The Clash of Ideologies on Persian Twitter

The Clash of Ideologies on Persian Twitter
June 2021
Pooya Azadi
Mohsen B. Mesgaran
Stanford Iran 2040 Project

Download Full Paper

In this paper, we provide a data-driven analysis of the Persian Twittersphere from a political perspective to demonstrate how the balance of power in the battle of ideas and ideologies has evolved over time, both in numeric terms and qualitatively. We employed an account selection method that ensured the inclusion of virtually all high-follower accounts (HFA)—otherwise referred to as influencers—and a representative sample of active ordinary accounts. A supervised machine learning model was developed to categorize users into three broad political groups: dissidents, neutrals (or nonpolitical), and pro-regime (encompassing the Islamic Republic’s officials, reformists, cyberbasij army, and the regime’s supporters abroad).

The analysis shows that, in rough terms, tweets in Farsi constitute 0.5% of all tweets currently sent daily across the world—a figure which is half the share of Iranians from the global population. A cross-language comparison of the trends in the number of tweets throughout the day suggests that an overwhelming majority of the Iranian users of Twitter are in Iran despite the technical and legal obstacles for using the platform. In addition to the sheer disparities between the audience size of the HFA and ordinary accounts, there exists a substantial degree of inequality within each of the two subsets as well (e.g., the top 10% of HFA accounts receive half of the total attention, corresponding to a Gini index of 0.57). We find that dissidents make up 23% of HFA and 7% of ordinary users while these shares for the pro-regime accounts are 34% and 11%, respectively. The total number of followers of dissidents’ accounts among HFA has expanded by an average of 38% per year, reaching 30 million by the time of this writing in 2021. On average, ­the pro-regime accounts are not only newer than those of dissidents but also enjoy a faster growth (upwards of 50%) in their audience. However, as we argue in the paper—contrary to the rosy picture that the analysis based on share of followers depicts for the pro-regime ideologies—both the actual trends in the public opinion of Iranians and the impact of Twitter as a communication platform have been unfavorable for the Islamic Republic regime and its supporters.

The prevalence of political astroturfing (coordinated practices intended to create a false impression of support from a widespread grassroot movement while masking the organization behind the effort) by the cyberbasij army artificially inflate quantitative measures of influence in favor of the regime. At the time of this writing, more than half of the pro-regime accounts are less than one year old while each of these accounts, on average, follows three times more users than dissidents’ accounts. In contrary, some risk-averse users among dissidents may not follow other dissident accounts to avoid unpredictable consequences that such action may cause and hence, classified as neutral by the model. Also, the apparent faster growth in followers of the pro-regime accounts is overestimated by the implicit assumption in our analysis that there has been no change in the political position of each individual user while, in reality, there is a considerable amount of anecdotal evidence that many who are now openly seeking regime change were advocates of gradual reforms in the past. This claim is supported by a sentiment analysis conducted on a selected number of figures and concepts associated with the reform movement which shows a significant shift toward negative expressions over the past decade.

To better describe the political landscape on Persian Twitter, we use a conceptual framework in which political groups are mapped against a two-dimensional matrix based on their degree of internal coordination and the scope of institutional change relative to the status quo which they advocate. In this framework, the highly organized cyberbasij, whose primary function is to suppress other agents of change, and the scattered dissidents among ordinary citizens with no political affiliations represent the two ends of the spectrum. We argue that it is the latter group who benefits immensely from Twitter through its indispensable role in the spread of ideas and information. In contrast, Twitter neither expands the regime’s communication bandwidth nor serves as a platform for new recruitment, regardless of the real or inflated share of the pro-regime accounts in the overall audience pool.

An informal review of the ordinary pro-regime users reveals that a significant share of these accounts do not seem to use their real identification despite lack of any threat from the regime, suggesting at least some degree of coordination, if not direct control, by organizations in the regime. Our data also show that the modus operandi of these accounts, surprisingly, mainly consists of activities that promote like-minded people and bots in their echo chamber (or botnet) rather than confrontation with dissidents. For example, replies to the tweets from dissidents constitute only 0.5% of the total replies by the pro-regime users, whereas replies to pro-regime tweets make up 5.1% of the replies by the opposition accounts. By and large, the coordinated accounts of the cyberbasij army can be seen as packs of wolves whose main function is to gain back the narrative through artificially boosting the visible numerical metrics of popularity and influence (e.g., number of favorites and retweets) rather than directly attacking the fort of what they refer to as the vicious enemy. It seems that the only truly beneficial use of Twitter for the pro-regime group is the activity of a small number (~50 users, 1% of HFA) of regime supporters among the diaspora who play an important role in defending the regime abroad.