This article belongs to the debate » The Rule of Law versus the Rule of the Algorithm
31 March 2022

Regulating News Recommender Systems in Light of the Rule of Law

Recommender Systems in the Online News Sector

Many online services – search, e-commerce, movie streaming, social media, and news – use recommender systems. Recommender systems are software programs that select items out of a bigger pool of items and then recommend these to users of online services. In this contribution, I focus on personalised recommender systems, which recommend different items to different individuals based on information about the individual.

I argue that it is largely unnecessary and, in any case, contrary to the rule of law to regulate how news media deploy recommender systems to select and rank the news for individual users. Instead, I consider an alternative for state regulation of news recommenders, should empirical research show that certain news recommender systems have harmful effects on individual rights and societies.

Democratic Role of News Media

The use of recommender systems by online news media is a particular case. News media are a cornerstone of democratic societies. They publish information on matters of public interest, act as a public watchdog over public bodies and powerful companies, and provide a forum for public debate. A well-functioning media sector contributes to the rule of law. At the same time, the rule of law requires media freedom. In this context, I distinguish news media from social media. People increasingly find news via social media, but these media are fundamentally different. News media produce and publish news themselves and exercise editorial control over the content they distribute, while social media only repost news produced elsewhere. Social media do not have media freedom, and the comments in this contribution apply solely to news media, not social media.

Online news media use recommender systems for commercial and journalistic goals. They try to obtain and retain subscribers with personalised recommender systems, while competing for attention with social media and other online platforms. In addition, news media, especially public service media, use recommender systems to realise journalistic ambitions of serving varied audiences and providing individuals with the news they need to become informed and critical citizens.

News recommender systems create opportunities and risks for individuals and societies. These personalised systems may help individuals find the news that is important to them and navigate the online information overload. At the same time, news recommenders’ risks for individuals include a loss of privacy and autonomy in making news choices. On a societal level, there are fears around filter bubbles and a decline of media diversity, an increase of polarisation, and the amplification of disinformation, which could have even more harmful effects when recommended to individuals who are specifically susceptible to false and fabricated information. While empirical research consistently shows that filter bubbles are much smaller and less prevalent than is commonly thought, news recommenders’ effects do warrant attention, seeing the news media’s democratic role.

Media Freedom Limits News Media Regulation

If news recommender systems are responsibly designed, implemented, and optimised, they can be used to benefit news media and audiences with limited risks for individuals and societies. Laws might help to ensure responsible recommender systems. For example, the European Commission’s proposal for a Digital Services Act (‘DSA proposal’) imposes an obligation on very large online platforms that use recommender systems to provide transparency about the main parameters used in their recommender systems and the options for individuals to modify or influence those parameters (article 29 of the DSA proposal). In the first reading of the DSA proposal, the European Parliament even included an obligation for very large online platforms that use recommender systems to provide at least one option for content presentation not based on profiling.

However, article 29 of the DSA proposal would not apply to news media, since the definition of ‘online platform’ does not cover them. Moreover, transparency only goes so far. Scholars have therefore commented that the DSA proposal should oblige very large online platforms to prioritize quality content.

A similar obligation, to prioritize specific types of content, cannot be imposed on online news media that use recommender systems. The rule of law requires media freedom. Any kind of regulation of news media should avoid restricting media freedom. Regulating by law how online news media use recommender systems would entail regulating how they select and present their news, which directly interferes with media freedom.

A Need to Regulate News Recommenders?

The principle of media freedom raises issues for the use of recommender systems by online news media. How can news users’ interests and fundamental rights be guaranteed, and how can public values be preserved if media freedom limits media regulation, specifically around decisions on what type of content to offer and how?

This question requires differentiation between public service media and commercial media. Commercial media, be it printed, broadcast, or online media, are not regulated in their decisions what stories to investigate, how to report on these stories, and which stories go on the front page. Nevertheless, these news media are subject to self-regulation. They adhere to journalistic codes of ethics and organise in press councils and are guided by strong ethical values, differentiating them from social media and other online platforms that are not led by similar standards. Many news media are highly aware of the risks of recommender systems, and not interested in building addictive services, where news users spend all day. Most news media use personalised recommender systems only on parts of their website or app to prevent filter bubbles. For instance, the mobile app of the Dutch news service NU.nl has three main sections in the bottom bar: ‘Front page’, ‘Just in’, and ‘My NU’. The app opens on ‘Front page’, where everyone sees the same news, while the ‘My NU’ section provides personalised recommendations. NU.nl wants ‘to be the place for everything you should know and everything you want to know. What you should know is what our editors believe is important, regardless of your interests […] What you want to know ties in to your interests’. The mission of providing quality content to news users and providing them with societally relevant news is at the heart of most news media, who might not need an explicit legal obligation to develop responsible news recommenders.

Public service media are regulated by national media laws. These laws usually formulate a remit for public service media to offer diverse and quality content. Thereby, these laws provide public service media an orientation point to decide what content to produce and how to distribute this. National legal provisions formulate the objectives for public service media on an abstract level, without the legislator dictating precisely what kind of content public service media can and cannot produce or what news goes on the front pages. Besides provisions on the public service media’s remit, national media laws mainly regulate their structure, organisation, funding, and supervision. Public service media that use recommender systems are guided by their remit, formulated in national media laws, and, like commercial media, by journalistic codes of ethics. These norms and standards could be enough to ensure that public service media’s recommender systems are oriented towards fundamental rights and preserve public values. For example, the BBC in the United Kingdom strives to ensure that their recommendations ‘reflect the breadth and diversity of our content and meet our editorial guidelines’ and are ‘informing, educating and entertaining’.

The above is not to say that recommender systems do not create challenges for public and commercial media. News media should update journalistic codes of ethics to the online environment, with more attention for the use of personal data, recommender systems, and other technologies such as artificial intelligence. Public service media need to ask how to translate their remit and values into recommender system design. Commercial media need to ensure financial motives do not lead in their recommender system choices. Still, it might be unnecessary and would, in any case, be contrary to the rule of law to regulate how news media use recommender systems to select and rank the news.

Supporting Instead of Regulating the News Media Sector

Even though many news media responsibly use recommender systems, not all news media fit this optimistic view. Some are less guided by journalistic ethics and more by profit. Tabloid-like media thrive on publishing sensational news and othering people. Mainstream news media sometimes play a role in spreading mis- and disinformation or (unwittingly) boost false content created by other accounts. Europe is also seeing a rise of highly partisan news media, which influence how citizens view the political system and are linked to low trust in mainstream media.

When these kinds of news media use recommender systems, the risks for individual rights and societies at large may increase. For instance, the German tabloid newspaper Bild has a fully personalised app called BUZZ. What is the effect on BUZZ users and their communities if they receive mainly personalised news recommendations via their mobile app, on topics such as the current covid-19 pandemic, German and international politics, and recent crimes? At the moment, we simply do not know the consequences of citizens mainly receiving news through recommender systems on tabloid or partisan news media, or media outlets containing disinformation. On the one hand, while filter bubbles are limited, as mentioned above, empirical research suggests that news recommenders can lead to selective exposure. If people are consistently exposed to news that confirms their views, they may become less tolerant to other views or groups. On the other hand, people usually use multiple news sources. The harmful effects of tabloid news recommender systems may be compensated by other, better news sources that citizens use, including public service media. Before law and policy makers take steps, we should empirically research whether recommender systems on less responsible news media threaten individual rights, democratic debate, and social cohesion.

If empirical research leads to the conclusion that news recommender systems on “less responsible” media should be addressed, the question is how to do so while respecting media freedom, which is crucial for the rule of law. In any case, the question of which media are less responsible than other media should be left to news users and the news sector themselves. It would infringe media freedom if the state made such a determination, and media freedom means we cannot subject some news media to regulation and others not. Likewise, standards of journalistic ethics and responsible news recommenders should be developed and enforced by the news media sector, not by the state.

One solution could be to support the news media sector instead of regulating news media and their recommender systems. We can take inspiration from approaches suggested to counter disinformation while respecting media freedom and freedom of expression. For example, the EU High Level Group on fake news and online disinformation recommended to fund independent quality news media and public service media as an antidote to disinformation spread by less trustworthy news media.

Financial support to sustain a reliable news media sector also contributes to the rule of law in two ways. First, financial support for trustworthy news media, to compensate for harm done by recommender systems on less responsible media, avoids direct regulation of news media and how they use recommender systems. Such regulation would interfere with media freedom and thereby diminish the rule of law. Second, if news media have the means to develop recommender systems that can compete with recommender systems used by social media and other online platforms, these news recommenders could facilitate the media in exercising its democratic role of informing citizens on matters of public interest. As such, responsible news recommenders may positively contribute to the rule of law.


SUGGESTED CITATION  Eskens, Sarah: Regulating News Recommender Systems in Light of the Rule of Law: Recommender Systems in the Online News Sector, VerfBlog, 2022/3/31, https://verfassungsblog.de/roa-regulating-news-recommender-systems/, DOI: 10.17176/20220331-131314-0.

Leave A Comment

WRITE A COMMENT

1. We welcome your comments but you do so as our guest. Please note that we will exercise our property rights to make sure that Verfassungsblog remains a safe and attractive place for everyone. Your comment will not appear immediately but will be moderated by us. Just as with posts, we make a choice. That means not all submitted comments will be published.

2. We expect comments to be matter-of-fact, on-topic and free of sarcasm, innuendo and ad personam arguments.

3. Racist, sexist and otherwise discriminatory comments will not be published.

4. Comments under pseudonym are allowed but a valid email address is obligatory. The use of more than one pseudonym is not allowed.