Searching inside YouTube Search
Organisation | Vrije Universiteit Brussel, imec-SMIT
My Role | Researcher, Author
How does YouTube shape and reflect online public discourse on key sociocultural issues?
This is an overview. Download the complete study here.
Note: The end-to-end research for this project was carried out as part of the Erasmus Mundus Joint Master’s Degree in Digital Communication Leadership (delivered by University of Salzburg, Vrije Universiteit Brussel and UCLA California).
Background.
There’s no question that YouTube is now widely regarded as part of the mainstream media landscape (Burgess and Green, 2009). Launched in 2005 and acquired by Google Inc. a year later, it is the second most visited site in the world, with more than a billion users watching hundreds of thousands of hours of content every day. Apart from being a huge resource for user-generated content, it has become one of the world’s most-visited platforms for content discovery and public engagement: a primary keeper of the cultural discussion (Gillespie, 2010).
Over the years, as broadband internet has become more pervasive the world over, as social media has become entrenched with everyday life, and as traditional broadcast television has taken a backseat, YouTube seems to have taken centre stage when it comes to public debate and discussion on hot topics. To understand their social impact and hold them accountable, scholars are exploring new methods to study their power.
This research proposes an alternative approach to examining what YouTube’s specific strain of ranking mechanism does, not only to understand their inner workings, but also to explore the socio-technical combinations at work in shaping public debate on an online platform and the broader forms of influence they wield.
Some content deemed more ‘relevant’.
The search function is an area where YouTube’s platform mediation become explicit.
For instance, the way that search results are displayed in rank-ordered lists makes visible that some content is deemed more relevant’ by the platform than others. Metadata, such as views, likes, and comment counts, also influences how information spreads on YouTube. In turn, users take these indicators into account when they view and share content, making them an important factor in how YouTube's algorithms rank videos.
This means that the ranking process is not a simple one, but rather a complex and ongoing process that requires in-depth study.
The search function is an area where YouTube’s platform mediation become explicit. For instance, the way that search results are displayed in rank-ordered lists makes visible that some content is deemed more relevant’ by the platform than others. Metadata, such as views, likes, and comment counts, also influences how information spreads on YouTube. In turn, users take these indicators into account when they view and share content, making them an important factor in how YouTube's algorithms rank videos.
This means that the ranking process is not a simple one, but rather a complex and ongoing process that requires in-depth study.
The search ranking process.
From Edward Snowden’s NSA revelations to the Cambridge Data Analytica scandal around Facebook, the question of data, for instance, has split the online discourse on issues related to agency, social media, consent, fairness, security, freedom of speech, democracy as well as regulation and anti-trust issues.
As a result, terms like ‘data privacy’ and ‘data protection’ serve as access points for a wide range of public issues each time a search is launched on YouTube.
Making
YouTube Search empirically visible.
Carried out in collaboration with the imec-SMIT-VUB research group, under the mentorship of Prof. Laurence Claeys, this mixed-methods research focuses on how YouTube’s search affordance influences the shaping and reflection of public discourse online on two key issues (1) Data Privacy, and (2) Data Protection. Working from Rogers’ (2009) theory of repurposing online objects for social and cultural research, the goal here is to use YouTube’s search function to study how a particular platform affordance organises the ranking of specific socio-cultural issues as a socio-technical assemblage of multiple factors.
Aims of the study.
To understand the complicated role that YouTube plays in algorithmically ranking viewpoints, I aim to:
-
Investigate the dynamics of YouTube search results ranking (highest ranking videos, N=20) in the context of the two above-mentioned issues across a specified timeline (20/11/2021 to 20/12/2021).
-
Qualitatively explore the content and contexts of these videos, and their bearing on the ranking system.
Utilizing APIs to harvest data from YouTube, my central idea here is to analyse what kind of discourse is assembled on this platform (co-curated by both algorithms and human agency) for anyone who seeks out content related to the topics of data privacy and/or data protection.
Methods.
Through rank visualisations, computational change metrics and in-depth qualitative analysis, the study emphasises how YouTube’s platform affordances are fundamental in how the search function (which provides an ordered list of videos in response to a search query) constantly defines and redefines the ‘ranking culture’ of certain issues across time.
In doing so, I illustrate how YouTube’s ranking contributes to shaping and reflecting online public discourse on two contemporary and politically charged issues (namely, ‘Data Privacy’ and ‘Data Protection’).
Findings.
Findings indicate the ranking morphologies of Data Protection and Data Privacy to be relatively stable over time, with no extraordinary or transformative public events hijacking the search volume or trend.
The most noticeable change in rankings happen within a set of low-ranking videos as opposed to a complete displacement of videos on the top of the list. To this end, the findings confirm that vanity metrics such as likes, comments and view counts are not the key factors that shape the ranking apparatus of YouTube: amateur videos with low view counts uploaded by little-known channels retain their high-ranking spots in both cases.
The study also illustrates how the issue of [Data Protection] is largely influenced by legal and institutional actors, whereas in the case of [Data Privacy], the online public discourse is driven by a wider variety of actors uploading content.
In the second step of the analysis, key video types were identified and analysed with close attention to similarities and differences in context, tone, and framing.
Findings.
Here, it was found that official news sources have a stronger impact on shaping online public debate around ‘data protection’ as opposed to ‘data privacy’, where the debate is largely driven by YouTubers and software firms.
That is, while the query [Data Protection] is mostly reflective of the regulations and legalities around data protection, [Data Privacy] is co-opted by several stakeholders (users, technology companies, civil society experts) as a heavily contested idea.
Furthermore, the dominance of YouTube-native content in the Top 20 results for the studied period provides more nuance to our understanding of how YouTube favours native content over external content.
The analysis also points to how certain YouTube subcultures can dominate search inputs as broad as [Data Privacy] and [Data Protection] both in terms of creator tactics and search ranking.
In summary.
In summary, the mixed methodological analysis makes partly visible YouTube’s ranking apparatus and gets us one step closer to the sociotechnical assemblage that organises public discourse on key social issues such as Data Protection and Data Privacy.
I close by arguing that the ranking of key sociocultural issues on YouTube is not a normative phenomenon that can be separated from the larger contextual circle of platform affordances, creator strategies and issue vernaculars.
Scaling this framework to a wider array of search queries and video genres with added emphasis on how linguistic differences affect search cultures will be an interesting way forward for future studies in platform politics.
Additionally, how a specific platform organises different meanings around an issue through other affordances (e.g., recommendations, subscriptions, cross-platform sharing, copyright enforcement) opens up a whole new site of ethnographic study. This could be extended as far as categorising broad subjects (e.g., productivity, fitness, music) according to how prone to ranking change they are.