Terms like algorithm, machine learning and big data seem to be frequenters in modern society and people’s life, especially online life. They are so common that people even take it for granted that algorithm and its derived productions are ought to be applied into various fields that are highly connected to the daily life. For instance, news, social media and so on. For some people, data and algorithm may be an arcane concept that is far away from them. Although the algorithm related technics and mathematical principles are complicated, the basic mechanic is simple. According to Just and Latzer (2017), Internet algorithm analyzes various data and information and relocates them. Make a vivid metaphor. The process is like classifying books in library that you need to differentiate the content of each book first and then put them into the right area. And this is how algorithm functions in our life that it examines the data before sending it to us through various channels.
However, when there is something wrong with the examination process, users are likely to be affected. In 2017, a London teenage girl died due to the action of self-harm and after years of investigation, police found out that her social media account has a mass of suicide, self-harm, and depression related content (Milmo, 2022). In this case, the possibility of social media negative guidance and recommendation algorithm mechanism problem have been taken into serious consideration. For now, people may be curious that how does this algorithm affect its users? Based on the research of Flew (2021), the whole process can be divided into three parts. At first, personal information will be collected and sent to the algorithm. Then, the algorithm will assign various content to its user after studying personal preference and habits in detail. Finally, when user starts to depend on the recommendation, the algorithm will transmit different information to reshape users invisibly. To be more precise, this blog will explain more about these three processes and then put forward possible ways to resist this reshape.
Data collection and consent
Before moving on, there is an important concept that needs to be explained. The concept is data and datafication. In light of Kitchin (2014, p.1), data is “measurements of a phenomena, such as a person’s age, height, weight, color, blood pressure, opinion, habits, location, etc.” and when these elements are measurable, it will be more convenient to store and carry the information at the same time. As for the datafication, it is a process to make personal information into data.
Apart from converting user information into data, there are also various ways for company to collect it especially in contemporary society when people are more rely on different kinds of online platforms to deal with different works such as eat, travel, and consumption. And before users can access to one specific platform, they have to agree with user term and regulation. Otherwise, they may be barred from using functions of platform. Normally, these user terms have hundreds of thousands of words or more and consist of legislation terms so that most people may not fully understand the meaning. In this case, many users will simply click the agree button without thinking. And this has given companies the freedom to collect information about their users. An early data leakage event can prove this. In 2006, a research about 1,700 college age Facebook users was conducted but the data was released to the public then, leading to the privacy disclosure of the students (Steinmann et al., 2015). However, as the “culprit”, Facebook has not suffered much from the incident due to its user term that users are agree with relinquishing their data to use Facebook. That is to say, this action is totally legitimate. The similar incidents happened a lot in the recent years. And these collected data will be used for trade with other companies without users’ permissions (Mejias & Couldry, 2019). And some companies even rely severely on the profit of user data. That is why Meta (parent company of Facebook and Instagram) was so anxious when apple allow its users to ban data collection on their equipment in 2021. And Meta may lose about 10 billion dollars of profit due to Apple’s new policy (Newman, 2022). We can conclude from these events that data is actually very valuable. But the problem is how do those companies make data into money? Next, we will explain about it.
Some people may think it as a convenient way for their consumption that they can easily get the information of something they plan to buy. However, the recommendation algorithm may also lead to the filter bubble condition. According to Pariser (2011), this term is used to describe the condition when algorithm filters information and data around people so that one can only access to the things he is interested in or the information that is deliberately prepared by the algorithm. And the latter is worse. As a result, the user will be gradually separated from the outer world as the cognition has been blocked by the filter bubble. Research about social media algorithm showed that YouTube is more likely to push similar content once the user watched a video, even it is an extreme content (Reed et al., 2021). And the reshape of algorithm on its user starts from it.
Back to the tragedy case mentioned above, the death of that teenage girl is suspected of being involved with the social media content she received. The recommendation algorithm kept sending negative content about suicide and the filter bubble would start to influence her mind and behavior unconsciously. Reed et al. (2021) claimed that the social media company should consider providing users with different content with alternative perspectives. If the platform could push some anti-suicide, positive content instead of extreme idea, the tragedy might not have happened. This is one aspect that people should worry about algorithm.
On the other hand, due to some unknown purposes, the algorithm may be designed to transmit prepared information. One of the famous algorithm controversies might be the United States president election between Trump and Biden. In light of CNN, the commonly used social media platform Instagram blocked negative contents and hashtags of Trump (Effron, 2020). Even though the company claimed that it was a bug in its algorithm, it still caused a wide attention and the election manipulation conspiracy started to appear. Voters were worried that they may be reshaped by the social media unconsciously. However, this concern does not come out of nowhere. Reuning et al. (2022) found out that media exposure of local parties in United States was manipulated by social media in 2018 and 2019. Because the algorithm involves the business secret, there is no direct evidence to prove the hypothesis of scholars. In 2021, Etter and Albu conducted research on social movement and social media (2021). They found out that social media does have the power to manipulate, to lead or even to misguide various social movement. The algorithm makes views and ideas be taken for granted by activists and communities through filter bubble and other ways. But what these activists do not know is that the companies behind may make profit through their movements. In other words, as what has been claimed by Etter and Albu (2021), those activists are shaped and exploited.
Now that we have a basic view of personal data and how people are influenced by algorithm, the solutions could also draw on the processes mentioned above. On the one hand, the users should be aware of the user term and user consent to protect their data use. Dr. Fred Cate expressed his concern on the compulsory user consent during a TED speech. According to him, the use should have the right to agree or disagree the company’s data policy (Cate, 2020). However, as what has also been referred in the previous part in this blog, the current user policies are too complicated for a normal user to understand. And the company can easily get away from being punished. In this case, Cate claimed that the attention should be paid on the stewardship rather than user consent. That is to say, companies are required to take responsibility while there is something wrong with the data privacy even if they have already reached an agreement with users. And with the regulatory measures of authorities, the collection and abuse of personal data should be curbed.
On the other hand, apart from the data collection, the algorithm should also be improved. In light of Reed et al. (2021), in addition to the positive and alternative content, the companies should also strengthen supervision over problematic, vicious content, especially those platforms of children and teenagers and keep greater transparency on the information distribution. This can help to decrease the risk that users may encounter harmful information. And the transparency policy can make user realize when and how their personal data is collected and used by who. For example, Apple’s new updated system can alert users that their data is being tracked.
So far, this blog has briefly explained the data and algorithm through the three processes of personal data industry. And then, this blog gave two possible solutions to the privacy data protection. However, both solutions are on the stance of company and government but not the user. Like what has been discussed in the speech, Dr. Cate claimed that in fact most users would not even glance at the user term before they click the agree button. The public awareness of their data privacy still needs to be strengthened because their data can be consumed invisibly. Considering Apple’s report that the father and the daughter may not even notice that their information is tracked, the issue of privacy security should arouse more public attention. In addition, the relevant research on how to better conduct data privacy protection should also be concerned.
- Apple. (2021, April). A Day in the Life of Your Data: A Father-Daughter Day at the Playground.https://www.apple.com/privacy/docs/A_Day_in_the_Life_of_Your_Data.pdf
- Cate, F. (2020, January 17). Data Privacy and Consent. YouTube. https://www.youtube.com/watch?v=2iPDpV8ojHA
- Effron, O. (2020, August 6). Instagram’s algorithm blocked negative hashtags of Trump, but not Biden. CNN. https://edition.cnn.com/2020/08/06/business/instagram-biden-trump-algorithm/index.html
- Etter, M., & Albu, O. B. (2021). Activists in the dark: Social media algorithms and collective action in two social movement organizations. Organization (London, England), 28(1), 68–91. https://doi.org/10.1177/1350508420961532
- Flew, T. (2021). Regulating platforms. Polity Press.
- Just, N., & Latzer, M. (2017). Governance by algorithms: reality construction by algorithmic selection on the Internet. Media, Culture & Society, 39(2), 238–258. https://doi.org/10.1177/0163443716643157
- Mejias, U. A. & Couldry, N. (2019). Datafication. Internet Policy Review, 8(4). https://doi.org/10.14763/2019.4.1428
- Milmo, D. (2022, September 20). ‘The bleakest of worlds’: how Molly Russell fell into a vortex of despair on social media. The Guardian. https://www.theguardian.com/technology/2022/sep/30/how-molly-russell-fell-into-a-vortex-of-despair-on-social-media
- Newman, D. (2022, February 10). Apple, Meta And The $10 Billion Impact Of Privacy Changes. Forbes. https://www.forbes.com/sites/danielnewman/2022/02/10/apple-meta-and-the-ten-billion-dollar-impact-of-privacy-changes/?sh=3dcb4aa472ae
- Pariser, E. (2011). The filter bubble : what the Internet is hiding from you. Viking.
- Reed, A., Whittaker, J., Votta, F., & Looney, S. (2021). Radical Filter Bubbles: Social Media Personalisation Algorithms and Extremist Content. Royal United Services Institute (RUSI). http://www.jstor.org/stable/resrep37297
- Reuning, K., Whitesell, A., & Lee, H. A. (2022). Facebook algorithm changes may have amplified local republican parties. Research & Politics, 9(2). https://doi.org/10.1177/20531680221103809.
- Steinmann, M., Shuster, J., Collmann, J., Matei, S. A., Tractenberg, R. E., FitzGerald, K., Morgan, G. J., & Richardson, D. (2015). Embedding Privacy and Ethical Values in Big Data Technology. In Transparency in Social Media. (pp. 277–301). Springer International Publishing. https://doi.org/10.1007/978-3-319-18552-1_15
- Kitchin, R. (2014). Conceptualising data. SAGE Publications Ltd, https://doi.org/10.4135/9781473909472