the company looked like a red flag. Like, I’m on the record swearing about it.

But once I started to really dig in, what I learned was so much gnarlier and grosser and more devastating than what I’d assumed. The harms Meta passively and actively fueled destroyed or ended hundreds of thousands of lives that might have been yours or mine, but for accidents of birth. I say “hundreds of thousands” because “millions” sounds unbelievable, but by the end of my research I came to believe that the actual number is very, very large.

To make sense of it, I had to try to go back, reset my assumptions, and try build up a detailed, factual understanding of what happened in this one tiny slice of the world’s experience with Meta. The risks and harms in Myanmar—and their connection to Meta’s platform—are meticulously documented. And if you’re willing to spend time in the documents, it’s not that hard to piece together what happened.

I started down this path—in this series, on this site, over this whole year—because I want to help make better technologies and systems in service of a better world. And I think the only way to make better things is to thoroughly understand what’s happened so far. Put another way, I want to base decisions on transparently sourced facts and cautiously reasoned analysis, not on assumptions or vibes—mine or anyone else’s.

What I want to promise you, my imaginary reader, is that I’ve approached this with as much care and precision as I can. I cite a lot of documentation from humanitarian organizations and many well-sourced media reports, and also a bunch of internal Meta documentation. What I’m after is maybe something like a cultural-technical incident report. I hope it helps.

This is the first of four posts in the series. Thank you for reading.

I’ve put notes about terms (Meta vs. Facebook, Myanmar vs. Burma, etc.) and sources and warnings in a separate meta-post. If you’re someone who likes to know about that kind of thing, you might want to pop that post open in a tab. If you spot typos or inaccuracies, I’m reachable at erin at incisive.nu and appreciate all bug reports.

Content warning: I’m marking some sections and some cited sources with individual warnings, but this is a story with a genocide at its heart, so be aware of that going in. I’ve also included a small selection of hateful messages in this post and throughout the series so that it’s clear what we’re talking about when we talk about “hate speech.” Some contain words that, used in specific contexts, constitute slurs. I put some notes about all of that in the meta post as well.

This is a story about tech

Even if you never read any further, know this: Facebook played what the lead investigator on the UN Human Rights Council’s Independent International Fact-Finding Mission on Myanmar (hereafter just “the UN Mission”) called a “determining role” in the bloody emergence of what would become the genocide of the Rohingya people in Myanmar.²

From far away, I think Meta’s role in the Rohingya crisis can feel blurry and debatable—it was content moderation fuckups, right? In a country they weren’t paying much attention to? Unethical and probably negligent, but come on, what tech company isn’t, at some point?

Plus, Meta has popped up in the press and before Congress to admit that they fucked up and have concrete plans to do better. They lost sleep over it, said Adam Mosseri, the person in charge of Facebook’s News Feed, back in March of 2018.

By that point in 2018, Myanmar’s military had murdered thousands of Rohingya people, including babies and children, and beaten, raped, tortured, starved, and imprisoned thousands more. About three-quarters of a million Rohingya had fled Myanmar to live in huge, disease-infested refugee camps in Bangladesh.

And Meta?

By that point, Meta had been receiving detailed and increasingly desperate warnings about Facebook’s role as an accelerant of genocidal propaganda in Myanmar for six years.

---

My big hope for the internet is that we handle the shit we need to handle to make sturdier, less poisoned/poisonous ways to connect and collaborate in the gnarly-looking decades ahead. I think that’s not just possible, but that it’s our responsibility to work toward it. Also, I’m pretty sure that despite the best intentions and the most transparent processes, we risk doing enormous harm if we don’t learn from the past. (Maybe even if we do.)

This series is for anyone who, like me—and despite everything not good about the tech world—has found themselves periodically heartened and sustained by open technology projects, online communities, and ways of being together even when we’re far apart.

And the thing I want you, and all of us, to remember about the sudden flowering of the internet in Myanmar in the 2010s is that in the beginning, it was incredibly welcome and so filled with hope.

Welcome to the internet

After decades of crushing state repression, a more democratic regime came into power in Myanmar in 2011 and gradually relaxed restrictions on the internet, and on speech more generally. Journalist and tech researcher Faine Greenwood (they/them) is working in Southeast Asia at the time, and gets caught up in the spirit of the moment, despite their skepticism about the benefits of the internet:

I’d connected with a Myanmar NGO dedicated to digital inclusion, and through them, I got a chance to meet and interview a number of brilliant and extremely online Burmese people, all of them brimming with long-suppressed, almost giddy, optimism about their country’s technological future.

It was hard for me not to share their enthusiasm, their massive relief at finally getting out from under the jackboot of a military regime that had tried to lock them away from the rest of their world for as long as they could remember. I came away from speaking with them with a warm, happy feeling about how online communication maybe, just maybe, really did have the power to unfuck the world.³

And online communication was coming in fast, as the price of SIM cards, which had been controlled by the ruling junta, dropped from the equivalent of $2,000 USD in 2009, to $250 in 2012, to $1.50 in 2014.⁴

Mobile adoption explodes, from less than one quarter of a percent of the population in 2011 to more than 90% in 2017.⁵ And smartphone use and internet uptake spikes with the mobile revolution instead of dragging along behind—a 2017 Bloomberg article does some flavor reporting to contextualize those numbers:

Thiri Thant Mon, owner of a small investment bank in the city, says she still remembers how magazines from the outside world used to arrive weeks late because censors needed time to comb through them.

“Suddenly because we’re on internet,” she said, “people realize what the rest of the world looks like. Now it’s like everybody on the street is talking about Trump. A few years ago, nobody knew what was happening in the next town.”⁶

In 2015, writer and photographer Craig Mod is working with a team doing internet ethnography in rural Myanmar on behalf of an organization that designs and builds hardware—including modern farming implements—to improve the lives of farmers, who make up the majority of Myanmar’s populace. (Disclosure: Craig and I have halfass-known each other for going on a couple of decades, and I’m a longtime fan of his work.)

Mod’s essay about his work is clearly the product of a sensitive eye and a nerd’s delight in the way new technology twines up with the unevenly distributed realities of daily life:

The village still lacks electricity although they’ve pooled funds and a dozen newly planted metal-power poles dot the fields, waiting to be wired up. Through our interpreter I ask, Where do you charge? Farmer Number Ten points to a car battery hanging in the corner onto which familiar USB wires are spliced.

The tech is all over the place, “cheap but capable,” and Craig wonders if the one-smartphone-per-human result would make Nicholas Negroponte happy. There are no iPhones, no credit cards, no data plans: “Everyone buys top-up from top-up shops, scratches off complex serial numbers printed in a small font, types them with special network codes into their phone dialers in a way that feels steampunkish, like they’re divining data. They feel each megabyte.”⁷

---

This is the point in the conversation when it becomes impossible not to talk about Facebook. Because in Myanmar, even back in 2012 and 2013, being online meant being on Facebook.

I’d read about Facebook’s superdominance in Amnesty International’s reporting, in long stories from Reuters and Wired and the NYT and the Guardian, in books, even in UN reports. But Mod’s notes bring to life both Facebook’s pervasiveness and the way Burmese people actually used it:

We ask about apps. The farmer uses [chat app] Viber and Facebook. He says he chats with a few friends on Facebook but mainly people he doesn’t know. Most of his Facebook friends are strangers. He tells us his brother installed the app for him, and set up his account. He doesn’t know the email address that was used. He gets most of his news from Facebook. The election looms and he loves the political updates.

Farmer Number Ten tells us he used to use radio for news but no more. He says he hasn’t turned the radio on in years. Other news apps—like one called TZ—use too much data. He’s data conscious. He uses Facebook mostly at night when the internet is fastest, and cheapest.

And speaking with the proprietor of a cellphone shop:

Facebook is the most popular app, he says. Nine out of ten people who come into the shop want Facebook. Ten months ago SIM prices dropped, data prices dropped, interest in Facebook jumped. I take note. Only half the people who come into the shop already have a Facebook account. The other half don’t know how to make one. I do that for them, he says. I am the account maker.

And what about other apps? He mentions a news app called TZ. Once popular, now less so. He brushes his hand aside and says it’s too data hungry. Everyone is data sensitive he says and reiterates: Facebook. Nobody needs a special app for their interests. Just search for your interest on Facebook. Facebook is the Internet.⁸

But how did Facebook get to be the internet?

In the early years, there’s a stripped-down version of the app kicking around that used less data than competitors’. But also, when Myanmar’s government opens up to two foreign telecom services, the stronger of the two, Norway’s Telenor, “zero-rates” Facebook.⁹ Essentially, zero-rating is a selective subsidy; it means that customers won’t be charged for the data they use for some parts of the internet, but would be charged for others. For all of Telenor’s customers, using Facebook is free.

I want to briefly flip ahead to something Meta whistleblower Frances Haugen says in a 2021 interview:

Facebook bought the privilege of being the internet for the majority of languages in the world. It subsidized people’s use of its own platform, and said “Hey, you can use anything you want, but you’ll have to pay for that—or you can use our platform for free.” As a result, for the majority of languages in the world, 80 or 90% of all the content in that exists in those languages is on Facebook.¹⁰

That this explosion of connectivity presented dangers as well as freedoms is immediately clear to civil society orgs in Myanmar, for two main reasons: The first is the state of comparative innocence with which the vast majority of Burmese people approach the internet. The second is that the political situation in Myanmar is a powderkeg at best.

There’s a third thing, though most people don’t know it yet, which is that Meta’s decision-making about Myanmar reflects no willingness to adjust for the first two. To get to that, we need to start at the beginning.

The blogger and the monk

A tiny canned history of modern Myanmar might go like this: In a series of East-India-Company–entangled conflicts, the British Empire took control of Myanmar, then called Burma. From the beginning of the wars in 1824 to 1948, Britain ruled Burma, with a big rupture during WWII when Japan invaded Burma.

After the war, the Burmese government the British left behind was too weak to withstand the combination of vigorous civil conflict with ethnic minorities on the frontiers and a 1962 coup by military leaders.

Beginning in 1962, Burma—renamed Myanmar in English by the ruling junta, but the two names come from the same root—is under incredibly tight military rule. Beginning in 2011, the junta relaxes their grip, and Myanmar begins a precarious transition to a more democratic form of government. Note: All through this entire period, Myanmar’s military, the Tatmadaw, is also waging a 70-year civil war with armed insurgencies associated with several different ethnic minorities.

Starting the year after Myanmar’s first quasi-democratic general election in 2010, the Burmese government begins granting mass amnesties to the country’s many political prisoners. Hundreds of people—journalists, activists, artists, religious leaders, and many more—are released over the next few years.

In the mass amnesty granted in January of 2012, two of the many prisoners released will go on to found organizations that play major and opposing roles in escalating crises of communal and military violence that Myanmar’s entry to the internet will fuel.

One is Nay Phone Latt, an early blogger and digital rights activist, who was jailed in 2008, basically for blogging about the 2007 Saffron Revolution demonstrations. Nay Phone Latt will go on to co-found MIDO, an organization dedicated to helping Myanmar’s ordinary citizens reap the benefits of the newly available internet. Over the next several years, MIDO will come up repeatedly in the intense struggle against online hate campaigns, especially on Facebook.¹¹

The second is Ashin Wirathu, a Buddhist monk jailed in 2003 for sermons inciting violence against Myanmar’s Muslim communities.¹² Wirathu will go on to digitize Myanmar’s hardline extremist Buddhist movement, which will play a major role in the coming waves of anti-Muslim violence—also especially on Facebook.

---

In 2013, Faine Greenwood goes back to Myanmar to write about the country’s first Internet Freedom Forum, “a gathering dedicated to helping Myanmar’s people take advantage of the new, liberated Internet.” Faine writes about the heady vibes:

Nay Phone Latt spoke at the conference, and so did a number of the other brilliant young Burmese tech enthusiasts I’d met before. The mood was still buoyantly optimistic as we circulated from one Post-It note-filled brainstorming session to the next, as we drank tea, discussed Internet freedom regulations and online privacy.

And yet, I could detect a slight edge in the air, a certain trepidation that had grown, mutated into new forms, in the few months since I’d been away […] During the conference, we talked about how hateful talk about the Rohingya was starting to pop up on Facebook, about how it was casting an ominous shadow over the good things about helping more people get online.¹³

The edge in the air and the ominous shadow aren’t just vague feelings—they’re connected to a surge in communal violence the same year of that critical mass amnesty, and to a parallel rise in online hate speech.

MIDO co-founder and program director Htaike Htaike Aung spoke on the online dynamics she and her colleagues encountered in the early days:

Unlike in countries where people gradually got used to the Internet and learnt how to find good content, thus learning what is bad content, for Myanmar this hype went straight up and people did not have the time to reflect on what the Internet is actually about. This perception can be summarized in the phrase: “Okay, if it’s on the Internet it must be right”. This really is dangerous, particularly if there are people who are using the Internet for the wrong agenda and propaganda.¹⁴

In 2014, Nay Phone Latt explains what he’s seeing and why he’s worried about it:

When we advocate for free speech, reducing hate speech is included.… Speech calling for hitting or killing someone is hate speech, and can spread hate among people and is a risk for society… It is the wrong use of freedom of speech. I am worried about that because it is not only spreading on social media but also by some writers and [Buddhist] monks who are spreading hate speech publicly. […]

If people hate each other, a place will not be safe to live. I worry about that most for our society. In some places, although they are not fighting, hate exists within their heart because they have poured poison into their heart for a long time [through hate speech]. It can explode in anytime.¹⁵

So what were Htaike Htaike Aung and Nay Phone Latt and their colleagues seeing that made them so worried? Here are some Facebook comments from all the way back in 2011, sparked by the BBC’s audacity in…referring to Rohingya people as a ethnic group that exists in Myanmar at all. The BBC’s offending infographic had been up for a year before the BBC’s Burmese Facebook page was flooded with comments:

Kick out all Muslim Kalar [Rohingyas, South Asians/Indians] from Burma. If this doesn’t work, then kill them to death. It’s time for Arakan to unite with each other.

Don’t assume that I won’t sharpen my knife. I am ready to make it sharp for the sake of protecting our nation, religion and races against those Bengali cheaters.

F@#$-ing Kalar, we will slap your face with shoes and cut your heads. Don’t criticize the god with little of what you know. We will set you on fire to death and turn the mosques into wholesales/retail pork markets…¹⁶

The New Mandala article about the controversy includes memes mocking BBC journalist Anna Jones using vintage 4chan trollface visuals, if that helps situate things for you.

So this is the landscape already in place before Nay Phone Latt was released from prison in 2012. And things don’t get any better as Facebook/internet adoption spikes.

We know now that behind the scenes, Burmese activists and organizations—and concerned westerners—spent this period desperately trying to get Facebook to act on the rapidly rising surge of anti-Rohingya hate speech. But neither those efforts nor public educational/digital literacy campaigns can keep up with what’s happening online.

The Rohingya

I need to make a brief detour, about the Rohingya. Here’s Médicins Sans Frontières, from their reference page about the Rohingya refugee crisis:

The Rohingya are a stateless ethnic group, the majority of whom are Muslim, who have lived for centuries in the majority Buddhist Myanmar, mostly in the country’s north, in Rakhine state. However, Myanmar authorities contest this; they claim the Rohingya are Bengali immigrants who came to Myanmar in the 20th century.

Described by the United Nations in 2013 as one of the most persecuted minorities in the world, the Rohingya are denied citizenship under Myanmar law. ¹⁷

It’s hard to overstate how contested those basic facts are within Myanmar itself, where successive governments have rejected any recognition of Rohingya existence, let alone legitimacy, referring to them instead as illegal “Bengali” immigrants.¹⁸ (The Rohingya Culture Center in Chicago has published a concise timeline of Rohingya history, and it makes a good, quick orientation.¹⁹)

Going deeper, things get immediately, bogglingly complicated. My background is inadequate to the task of explaining the entangled ethnic, religious, and political histories of Myanmar in even just Rakhine (formerly Arakan) State, where the Rohingya crisis takes place. What you need to know to understand this series is mostly just that Rohingya people do in fact exist and have long lived under crushing restrictions. Also that Myanmar’s Buddhist political mainstream has long been concerned with racial purity and security. And openly so: the official motto of the Ministry of Immigration and Population is: “The earth will not swallow a race to extinction but another race will.”²⁰

The rumors and the killings (2012)

Content warning: Violence against men, women, and children.

On the 28th of May, 2012, a Buddhist Rakhine woman called Ma Thida Htwe is killed in a village in Rakhine State. The next day, a newspaper reports that she was raped and murdered by “kalars” and calls the killing “the worst homicide case in Myanmar.” Almost immediately, graphic photos of her body begin circulating online. ²¹

The findings of UN Mission note that although Ma Thida Htwe was clearly murdered, both the rape allegation and ethnic origin of the suspected culprits remain in doubt, and that, “In the following days and weeks, it was mainly the rape allegation, more than the murder, which was used to incite violence and hatred against the Rohingya.”²²

For the people who’ve been working to whip up anti-Rohingya hatred and violence, the crime against Ma Thida Htwe is a perfect opportunity—and they use it, immediately, to suggest that it’s merely the leading edge of a coming wave of attacks by Rohingya terrorists. Four days after Ma Thida Htwe’s death, on June 1st, Zaw Htay, the spokesman of the President of Myanmar, posts a statement on Facebook warning that “Rohingya terrorists” are “crossing the border into Myanmar with weapons.” He goes on:

We don’t want to hear any humanitarian or human rights excuses. We don’t want to hear your moral superiority, or so-called peace and loving kindness.… Our ethnic people are in constant fear in their own land. I feel very bitter about this. This is our country. This is our land.²³

On June 3rd, five days after the Rakhine woman’s murder, a nationalist Buddhist group hands out leaflets in a village in southern Rakhine State stating that Muslims are assaulting Buddhist women. A bus carrying ten Rohingya men attempts to pass through the village that same day, and the villagers haul the men out, beat them to death, and destroy the bus. A few days later, Rohingya people in a nearby town riot, killing Buddhist villagers and burning homes. ²⁴

In an article published on the academic regional analysis site New Mandala only seven days after the Rohingya men are killed in Rakhine State, Burmese PhD candidate Sai Latt writes:

What actually triggered public anger? It may have been racial profiling by the ethnic Arakan news agency, Narinjara… When the rape incident took place, the agency published news identifying the accused with their Islamic faith. In its Burmese language news, the incident was presented as if Muslims [read aliens] were threatening local people, and now they have raped and brutally killed a woman. The words—Muslims, Kalar, Islam—were repeatedly used in its news reports. The news spread and people started talking about it in terms of Kalar and “Lu Myo Char [i.e. different race/people or alien] insulting our woman”.

Interestingly, Naranjara and Facebook users started talking about the accused rapists as Kalars even when the backgrounds of the accused were still unclear.²⁵

In the wave of military and communal violence that follows, members of both Rakhine (the dominant Buddhist ethnicity in Rakhine State) and Rohingya communities commit murder and arson, but the brunt of the attacks falls on the Rohingya. The UN Mission’s findings describe house-burning, looting, “extrajudicial and indiscriminate killings, including of women, children and elderly people,” and the “mass arbitrary arrests” and torture of Rohingya people by soldiers and police. More than 100,000 people, most of them Rohinyga, are forced from their homes. ²⁶

The next wave of violence comes in October and shows evidence of coordinated planning. According to Human Rights Watch, the fall attacks were “organized, incited, and committed by local Arakanese political party operatives, the Buddhist monkhood, and ordinary Arakanese, at times directly supported by state security forces.” The attacks included mass killings of Rohingya men, women, and children, whose villages and homes were burned to the ground, sometimes in simultaneous attacks across geographically distant communities.

In the worst attack that October, police and soldiers in Myanmar’s military, the Tatmadaw, preemptively confiscate sticks and other rudimentary weapons from Rohingya villagers, then stand by and watch while a Rakhine mob murders at least 70 Rohingya people over the course of a single day.

Human Rights Watch reports that 28 children are hacked to death in the attack, including 13 children under the age of 5.²⁷

Every spark is more likely to turn into a fire

According to many of Amnesty International’s interviews with Rohingya refugees, the 2012 episodes of communal violence mark the turning point of Myanmar’s slide into intense anti-Rohingya rhetoric, persecution, and, ultimately, genocide. Mohamed Ayas, a Rohingya schoolteacher and refugee, puts it this way:

We used to live together peacefully alongside the other ethnic groups in Myanmar. Their intentions were good to the Rohingya, but the government was against us. The public used to follow their religious leaders, so when the religious leaders and government started spreading hate speech on Facebook, the minds of the people changed.²⁸

In a short 2013 documentary, journalist and filmmaker Aela Callan interviews Richard Horsey, a Myanmar-based political analyst with decades of experience in the country and (now, at least) advisor to The International Crisis Group.

Horsey’s take on the situation in Myanmar accords with the Amnesty interviews:

We’ve seen violence in previous decades by Buddhist populations against Muslim populations, but what’s new is that this information is readily available and transmissible. People are using Facebook and mobile phones, and so you get a much greater resonation every time there’s an issue. Every time there’s a spark, it’s much more likely to turn into a fire.²⁹

In the months between the June and October waves of violence in Myanmar, dehumanizing and violence-inciting anti-Rohingya messages continue to circulate on Burmese-language Facebook. Western observers who fly in to conduct investigations emphasize town-hall meetings and pamphlets spreading hateful and violent rhetoric, but—while noting that it was also surging online—write off the internet as a major influence on the events because internet access remains unusual in rural Rakhine State.³⁰

The very early and very strong concerns of Burmese observers and activists I’ve cited above—Htaike Htaike Aung, Nay Phone Latt, and Sai Latt—about hate speech on Facebook specifically suggest to me that it’s possible that western observers may have underestimated the role of secondhand transmission of internet-circulated ideas, including printed copies of internet propaganda, which are cited as a vector in an interview with another MIDO co-founder, Phyu Phyu Thi. (This exact dynamic—the printed internet—shows up in Sheera Frenkel’s 2016 BuzzFeed News article, which cites “print magazines called Facebook and The Internet.”)³¹

In any case, online and off, hardline Buddhist monks, members of the government, and what appear to be ordinary people are all echoing a few distinct ideas: They claim that the Rohingya are not a real ethnic group, but illegal “Bengali” immigrants; subhuman animals who “outbreed” Buddhists; indistinguishable from terrorists; an immediate danger to the sexual purity and safety of Buddhist women and to Buddhist Myanmar as a whole. ³²

Sai Latt writes in June 2012 that the anti-Rohingya hate campaign is “a public and transnational movement orchestrated openly on social media websites,” which includes thousands of people from the Burmese diaspora:

The campaign is so severe that such comments and postings have been littered on thousands of Facebook walls and pages. Thousands of Burmese-speaking Facebook users are exposed to them every single day. Many Facebook groups such as “Kalar Beheading Gang” appeared one after another. While it is possible to report abuse to Facebook, there were few Burmese Muslim Facebook users who would report to Facebook. Even when they did successfully, new groups keep appearing frequently.

That Facebook Page, “Kalar Beheading Gang,” shows up in the international press as well. It features “an illustration of a grim reaper with an Islamic symbol on its robe on a blood-spattered background,” and a stream of graphic photos presented as evidence of Rohingya atrocities. By the time the Hindustan Times article goes to press on June 14, 2012, the Page already has more than 500 likes.³³

When Sai Latt says that what he’s seeing on Facebook is an orchestrated movement, he’s right. Remember Wirathu, the Buddhist monk released in the same amnesty as MIDO’s Nay Phone Latt? Soon after his release from prison, he’s one of the people doing the orchestrating.

“A faster way to spread the messages”

“We are being raped in every town, being sexually harassed in every town, being ganged up on and bullied in every town.… In every town, there is a crude and savage Muslim majority.” —Wirathu, 2013 ³⁴

Ashin Wirathu is very, very good at Facebook. He spoke with BuzzFeed News about his history on the platform:

[Wirathu’s] first account was small, he said, and almost immediately deleted by Facebook moderators who wrote that it violated their community standards. The second had 5,000 friends and grew so quickly he could no longer accept new requests. So he started a new page and hired two full-time employees who now update the site hourly.

“I have a Facebook account with 190,000 followers and a news Facebook page. The internet and Facebook are very useful and important to spread my messages,” he said.

On the dozens of Facebook pages he runs out of a dedicated office, Wirathu has called for the boycott of Muslim businesses, and for Muslims to be expelled from Myanmar.³⁵

In a 2013 video interview with the Global Post, Wirathu speaks placidly about the possibility of interfaith problem-solving and multi-ethnic peace—and then, with no transition or change of tone, about Muslims “devouring the Burmese people, destroying Buddhist and Buddhist order, forcefully taking action to establish Myanmar as an Islamic country, and forcefully implementing them.” To these absolute fantasies—Muslims make up maybe five percent of Myanmar’s population at this time—Wirathu adds that, “Muslims are like the African carp. They breed quickly and they are very violent and they eat their own kind.”³⁶

In 2013, Time puts Wirathu on the cover of its international editions as “The Face of Buddhist Terror.” Time reporter Hannah Beech quotes him in the cover story explaining that 90% of Muslims in Myanmar are “radical bad people,” and that Barack Obama is “tainted by black Muslim blood.“ ³⁷

Cover of Time's international editions for July 1, 2013. The photo shows Ashin Wirathu, a Burmese man wearing the crimson robe of a Buddhist monk. His head is shaved and his expression is thoughtful. The cover text reads The Face of Buddhist Terror, in progressively large type, and underneath, How militant monks are fueling anti-Muslim violence in Asia, by Hannah Beech.

Photo by Adam Dean/Panos

It would be hard to overstate Wirathu’s influence in Myanmar during the years leading up to the beginning of the Rohingya genocide. One NGO writes that, “It is noteworthy that almost every major outbreak of communal violence since October 2012 in Rakhine state has been preceded by a 969-sponsored preaching tour in the area, usually by [Wirathu] himself.”³⁸ In fact, the same thing is true of the 2012 violence, though from a little further away: Wirathu led a rally in Mandalay in September of 2012 in support of a proposal by then-President Thein Sein to deport the Rohingya en masse. The worst of the violence in Rakhine State took place the very next month.³⁹

I think it’s useful to note that the 969/Ma Bha Tha movement Wirathu is associated with is also a force for practical and immediate good among Myanmar’s Buddhist majority—they run or support lots of Buddhist religious education for children, legal aid, donation drives, and relief campaigns. And these things, along with the high esteem in which monks are generally held in Myanmar, makes them seem exceptionally credible.⁴⁰

Wirathu himself is beloved not only for his incendiary sermons that “defend” Buddhism by demonizing Muslims, but also for his direct involvement in community support. If the project of supporting and defending Buddhism is extended to eliminating the Rohingya, maybe those two kinds of action blur. In Callan’s documentary, Wirathu holds office hours to help ordinary community members solve problems and disputes. His attention flicks constantly between the petitioners kneeling in front of him and the smartphone cradled in his hand.

A few years later, Wirathu—by then running a Facebook empire with nearly 200,000 followers, dozens of Pages, and a full-time staff—sums up the situation: “If the internet had not come to [Myanmar], not many people would know my opinion and messages like now,” Wirathu told BuzzFeed News, adding that he had always written books and delivered sermons but that the “internet is a faster way to spread the messages.” ⁴¹

A quick aside

I don’t want to spend much time on my process, but I do want to make a note about the provenance of a document that would otherwise have unclear sourcing.

Several accounts of Meta’s role in Myanmar’s persecution of the Rohingya include partial lists of the many warnings people on the ground in Burma—and concerned international human-rights watchers—delivered to Meta executives between 2012 and 2018. Taking on the implications of that litany of warnings and missed chances is what first gave me the first taste of the sinking dread that has characterized my experience of this project.

Because of citations in Amnesty International’s big Meta/Myanmar report, I knew there was a “briefing deck” floating around somewhere that had more details, but the closest reference I could find led to a dead URL that Archive.org didn’t have, and emailing around didn’t turn anything up.

Eventually, I found a version of it on a French timeline-making site in the form of a heavily annotated timeline, and cross-referenced that version with tweets from Htaike Htaike Aung and Victoire Rio, the women credited in the Amnesty report with making the deck. The URL they link to in the tweets is the dead one, but the preview image in their tweets is an exact match for the timeline I found. I cite this timeline a lot below.

Edited Oct. 16 2023 to add: The original material is now back online in one form at the Myanmar Internet Project website, and someone affiliated with the project also passed on a PDF form with attached case studies, which I’ve archived via Document Cloud: “Facebook and the Rohingya Crisis.”

The warnings

The timeline showing Burmese civil-society and digital-rights leaders' attempts to warn Meta about what it was enabling in Myanmar.

Screenshot of the timeline discussed above.

All the way back in November of 2012, Htaike Htaike Aung, MIDO’s program director, speaks with Meta’s Director of Global Public Policy and Policy Director for Europe about the proliferation of hate speech on Facebook at a roundtable in Azerbaijan. Eleven months later, in October 2013, Htaike Htaike Aung brings up her concerns again in the context of “rising inter-communal tensions” at a roundtable discussion in Indonesia attended by three Meta policy executives including the Director of Global Public Policy.⁴²

At the same time, activists and researchers from MIDO and Yangon-based tech-accelerator Phandeeyar “followed up over email to ask for ways to get Facebook to review problematic content and address emergency escalations.” Facebook didn’t respond. MIDO and Phandeeyar staff start doing “targeted research on hate speech.”⁴³

In November of 2013, Aela Callan (whose documentary work I’ve cited) meets with Facebook’s VP of Communications and Public Policy, Eliot Schrage, to warn about anti-Rohingya hate speech—and the fake accounts promoting it—on Facebook.⁴⁴ After her visit, Meta puts Callan in touch with Internet.org—the global initiative that would eventually cement Facebook’s hegemony in Myanmar, among other places—and with Facebook’s bullying-focused “Compassion Team,” but not with anyone inside Facebook who could actually help.⁴⁵

In March 2014, four months before deadly, Facebook-juiced riots break out in Mandalay, Htaike Htaike Aung travels to Menlo Park with Aela Callan after RightsCon Silicon Valley. The two women met with members of Meta’s Compassion Team to try once again to get Facebook to pay attention to the threat its service poses in Myanmar. ⁴⁶

In the same month, Susan Benesch, head of the Dangerous Speech Project, coordinates a briefing call for Meta, attended by half a dozen Meta employees, including Arturo Bejar, one of Facebook’s heads of engineering.

On the call, Matt Schissler, a Myanmar-based human-rights specialist, delivers a grim assessment describing dehumanizing messages, faked photos, and misinformation spreading on Facebook. In An Ugly Truth: Inside Facebook’s Battle for Domination, their 2021 book about Facebook’s dysfunction, reporters Sheera Frenkel and Cecilia Kang describe Schissler’s presentation:

Toward the end of the meeting, Schissler gave a stark recounting of how Facebook was hosting dangerous Islamophobia. He detailed the dehumanizing and disturbing language people were using in posts and the doctored photos and misinformation being spread widely.⁴⁷

Reuters notes that one of the examples Schissler gives Meta was a Burmese Facebook Page called, “We will genocide all of the Muslims and feed them to the dogs.” ⁴⁸

None of this seems to get through to the Meta employees on the line, who are interested in…cyberbullying. Frenkel and Kang write that the Meta employees on the call “believed that the same set of tools they used to stop a high school senior from intimidating an incoming freshman could be used to stop Buddhist monks in Myanmar.”⁴⁹

Aela Callan later tells Wired that hate speech seemed to be a “low priority” for Facebook, and that the situation in Myanmar, “was seen as a connectivity opportunity rather than a big pressing problem.”⁵⁰

Then comes Mandalay.

“Facebook said nothing”

I don’t have space to tell the whole story of the 2014 Mandalay violence here—Timothy McLaughlin’s Wired article is good on it—but it has all the elements civil society watchdogs have been worried about: Two innocent Muslim men are falsely accused of raping a Buddhist woman, sensationalist news coverage explodes, and Ashin Wirathu exploits the story to further his cause. Riots break out, and all kinds of wild shit starts circulating on Facebook.

This is exactly the kind of situation Htaike Htaike Aung, Callan, Schissler, and others had been warning Meta could happen—but when it actually arrives, no one at Meta picks up the phone.

As the violence in Mandalay worsens, the head of Deloitte in Myanmar gets an urgent call. Zaw Htay, the official spokesman of Myanmar’s president, needs help reaching Meta. The Deloitte executive works into the night trying to get someone to respond, but never gets through. On the third day of rioting, the government gets tired of waiting and blocks access to Facebook in Mandalay. The block works—the riots died down. And the Deloitte guy’s inbox suddenly fills with emails from Meta staffers wanting to know why Facebook has been blocked.⁵¹

The government wasn’t the only group trying to get through. A few months before, a handful of Facebook employees—including members of policy, legal, and comms teams—had formed a private Facebook group, to allow some of the civil-society experts from Myanmar to flag problems to Facebook staff directly.

With false reports and calls to violence beginning to spread on Facebook in Mandalay, Burmese activists and westerners in the private group try to alert Facebook. They get no response, even after the violence turns deadly. A member of the group later tells Frenkel and Kang:

When it came to answering our messages about the riots, Facebook said nothing. When it came to the internet being shut down, and people losing Facebook, suddenly, they are returning messages right away. It showed where their priorities are.⁵²

A couple of weeks after the violence in Mandalay in 2014, Facebook’s Director of Policy for the Asia-Pacific region, Mia Garlick, travels to Myanmar for the first time. In a panel discussion, Garlick discusses Meta’s policies and promises to…speed up the translation of Facebook’s Community Standards (its basic content rules) into Burmese. That single piece of translation work—Facebook’s main offering in response to internationally significant ethnic violence about which the company has been warned for two years—takes Meta fourteen months to accomplish and relies, in the end, on the Phandeeyar folks in the private Facebook group.⁵³

Also in 2014, Meta agrees to localize their reporting tool for hate speech and other objectionable content. Meta employees work with MIDO and other Burmese civil-society people to translate and fine-tune the tool, and they launch it by the end of 2014.

There’s just one problem: As the Burmese civil society people in the private Facebook group finally learn, Facebook has a single Burmese-speaking moderator—a contractor based in Dublin—to review everything that comes in. The Burmese-language reporting tool is, as Htaike Htaike Aung and Victoire Rio put it in their timeline, “a road to nowhere.”⁵⁴

Yet more warnings

Heading into 2015, the warnings keep coming:

February 2015: Susan Benesch, founder and director of the Dangerous Speech Project at the Berkman Klein Center (that’s the Berkman Center as was, for my fellow olds) gives a presentation called “The Dangerous Side of Language” during Compassion Day at Facebook. According to this legal filing for a class-action suit against Meta on behalf of the Rohingya, the presentation explains “how anti-Rohingya speech is being disseminated by Facebook.”⁵⁵

March 2015: Matt Schissler travels to Menlo Park to meet with Facebook employees in person. Reuters reports that he “gave a talk at Facebook’s California headquarters about new media, particularly Facebook, and anti-Muslim violence in Myanmar. More than a dozen Facebook employees attended, he said.” Frenkel and Kang describe his presentation as documenting “the seriousness of what was happening in Myanmar: hate speech on Facebook was leading to real-world violence in the country, and it was getting people killed.”⁵⁶

After his meeting at Meta, Schissler has lunch with a few Meta employees, and one of them asked Schissler if he thought Facebook could contribute to a genocide in Myanmar, to which Schissler responded, “Absolutely.” Afterwards, Schissler tells Frenkel and Kang, one Facebook employee loudly remarks, “He can’t be serious. That’s just not possible.”⁵⁷

May 2015: Another expert comes to Menlo Park to warn Facebook about the dangerous dynamics it was feeding in Myanmar: David Madden, founder of Phandeeyar, the technology accelerator in Myanmar. In an interview with Amnesty International, Madden reported that “those of us who were working on these issues in Myanmar had a sense that people in Facebook didn’t appreciate the nature of the political situation in the country.”⁵⁸

In his meeting with Meta, Madden discusses specific examples of dangerous content and drew a direct analogy between radio news in Rwanda and Meta’s role in Myanmar, “meaning it would be the platform through which hate speech was spread and incitements to violence were made.”⁵⁹

September 2015: Mia Garlick, Facebook’s Director of Public Policy, Asia-Pacific, comes to Myanmar to finally launch Facebook’s Burmese-language Community Standards. Phandeeyar puts together a group of “15+ civil-society leaders from across Myanmar” to brief Garlick “on specific incidents and actors.”⁶⁰

In an interview with Amnesty International, Victoire Rio reports that several of these leaders specifically tell Garlick that Facebook’s Community Standards aren’t being enforced in Myanmar.⁶¹

Which, of course they aren’t: Reuters reports that in 2015, Facebook employed a total of two (2) Burmese-speaking moderators, expanding to four (4) by the year’s end.⁶²

---

To sum up a little bit, by the end of 2015, Meta knew—as much as any organization can be said to know—that both international civil society experts and the government of Myanmar believe Facebook had a significant role in the 2014 Mandalay riots.

And they’d been warned, over and over, that multiple dedicated civil-society and human-rights organizations believed that Facebook was worsening ethnic conflict.

They’d been shown example after example of dehumanizing posts and comments calling for mass murder, even explicitly calling for genocide. And David Madden had told Meta staff to their faces that Facebook might well play the role in Myanmar that radio played in Rwanda. Nothing was subtle.

After all that, Meta decided not to dramatically scale up moderation capacity, to permanently ban the known worst actors, or to make fundamental product-design changes to reliably deviralize posts inciting hatred and violence.

Instead, in 2016, it was time to get way more people in Myanmar onto Facebook.

That’s next, in Part II: The Crisis.

---

1. Weaponizing Social Media: The Rohingya Crisis, CBS News, February 26, 2018 (YouTube lists the date when the documentary was uploaded, not when it was initially broadcast). Also available at the CBS News site, where you will have to watch multiple ads for Pfizer. Content warning: First-person accounts of atrocities.↩︎

2. “U.N. Investigators Cite Facebook Role in Myanmar Crisis,” Reuters, March 12, 2018.↩︎

3. “Facebook Destroys Everything: Part 1, Faine Greenwood, August 8, 2023.↩︎

4. “Myanmar Cuts the Price of a SIM Card by 99%,” Sam Petulla, Quartz, April 3, 2013.↩︎

5. “Digital 2011: Myanmar,” Simon Kemp, Datareportal, December 28, 2011; “Digital 2017: Myanmar,” Simon Kemp, Datareportal, February 1, 2017, both cited in Amnesty International’s report, “The Social Atrocity,” cited elsewhere.↩︎

6. “The Unprecedented Explosion of Smartphones in Myanmar,” Philip Heijmans with assistance from Jason Clenfield, Grace Huang, Masumi Suga, and Mai Ngoc Chau, Bloomberg, July 31, 2012.↩︎

7. “The Facebook-Loving Farmers of Myanmar,” Craig Mod, The Atlantic, January 21, 2016.↩︎

8. “The Facebook-Loving Farmers of Myanmar,” Craig Mod, The Atlantic, January 21, 2016.↩︎

9. “Telenor Brings Zero-Rated Wikipedia and Facebook Access to Myanmar,” LIRNEasia, November 3, 2014.↩︎

10. “Frances Haugen DW News Interview,” DW Global Media Forum, November 10, 2021 (Haugen’s sentence begins at about 7:39 in the video).↩︎

11. “Imprisoned Burmese blogger Nay Phone Latt to receive top PEN honor,” PEN America, April 14, 2010; “‘I Am Really Worried About Our Country’s Future’,” Jessica Mudditt, Mizzima, September 21, 2014 (archived at Burma Link).↩︎

12. “‘It Only Takes One Terrorist’: The Buddhist Monk Who Reviles Myanmar’s Muslims,” Marella Oppenheim, The Guardian, May 12, 2017.↩︎

13. “Facebook Destroys Everything: Part 1,” Faine Greenwood, August 8, 2023.↩︎

14. “‘If It’s on the Internet It Must Be Right’: An Interview With Myanmar ICT for Development Organisation on the Use of the Internet and Social Media in Myanmar,” Rainer Einzenberger, Advances in Southeast Asian Studies (ASEAS), formerly the Austrian Journal of South-East Asian Studies, December 30, 2016.↩︎

15. “‘Hate Speech Pours Poison Into the Heart’,” San Yamin Aung, The Irrawaddy, April 9, 2014.↩︎

16. “BBC under fire on Rohingyas,” Sai Latt, New Mandala, November 3, 2011.↩︎

17. “The Rohingya: Persecuted Across Time and Place,” Médecins Sans Frontières, resource undated.↩︎

18. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 460 — the “report landing page includes summaries, metadata, and infographics. Content warnings apply throughout.↩︎

19. “History of the Rohingya,” Rohingya Culture Center, resource undated.↩︎

20. “‘An Open Prison without End’: Myanmar’s Mass Detention of Rohingya in Rakhine State,” Human Rights Watch, October 8, 2020↩︎

21. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 704. Content warnings apply throughout.↩︎

22. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 625. Content warnings apply throughout.↩︎

23. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 705. Content warnings apply throughout.↩︎

24. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 627. The New Light of Myanmar article cited in the findings is archived at https://www.burmalibrary.org/docs13/NLM2012-06-05.pdf. Content warnings apply throughout.↩︎

25. “Intolerance, Islam and the Internet in Burma,” Sai Latt, New Mandala, June 10, 2012↩︎

26. “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018, paragraph 628. Content warnings apply throughout.↩︎

27. “‘All You Can Do is Pray’: Crimes Against Humanity and Ethnic Cleansing of Rohingya Muslims in Burma’s Arakan State,” Human Rights Watch, April 2013 (PDF version). Content warnings apply throughout.↩︎

28. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022.↩︎

29. Myanmar: Freedom from Hate, Aela Callan, Al Jazeera English, September 5, 2013.↩︎

30. “Symposium on Myanmar and International Indifference: The Rohingya Genocide — Warning Signs, International Inaction, and Missteps,” Matthew Smith, OpinioJuris, August 29, 2022; “‘All You Can Do is Pray’: Crimes Against Humanity and Ethnic Cleansing of Rohingya Muslims in Burma’s Arakan State,” Human Rights Watch, April 2013 (PDF version); “Myanmar Conflict Alert: Preventing Communal Bloodshed and Building Better Relations,” International Crisis Group, June 12 2012.↩︎

31. “‘If It’s on the Internet It Must Be Right’: An Interview With Myanmar ICT for Development Organisation on the Use of the Internet and Social Media in Myanmar,” Rainer Einzenberger, Advances in Southeast Asian Studies (ASEAS), formerly the Austrian Journal of South-East Asian Studies, December 30, 2016; “This Is What Happens When Millions Of People Suddenly Get The Internet,,” Sheera Frenkel, BuzzFeed News, November 20, 2016.↩︎

32. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022; “Report of the Detailed Findings of the Independent International Fact-Finding Mission on Myanmar,” United Nations Human Rights Council, September 17, 2018 (many paragraphs over the report).↩︎

33. “Web users vent rage over Myanmar unrest,” Hindustan Times/AFP June 14, 2012; “How the Word ‘Kalar’ is a Depressing Indictment of Myanmar Society,” Myanmar Mix, April 22, 2020.↩︎

34. “Buddhist Monk Uses Racism and Rumours to Spread Hatred in Burma,” Kate Hodal, The Guardian, April 18, 2013.↩︎

35. “This Is What Happens When Millions Of People Suddenly Get The Internet,,” Sheera Frenkel, BuzzFeed News, November 20, 2016.↩︎

36. “A Burmese Journey Q&A With Ashin Wirathu,” Tin Aung Kyaw, Global Post, June 2013 (archived at the Rohingya Video News YouTube channel); note: the YouTube comment section is now filled with people cheering for Wirathu’s anti-Islamic positions. (Related article about this interview, from which I draw the “African carp” translation.)↩︎

37. “The Face of Buddhist Terror,” Hannah Beech, Time, July 1, 2013.↩︎

38. “Hidden Hands Behind Communal Violence in Myanmar: Case Study of the Mandalay Riots, Justice Trust Policy Report,” Justice Trust, March 2015 (archived at Burma Library).↩︎

39. “‘It Only Takes One Terrorist’: The Buddhist Monk Who Reviles Myanmar’s Muslims,” Marella Oppenheim, The Guardian, May 12, 2017.↩︎

40. “Misunderstanding Myanmar’s Ma Ba Tha,” Matthew J Walton, Asia Times, June 9, 2017.↩︎

41. “This Is What Happens When Millions Of People Suddenly Get The Internet,” Sheera Frenkel, BuzzFeed News, November 20, 2016.↩︎

42. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022, p 51; “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022.↩︎

43. “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022.↩︎

44. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022.↩︎

45. “Hatebook: Why Facebook Is Losing the War on Hate Speech in Myanmar,” Reuters, August 15, 2018. This piece is part of Reuters’ extraordinary, Pulitzer-winning coverage of the Rohingya genocide, Myanmar Burning, which landed two Burmese Reuters reporters in prison in Myanmar for documenting a mass killing of Rohingya people; both men were released in 2019. “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022.↩︎

46. “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022.↩︎

47. An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021—I read the ebook edition, so I have no page numbers to cite, but this comes from Chapter Nine: “Think Before You Share.”↩︎

48. Weaponizing Social Media: The Rohingya Crisis, CBS News, February 26, 2018 (YouTube lists the date when the documentary was uploaded, not when it was initially broadcast). Also available at the CBS News site, where you will have to watch multiple ads for Pfizer. Content warning: First-person accounts of atrocities.↩︎

49. An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

50. “Hatebook: Why Facebook Is Losing the War on Hate Speech in Myanmar,” Reuters, August 15, 2018; “How Facebook’s Rise Fueled Chaos and Confusion in Myanmar,” Timothy McLaughlin, Wired, July 6, 2018.↩︎

51. “How Facebook’s Rise Fueled Chaos and Confusion in Myanmar,” Timothy McLaughlin, Wired, July 6, 2018.↩︎

52. An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

53. “How Facebook’s Rise Fueled Chaos and Confusion in Myanmar,” Timothy McLaughlin, Wired, July 6, 2018; An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

54. “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022; “How Facebook’s Rise Fueled Chaos and Confusion in Myanmar,” Timothy McLaughlin, Wired, July 6, 2018; An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

55. Class-action complaint and demand for jury trial,” filed in the Superior Court of the State of California for the County of San Mateo by Rafey S. Balabanian and Richard Fields of Edelson PC and Fields PLLC, The Guardian, December 6, 2021. (I got the date from Reuters.)↩︎

56. An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

57. An Ugly Truth: Inside Facebook’s Battle for Domination, Sheera Frenkel and Cecilia Kang, HarperCollins, July 13, 2021 (Chapter Nine: “Think Before You Share”).↩︎

58. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022.↩︎

59. The Facebook Dilemma, Frontline, October, 2018; the full interview with David Madden is available as well.↩︎

60. “Rohingya and Facebook,” Htaike Htaike Aung, Victoire Rio, possibly others, August 2022.↩︎

61. “The Social Atrocity: Meta and the Right to Remedy for the Rohingya,” Amnesty International, September 29, 2022.↩︎

62. “Hatebook: Why Facebook Is Losing the War on Hate Speech in Myanmar,” Reuters, August 15, 2018.↩︎

28 September 2023

Meta Meta

I wrote so many posts that that my posts needed a post. Sorry about that.

The series this note accompanies is a work of synthesis, in which I attempt to collect up the hundreds of pieces of the overall story of Meta’s actions (and inactions) in Myanmar before and after the peak of the genocide of the Rohingya people in 2016 and 2017, and make them intelligible to a readership of English-speaking people who want to make or use an internet that doesn’t replicate these dynamics.

If you spot typos or inaccuracies, I’m reachable at erin at incisive.nu and appreciate all bug reports. Someday I may even fix up the font loading on this website.

Limitations (mine)

I’m the furthest thing from a scholar of Burmese history or politics, and although my background has included working with atrocity materials in other contexts, I’m also not a genocide scholar. My aim with this series is to give mostly-western makers and users of social technology a sense of one US-based technology company’s role in what happened to just one group of people in just one place over a very limited time range. As enormously long as this series is, I’ve left out vastly more than I’ve put in, especially about the complexities of both colonial and post-colonial violence and repression in Myanmar.

If you want to get a sense of Myanmar’s history both generally and as it relates to oppression of the Rohingya and other ethnic minorities, I really like Carlos Sardiña Galache’s The Burmese Labyrinth (Verso, 2020) and Thant Myint-U’s The Hidden History of Burma (WW Norton, 2019).

Content warnings

This is a series about a US technology company’s role in a genocide, so it describes the events leading up to and constituting the genocide.

I’ve tried to be really careful about what makes it into the text, and I haven’t included any graphic descriptions or images—or any images of traumatized fleeing refugees, for that matter.

Within the text of my posts, I warn readers before getting into particularly rough sections dealing with events in Myanmar. But I do say, in general terms, what happened, which was about as bad as anything can be. The sources I cite range from high-level and abstract to literal catalogs of atrocities—and even some of the regular newspaper reporting includes images that, if you look at them closely, may haunt you. If you decide to get into the source documentation and you don’t have experience working with atrocity material, you might want to keep the Tetris intervention handy.

I also quote examples of anti-Rohingya and anti-Islamic hate speech throughout the series, some of which include terms that qualify as slurs and some of which are just horrible things to say.

Terms

It’s been really hard to figure out what to call things in a story about Meta’s actions in Myanmar—or should that be Facebook’s actions in Burma?

When I talk about the platform, I say “Facebook,” but when I talk about the company that runs Facebook, I say “Meta,” no matter what they were called at the point in history I’m looking at. There’s a good argument to be made for just calling them Facebook all the way through.

Is it silly to talk about Meta in 2013, when the company was still called Facebook? Probably! But so are corporate rebranding exercises. Also, we do this with humans all the time, and I’m pretty committed to calling people and entities what they ask to be called.

Let this be my tiny part in making sure that the construct called Meta can’t escape responsibility for what the construct called Facebook did.

When I talk about Myanmar, I say “Myanmar,” not “Burma,” because that’s what the country’s government calls the country in English and what most other countries and international bodies use. There are some linguistic complexities and tangled history, and many people from Myanmar still use “Burma” in English-language text. As a white American, I think using colonial-period names in my work on principle would be a little weird, so I’m erring on the side of “use the name the entity requests.”

My own government refuses to write “Myanmar” in official documents because the Burmese government that switched from using “Burma” to “Myanmar” as the official English name was a military government that took control of the country in a coup in 1962. Given the United States’ history of supporting both military governments and coups elsewhere in the world, I’m…not going to say anything else about that.

Things get trickier, though, in referring to people from Myanmar, and to the most commonly spoken language. Technically, “Myanma” is the official term for people who live in Myanmar, inclusive of both the Bamar ethnic majority and the officially recognized ethnic minorities, but in all my reading for this series, I encountered that exact usage about three times, total. Similarly, the most commonly spoken language in Myanmar is officially “the Myanmar language,” but in practice, the usage is vanishingly rare, so I use “Burmese” in both cases.

More place and ethnicity names: The city formerly known as Rangoon is now Yangon. Arakan State and the Arakan ethnicity are now Rakhine State and the Rahkine ethnicity, but my sources are all over the place, so some quoted material still uses the old terms.

Personal names: Burmese-language personal names can be tricky for western readers—there are multiple formulations of many names, and a person’s name isn’t divided into given name and surname/family name. I’ve given the most common formulation of Burmese people’s names, and provide them in full each time they come up, because shortening them to a faux-surname would be weird.

The “Ashin” in “Ashin Wirathu” is an honorific used for monks and certain other honored figures, so I’ve dropped it after the first reference. In quotations and cited sources, you’ll also see “U Wirathu”‚ and “U” here is an honorific use of “uncle.” (Relatedly, the “Daw” in “Daw Aung San Suu Kyi” is the “auntie” honorific.)

Sources and citations

I’ve cited sources very heavily in this series because I want to provide a highly explorable web of information, and because I think the subject demands attention to sourcing. And because all of this stuff came from other people and they deserve credit. I’ve done a lot of plain old hyperlinking as well, but the footnotes in each post provide the most complete set of references. I use a roughly Chicago-style citation method, but with titles first instead of authors because a lot of my references lack designated authors and also it’s my website and I can do whatever I want.

I quote from a lot of US- and UK-based media sources. When I do, I lean on in-depth investigative reporting rather than more general explainery coverage. I also rely heavily on the work of the UN Human Rights Council’s Independent International Fact-Finding Mission on Myanmar, and on reporting and research from Amnesty International, Human Rights Watch, and Médicins Sans Frontières.

A sidenote: One of the things I did to get up to speed for this series was watch a lot of video, including interviews with physically scarred Rohingya women and children in refugee camps, as well as footage of incidents of violence in Rahkine State. I also read detailed reports on specific incidents in Myanmar, because I wanted to make sure that I understood the nature of the evidence before making claims. I’m not going to discuss any of that except to say that what I saw and read provided enough evidence for me to feel very secure citing the sources I’ve cited.

Especially in the third and fourth posts in this series, I use a lot of material from internal Meta documents disclosed by whistleblowers. Meta responds to all whistleblower disclosures with blanket denials, and sometimes frankly absurd statements meant to act, I think, as radar chaff. Having read a bunch of these responses, I don’t think they actually add anything at all to the conversation, so I rarely quote from them, but I do link to some in footnotes.

Also, I’ve included views from as many Burmese civil-society and digital-rights people as I could cram in, along with first-hand perspectives from several western journalists and tech people who spent time in Myanmar. I’ve tried to rely on local, Burmese-speaking sources for ground truths about the interaction of the internet with communal and state violence, and include western perspectives for backup and a sense of the vibes.

The Rohingya

I deal with pieces of the history of the Rohingya at various points throughout the series, so here I will confine myself to terminology and the basic facts of Rohingya existence.

The most deeply researched accounts I’ve read about the history of the Rohingya in Rakhine State agree that there have been population flows across what is now the border between Myanmar and Bangladesh for centuries, and that the pre-2016 Rohingya population in Rakhine has roots reflecting both longstanding local populations and periodic inflows from what is now Bangladesh. These inflows included people “originally” from what is now Rakhine State in Myanmar, who’d been forced to migrate north and later returned, as well as people “originally” from the area now known as Bangladesh.

Claims that the Rohingya are “illegal Bengali immigrants” were widely accepted in Myanmar before the 2021 coup, but they’re best understood as political maneuvering intended to demonize an ethnic group, not statements of fact.

While we’re on the subject, it’s my opinion that borders are not a coherent way of describing ancestry and belonging, or of parceling out the obligation to treat people like humans.

Changelog

Friday, September 29: Fixed some typos in this post, expanded content warnings in this post and Part I to explicitly include hate speech and ethnic slurs, and refined the language in the section in this post that deals with using “Burma” vs. “Myanmar.” I’ve also reorganized some of the reading recommendations in this post and tried to make explicit some of the many things that I’m necessarily leaving out. Thank you to readers for raising flags about these issues.

Saturday, September 30: Posted Part II, fixed a couple of typos throughout and one inexplicable word substitution in Part II (a random insert of “Malaysian,” which shouldn’t be in the post at all). Thank you to everyone who reported!

Friday, October 6: Posted Part III. Fixed several typos in the other two parts and changed some ambiguous phasing to clarify Arturo Bejar’s position in the Facebook org chart. Thank you to all reporters of errors! I’m sure I’ve made several more.

Friday, October 13: Posted Part IV. Fixed yet more typos.

Monday, October 16: Posted the series home and acknowledgements post. Fixed yet more typos, probably.

28 September 2023

Mastodon Is Easy and Fun Except When It Isn’t

Ed. note: This post is from July, 2023. It circulates every month or so on social networks and a lot of people think it’s new, probably because I run dates at the bottom of each post intead of just under the title. Until I get around to changing that, this is your temporal location-finder.

After my last long post, I got into some frustrating conversations, among them one in which an open-source guy repeatedly scoffed at the idea of being able to learn anything useful from people on other, less ideologically correct networks. Instead of telling him to go fuck himself, I went to talk to about fedi experiences with people on the very impure Bluesky, where I had seen people casually talking about Mastodon being confusing and weird.

My purpose in gathering this informal, conversational feedback is to bring voices into the “how should Mastodon be” conversation that don’t otherwise get much attention—which I do because I hope it will help designers and developers and community leaders who genuinely want Mastodon to work for more kinds of people refine their understanding of the problem space.

what I did

I posted a question on Bluesky (link requires a login until the site comes out of closed beta) for people who had tried/used Mastodon and bounced off, asking what had led them to slow down or leave. I got about 500 replies, which I pulled out of the API as a JSON file by tweaking a bash script a nice stranger wrote up on the spot when I asked about JSON export, and then extracted just the content of the replies themselves, with no names/usernames, IDs, or other metadata attached. Then I dumped everything into a spreadsheet, spent an hour or so figuring out what kind of summary categories made sense, and then spent a few more hours rapidly categorizing up to two reasons for each response that contained at least one thing I could identify as a reason. (I used to do things like this at a very large scale professionally, so I’m reasonably good and also aware that this is super-subjective work.)

None of this is lab-conditions research—sorry, I meant NONE OF THIS IS LAB-CONDITIONS RESEARCH—and I hope it’s obvious that there are shaping factors at every step: I’m asking the question of people who found their way to Bluesky, which requires extra motivation during a closed beta; I heard only from people who saw my question and were motivated to answer it; I manually processed and categorized the responses.

I didn’t agonize over any of this, because my goal here isn’t to plonk down a big pristine block of research, but to offer a conversational glimpse into what real humans—who were motivated to try not one, but at least two alternatives to Twitter—actually report about their unsatisfactory experiences on Mastodon.

Lastly, I’ve intentionally done this work in a way that will, I hope, prove illegible and hostile to summary in media reports. It’s not for generalist reporters, it’s for the people doing the work of network and community building.

A note on my approach to the ~data and numbers: It would be very easy to drop a bunch of precise-looking numbers here, but that would, I think, misrepresent the work: If I say that I found at least one categorizable reason in 347 individual replies, that’s true, but it sounds reassuringly sciency. The truth is more like “of the roughly 500 replies I got, about 350 offered reasons I could easily parse out.” So that’s the kind of language I’ll be using. Also, I feel like quoting short excerpts from people’s public responses is fine, but sharing out the dataset, such as it is, would be weird for several reasons, even though people with a Bluesky login can follow the same steps I did, if they want.

got yelled at, felt bad

The most common—but usually not the only—response, cited as a primary or secondary reason in about 75 replies—had to do with feeling unwelcome, being scolded, and getting lectured. Some people mentioned that they tried Mastodon during a rush of people out of Twitter and got what they perceived as a hostile response.

About half of the people whose primary or secondary reasons fit into this category talked about content warnings, and most of those responses pointed to what they perceived as unreasonable—or in several cases anti-trans or racist—expectations for content warnings. Several mentioned that they got scolded for insufficient content warnings by people who weren’t on their instance. Others said that their fear of unintentionally breaking CW expectations or other unwritten rules of fedi made them too anxious to post, or made posting feel like work.

Excerpts:

• Feels like you need to have memorized robert’s rules of the internet to post, and the way apparently cherished longtimers get hostile to new people
• i wanted to post about anti-trans legislation, but the non-US people would immediately complain that US politics needed to be CWed because it “wasn’t relevant”
• I don’t know where all the many rules for posting are documented for each instance, you definitely aren’t presented them in the account creation flow, and it seems like you have to learn them by getting bitched at
• Constantly being told I was somewhat dim because I didn’t understand how to do things or what the unwritten rules were.
• I posted a request for accounts to follow, the usual sort of thing, who do you like, who is interesting, etc. What I got was a series of TED Talks about how people like me were everything that was wrong with social media.
• sooooooo much anxiety around posting. i was constantly second-guessing what needed to be hidden behind a CW
• the fact that even on a science server, we were being badgered to put bug + reptile stuff behind a CW when many of our online presences are literally built around making these maligned animals seem cool and friendly was the last straw for me

What I take from this: There obviously are unwelcoming, scoldy people on Mastodon, because those people are everywhere. I think some of the scolding—and less hostile but sometimes overwhelming rules/norms explanation—is harder to deal with on Mastodon than other places because the people doing the scolding/explaining believe they have the true network norms on their side. Realistically, cross-instance attempts to push people to CW non-extreme content are a no-go at scale and punish the most sensitive and anxious new users the most. Within most instances, more explicit rules presented in visible and friendly ways would probably help a lot.

In my experience, building cultural norms into the tooling is much more effective and less alienating than chiding. The norm of using alt-text for images would be best supported by having official and third-party tools prompt for missing alt-text—and offer contextual help for what makes good alt text—right in the image upload feature. Similarly, instances with unusual CW norms would probably benefit from having cues built into their instance’s implementation of the core Mastodon software so that posters could easily see a list of desired CWs (and rationales) from the posting interface itself, though that wouldn’t help those using third-party apps. The culture side of onboarding is also an area that can benefit from some automation, as with bots on Slack or Discord that do onboarding via DM and taggable bots that explain core concepts on demand.

couldn’t find people or interests, people didn’t stay

A cluster of related reasons came in at #2, poor discoverability/difficulty finding people and topics to follow, #4, missing specific interests or communities/could only find tech, and #7, felt empty/never got momentum. I am treating each group as distinct because I think they’re about subtly but importantly different things, but if I combined them, they’d easily be the largest group of all.

People in the “poor discoverability” group wrote about frustration with Mastodon features: how hard it was to find people and topics they wanted to follow, including friends they believed to already be on Mastodon. They frequently also said they were confused or put off by the difficulty of the cross-server following process as secondary reasons. Several people wrote about how much they missed the positive aspects of having an algorithm help bring new voices and ideas into their feeds, including those that they wouldn’t have discovered on their own, but had come to greatly value. Another group wrote about limited or non-functional search as a blocker for finding people, and also for locating topics—especially news events or specialist conversations.

The “missing specific interests or communities” group wrote about not finding lasting community—that the people and communities they valued most on Twitter either didn’t make it to Mastodon at all, or didn’t stick, or they couldn’t find them, leaving their social world still largely concentrated on Twitter even when they themselves made the move. Several also noted that tech conversations were easy to find on Mastodon, but other interests were much less so.

The “felt empty” group made an effort to get onto Mastodon, and in some cases even brought people over with them, but found themselves mostly talking into a void after a few weeks when their friends bailed for networks that better met their needs.

Excerpts:

• For me, it was that Mastodon seemed to actively discourage discoverability. One of the things I loved most about Twitter was the way it could throw things in front of me that I never would have even thought to go look for on my own.
• I feel like every time I try to follow a conversation there back to learn more about the poster I end up in a weirdly alien space, like the grocery store on the other side of town that’s laid out backwards
• It seemed like it needed to pick a crowd, rather than discover new ones. Fewer chances at serendipity.
• I also remember trying to follow instructions people posted about “simple” ways to migrate over your Twitter follows/Lists, & none of them really worked for me, & I got frustrated at how much time I was spending just trying to get things set up there so I wasn’t completely starting from scratch
• Mastodon was too isolating. And the rules made me feel like the worst poster.
• Quote-replies from good people giving funny/great information is how I decide are important follows.
• Discoverability/self promo is limited & typing out 6 hashtags is annoying. # being in the actual posts clutter things (unlike cohost/insta).
• Difficulty in finding new follows was high up for me. But even once I got that figured out, it was a pain to add new people to follow if they weren’t on my instance.
• finding people you want to follow is hard enough. Adding in the fact that if you joined the wrong server you might never find them? Made it seem not worth the trouble.
• I couldn’t really figure out how to find people and who was seeing what I posted; I was never sure if I had full visibility into that
• the chief problem was an inability to find a) my friends from Twitter who were already there and b) new friends who had similar interests, both due to the bad search function
• Just didn’t seem active enough to feel worth learning all the ins and outs.

What I take from this: Mastodon would be much friendlier and easier to use for more people if there were obvious, easy ways to follow friends of friends (without the copy-paste-search-follow dance). Beyond making that easier, Mastodon could highlight it during onboarding.

Making it easy to search for and find and follow people—those who haven’t opted out of being found—would also be tremendous help in letting people rebuild their networks not just when coming from elsewhere, but in the not-that-rare case of instances crashing, shutting down, or being defederated into oblivion, especially since automatic migration doesn’t always work as intended.

Missing replies also feed into this problem, by encouraging duplicate responses instead of helping people find their way into interesting conversations and notes—a social pattern that several people mentioned as something they prize on more conversationally fluent networks.

too confusing, too much work, too intimidating

The next big cluster includes group #3, too confusing/too much work getting started, group #5, felt siloed/federation worked badly, and group #7, instance selection was too hard/intimidating.

A lot of people in the responding group found the process of picking an instance, signing up, and getting set up genuinely confusing. Others understood how to do it, but found it to be too time-consuming, or too much work for an uncertain return on investment. A couple of people had so many technical errors getting signed up to their first instance that they gave up. Several mentioned that they were so flooded with tips, guides, and instructions for doing Mastodon right that it seemed even more confusing.

Many found the idea and practice of federation to be confusing, offputting, or hostile; they cited difficulties in selecting the “right” instance and shared stories about ending up on an obviously wrong one and then losing their posts or having migration technically fail when they moved. Several explicitly used the words “silo” or “siloed” to describe how they felt trying to find people who shared their interests and also, I think crucially, people who didn’t share special interests, but who would be interesting to follow anyway. (This is obviously intimately tied to discoverability.)

Several brought up patchwork federation and unexpected or capricious defederation. Side conversations sprang up over how difficult people found it to pigeonhole themselves into one interest or, conversely, manage multiple accounts for multiple facets of their lives.

Excerpts:

• My Twitter friends joined various Mastodon servers that didn’t talk to each other and I gave up on trying to figure it out.
• I’m tech savvy and have found mastodon simply opaque. I’ve set up 4 accounts, each on a different server, and don’t know how to amalgamate all the people I’m following everywhere (assuming all those servers federate with each other).
• It was the thing where people had to make whole twitter threads just to explain how to sign up
• the federation model is a mess and it’s impossible to use. i’ve been using computers all day every day since the 90s and mastodon makes me question whether i’m actually good at them
• discovered I was on some kind of different continent from my friends, and could not follow them, nor they me. Immediately felt frustration and disgust and never looked back.
• I was told picking a server didn’t matter. Then it turned out it actually mattered a great deal for discoverability. Then I’m told ‘migrating is easy’, which is just a straight up lie.
• Just 100 tiny points of friction for little return

What I take from this: I agree with these people, and I think all fedi projects meant for a broad audience should focus on fixing these problems.

too serious, too boring, anti-fun

People in this category talked about a seriousness that precluded shitposting or goofiness, and a perceived pressure to stay on topic and be earnest at all times.

• It felt like the LinkedIn version of Twitter - just didn’t have any fun there
• It feels overly earnest and humorless — I don’t consider myself a particularly weird or ironic poster but I want some of those people around saying funny stuff, you know?
• And in the occasional moments where I do feel like being a little silly & humorous, I want to be in a crowd that will accept that side of me rather than expecting a constant performance of seriousness!
• it just didn’t have as much fun or joy as early Twitter and Bluesky
• ultimately, I just bounced off of the culture, because it wasn’t banter-y and fun. It feels too much like eating your vegetables.

What I take from this: Honestly, I think this is the most obvious culture clash category and is less something that needs to be directly addressed and more something that will ease with both growth and improved discoverability, which will help people with compatible social styles find each other. I think the other piece of this is probably the idea of organizing people into interest-based instances, which I think is fundamentally flawed, but that’s a subject for another time.

complicated high-stakes decisions

There’s a meta conversation that is probably unavoidable, and that I’d rather have head-on than in side conversations. It’s about what we should let people have, and it shapes the discourse (and product decisions) about features like quote posts, search, and custom feeds/algorithms—things that are potentially central in addressing some of the problems people raised in their replies to my question on Bluesky.

Broadly speaking, in the landscape around and outside of the big corporate networks, there are two schools of thought about these kinds of potentially double-edged features.

The first, which I’ll call Health First, prefers to omit the features and affordances that are associated with known or potential antisocial uses. So: no quote-posts or search because they increase the attack surface afforded to griefers and nurture the viral dynamics that drive us all into a sick frenzy elsewhere. No custom algorithms because algorithms have been implemented on especially Facebook and YouTube in ways that have had massive and deeply tragic effects, including literal genocide affecting a million adults and children in Myanmar whose lives are no less real than yours or mine.

The second, which I’ll call Own Your Experience, states that people, not software, are responsible for networked harms, and places the burden of responsible use on the individual and the cultural mechanisms through which prosocial behavior is encouraged and antisocial behavior is throttled. So: yes to quote-posts and search and custom feeds, and just block or defederate anyone using them to do already banned things, like harassment or abuse or the kind of speech that, given the right conditions, ignites genocide.

At their simplest, I don’t like either of these positions, though they both get some things right. The Own Your Experience school doesn’t really grapple with the genuinely terrifying dynamics of mass-scale complex systems. And I don’t think the Health First school has come to terms with the fact that in an non-authoritarian society, you can’t make people choose networks that feel like eating their vegetables over the ones that feel like candy stores. Even most people who consciously seek out ethically solid options for their online lives aren’t going to tolerate feeling isolated from most of their peers and communities, which is what happens when a network stays super niche.

From where I stand, there are no obvious or easy answers…which means that people trying to make better online spaces and tools must deal with a lot of difficult, controversial answers.

If I had to pick a way forward, I’d probably define a target like, “precisely calibrated and thoughtfully defanged implementations of double-edged affordances, grounded in user research and discussions with specialists in disinformation, extremist organizing, professional-grade abuse, emerging international norms in trust & safety, and algorithimic toxicity.”

If that sounds like the opposite of fun DIY goofing around on the cozy internet, it is. Doing human networks at mass scale isn’t a baby game, as the moral brine shrimp in charge of the big networks keep demonstrating. Running online communities comes with all kinds of legal and ethical obligations, and fediverse systems are currently on the back foot with some of the most important ones (PDF).

this post is too long, time to stop

Right now, Mastodon is an immense achievement—a janky open-source project with great intentions that has overcome highly unfavorable odds to get to this point and is experiencing both growing pains and pressure to define its future. If I were Eugen Rochko, I would die of stress.

I don’t know if Mastodon can grapple with the complexities of mass scale. Lots of people would prefer it didn’t—staying smaller and lower-profile makes it friendly to amateur experimentation and also a lot safer for people who need to evade various kinds of persecution. But if Mastodon and other fedi projects do take on the mass scale, their developers must consider the needs of people who aren’t already converts. That starts by asking a lot of questions and then listening closely and receptively to the answers you receive.

28 July 2023