책冊

The Invisible Information Prison Created by Algorithms

Today, we live exposed to more information than in any previous era. Just opening a smartphone unleashes an endless stream of news, YouTube videos, social media posts, and search results. On the surface, it seems like we are encountering diverse perspectives amidst a flood of information. However, the exact opposite is happening. We are increasingly exposed only to the information we want to see, and we are trapped in an environment where we cannot see opinions different from our own. This phenomenon was named 'Filter Bubble' by Eli Pariser, an American internet activist and author.

A filter bubble refers to the phenomenon where internet platforms analyze users' preferences and behaviors to provide personalized information, thereby confining users within a specific information environment. Without realizing it, users repeatedly encounter only information that aligns with their preferred opinions, while opposing information or different perspectives gradually become hidden. It's like being trapped inside a soap bubble, unable to properly see the outside world.

In the past, when newspapers and broadcasting dominated information distribution, everyone was exposed to relatively similar news. Conservatives and liberals could discuss the same front-page newspaper article or the same evening news. However, in the era of internet platforms, the situation has completely changed. Even if the same event occurs on the same day, different information is provided to different people. One person's YouTube screen might recommend only videos of a certain political leaning, while another person's screen shows only content of the opposite leaning. Consequently, even though they live in the same society, they experience different realities.

This phenomenon occurs because of the recommendation algorithms used by platform companies. The goal of platforms like Google, YouTube, Facebook, Instagram, and TikTok is to keep users engaged for as long as possible. The longer users stay, the more ads they see, and the company's revenue increases. Therefore, algorithms continuously recommend content that users are likely to enjoy.

For example, let's assume a user clicks on a video with a specific political leaning once. The algorithm identifies this as the user's interest. Subsequently, it starts recommending more videos of a similar nature. If the user watches these, the algorithm learns even more strongly. Eventually, the user's screen becomes filled with similar content. Over time, the user mistakenly believes that the information they are seeing represents the entire reality.

The problem is that filter bubbles go beyond matters of taste and seriously impact democracy and social cohesion. Democracy is a system where citizens with different opinions debate based on common facts. However, within filter bubbles, people come to believe different facts. Regarding the same event, one side considers it an undeniable truth, while the other believes it to be a fabricated lie. A common understanding of reality disappears.

This phenomenon further deepens political polarization. Humans inherently have a confirmation bias, preferring information that aligns with their own beliefs. Algorithms maximize this human psychology because they repeatedly supply only the information users like. As a result, people become increasingly convinced of their own opinions and are unwilling to understand opposing views. In fact, several studies report that the longer people spend on social media, the more political polarization tends to worsen. Content that incites anger, fear, and hatred, in particular, often gets actively spread by algorithms because it increases user engagement. Sensational and extreme claims receive more attention than calm and balanced analyses.

The problem of fake news is also closely linked to filter bubbles. People tend to be less critical of information that aligns with their beliefs. Algorithms exploit this characteristic by prioritizing content that users are likely to prefer. Therefore, even if information is of questionable veracity, it can spread rapidly if it aligns with the user's existing beliefs. Consequently, false information transforms into stronger convictions, and information that refutes it never appears on the user's screen.

The interesting thing is that the filter bubble does not operate coercively. In the past, censorship by authoritarian regimes involved the state directly blocking information. In contrast, modern filter bubbles are created by a combination of user choices and algorithms. People think they are making their own choices, but in reality, they are often choosing within an information environment designed by algorithms. This makes filter bubbles even more powerful and dangerous, because people don't even realize they are trapped.

The most prominent example of a filter bubble in Korean society can be found in the political YouTube ecosystem. YouTube is structured to continuously recommend similar content by analyzing users' viewing history and interests. The problem is that this recommendation system can act as a very powerful propaganda tool in the political sphere. For example, if a user watches a few videos of a certain political leaning, the YouTube algorithm will continue to recommend similar political channels and videos. Over time, the user's screen becomes filled only with the arguments of a specific camp. Opposing views or balanced analyses gradually disappear, and only the political messages they prefer are repeatedly exposed. Users think they are encountering diverse information, but in reality, they are consuming information only within a single political world created by the algorithm.

A bigger problem is that political YouTube channels generate profit by stimulating emotions rather than delivering information. Due to YouTube's revenue structure, where views and watch time are crucial, content that incites anger, fear, and hatred is far more advantageous than calm analysis. This is why extreme expressions such as 'the country is collapsing,' 'the election was rigged,' 'the opposing camp are all traitors,' and 'a dictatorship has begun' repeatedly appear. Emotional mobilization generates more clicks than factual accuracy. In this process, some political YouTubers present unverified claims from the media or academia as facts. Users often accept these without critical examination because the content aligns with their already trusted political views. Within a filter bubble, even false information can be perceived as truth through repeated exposure.

This phenomenon is particularly evident in Korean politics, observed in both progressive and conservative camps. People who only watch YouTube channels from a specific camp lose the opportunity to encounter the arguments of the opposing side. Consequently, they come to perceive the other side not as political competitors to be understood, but as enemies to be eliminated. What is needed in a democratic society is a culture of debate that acknowledges differences of opinion, but filter bubbles and political YouTube often amplify hostility and hatred instead. More concerning is that this information environment also influences actual political behavior. Some users come to believe content repeatedly encountered on YouTube as objective fact, and their judgments on elections, policies, and social issues also depend on such information. Ultimately, there is an increased risk that political decision-making will be swayed by emotional content provided by algorithms rather than by facts and evidence.

Political YouTube itself is not the problem. On the contrary, it also plays a positive role in providing diverse perspectives that traditional media cannot cover. However, when users start to rely on only one channel or one political camp, the filter bubble rapidly intensifies. This does not expand the diversity of information but rather reinforces a specific political worldview.

Today, the filter bubble extends beyond politics to influence the economy, culture, and society as a whole. Investors only encounter investment opinions they believe in, consumers are only recommended products that match their tastes, and citizens repeatedly confirm only the values of the groups they belong to. The development of information technology is paradoxically weakening social diversity.

So, what is the way to escape the filter bubble? The most important thing is to intentionally make an effort to encounter diverse perspectives. One should read media outlets they don't prefer and examine the arguments of people with opposing views. Furthermore, instead of consuming only information recommended by algorithms, one should develop the habit of actively searching and comparing. Platform companies should also design algorithms in a way that considers information diversity and public interest, not just user engagement time.

The filter bubble is not a technological problem. It is a structural problem of modern society created by the combination of human confirmation bias and platform capitalism. We live in an age of information overload, but paradoxically, we may be thinking within an increasingly narrow world. True information freedom comes not from seeing a lot of information, but from the ability to encounter different perspectives. Understanding the filter bubble is the first step to understanding modern democracy and the media environment.