A new memory ecology, social media, and algorithmic power

In Hoskins’ (Reference Hoskins and Hoskins2017a) discussion of a new memory ecology, two concepts are of particular interest for this article. The first is ‘sharing without sharing’. With this concept, Hoskins argues that people feel active when they post, record, and react to digital content, and consequently, many do so compulsively. Sharing has become a digitally fostered and unconscious ‘obligation to participate and to reciprocate’, which entails that digital memory practices happen without any real intent or goal. They become ‘a continuous end in itself’, with no ‘prospect of arrival, settlement, or completion coming into view’, Hoskins writes (Reference Hoskins and Hoskins2017a, 29). The idea of a sharing without sharing is closely connected to the concept of ‘grey memories’. Hoskins argues that in the new memory ecology, how we experience our personal memories has become blurred and obscured. Our memories have turned grey, as we have lost control over where they are, how they are displayed, and to what end (Hoskins and Halstead Reference Hoskins and Halstead2021). The concepts of sharing without sharing and grey memories offer a critical and also rather pessimistic understanding of what memory culture is in a present that is ‘emptied of meaning’ (Hoskins Reference Hoskins and Hoskins2017b, 102). Hoskins’ account of contemporary memory culture, however, has not gone unchallenged (Halstead Reference Halstead2021), and this article is a further attempt to offer an empirically based assessment of the ideas of a sharing without sharing and grey memories.

In the growing body of research on digital memory, memory work on social media platforms has garnered much attention (e.g. Garde-Hansen Reference Garde-Hansen, Garde-Hansen, Hoskins and Reading2009; Smit Reference Smit2018; van Dijck Reference van Dijck, Bond, Craps and Vermeulen2017). This research shows that memory practices on social media offer a sense of belonging and meaning (e.g. Ekelund Reference Ekelund2022a; Lagerkvist Reference Lagerkvist2014); however, social media platforms push their users towards less insightful memory work (Ekelund Reference Ekelund, Glaser, Håkansson, Lund and Lundin2022b; MacDonald Reference MacDonald, Arnold-de Simine and Leal2020). Studies have also argued that Facebook in particular has become a platform characterised by affective and nostalgic memory work (Davalos et al. Reference Davalos, Merchant, Rose, Lessley and Teredesai2015; Ekelund Reference Ekelund2022c). All in all, this illustrates the complexity surrounding how memories are used and practised in the new memory ecology.

Social media platforms are increasingly viewed as socio-technical actors that enable, shape, and interfere in memory practices (Smit et al. Reference Smit, Heinrich and Broersma2018; van Dijck Reference van Dijck, Bond, Craps and Vermeulen2017). Within this research, a critical perspective on how algorithms work is integral. Studies have discussed how algorithms encourage us to remember as well as forget (Jacobsen Reference Jacobsen2021; Makhortykh Reference Makhortykh2021), and in addition, how algorithms sort, select, and display certain memories (Jacobsen and Beer Reference Jacobsen and Beer2021a). This leads to the quantification and automatisation of ‘likeable’ memories (Jacobsen and Beer Reference Jacobsen and Beer2021b) by which social media platforms are able to promote an immediacy and a rapidness in memory practices (Ekelund Reference Ekelund, Glaser, Håkansson, Lund and Lundin2022b; Kaun and Stiernstedt Reference Kaun and Stiernstedt2014).

With a few exceptions (Jacobsen and Beer Reference Jacobsen and Beer2021a), existing research on digital memory has thus far not studied closely how users of social media platforms experience and deal with the affordances of the interfaces and the algorithms. This article contributes to the existing body of research by investigating the experiences and memory practices of members of retrospective Facebook groups and is informed by previous research on algorithmic media and algorithmic power (Beer Reference Beer2017; Bucher Reference Bucher2018). These studies illustrate how algorithms produce the threat of invisibility for users (Bucher Reference Bucher2012) by shaping what is seen and what is made invisible on social media platforms (Jacobsen Reference Jacobsen2023). Furthermore, research shows that posts sometimes disappear on Facebook due to algorithmic interventions (Villegas Reference Villegas2016), and that Facebook produces ‘digital voids’ (Jacobsen Reference Jacobsen2021) by classifying, sorting, and selecting which posts should be resurfaced as happy memories and which could potentially be upsetting and unsettling. These results strengthen Hoskins’ argument that the connective turn has brought with it a greying effect. With concepts such as ‘algorithmic imaginary’ (Bucher Reference Bucher2017) and ‘algorithmic literacy’ (Swart Reference Swart2021), studies have also investigated how people experience and make sense of algorithms on social media. A central point in these studies is that algorithms become most visible for users when they produce irrelevant or uncanny results.

Theory and method

Theoretically, the article draws from Deborah Lupton's (Reference Lupton2020) concept of ‘data selves’. With this concept, Lupton builds an understanding of the relationship between people, data, and technology. She writes (Reference Lupton2020, 12–13),

In this approach, the entanglements of people and technology become an existential question, and experiences of ‘doing data’ – or in this article, ‘doing memory data’ – become practices that are simultaneously entangled, shared, and personal. ‘Doing data’ (Lupton Reference Lupton2020, 74), which is a particular perspective in Lupton's theory, focuses on understanding the ‘performances, enactments, and sense-making’ when people engage with data in relation to digital settings. Therefore, with the concept of ‘data selves’ and its perspective on ‘doing data’, our interest lies in exploring what people do with data and what data does to them (e.g. Jacobsen and Beer Reference Jacobsen and Beer2021a).

Lupton opposes the common view of digital data as immaterial, invisible, and intangible (Reference Lupton2020, 19), arguing instead that digital data are highly material given that they are generated through material devices, materialised in ways and formats that ‘invite human sensory responses’, and have ‘material effects on human bodies’. This thinking moves us away from perceiving digital data and practices of producing digital data as abstractions to doing data within human–data assemblages as an affective, bodily and material process. Using Lupton's concept of data selves, the article's analysis highlights everyday embodied experiences of doing memory data within a certain human–data assemblage. Furthermore, this leads to an analytical interest in the hybridity of contemporary memory practices. While much research on digital memory practices focuses primarily, if not exclusively, on the interactions that take place within the platform and its interface (e.g. Davalos et al. Reference Davalos, Merchant, Rose, Lessley and Teredesai2015; Robards et al. Reference Robards, Lincoln, Pinkard, Harris, Dobson, Robards and Carah2018; Smit et al. Reference Smit, Heinrich and Broersma2018), an interest in how people understand, experience and deal with memories in relation to the affordances of the platform calls for a hybrid approach (Merrill Reference Merrill, Wang and Hoskinsforthcoming/2024). Such an approach studies not only the online interactions but also the practices that take place offline in order to understand how these intersect and reroute each other (Gajjala Reference Gajjala2019). The method used in this article is semi-structured qualitative interviews (Brinkmann and Kvale Reference Brinkmann and Kvale2018) with nine members of retrospective Facebook groups (see Reference EkelundEkelund forthcoming/2024 for a more elaborate methodological discussion). The primary selection criterion was that the member should be active in their group in terms of making, reacting to, and commenting on posts, as this shows they have enough relevant experience of the group. Seven of the interviewees are members of local history groups that focus on sharing memories from a particular town or city. The remaining two interviewees are members of groups that focus on sharing memories from a specific decade when they were young – more specifically, the 1960s. Many of the memories the interviewees (as well as other members of these groups) share are directly connected to themselves or their family, most often in the form of photos from family albums, but they also strive to find memories and facts of a more general character to be shared in the groups, for example postcards and newspaper clippings. The interviewees were selected to represent the overall gender and age balance of the groups, resulting in interviews with five men and four women, between 64 and 71 years old. The age category is important, as previous research indicates that older people tend to have less algorithmic literacy (Swart Reference Swart2021), which means that the age of the interviewees shapes the nature of the hybrid practices (and the kind of hybridity) that the article investigates.

The interviews focused on three topics: (1) what the groups are about and how they work, (2) the interviewees’ practices within or related to the groups, and (3) Facebook as a platform for group-based memory practices. During the interviews, I used the experiences and insights gained from my netnographic fieldwork in retrospective Facebook groups (Ekelund Reference Ekelund2022c; Reference EkelundEkelund forthcoming/2024) to phrase the questions and introduce the topics. This helped the interviewees to describe aspects such as how they experienced the platform in more elaborate detail. It also helped to make the hybridity of online and offline practices more visible. The interviews lasted between 1 and 2 h and then were transcribed in full and coded into themes. The interviews followed the relevant ethical guidelines, including informed consent (Swedish Research Council 2017). All the interviewees’ names have been pseudonymised.

The first step in analysing the data was coding and thematising. In this phase, much of what the interviewees talked about could be related to two overarching themes, namely, aspects of finding and sorting, as they talked extensively about their process of locating and selecting ‘relevant’ memories to share within their group. The themes of finding and sorting also correlate with Hoskin's discussion of how the connective turn has resulted in new ways of finding and sorting (Reference Hoskins and Hoskins2017a). In the second step, the analysis zoomed in on these themes, and the results are presented in the article's four subsections: the first two focus on finding and the latter two on sorting. Using these themes to structure the analysis, the article provides empirically based insights, as well as nuances, to the discussion on how the connective turn affects contemporary memory practices.

Finding retrospective Facebook groups

The interviewees all aspire to actively contribute to their respective groups, share relevant content, and participate in the discussions. Therefore, they need material to post and share. For eight of the nine interviewees, this is not a problem – at least not in the beginning. They talk about having inherited photo albums, folders, binders, and boxes of postcards, documents, and newspaper clippings from their parents and ancestors. Bertil describes how he has 30 moving boxes of ‘stuff’ from his father and grandfather stored in his basement. On an individual level, this shows an abundance that is not linked immediately with the connective turn. ‘Everyone sits and has old binders. We have all inherited binders’, Bertil explains, as if trying to convince that this type of inheritance is common for people engaged in retrospective Facebook groups. Judging from the interviews, this seems correct. A common thread is that the interviewees have been given boxes and boxes worth of material that they at first did not know what to do with (c.f. Gudmundsdottir Reference Gudmundsdottir2022). ‘I was almost on the verge of starting to throw [the moving boxes] out’, Bertil explains, ‘but then Facebook appeared. Okay, so I can share the pictures there, and then I can throw them away’. This shows that the initial problem was not about finding data but rather finding somewhere to put it – storing 30 moving boxes takes up a lot of space – and more importantly, deciding what to do with the data. Lotta talks about the same issue when describing how she inherited a large collection of postcards, photos, and books from her father.

What Lotta as well as several other interviewees describe when talking about the material they have inherited can be interpreted as having been given a burden, a weight, and a responsibility. The boxes contain data that connects to the past – mnemonic objects that illustrate and tell stories. Kept in the boxes, these memories are mute. Even when scanned and made into books for oneself and one's near family, the memories can only resonate so far. Listening to the interviewees, it appears that the materiality of the objects – the binders, albums, and indeed, the large size of the boxes that take up considerable space in their houses and flats – have a sensory and material effect on them and their perception of the world (c.f. Lupton Reference Lupton2020). Doing data memories, therefore, did not start by joining a social media platform and becoming a member of a retrospective group, but rather the Facebook groups provided the interviewees with an outlet – an opportunity to do something with the material they already had, to share it and thereby give life to the memories previously contained in boxes. The retrospective groups thus gave them a way to unburden themselves, enabling them to carry the responsibility they felt they had been given when inheriting the boxes.

Once using Facebook was decided, however, it meant that the interviewees were engaging with a socio-technical actor whose interface and algorithms not only enable but also shape and interfere. Among the research that has studied Facebook as a platform for memory work, an important finding is how social media platforms promote the immediacy of shared content and interactions. Kaun and Stiernstedt (Reference Kaun and Stiernstedt2014) have discussed Facebook in terms of ‘social media time’, meaning that the logic of the platform prioritises newness and immediacy in order to engage users longer and more often. They argue that this results in a constant production and collection of new data to be posted. Research on memory communities on Facebook have likewise stressed that the affordances of the platform produce a hurried and fragmented form of memory practices – a production of instant memories that are selected and shared for an instantaneous response from others (Ekelund Reference Ekelund, Glaser, Håkansson, Lund and Lundin2022b). This connects with Hoskins’ (Reference Hoskins and Hoskins2017a) concept of sharing without sharing, and the idea that digital technologies make people share memories compulsively and unconsciously rather than pursuing memories with an active and actual awareness. In the interviews, it is clear that the interviewees experience the rapidness promoted by the platform. Having started with boxes and boxes of stuff and seemingly an abundance of data, this plenitude, however, soon faded after the constant posting of content in the Facebook groups. For example, Peter explained that some time ago ‘they’ – meaning he and the two other group administrators – ‘had emptied all the archives’. As a result, they now had to try to find new data. Of the interviewees, only Bertil still has boxes of material to go through (approximately 10 more moving boxes of the 30 he started with).

Many of the interviewees first turned to using Facebook groups to be able to do justice to the memories and stories contained in the boxes they inherited and had thus fulfilled their goal. However, the weight and responsibility of having inherited boxes of memories seems to have been replaced by the weight and responsibility of keeping the retrospective Facebook groups going. The interviewees talk about feeling ‘committed’ to the groups, being ‘afraid’ that the groups will be ‘exhausted’ or ‘irrelevant’, afraid that the memories, the knowledge, and the stories that they and others have shared in the groups could ‘disappear’. Engaging in sharing memories in the groups has thus resulted, for several of the interviewees, in an emotive relationship with not only the memories and connections with the past but also with the constant flow of data in the groups.

The interviewees talk about their constant and active search for new data to post in the groups. A couple of them even describe themselves as ‘detectives’ in their ‘hunt’ for, most often, photos and postcards to be shared. Hans describes himself and other members of the group that he is engaged in as ‘photography detectives’. Peter talks about some members as ‘time detectives’: they have become ‘brilliant’ at ‘looking at clothing and other details’ in order to put a date on specific photos. He also talks about himself as ‘perpetually on the lookout for “accidental findings”’, meaning that he always hopes to find material that can be shared in the group, even when he is not actively searching for it.

The interviewees thus seem to be consumed by their engagement in the groups and by their feelings of responsibility towards the flow of content in the groups. It takes up a large part of their everyday life, even when they are not on the platform. Again, this connects to Hoskins’ (Reference Hoskins and Hoskins2017a) idea of a sharing without sharing – that the connective turn has meant that digital memory practices become ‘a continuous end in itself’. The interviewees have reached a goal in sharing the data and the memories they first envisioned, but the affordance of the platform and the affective relationships they have built with the groups means that they keep on sharing without any real prospect of completion.

It is also interesting to take note of the interviewees’ use of language. They describe themselves and other members of the retrospective groups as ‘searching’, ‘hunting’, ‘being on the lookout’, and doing ‘detective work’. Hereby, they articulate themselves as highly active, engaged, and, most importantly, conscientious of their memory practices. While the platform pushes them towards immediacy, their response is to try and keep up the tempo with clear intent and purpose and an awareness about what they are pursuing. This nuances the idea of a sharing without sharing (Hoskins Reference Hoskins and Hoskins2017a), and several of the interviewees also emphasise that it is important to them that the groups are not, as Lena phrased it, ‘open to any kind of content’ and that it is crucial that the groups ‘keep [their] narrow focus’. Thus, the very character of the groups – that they are explicitly centred on often very specific aspects of the past – helps the interviewees maintain a focus in their memory practices.

Finding more memory data (and then more…)

As Facebook promotes a newness principle (Kaun and Stiernstedt Reference Kaun and Stiernstedt2014), the platform becomes a catalyst for how its users engage and interact, and this also applies to retrospective groups (Ekelund Reference Ekelund, Glaser, Håkansson, Lund and Lundin2022b). It prompts members of the groups to search for new content; for example, the interviewees speak of how they go visit the library, local folklife archives, city and town archives, local museums, and memorial sites in search of data. They also browse the internet, search Google, visit digital collections made available by museums and individual enthusiasts, and they buy and trade old postcards on online marketplaces. Often, these activities are focused on finding relevant image material. This shows that the operative logic of Facebook not only promotes immediacy but also interactions between its users that are image focused. Posts that include, or focus on, photographs are also the posts that receive the most likes and comments (Malik et al. Reference Malik, Hiekkanen, Dhir and Nieminen2016), as they are coded as more relevant and thus made more visible by the platform (Bucher Reference Bucher2018). It is therefore not surprising that most of the activities involving the interviewees’ memory practices are image centred, given that this is what the digital platform encourages.

The search for sharable photographs took an unexpected turn for one of the interviewees, Monica, when other members of her Facebook group, which she created and administers, began to contact her, offering access to their family albums. As a result, at least once a month, she packs her digital scanner into the trunk of her car and drives to those who have contacted her. Often, they are elderly people who have neither the equipment nor the expertise to scan and digitally share their photos but have heard of Monica and the Facebook group she created:

The constant hunt for new photos has thus produced a local and micro-level memory economy where Monica gets new material for the Facebook group, while others get a copyable, sharable, and improved digital version of their physical photo albums. The hunt for old photographs that is produced, or at least seems to be sped up, by engaging in retrospective Facebook groups, produces not only intensified interactions between group members online but also between people outside of social media platforms in offline social interactions.

Doing data memories is always a hybrid phenomenon where the old photographs are as crucial as the digital technology of the scanner or smartphone. The latter helps translate the former into a digital existence. In addition, there is the bicycle that Erik rides to the local archive, Monica's car, the stairs to Bertil's basement, the computer screen, the keyboard, and the mouse used by Peter when he visits digitised archives. These material aspects of the hybrid memory practices have an affordance in the same sense as the operative logic of the digital platform (c.f. Lupton Reference Lupton2020). While Facebook promotes speed and immediacy, the physical attributes of the scanner, the bike, the car, and the bodies of the people who engage in the retrospective groups produce a slowness and a friction that contrasts with the seemingly uninterrupted speed of digital existence. This slowness of material existence seems to help and encourage the interviewees in continually reflecting on their doings of data rather than making their memory practices blurred, obscured and grey (Hoskins and Halstead Reference Hoskins and Halstead2021) – even though they have no control over what happens with the memory data once they share it online.

In addition, several of the interviewees stress that they are constantly searching for information and facts about the past in question. Anders describes how he makes visits to the library, not to search for photos but rather the small facts and details around a particular street, a date, a company, or a person. Another interviewee, Madeleine, has taken several guided tours in her home city, and she has visited open lectures organised by the church and the local cultural history association. Peter explains that he uses Wikipedia often and that he has ‘gone into the church records’ on other occasions, and Erik speaks of visiting local second-hand shops in search of books about the history of his hometown. He also emphasises how he has saved all the information leaflets that were sent out to his community when a new district was built nearby, for future use and for the possibility of finding interesting information about the area. ‘I collected every single one of those’, he enthusiastically acknowledges.

The engagement in retrospective Facebook groups thus seems to produce a constant search for and collection of data. The search is primarily for image material, for example, old photographs and postcards, which is what the platform promotes, but it can also be in the form of knowledge about the past that is discussed in the groups. Peter describes how this has evolved in the group that he is a member of:

While the logic of the platform promotes a constant hunt for new data memories, which connects with the idea of a social media time and an instantaneousness (Ekelund Reference Ekelund, Glaser, Håkansson, Lund and Lundin2022b; Kaun and Stiernstedt Reference Kaun and Stiernstedt2014), the actual interactions in the groups produce reflections and queries that prompt the interviewees to become more conscious of what it is they are doing. This reflective process is sped up by the constant flow of new content in the groups, but the memory practices that take place in connection with the retrospective groups still nuance the idea of a sharing without sharing (Hoskins Reference Hoskins and Hoskins2017a).

Sorting memory data

One of the main characteristics of Facebook's operative logic is that it curates the flow of content by personalising and constantly redistributing it based on what it understands as relevant, interesting, and ‘hot’ (Bucher Reference Bucher2018; van Dijck Reference van Dijck, Bond, Craps and Vermeulen2017). The default setting when browsing a Facebook group is that the interface will present the ‘most relevant’ posts at the top of the timeline. This curation is based on whether a single post is new, if it recently has received many likes, reactions, or comments, and if it is posted by a member whose posts you yourself usually react to, comment, or like. This means that the groups’ timelines do not sort themselves chronologically or collectively, but rather the structure and order of the timeline presents itself as constantly changing and as personalised. This connects with Hoskin's concept of grey memories, in that individuals have no control over how the content they share is displayed and to what end (Hoskins and Halstead Reference Hoskins and Halstead2021). How, then, does this affect the interviewees’ memory practices?

When asking this, it is important to take note of how the interviewees describe the retrospective groups. Several of them talked about the groups as ‘archives’. Ulla, for example, described the group that she is a member of as ‘a wonderful little archive of pictures from the past of the city’. Hans and Anders both referred to the practices in their respective groups as ‘archival work’, and the groups were also described as ‘collections’. In the interview with Christina, she also talked about the group as ‘a collection of memories and stories from generations back’. Peter used the same concept when he talked about how he and others are ‘adding to the collection of information and memories’ of their city.

The concepts of an archive and a collection are common in memory practices and memory work. They have also been used to discuss how Facebook and other social media platforms affect contemporary memory culture (e.g. Garde-Hansen and Reading Reference Garde-Hansen, Garde-Hansen, Hoskins and Reading2009). Nevertheless, the notion of the archive and collection harken back to a pre-social media age and an idea of a systematic order and categorisation (Bowker Reference Bowker2005). The use of these concepts to describe the retrospective groups are central to interpret not only how the interviewees understand the groups but also to construe how they make sense of doing memory data in relation to a digital platform built on an operative logic of newness and personalisation.

When I asked the interviewees how they have experienced the interface and the arranging of the groups’ timelines, several of them said that they had not noticed or thought about this, which indicates low algorithmic literacy (Swart Reference Swart2021). This is not surprising, as this is how the algorithms are supposed to work on social media platforms – they are ‘barely noticeable’ (Bucher Reference Bucher2018, 94). Nevertheless, when explicitly asked about it, the interviewees quickly recognised that the group content and its timeline often seemed elusive to them. For example, Christina described how she, almost subconsciously, had developed a technique where she takes screenshots of posts and photos that she finds interesting: ‘I've learned that, if I don't take a screenshot, I won't find it again. […] So that's what you have to do if you want to have a chance to go back [and find a post].’

Another example is Hans’ reaction when I ask his opinion about how Facebook presents the group content on the timeline:

Even though a couple of the interviewees said that they had not actively noticed or reflected on how the interface of the platform curates the groups’ data, they, as in the cases of Christina and Hans, still seemed to have developed strategies and methods for dealing with and making sense of how the platform works. Most importantly, their strategies and methods seem to be about being able to sort the data. Christina's many screenshots are not only a technique to save specific content, but they are also a way for her to categorise and organise the data. The screenshots allow her to sort the selected content in specific albums on her smartphone so she can retrace specific themes and discussions in a way that is not possible within the interface of the retrospective group. The interviews thus shed light on how individuals, consciously or not, deal with the greying effect produced by the operative logic of the social media platform.

Another of the interviewees, Erik, also said that he had not thought about how the content shared in his group is arranged by the interface, although he had noticed that it was difficult to maintain the systematicity that he envisioned for his posts. In the interview, he emphasised that he wanted to ‘keep a systematic order’ with his posts. This, he described, included both a thematic and a chronological selection of the postcards he was currently going through. This systematic order, however, was nullified as soon as his postcards were shared, he explained. While Erik did not seem worried or upset about this – ‘This is just the way it works on Facebook’ – he stressed that the sorting of photos and postcards took up a lot of time and was important to him. During the interview, he showed me bookshelves and cabinets filled with binders, folders, and albums of all the material he had carefully gone through and sorted.

This act of carefully sorting the physical data, of going through all the boxes one had inherited, of categorising the ‘new’ data one had acquired, and of putting scanned material in specific folders on one's computer was something all the interviewees talked about as important to them. Sorting is thus an integral aspect of doing memory data, but this happens primarily outside of the retrospective groups and the digital interface of the platform. Thus, studying the memory practices that take place in connection with digital platforms as a hybrid phenomenon illustrates how individuals strive to maintain control over where their memories are stored in the real world, how they are displayed, and to what end, even if they have no control over this within the digital platform.

Sorting the timeline

While several of the interviewees said they had not thought about how Facebook arranges the group content, others stressed that they not only thought about it but had actively tried to create more order and find a more structured way of approaching the groups’ timelines. Two of the interviewees described that they had started using the group search function to handle the increasing quantity of data in their groups. Bertil talked about the search function as something he tried to promote within the group. He also described how the messiness of the group content meant that some photos were shared repeatedly and that some questions were constantly asked. The group search function, he argued, could be a solution to the problem:

In Facebook groups, members can search for content that has been shared in the group by clicking a search button. A window pops up asking you to type in your search query for ‘posts’, ‘comments’, or ‘members’ within the group. The user is then taken to a page where you can scroll through the search result. The ranking of the search results seems to be based on a relevance principle, but users can also choose different ‘filters’ to narrow down the selection. One can click to include only the posts one has already seen to only see the ‘most recent’ posts, and one can select criteria for ‘posted by’, ‘tagged location’, and ‘date posted’. Based on the interviews, however, the search function does not seem to be used much. As Facebook promotes a focus on images, as discussed above, this means that a text-based search function may not be as useful as it first seems. To find a specific post, one must know either who made the post, roughly when the post was made, or which specific words were used.

The interview with Peter kept coming back to how Facebook's interface makes it difficult to navigate the group's content. He would like the search function to be more useable:

Peter's thoughts illustrate how some of the interviewees struggle with navigating and making sense of the groups’ timelines. As Facebook is interested in its users continuously making posts – regardless of whether the content of the post is new or it has already been shared – the platform is also less interested in its users being able to successfully relocate old posts. Therefore, the difficulties in using the group search function can be understood as serving the purposes of the operative logic of the platform rather than the interests and logic of the members of retrospective groups.

Another strategy that two of the interviewees spoke about is ways to bundle content, for example, by labelling posts with hashtags. Monica talked about having spent hours and hours in front of the computer working on ‘cataloguing’ the group content by adding hashtags to posts. Hereby, she had arranged posts in terms of school photos, aerial photos, portraits of people, photos of buildings, etc. The result is that members of the group can click on a hashtag to see all the other posts with the same tag. ‘I want it to be like at the library’, Monica said. Then she laughed before she continued: ‘I want to be able to categorise, to catalogue. Like you would if you worked at a research archive or a library.’

Peter also emphasised that he had found the ability to tag posts and photos as ‘probably the best way to create some kind of order’ in his group. During our interview, however, he explained that he had read that Facebook was going to disable this feature.

Peter seemed resigned when talking about this. His tone of voice and body language expressed powerlessness and sadness. All the work that he had put into sorting and categorising the data, for himself and the other group members, was to be obliterated. He returned to the subject later in the interview: ‘That function will disappear. And why they do it, I don't know. They haven't written why. It just popped up, saying “we're going to remove this”.’ When interviewing Monica a week later, I shared what Peter had told me. She was perplexed: ‘I haven't heard that, but it doesn't sound good. It would be disastrous. Even if it's not a good way to sort photos, it is still a way.’

The interviewees’ memory practices thus deal with what Bucher (Reference Bucher2012) has discussed as a threat of invisibility and what Jacobsen (Reference Jacobsen2021) has phrased as a sculpting of digital voids. The elusive character of the interface and the algorithmic curation of content seem to give the interviewees, Peter and Monica in particular, a feeling that the memory data they spend so much time to find, sort, and share can become invisible and unusable. This illustrates how the interviewees struggle with the greying effect of the digital platform and how the doing of data memories in connection with Facebook can become a practice that causes anxiety.

In addition to using hashtags, Monica had tried other ways to catalogue and curate the group content: ‘I have made “package posts”’ she explained, meaning that she had taken old posts, together with new content, and posted it as one post – a ‘package’ including several photos. This, she explained, lets other group members view the content as thematic photo albums. However, this strategy seemed to work only temporarily:

Just like Peter in the previous example, Monica was resigned when talking about this. Again, it is the logic of the interface that becomes problematic. If a photo that has been posted as part of a package receives a comment, it is understood by the interface as ‘more relevant’ and is therefore displayed on its own. Monica describes this as violent and destabilising. The photos are ‘torn apart’ and ‘come loose’. In a sense, then, she feels that both her memory work, and the memories that she wishes to share, becomes blurred, obscured, and greyed (Hoskins and Halstead Reference Hoskins and Halstead2021). Her use of words illustrates both the resignation she feels as well as how the disordered and fragmented character of the interface violates her ideas of order and systematicity. However, understanding that this way of curating the group data does not work as intended, Monica also learnt something about how the interface works and how she can manipulate it for the purpose of the group:

By ‘lifting’, Monica means that she locates old posts that she thinks will be relevant for other members of the group, and by making a comment as brief as a punctuation mark on the post, she tricks the interface into thinking that it has become more relevant again, and it is thus displayed in the top sections of the timeline. In a sense, then, Monica is working with the algorithms and in the same way as algorithms work. She manually sorts and selects certain memories to be displayed, but while the automatisation of algorithms produces a quantification of memories (Jacobsen and Beer Reference Jacobsen and Beer2021b), her manual process is intended to produce a qualification of memories. However, this technique does not mean that she can produce a systematic catalogue of content in the way that she would like, but rather it means that she adapts to the fragmentary and messy character of the platform's operative logic and its greying of memories. The aspect of sorting, which is integral in the interviewees’ memory practices, thus seem to be a constant uphill battle when it takes place within the interface and when it must work together with the shaping power of the algorithms.