Sunday, April 12, 2015

Algorithms: secret gatekeepers on the internet


The terrorist attacks of January in France lead to a great demonstration in Paris celebrating freedom of speech. Ironically, instead of creating new rights, the antiterrorist law resulting from these events could arm investigation journalism and citizen right to privacy.

The new intelligence bill and the right of privacy protection

After the Charlie Hebdo and Hypecasher events, the shock was very important in France. A new law concerning surveillance was announced by the Prime Minister Manuel Valls a few weeks after the attack. This new intelligence bill is currently examined by the National Assembly (one of the chambers of the French Parliament). It prescribes a lot of measures which increase the powers of secret services, and it raises a tense debate in French society.

The most criticized measure is the use of “black boxes”, which will be installed at French telecom operators. These systems are algorithms, which reveal automatically people having a suspicious behavior on the internet, such as visiting or participate on jihadist websites, having contacts with presumed terrorists, etc. Once these people are identified, secret services would be able to access their private information. This law project is very unwelcomed, to say the least. Some French host companies are opposed to the bill and threaten to leave the country if it is adopted. Even Charlie Hebdo, the newspaper used by the government to justify such a restrictive measure, strongly criticized this use of “black boxes”. The project is not only perceived as inefficient, but also as intrusive in citizens’ private life. Anyone could be suspected and watched by secret services, even those who study terrorism in order to fight it.
 The paradox of the post-Charlie France: dialogue between opponents and partisans of a more watched environment.

Indeed, algorithms are not very clever. These systems are used to have a look on a huge content of information. An algorithm acts mechanically and cannot replace human expertise. It won’t make any difference between an actual potential terrorist and a journalist investigating on these issues. If journalists are watched, it threatens the secret of their sources and makes their work more complicated. More generally, this could be a threat for citizens who maybe were just searching for deep information, who will have their right to protection of private life and private information restricted.

Social networks and the protection of freedom of speech

Algorithms are often used in internet regulation; it is thus already possible to see what their limits are and foresee the problems that could face secret services in their struggle against terrorism through algorithms. Social networks use algorithms to control huge amount of content, which is used, produced and distributed by users on these platforms. Social networks are often criticized for blocking content mechanically, and it arms sometimes freedom of speech. Facebook is known to block nudes, which sometimes leads to suppression of art pictures (Facebook will go to court after blocking the account of a French professor who posted the picture “The origin of the world” by Courbet) and even breast cancer campaign, because algorithms automatically recognized it as pornography. It is also difficult to suppress flags of IS thanks to these systems, because the risk is to also block content such as news, articles, or people posts denouncing the organization for example. That’s why it is also impossible to use algorithms to block offensive hashtags, because reused by people who denounce the ones who used them in the first place, as it was the case with #jesuiskouachi, tweeted more by opponents to the attack than actual supporters of the Kouachi brothers, the authors of the Charlie Hebdo massacre. We see here that algorithms have their drawbacks and cannot be as effective as case by case examination.

”The origin of the world” by the realist painter Courbet was censored by Facebook.

The biggest issue of internet regulation is here. Is has become very difficult to control information online because of the evolution of new media. Now, as Axel Bruns stresses, internet users become also producers of content, which leads to a huge amount of different information and to blurred lines between amateurs and specialists, or journalists. It is particularly the case on social networks, were user produce, use and share contents. Algorithms are used to regulate this huge amount of information. This is a big change compared to what was regulation of content in traditional media. The specific context in which information could be consumed and the knowledge of who constitute audiences lead to regulation according to a specific context, namely if the information would be consumed in private or not, if children are susceptible to watch, etc. This context now changed completely, because it is possible to have access to content on different platforms, an aspect of the convergence of media phenomena explained by Henry Jenkins, and possible to access that content anytime thanks to internet. Moreover, newspapers, radio stations or TV channels take time to justify the content they publish, and explain their choice according to ethics of journalism. Social networks refer to their guidelines but won’t always justify why they suppress some content or not. Due to professional secret, we don’t know how their algorithms works, which content they target according to which ethics and moral principles.

The French media specialist Olivier Ertzscheid explains that social networks’ guidelines work  as the only law online and denounces the obscurity around algorithms used to make this rules respected. Who finally decides the content targeted? It is difficult to imagine that so much power on information received by more and more people around the globe can be decided by some unknown actors according to obscure moral principles. This could possibly arm freedom of speech, as Instagram illustrated recently by censoring a picture of a girl having period blood on her clothes and bed. This picture belonged to the project of the artist Rupi Kaur whose aim was to demystify periods and denounce the taboo around it, and was finally allowed by the social network which recognized its “mistake”.


The algorithms raise the questions of the issue of freedom of speech concerning social networks and the issue of privacy concerning French new intelligence bill. The extension of information we are exposed to makes it difficult for public powers to find efficient ways to fight terrorism or, for example, pedophilia, without any restriction of fundamental freedoms. In France, the system of algorithms used by social networks to spot unappropriated content would be used by secret services to spot “unappropriated behavior” through French operators. The limits would be the same, as Facebook interprets breast cancer campaign as pornography, these algorithms would be likely to commit mistakes too. And French secret services actually don't need more targeted surveillance to commit mistakes: before the Charlie Hebdo attack, secret services confused the address of one of the Kouachi brothers with the one of an homonym who was…a 81 year-old man. 

No comments:

Post a Comment