Migration,non-use,and the ‘Tumblrpocalypse’: Towards a unified theory of digital exodus

Introduction

In this article, we use the ‘2018 Tumblr migration’ to explore how (1) platform ‘non-use’ and (2) platform ‘migration’ are overlapping aspects of a broader phenomenon of digital ‘leaving’ or exodus. A unified analysis of this phenomenon has important consequences for understanding emergent relationships between individual agency, platform structure, and online culture.

From classic sociological and anthropological texts to the study of new media, scholarship has shown how ostensibly distinct domains of culture in fact intersect and shape each other. Unified analyses such as Collier and Yanagisako’s (1987) feminist anthropological reading of gender and kinship allow for exploring overlapping phenomena, such that the analytical framework is not comparative but intersectional. Such analysis can work diachronically as well: as its name indicates, the study of ‘new’ media can be informed by the influence of cultural logics that arose with older media (e.g. Kittler, 1999). By reading platform non-use and migration as interrelated under the umbrella of digital exodus, we begin a conversation on how we might create more unified analyses of interlocking digital systems, concepts, and practices.

Online cultures present myriad novelties, but enduring theoretical dilemmas as well. For instance, they are fundamentally shaped by forms of movement, physical and virtual. They bring together domains of experience and practice (for instance, work and play) in ways that suggest unified analyses in response. They are shaped by dialectics of individual/collective and continuity/change, social structures deeply shaped by the power structures of technological capitalism.

All these dynamics of culture and power were evident on 3 December 2018, when the CEO of Tumblr, Jeff D’Onofrio, claimed to introduce ‘a better, more positive Tumblr’ (D’Onofrio, 2018). Tumblr is a microblogging site where users obtain a free blog page they can fill with personalized content. The site is designed to emphasize images (though videos and text are common as well). Through the ability to ‘follow’ specific users, to reply to and ‘like’ individual postings, and to repost or ‘reblog’ content, Tumblr explicitly seeks to bridge blogging and social networking. The affordances and governance practices of the platform thus present Tumblr as an intermediary, a venue for users to create and share content, despite the nearly unlimited power of the corporate entity controlling the platform (Gillespie, 2010). D’Onofrio’s post clarified that behind his euphemistic language of improvement and positivity was the banning of ‘adult content’, specifically ‘photorealistic imagery or photography – images, videos, or GIFs – with real humans that include exposed genitals or female-presenting (yeah, we know you hate this term) nipples or depict sex acts’ (D’Onofrio, 2018).

This announcement, soon dubbed ‘the porn ban’, forms the primary context of the fieldwork on which our analysis is based. Methodologically, this fieldwork draws on participant observation and the analysis of texts and images produced around the ban (these texts and images themselves formed part of the online social context under analysis). As discussed by Boellstorff et al. (2012), participant observation involves ‘step[ping] into the social frame in which activity takes place. . . [and] becoming directly involved in the activities of daily life’ (p. 65). In this regard, the first author observed, studied, and blogged alongside Tumblr users from 2017 to 2020, as well as participating as a casual user for a significantly longer period. ¹

With regard to our argument, the most significant effect of D’Onofrio’s post was that from the ban’s introduction in February 2019, monthly page views fell by 150 million, a 25% drop (Jackman, 2019). This represented users deleting their blogs, often termed the ‘Tumblr Purge’, and users moving content to other platforms, often termed the ‘Tumblr Migration’ (Tumblr NSFW Content Purge, 2019). While Bronstein (2020) has discussed Tumblr’s lifecycle in relation to pornography, this change in Tumblr governance was generally interpreted as a ban on ‘adult content’ even when not pornographic. Many users thus saw at stake issues of erotic sociality and free speech rather than pornography narrowly construed (Notopoulos, 2018). Queer and trans communities especially have had a significant presence on Tumblr for many years (Fink and Miller, 2014) and the curtailing of sexual expression was seen as particularly detrimental for displays of non-normative gender and sexuality (Braidwood, 2018). The unique permissiveness that Tumblr had otherwise displayed towards adult content was seen as a key feature of the platform, part of a culture of diversity in place of the ‘unattainable cookie-cutter lifestyle’ seen on many other social media sites (McNicoll, 2020). This is one reason why these sociotechnical practices of leaving link up to the fundamental questions around movement, selfhood, and power with which we opened this article.

Non-use as social act

Platform affordances typically imply a clear break of unsubscription or deletion: one is in or out. However, a growing body of work underscores ‘non-use’ as a set of social acts and relationships, a ‘leaving’ that is multivalent and processual. Non-use can encompass a range of partial and contextual disengagements, as well as forms of occasional reconnection, lurking, or transfer to new online environments. For instance, in their study of Grindr, a location-based dating and hookup app for gay men, Brubaker et al. (2016) concluded that leaving involved ‘layered social and technical acts’, and that viewing it as such ‘[exposed] a set of relational possibilities and spatial arrangements within and around which people move’(p. 373).

Tumblr non-use can be seen as overlapping and interacting sets of social decisions largely (and perhaps unsurprisingly) disincentivized by the technical affordances of the platform. Three forms of non-use reflecting these dual social and technical pressures are (1) breaking of patterns of use; (2) logging out as social protest; and (3) archiving and deleting one’s blog.

The first of these forms of non-use is best understood as an active technique of decreasing usage, including to the point of permanently logging out. While in a sense non-use can thus be quantified – fewer minutes, to the point of zero, in a particular online context – non-use always takes culturally specific forms shaped by that online context and the agentive practices of those involved. In this sense ‘non-use is not an absence or a gap. . .. Non-use is, often, active, meaningful, motivated, considered, structured, specific, nuanced, directed, and productive’ (Satchell and Dourish, 2009: 15).

One aspect of this involves the ‘use’ in question: non-use of email, Fortnite, and Yelp cannot be reduced to a single temporal metric. Traffic data indicates that Tumblr non-use rose after the content ban (Jackman, 2019), recalling the ‘leaving as disuse’ seen on Grindr (Brubaker et al., 2016: 381). One user directly addressed Tumblr staff saying ‘I’ll be honest – I’m not sure how much longer I’m going to be using Tumblr. If my friends all leave because of your unfair censorship policy, I’ll have no reason to actually stay here and I’ll pack up and move’. Another discussed having ‘cleared out’ posts and being engaged in the process of deciding which blogs to keep active for personal or professional purposes. Despite being ‘still angry at Tumblr’, another user explained that they were ‘just going to be posting here anyway, at least for now, but encouraging people to move elsewhere’. Tumblr’s non-use was for these users incomplete.

The use of Tumblr is tied to the social performances and connections it allows: non-use, the decision to ‘pack up and move’ or just keep posting ‘for now’, also required active negotiation of social connections and consensus. These forms of non-use were complicated by technical affordances of Tumblr such as the default ‘infinite scroll’ form of content consumption and the ‘always on’ relationship to the platform that that encourages many hours a week of engagement in a refresh-scroll-refresh cycle (Hillman et al., 2014: 779). One user, after having moved to another platform, said they ‘may have kicked it [Tumblr usage] for good this time’. Technical affordances and social factors appear to have worked together to hinder the breakdown of old patterns of use.

One concrete example of non-use is logging off. Tumblr users, in protest of the ban, circulated posts calling users en masse to log off on Monday, 17 December 2018 (Radulovic, 2018b). Due to Tumblr’s technical features, logging off as an active form of social protest made sense, at least to some users. Tumblr’s log-in system automatically generates a persistent cookie that only expires after a year of use. This means there is no opt-in ‘remember me’ feature: the platform encoded an assumption of continuing use, facilitating an ‘always on’ dynamic where users would check the platform for updates frequently. Logging out was therefore an intentional act, not something that took place automatically when closing a browser. Furthermore, the logout button was hidden on a submenu in a small, greyed-out font. By separating it from the main interface – visually and at the level of information architecture – the platform thus framed logging out as disrupting the flow of normative use. In such a context, the act of logging out is technologically framed as non-use, the consequence of active intention rather than passive discontinuation.

Another form of non-use was to delete one’s Tumblr blog entirely. This was common in the wake of the 2018 pornography ban (Notopoulos, 2018). While from the affordances of the platform blog deletion was an individual act, blog deletion took a deeply social form as a response to the ban. First, users discussed blog deletion as distinct – socially and technically – from migrating a blog to another platform. ‘You can find me on Twitter’, one user said, with their Twitter handle and the addition ‘although I’m not deleting my Tumblr or anything like that. I want to see what’s gonna happen on here’. The shift from primary site of activity was not equal in weight to actually deleting one’s Tumblr blog. Another post was quoted by a news site as arguing the porn ban was a ‘removal of content, not people’ and emphasized that Tumblr was not kicking people off (Radulovic, 2018a). These responses demonstrate an awareness of the partiality of leaving compared to the finality of blog deletion.

Bound up in the process of deleting or moving one’s blog were discussions where users sought advice regarding how to archive their Tumblr blogs. ‘The built-in backup is shit’, one user said, before explaining their process for producing a bootleg backup. ‘I’m nervous Tumblr might suddenly disappear, so I exported my main blog’, another post said. In these cases, users were not seeking to recreate their blog (and its social network) on another platform. Instead, archiving was part of the process of deleting and the successful enactment of non-use. The issues at play were not related to migration, but to archiving cultural artifacts (from memes to art to personal posts) that their Tumblr account had contained. In these cases Tumblr users emphasized archiving their blogs for the personal meaning those archives would represent. ‘I’m sentimental so I like saving things’, a user wrote when updating their followers on the efficacy of a bootleg backup method. Another wrote ‘All I really need is a copy for myself’.

These cases of Tumblr blog deletion in the wake of the 2018 pornography ban involved not just individual decisions, but technical practices. Because user base is a crucial metric for platforms, affecting the corporate bottom line, platforms like Facebook typically make it difficult to delete one’s account entirely, often offering forms of logging out or temporary account suspension as alternatives (Karppi, 2011). When accounts are so difficult to delete, users may instead use various means to ‘delete’ the platform from their daily lives without technically doing so – for instance, by anonymizing profiles and removing content (Brubaker et al., 2016: 380). In contrast, deletion practices on Tumblr as it was structured in 2018 were relatively easy. The first blog a user would make on Tumblr would be identified as the main blog for that account, and users could add additional blogs (known as ‘sideblogs’) under the same account. Deleting a sideblog was as simple as clicking ‘delete blog’ and signing in again to confirm. Deleting an account entirely was a similar process, distinguished only by a final ‘delete everything’ button. Once the 2018 ban took place, these affordances made the choice of non-use an almost entirely social rather than a technical action.

While ascertaining all motivations for Tumblr non-use following the pornography ban is not possible – not least because those motivations may be implicit and overdetermined – analyzing some discussions around forms of non-use reveal that for these users technical steps in enacting non-use (such as logging out, archiving, and deleting) acted as complicating factors in the social decision to leave Tumblr.

Migration as social reconfiguration

In the previous section, we outlined forms of non-use of Tumblr in the wake of the 2018 pornography ban. We now turn to forms of migration away from Tumblr, but treat such migration as a form of ‘leaving’ that can be interwoven with the analysis of non-use.

The 2018 pornography ban had a massive impact on Tumblr users, but it did not occur in a digital vacuum. By that time, thousands of other platforms had been shut down, had mass exoduses, or undergone other patterns of user upheaval. And in many cases the response involved not simply ‘non-use’ but movement to other platforms. For instance, the non-use of Grindr often involved a movement to Scruff or OkCupid (Brubaker et al., 2016: 382). Other scholars have explored the movement from MySpace to Facebook in the 2006 to 2007 period in terms of ‘white flight’, or looked at the ‘Uru diaspora’, in which members of a virtual world that was shut down worked to recreate it in other virtual worlds (Boyd, 2012; Pearce, 2009). Indeed, the rise of Tumblr was linked to users leaving Livejournal after that platform engaged in bulk content deletion in 2007 (Dreyfuss, 2018). From examples like these we see migrations in which users are ‘leaving’ a platform or ‘being left behind’ by a platform. In the case of Uru, residents were left behind by a shifting virtual landscape, forced to forge alternative digital places. Livejournal and MySpace did not shut down, but governance shifts (e.g. the banning of content) or cultural changes (e.g. the framing of a platform in terms of race and class) shaped significant collective abandonment.

For the 2018 Tumblr migration, we can explore movement to other platforms by examining posts designed to facilitate reconnection. In the wake of the pornography ban, many users began posting their profiles, usernames, and accounts on other social media platforms. One posted: ‘In case Tumblr decides to nuke itself you can find me here’ followed by their handle on another platform. Another lamented that ‘starting on a new platform is very tiring, not to mention stressful’ but then listed some other places they could be reached and ended by saying ‘if and when I decide to migrate, I’ll be sure to make a post so that I can keep in touch with all of you wonderful people’. One emergent social practice on Tumblr was known as ‘mutuals’, which referred to users who reciprocally followed each others’ blogs. This ‘mutuals’ relationship was typically seen as form of friendship more meaningful than the unidirectional ‘follower’ relationship, even when direct messaging was infrequent. Illustrating this, one user posted a humorous internal dialogue that began ‘Me: *wants to talk to my mutuals because, in my head, we’re friends*’ and continued, ‘Also, me. Just thinking of dm-ing [direct-messaging] anyone:’ followed by a gif of a popular internet creator screaming in mock terror. Alongside posting information about other platforms, another common practice was for users to inform their mutuals that they should message the user in order to decide how to stay in contact after the user had left Tumblr. This practice is strongly linked to the process of migration because it involves the attempt to forge a Tumblr-like sociality beyond the platform in a manner recalling aspects of the Uru diaspora. Unlike the Uru diaspora, however, the 2018 Tumblr pornography ban did not involve the wholesale shutting down of a platform, but a change in governance practice that led many (but not all) users to conclude the platform had become incompatible with their social and expressive goals.

Tumblr, as a corporate-owned site of culture making, has an experience tied to both the technical policies and affordances of the platform as well as the social context of the ‘dashboard’, the feed of followed blogs displayed to users when logged in. Posts leading followers and mutuals to new venues for interaction indicates that Tumblr users were not primarily leaving the people they had met or the communities that they had formed on Tumblr. Instead, Tumblr users were leaving the platform in an act of shared defiance against a corporate decision. Network effects, generally used to model adoption and diffusion of technology, could help explain patterns of virtual migration as well (Baumer et al., 2013). Even as many Tumblr users signaled their willingness to reconnect on other platforms, research on platforms like Facebook suggests that users sometimes hesitate to reestablish connections through other means (Baumer et al., 2013: 3262). Socialities are not necessarily platform agnostic. One of the advantages many users saw to Tumblr was that it was isolated from other forms of online sociality, especially those associated with their offline identity. One user, discussing reasons for staying on Tumblr, argued that it was ‘the only godforsaken place where I can be open online without my f***ing family finding it’. From the disconnect between Tumblr and ‘real life’, it follows that many people who communicated primarily through Tumblr may have had hesitated to reconnect on other platforms, and therefore been more likely to lose touch. Such fracturing of social networks can sometimes be seen as a positive consequence of platform migration. Birnholtz (2010) studying students who decreased or stopped usage of instant messenger (IM) technologies after entering college, found that one of the problems of using the IM system for a long period is that there was no ‘graceful degradation of social ties’ (p. 1432). At issue is that the affordance of some platforms leads to a sociality in which users are as visible to acquaintances and former friends (weak social ties) as to current friends (strong social ties). This can lead to tensions when users find themselves spending time on connections that are not priorities for them (Birnholtz, 2010: 1431).

Birnholtz’s analysis can inform our own in addressing how decreases in social contact due to Tumblr migration might have been experienced as beneficial. Tumblr has included an IM feature since 2015 which includes a green dot indicating if the user’s account has had any recent activity (Perez, 2015). Birnholz’s findings about IM applications can be applied to Tumblr users who are also likely to may have had shifting levels of closeness with followers and mutuals. While the activity indicator can be turned off, Tumblr’s default settings and social norms are such that any visible posting or reblogging can signal online or available status. Posting can thus open Tumblr users to the same kind of distracting interjections from acquaintances that Birnholz saw with his participants.

Based on posts from Pillowfort, a small-scale Tumblr alternative that received a boost in attention and traffic following the pornography ban (Alexander, 2018), former Tumblr users were encouraged to leave behind old feuds and ‘start from scratch’ when coming across people on new platforms. Pillowfort hosts over a dozen communities related to the Tumblr ‘exodus’, many of which directly refer to their members as ‘Tumblr refugees’. One such community has over 1000 members and includes a post calling Pillowfort ‘home’ now. Thus, even the visible signs of communal migration can contain unforeseen erosions of the networks that Tumblr as a platform once supported.

A unified approach to non-use and migration

If we consider non-use as a series of socially-situated, technically-enabled decisions and migration as a shifting and reconfiguring of platform-specific social relations, we can begin to develop a unified approach by reading them across one another. That is, by applying the analytical insights from study of non-use to migration and vice versa, we can understand them as unique but overlapping aspects of digital exodus.

Considering migration through the lens of non-use elucidates in particular two aspects of migration: one, the degree to which digital migration is a series of social actions; and two, the ways migration is disincentivized by the technological affordances of the platform. Migration, like non-use, is not a single decision. It is an ongoing and often partial process wherein users practice breaking patterns of use (‘kick[ing] it for good’) and attempt to build a new social ‘home’ on the replacement platform. The process of ceasing to use one platform (non-use) is technologically disincentivized, as already discussed. But building new spaces for ‘refugees’ post-migration is also made technically difficult. Tumblr users, during the move to other platforms, discussed hashtags and other tools to reconnect on other platforms such as Mastodon. However, 2 years later, the Tumblr migration’s half-life on other platforms seemed relatively short. On Pillowfort, the Tumblr refugee communities, despite having large numbers of members, were largely inactive by 2020. For instance, none of the discussion posts regarding the largest ‘Tumblr refugee’ community had any activity from February to June 2020. During this same period, searching for discussions of Tumblr’s migration on Twitter resulted in multiple small threads lamenting the effect new Tumblr emigrants had on Twitter, but no comments from users themselves. In a sense, digital platforms reinforce a certain amount of cultural assimilation, a constant presentism due to the lack of infrastructure to signal previous platform allegiances.

Non-use, like migration, can reconfigure platform-specific social processes, even when no replacement platform is used. Just as weak ties are often left behind when changing platforms, non-use can have the benefit of trimming unwanted social relationships. Baumer et al.’s (2013) findings on connections lost when leaving Facebook – that is, that users are hesitant to reconnect with platform-specific friends when they move to other digital spaces – are just as true of strict non-use as of virtual migration. Non-use, like shifting usage, is platform specific: the enactment of non-use varies depending on the platform in question. For Tumblr users, signing out in the wake of the content ban was seen as a significant mode for communicating displeasure with corporate decision-making. At the same time, the mass abandonment or closing of a platform, if not accompanied by a shift to a new digital space, represents the end of a platform’s unique culture. Thinking about non-use through migration shows starkly that non-use is an end, a death of a particular social formation. Migration, be it virtual or physical, always has both an origin and a destination. Non-use, on the other hand, is a removal without replacement.

A unified approach to digital exodus may thus concern itself with several things. It can address the degree to which specific technological capacities of platforms shape how users leave and to where they go. Similarly, if a platform’s culture shapes its use, we may consider how that unique social formation lends itself to either dissolution or reconfiguration. Another question involves how, if at all, new platforms both technically and culturally retain traces of previous digital spaces – or if they mandate assimilation into the new platform’s paradigm.

Conclusion

Human agency, structural or infrastructural limitations, and existing social relations all play into shifts in physical geographies. Virtual migration and exodus are likewise shaped by personal desires, corporate ownership, and the cultural milieu of the original and replacement platforms. The various acts and behaviors that constituted leaving Tumblr offer different insights than previously studied types of technological non-use, abandonment, or migration. Specifically, we have read the multiple sociotechnical actions of non-use against and across the reconfiguration of platform-informed sociocultural formations. This unified approach to digital leaving highlights the way technical affordances, platform particularities, online culture, and individual sociality all interact when pressed by changing digital geographies.

In the 18 months after Tumblr’s porn ban, the site went through various ‘shrinking pains’, including a significant market devaluation and a sustained dip in traffic and app downloads (Leskin, 2019). Tumblr’s 2019 sale to Automattic for less than $3 million became another meme on Tumblr and other platforms. There are, of course, people still using Tumblr, the first author among them. However, the popular press narrative, as well as the assumption on many other social media sites, is that Tumblr is essentially dead. Previous work has explored the ‘digital afterlife’ or ‘legacy’ of individual users of social media platforms (Brubaker et al., 2014) but Tumblr raises the issue of how to study the digital afterlife of platforms themselves and the impacts of different leaving processes on both desist-ers and persist-ers. As Bucher (2020) so poignantly argues, we must ‘disconnect from the seductive idea of individual and voluntary non-use’, and instead question ‘what it means to be many’ and thus socially and collectively seek digital disconnection in a ubiquitous digital age.

The data suggest that forms of digital leaving, like older forms of migration, are often incomplete and imperfect disconnections. And non-use, unlike death, is often partial and always revocable. The social and cultural changes resulting from the Tumblr migration are difficult to assess but they raise questions about the constantly shifting sands of digital spaces and the tendency away from digital vagrancy. There are no affordances for retaining an individual or collective social media lineage as platforms change and die. And yet, the digital spaces we occupy at different times, our online ‘homes’, have huge effects on us. From our friend groups and self-conception to the memories and archives we retain of our earlier selves, a platform’s technicality and culture shape us the more we use it.