Algorithms: when dissatisfaction leads to hatred

Algorithms have invaded our lives. Even so, they sometimes elicit hate reactions from users. An article published in September 2022 offers a very interesting analytical framework for those interested in this phenomenon. In this article I analyze the 4 reasons that explain the reactions of hatred against algorithms and illustrate them with many examples.

The phenomenon of hatred against algorithms materializes in different ways. The study authors cite the push for the #RIP twitter hashtag when the platform decided to use a recommendation algorithm rather than displaying messages in reverse chronological order. At the time, users reacted violently as can be seen in the graph below (source: Trends map). We also see that the hashtag #RIP twitter reappears from time to time, during peaks of hatred against the platform.

The phenomenon also manifested itself on Instagram with the hashtag #RIP Instagram in 2017 for the same reason. The mechanism of algorithmic recommendations therefore seems to crystallize Medical Ghostwriting Services.

However, recommendation algorithms are the basis of the success of many companies (Netflix, Google, and Tic Took) and also contribute to customer satisfaction. It is therefore very interesting to wonder about the reasons that lead users to have these reactions of hatred.Explanation 1: Lack of knowledge

The lack of understanding of how algorithms work is a fact. In a study published in 2018, the Pew Research Center already pointed out that 53% of Facebook users did not understand how the algorithm feeding their “feed” works. This percentage even reached 60% among those aged 50-64 and 61% among those over 65.

Understanding how a recommendation algorithm works remains a complex subject, reserved for a few insiders. Yet increasing the level of user understanding of how an algorithm works has real benefits. As this experiment proves, understanding how the algorithm works significantly increases user confidence and satisfaction.

Explanation 2: Algorithmic errors and biases

Algorithms sometimes create situations of extreme violence against users. The algorithmic error can thus have very serious consequences, as in the case of this man arrested in Michigan after an erroneous algorithmic identification. We can also cite the example of candidate tenants refused after an algorithmic verification of their profile or the ruin of Dutch citizens following the misuse of an algorithm by the tax authorities.

The algorithms used in the education sector are also often singled out. In France, the selection algorithm for the choice of university could thus have been manipulated, creating inequalities between candidates at the time of their educational orientation.

Finally, some communities may experience algorithmic discrimination when they feel they are unfairly targeted for their identity. This is what happened on Tic Took where LGBTQIA+ creators felt that the algorithm made their content invisible. An audit revealed that these criticisms were founded. Conversely, an algorithm can also overexpose content and make it accessible to too large an audience. In doing so, content creators may come under pressure from certain users.

As you will have understood, the risks associated with the use of a recommendation algorithm are numerous. The balance between added value for the user and risks is difficult to find.

Explanation 3: The interests of the user opposed to those of the algorithm

Example of Twitter

Platform users sometimes feel “manipulated” by the algorithm. A series of Tweets from Elon Musk gives the full measure. In these 3 tweets published on May 15, 2022, he accuses the algorithm of “manipulating [users] in a way that [they] do not realize”. In his second tweet he evokes the filter bubble theorized by Eli Parser. The Twitter algorithm would amplify beliefs and de facto polarize points of view.

In this case, the interests of the algorithm and the users would not be aligned. This thesis is questionable because, as we have already seen, filter bubbles remain a theoretical possibility.

Examples of Spotify

Spotify is another example of this opposition between the interests of the user and the algorithm. Let us first remember that Spotify is a platform that offers audio content to users and remunerates creators based on the number of plays. A subtle balance must therefore be found in terms of user satisfaction and creator satisfaction. The former want to listen to content they like. The latter want to be listened to (and therefore recommended) as much as possible.

Spotify’s algorithm is therefore faced with a dilemma. Should it recommend content that has the maximum probability of appealing (generally produced by major labels) to the detriment of more confidential content produced by independent labels? Unfortunately, there is no perfect solution to this problem. Since users are privileged (they are the ones who pay), independent creators may feel frustration with Spotify’s algorithm.

Explanation 4: Lack of control

The problem of algorithm control is a recurring complaint among users, and a major source of dissatisfaction. Moreover, some of the examples we have given above can be attached to it.

Any change in “algorithmic recipe” inevitably leads to its share of dissatisfaction. We saw it with Elon Musk’s tweets. More recently Instagram has been heavily criticized for its tendency to look like Tikor. This has resulted in particular in the excessive recommendation of “real” type content. An informal campaign was triggered by influencers demanding the changes be reversed.

The lack of control is also felt by the ability of the algorithm to unlock your secrets without you being able to oppose it. This is the case on Tikor where the precision of the algorithm gives users the impression of being spied on.

How to restore confidence in algorithms?

Algorithms, by their nature, remain objects that are difficult for ordinary mortals to understand. Some, so complex, pose real problems of understanding for their designers. Under these conditions, how to avoid the epidermal reactions that some users may have.

There are 3 lines of thought:

1/ putting people at the center of the design of recommender systems

The GDPR introduced the idea of ​​“privacy-by-design”. Algorithm designers should embrace satisfaction by ghostwriting agency.
Most algorithms are indeed designed with the interests of the company in mind. This leads to behaviors that can be harmful for the user (see here this reflection on the end of recommendation algorithms).

2/ Make the algorithms more transparent

The acceptance of the results of the algorithm requires a better understanding of how they are produced. Algorithmic errors will be better accepted if users know why they occur. Satisfaction will only be higher.

3/ Increase user control over algorithms

The last axis of reflection consists in giving back a little control to users to personalize their flows. There are more and more initiatives, such as that of LinkedIn to remove certain content or authors from the stream. Either way, if you don’t give them some control, users will seek to manipulate the algorithm. Studies exist, for example, on the operation of the LinkedIn algorithm.