Open menu

Platforms like YouTube, Facebook and Twitter provide you with personalized feeds and recommendations to make it easier for you to discover content that you like. In a previous blog post, I already argued that these recommendations might not always serve you well: arguably, YouTube's aim is not to provide you with that one educational video that will change your life, but to stimulate you to continue watching other movies. This goal is far easier reached with funny cat movies that entertain you, but that are not actually intellectually challenging. Whether this is a good or a bad thing depends on your state of mind and ambitions: if you want to relax after a long working day, cat movies might be just what you need, but if you actually aim to learn something, this focus on entertainment is counterproductive.

However, there is a particular type of recommendations that obviously does not primarily serve the user's goals: personalized advertisements are specifically designed for the benefits of advertisers and for the platform to make money. There are serious concerns regarding the collection of personalized data for this purpose (as stated in a popular Forbes article, if you’re not paying for it, you become the product), but that is not the focus of this post. In my opinion, there is another increasing problem: advertisements are often not recognizable as such, and recommendations for seemingly independent articles actually lead to advertorials that intend to influence you without you being aware of it.

Influencers

In the early years, people were rather amused by influencers, people who are paid to promote products on social media, pretending that their endorsement was completely independent and genuine. I once saw a documentary about an offline influencer, who actually walked around at parties and started conversations always ending up talking about a specific beer brand. It was completely obvious that this woman was promoting something. In the meantime, however, influencer marketing has become a mature industry. Not only celebrities make money by promoting brands: there are people who become celebrities and earn a lot of money just because of the fact that they are successful influencers. It has even come to the point where people pretend to be an influencer, hoping to become one.

beautyblogger

Dutch "beauty blogger" and influencer Debbie Zwiers (with an interesting variety of pictures on Instagram)

As a result, we often cannot be sure whether a can of Coca Cola appears on the picture by coincidence or whether it was purposefully placed in exchange for money. This might seem innocent enough, but exactly the same happens with actual content.

Advertorials

Many reviews and articles are actually advertorials without being labeled as such. According to Wikipedia, advertorials are "usually written to resemble an objective article and designed to ostensibly look like a legitimate and independent news story". In many countries, there are laws and regulations that require an advertorial to be clearly marked and recognizable as such. In practice, advertisers (obviously) prefer the label to be as invisible as possible. The Huffington Post is well-known for its sponsored links to seemingly independent articles (scroll down this article until you see "You may like"). You might argue that the links to parties like Babbel, Throne Free Online Games and Trips Shop are still recognizable as such (and actually, there is a text "sponsored by", which I first did not notice, to be honest), but do realize that you - like most readers of this blog post - are probably more conscious about this than the average web user.

sponsoredarticles

Sponsored articles that I received on the Huffington Post

So, how does this work behind the screens? As you probably know, I am site admin and chair for the Dutch vegetarian website Vegatopia. In order to finance our hosting costs and other expenses, we depend on advertisements. However, increasingly, we get requests for advertorials that ideally would be presented as regular articles. About two-thirds of all requests are withdrawn once we have made clear that we insist that all advertorials are labeled as such. Note that Vegatopia is far from being an influencer and that we apparently are very conservative in our advertising policies. An interesting example of an advertorial is this article on pancake raclette (written by myself, I recommend the activity), which is actually written completely independently and the only thing that we were being paid for was the placement of a link, without further explanation or endorsement.

preipannenkoek pan

Pancake raclette, a family tradition

What does this mean for recommendations and news feeds?

There are many valid reasons for recommendations to influence your behavior (a bit). For example, I wouldn't mind if YouTube would recommend me some educational videos to motivate me to stop watching cat movies and do something useful. I also think that Google News should not limit my personal newsfeed to topics that I am interested in: it should also inform me about important events, even if they are related to soccer (at this point, I start wondering whether it is possible for soccer events to be important at all). Finally, I think it's also valid that Facebook feeds are not only meant to please you, but also to keep you returning to Facebook. After all, these recommendations serve multiple parties.

In a study on creepy advertisements (about which I will write soon), we discovered that sponsored recommendations that are intertwined in actual recommendations are actually noticed as soon as there's a mismatch between what a user expects and the items that are being recommended. Once this happens, users may start to distrust all recommendations. And perhaps for good reasons.

facebookad facebookadexplanation

Facebook ad in a user's newsfeed and the corresponding explanation

In sum, it is clear that actual, objective content is increasingly being mixed with sponsored, targeted content that is meant to influence you one way or another. In addition, recommendations may also be intended to influence you or to expose you to sponsored content. This practice is largely invisible to users and works in several layers that reinforce one another:

  • For users, it is often hard or impossible to figure out which articles, posts, blogs and videos are 'genuine' and which are 'sponsored, as the latter category is deliberately often not labeled as such.
  • For users, it is also often hard to find out which items are recommended to serve their interests, goals or queries, and which items are recommended for other reasons (varying from promotion of items or sponsored content, to more beneficial reasons such as diversification).

However, these two layers most likely interact with one another and pose new questions to be dealt with:

  • If users cannot spot the difference between 'real' and 'sponsored' content, how could 'genuine' recommender systems separate between the two categories?
  • How bad or undesirable is it to recommend sponsored content that actually serves the users' needs or interests? Note that media independence is considered a normative principle for free press.
  • As many advertorials are disguised as actual content (and often are actually informative), is there a real, clear separation? What (as is the case with my sponsored article on pancake raclette) if an article is written independently, but the choice of topic influenced or dictated by a sponsor?
  • Given the above questions, will it ever be possible to design fair and transparent recommender systems that work in a daily context, with most content, services and platform provided by (commercial) parties with their own interests?

As the line between independent and sponsored content and recommendations probably will increasingly be blurred, I am curious to see which policies, regulations, technologies and other solutions will be explored and developed. I hope this blog post convinced you of the need for them.

eelcoherder 256px

Privacy Engineering, User Modeling, Personalization, Recommendation, Web Usage Mining, Data Analysis and Visualization, Usability, Evaluation

Dr. Ir. Eelco Herder
Associate Professor
Human-Centered Computing Group
Information and Computing Sciences
Utrecht University

Buys Ballotgebouw
Princetonplein 5
3584 CC Utrecht
The Netherlands

Email: e.herder@uu.nl

linkedin

facebook