The Washington PostDemocracy Dies in Darkness

Saudi Twitter blew up with support for the crown prince. How much of it is genuine?

Tweets in support of Crown Prince Mohammed bin Salman racked up after a U.S. report implicated him in the killing of journalist Jamal Khashoggi

A man displays information on his phone from the Tawakkalna app, launched by Saudi authorities to track people infected with the coronavirus, as he enters the Al-Othaim market in Riyadh, Saudi Arabia, on Feb. 22. (Ahmed Yosri/Reuters)

Last month, the Biden administration released a long-awaited intelligence report on the killing of Saudi journalist and Washington Post contributing columnist Jamal Khashoggi. The report’s assessment that Crown Prince Mohammed bin Salman personally ordered the assassination triggered an immediate backlash from Saudi-affiliated Twitter accounts.

The prevailing narrative about Saudi Twitter is that much of the discourse is manufactured by “bots” — inauthentic, centrally controlled accounts. The Saudi regime has invested substantial resources over the past decade to turn Twitter into a tool of control. In February, Marc Owen Jones and other researchers quickly identified thousands of bots amplifying pro-Mohammed hashtags while attacking the Khashoggi report.

There’s something else going on, though — the Twitter barrage isn’t just the result of electronic bot armies. Our research points to the critical role of real-life influencers who prompt what appear to be authentic pro-government rallies of real users inside Saudi Arabia. While pro-government bots are indeed ubiquitous, humans are still the bedrock of “digital authoritarianism.” Even if Twitter were able to identify and eliminate all Saudi bots, our research suggests that pro-government mobilization would still be substantial.

Saudi Arabia just lifted Qatar’s 43-month blockade. How did this rift end?

Twitter in Saudi Arabia a “parliament of the people”?

While Saudi Arabia severely restricts opportunities for political debate, the rapid adoption of Twitter by Saudi citizens raised hopes that it could serve as a kind of “parliament of the people” — a relatively open forum where citizens could share political views, criticize the government and even solve collective action problems. While the regime initially took a more lenient approach to online critics, researchers document a growing number of crackdowns and efforts to shape Twitter discourse directly.

Internet watchdogs at Oxford University and Bellingcat, as well as academic researchers, have noted the regime’s use of inauthentic coordinated networks (bots) to manipulate social media, while Twitter has suspended accounts based in Saudi Arabia for the same reason. These bot accounts harass government critics, boost pro-government narratives and fabricate a sense of popular support in Saudi Arabia and elsewhere in the world. Before his death, Khashoggi even talked with Saudi dissident Omar Abdulaziz about countering these pro-regime “flies” with a comparable army of opposition “bees.”

The fascination with bots can hide the role of “influencers” — high-profile users whose power has been recognized in other online communities, such as far-right extremists in the United States. Many top Saudi influencers appear to be embedded within Saudi government institutions or major media outlets, while others have built sizable Twitter followings out of tweets praising Saudi leadership and attacking the kingdom’s enemies. Some no doubt act at the direction of the Saudi government, but this is far from clear in other cases.

How we researched what’s happening on Twitter

To assess the relative role of bots and Saudi influencers, we downloaded a number of politicized hashtags relevant to Saudi Arabia in near-real time. We focused primarily on pro-government mobilization, such as a wave of Twitter comments that tarred women’s rights activists as “agents of the [foreign] embassies” following their arrest in May 2018 (over 370,000 tweets).

For comparative purposes, we also collected tweets from mobilization against government institutions (such as conservative backlash to a government-hosted Russian circus) and tweets related to sudden, high-profile news events such as a rocket attack on Riyadh in 2017 (as a benchmark for high-volume, spontaneous tweeting).

For a broader perspective on bot prevalence, we tracked 276 trending Arabic-language political hashtags during the eventful October 2019 to January 2020 period, including uprisings in Lebanon, Iraq and Iran, as well as the assassination of Iranian military leader Qasem Soleimani.

Will Soleimani’s death change Iran’s relationship with the Revolutionary Guard?

Bots: a ubiquitous minority

Inauthentic, coordinated networks are ubiquitous on Middle East Twitter, but our broad sweep of political hashtags suggests that bots are more prevalent on more marginal hashtags. By our estimates, typically only 5 to 9 percent of users on political hashtags in the region are bots. In separate analysis of the largest hashtag reacting to the Khashoggi report, #WeAreAllMohammedBinSalman (in Arabic), we find that no more than 4 percent of accounts — around 8,600 — belonged to coordinated networks, out of a quarter-million total participants. That figure is in line with Twitter’s suspension of just 3,500 accounts in the wake of this campaign.

When we compared the “agents of [foreign] embassies” campaign to other 2018 campaigns, those participating were less likely to be suspicious accounts (i.e., a bot account, or an account then flagged by Twitter for suspension) than those participating in the “grassroots” campaign against the Russian circus. Likewise, tweets attacking the activists came no more fast-and-furious than those following the rocket attack on Riyadh. These findings suggest little evidence of bots acting at superhuman speed.

Are influence campaigns trolling your social media feeds?

Inequality and intimidation are a bigger problem

Two other dimensions shape Saudi Twitter: inequality and intimidation. Hashtags are extraordinarily unequal, with just a handful of participants garnering the lion’s share of retweets. The #WeAreAllMohammedBinSalman hashtag showed this same pattern, with just over 1 percent of users — a who’s who of pro-government Saudi Twitter — garnering 80 percent of all retweets.

Inequality in turn facilitates intimidation, with influencers becoming the primary targets of coercion or co-option. Besides making anti-regime social media posts a punishable offense, the Saudi government in recent years reportedly hacked the phones of opposition activists like Omar Abdulaziz, and recruited spies within Twitter to obtain private data on prominent users. In 2017, Khashoggi already believed these moves had a chilling effect on the willingness of Saudi individuals to publicly voice critical views — leaving behind mostly loyalist Twitter traffic.

Our research suggests regimes like the Saudi monarchy derive their narrative power from prominent users who engage in the cut-and-thrust of the “Internet of beefs” — not just from anonymous bots that buzz around the margins. And this, ultimately, may be the deeper deception of Saudi Twitter: behind the apparent scale and automation, the political discourse hinges on the unique emotions, ambitions, creativities and even cruelties of a small number of influencers.

Professors: Check out TMC’s expanding list of classroom topic guides.

Andrew Leber is a PhD candidate at Harvard University’s department of government. Find him on Twitter @AndrewMLeber.

Alexei Abrahams is a data scientist with the Technology and Social Change (TaSC) Project at Harvard University’s Shorenstein Center. Find him on Twitter @abulkhaezuran.