Online user misconduct and an evolving infrastructure of practices: a practice-based study of information infrastructure and social practices

Introduction

Moderating user-generated content online has become increasingly important among news outlets which strive to engage users in serious and responsible debate (Helberger, Pierson and Poell, 2018). While the practice of offering the public a space to participate in the discussion of news is very much in line with the Western idea of news media’s democratic role in society (Russell, 2011; Robinson, 2014), it can also facilitate contradictory practices among users in the form of various antagonistic activities (Hille and Bakker, 2014; Webb et al., 2015, Helberger et al., 2018). Just as the practice of engaging with users over news online brings forth a need for practices that deal with the disruptive user-generated content generated by this engagement, it is also shaped by infrastructural arrangements and the spatial and temporal connections that it enables (Shove, Watson and Spurling, 2015).

This study focuses on an information infrastructure which is composed of the practices of managing the comments sections below online news articles. Instead of framing tools (e.g., websites) and agents (e.g., social media providers or news organisations) as the architects of practices and networks, this paper emphasises the role which practices play in producing new actors and structures. By framing practices as the building blocks of information infrastructure, i.e. shared, open and heterogeneous networks consisting of human as well as technological components (Hanseth and Lundberg, 2001), we can acquire knowledge on how socio-technological change comes about. This study seeks to understand how fundamentally different practices for dealing with the highly relevant problem of online user misconduct are intermeshed into a heterogeneous and dynamic online media structure.

In this study, practice theory is used as a lens through which to gain a better understanding of the information infrastructure that is formed by a complex of practices for acting on user-generated misconduct online. It explores how a number of the practices for acting on user-generated activities which are perceived to be abusive are interconnected.

This study does not intend to explore practices such as trolling or cyberbullying; neither does it seek to label the different hostile activities that users may carry out online. While there is much research on the various forms which antagonistic user practices take online and on the kind of counter activities that online communities undertake to combat them (e.g., Binns, 2012; Golder and Donath, 2004; Hardeker, 2010; Buckels, Trapnell and Delroy, 2014), there is a noticeable lack of studies framing the kind of supporting structure that has grown up around the issue of dealing with user activities online which are considered offensive. Exploring the social role of the information infrastructure that interconnects many of the practices performed to deal with one of the urgent problems of our wired world may help us to understand how modern society is structured through socio-technical arrangements.

Information infrastructures and social practices

‘Infrastructures are built networks that facilitate the flow of goods, people, or ideas and allow for their exchange over space’ (Larkin, 2013, p. 328). When infrastructures are designed, or ‘grown’ (Edwards, Bowker, Jackson and Williams, 2009, p. 369), their intended purpose is that of a supporting structure and their social implications are rarely predicted (Rolstadsås, Henriksen and O’Sullivan, 2012). However, as Edwards et al. argue, ‘[w]e are transformed by our infrastructures at all social and organisational levels’ (Edwards et al., 2009, p. 372).

The technological achievements of recent decades have prompted a number of studies on the infrastructures that have been brought into being by the Internet. The advent of new information technology has given rise to a body of research on information infrastructures (e.g., Hanseth and Monteiro, 1997; Bygstad, 2010; Henfridsson and Bygstad, 2013). Hanseth and Lyytinen (2010) mention the Internet and industry-wide electronic data interchange (EDI) networks as examples of information infrastructures and argue that information infrastructures are ‘recursively composed of other infrastructures, platforms, applications and IT capabilities […]’ (Hanseth and Lyytinen, 2010, p. 1). However, contrary to what is often, and historically has been, the focus of many of these studies, infrastructures are not just composed of technical performance (Larkin, 2013). Sundin and Carlsson (2016) explore school teachers’ experience of supporting pupils searching for and assessing information when using search engines, and they understand information infrastructures ‘as networks constituting the conditions for knowing’ (Sundin and Carlsson, 2016, p. 990).

Just as technical systems and networked appliances may have the ability to compose infrastructures, this article argues that this may also apply to professional and everyday practices. In their study of a collaborative software system for a geographically dispersed community of geneticists, Star and Ruhleder (1996) conducted influential research on the relation between infrastructural arrangements and information processes. They explore the relationship between practice and technology, ask when, not what, is an infrastructure and argue that infrastructures transpire for people whose practices are linked through activities and structures (Star and Ruhleder, 1996).

More recently, social practices have received attention from researchers exploring different kinds of infrastructural arrangement. Simone (2004) extends the notion of infrastructure to people’s activities, e.g., economic collaboration, in deteriorated urban city spaces in Johannesburg, and in Elyachar’s study (2010) of social practices among economically disadvantaged women in Cairo, it is practices such as visiting friends which lay the foundation for a supportive and empowering network. In a study contributing to research on the problems causing climate change, Shove et al. (2015) analyse car dependence while exploring how infrastructures and complexes of social practices connect across time and space.

While there is a growing number of studies framing the infrastructural properties of practices, the information infrastructure research within information studies still tends to emphasise the materiality of technical systems rather than practices and performances (e.g., Ioannidis et al., 2005; Hanseth and Lyytinen, 2010; Bygstad, 2010; Henfridsson and Bygstad, 2013). The study of information processes in general tends to view technical objects as generators of practices rather than the other way around. However, although technical objects and practices are interlinked, there are advantages to framing the practices rather than the technical solutions in order to understand the complexity of an information infrastructure which connects different interactive practices within a participatory media environment.

Theoretical framework

Rivera and Cox (2014) assert that one of the goals of practice theories is to offer an explanation of how social actions support social structures. In this study, the application of practice theory is inspired by Schatzki’s (2002) notion of the social as a setting in which people coexist and interact with each other and with the technical solutions they use to communicate. As in Reckwitz (2002) and Shove, Pantzar and Watson (2012), material entities are conceptualised as being mobilised alongside different forms of competence when engaging in practices (Shove et al., 2015).

In the context of this study, material entities can be technological artefacts such as computers but also websites, online forums and the interface of online participatory platforms. The notion of practice as a pattern made up of an assemblage of distinct and often exclusive actions (Reckwitz, 2002; Cox, 2012) represents the arrangements of activities which people (e.g., journalists) and news organisations engage in when acting on aggressive user posts online.

Social practices are organised arrays of doings and sayings (Schatzki, 2002; 2010), and all social affairs embrace people who are interrelated to each other in infrastructures of interwoven timespaces (Schatzki, 2010). According to Blue and Spurling, ‘[t]echnological infrastructures bring practices together in ways that allow their mutual influence’ (Blue and Spurling, 2017, p. 28). However, as will be discussed in this study, practices also form infrastructures and allow technological change. But when is an infrastructure (Star and Ruhleder, 1996)? When do human performances connect and form a structure? Clearly, an infrastructure does not simply materialise into our lives (Niewöhner, 2015). Instead, it is a product of social practice within a to-and-fro dynamic (Karasti, Millerand, Hine and Bowker, 2016). Traits which define an information infrastructure are, for instance, its embeddedness into other structures, its transparency and its reach beyond singular exclusive events (Star and Ruhleder, 1996).

Data collection

The empirical material from which this study draws its findings is based on interview material. Ten semi-structured interviews with intermediaries from nine different public-service and private news organisations in four different countries (Sweden, Denmark, Germany and Canada) were conducted in the spring and autumn of 2017.

The sample incorporates the two Swedish national public-service television and radio companies Sveriges Television (SVT) and Sveriges Radio (SR), the two major Swedish newspapers Dagens Nyheter (DN) and Sydsvenska Dagbladet, the Danish public-service radio station Danmarks Radio (DR), the German public-service news outlet Tagesschau, the major German news website Spiegel Online (SPON) of SPIEGEL.net, a subsidiary of the publisher Spiegel Verlag, the local Canadian Newscap radio station VOCM of St. John’s and the daily Canadian newspaper The London Free Press.

The reason for conducting interviews in multiple countries was to be able to inquire into an infrastructure that is very much global rather than local. An interview guide was prepared to structure the interviews. At the start of every interview, the interviewees were requested to introduce themselves and talk about their work tasks at the news organisation for which they were working. Following this, a line of follow-up questions had been prepared about each news organisation’s Web presence and online user participation strategies before and after the introduction of social media. Particular attention was dedicated to questions regarding the transition of interactive online activities with users from the news organisation’s own website to social media sites such as Facebook. Next, the interviewee was questioned about various past and present undertakings carried out by them and their news organisation to ensure a smooth interaction with their users online (e.g., various forms of in-house moderation techniques, outsourcing etc.).

In order to avoid getting simplified descriptions of work procedures and encourage representations of in-depth experiences, the interviewees were, in addition to being asked to provide examples of techniques and strategies actually used, also asked what they perceived as a smooth interaction with users. Questions were also asked about the wishes and desires that the interviewees might have for a user interaction in general and how they experienced user comments that they perceived as aggressive or abusive. Although all the interviews were conducted following a similar basic structure, all the interviewees were encouraged to talk as freely as possible about their work and the challenges they face when managing user comments. Every interview was completed by asking the interviewees to add anything relevant they could think of, a possibility which was used by all but one interviewee.

The interviews lasted between 30 and 50 minutes. They were all recorded and transcribed by the author. The transcriptions are literal. However, when quoting the interviewees, a certain amount of linguistic editing was conducted and speech disfluencies such as fillers and non-lexical vocables have sometimes been edited out. When quoting participants from interviews which were not conducted in English, the quotes have been translated into English from Swedish, Danish and German by the author. All the participants were informed of the nature and purpose of the study. Their participation was entirely voluntary and any personal information has been anonymised.

The analysis was conducted through repeated reading of the transcripts. Activities aiming to counteract perceived misconduct which were mentioned in the interviews were highlighted and organised into Word tables where each column represented a practice encompassing a complex of activities. Then, by going back to the manuscript and through recurrent listening to the recordings, relations and connections between the extracted practices, as discussed in the interviews, were identified.

Findings

In the following, different activities which aim to enable the smooth participation of users online will be addressed in five sections: 1.) Practising user participation. In this section the news media’s practice of offering space online for users to participate in discussions of news will be examined. 2.) Engaging in the practice of content moderation. This section covers the explicit practice of content moderation which is carried out in order to keep comments sections free from any content considered to be offensive. In the sections 3.) Engaging in the practice of hashtagging and 4.) The practice of outsourcing content moderation two different forms of practices, i.e. hashtagging as a way to raise awareness about content that is considered abusive or otherwise improper and the practice of outsourcing as a way for some news media organisations to manage user participatory online spaces will be examined. Finally, in section 5.) Engaging users to keep online discussions civil, the practice of posting comments in online news media comments sections as a way of maintaining a civil online tone will be considered.

Practising user participation

Recent decades have seen the news media relocate much of its news coverage onto digital platforms (Mitchelstein and Boczkowski, 2010; Slaatta, 2015; Pew Research Center, 2016), and today many news sites also offer space online where readers may comment or contribute with their thoughts and ideas (Hermida and Thurman, 2008; Robinson, 2014; Lee, Lindsey and Kim, 2017). This interaction may take place on, for example, a news outlet’s own website or on social media sites such as Facebook and Twitter.

Many of the participants who, in the interviews, mentioned the early days of online user participation referred to it as a time of change. It was a time when they had to work out how they could best engage in new practices.

I used to work as editor until 2008. After 2008, I’ve been in charge of […]. That includes working with how we interact with the public. When I started in 2008, that meant only comments sections. […] Then we didn’t have the social, or we didn’t call it ’social media’. Later it has become [social media].

(Interview 5, March 2017)

When the news outlets first maintained comments sections for users to interact in on their own websites, they had to establish how these practices should be carried out, and it was a process which involved some trial and error.

There happened to be an error with the code and, instead of having to turn on [a comment function connected to articles on the website], every article was full of comments. It was pretty wild. It was possible to comment on everything on the site and the amount of comments we got that day was incredible.

(Interview 8, August 2017)

Later, when social media came onto the scene and the news outlets started to have a presence on sites such as Facebook and Twitter, the user interaction practices which had been applied in the early years of Web 2.0 had to be renegotiated once again.

To begin with, we used to be present on Facebook and that was, well, everyone started there and there wasn’t anyone who knew exactly what we were supposed to do or why. Then, it took a year or two before we found a strategy for how we could work.

(Interview 8, August 2017)

After a while, the news outlets established routines for their activities on social media. With the growing significance of social media, social media providers such as Facebook became important in the spreading of news (Hong 2012; Pew Research Center, 2016; Weber 2014). With a fundamentally interactive concept such as Facebook, which encourages set-ups with built-in space for interaction, it is hardly surprising that the journalists participating in this study found the idea of news stories on Facebook meaningless without their supplementary comments sections. When asked whether they would ever consider shutting off the plugins which allow readers to comment on certain articles, the participants reacted with scepticism.

But that’s the nature of that beast, right? If you’re on Facebook. The whole machine is ‘let’s have a discussion!’ Like, it’s all discussion based. The notion that you would say to the reader ‘not going to let you say anything on this story!’ That would look muzzling and so negative, right?

(Interview 10, December 2017)

Although some of the news outlets participating in this study still had comments sections on their own websites which were alive and thriving, the comments those discussions generate are fewer in number than those on Facebook. Therefore, it is hardly an overstatement to argue that the practice of reading and commenting on news on social media has, while overlapping the practice of reading on news outlets’ own websites, actually overtaken this.

Engaging in the practice of content moderation

Whether or not the news outlets decided to drop user interaction on their own websites when social media started to take over, the practices for allowing readers to participate produced a need for organisational change and involved the setting up of social media teams and the hiring of social media editors.

I’ve been in charge of social media for about two and a half years now. That position didn’t exist before me. In those days, it was the job of the Web editors to handle social channels. But later we created more and more accounts and there are more channels so [the situation] became pretty unsustainable.

(Interview 4, March 2017)

As the number of readers with a presence on social media increased, the need to adjust the clearly insufficient content moderation practices applying to comments sections grew, and practices for managing users online were developed further. Codes of conduct which had been established for interactive discussions on websites prior to social media were reinvented and incorporated into those dictated by social media providers such as Facebook.

When changes such as the moving of comments sections to a new platform occur, they affect the practices of everyone involved, and this contributes to the evolution of the practices that they engage in. While users could post comments anonymously on news outlets’ own websites, the regulations of Facebook prevent this; unless a user operates a fake account, their comments are signed with their own name. These regulations may not necessarily lead to an improved debate climate, but they may enable other practices such as the blocking of users by moderators.

What we noticed was, some years ago, when we swapped over to Facebook’s system of comments. And then we noticed that, you know, it became another tone. We saw that the comments on our posts on Facebook usually were much more relevant and interesting.

(Interview 8, August 2017)

Whether or not the assimilation of Facebook’s standards and practices for user interaction actually improved the tone in the discussion forums, they allowed for adjustments in how to respond to aggressive posts made by readers. If a reader violated the rule of posting under his or her name, the post could immediately be removed.

Often, when you notice someone who behaves badly in a thread, we go in and check [the person’s FB account] and, like. Hey! This account was created today! And then we’ll block that person.

(Interview 7, August 2017)

In this technologically fast-paced environment, there is a constant need for new practices to rectify the problems that arise when other practices fail. Disruptive individuals have been surfing the Internet since the very beginning (Brodnig, 2016; Donath, 1994). However, as more and more people went online and with the interactive environment of Web 2.0, the need for practices and stricter rules on how to deal with offensive activities became more urgent. With the wider range that news outlets gained through social media, they also gained new potential troublemakers and the situation created the need for new practices for managing the people whose activities were unwanted.

Engaging in the practice of hashtagging

Reacting against an aggressive online culture by engaging in the practice of hashtagging as a way to raise awareness and inspire thoughtfulness is a practice which has emerged through the social media society (http://www.jagarhar.se). One example of a more or less organised form of practising hashtagging is the non-profit organisation #JagÄrHär [#IamHere], which was founded by the Swedish journalist Minna Dennert in 2016. With the launching of sister organisations in other countries, e.g., Germany, where #IchBinHier was founded shortly after, it has gained in popularity and influence.

The organisation strives to improve the debate climate and tone of discussion online while arguing that hate and abuse undermine freedom of speech. #JagÄrHär describes its own work as an effort to change attitudes (http://www.jagarhar.se). Whenever any of their members come across a post which is considered aggressive and abusive, they make their presence known by hashtagging it.

The practice which #JagÄrHär activists were engaging in soon caught the attention of those journalists working with content moderation.

What we noticed was that when those #WeAreHere [referring to #JagÄrHär, which correctly translates as #IamHere] appeared. Yeah, the first days were just like, hey! Hang on a minute! Now, they’re discussing gender here?! I was a little puzzled. But then they also began spamming and then we sometimes had to remove them as well.

(Interview 7, August 2017)

The practice of hashtagging for a better tone in online discussions overlapped and connected with the practices of journalists engaging in content moderation.

[S]ince we started this [conducting stricter content moderation], we have an increased interaction. […] But then, of course, there could also be other reasons for that […]. One such reason could be the group #JagÄrHär. It’s quite a new thing […]. Their presence has the effect that others dare to speak up.

(Interview 4, March 2017)

Although there is no organised collaboration between news sources and #JagÄrHär, the presence of #JagÄrHär adds to the efforts already made by news organisations to improve the climate of the discussion in comments sections. Certainly, the results of #JagÄrHär’s efforts may not always be to the taste of the news organisations, yet it’s a practice which at some levels overlaps and relates to that of professional moderators.

The practice of outsourcing content moderation

If the practice of hashtagging, which #JagÄrHär activists engage in, is not a deliberate attempt to help the news organisations with their work, the practice of hiring the services of external and private companies to conduct content moderation in online forums provided by the news organizations is.

An example of one such private company is Interaktiv Säkerhet [meaning Interactive Security], which was founded in 2006, when social media platforms such as Facebook were growing rapidly (Medievärlden…, 2009).

The rules for how to conduct content moderation applying to companies such as Interaktiv Säkerhet generally differ slightly to those which apply to the news media organisations, and there are strict guidelines as to what such companies are allowed or not allowed to do.

We have a deal with Interaktiv Säkerhet. They are supposed to audit all the comments and any comment that breaks our rules should be removed within an hour. That is their responsibility, but they never go in and answer [a comment made by a user].

(Interview 4, March 2017).

We try to differentiate between legal and journalistic moderation. Legal moderation is BBS-legislative [Bulletin Board System]. That is the law. They [Interaktiv Säkerhet] are supposed to handle that. When it comes to journalistic moderation, which is everything from the posting of follow-up questions to the supporting of ideas […] The editorial teams should handle that themselves as far as it goes.

(Interview 5, March 1017)

Although many news outlets could see the benefits of hiring external services for content moderation in order to improve the discussion climate amid a growing level of user interaction, some of the news outlets participating in the study did not employ external companies to manage content moderation.

No, we don’t use that [content moderation by an external party]. We do our own [content moderation].

(Interview 6, August 2017)

Both inside and outside Scandinavia, there were those who reacted quite strongly against the idea of engaging a third party.

Oh, no! We have nothing like that! […] The notion that you would have a private company quietly going around killing public commentary. No way!

(Interview 10, December 2017)

Whether or not a news outlet had decided to engage the services of an external company for content moderation, they often still had to position themselves by making a decision on whether or not to engage in the practice of outsourcing.

We made a decision in 2011 not to outsource [the moderation of] our reader comments.

(Interview 8, August 2017)

Although making the decision not to engage in the practice of outsourcing, as the news outlet in the quote above had, they still had to relate to the fact that this practice is part of the same infrastructure and that it is a practice engaged in by many of their competitors on the market.

Engaging users to keep online discussions civil

So far, only organised practices for counteracting user misconduct have been considered. However, the cooperation and mostly spontaneous performance of the regular user are essential for a pleasant tone in the comments sections. As much as there are users who can be held accountable for the necessity of practices for managing online misconduct, there are also users whose actions have a very positive effect on overheated debates. Many interviewees testified to the calming and pacifying effect that the online practices of users posting comments can have on inflamed debates.

A large number of those who follow us are ordinary people who like […] the Eurovision Song Contest, and that means that our crowd isn’t much fun to hang out with [for those being disruptive] because you’re confronted.

(Interview 3, March 2017)

The average user has always been essential for the emergence and reinvention of practices for how to manage insults and threats in comments sections. In spite of having to put up with fellow debaters expressing themselves in offensive ways, the users keep on engaging in the practice of discussing and debating news online. In fact, as one interviewee remarked, when it comes to upholding a good tone of discussion in the comments sections, the users’ contribution is crucial.

Discussion

The interactive environment that, with the appearance of Web 2.0, emerged on news outlets’ own websites and later on social media platforms has grown into an environment in which new practices for managing offensive users have evolved and overlapped with other practices (Binns, 2012; Hille and Bakker, 2014).

Some of the practices which have evolved around the issue of counteracting an antagonistic online climate have seemingly little in common. For example, the practice of moderating comments sections by simply removing any perceived improper content is a very different way of acting than responding by hashtagging it. Organisations such as Interaktiv Säkerhet and #JagÄrHär do not rely on each other when they engage in their own practices, and they are driven and motivated by fundamentally different factors. Yet, as their practices meet and overlap in online comments sections, they are acting in relation to each other, and they were both created in reaction to the practice of users posting offensive comments in online comments sections.

Information infrastructures are constantly changing and changed. They are products of socio-technical actions (Henfridsson and Bygstad, 2013) and the more efficiently they work, the less perceptible they are to those embraced by them (Sundin and Carlsson, 2015). While keen observers might be aware of specific practices, they are often ignorant of the multi-layered structure of which their own actions are part. For example, while regular users who engage in discussions in online comments sections are likely to be aware of the interaction they are engaged in with the newspaper and any other users participating in the same discussion, they might be oblivious to some of the moderating practices which are or are not undertaken by the newspaper or any other actors (e.g., a private company whose services have been hired) participating in the interaction.

Although the appearance of new entities such as news organisations, online forum providers or non-profit organisations such as #JagÄrHär can clearly bring about changes at all levels within a network, this study has drawn attention to the advantages of shifting the focus from the practitioners to the practices they engage in. By framing the relationality between tools such as technical systems and human agents or organisations as the only breeding ground for practices and infrastructures, an overview of the powerful role that practices play in the formation of structures may be overlooked.

Although this study does not aim to contest the importance of established agents such as companies and organisations in the evolution of infrastructures, it challenges a perspective which has a tendency to look at them and the relations between them as the sole generators of change. Instead, it advocates a perspective which understands practices as an important source of change. The practice of hashtagging was not invented by the organisation #JagÄrHär, nor was it a practice which it had to take up and adjust to. Instead, the already publicly popular practice of hashtagging paved the way for the emergence of #JagÄrHär. Rather than that change coming about through a network of actors (Law, 1999), it is the result of a multi-layered web of practices which has the power to produce new practices.

The growing number of users engaging in the practice of discussing news online and posting comments in comments sections required new practices for managing offensive comments made by users. New entities such as #JagÄrHär and private companies such as Interaktiv Säkerhet emerged and engaged in practices such as hashtagging and content moderation.

Infrastructures are embedded into other structures and social arrangements (Star and Ruhleder, 1996), and, as social practices take shape, overlap and replace other practices, they form an infrastructure in which professional and everyday practices meet and become deeply embedded into a system which is taken for granted as the generally acknowledged way of how things are done.

Conclusion

The practices in which a variety of organisations and individuals engage when acting on online user misconduct are entangled and overlapping, and together they create an information infrastructure for managing online misconduct. By exploring the overlap of practices for counteracting the problems of user-generated misconduct, this study has underlined the key role of practices rather than and in addition to the practitioners as the basis of an information infrastructure that supports the work of managing perceived online misconduct.

As the intermingling of practices such as hashtagging and outsourcing of content moderation shows, infrastructures are often more complex than they appear at first glance. Information infrastructures are multi-layered and embrace a variety of entities, interests and agendas (Öbrand, 2015). Entities on different levels act in relation to each other, and together they form a heterogeneous and constantly evolving network (Hanseth and Monteiro, 1997). However, they are also, as this study has shown, constituted by manifold activities performed in the form of practices. Practices have the power to produce further practices (Schatzki, 2002) which are engaged in and taken up by people and organisations.

Acknowledgements

I would like to thank my supervisors Professor Isto Huvila and Associate Professor Ulrika Kjellman at Uppsala University and Associate Professor Jutta Haider at Lund University for their support and advice. I am also grateful to the anonymous reviewers of this paper for their helpful comments and suggestions.

About the author

Amalia Juneström is a PhD candidate at the department of archival studies, library & information studies and museum studies at Uppsala University, 75236 Uppsala, Sweden. In her thesis work, different aspects of the information processes which have occurred around the problems the news media face in terms of online user misconduct are explored. She can be contacted at amalia.junestrom@abm.uu.se.

How to cite this paper