TikTok

TikTok is a mobile phone based app, owned by ByteDance, that rose exponentially in popularity in 2020. The app's global user-base posts short videos, up to three minutes long, that appear to an audience of followers and new viewers. Which, and how many, new viewers are reached is determined by an innovative relevance algorithm similar to those used by Instagram and Youtube. The app's design encourages users to encounter new content instead of searching for it, though search functions do exist for those users looking to explore specific types of content.

TikTok's Search Functions


On the app, the user must navigate to the “Discover” button at the bottom of the phone screen to access the search function. The search bar is located at the top of the screen.


When a user taps into the search bar, their past four previous searches appear, along with algorithmically generated “Suggested Searches”.

Search results are categorized, along the top, underneath the search bar, as the following:

  • Top

  • Users

  • Videos

  • Sounds

  • LIVE

  • Hashtags

On Desktop: The search bar appears at the top of the landing page, and every other page, on the desktop version of TikTok. When you type in a keyword, phrase, or term, the search results include relevant keywords/phrases and accounts. Users cannot search by Sound on Desktop.


Searching by Keyword/Phrase


Suggested Searches are made up of key words/phrases. Users can search by keyword or phrase.


If a user chooses one of the Suggested Searches, i.e. “Sound Healing”, after they scroll they will see an “Others searched for” section with suggestions.


After tapping “Sound Healing”, then navigating back to the search window, the user’s Suggested Searches refresh.

Searching by Hashtag


Users can also search by hashtag.


When a user searches #soundhealing, they recieve suggested searches for both hashtagged words and key phrases. TikTok appears to be encouraging users to search by keyword/phrase.


Creators overwhelmingly continue to hashtag their videos, however, they may not always be tagging the most relevant key terms or phrases. Instead, creators are compelled to use branded and popular hashtags as a means to monetize their work (Hauschild, Smith).

Searching by Sounds


If the user knows the title of the sound, they can search its title and the sound will appear.


However, the app’s design encourages users to encounter new sounds on their For You/Following streams instead of centering them on the “Discover” page (Hauschild, 2020; Smith, 2021). Creators and users alike are compelled to follow creators who produce their own viral original sounds, as these trendsetters literally set the tone for the app at any given moment.


It is possible for the creator to categorize their video using a sound that cannot be heard by the viewer. Creators often approach sounds similar to hashtags – utilizing popular sounds will allow their content to reach a broader audience, even if the sound is not audible in the video.

Searching by Handles


Creators identify their accounts with handles (aka usernames). Users can search for specific creators by their handle or username. They are not able to search by creator’s first/last name, nor with other information provided by the creator in their bio.

How does search work?

When users search a term, such as “books” for example, they first land on a “Top” page. One would expect these videos to be organized by the most to least number of likes, but that is not the case. As of an 04/25/2022 search, the first two video results that appear have 262.4K and 291.0K likes, respectively. Then, the following sponsored video has only 2 likes. This is followed by a video with 239.6K likes. Once the user scrolls past the “Other users searched for” box, they see videos with 437.0K and 1.6M likes. Predictably, those videos with more likes were uploaded by creators with more followers. The majority of the results that appear under “Top” have over 200K likes.

Organization of Search Results by Other Popular Features

  • “Users” appears to be organized by relevance, with those users I am following appearing first.

  • “Videos” yields similar results as “Top”, with many of the same videos appearing on page one of the search.

  • “Sounds” yields those sounds with the highest prevalence of the search term in the title.

  • “LIVE” yields live videos related to the search term - though it seems the search term does not have to appear in the user’s handle.

  • “Hashtags” also appears to be organized by relevance.

TikTok’s official website and help page does not offer any information as to how a video’s relevance is determined. A Google Search reveals blog writers speculating, often anecdotally, on how to respond to the algorithm to reach a wider audience, and therefore receive more likes and views.

Problematizing the TikTok Algorithm

When contacted by the Australian Strategic Policy Institute (ASPI) in 2020 amidst concerns around censorship and shadowbanning on the platform, a TikTok spokesperson revealed that hashtags and terms such as “ “#гей (“Gay” in Russian)” and “# متحول جنسياً” (“Transgender” in Arabic - edited for correct translation) had been “partially restricted due to relevant local laws.” Further, they noted other hashtags and phrases had been “incorrectly moderated.” (Ryan, Fritz, & Impiombato, 2020)

The report also notes the English acronym “acab (all cops are bastards)” has been shadowbanned by the platform during protests in response to police brutality in the US. Using a conglomeration of view numbers, ASPI concluded, “search results for #acab…were suppressed in the early days of the George Floyd protests.” This pattern was noted by users, who raised their voices against the censorship. On May 29th, 2020, “TikTok backed down, making several hashtags related to the protest, including #acab, available.” However, the ASPI found that “three months later, as anti-racism and anti-police-brutality protests took place in Kenosha, Wisconsin, following a police shooting of black man Jacob Blake, the #acab hashtag was censored once again after media scrutiny subsided.” (Ryan, Fritz, & Impiombato, 2020)

TikTok’s nebulous organization of search results and lack of transparency concerning how "relevance" is determined - in search functionality and in the For You stream alike - allows the platform plausible deniability when it comes to issues of shadowbanning (particularly creators of color and activists) and may lead to de facto censorship at a global scale.

Takeaways

For the average and informed user alike to maintain confidence in navigating the platform ethically, TikTok's must be willing to more openly discuss the mechanisms of their "relevance" algorithm. The brand's hesitance to release what it deems to be a proprietary trade secret is apparent, however, reputational damage to the brand will continue if marginalized users continue to experience shadowbanning with no clarity as to why.