YouTube flow and the transmission of heritage: The interplay of users,content,and algorithms

Theorizing YouTube, diversity, and flow

Social media platforms, like the Internet in general, were initially championed as democratic and egalitarian media. The progressive potential of social media sites such as YouTube emerges through processes that enable users and audiences to produce culture (Bruns, 2008; Jenkins, 2008: 53). Jenkins’ (2008: 268) stance on participatory culture, including social media, outlines how platforms such as YouTube promote cultural diversity, thereby countering dominant perspectives through the creativity of users. YouTube in turn enables the creation and dissemination of grassroots media (Jenkins, 2008: 274). On the grounds that Jenkins’ (2008: 268) analysis overlooks the ownership and business models of YouTube and does not take into account that the culture of social media is part of a cultural industry embedded within a political economy, Fuchs (2014: 57) is critical of Jenkins’ approach to YouTube as participatory culture that advances diversity. A political-economic analysis of YouTube focuses on the interconnections between the platform owners, labor, consumers, and advertisers as well as on the power relations embedded in these interconnections. It addresses how the platform receives profits through the exploitation of users’ labor to benefit financially the dominant class of its corporate owners through monetization geared to individual user data and participation (Fuchs, 2014: 105). According to Fuchs, Jenkins bypasses the corporate ownership that turns user participation and creativity into profits. For Fuchs, the possibility that YouTube can produce divergent cultural representations through user involvement is always curtailed by user exploitation and monetization.

Rogers’ (2013: 85, 119) work on Google’s search engine, a part of his digital methods for analyzing the Internet, suggests that ‘source hierarchies and exclusionary practices’ in the working of search engines render unlikely the divergence of representations disseminated by search engine results. Even though YouTube’s search engine differs from that of Google, Rogers’ assessment of Google’s search engine results is applied in a general sense here: both engines privilege content at the top of a search engine result page (SERP). According to Rogers (2013: 86), Google’s search engine does not offer ‘a collision space for alternative accounts of reality’. Having analyzed the top of specific SERPs over a period of time, Rogers proposes that search engine results relate a type of story that does not provide a range of viewpoints. This story is ‘that of the current status of the topic or issue in question through the organizations currently representing it, on the record, in the engine returns’ (Rogers, 2013: 86). Search engines produce cultural hierarchies and suppress democratic perspectives in favor of dominant perspectives and actors. Rogers’ (2013: 21) digital method focuses on the analysis of digital data as a vehicle ‘to diagnose cultural change and societal conditions by means of the Internet’. As Rogers (2013: 21–23) notes, ‘Knowledge claims may be made on the basis of data collected and analyzed by devices such as search engines’.

Each of the media theories outlined – the participatory theory of Henry Jenkins, the political-economy theory of Christian Fuchs and the digital methods theory of Richard Rogers, rooted in the epistemological potential of digital data analysis – approaches the issue of diverse cultural representation in digital media from a partial frame of reference. Theorizing the partiality of these approaches here is inspired by van Dijck’s (2013a: 128–129) stance that conceptual gaps arise when social media are analyzed from a singular vantage point. Henry Jenkins’ approach can provide only a limited understanding of YouTube because it fails to take into account YouTube’s business strategies and corporate ownership (van Dijck, 2013a: 128). Political-economic approaches that emphasize the corporate structure and business models of social media tend to overlook the impact of technology and underestimate the impact of users, content, and cultural form (van Dijck, 2013a: 128–129). In the analysis of the stories and topics that are conveyed by digital media, the digital methods approach does not integrate in-depth qualitative analysis of these stories or topics outside the source of the Internet or beyond the analysis of the digital data (Rogers, 2013: 23). Rogers’ (2013: 21) digital methods could lead to superficial conclusions when used to realize the goal of his method to attain knowledge of cultural change and societal conditions via the Internet.

José van Dijck provides a comprehensive approach to analyzing YouTube that interrelates the platform’s business models, corporate ownership, users, cultural form, content, and digital technologies. van Dijck’s (2013a: 119) analysis of the content of YouTube videos deals with categories that characterize YouTube videos in general, including a video’s status as user-generated content, as professional-generated content or as a ‘snippet’ of content designed for recycling, as well as its role in ‘storing, collecting, and sharing’. Her multifaceted approach to social media informs the theoretical stance of the present research, which broadens the attention to content through in-depth analyses of video content. The specific focus of this study diverges from van Dijck’s approach in its emphasis on the actual content of individual videos tabulated into lists. The hierarchical ordering of videos into lists is governed by factors integrated within her approach, namely business models, audience use patterns, and technologies, including algorithms. The privileging of video content, which determines what is actually represented in videos, must be considered in assessing whether YouTube can enable diverse representations to come into contact, creating relational juxtapositions that become meaningful through human interpretation. That YouTube brings together divergent heritage narratives, including grassroots perspectives that counter official renditions, emerges in the ways in which television flow is applied and revised within the context of the platform.

At the beginning of the 1970s, movement appeared so essential to television that the cultural theorist Raymond Williams (1974) coined the term ‘flow’ not only to describe the movement of the camera within television programs themselves but also to depict the way that programs merged one into another through scheduling and audience consumption patterns (Spigel, 2001: 62). This movement that Williams identified in television resonates in the dynamism and motion within the transmission and reception of YouTube videos. However, flow on YouTube also diverges from television flow and its adaption to shifting technologies. YouTube flow is characterized by a stream of continuous videos that combine ‘relevant’ videos, recommended videos, and advertisements that appear in the lists of videos, including lists of up-next videos under particular search requests. Thus, flow on YouTube differs from Raymond Williams’ concept, as the content of the YouTube stream is not a product of broadcasting as ‘one-way and unconditional for anyone as someone anywhere anytime’ (Scannell, 2014: 44, original emphasis). Broadcasting is circumscribed to a central sphere and organization that directs the provision and dissemination of signals (van Dijck, 2013b: 149). YouTube does not have a programmed flow defined as an unbroken stream of programs and commercials whose organization is designed to keep the audience fixed to the screen. That the owners of YouTube do not determine the videos that viewers are able to view at particular times via a programmed flow does not imply that the stream of content on YouTube is not mediated (van Dijck, 2013a: 113). The flow of this content is organized and steered by search engines and ranking algorithms. YouTube guides the traffic of videos through an information management system that directs the navigation of users and designates the content that is to be offered and advanced (van Dijck, 2013a: 113). Just as viewers have been able to control the flow of television programming through devices such as the Remote Control Device (RCD) and Video Cassette Recorder (VCR) (Uricchio, 2005: 239, 242), the viewing path that users take through a SERP is determined by users’ choices. YouTube users decide which videos they want to watch under a particular search heading, whether they want to look at the advertisements and whether they want to view the up-next videos. Nonetheless, the videos from which users can choose are provided by YouTube’s information management system, which is composed of search algorithms, ranking systems, and referral mechanisms.

The flow of YouTube content has been described by van Dijck (2013b: 152) as ‘staccato flow’, in reference to the ‘self-selected short videos sequenced by [the] user’s clicks’. This concept of staccato flow has become less relevant since YouTube currently provides the possibility of uploading videos that well exceed the original 15-min limitation, the maximum being 11 h. Despite this substantial increase, the average video length for most popular videos is 4.4 min (comScore, 2014). Furthermore, for a video of 4–5 min, less than 60% of viewers will watch the entire video, in contrast to 75% of viewers who will watch a video of 1–2 min in its entirety (Ruedlinger, 2012). The ways that users move around on YouTube, which include often viewing videos for only 1–2 min, skipping over videos, skipping or streaming the short ads before videos, glancing at lists to make a video selection and providing brief comments, still render the flow on YouTube disconnected. User patterns produce a ‘staccato’ feel that diverges from the more connected sense of flow initially put forward by Raymond Williams (1974).

In March 2015, YouTube added a new autoplay feature that reinforces a more continuous flow. With the autoplay, the up-next video automatically plays once a selected video has ended. Nonetheless, users can ignore this feature and skip to a video of their choice or completely disable the blue autoplay slider. Even though users may skip over the auto-played up-next video and other videos on the up-next list, these videos often appear again as users move through the site under the same or related search headings. The choices of listed videos, including the reappearance of up-next videos that have been skipped over by users, are determined by a management system that governs navigation by users and directs the content to be promoted. As Geert Lovink (2008: 12) elucidates, ‘it is inherent in the interface that we keep going and going and the clip chain continues forever’. Because the algorithms that rank videos under particular headings integrate audience use patterns, user behavior influences the working of algorithms that are designed to monetize the labor of YouTube users. In this sense, YouTube flow resembles the way that digital technology has changed television flow from viewer controlled to algorithm controlled or to a combination of the two (Andrejevic, 2009: 35).

The continuous presence of new tabulated lists of videos, produced through the combined effort of user-generated content and algorithms, constitutes the platform’s flow. This flow emulates Berry’s (2011) idea that the notion of pages, through which the Internet has typically been conceptualized, is giving way to the concept of ‘streams’. The retrieval of information via a specific machine to obtain data in accordance with required demands is being supplanted by ‘an ecology of data streams that forms an intensive information-rich computational environment’ (Berry, 2011: 143). Berry conceptualizes real-time streams as dispersed narratives that, despite their disconnected and fragmented nature, are moving across and within multiform media. The user in turn is part of this process. The activity and communication of a user are turned into data, or a stream that plugs its narrative information into the cloud. This stream is evaluated, combined, and rechanneled back to the user as well as other users in the form of patterns of data (Berry, 2011: 144). Therefore, audience use patterns feed into the data flow of YouTube by producing content as well as responding to and interpreting it. The streams of YouTube’s flow, enabled and produced by audience use patterns, could lead to interpretations of content that yield a diversity of cultural representations.

Audience use patterns can also impact organic rankings of videos under particular search terms, which steer the organization of videos and thus guide the way that users navigate the platform. Organic search results are obtained without paying for video ranking through advertisements. Watch time is considered to be the top factor determining the ranking of a video, followed by its total number of views, its title, its description, and its ratings based on the number of likes and dislikes. Videos have a better chance of ranking well in a YouTube search if the video description is at least 350 words and if it includes at least three or four keywords from the title and tags. Nonetheless, users can still find free tools, including the Keyword Tool (2015), which provides 750 YouTube keyword suggestions at no cost. Tags that used to be public until August 2012 contribute to the organic (and paid) discoverability of videos (Brian, 2012). The textual information, or metadata, attached to a video is essential, as YouTube’s ranking algorithms are unable to listen to or watch videos. Textual information is used by video crawlers to index videos, as sound and images are invisible to crawlers. Besides the key factors – watch time, views, title, description, and ratings – a host of other factors influence how a YouTube video will rank in terms of a particular search term, including playlist addition, flagging, age of video, freshness, channel authority, channel views, and subscribers. For a YouTube video to rank favorably, it must also engage the community within YouTube and outside of the platform through shares, embeds, inbound links, and comments.

The potential of user input to influence the ranking of videos is nonetheless curtailed by Google’s algorithms. Online analyses of the Sema reveal that videos that do not have many of the dominant characteristics that are needed to rank well organically, including a large number of views and likes, can still be tabulated on the top of the first SERP under a particular search term. The phenomenon of the unexplainable ranking of videos on the first SERP on YouTube indicates that a thorough knowledge of the algorithmic processing used to rank videos organically on YouTube remains unattainable; indeed, Google keeps the platform’s ranking algorithms a secret. Google’s algorithms are simultaneously hierarchical and often favor the more popular uploaders in their ranking mechanisms (van Dijck, 2013a: 116). The popularity principle underlying Google algorithms on YouTube is designed to monetize the labor of YouTube users. The hierarchical structure of YouTube privileges certain channels over others, leading to a predominance of views for videos that are better ranked by the platform’s search engine (van Dijck, 2013a: 116, 117). YouTube’s Partner Program currently enables creators to monetize content on YouTube through advertisements, paid subscriptions, and merchandise. Videos can obtain better rankings as promoted videos through paid searches that are connected to keywords in search queries. With the tendency of videos from YouTube’s partners to obtain more views and to be more prevalent in the top 10 search results, these videos are favored by search algorithms (van Dijck, 2013a: 126).

YouTube flow produces competing results. Google’s algorithms and business models create a hierarchical structure that often consistently promotes select videos at the top of SERPs. This flow of content could potentially homogenize cultural representation by privileging certain viewpoints and actors. At the same time, the flow on YouTube, steered by its search engine and ranking algorithms in combination with audience use patterns, has the potential to challenge the homogenizing of culture. This potential for diversity results as users decide which viewing path they want to take through a SERP. The actual content of videos provides the information that enables users to determine whether YouTube flow yields multiformity or homogeneity in cultural representation. Two questions are raised that highlight the content of videos in relation to the competing cultural results of YouTube flow. First, how can in-depth analysis of video content demonstrate that YouTube flow, produced through algorithms and audience use patterns, fosters varied representations of culture? Second, how can in-depth analysis of video content at the top of a SERP undermine the claim that YouTube flow results in the homogenizing of culture, as promoted by the hierarchical structure of both Google’s ranking algorithms and its business models?

Methodology of the case study

These two questions underlie the methodology of the case study. To answer the first question, about how content analysis illustrates how YouTube fosters diverse representations of culture, an online method referred to here as the ‘first method’ has been undertaken. This first method consists of analyses of YouTube videos as well as the lists, metadata, and comments connected to videos under the search term ‘Mevlevi Sema ceremony’. This type of analysis has been conducted over various periods of time since 2010 to capture the way that audience use patterns in conjunction with algorithms produce flow. This search term, which is the title of UNESCO’s (2009) video promoting the official Sema narrative, is used to analyze whether unofficial versions are also ranked under the UNESCO title. This method is inspired by the virtual ethnography that Hine (2000) has developed, which examines digital culture by analyzing and following websites and discussion forums. Therefore, YouTube’s flow of videos, in addition to lists, metadata, and comments, is tracked and analyzed in conjunction with the taking of ‘fieldnotes’ like those compiled in an actual ethnography. The first method enables the tracking of the flow of YouTube videos, including the metadata, comments, and lists embedded with the videos, to illustrate how the sound and images within videos provide diversified representations of the Sema.

To answer the second question, about how the analysis of content can undermine the claim that YouTube flow promotes the homogenizing of culture, a second online method has been used that is composed of two sections referred to here as ‘part A’ and ‘part B’. In part A, a recording and analysis of the videos that appear at the top of the first SERP under the key term Mevlevi Sema ceremony have been conducted at intermittent periods since 2012. This methodology builds upon Rogers’ (2013) digital method of specifically tracking the videos that appear at the top of SERPs over a period of time. Part A tracks the top of SERPs in combination with an in-depth analysis of the video content of prioritized videos to ascertain how video content itself undermines the homogenizing of culture promoted by the hierarchical structure of Google’s ranking algorithms. In part B, an analysis was conducted in March 2015 of the first list of videos tabulated under a combined Turkish and English keyword term that deliberately forces a meeting between official and unofficial versions of the Sema in YouTube’s flow. The search term used combines the UNESCO title Mevlevi Sema ceremony in English with a Turkish translation of ‘women whirling dervishes’: kadın semazenler. Part B demonstrates that the homogenizing of culture promoted by the hierarchical structure of Google’s ranking algorithms can be challenged through the conjoining of differing keywords in one search term. It further illustrates that in-depth analysis of video content provides an understanding of the juxtapositions of official and unofficial heritage produced through contrasting keywords.

Research on the Sema has been conducted through the prism of UNESCO practices and the Turkish context. This research enriches the first and second online methods, which analyze the way that YouTube’s flow of content brings together disparate representations of the Sema, fostering divergent accounts of the practice. Interviews were conducted in July 2012 with members of UNESCO’s Intangible Heritage Section in Paris in regards to YouTube and heritage. Research has also been conducted on UNESCO policies related to the 2003 Convention. For instance, protocol dictates that a Candidature File outlining a specific national practice must be submitted to UNESCO’s Intergovernmental Committee for the Safeguarding of the Intangible Cultural Heritage (Committee). The Committee is composed of representatives of nation-states who have the authority to examine whether this intangible practice, as outlined in detail in the Candidature File, fulfills the necessary criteria to be officially recognized by UNESCO and inscribed on its Lists of Intangible Heritage. Moreover, the UNESCO Candidature File produced by the TMCT in 2004 through the affiliated organization the International Mevlana Foundation (2004: 4) has been extensively analyzed. Findings reveal that both the TMCT and Mevlana Foundation are backed by the governing AK Party. Furthermore, this Candidature File does not officially recognize EMAV as a legitimate Mevlevi community, primarily because its practices integrate women and colored robes in public ceremonies. In the Candidature File, the EMAV community is therefore not included within the safeguarding strategies of the Turkish practice as officially recognized by UNESCO. Even though the Convention mandates that nation-states must consult all the communities involved when safeguarding a practice (UNESCO, 2003), performances by EMAV remain invisible within the context of Turkey’s national safeguarding. The Candidature File explains that EMAV is not recognized because it has modernized the ceremony in ways that counter traditional renditions (International Mevlana Foundation, 2004: 4). Additional research on the Sema within the Turkish context sheds further light on EMAV’s nonrecognition as a purveyor of official heritage within the Candidature File. A political agenda lies within the former ruling party’s dismissal of EMAV’s contemporary interpretations of the ceremony. Research on the safeguarding of the Sema reveals that this ceremony was deployed by the governing party as a symbol of its political goal to reinstate Sunni Islam within a once officially secular nation and to establish Sunni Islam as the emblem of a renewed Turkish identity (Aykan, 2012: 57). The male-exclusive Sema performance therefore serves as a hallmark of Sunni Islam.

To analyze EMAV videos, an actual ethnographic study was conducted on the practices of EMAV in Istanbul during intervals from June 2011 to August 2012. This study included attending Sema performances and conducting interviews with EMAV’s spiritual head, Hasan Dede, as well as with women dervishes. Additional research on official and unofficial versions of the Sema included consulting scholarly journals, EMAV promotional materials, and Turkish newspaper and magazine articles. For instance, that Hasan Dede has allowed women to perform in public EMAV Sema ceremonies since 1993 was first ascertained through the Turkish popular press (Ayman, 2004). The content of Sema videos is interpreted from research conducted within and beyond the borders of YouTube, enabling a comprehensive analysis of the versions of the Sema disseminated through YouTube flow. Combining background knowledge of the Sema with these two online methods avoids producing superficial analyses that could arise by using YouTube and digital data as the only sources of knowledge.

Analysis of examples from the case study

The two methods informed by background research have been used to analyze the flow of Sema videos on YouTube. First, an example from the case study investigated through the first method demonstrates that in-depth analyses of video content illustrate how YouTube flow, produced through algorithms and audience use patterns, counters official heritage narratives. This example draws from a YouTube analysis conducted on December 8, 2014, that builds upon previous online analyses of YouTube heritage videos undertaken intermittently since 2010 (Pietrobruno, 2013, 2014). On December 8, 2014, under the search heading Mevlevi Sema ceremony, the first SERP appeared with 20 videos (out of a count of 3510), the position of which may shift in accordance with audience use patterns and algorithms. The majority of the videos on this first list reflect Turkey’s official national narrative by depicting this practice as exclusively performed by men in white robes. There are a few videos on this list that feature performances by EMAV. In these EMAV videos, women dervishes dressed in colored robes or white robes whirl alongside men. EMAV’s public ceremonies challenge the Sema as safeguarded by the Turkish nation-state and sanctioned by UNESCO.

This countering emerges on YouTube through user-generated videos uploaded by channels representing individuals. These videos are part of the personal collections of individuals who have attended EMAV ceremonies in Istanbul and have made them public documents through their dissemination on YouTube. The channel Delgado (2012) provides an example, entitled ‘Mevlevi Sema Ceremony at Orient Express Station – Istanbul’. This channel has uploaded this video from a selection of travel videos from around the world. This video has been on the first SERP under the search term Mevlevi Sema ceremony since January 2013. The uploaded video does not mention that this performance is by EMAV.

If a user clicks on Delgado’s video, the up-next videos are listed in accordance with user-generated content and algorithms that forge relational juxtapositions in regards to gendered heritage narratives. A look at the content of the up-next videos indicates that most of the tabulated YouTube videos again reinforce the male exclusivity of the Sema. Yet, there are a few that counter this male exclusivity, particularly the promotional video of EMAV, uploaded by a representative of this community (Ayar, 2010). The voiceover in the promotional video of EMAV clearly states that women are and should be part of the Sema. Juxtaposing the EMAV video with male-exclusive representations of the Sema puts into question whether there are women dervishes in Istanbul. As online research has shown, this existence of women dervishes within a genuine spiritual context is often debated in posted text comments. These relational juxtapositions emerge via the path that users take as they move through the flow of texts and images.

To conduct this research on YouTube, the Chrome browser is cleared of cookies and search history, and the YouTube user is not signed in. Lists of videos that would normally be recommended to an Internet protocol (IP) address do not appear even though the IP address is still used by YouTube to draw assumptions about language, location, and time of day. This limited ‘cleaning’ of the slate for experimental purposes is not the situation for the majority of YouTube users. If users are searching for videos with the keywords Mevlevi Sema ceremony, recommended videos along with related videos will appear in the lists of up-next videos. Targeted advertisements will also appear. Therefore, the manner in which YouTube can counter dominant heritage narratives is influenced by audience use patterns that impact the flow of content on the platform.

Users may encounter videos that counter dominant heritage narratives as they move through the flow of YouTube content. Ultimately, it is the users who decide the path that they will take as they navigate from one video to the next in response to the up-next lists produced by YouTube’s algorithms in conjunction with the platform’s information management system. The way that relational juxtapositions come together on YouTube through human interpretation to create heritage narratives that counter those proposed by nation-states echoes the narrative-generating potential of hypertexts discussed in early research on the Internet in the 1990s. At that time, the web was envisioned as a space for the surfer to create a journey that could be translated into a story (Rogers, 2013: 85). In hypertext analysis, the stories generated by moving from one hypertext to another were seen as by-products of a random flow of information.

This extended example illustrates that YouTube’s display of listed videos is not completely accidental or unplanned. The relevance of the videos featured under particular search terms is determined by algorithms designed to monetize the labor of YouTube users, but searches can still produce ‘chance linkages’ that are not completely predicted by Google’s algorithms (Ippolita, 2013: 74) and can lead to the creative combining of information (Mahnke and Uprichard, 2014: 260). Relational juxtapositions rendered possible by algorithms and user activity are dependent upon audiences’ interpretation strategies and users’ subjective perceptions of the search results (Ippolita, 2013: 74). The idea that users or audiences can decipher the meanings of search engine results on YouTube in different ways relates to the reading of the televisual message (Hall, 1999). Consequently, YouTube’s flow may bring together videos that contradict the male exclusivity of the Sema, but this connection challenges the dominant hegemonic view of the practice only if the audience chooses to interpret it in this manner (Hall, 1999). Even within the business model of this platform, which governs the mediation of algorithms and user involvement, lies the potential for dominant heritage narratives to be countered by the interpretative work of audiences.

Second, an example from Part A of the second method demonstrates how in-depth analysis of video content can undermine the cultural homogenization of YouTube flow promoted by the hierarchical structure of the ranking algorithms of Google’s business models. A recording and analysis of the videos that appear at the top of the first SERP under the term Mevlevi Sema ceremony have been conducted at intermittent periods since 2012. In a particular instance, on December 8, 2014, the first video on the list was entitled ‘Sema ceremony – whirling dervishes’, uploaded by the channel mevlanafoundation (2011). This video has stayed at the top of the first SERP since October 2012. As mentioned, the International Mevlana Foundation (2004) is the organization affiliated with the TMCT that was responsible for producing the Candidature File for UNESCO. The second video at the top of the first SERP on December 8, 2014, entitled ‘The Mevlevi Sema Ceremony’ (UNESCO, 2009), which has stayed in this rank since October 2012, was uploaded by the UNESCO channel and is featured on the UNESCO website of the Lists of Intangible Cultural Heritage (UNESCO, 2012). This video is a shortened and revised version of an approximately 10-min video submitted along with the Candidature File in 2004 and produced by Ankara Productions for the TMCT (albaKultur, 2009). Sicard (2012), the knowledge management specialist of UNESCO’s Intangible Heritage Section in Paris, proposes that Google may privilege the ranking of heritage videos uploaded by institutions. Analyzing the top videos under the specific search term Mevlevi Sema ceremony reveals the official national narrative of this heritage practice proposed by Turkish governmental institutions and sanctioned by UNESCO. An analysis of the top of the first SERP does coincide with Rogers’ (2013: 86) work on digital methods, which claims that search engine results cannot provide a diversity of representations but instead reproduce the dominant narrative of the organizations currently representing a practice.

Despite promoting official heritage narratives, the top results on YouTube search engine lists may also be resisting them by exposing alternative accounts of heritage. This possibility emerges through the visual content of videos. This research method that constitutes a form of resistance is grounded in user practice. It is users who might find alternative heritage narratives by watching a video and interpreting its images through the prism of YouTube’s flow. The nature of these counter readings is further steered by the cultural horizon and knowledge of users. Once Part A is intertwined with more extensive analysis of video content, informed by research undertaken beyond the digital data itself that may also coincide with the cultural knowledge of viewers and uploaders of Sema videos, the potential for YouTube to produce a diversity of heritage narratives becomes apparent even within the top results of SERPs. An example is the video entitled ‘Sema ceremony of the whirling dervishes’ (mevleviyya, 2006), uploaded by the Mevlevi Order of America, which appeared in fifth position on December 8, 2014, on the first SERP. The Candidature File recognizes non-Turkish Mevlevi communities such as the Mevlevi Order of America as legitimate organizations even though these non-Turkish communities based in Europe and the United States include women dervishes in public performances. The Candidature File notes that when these non-Turkish groups perform in the Galata Mevlevi Lodge in Istanbul, only male dervishes participate in public performances (International Mevlana Foundation, 2004: 6). Built in 1491, the Galata Mevlevi Lodge is a celebrated building in which Mevlevi ceremonies are held for public viewing. Despite claims made in the Candidature File that women dervishes from officially recognized non-Turkish communities do not perform with male dervishes in Istanbul, videos on YouTube tell another story. The video of the Mevlevi Order of America features a performance by members of this organization in the Galata Mevlevi Lodge in which women and men together performed in a public ceremony in 2005. The presence of women in the performance is also discussed in YouTube comments. In response to a question about whether it is just men who participate in this practice, the channel mevleviyya (2006) writes, ‘Women and men are working with this practice. There are women involved in this ceremony shown in this clip’.

This example illustrates how the flow of content on YouTube at the top of SERPs can challenge claims made by official heritage authorities supported by national governments. The content of videos themselves offers an evasion of corporate surveillance whose algorithms may be reinforcing the positioning of institutions at the top of search engine lists. This research demonstrates that the content of videos or the actual stories recounted through images can challenge dominant heritage narratives by escaping to a certain extent the ranking and sorting performed by the algorithms of YouTube’s search engine. As mentioned, the visual data in videos are invisible to search engines, which rely on textual metadata to classify and rank videos on YouTube. As these annotations are produced by users, they may be unreliable and hence insufficient in their ability to index images within videos (Morsillo et al., 2010: 358). Moreover, as a video is composed of proliferation of meanings conveyed by simple categories such as colour and shape to more complex elements such as elaborate events and cultural contexts, even reliable user-provided metadata is unable to annotate this broad span of interpretations (Shirahama and Grzegorzek, 2016: 307). The metadata attached to a video is always partial and incomplete and does not fully capture the depth of the visual data that users may ascertain from merely viewing a video through the work of the human eye and mind. Viewers can see and interpret images in videos in ways that bypass their indexing through textual metadata. Therefore, the flow of visual data on YouTube, which brings audiences into contact with images that can evade the indexing of search engines, may become part of the narratives forged by audiences as they find meaning in the juxtaposition of images and texts that can counter dominant heritage narratives.

Third, an example from Part B of the second method further demonstrates how in-depth analysis of video content can undermine YouTube flow’s homogenizing of culture. Part B differs from part A by exposing that the homogenizing of culture promoted by the hierarchical structure of Google’s ranking algorithms can be manipulated through the use of divergent search terms. When the search term ‘Mevlevi Sema ceremony + kadın semazenler’ is inserted (with the Chrome browser cleared of cookies and search history and without signing into YouTube), a list of 20 videos appears on the first SERP. An examination of this first SERP under the above search term recorded on March 30, 2015, reveals that 12 of these videos (out of a count of 108) depict this practice in accordance with the safeguarding agenda of the TMCT, which sanctions male-only ceremonies with dervishes donning white robes. Among these 12 videos on the first SERP are those of UNESCO (2009) and the channel mevlanafoundation (2011), both previously discussed. A selection of other videos on the first SERP captures representations that counter those sanctioned by UNESCO through the TMCT in terms of the rising influence and presence of women in the Sema.

The video that ranked in eleventh place on the first SERP, entitled ‘ERKadinlarAyini’ and uploaded on February 27, 2015, reveals a significant change in the status of women dervishes in EMAV ceremonies (Baydar, 2015). This video depicts a roughly 20-min segment of a recent Sema performance by EMAV in which men and women perform together. An analysis of the content of this video indicates that a woman dervish is leading the ceremony as a stand-in, or postnisin, for Hasan Dede, the male spiritual leader of the community. The postnisin has always been a man in YouTube videos previously analyzed as well as in performances attended during actual ethnographic work in Istanbul and documentation produced by the community (Emir, 2010: 32). This video illustrates that the contemporary practices of EMAV are further challenging the official version of the Sema sanctioned by UNESCO by placing women as spiritual leaders in public ceremonies. This recently uploaded video, which has only 212 views, is ranked on the first SERP. One of the primary reasons for this ranking, besides some textual relevance of the video description to the search term, including the words kadin semazenler and ‘Sema’, could be YouTube’s ranking algorithms, which favor ‘freshness’. The freshness factor, which gives precedent to new uploads, may enable recent changes in the practices of the Sema captured in YouTube videos to become part of the first SERP under specific search headings. The flow of YouTube videos determined by YouTube’s algorithms provides an indication of the way that the Sema is developing in everyday practices. This integration of content and algorithms is further challenging the official representations of this practice as concretized by the TMCT through UNESCO’s video. The explicit triggering of relational juxtapositions between official and unofficial versions of the Sema through the strategic use of disparate search terms reveals that YouTube flow as managed by hierarchical algorithms can also provide a space for alternative representations of the Sema. This possibility occurs within YouTube’s ultimate goal to monetize the labor of users by prioritizing specific videos on SERPs.

Flow, diversity, and audience engagement

The case of the Sema illuminates how, forged through the combined work of users and algorithms, the flow of videos embedded alongside advertisements, lists, metadata, and comments enables varied representations of heritage to come into contact. This junction creates relational juxtapositions that become meaningful through human interpretation. Audience engagement is integral to the paradoxical nature of YouTube flow, which both fosters and hinders diversity in video content. The juxtaposition of official and unofficial heritage versions of the Sema is produced by YouTube flow as users move through the platform, choosing their own paths under particular search terms. The reading and analysis of juxtapositions of video content by audiences are fundamental to the ways that YouTube can produce divergent narratives of culture. Furthermore, audiences can impact the tabulation of videos that comprise YouTube flow. The appearance of certain videos under a search term within YouTube flow results from the combined workings of algorithms and audience use patterns; user input of metadata can influence to a certain extent the organic rankings of videos under particular search terms, impacting YouTube flow. At the same time, user influence upon the ranking of videos is constrained by algorithms designed to monetize the labor of users. Google’s algorithms and business models forge hierarchical structures that promote specific videos at the top of SERPs, potentially thwarting a mélange of content. YouTube’s ranking and sorting through the algorithms of its search engine are unable to index or classify the actual visual and sonic data of videos. Being able to listen to and watch videos, audiences can interpret their data at a deeper level than Google’s indexing by video crawlers. Through an in-depth reading of the videos ranked at the top of SERPs, audiences may find alternate narratives within the seemingly homogenous versions given top ranking by search algorithms that privilege dominant actors and singular perspectives. Audience engagement in the transmission of the diversity of heritage narratives through YouTube flow takes shape as users read and analyze videos, choose the viewing path, upload videos, and provide metadata.

Because YouTube is a commercial platform, it is often considered to be a nonpolitical form of entertainment (Fuchs, 2014: 99). When YouTube is regarded as a political platform because it democratizes cultural representation, the platform’s political potential is relegated to addressing the concerns of minorities and thus minimized (Fuchs, 2014: 61). An analysis of videos on YouTube illustrates that user involvement in the platform can lead to interpretations of video content that are politically significant. The case study of the Sema indicates that without an in-depth analysis of video content, the videos of the Sema in the Turkish context may appear to address a minority’s concerns or to be a nonpolitical form of entertainment. Images of women whirling in colored and white robes or leading Sema ceremonies can be simply trivialized as a mere curiosity without wider significance. The diversity of heritage representations, produced through the ways that flow on YouTube juxtaposes competing practices of the Sema, is far from an issue of relevance only to a minority. The participation of women dervishes in Istanbul brings to light the role of women in Mevlevi Sufism and the changes that are taking place in Turkey that counter official versions of the Sema. The depiction of women dervishes in both white and colored robes illustrates how struggles over definitions of legitimate religious expression are embodied in seemingly trivial micropractices that have meaning related to the political goals of the Turkish government.

YouTube is packed with videos that appeal to niche audiences smaller than those of the most popular videos and that may appear trite to audiences unfamiliar with their contents. Regarding YouTube videos as nonpolitical forms of entertainment is the preferred stance of social media scholars, including Fuchs (2014), who analyze YouTube as a system from a political-economic perspective, emphasizing corporate structures and business models. This approach does not combine in-depth analyses of less popular videos with a consideration of the technologies of search algorithms. An analysis of both YouTube’s flow, which is produced by algorithms entrenched in corporate structures and business models as well as in audience use patterns, and the content of seemingly inconsequential videos leads to a deeper understanding of the dissemination of culture on the platform. YouTube’s flow brings divergent representations of culture into contact, potentially producing political meaning through user interpretation.