The complex legal work of judging the legality of content is now delegated to digital companies, which largely automate it with imperfect software.
Each day its scandal, more or less serious, affecting moderation on social networks. This weekend, the dissemination of links to the site "Political Porn", which hosted the intimate videos of Benjamin Griveaux, on Facebook and Twitter, provoked a flock of critics. On Monday, it was, conversely, the deletion of the Twitter account of the Syndicat des Avocats de France (SAF) which sparked outcry in the political class.
Social media moderates too much, or not enough. In any case, never as we would like. But beyond the shortcomings - very real - of the large platforms, of their inability to manage the monsters that they themselves have created, these successive fiascos should question us about our own role. The user who skids is always the other; the one whose freedom of expression is flouted, always us. And despite the multiplication of laws aimed at "framing" or "regulating" online speech, the State has largely abandoned the very idea of arbitrating what can or cannot be said on social networks, by delegating to large digital platforms take care to make these difficult choices.
Predictable side effects
The bill on hate content on the Internet, known as the "Avia law", under parliamentary discussion, provides for severe penalties for social networks that would not remove illegal content in less than twenty-four hours. But in practice, it is Facebook and Twitter that are responsible for defining the jurisprudence against hate online. It is their moderators who must, in a few tens of seconds, decide whether a text constitutes a call to hatred or the legitimate expression of an opinion. We have collectively delegated complex legal work to underpaid "click workers" and, worse, to automated systems.
Because, if our legal system cannot absorb hundreds of thousands of daily complaints, the human moderators of these platforms cannot deal with each case either. The overwhelming majority of the moderation of Facebook or Twitter today is done by algorithms, which determine, sometimes even before a message has been read by a human, whether a content constitutes an apology for terrorism or a insult. We also pay, on a daily basis, for the errors of these necessarily imperfect tools, which we are asked to arbitrate between freedom of expression and necessary moderation when we ourselves are sometimes unable to decide.
These automated tools have many very predictable side effects. The FAS Twitter account was temporarily deactivated because it called out a large number of elected officials, a behavior which, for a machine, has all the attributes of characterized harassment. Elected officials who urge social networks to take responsibility are very often the same people who have asked these platforms of expression to take, in place of the State, the responsibility of administering a form of justice in line.
The Internet is not an anonymous wild west
Despite all the legitimate criticisms that can be made of it, there is a point of principle on which it is difficult to contradict Mark Zuckeberg, the boss of Facebook: it is not for a private company to determine what can, or does not can not, be written on a social network. He repeated it again last Monday in Brussels.
The rules set by the networks themselves cannot replace the law. But good law is also applicable law. It is not enough to declare that such or such type of content is illegal for it to magically disappear from the Internet, without consequences for other "legitimate" content.
Part of the moderation problems that we see are also the direct result of the requests we made to the leaders of social networks, to whom, paradoxically, we entrust more and more quasi-regal powers while criticizing their ability to place themselves at the - above the laws. But the political argument remains unstoppable: by presenting social networks as “spaces of lawlessness”, where only “torrents of mud” would flow and that it would be necessary to “regulate” with force, candidates and elected officials give , at little cost, the impression that they act. And since the "torrent of mud" is always published by the other, never by us, it is easy to win agreement on these simplistic statements.
It does not matter that, ultimately, these spaces are already very largely framed by law: the rapid investigation of the alleged broadcasters of Mr. Griveaux's intimate videos provides yet another demonstration that the Internet is not - and has not never been - the anonymous wild west that many critics like to describe.
To go further would imply following the example of authoritarian regimes which, in the arbitration between moderation and freedom of expression, resolutely chose the first. Despite all their shortcomings, “open” social networks remain a sign of good democratic health. Reporting their abuses is one thing; clearly wanting to control all forms of popular expression is another.
Sources of article: