The debate over censorship in social networks, a double-edged sword

President Andrés Manuel López Obrador in the National Palace.Hector Vivas / Getty

On the night of July 1, 2018, when he went out to celebrate his triumph in the presidential elections, Andrés Manuel López Obrador mentioned his gratitude to the “blessed social networks” that had allowed him to communicate with his voters.

Two and a half years later, in January 2021, the Mexican president said that platforms such as Twitter had become a kind of “Holy Inquisition” against freedom of expression and that it is not possible for private companies to decide on the right to broadcast messages .

Between one moment and another, the only thing that seems to have remained intact is the sacred character that the president attributes to the networks: from blessed to holy inquisitors, large technological platforms have become a problem for the world's democracies and a concern for political leaders, who look at them with benevolence or suspicion depending on the context and the electoral situation in which they find themselves.

The recent offensive by the Mexican government against the networks began in early January, when Twitter and Facebook decided to suspend Donald Trump's accounts for inciting violence on Capitol Hill.

López Obrador described the measure as censorship and said that he would order the study of a plan to create a national social network, in order to "guarantee communication and freedom of expression."

What began as a sovereign crusade, however, was leading to an arm wrestling from which it seems impossible to remove the horizon of midterm elections next June: at the end of January, the president pointed out to the director of public policies of Twitter for being "sympathetic or very close militant ”of the National Action Party, said that he expected him to do“ his job in a professional way ”and not to promote“ the creation of bot farms ”.

The following days, Twitter blocked content or suspended various users, including three accounts related to the Government with tens of thousands of followers, on the grounds that they violated its rules of use on

spam

.

The president of the ruling party, Mario Delgado, said that these users had been censored.

"I'm worried about that speech of 'as long as they favor me, the networks are wonderful, but if they don't they are horrible.'

This speech is very worrying in the face of the elections ”, says Rossana Reguillo, researcher, teacher and coordinator of Signa Lab, a laboratory of the Jesuit University of Guadalajara that has analyzed the #RedAMLOve network, a network of followers on Twitter that amplifies the López Obrador's speech.

Many users have passed through the #RedAMLOve but also

automated

bots

to give more echo to the president's attacks on the press, which he has described several times as bourgeois, malicious, or corrupt.

"During January and February 2019 they became several times trends that monopolized much of the online discussion," says one of the studies on the network.

But the López Obrador government, Reguillo points out, has lost influence in networks in recent months.

“In the laboratory we have been studying this phenomenon since January 2019, and what we have seen is how during 2019 and the first third of 2020, the so-called 4T or #RedAMLOve won in narrative, they overwhelmed.

This has decreased, so we think that either it was initially artificial or financing for this ended ”.

Concern about the growing power of the networks, especially in electoral times, is an issue that reveals experts and activists all over the planet.

“This is a problem where the government has totally lost power.

He lost control and feels defenseless and he is not the only one who feels that way, ”says Guatemalan lawyer Renata Ávila, a specialist in technology and human rights and a member of the legal team that defends Julian Assange and Wikileaks.

The same concern raised by the Mexican government can currently be heard in countries like Germany or France, Avila assures, and points out that this situation stems from a problem common to all: the negligence and comfort with which governments of any ideological orientation left in the hands private “the critical digital infrastructure with which they communicate with their citizens”.

For this reason, Ávila was excited when he read that López Obrador proposed to develop a national social network in a country like Mexico, "which has all the technical capacity to carry out a project of this magnitude," he says, as long as there is political will and investment. .

But as the days go by, the statements about the creation of its own platform have disappeared from the official discourse and have given rise to a regulation project led by the ruling Senator Ricardo Monreal, who seems to seek a quick solution to a complex problem.

"The networks use public inputs such as the radioelectric space or the fiber optic of the Internet and we must guarantee that the right to say what you say is respected, they do not censor you," Monreal told EL PAÍS, and assured that the regulation (a modification to the Federal Telecommunications and Broadcasting Law) would be ready in the next three months.

Before the June 2021 elections.

Failed experiments and real problems

Last week, Senator Monreal explained in an interview that his proposal essentially consisted of a constitutional body deciding “whether content or an account affects social stability or calls for illegality”, and not “the owners of the networks ”.

For that, he said, they proposed to expand the powers of the Federal Telecommunications Institute (IFT), responsible for controlling radio and television, and anticipated that this week he would speak with research institutes and academics.

However, just four days later, the legislator published a text with his proposed regulation.

According to this project, social networks with more than a million users —that is, not only giants like Facebook but others much smaller—, which are defined as “relevant social networks”, would need authorization from the IFT to provide their service in Mexico and, in addition, an authorization from the same body to establish "the terms and conditions" of its service.

The text of the initiative indicates that the networks must ensure that "the spread of false news" or the "spread of hate messages" is avoided, two expressions whose vagueness has alerted the experts.

"This is something that the different rapporteurs for freedom of expression have repeatedly pointed out: we cannot have such broad, vague terms, without any kind of clarity," Vladimir Cortés, official of the digital rights program, told El PAÍS. the organization Article 19. This ambiguity, Cortés warns, may lead to the elimination of "legitimate messages that could be categorized as hate messages, or criticism, or dissent, or journalistic investigations that are classified as false news."

For Luis Fernando García, director of the Network in Defense of digital rights in Mexico, this is the most dangerous point of the initiative: "It is a covert censorship mechanism," he wrote on Monday night, shortly after Monreal released the text of your initiative.

According to defenders of freedom of expression who have followed the bill, the legislator seems not to have considered the profound complexity that this type of regulation would imply.

“No democratic country requires authorization from a platform or an internet page to be able to disseminate information on the internet.

Freedom of expression is clearly unconstitutional and violent, ”says García.

The project released by Monreal also establishes that, in the event that a network decides to close an account or delete content, and the same network does not resolve its user's claim in less than 24 hours, the user can go to the IFT to "File the corresponding complaint."

This key point on content regulation assumes that all networks have a centralized complaints system, such as Facebook or Twitter, without considering networks such as Wikipedia in which moderation works in a more decentralized way.

"There is a lack of knowledge about both the digital space and internet governance," says Cortés, regarding the limited control that IFT may have over global networks.

Like other experts consulted, he has not seen that there is a political party that is actively censoring a particular group on the networks, but he is concerned about the lack of transparency of companies when deciding when to close an account or not.

"But let's be very clear: the decision whether to suspend an account should not rest with the state."

Being clear that the State is not the best arbiter of the debate on social networks does not solve the fundamental problem of how to build a healthier digital debate that is not plagued by disinformation, manipulations or hate speech.

International regulation on freedom of expression is very clear about what content is not debatable, such as child pornography or xenophobic discourses that promote the murder of minorities.

But it is less easy to define what is misinformation or what is

bullying

, and control it in millions of users around the world who post every second.

"The experts in this battle have been at least since 2016 trying to describe what is

fake news

, what is disinformation, and we have not been able to" says Cristina Tardáguila, deputy director of the International Information Verification Network (IFCN, for its acronym in English). ), the network that brings together 80 organizations that verify information around the world.

"Either we fall short, with a simple definition that reveals a lot of cases, or a very long definition that puts humor, art, or jokes at risk."

To resolve the dilemma, the countries that give the Executive more power may end up criminalizing users who share information that they consider false, or contrary to the interests of the Government.

This is the case of Nicaragua, where a cybercrime law approved in 2020 allows penalizing - with fines or imprisonment - users who publish information that generates "anxiety in the population", or puts at risk "economic stability, public order , public health or sovereign security ”.

Other cases of content moderation in Latin America seem more interesting experiences.

The Congress of Brazil, for example, approved in 2014 an Internet Civil Framework that gives more legal security to users and platforms.

"The keystone of this model is that a person cannot be sued for the content he publishes, and the platform cannot be sued if he has not previously received a court order to remove content," the director of the organization explained from São Paulo InternetLab, Francisco Brito Cruz.

"The platform is only legally responsible if it did not remove content after a court order ordering it, after studying the case."

One risk of this exit is that judges may end up inundated with lawsuits, many times more from political actors than from citizens.

The digital company CTRL + X, which tries to follow the trail of these litigation in Brazil, has identified that almost three thousand of 5,242 legal actions are made by politicians.

Of the total, 77% of the cases are a defamation charge.

"In the last five years we have reached record numbers of petitions in the courts to remove content, and especially many government petitions," says Britto Cruz.

There is also the risk that judges will be co-opted by a political party, and that they will fall back into the initial problem: that it is a party that decides what deserves to be censored, and what does not.

Since 2018, the International Information Verification Network has been monitoring regulatory attempts in more than 50 countries, and has been developing a guide based on these cases.

"Based on this guide, which has been in place for three years," says Tardáguila, "I can guarantee you that there is no experience that comes from laws that have resulted in success in reducing misinformation."

Responsibility of networks

After the 2016 presidential elections in the United States, in which disinformation campaigns on social networks played a central role, platforms such as Twitter or Facebook have tried to moderate their content with algorithms that allow identifying what content may be offensive or violent.

But the use of artificial intelligence has also led to conflicts and cases of content suppression reported by users.

In October 2020, for example, Facebook deleted a campaign against breast cancer in Brazil because the photographs showed female nipples and the algorithm identified that they violated “Facebook's community norm on nudity and adult sexual activity” (after a process appeal, the content was restored).

One of the strategies developed by Facebook to respond to the criticism was to create a Content Advisory Council: a kind of Supreme Court, independent of the company, in which its

magistrates

rule in emblematic cases in which content has been removed from the platform .

Last month they published their decision in the first five cases (a post with hate speech against the Azerbaijani population, another with possible misinformation about cures against the coronavirus, or the case of breast cancer in Brazil).

"These were cases in which it was very important to check whether Facebook was properly applying international human rights law," one of the members of the Advisory Council, Colombian lawyer Catalina Botero, a specialist in freedom of expression, told EL PAÍS. .

"They allowed us to identify if the way Facebook was doing content moderation was adequate, or if it merited a change not only in the specific decision, but also in the policy."

For each case, the Advisory Council decides whether or not it was appropriate to delete the post, and Facebook is obliged to keep it deleted or to restore it.

But the company is not obliged to change its usage policies according to the recommendations given by the Council, and that opaque control that the social network continues to have over the rules of the game to maintain a post or not (or on how its algorithm works that privileges one information over another) has led many to think that networks should be regulated as media that make editorial decisions.

"I think we lose the democratizing potential of social networks," says Botero about this possibility, that of regulating networks as media.

“When a platform has given 3,000 million people the opportunity to interconnect, to tell their stories, of course there are enormous risks.

But if we ask that platform to carry out the editorial work of a media outlet with respect to what the 3,000 million users say, the platform disappears.

It is literally impossible to fulfill this editorial function ”.

Another Facebook strategy started in 2017, when it began outsourcing the verification of its information by hiring some non-partisan media to review viral information on its platform.

"We cannot generate a direct consequence against misinformation, but indirectly we can, because Facebook takes as a signal what we publish so that they determine what decision to make in front of a post," explains Pablo Medina, director of ColombiaCheck, one of the 10 independent media that have contracts with Facebook in Latin America to verify information.

In any case, as with the Advisory Council, Facebook is not obliged to take the recommendations of these media.

"I do believe that digital platforms should have more legal responsibility for their content," says Medina.

Digital sovereignty or how to live without Silicon Valley

More than greater interference by the State, Medina and other experts consulted by EL PAÍS believe it is essential to create “a public-private mechanism in which there are representatives of the State, civil society and platforms, and that is where they deliberate on how to implement policies on content moderation ”.

An independent body in which the rules of the game are not determined by Facebook or the government of the day.

Renata Ávila, who in addition to her research and advisory work directs the Intelligent Citizenship Foundation, agrees that a State regulation mechanism does not make sense in the current situation, and that greater citizen, academic and technical participation is needed for this debate.

"Any regulation that the congressmen may be thinking of is nothing more than aspirin to kill cancer," he says.

But the proposal that Andrés Manuel López Obrador initially proposed to create an independent social network seems more interesting to him to return to a larger debate on digital sovereignty and the end of the Silicon Valley monopoly.

"When Lula was in charge of the presidency, the powerful were the

free

software

groups

in the region, the bad guy in the movie at that time was Microsoft," he explains.

"That was an interesting time because Latin American countries were very focused on building their own digital infrastructures."

These national projects, which promoted an incipient digital sovereignty, did not continue at the regional level, and the power of private social networks grew freely in the region during the last decades.

Even in Cuba "they began to use the social networks of the United States," he says.

"It was a total neglect of the governments of the region, having the capacities, and having the people, not having developed their own regional platforms."

Ávila does not agree with new social networks controlled by the State, but sees in this crisis about censorship and content an opportunity.

"If there is a real, technical citizen appropriation, and it does not become a political vehicle, but rather a common project that Mexicans are proud of, I think that something very interesting can be done there."

A decentralized experiment with citizen control in which, for now, the Mexican government does not seem to be interested.

(Darinka Rodríguez contributed to the reporting of this article)