Tunisia under the spell of algorithms

Algorithms, the invisible entities of politics

The question of the effects of this embedding goes far beyond the libertarian utopia that, at the turn of the 2000s, celebrated the horizon of the Internet: a horizontal, egalitarian communication, freed from censorship, transcending borders. Digital platforms have established a near-total monopoly over this new civic space. Their economic model rests on the massive collection of user data, in order to feed algorithms whose primary purpose is to build profiles that allow them to select content that captures attention, drives interaction, and prolongs the time spent on social networks, so as to maximize exposure to targeted advertising messages.

The formula of these algorithms is so strategic that it is kept secret, protected as a trade secret behind a wall of confidentiality. And as the time spent in front of screens increases, our perception of the world and our mutual relationships are increasingly shaped by this "genie," invisible yet omnipresent. Algorithms helped shape the course of the events that began in December 2010 in Tunisia.

When Mohamed Bouazizi set himself on fire on 17 December 2010, and protests erupted in the wake before the governorate of Sidi Bouzid, dozens of phones captured images of the popular reaction and then of the police crackdown. A new generation of cyber-activists quickly spread the videos on Facebook, which Al Jazeera picked up and rebroadcast. Those images stirred the embers of a whole generation's latent anger; the uprising broke out and spread like wildfire. Then came the footage of the wounded and the dead at Kasserine hospital on 6 January 2011, producing a turning point in the public's conscience: already seen as corrupt, Ben Ali's regime now appeared bloody. At the same time, a number of cyber-activists launched the rumor of "the general who said no," according to which the army's chief of land staff, General Rachid Ammar, had refused to order soldiers to fire on demonstrators. Entirely unfounded, the rumor seemed plausible and convinced Tunisians that they had nothing to fear from an army that would stand on their side against the security forces. The idea even spread through the ranks of the military institution itself and redrew the protagonists' expectations, in a chain of events whose conclusion was the irreversible flight of Zine el-Abidine Ben Ali⁽²⁾.

There had already been talk of a "Twitter revolution" in the Moldovan and Iranian uprisings of 2009. Tunisia, for its part, is said to have had a full-fledged "Facebook revolution." And while it is utterly naive to see Facebook as the cause of the Tunisian revolution's success ⁽³⁾, the role of social networks in it is undeniable ⁽⁴⁾. By breaking censorship, circulating information, easing individual isolation, spreading slogans ("the people want"), and encouraging people to step outside the circle of fear… all virtues then celebrated with enthusiasm. These networks appeared as new spaces of expression and citizenship, transcending the limits of the traditional frameworks of political action—parties and unions—and the renewal they carried promised a renewal of collective action at a moment when the grand narrative of Western democracy, shaken by the Iraqi quagmire, seemed to be regaining its luster.

Fifteen years later, the facts have refuted that naive vision with striking clarity. The social networks that were promised to undermine the foundations of authoritarian regimes, to become an engine of mobilization, a vector of transparency and informational pluralism, and a space of equality, have produced a genuine informational disorder exploited for the purposes of mass manipulation.

This is due, on the one hand, to the nature of algorithms designed to favor divisive, provocative, and sensationalist content; and, on the other—and this is the most important—to the fact that states and other actors have seized upon these tools amid a staggering disproportion in technical and financial means. Sometimes with the explicit support of the digital platforms themselves, which, far from the Internet's original neutrality, have their own ideological orientations. We have seen the X platform's commitment to supporting the Trumpist "MAGA" movement, as well as Meta's tendency to comply with Israeli demands to bury Palestinian content or delete particular accounts. The manipulation potential offered by social networks has given birth to a sprawling and powerful influence industry.

This new sector has been taken up by private companies, sometimes connected to the world of intelligence. The art of these manipulation engineers consists in deploying the analytical capabilities of algorithms—originally designed to target advertising messages—in order to identify the targets of the propaganda they are tasked with promoting, and to boost the visibility of falsified or steered content by artificially creating an effect of mass.

To carry out their manipulation strategy, they have developed an integrated operational infrastructure ⁽⁵⁾. Foremost among these are the "troll farms," whose organization recalls that of call centers: under the supervision of a manager, operators—most often highly qualified young people finding an outlet for work amid structural unemployment ⁽⁶⁾—run a large number of fake accounts or fake Facebook pages, to comment on posts, spread doctored images and clips, respond to the posts of designated profiles, disparage targeted opponents through hundreds of false accounts by broadcasting defamatory accusations, and systematically report content in order to force the platforms to downrank or delete it. In addition to these "real-fake" accounts, troll farms can deploy "bots"—programs that automatically generate or relay messages under the guise of genuine human profiles ⁽⁷⁾. To this arsenal may be added "fake news" sites disguised as news outlets, and quasi-academic publications, allowing the diffusion of distorted narratives and the subjection of artificial-intelligence results to them.

Independent users complete this structure: either "mercenaries," like the influencers recruited by Israel's National Public Diplomacy Directorate (the Hasbara), via private agencies, to relay its propaganda in exchange for generous remuneration ⁽⁸⁾; or opportunistic activists working in coordination with states or parties; or independent believers who echo the rhetoric of the regime or party they support. The message carried by this artificially amplified flow—whose flame the algorithms automatically fan—turns into an irrefutable truth and an obligatory theme of public debate, in an echo chamber that literally swallows the public whole ⁽⁹⁾.

The recourse to Cambridge Analytica—which had developed its own algorithms from the data of millions of Facebook accounts to support the Brexit and Trump campaigns in 2016—was the first intervention that became a benchmark for this industry. The Arab world followed suit, particularly in Saudi Arabia and the Emirates ⁽¹⁰⁾. These authoritarian regimes discovered in social networks a formidably effective tool to ensure their political survival, break the impulses of democratic uprising in the region, and maximize their regional influence ⁽¹¹⁾. The researcher Marc Owen Jones described these two states as having become veritable "exporters of disinformation" ⁽¹²⁾.

Although Tunisia lies on the periphery of this regional epicenter of manipulation, it has not been spared this transformation. The deleterious effects of social networks and algorithms have weighed on its political trajectory since 2011.

In the first months after the fall of the dictatorship, the public sphere thawed after decades of censorship, particularly on the identity question inflamed by the attempt at authoritarian "modernization" pursued since independence. In that fluid moment of social norms, the Islamists sought to shift the lines toward greater Islamization, while the Ennahda movement's integration into the political scene and the emergence of jihadist Salafist groups awakened old fears. The sore losers of the revolution spread planted and exaggerated information (such as the rumor of cholera's appearance in Tunisia in the summer of 2012) to discredit the democratic process and stir nostalgia for strong rule. Facebook and Twitter were the echo chambers of this ferment.

The great stakes raised by the revolution—the economic model, the modalities of social and territorial redistribution, the reform of security and judicial institutions—were thus eclipsed in this ceaseless accumulation of polemics that consumed the first three years of the transitional period. Instead of being a shared space for deliberation and the rationalization of debate, the social networks instead inflamed questions that warranted calm reflection, and fragmented the public into mutually deaf compartments. This polarization brought the transition process to the brink of collapse in the summer of 2013, before it was relaunched by a compromise between two major political forces: Nidaa Tounes and Ennahda.

Yet in the following years, the political forces went on feeding this identity controversy in order to distinguish themselves and mobilize their respective electorates, while spending most of the first legislature (2014–2019) in inter-apparatus and behind-the-scenes disputes over the division of power. These wasted years alienated the Tunisian majority from political life, discredited the parties, wore out the very idea of democracy, and deepened resentment toward the "rentiers" of the transition.

On this fertile ground, the presidential election of November 2019 marked the effective entry of the influence industry into Tunisia. As early as January 2019, Facebook pages began multiplying to expose the corruption and incompetence of the political class: "The parasites of Tunisia," "Anyone but Chahed" (Youssef Chahed being the head of government at the time and a presidential candidate), "Fed up with politics and politicians"… Strikingly, eleven such pages share a single exception: Nabil Karoui, a media magnate (founder of the Nessma television channel in Ben Ali's era), leader of the "Qalb Tounes" movement and presidential candidate. Following a Facebook investigation, it emerged that these pages were linked to an Israeli company, "Archimedes Group," active in several African electoral campaigns ⁽¹³⁾. On 15 May 2019, Facebook shut down 265 accounts linked to it across thirteen countries, including the Tunisian pages. In a record and unprecedented span of time, they had amassed 500,000 subscriptions, posted 36 clips viewed 7.8 million times, and their 359 posts had generated more than a million interactions ⁽¹⁴⁾.

In June 2020, an investigation by the DFRLab revealed that another company specializing in cyber-influence, "UReputation," owned by a Franco-Tunisian businessman, Lotfi Bel Hadj, had conducted an operation dubbed "Operation Carthage" to support Nabil Karoui's candidacy, relying on a fact-checking site, "Fake News Checking," and fake Facebook and Twitter accounts ⁽¹⁵⁾. It turned out that its activity extended across the entire African continent. On 5 June 2020, Facebook shut down hundreds of accounts, pages, and groups created by the company.

It is difficult to measure the impact of these operations on Karoui's result. They did not enable him to win, but they propelled him to the front rank and cemented his place in the electoral race as a genuine contender for Carthage. The actual winner was an outsider: Kaïs Saïed, who masterfully capitalized on the frustrations of the transition and won the second round with 73% of the vote. No one was able to establish whether a company, a faction within the Tunisian state, or a foreign state lay behind this success, but his use of social networks owed nothing to chance. His style was striking and matched the image he wished to project: a man of the people answering the citizens' call. He was the only one without an official page or website; yet the Court of Audit's report after the elections identified at least 30 "independent" support pages run by 120 administrators, totaling three million subscribers (i.e., five times more than Karoui's support pages!).

After his election, Saïed appeared to lose the initiative as the social, financial, and health situation deteriorated at speed. In reality, he was letting time work in his favor. At the deadliest and most chaotic point of the epidemic waves, in July 2021 ⁽¹⁶⁾, a campaign took off on social media—through Facebook pages, Twitter accounts, and hashtags all carrying a single message: the corruption of politicians—demanding that the government and parliament be "ousted" and calling on Kaïs Saïed to act. The campaign targeted in particular the head of government Hichem Mechichi, and even more so the president of Ennahda and Speaker of Parliament, Rached Ghannouchi. A page titled "No to compensation for the Nahdhaouis" gathered 700,000 subscribers. The Islamist party's leadership had indeed, at a most inopportune moment amid acute crisis and rising popular hostility, called on its members to demand the payment of the indemnities mandated by transitional-justice rulings. The announcement of a Facebook event titled "The Revolution of 25 July: youth reclaim power" garnered 52,000 interactions.

The demonstrations of 25 July in dozens of cities echoed the slogans ignited by social media during the preceding two weeks: "The people want the dissolution of parliament," "Mechichi out! Ghannouchi out!"… In the capital, a thousand people gathered before the parliament building, while several Ennahda offices were attacked across the country. Then came Saïed's announcement, at the end of the day, of his activation of Article 80 of the Constitution, his assumption of all power, the freezing of parliament, and the dismissal of the government—appearing as a response to the people's call and to the scenes of violence.

Researcher Marc Owen Jones's investigation established the role of the Gulf monarchies in glorifying Saïed's coup: "Shortly after Saïed's decision, the hashtag 'Tunisians revolt against the Brotherhood' gained accelerating momentum on Twitter. An analysis of 12,000 tweets from 6,800 accounts suggests that this was a coordinated influence campaign launched mainly from the Gulf, aiming to portray the president's decisions as a Tunisian popular revolt against Ennahda, and to justify the return to authoritarianism under the pretext of overthrowing it. The ten most influential accounts on the hashtag were based in Saudi Arabia or the Emirates." Fake accounts programmed to retweet these Gulf influencers ensured maximum reach in Tunisia ⁽¹⁷⁾.

As power concentrated in Saïed's hands, the president became the sole and final decision-maker on all important matters, and the algorithms became a central instrument in managing them: mobilizing Internet users in preparation for radical measures and subsequently justifying them—such as the dissolution of the Supreme Judicial Council in February 2022, or the dissolution of parliament when deputies announced their intention to hold an online session on 30 March 2022. These voices now demand the banning of civil-society organizations and of foreign funding—something effectively being applied on the ground.

Day after day, the algorithms present to the watching public the figure of the "corrupt," the "profiteers," the "unpatriotic" officials, and the "traitors" beholden to foreign interests—the recurring descriptors of Saïed's discourse. The digital attacks often precede arrest campaigns. Conversely, on the pages of the opposition and the accounts supporting political prisoners, every post is flooded with millions of mocking emojis posted from fake accounts that support the president (with almost no friends and no posts), often bearing improbable origins (Sri Lanka, Brazil…).

Since late 2022, a new obsession has invaded the Tunisian digital scene: claims of a conspiracy to settle sub-Saharan migrants, who are reduced by a new acronym that turns them into a single bloc—the "ajassiyn"—a term popularized on social media. It all began with a hashtag, #Tunisian_nationalism, launched by a fringe political party, the "Tunisian Nationalist Party" (PNT), on its TikTok account on 18 June 2022. Through it, the party promotes clips containing hate speech and lies that incite violence against migrants, and claims—based on out-of-context quotations—the existence of a project to conquer Tunisia by Black Africans. To bolster its claims, the party published a report at the end of 2022 that it submitted to the head of state. As the campaign swelled on social media, the president—strikingly attentive to Facebook polemics—took up the report's content and reworked it in his speech of 21 February 2023, claiming that the presence of sub-Saharan migrants in Tunisia was nothing but the fruit of a conspiracy aimed at altering its demographic composition and erasing its Arab-Islamic identity ⁽¹⁸⁾.

The very next day, groups of citizens formed to hunt down migrants in the streets and drive them out of their homes. From 28 January to 27 February, the party's TikTok account was viewed 11 million times, 9 million of them in the days following Saïed's statement. "We rejoice at gathering 500 people for a march against racism, while every clip by the party racks up two million views!" laments one of the organizers of a protest march held in Tunis a few days later. The wave of violence subsided after two weeks, but the subject continues to live on social media. The associations and figures defending migrants' rights—most of them behind bars—are threatened online.

One of the most active pages on this subject ("Resistance to African colonization" — with 110,000 followers). Its reels and visual content are viewed thousands, even tens of thousands, of times. Its latest target is the Arab Institute for Human Rights, which it accuses of having defended the adoption of an asylum law in Tunisia. The page's administrator sees in this proof of the conspiracy and its source, convinced that he can discern the outline of the Israeli flag in the organization's logo. The page's profile picture is a portrait of Hannibal, the Carthaginian general, a recurring visual for several years now and emblematic of this new Tunisian nativism. This persistent campaign now has its own dynamic, but the question of whether it is orchestrated and funded to produce and amplify content remains unanswered to this day. The fact is that this discourse serves a power that reaps political benefits from the European Union, and from Italy in particular, on account of its role as a "safety valve" against sub-Saharan migration flows toward Europe.

The immigration file may be seen as the product of Europe's exporting of its own crisis in managing migration. With the cross-border circulation of digital content, Tunisia has also imported the tropes of the European identity-based and xenophobic backlash, such as the "great replacement," the dread of morbid contagion, as well as the linking of immigration to unemployment and crime, and so on. But this fever also crystallizes affects recurrent since 2010: the sense of lost sovereignty and of identity eroded under the weight of external factors, the fear of a European (even Israeli) conspiracy to enslave the nation, the desire to reassert a "true people" in the face of the threat of identity dissolution, the need for authority and order, and so forth.

as Giuliano da Empoli notes ⁽¹⁹⁾. Tunisia is a living illustration of this. At its core, two phenomena combine. The first is a political "disintermediation," the consequence on the one hand of the crisis of trust in parties, unions, traditional media, and the mechanisms of representative government, which strip citizens of their agency and tend to confine political life to the hands of an elite; and on the other hand, the consequence of the executive's desire to apply its vision without having to negotiate with counter-powers, perceived as interest groups. The second is a cruel paradox: that this space of freedom and of the diffusion of independent, critical speech, opened up by the Internet, is in the hands of private companies whose profitability rests on mechanisms that distort public discourse, elevating the content that magnetizes debates by exploiting affects.

This combination produces the conditions conducive to the emergence of an "authoritarian populism." Populism, before being an exogenous evil from outside, is first the product of a demand to reclaim popular sovereignty when the vote comes to seem a ritual without power and parties micro-enterprises in the service of the ambition of a few.

Social networks offer a response within everyone's reach to this feeling of dispossession: "The algorithms devised by the engineers of chaos give everyone the impression of being at the heart of a historic uprising, and of having at last become an actor in a history they thought they were condemned to live through passively," holds da Empoli. Posting stories, commenting, reacting, addressing remote figures directly—the kind of engagement that social networks allow is certainly within everyone's reach, but it is an engagement atomized in its practice and massified in its capacities for manipulation. Most of the time, it is a virtual, if not fictitious, participation, steered or co-opted toward ends decided by a minority.

The "people" that the participants in this exchange flow believe they constitute through these algorithmic communities is a virtual reality, in the absence of a framework for collective action and a space for the exercise of citizenship in which the people manifests itself politically and organically. The absence of structured social and intellectual frameworks creates a climate conducive to the spread of viral information without its encountering any antibodies. Informational contaminations (rumors, collective panics) have always played a role in history, but with the ubiquity and instantaneity of social networks and the propensity of algorithms to amplify polemics, a threshold has been crossed, as Yves Citton reminds us. Public debate tends to become "a muddy brawling ground from which every elementary rule of civility and good faith has been banished" ⁽²⁰⁾.

In reality, from the multitude worked upon by these affects—artificially created or distorted by digital virality—it hears only what serves its designs and visions. But this discourse, beyond its judicial effects, sustains nothing but its own virality on the networks. Politics locked within the loop of algorithms produces a power that grasps of reality only the image it fashions of itself, and a society that the digital viralities now represent only in distorted fashion, growing ever more external to this discursive bubble.

Tunisia does not hold the monopoly on this configuration. It is spreading, even into long-established democracies worked upon by the same democratic fatigue, the same identity-based withdrawals, where the state of security and economic exception tends to become the norm.

As during the outbreak of the Arab revolutions, the role of technology should not be overstated. There remains space and resources for the forces opposing this wave. The recent electoral defeats of Viktor Orbán in Hungary and of Giorgia Meloni in her Italian referendum, and the ferocity of the resistance against Trump in the United States, give hope that the public still retains the capacity to break free from the spell of algorithms.

But the horizon is hardly reassuring. The alliance between populisms and technology may be only a turbulent phase of a far more formidable alliance. States, democratic or not, and digital platforms share a single passion: amassing as much information as possible. Everything has been designed to make easy and comfortable the use of tools that record the slightest trace of our activities. Algorithms coupled with artificial intelligences are acquiring the capacity to make decisions concerning us without human intervention, and we have seen in Gaza that this extreme has already become a tangible reality. For the major digital platforms, organically bound to the interests of the great powers and to the extractive, militaristic order, appear utterly remote from the libertarian utopias of their beginnings. The horizon that a world governed by these companies' algorithms opens up goes entirely beyond authoritarian populism: it approaches what is called—for want of a more precise term—techno-fascism ⁽²¹⁾.