skip to main content


Title: "Give Everybody [..] a Little Bit More Equity": Content Creator Perspectives and Responses to the Algorithmic Demonetization of Content Associated with Disadvantaged Groups
Algorithmic systems help manage the governance of digital platforms featuring user-generated content, including how money is distributed to creators from the profits a platform earns from advertising on this content. However, creators producing content about disadvantaged populations have reported that these kinds of systems are biased, having associated their content with prohibited or unsafe content, leading to what creators believed were error-prone decisions to demonetize their videos. Motivated by these reports, we present the results of 20 interviews with YouTube creators and a content analysis of videos, tweets, and news about demonetization cases to understand YouTubers' perceptions of demonetization affecting videos featuring disadvantaged or vulnerable populations, as well as creator responses to demonetization, and what kinds of tools and infrastructure support they desired. We found creators had concerns about YouTube's algorithmic system stereotyping content featuring vulnerable demographics in harmful ways, for example by labeling it "unsafe'' for children or families -- creators believed these demonetization errors led to a range of economic, social, and personal harms. To provide more context to these findings, we analyzed and report on the technique a few creators used to audit YouTube's algorithms to learn what could cause the demonetization of videos featuring LGBTQ people, culture and/or social issues. In response to the varying beliefs about the causes and harms of demonetization errors, we found our interviewees wanted more reliable information and statistics about demonetization cases and errors, more control over their content and advertising, and better economic security.  more » « less
Award ID(s):
2040942
PAR ID:
10391666
Author(s) / Creator(s):
; ; ; ;
Date Published:
Journal Name:
Proceedings of the ACM on Human-Computer Interaction
Volume:
6
Issue:
CSCW2
ISSN:
2573-0142
Page Range / eLocation ID:
1 to 37
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. To manage user-generated harmful video content, YouTube relies on AI algorithms (e.g., machine learning) in content moderation and follows a retributive justice logic to punish convicted YouTubers through demonetization, a penalty that limits or deprives them of advertisements (ads), reducing their future ad income. Moderation research is burgeoning in CSCW, but relatively little attention has been paid to the socioeconomic implications of YouTube's algorithmic moderation. Drawing from the lens of algorithmic labor, we describe how algorithmic moderation shapes YouTubers' labor conditions through algorithmic opacity and precarity. YouTubers coped with such challenges from algorithmic moderation by sharing and applying practical knowledge they learned about moderation algorithms. By analyzing video content creation as algorithmic labor, we unpack the socioeconomic implications of algorithmic moderation and point to necessary post-punishment support as a form of restorative justice. Lastly, we put forward design considerations for algorithmic moderation systems. 
    more » « less
  2. Social media users have long been aware of opaque content moderation systems and how they shape platform environments. On TikTok, creators increasingly utilize algospeak to circumvent unjust content restriction, meaning, they change or invent words to prevent TikTok’s content moderation algorithm from banning their video (e.g., “le$bean” for “lesbian”). We interviewed 19 TikTok creators about their motivations and practices of using algospeak in relation to their experience with TikTok’s content moderation. Participants largely anticipated how TikTok’s algorithm would read their videos, and used algospeak to evade unjustified content moderation while simultaneously ensuring target audiences can still find their videos. We identify non-contextuality, randomness, inaccuracy, and bias against marginalized communities as major issues regarding freedom of expression, equality of subjects, and support for communities of interest. Using algospeak, we argue for a need to improve contextually informed content moderation to valorize marginalized and tabooed audiovisual content on social media.

     
    more » « less
  3. How social media platforms could fairly conduct content moderation is gaining attention from society at large. Researchers from HCI and CSCW have investigated whether certain factors could affect how users perceive moderation decisions as fair or unfair. However, little attention has been paid to unpacking or elaborating on the formation processes of users' perceived (un)fairness from their moderation experiences, especially users who monetize their content. By interviewing 21 for-profit YouTubers (i.e., video content creators), we found three primary ways through which participants assess moderation fairness, including equality across their peers, consistency across moderation decisions and policies, and their voice in algorithmic visibility decision-making processes. Building upon the findings, we discuss how our participants' fairness perceptions demonstrate a multi-dimensional notion of moderation fairness and how YouTube implements an algorithmic assemblage to moderate YouTubers. We derive translatable design considerations for a fairer moderation system on platforms affording creator monetization. 
    more » « less
  4. null (Ed.)
    Lecture videos are rapidly becoming an invaluable source of information for students across the globe. Given the large number of online courses currently available, it is important to condense the information within these videos into a compact yet representative summary that can be used for search-based applications. We propose a framework to summarize whiteboard lecture videos by finding feature representations of detected handwritten content regions to determine unique content. We investigate multi-scale histogram of gradients and embeddings from deep metric learning for feature representation. We explicitly handle occluded, growing and disappearing handwritten content. Our method is capable of producing two kinds of lecture video summaries - the unique regions themselves or so-called key content and keyframes (which contain all unique content in a video segment). We use weighted spatio-temporal conflict minimization to segment the lecture and produce keyframes from detected regions and features. We evaluate both types of summaries and find that we obtain state-of-the-art peformance in terms of number of summary keyframes while our unique content recall and precision are comparable to state-of-the-art. 
    more » « less
  5. Abstract. With TikTok emerging as one of the most popular social mediaplatforms, there is significant potential for science communicators tocapitalize on this success and to share their science with a broad, engagedaudience. While videos of chemistry and physics experiments are prominentamong educational science content on TikTok, videos related to thegeosciences are comparatively lacking, as is an analysis of what types ofgeoscience videos perform well on TikTok. To increase the visibility of thegeosciences and geophysics on TikTok and to determine best strategies forgeoscience communication on the app, we created a TikTok account called“Terra Explore” (@TerraExplore). The Terra Explore account is a jointeffort between science communication specialists at UNAVCO, IRIS(Incorporated Research Institutions for Seismology), and OpenTopography. Weproduced 48 educational geoscience videos over a 4-month period betweenOctober 2021 and February 2022. We evaluated the performance of each videobased on its reach, engagement, and average view duration to determine thequalities of a successful video. Our video topics primarily focused onseismology, earthquakes, topography, lidar (light detection and ranging),and GPS (Global Positioning System), in alignment with our organizationalmissions. Over this time period, our videos garnered over 2 million totalviews, and our account gained over 12 000 followers. The videos thatreceived the most views received nearly all (∼ 97 %) oftheir views from the For You page, TikTok's algorithmic recommendation feed. Wefound that short videos (< 30 s) had a high average view duration,but longer videos (> 60 s) had the highest engagement rates.Lecture-style videos that were approximately 60 s in length had moresuccess in both reach and engagement. Our videos that received the highestnumber of views featured content that was related to a recent newsworthyevent (e.g., an earthquake) or that explained location-based geology of arecognizable area. Our results highlight the algorithm-driven nature ofTikTok, which results in a low barrier to entry and success for new sciencecommunication creators. 
    more » « less