Online Fandoms

Sarah Jiminez is one of many fans that utilizes social media to not only support their artists, but to collaborate with other fans around the world. The Korean pop (K-pop) fandom is a super community composed of fans of numerous K-pop groups such as BTS, Blackpink, and Girl’s Generation. It is estimated that the K-pop stan community has roughly 100 million people, with BTS ARMY, fans of BTS, the most popular K-pop group, making up 50 million alone.

In the wake of the protests surrounding the George Floyd murder, on June 3, 2020, K-pop fans collaborated and overtook white supremacist hashtag “#WhiteLivesMatter” on Twitter, flooding it with fan recorded videos and memes, making it a K-pop trending hashtag.

K-pop’s remarkable rise and global impact bring up the questions of: how can we effectively quantify and compare collaboration among fan groups on different online platforms? What distinguishes K-pop fans’ collaborative movements from others?

Comparing Fan Activity during Album Release

We primarily explored the online presence and activity of BTS fans compared to that of two popular Western artists, Taylor Swift and Justin Bieber. Over the years, both Swift and Bieber have developed devoted fanbases, who identify themselves as Swifties and Beliebers respectively. For the purposes of this analysis, we used BTS fans as a proxy for K-pop fans in general as BTS ARMY are the largest subgroup of the K-pop community at 50 million strong.

We specifically focused on the period prior to and following an album release due to increased fan participation from new content. Fans can share their first impressions of the new album and spread the word about new music, hoping to push their artist’s music to popular record charts such as the Billboard Hot 100 Song. To account for technical and social developments, such as the rise of Twitter’s popularity, we chose albums from each artist that were released roughly around the same time. By comparing these three artists and fan participation levels during three of their releases, we gain a more quantified understanding of the social impact of musical fanbases online.

Our Data Sources

For a wider scope of fan activity and artist popularity, we explored three different online platforms: Google Trends, Twitter, and Wikipedia. These data sources give us multiple facets for collaboration to compare with varying levels of required commitment, from easiest (Twitter likes) to hardest (Wikipedia revisions).

Google Trends

Google Trends analyzes the popularity of search queries inputted into Google’s search engine over a given period of time for various locations and languages. The data is an unbiased sample of Google search data that is anonymized, categorized, normalized, and aggregated to the time and location of a query. Search data can be filtered by category, such as “Arts and Entertainment” and “News”.

Google Trends then computes and outputs a scaled popularity score on a range of 0 to 100, with 100 being the maximum possible search interest, by comparing the volume of the term’s searches to the overall volume of site searches for the set period and geography.

Monthly Google Trends Arts and Entertainment search interest since 2004:

Monthly Google Trends General search interest since 2004:

When plotting both our Arts and Entertainment and general search, we see that BTS has had a steadier increase in interest compared to Swift or Bieber, who became “overnight sensations” and had a large spike in interest during their early career. On our Arts and Entertainment graph, it appears that Taylor Swift and Justin Bieber are on a downtrend in popularity, with interest spikes during album releases, after peaking in the early-mid 2000/2010s, with BTS overtaking Bieber in popularity around August 2018. Currently, all three artists have approximately the same level of Arts and Entertainment interest, with Swift having a slight lead over BTS and Bieber at the bottom of the three.

Our general search plot shows a lower general interest rate for Taylor Swift and an increased interest in Justin Bieber and BTS. However, this plot shows a sharper decline in popularity for Bieber and higher increase for BTS in comparison to the Arts and Entertainment trends. Despite Swift and Bieber’s established popularity, newcomer BTS has been able to unseat both of these popular Western artists in general search interest before the release of their second album.

Twitter

To obtain our Twitter data, we used the Python API twint to scrape tweets without the limitations of the official API, which only allows for scraping up to seven days before the usage date. We determined “relevant” tweets during the aforementioned time periods by scraping Twitter for two hashtags per album: the most popular artist hashtag and the most popular album hashtag. For example, for BTS’s first album Dark & Wild, we scraped all tweets that contained the “query” hashtags #BTS and #DarkAndWild during the time period.

From twint, we are only able to scrape original tweets; we unfortunately cannot access retweets or replies to tweets themselves. However, each tweet contains information about the number of likes, retweets, and replies that specific tweet received. By using that data, we are able to estimate how many total tweets there were during a given timeframe.

Summary Statistics for Twitter album data

By sheer numbers, BTS Twitter fans appear to carry more weight per individual which more than compensates for their lower overall size throughout the three albums. By the third album, there are about 1.1 total tweets (including replies) per follower for BTS in that time period, while Taylor Swift and Justin Bieber average 0.020 and 0.0035 tweets per follower. Though BTS has a relatively smaller fanbase size as defined by the number of followers, its members are far and beyond more actively engaging on Twitter.

However, sheer numbers alone do not conclusively indicate collaboration. It is possible that the original tweets in our dataset could largely be independent individuals tweeting about their favorite artists/albums. As a specialty of fandom Twitter—not counting artist and public relations (PR) tweet contributions—there are two main subsets of users: content generators and amplifiers.

Naturally, content generators make up a very small portion of total Twitter users (as seen in the figure below), but account for the majority of engagement. With the amount of engagement the content they push out receives, we consider them a primary driving force behind Twitter collaboration, since they enable mutual engagement from their posts. For example, if a content generator posts a tweet asking for fans to report to spread the release of an album for more people to listen to it, fans collaborate by liking (increasing visibility), retweeting (sharing), and replying (adding their own content). So, by studying these content generators, we are better able to study the mechanisms behind collaboration on Twitter, opposed to just looking at raw numbers and assuming collaboration.

BTS Be Non-Artist/Non-PR Engagement Distributions

To fairly compare the role and power of content generators across the three artists and their albums, we normalized each of the albums to compare how the fanbases changed over the years. Since there is no simple way of classifying content generator users, we chose to analyze a curve borrowing the concept of the Lorenz Curve to calculate and visualize the A percentage of these few users (content generators) that account for B percentage of engagement (likes/retweets/replies). Mathematically:

To apply the Lorenz Curve concept, we looked at original tweets, since content generator tweets are all original tweets by definition. Visualizing and analyzing curves through that range allows us to effectively envision what content generator impact looks like.

Wikipedia

To obtain our Wikipedia revision history, we used the MediaWiki API to scrape the complete revision history in XML format of relevant articles, including their Talk pages. We scraped the main pages for each artist, and the pages for the albums of interest. (For example, for BTS, we scraped "BTS", "Talk: BTS", "Dark & Wild", and "Talk: Dark & Wild" among other albums.) In addition, we used the MediaWiki API to collect Wikipedia pageviews, which contain information about how many people viewed a certain Wikipedia page.

Wikipedia Summary Statistics for Relevant Pages

It’s important to note that the time that these Wikipedia articles have been active may play a factor into how much traction it received. Overall, the amount of edits on Wikipedia has been decreasing, with a 6.90% decrease from 2019-2020 alone. Taylor Swift’s article was created on June 4th, 2006; Justin Bieber’s on April 22nd, 2008; and BTS’s following their debut on July 4th, 2013. To account for these differences, we normalized by the number of months since the article was released in Figure X to accurately compare the three articles.

Overall, it appears that the trend in Wikipedia page views stays relatively constant with regular spikes. Notably, these large spikes can be correlated with recent album releases or other major career developments.

For Taylor Swift, the obvious spikes in the pageviews correspond to the following events:

August 2017: Release of studio album ‘Reputation’

August 2019: Release of studio album ‘Lover’

December 2020: Release of studio album ‘Evermore’

For Justin Bieber, the two biggest spikes correspond to the release of ‘Despacito’ in April 2017 and the release of “No Brainer” in July of 2018. For BTS, the growth in pageviews begins in 2020, when their singles “Dynamite” and “Life Goes On” reach number one on US Billboard Hot 100.

Revisions on Wikipedia give us a more direct measuremeant of collaboration on the platform. Based on revisions alone, it seems that editors who revise Taylor Swift’s Wikipedia page participate in a larger scale of collaboration than the other editors.

With much rarer spikes in the data, it’s difficult to attribute major album release dates with more revisions on Wikipedia. Instead, we speculate that collaboration in the form of revisions most likely happens after an event that hasn’t been pre-announced occurs (for example, winning a Grammy’s award). We theorize that Taylor Swift’s first spikes in page revisions are correlated with the following events:

Late 2007: Hersingles “Our Song” and “Should’ve Said No” reached number one on iTunes

Early 2012: Swift receives 2 Grammy awards at the 54th Annual Grammy Awards

Revision length also gives us a deeper understanding into the rate of collaboration within each revision. Justin Bieber’s revision length history has noticeably more sudden dips, which might hint that editors are having more disagreements and thus need to delete each other’s work. BTS’s revision length history has the fastest growth of the three artists’ pages, suggesting that editors on BTS’s Wikipedia page collaborate on a faster timeline, contributing to more page content in a shorter amount of time than that of Taylor Swift’s or Justin Bieber’s.

Number of Contributions per Wikipedia Editor

Editors on Wikipedia typically don’t edit many pages. Most editors in an article’s edit history only make 1 or 2 revisions, as evident in BTS and Justin Bieber’s editor contributions. However, when plotted on the same scale, there are much more editors on Taylor Swift’s page, with a larger number of editors making more than 1 contribution.

Google Trends: BTS becomes increasingly popular.

For our album-specific trend analysis, we examined both Arts and Entertainment related and general trend data to eliminate any possible misclassifications and attempt to understand the relative interest in our three musical artists in relation to the entertainment industry as well as general topics. Interest is defined daily rather than monthly as in our previous historical search.

Google Trends Arts and Entertainment (left) and General (right) Search Data for three albums (top to bottom)

As we progress through the albums, it is evident that interest in BTS is rising while interest in Swift and Bieber has generally decreased in comparison to their earlier releases.

Twitter: BTS leads in Twitter engagement across all albums.

As previously mentioned, our Twitter analysis primarily focused on identifying dedicated content creators responsible for the promotion and encouragement of Twitter engagement and subsequent collaboration during album releases. Prior to analyzing engagement content, we plotted the number of original (non-retweet) tweets per day during the album release periods to visualize the general creation trends.

Original Tweets per Day for each Album

From these plots, we see an expected trend: the number of tweets per day increases until peaks around the time of the album release before declining throughout the week following. These time-series give us an additional look at the sheer volume of tweets BTS fans are pumping out compared to Taylor Swift and Justin Bieber fans.

In our content generator engagement analysis, we are looking to see which fanbase has the greatest percentage of content generators relative to the total engagement. From applying the content generator analysis across the three album releases for each artist, we are able to estimate the percentage of the total amount of likes, retweets, and replies that were accounted for by them for each of the three albums.

Plots for 0-5% of Users Accounting for B% Engagement for the First Albums

During the time period of the first album, BTS had roughly 255k followers compared to the tens of millions that Taylor Swift and Justin Bieber had. Despite this, there were still a similar amount of unique tweets during the time period as seen in the original tweets figure above. BTS largely comes out on top of all three artists in these curves, meaning that their content generators were responsible for more of their total engagement.

Plots for 0-5% of Users Accounting for B% Engagement for the Second Albums

For the second albums, we see that BTS has begun to pull away, notably leading in both likes and retweets. BTS also has far higher Twitter engagement specifically for original tweets. From our analysis earlier, we see that there are significantly more BTS-related tweets per day, up to double to triple the amount for Swift or Bieber. We expect the denominator for our content generator percentage calculation to be that much larger since there are many more original tweets to be engaged with, but BTS content generators still consistently account for a higher majority of engagement than the other two artists.

Plots for 0-5% of Users Accounting for B% Engagement for the Third Albums

By the time we reach the third album, we again see that the sheer number original tweets per day about BTS has reached an entire order of magnitude higher than totals for either Taylor Swift or Justin Bieber. Despite this, BTS content generators still outperform the other artists’ content generators across the board for our percentage ranges. Through our analyses for these three periods of time, we see that the numbers support our hypothesis: BTS content generators make up more of the total engagement and better enable collaboration between members of the fanbase on Twitter.

Wikipedia: Interest & collaboration on Taylor Swift related pages are higher than others

Album Comparisons for Wikipedia Pageviews during Album Release

We began our comparisons with albums that released after 2015, which is when the MediaWiki API began to collect page view data. We see that for our album two comparisons, the Taylor Swift and Justin Bieber album pages lead in views each day. There’s a general trend where the number of views increases prior to album release, peaks at album release, and decreases in the days after. However, Taylor Swift’s page views on her album pages have quadrupled that of BTS’s, suggesting that interest in her album, at least on Wikipedia, significantly outweighs the interest in BTS’s album and Justin Bieber’s album.

Album Comparisons for Wikipedia Revisions during Album Release

The figure above compares the number of revisions for albums released during a similar time period for the three artists: BTS, Taylor Swift, and Justin Bieber. Although we expected more collaboration on the BTS wikipedia page from our Twitter analysis, the results actually show the opposite. Taylor Swift’s album page leads in the number of revisions, especially on the day of the album release. Revisions on Justin Bieber’s album page show a similar pattern, peaking on the day of his album release and slowly decreases as time goes on. This trend doesn’t appear with BTS’s album page as the number of revisions stays low throughout the 12 day interval.

Album Comparisons for Wikipedia Revision Length during Album Release

In addition, we also explored how the revision length changed over time for these album pages. The same upward slope is seen in all album pages (except for Taylor Swift's Folklore), suggesting a slow but steady effort in collaboration among Wikipedia editors. The article for Taylor Swift’s Folklore has a much steeper upward slope in revision length, suggesting that editors in Taylor Swift’s Folklore page were contributing more content with each revision.