Reading with Algorithms
Classification systems are fundamental to how readers discover books and stories, both on and offline. Traditional knowledge organisation systems, such as the Dewey Decimal Classification (DDC) system and Library of Congress Classification (LCC) system, are built around and support linear, hierarchical, and mutually exclusive relationships between subjects according to a tradition of Aristotelian logic.1 In the publishing industry, Book Industry Standards and Communication subject headings, more commonly known as BISAC codes, likewise structure book categories in a hierarchical way, but one that follows a more commercial logic centred around reader-interests.2 Newly introduced platform-based classification systems often adopt and adapt some traditional systems of knowledge organisation; however, they are intrinsically shaped by their unique computational, and often commercial, contexts. Book-related sites and platforms — such as Wattpad, Amazon, Archive of Our Own, Goodreads, Radish, and so on — each develop systems to classify books in line with their specific purpose (e.g., book buying, reading, reviewing), and their individual technological architecture, algorithms, commercial imperatives, and user-behaviours. Wattpad, one of the largest social media platforms for reading and writing, combines a top-down subject heading system with a folksonomy (a user-generated system of classification that uses metadata such as electronic tags) to categorise and organise stories posted to its site.
How books are categorised determines how they may be discovered. Book classifications determine what we do with those books, where they go, who they are supposedly made for, and who can access them. As Janice A. Radway argues, books "do not appear miraculously" in the hands of readers, but rather are the "end product of a much-mediated, highly complex, material and social process."3 On Wattpad, as with other platforms used for publishing, buying, and reading books, this mediated process is augmented by the collection and deployment of personal data through opaque algorithms that shape reading practices on the platform.
The datafication of writing and reading on Wattpad has transformed classification and discoverability into processes structured around personalisation. This personalisation occurs at the level most associated with algorithmic culture: that of personalised recommendations based on users' individual data histories. However, personalisation also occurs through the development of genre assemblages centred around individual users. These personalised genre assemblages, which are created through tags, have particular importance for the visibility of stories by and about historically marginalised communities and contribute to the rise of microgenres that shape reading behaviours. Microgenres have proliferated with digital publishing and, on Wattpad, tend to organise content around personal experience, mood and affect or connected to cultural phenomena.4 The datafication and personalisation of classification algorithms, which in turn affect the discoverability of stories, are intricately intertwined with a third process: commercialisation. The classification systems and discoverability algorithms that shape readers' experiences and enhance user agency on Wattpad are also baked into the company's platform capitalist business model, flooding its revenue streams and feeding its corporate profit.
Wattpad launched in 2007 as a social media reading and writing platform, branding itself as the YouTube for books. Over the past decade and a half, Wattpad has grown into a multi-platform entertainment company that uses its creative writing and reading social media site as a data-driven discoverability platform. Intellectual property and user-data from the site are developed and commercialised through in-house multimedia programs, including its publishing imprint Wattpad Books and entertainment arm Wattpad Studios, as well as partnerships with other producers and publishers.5 The folksonomy developed through authors' use of tags underpins Wattpad's commercial enterprises and revenue streams as they provide nuanced looks at the stories, authors and genres that are currently trending. When combined with popularity metrics, they help inform which texts are adapted, republished or brokered, or constitute the data which is on-sold. While Wattpad ultimately controls how classifications structure visibility on the platform, which heavily regulates the popularity metrics that inform commercial decisions, free-text tags are a mechanism by which authors retain some agency concerning how their stories are named and interpreted. The specificity ascribed to most tags not only assists in more reliable search results and complements the top-down platform categories that generally reflect typical publishing genre categories (action, adventure, fantasy, historical fiction, mystery and thriller, non-fiction, romance, science fiction, poetry, teen fiction) and some subgenres (paranormal and vampire, for example). Importantly, tags have become an essential mechanism for the classification and discoverability of diverse and inclusive stories.
While there is somewhat of a common lexicon of tags developed by Wattpad's user community, tags present a radical departure from the hierarchical and relational way identity is represented in traditional classification systems that tend toward social and political marginalisation. Intersectional feminist approaches to knowledge organisation demonstrate how DDC and LCC, for example, both construct minority subjects against a hegemonic (white, cis, male, heterosexual) norm, and position books by and about historically marginalised groups under subject headings that are lower in the hierarchical structure.6 This structure is somewhat mirrored in BISAC subject headings, which relationally organises books by and about groups that are marginalised in Western society based on their gender, sexuality, race, age, ability, ethnicity, language, and/or religion. While the explicit naming of marginalised groups within book culture can also be a productive discoverability mechanism — listing titles under subgenre categories relating to diversity and inclusion, such as African American & Black romance or LGBTQIA+ fantasy, can be a useful marketing strategy for authors to better reach readers who are looking for these types of stories, for example — the relational position of these categories against a "general" (white, cis, het, male) norm as well as their normative treatment in bookstores and online bestseller lists often contributes to their marginalisation. Many publishing platforms, including Amazon, reinforce this system of subjugation through their top-down classification systems.7 On Wattpad, these classifications are removed entirely from its top-down categories, but not its discoverability menu. There are no classifications relating to historically marginalised identities such as "LGBTQIA+" or "Diverse Lit" in its platform-selected, drop-down category list, though both appear in the browse menu on the Wattpad's homepage.
In their location on the landing pages of Wattpad stories, tags mediate readers' experience by also acting as interpretive frameworks reflective of personalised genre assemblages constructed by the author. In this way, tags may be viewed as a form of digital paratext, the elements that "surround" a text and present it in a particular way.8 As Maria Lindgren Leavenworth notes in relation to tags used on fan fiction sites, while some "work similarly to how fictions are arranged in a bookstore, signalling a connection to known genres, others are specific to the [. . .] text form and as they are placed in close proximity to the text itself, signal both how the text should be read and how it should not be read."9 As with fan fiction sites, tags on Wattpad stories signal the type of text (the prose or poetic form), what genres and subgenres it belongs to, and how it might be approached. They are also often used by authors as meta-commentary and as expressers of emotion and affect.10 Several Wattpad authors use tags that are not widely used or searched for on the platform, including tags highly idiosyncratic to and often illegible beyond the context of stories and/or authors. For example, one author interviewed included the tag "uselesslesbian" on her apocalyptic story Wish Granted as a reference to her "hopeless romantic klutz" main character.11 At the time of writing, only six other stories used this tag on Wattpad and thus better reflects the meta-commentary use of tags rather than their role in aiding in discoverability. This metacommentary, in turn, also contributes to personalised genre assemblages within the platform's curatorial systems, which are often sorted into microgenres by the platform and often reflective of experience, affect, feeling, and vibes.
While the exclusion of categories relating to historically marginalised groups in Wattpad's categories can limit readers' ability to find books by and about people from these communities, the user-agency afforded in a free-text tagging system enables authors not only the opportunity to include representative and personal categories to their work, but do so in a way that is open-ended and non-hierarchical. Although a small sample, the tags on stories by authors I interviewed demonstrate what this looks like for some marginalised identities and representation. Across the 21 stories by 7 Wattpad authors I interviewed as part of a larger research project into the platformization of publishing, the most common tags used included: Young Adult, Love, Romance, Humor, Teenfiction, Adventure, Life, Drama, and Filipino. Several tags, and variations of similar tags, are used to indicate diversity and inclusion, including: celebrateblackwriters, civilrights, diverse, diversity, weneeddiversebooks, poc, interracial, arab, islam, philippines, filipino, westphsea, taglish, tagalog, bisexual, bi, girlxgirl, gxg, lesbian, gay, gaydisaster, lgbt, lgbtq, lgbtpride, writtenwithpride, and lgbt-themed. Some of these fit within relatively established lexicons of representation in book culture, such as weneeddiversebooks and gxg, the former a term popularised by the #WeNeedDiverseBooks social media campaign that began in 2016 by a non-profit children's literature organisation, and the latter a popular shorthand for girl meets/with/on girl or lesbian-themed literature in fan fiction and anime genre communities. Others are idiosyncratic to certain authors and texts. The openness of tags that allow authors to highlight diverse and inclusive characters and themes provides authors with greater autonomy to articulate preferred interpretive vocabularies in relation to their or their characters' identities. The variety of tags listed above that denote diversity and inclusion, with many variations often appearing on the same story (i.e. gxg and lesbian), show that writers are embracing the opportunity to classify their work with non-hegemonic and non-hierarchical descriptors. Wattpad's provision of technical features allows authors to prioritise aspects of the text that they feel are important but are not available in the platform's more rigid categorization system.
This process must be contextualised by the overwhelming white logic that underpins Wattpad's business practices to date, limiting the potential for greater racial, linguistic, and geographic diversity from this global platform. As one author described to me in an interview, "Wattpad has always been a predominantly white platform. Readers are from all over, but the stories that have gotten popular have been white, straight romance novels." The first book published by Wattpad Books to feature a non-white main character was written by a white author. Anecdotally, books by white authors published by Wattpad Books have received larger advances, reflecting pay discrepancies in the traditional publishing industry as highlighted by the viral Twitter hashtag #PublishingPaidMe in 2020. Wattpad's participation in the global book market, through its platform and social media platform, likewise reinforces the hegemony of Western Anglophone writing, demonstrating its seeming disinterest in developing non-U.S. and non-Anglophone book markets despite having active readerships there. The acquisition of Wattpad by Naver, a South Korean technology company that also owns Webtoon, in 2021 may change Wattpad's current course. However, exclusion is baked into its organisational operations. Despite having a Trust and Safety team, Wattpad reportedly does very little to protect authors of colour from harassment by readers and other authors on the platform. In 2021, several authors left the Stars program — the community of authors on Wattpad that benefit from the company's commercial programs — due to the marginalisation of authors of colour and queer authors, including inadequate promotion of their stories and limited recourse against harassment and microaggressions.
The agency afforded to authors to determine identity-based classifications on their own terms is thus tightly refereed by the platform's control over how these tags factor into discoverability. The convergence of production and reception spaces on Wattpad results in stories on Wattpad being subject to two categorisation systems that operate relationally. As just described, first authors classify their stories as they begin the publication process. Second, upon publication, Wattpad translates back-end author-chosen metadata to determine how books are organised for readers to discover. The interpretive framework that creates links between community-generated tags is contingent on an algorithmic, top-down vocabulary that identifies tags as denoting what community authors mean. In other words, to classify stories as "diverse lit" in Wattpad's recommender algorithms, authors must still use tags that Wattpad would deem diverse. The relationship between tags and recommender systems is dependent on opaque algorithms that mediate possible connections between authors, their content, and readers.
The combination of structured and unstructured categories serves authors and readers by enabling a more representative, community-driven system of naming content, and readers in discovering niche content that fits their interests. Folksonomies enable user agency in a way that contributes to collaborative popularity metrics and records evolving language and tag behaviour, from niche trends to knowing "the next big thing." Popularity, though, is not a neutral or natural occurrence online; it is situated in how content is classified and mediated through discoverability algorithms that determine what users see and when — both of which are ultimately controlled by the platform. A publisher looking for the next big romance novel is not simply interested in the popular stories on Wattpad, but rather the most popular romance stories on Wattpad. Classification and discoverability algorithms thus shape the popularity metrics that underpin Wattpad's revenue streams, including the on-selling of trend data to third-parties, facilitation of personalised advertising, and intermedial adaptation through in-house production divisions and partnerships.12 The tagging system therefore works for Wattpad as much, if not more, as it does for authors and readers.
Wattpad's homepage, the first point of call for readers to discover new stories, has become increasingly curated over the last few years to benefit the platform's profit-making initiatives. For the first five years or so after its launch, Wattpad's homepage included stories that had topped crowdsourced popularity metrics. The top-level menu option on the homepage has gone from housing links to forums to a single link to the Watty Awards page, an award that itself has gone from being fan-voted to being judged by a selection of Wattpad staff. This minimisation of possible user engagement, seemingly at odds with a platform whose business model is based on profiting from user-generated data, represents the evolution of the platform through the refinement of the platform's business model. Over the past few years, Wattpad has moved away from a revenue stream dependent on advertising to one focused around its pan-entertainment enterprises, including Wattpad Books, Wattpad Studios, and content development for brand partnerships. As Wattpad's pan-entertainment system has grown, the homepage has increasingly featured platform-curated stories that ultimately feed into the site's monetisation programs. As seen in Figure 1, several carousels on the homepage are dedicated to "Bestsellers," "Editors' Picks," "From our Stars," "Paid Stories," "Available in Bookstores," "Wattpad Studio Hits," and "Wattpad Books."
Tags and featured carousels on Wattpad's homepage also tend to reflect the broader popular culture and media environment in which stories are written, which further emphasises the commercial context in which Wattpad is increasingly attuned. In addition to internal commercial platform endeavours, intertextual media tags also proliferate as writers are encouraged to include tags of bands, media texts, and celebrity names (for fan fiction and other genres), especially if celebrities' likenesses have been cast onto the characters (i.e., if characters are meant to look like celebrities even if they do not portray them).
Wattpad also connects stories to popular culture and media events through curated categories, or "handpicked" in Wattpad's vernacular, on the homepage. In January 2021, when the historical romance series Bridgerton was trending on Netflix, Wattpad featured a carousel of historical romance fiction stories under the headline "In The Mood For Historical Fiction?" (Figure 2). Experience-based categories are also evident in seasonal featured lists, such as "Spooky SZN" (October) and "Scorpio Season" (October-November). This reflects a broader trend visible on creative content platforms (from creative writing to video streaming) to organise content with a focus on affect and experience through microgenres. Other microgenres are reflective of specific tropes, such as "Celebrating Strong Women" (around International Women's Day), or vibes, such as "Spring Is The Backdrop of Love." This latter carousel microgenre appeared on my homepage, that of a person who lives in the southern hemisphere, in April (our Autumn), demonstrating the sometimes-blunt application of microgenres on platforms; they may be reflective of a person's experiences, but aren't necessarily personalised recommendations.
In publishing studies, genre classifications are primarily understood as a means for communication between producers and consumers, created through marketing, textual form, and reception.13 The affective, experience-based categories on Wattpad, as well as other attention economy platforms that are premised on good user experiences (such as Netflix), highlight the degree to which consumers have become central to the construction of genres. In other words, these categories demonstrate the increasing interconnection between marketing, textual form, and reception in the organisation of content on digital platforms. These genre processes do not happen in isolation of media sectors though. As is evidenced by the Bridgerton-inspired categorisation example on Wattpad, there is a growing intertextuality between media sectors in the formation and understanding of affective, experience-based genres. This interrelationship demonstrates the growing connection of popular culture across platforms and media spheres in the entertainment ecosystem, which may only be captured by an integrated platform, publishing, and media studies approach.14
There is a well-established relationship between book classifications, genre assemblages and reading practices. The relational power between the two is augmented on digital platforms as classification systems contribute to recommendation and discoverability algorithms and, in turn, reading practices and aesthetics. Wattpad's bipartite classification system, which combines a top-down taxonomy and user-driven folksonomy, represents a more heterogeneous approach to categorisation in book culture. In one way, this system serves authors and readers of inclusive fiction by allowing autonomous modes of classification and identification. This power is curtailed somewhat by the platform capitalist context in which these processes now take place. Reading through Wattpad's classification and discoverability algorithms shows the impact of three interlinked processes on publishing platforms: datafication, of what and how people are reading; personalisation, through sometimes hyper-specific personal genre assemblages and microgenres; and the commercialisation of this content as well as user-behaviours. The opportunities and practices described here — of autonomy with regards to representation, personalised genre assemblages, and the deployment of affective and intertextual microgenres — highlight the productive tension between user agency and (corporate) platform prerogatives, and what is at stake when the balance tips too far in favour of the latter.
Claire Parnell (@cparnell_c , @claireparnell.bsky.social) is a lecturer in digital publishing at the University of Melbourne. Her research focuses on the platformisation of publishing and cultural inclusion in book culture.
References
- Hope A. Olson "How We Construct Subjects: A Feminist Analysis," Library Trends 56, no. 2 (2007), 510. [⤒]
- Daniel Martínez-Ávila, Rosa San Segundo, and Hope A. Olson, "The Use of BISAC in Libraries as New Cases of Reader-Interest Classifications," Cataloguing & Classification Quarterly 52, no. 2 (2014), 137-55. [⤒]
- Janice A. Radway, Reading the Romance: Women, Patriarchy, and Popular Literature (Durhman, NC: University of North Carolina Press, 2009), 12. [⤒]
- Beth Driscoll "The Rise of the Microgenre," Pursuit, (2019). [⤒]
- Claire Parnell "Mapping the Entertainment Ecosystem of Wattpad: Platforms, Publishing and Adaptation," Convergence 27, no. 2 (2021), 524-38. [⤒]
- Melissa Adler and Lindsey M. Harper "Race and Ethnicity in Classification Systems: Teaching Knowledge Organization from a Social Justice Perspective," Library Trends 67, no. 1 (2018), 52-73. [⤒]
- Claire Parnell "Independent Authors' Dependence on Big Tech: Categorization and Governance of Authors of Color on Amazon," AoIR Selected Papers of Internet Research (2021). [⤒]
- Gérard Genette, Paratexts: Thresholds of Interpretation (Cambridge: Cambridge University Press 1997). [⤒]
- Maria Lindgren Leavenworth "The Paratext of Fan Fiction," Narrative 23, no. 1 (2015), 47. [⤒]
- Ludi Price and Lyn Robinson "Tag Analysis as a Tool for Investigating Information Behaviour: Comparing Fan-Tagging on Tumblr, Archive of Our Own and Etsy," Journal of Documentation 77, no. 2 (2021), 320-58. [⤒]
- Sol Tuberosum 2020. Interview. [⤒]
- Parnell, "Mapping the Entertainment Ecosystem." [⤒]
- Karl Berglund "Genres at Work: A Holistic Approach to Genres in Book Publishing," European Journal of Cultural Studies 24, no. 3 (2021), 757-76; Claire Squires Marketing Literature (London, UK: Palgrave Macmillan UK, 2007). [⤒]
- Parnell, "Mapping the Entertainment Ecosystem." [⤒]