The Israel-Hamas War on YouTube and TikTok: Visibility of videos and their use for humanitarian organisations

Aleks Berditchevskaia, Lequisha Girjasing, Yu Pei, Yushan Sandy Dai, Charlotte Spencer-Smith

Summary of Key Findings

The Israel-Hamas war starting in 2023 has precipitated a surge of social media content directly from and commenting about the conflict, offering various perspectives on the unfolding events. The aim of this project was to identify what kinds of content and which actors are particularly visible on YouTube and TikTok, two popular video platforms. Furthermore, this project aimed to assess to what extent content uploaded to these platforms could help humanitarian organisations identify the needs of civilians in Gaza affected by the war.

The project found that mainstream media organisations were highly visible on both YouTube and TikTok, but alternative media played a greater role on YouTube, while private individuals played a greater role on TikTok. The study utilised a comprehensive search methodology on YouTube and TikTok, focusing on the water crisis in Gaza between October 7th and December 31st, 2023, revealing that YouTube results were older, TikTok content was more recent, and common bigrams on both platforms included "million people" and "Gaza strip," with top TikTok content directly addressing the water crisis whereas top YouTube content referred to US political content. We found that citizen journalists started posting more at the end of November, suggesting that TikTok might be an emerging platform for frontline reporters. The content they post has relevant information for humanitarians, for example, images of damaged infrastructure.

Introduction

Since the Hamas-led attack on Israel on October 7, 2023, a violent conflict has started between Israel and Hamas-led militant groups in the Gaza Strip. As part of the broader Israel-Palestine conflict, this war has sparked a surge of user-generated content on social media platforms. While some of this content shows footage directly from the conflict, other content comments on it or expresses support or opposition to the participants in the war. More footage still highlights the plight of affected civilians living in the Gaza Strip. In particular, videos show audiovisual media and can convey their messages in a persuasive and powerful way. The widespread use of smartphones has allowed users to film and publish their own videos in the Syrian Civil War (2011-), the Russo-Ukrainian War (2014-) and since the Russian invasion of Ukraine in 2022 (Smit et al. 2017, Specia 2022). YouTube and TikTok are globally the two most widely used social media platforms that specialise in video content, with 2.5 and 1.2 billion monthly users, respectively, and therefore have particular power to shape public discourse on the conflict (Statista 2023). There is also potential for humanitarian organisations to find footage directly from the Gaza Strip on social media to gain an initial understanding of the needs of civilians. Considering this context, it is important to understand (a) who and what content is dominating the discourse about the conflict on YouTube and TikTok, as Glaesener (2023) states that mainstream media is present on YouTube, but how present can it be? and (b) to what extent social media has utility for humanitarian organisations seeking to assess civilian needs in the conflict zone.

Initial Data Sets

The initial data sets collected for this research consist of social media metrics, including metadata from YouTube videos related to the conflict. This metadata involves tags, descriptions, view counts, and dates of publication. This provides a temporal context for the impact and reach of the contents. Similar to TikTok video metadata, focusing on the platform’s engagement features, including hashtags and view counts. For textual analysis, we have included video transcripts from both platforms. To identify the common themes and sentiments.

Further, to gain insights into the sources of the narratives, content creator data was collected. This dataset includes YouTube channels and TikTok accounts known for posting content that is relevant to the conflict, the number of videos they posted related to the conflict, and their overall reach. The groups were then categorised into mainstream media, alternative media, citizen journalists, and/or individual influencers who interpret and publish information through their scopes. Besides, this study focuses on specific videos that directly address the humanitarian issues in the region. Those that highlight issues such as access to clean water, food crises, and infrastructure damage. Contents that explicitly aim at raising awareness or providing real-time updates on the humanitarian situation have been observed.

The datasets were set to include the most active period of the conflict, and the data were filtered to only include the content published between October 7 and December 31, 2023. This gives us a focused analysis of the conflict’s peak and its aftermath. The periodization is to evaluate the development of the online discourse as the situation on the ground develops. The extraction of bigrams from video transcripts of both YouTube and TikTok provides a scrutinised view of the discussion, indicating the most recurring topics and concerns.

Research Questions

Which accounts and videos are highly visible in the discourse around the Israel-Hamas War on YouTube and TikTok?

Can humanitarians find relevant information in the videos posted on YouTube and TikTok:

1. by using narrow search terms relevant to humanitarian concerns?
2. by following a curated list of local citizen journalists?

Methodology

To determine the top accounts and videos, we performed a series of relevant keyword searches on YouTube and TikTok using terms such as "Israel," "IDF," "Gaza," "Palestine," "Hamas," and “Israel Palestine.”. These terms were selected to reflect the mainstream discourse surrounding the conflict and are likely to be encountered by average users without specialised knowledge of the subject. In the next step, the metadata for the top 50 results for each search term was scraped for each platform. For YouTube, this was performed using YouTube Data Tools. While for TikTok, search results were scraped using the Zeeschuimer browser extension developed by the Digital Methods Initiative.

As YouTube Data Tools pulls data directly from a YouTube API, Zeeschuimer operates through a web browser; thus, it was necessary to set up a clean research browser for collecting TikTok data to minimise the influence of TikTok’s personalisation algorithms on the search results. The data was then exported to the 4CAT Capture and Analysis Toolkit, a software suite for analysing data also developed by the Digital Methods Initiative. The scraped data was downloaded as a CSV file and aggregated into one dataset for YouTube and one dataset for TikTok.

A co-tag network analysis was performed on this data in Gephi to identify further popular relevant search terms that could be used to collect more relevant data. As a result of this, the search term “#freepalestine” was added to the TikTok search terms. The co-tag analysis of the YouTube data did not reveal any further popular relevant search terms. Because sometimes the same videos were recommended across different searches for YouTube and TikTok individually, any duplicates were removed from the aggregated data. In this data, we identified the top 25 accounts whose videos had generated the most views for each platform. As a broad categorisation, these accounts were grouped by 'mainstream media’, ‘alternative media’, ‘state media’ and ‘other’. The accounts that had accrued the most views through the videos found through the search results were then identified to show who had the most reach in the discourse on the Israel-Hamas conflict.

To understand if the platforms returned relevant results for humanitarians, we focused on the water crisis in Gaza, examining content from YouTube and TikTok between October 7 and December 31, 2023. We developed 5 targeted search queries: “Gaza water crisis”; “Gaza clean water access”; “Gaza clean water access, water crisis”; “Gaza eyewitness, clean water access, water crisis”; “Gaza press heroes, clean water access, water crisis.” We collected TikTok content generated by citizen journalists using a curated list shared by user @designbydeda on Instagram. We employed YouTube Data Tools and Zeeschuimer to collect data. The transcripts for the Tiktok videos were generated using audio transcription with OpenAI’s Whisper model, and for the YouTube transcripts, we used YouTube Data Tools. We created word trees from the transcripts (with a maximum of 3 branches and a window size of 5) and generated bigrams to understand dominant discourses. Video thumbnails were analysed using OpenAI’s CLIP model and visualised by category. All analyses were implemented through modules within the 4CAT data analysis suite.

Findings

Comparing high-visibility content on YouTube and TikTok

Mainstream media is highly viewed on both YouTube and TikTok

The analysis of mainstream media presence on YouTube and TikTok during the Israel-Gaza conflict reveals distinctive trends in content viewership. As was expected, mainstream media was particularly well represented in the YouTube data and generated a significant proportion of the highest views. This is also reflected in particularly prominent videos. A notable example includes a mainstream media video featuring a Hamas fighter shooting at Israeli soldiers, which had the highest number of views in the YouTube dataset. Interestingly, a ‘YouTube Short’ depicting the rescue of a baby from rubble in Gaza appeared as the most frequently recommended piece across various search terms. This trend aligns with the platform’s previous statements about wanting to promote authoritative news sources (Porter 2023). Similar to YouTube, the results from TikTok were dominated by mainstream media and the accounts of private individuals. The Daily Mail generated views from multiple videos, particularly a video about Taylor Swift’s bodyguard returning to Israel to serve in the army.

Alternative media are more important on YouTube than on TikTok

In contrast to YouTube, TikTok’s data set shows a noticeably lower representation of alternative media. On YouTube, alternative media sources, including religious news sources like TBN Israel and Muslim Central, as well as shows like Piers Morgan Uncensored, were prominently viewed. However, this contrasts with the TikTok data, where alternative media did not play a role in highly-viewed content.

Graph 1: Most viewed videos and accounts on TikTok search results for the Israel-Hamas war

Graph 2: Most viewed videos and accounts on YouTube search results for the Israel-Hamas war

Private individuals have greater impact on TikTok than on YouTube

In comparison to YouTube, the accounts of private individuals played a significantly greater role, with two ‘viral videos’ from two separate private accounts generating the highest numbers of views in the dataset. The first video garnered 44 million views and showed a female Israeli soldier performing a well-known TikTok meme to express support for Israel. The second video achieved a similar number of views and showed a woman in a hijab performing the same meme in response to the first video, albeit expressing opposition to Israel. A Palestinian user has also accumulated views through multiple videos in which he challenges Israelis he encounters on a video chat app.

Image 1: Thumbnails viral videos

Relevant information for humanitarians on YouTube and TikTok

The recency of the content on YouTube and TikTok

To assess the recency of the material found for our specific search terms, we categorised the videos into weekly groups. Notably, YouTube search results consisted mainly of videos from the beginning of the crisis, while TikTok showed a more even distribution, with the exception of a notable spike in week 6 (6/11/2023–12/11/1023). An interesting trend emerged when citizen journalists started producing more content in late November and posting to TikTok, indicating an emerging TikTok channel for on-the-ground reporting. These temporal differences highlight the evolving nature of content creation and consumption on these platforms over the observed period.

Graph 3: % of total videos returned by search on YouTube

Content on YouTube leans towards politics; content on TikTok is about the water crisis

The transcript analysis revealed a notable difference in focus between YouTube and TikTok. YouTube's content focused primarily on the Gaza water crisis, while TikTok's content focused more on the border crisis. A word tree generated using the search query "water" highlighted the differences between the two platforms. While both platforms contained relevant information about the water crisis, looking at the top 10 bigrams revealed a striking difference. The top 10 results on YouTube were mainly related to political contexts, often referencing US politics. In contrast, the top 10 results on TikTok were directly related to the water crisis, with common bigrams such as “million people '' and “Gaza strip,” highlighting the diverse narrative nature of the video content on each platform.

Image 2: Word tree after querying YouTube

Image 3: Word tree before querying TikTok

Image 4: Top 10 bigrams from video transcripts

Valuable insights on TikTok for humanitarians

A carefully curated TikTok scrape of content from citizen journalists in Gaza revealed a significant concentration of images depicting damage to infrastructure. This collection added a visual dimension to understanding the crisis. In our final research, we focused on the visual content of videos produced by citizen journalists, limiting our analysis to thumbnail images due to time constraints. We have categorised these images into pre-selected categories that are relevant to humanitarian issues: rubble, roads, water, food, and tents. The findings indicate that by following citizen reporters, humanitarians can gain valuable insights into the damage suffered by local infrastructure. However, the visual content provides less information on aspects related to food and water in the examined dataset.

Image 5: Categorised TikTok thumbnails

Discussion

In assessing the digital discourse surrounding the Israel-Gaza conflict, this project encounters certain limitations that reflect the broader challenges faced in digital methods and platform research. One of the limitations is the focus on analysing high-visibility content, which overlooks less visible content that is still impactful on the discourse around the Israel-Gaza conflict. These contents can significantly influence user groups through platform personalization algorithms; as mentioned by Rieder et al. (2018) and Brennan (2020), both YouTube and TikTok are well-known for their personalised recommendation algorithms. Indeed, YouTube claims that its recommendation algorithm drives 70% of watch time (Solsman 2018).

Despite our best efforts to neutralise bias searches by utilising a clean research browser, we recognize that our method cannot perfectly mirror the dynamic and personalised content ecosystem that most users navigate. Typically, users do not just stumble upon content through deliberate searches but are led by the invisible hand of algorithms that curate a unique digital experience (Boeker & Urman, 2022). For this reason, while the study of popular, high-visibility content is important to public discourse, this project has not taken into account the impact of content with individually lower views that nonetheless has a lot of impact, either in aggregate or in personalised user experiences. Future research could seek to address this issue by simulating particular user journeys or exploring particular sub-discourses.

Moreover, this project relied on search results to measure content visibility, bringing us face-to-face with another limitation: the possibility of overlooking content that, while highly relevant and viewed, does not make it to the search result spotlight. Without access to tools like earlier versions of Facebook’s Crowdtangle, which enabled researchers to find the most popular content directly through a platform API, this project has relied on search results. It is possible, and even likely, that there is high-visibility, high-view count content on both YouTube and TikTok relevant to the Israel-Gaza conflict that is not being shown in search results.

This points to a central problem of platform research: it is not possible to know with certainty which content is most popular on platforms as long as these platforms restrict data access (McNulty, 2023). While innovative projects like the TikTok Global Observatory (https://tkgo.aiforensics.org/) work around this using digital methods, this can only give us a very top-level and approximate impression of which content is popular. The lack of data access has significant implications for monitoring the kinds of public discourse that reflect and inform public opinion on geopolitical issues, including the Israel-Gaza conflict.

Conclusion

The exploration of high-visibility content on YouTube and TikTok during the Israel-Gaza conflict reveals intriguing patterns. The influence of mainstream media, alternative sources, and private individuals contributes to the varied nature of information dissemination on these platforms. Unravelling these dynamics adds nuance to the multifaceted discourse surrounding geopolitical events. The digital landscape continues to shape how we perceive and engage with global issues, highlighting the need for ongoing research to understand the evolving nature of online information dissemination. As anticipated, mainstream media holds a commanding presence on both YouTube and TikTok. The prominence of mainstream media in generating high views underscores its influence in shaping the narrative. Notably, a YouTube video depicting a Hamas fighter garnered the highest views, showcasing the platform's role as a repository for varied perspectives. TikTok, aligning with its commitment to authoritative news sources, prominently featured mainstream media. Mainstream media organisations can extend and dominate their influence on both offline and online platforms, but the relevance algorithm on YouTube may be biassed toward mainstream media (Glaesener, 2023).

A notable contrast emerges concerning the significance of alternative media. While YouTube witnessed the influence of alternative sources like religious news channels and Piers Morgan’s uncensored talk shows, TikTok remained relatively untouched by such content in the high-visibility spectrum. This divergence underscores the platforms’ unique ecosystems and user preferences, reflecting a nuanced interplay between content and audience. The role of private individuals in shaping discourse is pronounced on TikTok. Unlike YouTube, where private channels play a limited role, TikTok showcases the power of individual creators. Viral videos from private accounts expressing both support and opposition to Israel underscore the platform’s potential for individualised narratives, adding a personal dimension to the broader geopolitical discourse. Two distinct private accounts secured the highest click-through rates in our dataset. The initial video, which amassed 44 million views, featured a TikTok meme where a female Israeli soldier expressed support for Israel. A subsequent video, garnering comparable viewership, showcased a woman in a hijab responding to the first video with the same meme but expressing opposition to Israel. Additionally, a Palestinian user garnered significant engagement by challenging Israelis in multiple videos through a video chat app.

Our research into YouTube and TikTok content related to the Gaza water crisis provides valuable insights for humanitarian workers. The temporal analysis shows a notable difference in recency between the platforms, with YouTube dominated by past crisis videos, while TikTok shows a more even spread of content, indicating an emerging channel for on-the-ground reporting. This highlights the dynamic evolution of content creation and consumption on both platforms over the observed period. Furthermore, a significant deviation in content focus was observed between YouTube and TikTok. YouTube focuses primarily on the political aspects of the Gaza water crisis, while TikTok focuses more broadly on the crisis itself, as evidenced by the top 10 bigrams that highlight the diverse narrative nature of the platforms. This nuanced understanding of the substantive differences is essential to understanding the multifaceted perspectives of these platforms.

TikTok in particular is emerging as a valuable resource for humanitarians, as a carefully curated collection of content from citizen journalists reveals a plethora of images depicting damage to infrastructure. While this visual dimension provides crucial insights into the crisis, our examination of the thumbnails indicates that there is relatively little information on aspects related to food and water. So, while TikTok offers unique perspectives on the crisis, there is room to expand its coverage to provide a more comprehensive humanitarian view.

While our study offers valuable insights, it does come with certain limitations. The concentration on high-visibility content may capture a segment of platform dynamics, yet it falls short of encapsulating the complete spectrum, especially given the prevalence of personalised algorithms on both YouTube and TikTok. Looking forward, future research endeavours should extend their purview to encompass lower-visibility content, unravelling its influence on user experiences and contributing to a more holistic understanding of the evolving digital dialogue.

In conclusion, our research navigates the multifaceted landscape of YouTube and TikTok, illuminating the distinct roles played by mainstream and alternative media, as well as individual creators. As these platforms shape the narrative surrounding the Israel-Gaza conflict, acknowledging their unique dynamics becomes pivotal for comprehending the intricate interplay between content, platforms, and user engagement. By balancing these insights and limitations, future studies can further untangle the complexities of online discourse, providing a roadmap for navigating the evolving landscape of digital communication.

References

Boeker, M., & Urman, A. (2022). An Empirical Investigation of Personalization Factors on TikTok Proceedings of the ACM Web Conference 2022, Virtual Event, Lyon, France. https://doi.org/10.1145/3485447.3512102

Brennan, M. (2020). Attention factory: The story of TikTok and China's ByteDance. China Channel.

Glaesener, T. (2023). Who informs Germans about the Russia-Ukraine War on YouTube? The Journal of Social Media in Society, 12(1), Article 1. https://thejsms.org/index.php/JSMS/article/view/1191

McNulty, T. (2023). What’s on top of TikTok. Public Books. Retrieved 16.01.2024 from https://www.publicbooks.org/whats-on-top-of-tiktok/

Porter, J. (2023). YouTube wants to get you watching more news from ‘authoritative sources’. The Verge. Retrieved 16.01.2024 from https://www.theverge.com/2023/10/18/23922070/youtube-news-immersive-watch-page-experience-shortform

Rieder, B., Matamoros-Fernández, A., & Coromina, Ò. (2018). From ranking algorithms to ‘ranking cultures’:Investigating the modulation of visibility in YouTube search results. Convergence-the International Journal of Research into New Media Technologies, 24(1), 50-68. https://doi.org/10.1177/1354856517736982

Smit, R., Heinrich, A., & Broersma, M. (2017). Witnessing in the new memory ecology: Memory construction of the Syrian conflict on YouTube. New Media & Society, 19(2), 289-307. https://doi.org/10.1177/1461444815604618

Solsman, J. (2018). YouTube's AI is the puppet master over most of what you watch. Cnet. Retrieved 12.10.21 from https://www.cnet.com/news/youtube-ces-2018-neal-mohan/

Specia, M. (2022). ‘Like a Weapon’: Ukrainians Use Social Media to Stir Resistance. New York Times. Retrieved 12.01.2024 from https://www.nytimes.com/2022/03/25/world/europe/ukraine-war-social-media.html

Statista. (2023). Most popular social networks worldwide as of October 2023, ranked by number of monthly active users. Retrieved 12.01.2024 from https://www.statista.com/statistics/272014/global-social-networks-ranked-by-number-of-users/

-- Main.BernRieder - 19 Feb 2024