Platform Configuration and Digital Materiality: How News Publishers Innovate Their Practices Amid Entanglements with the Evolving Technological Infrastructure of Platforms

ABSTRACT Technology companies and their platforms are important digital intermediaries for news publishers. Platforms as technological infrastructures have digital materiality and how publishers organise their innovations have become entangled with the digital materiality of platforms. This six-year longitudinal study (2015–2021) examines this entanglement by analysing how and why a news publisher innovated its practices vis-à-vis the changing digital materiality of Facebook and Instagram. Employing practice theory and the platform configuration framework, this study assesses the publisher’s configuration of its editorial activities in relation to shifts in the digital materiality of Meta’s platforms and the reflexivity of newsroom staff. Using a mix of qualitative methods, this study finds that the publisher both embraced and resisted the platforms’ algorithmic changes. While it enhanced video production and distribution to align with the platforms’ preference for audio-visual content, the publisher also developed new practices to counterbalance its reliance on Facebook for news distribution and digital advertising revenue, especially after Meta’s downranking of publishers’ content on News Feeds and the introduction of restrictions for Instant Articles. This study concludes that while publishers are circumscribed by the changes in the platforms' digital materiality, they strategically innovate specific practices to both leverage platforms and reclaim or maintain independence from them.


Introduction
Global technology companies and their platforms (e.g., Alphabet/Google, Bytedance/ TikTok, Meta/Facebook/Instagram, Twitter, etc.) have come to dominate the digital media environment with their search engines and social networking services (Nielsen and Ganter 2022;Poell, Nieborg, and Duffy 2022a).They have had a significant impact on daily life and work practices-and news organisations (hereafter referred to as "publishers") are no exception.Publishers have found themselves in a quagmire.On the one hand, they rely on these global technology companies, especially Alphabet and Meta, for digital news production, dissemination and monetisation (Nielsen and Ganter 2022;Steensen and Westlund 2021), and have been innovating their practices to capitalise on the functionalities and services of platforms.Publishers have been using platforms to carry out background research, source leads and generate story ideas (Belair-Gagnon 2015), and have incorporated platforms in their editorial decisions and audience engagement (Ekström, Ramsälv, and Westlund 2022;Hanusch 2017).Commercially, publishers rely on platforms for driving online traffic to their websites to generate digital advertising revenue and subscriptions (Meese and Hurcombe 2020;Myllylahti 2021), and turn to them for training and funding (Bell 2021).On the other hand, publishers have become frustrated by the loss of control over their data, audiences and revenues to platforms and the need to keep pace with their incessant policy and technological changes (Cornia et al. 2018;Nielsen and Ganter 2022;Sehl, Cornia, and Nielsen 2021).Consequently, publishers pursue innovations in their practices to strategically reduce their dependence on platforms and regain control of their journalism and business.Ultimately, the dominance of platforms over publishers has led to multiple asymmetries that largely favour platforms, leading to concerns about the ability of news organisations to remain editorially independent and financially autonomous while serving civic needs (e.g., Poell, Nieborg, and Duffy 2022a;Toff and Mathews 2021;Westlund, Myllylahti, and Chua forthcoming).This concern has prompted researchers to call for greater scrutiny of the interplay between publishers' practices and platforms.However, they also advocate against simplistic understandings of publishers' dependency on platforms (e.g., Poell, Nieborg, and Duffy 2022b;Westlund, Krumsvik, and Lewis 2021).For example, Poell, Nieborg, and Duffy (2022b) call for an examination of the "spaces of negotiation" between publishers and platforms, which refers to the "opportunities available to news organisation to determine how they produce, distribute, and monetise content vis-à-vis platforms" (2).Similarly, Zelizer (2019) asserts that studying how journalism innovates its practices with careful consideration of its relationship with technology is essential to analysing its raison d'être.This allows for the posing of vital questions about whether and to what extent publishers, as cultural producers, can independently create and disseminate information that shapes our perception of the world.
This study examines the interplay between the innovation of publishers' practices and platforms by considering platforms as technological infrastructures that consist of digital materiality.This study postulates that publishers' continuous innovations in their practices are entangled with the evolving digital materiality of platforms.Amid this entanglement, it is vitally important to gain nuanced insights into how platforms, through their digital materiality, exert influence over publishers' practices and how publishers negotiate their positions in relation to these platforms.
This study longitudinally examines how a Singaporean publisher configured its innovation of practices in relation to the evolving digital materiality of Facebook and Instagram between 2015 and 2021.Practice theory (Ahva 2017;2019) and platform configuration (Chua and Westlund 2022) are used in this study's theoretical approach and analytical framework, respectively.Practice theory conceptualises innovation as constituted by three interconnected elements: activities, materiality and reflexivity, while platform configuration sees innovation as involving the adding of something new into use, and/or the modifying or removing of something existing.In this study, practice theory and platform configuration guide the examination of how the publisher practised platform configuration by adding, modifying and/or removing certain editorial activities with respect to shifts in the digital materiality of Facebook and Instagram.Additionally, this study examines why the publisher practised platform configuration by analysing the newsroom staff's reflexivity towards their activities and the platforms' digital materiality.These concepts are further discussed in the section on theoretical approach and analytical framework.This study uses a triangulation of qualitative methods, including in-depth interviews, newsroom observation and the tracking of key changes introduced by the platforms.
This study provides valuable contributions to both the research and practice of journalism.From a research perspective, it offers three contributions.Firstly, it advances knowledge in a vital area of scholarship that examines the dynamics between publishers and platforms.Specifically, this study sheds light on how a newsroom innovated its practices in relation to changes in Meta's platforms.Secondly, this study uses a longitudinal approach that spans a six-year period, which provides an in-depth analysis of the configurations made by the publisher.This approach is essential for understanding the nuances of how platform changes affect publishers' creation and monetisation processes, and how publishers respond over time.To the best of my knowledge, there has yet been a longitudinal case study on a single publisher over six years that examines its engagement with the shifts in the digital materiality of platforms.Thirdly, this study adds to the body of research on the relationship between publishers and Facebook and Instagram, two important digital intermediaries for news (Newman et al. 2022).Furthermore, this study contributes to research on publishers' use of Instagram, which is a relatively underresearched platform in the journalistic field (Al-Rawi, Al-Musalli, and Fakida 2021;Hermida and Mellado 2020).Practically, this study highlights the opportunities and challenges that arise from the entanglement between publishers and the changes in the technological infrastructures of platforms.This knowledge enables publishers and journalists to better navigate the complex platform-dominated news landscape and formulate effective approaches.
This article consists of six main sections excluding the introduction.The second section discusses the relevant literature, while the third section presents this study's theoretical approach, analytical framework and research question.The fourth section describes the methods and materials, and the findings are presented in the fifth section.The final section offers a discussion and conclusion of this study's results.

Literature Review
This section consists of two parts that discuss the relevant literature in light of practice theory's three elements (Ahva 2017;2019).The first part discusses the overarching themes in journalism research regarding publisher's activities and reflexivity towards platforms.The second part explains this study's approach towards the digital materiality of platforms, drawing on platform studies literature.

Publishers' Activities and Reflexivity towards Platforms
Journalism scholarship that examines the interrelationship between publishers and platforms reveals two overarching themes that capture publishers' activities and reflexivity towards platforms.One theme highlights how publishers, often adopting a pro-platform attitude, focus their activities on building platform presence (Steensen and Westlund 2021) by leveraging on the positive value that platforms supposedly provide.Another theme underscores how publishers are engaging in strategic countermeasures to reduce or (re-)balance their dependence on platforms-activities that are geared towards platform counterbalancing (Chua and Westlund 2019).
With respect to Facebook and Instagram, studies on publishers' use of these platforms demonstrate their focus on "building platform presence".In terms of news production, studies indicate that publishers use Facebook for news monitoring, sourcing, generating story ideas and verifying information (Humayun and Ferrucci 2022).For example, Lischka (2021) found that Swiss and Finnish publishers monitor Facebook engagement metricsshares, likes and comments-to assess the popularity of user-generated content and produce stories, while Ferrucci and Wolfgang (2021) found that publishers continue to use Facebook comments for lead-sourcing despite concerns over the reliability of information on the platform.
Research suggests that publishers are also producing and distributing more videos and live-streaming content on Facebook and Instagram following Meta's decision to algorithmically privilege videos on users' news feeds.A study of 19 newsrooms in the US, the UK and Germany found that their investments in online news videos were shaped by their perception of the importance that platforms were giving to videos and the accompanying revenue potential they offered (Kalogeropoulos and Nielsen 2018).Other studies likewise have observed that publishers devote resources to video production to garner higher audience engagement metrics on platforms, which in turn increases their ranking and exposure, and facilitates traffic to their news sites (Saks and Walck 2022;Tandoc and Maitra 2018).Despite significant investments in videos, publishers have been found to be struggling to monetise their platform-related commitments and deeply understand their audiences' preferences as the data provided by Meta is "woefully inadequate" (Nielsen and Ganter 2022, 141).Poell, Nieborg, and Duffy (2022a) suggest that when platforms algorithmically prioritise videos over text, as Meta has done, publishers invariably become more platform-dependent as they rely on platform-provided resources to create and distribute audio-visual media.
Publishers also use Facebook and Instagram for audience engagement.Separate studies of Norwegian and US publishers both demonstrate how they construct Facebook posts to address audiences directly and interpret stories for them (Guo and Sun 2022;Hågvar 2019).Another study of international news broadcasters found that publishers curate content for Instagram based on aesthetic qualities and enhance them to encourage audience engagement (Al-Rawi, Al-Musalli, and Fakida 2021).Similarly, a multicountry study on publishers' Instagram practices found that publishers tried new ways to leverage the platform's new content features, especially the novel video and photo content formats (e.g., Instagram Stories, Reels and Live), video and photo editing tools, filters and effects, and functions that extend the lifespan of ephemeral content (Vázquez-Herrero, Direito-Rebollal, and López-García 2019).
However, recent research indicates that publishers are increasingly engaging in platform counterbalancing, especially with respect to Facebook.These studies highlight publishers' frustration towards Facebook's immense influence over their editorial and commercial activities, and the lack of returns on their platform commitments.Data from different studies of publishers in Australia, Europe and the US show that they had witnessed a fall in internet referral traffic and programmatic advertising income following Facebook's News Feed algorithm changes that deemphasised news content from publishers' pages (Bailo, Meese, and Hurcombe 2021;Cornia et al. 2018).As such, these newsrooms reduced their reliance on Facebook for audience traffic and data by diversifying their news distribution, including publishing news on other platforms and recommitting resources to innovating their proprietary products and services (Chua and Westlund 2019).
Studies also highlight the misalignment of objectives between publishers and platforms as a reason why publishers are engaging in activities to counterbalance their relationship with Facebook.Nielsen and Ganter (2022) cite Facebook's Instant Articles (IA)-a publishing feature within Facebook that natively hosts and distributes publishers' content and provides advertising revenue-as an example of this misalignment.They write that while IA, which Meta discontinued in April 2023 (Meta 2023), promised publishers several editorial and commercial incentives, Facebook's goal is to keep users in its digital ecosystem.This is at odds with publishers' aim to drive traffic to their websites where they have greater control over editorial experiences, data and revenue.Scholars also argue that publishers have been hesitant to adopt or have stopped using IA due to the platform's overly restrictive specifications and disappointing returns on investment (e.g., Meese and Hurcombe 2020;Myllylahti 2018).
Furthermore, publishers often organise their activities by considering the costs and benefits of using Meta's platforms.A study of 20 European public and commercial news organisations found that they capitalise on specific aspects of Facebook that enabled them to meet certain editorial or commercial objectives, despite the potential of becoming over-reliant on the platform (Sehl, Cornia, and Nielsen 2021).The authors conclude that publishers adapt their platform strategies by balancing platform risks and opportunities to meet short-and long-term goals.Other studies reveal that certain publishers selectively rely on Facebook for functions that they lack, such as Meta's artificial intelligence applications (Simon 2022) or audience data (Nielsen and Cherubini 2022).
Overall, the two overarching themes in the research findings reflect how publishers organise certain activities in relation to Facebook and Instagram, as well as capture their reflexivity towards the platforms.However, to go beyond the literature and elucidate a more nuanced understanding of the dynamics between publishers' practices and platforms, it is imperative to parse how their activities and reflexivity interrelate with the platforms' digital materiality, which is further discussed in the next section.

Platforms and Their Digital Materiality (Technological Infrastructures)
This article sheds light on how the digital materiality of platforms evolve over time by drawing on insights from platform studies.This approach corresponds with journalism scholars whose research has taken a similar path.Two examples are Nielsen and Ganter (2022), who build on platform studies to critically interrogate how publishers have responded to the infrastructures, logics and economics of platforms, and Riedl (2022), who argues that perspectives from platform studies can open a vista for media scholars to strengthen their investigations into the interplay between the materiality of platform algorithms and newsrooms' editorial decisions.
One key concept from platform studies describes platforms as technological infrastructures that host, organise and circulate content produced or shared by users, and "process data (content, traces, patterns of social relations) for customer service and profit" (Gillespie 2018, 254).Platform scholars also define platforms as "(re)programmable digital infrastructures that facilitate and shape personalised interactions" (Poell, Nieborg, and Van Dijck 2019, 3) among users.Importantly, this understanding of platforms foregrounds three characteristics of their digital materiality.
Firstly, seeing platforms as technological infrastructures acknowledges their digital materiality, which consists of features that are front-end (e.g., graphical user interface, terms and conditions, etc.) and back-end (e.g., application programming interface, algorithmic processing, etc.).Importantly, this digital materiality affords and constrains activities related to their use (Duffy, Poell, and Nieborg 2019), which Plantin et al. (2018) describe as the "dual nature of platforms" (297).In reviewing the platform studies scholarship, Plantin et al. (2018) emphasise that researchers have demonstrated how platforms and their affordances support participation, innovation and creativity, and yet simultaneously constrain them, often for the profit of the platforms' owners.
Secondly, platforms are mutable technological infrastructures, meaning that their digital materiality is centrally designed, controlled and consistently reprogrammed by their owners who exert considerable influence over how they are used (Chua 2023).Helmond, Nieborg, and Van der Vlist (2019) illustrate this point by tracing how Meta gained dominance in cultural industries by evolving as a (re)programmable technological infrastructure (see also "platform evolution" in Poell, Nieborg, and Duffy 2022b).The authors suggest that Meta progressively designed its software to encourage businesses to build applications that integrate with its platforms' functionalities and logics.The platforms' design, together with strategically forged corporate partnerships, allows Meta "to create institutional dependencies among its vast network of partners" (124).In other words, platform practices, including those of publishers, are technologically shaped by platform companies (Poell and Van Dijck 2014).
Thirdly, the infrastructural dimension also emphasises platforms as data infrastructures.This highlights not only the ability of platforms to gather and process usage patterns from substantial numbers of users in the form of analytics, but also their capacity to host, organise and distribute extensive amounts of information that are offered as commercial services to businesses.Poell, Nieborg, and Van Dijck (2019) argue that a small number of powerful platforms have capitalised on their capability to "endlessly recombine and reuse" immense amounts of data flows and transform them into "business models based on the trade of (mostly free) services for (mostly user-generated) data" (9).Platforms scholars also highlight how publishers are optimising their production and distribution, consumer/audience engagement and monetisation activities to suit the logics of the data infrastructures owned and controlled by platform companies (Poell, Nieborg, and Van Dijck 2019;Van Dijck and Poell 2013).Essentially, platforms exert control over publishers who become infrastructurally captured (Nechushtai 2018).
In sum, seeing platforms as technological infrastructures provides a conceptual lens for understanding their evolving digital materiality.This understanding, in turn, provides a path to examine how publishers configure their practices, specifically their activities and reflexivity, in relation to the digital materiality of platforms.Concomitantly, it sheds light on how platforms, through the control of their digital materiality, exert their influence by establishing conditions for how publishers may engage with them.

Theoretical Approach and Analytical Framework
Drawing on the findings of previous studies discussed in the preceding section, this research employs practice theory and the platform configuration framework to understand how and why a Singaporean publisher configured the innovation of its practices in relation to the evolving digital materiality of Facebook and Instagram.This section introduces the key concepts of practice theory and platform configuration, explains their relevance to this study and concludes with the research question.
Practice theory espouses that innovation entails three interconnected elements: (1) activities (or constellation of activities)-how actors behave, do things or move about; (2) materiality-objects, tools, technologies or geographies that are part of practices; and (3) reflexivity-symbolic meanings that actors give to their actions and the objects they use (Ahva 2017;2019).Drawing on this understanding of innovation, platform configuration suggests that publishers innovate their practices in relation to platforms by continuously configuring what they do (activities) in relation to specific platforms and their technological infrastructures (materiality), based on their ongoing individual and collective assessments (reflexivity) of their resources and intentions for achieving certain organisational goals (Chua and Westlund 2022, 67).
Activities refer to what publishers do when they configure their practices in relation to platforms.This study pays attention to the publisher's editorial activities (news production, distribution and promotion), while being cognisant that these often intersect with technology and commercial activities (see discussion in Chua and Westlund 2022;Lewis and Westlund 2015).Platform configuration posits that publishers configure these activities in relation to platforms via three processes: (1) adding, (2) modifying and (3) removing.Adding refers to instances when publishers add new activities and is connected to "building platform presence" (Steensen and Westlund 2021).Modifying refers to instances when publishers alter existing activities and entails either an increase or a decrease in activities.An increase is linked to "building platform presence", while a decrease is geared towards "platform counterbalancing" (Chua and Westlund 2019).Removing specifies how publishers discontinue existing activities and is also linked to platform counterbalancing.These processes provide a basis for analysing how the publisher in this study continually configured its activities in relation to the changes in the digital materiality of Facebook and Instagram.
Digital materiality pertains to platforms as technological infrastructures that enable or constrain publishers' activities (see Djerf-Pierre, Ghersetti, and Hedman 2016; Graves and Anderson 2020).For example, Facebook and Instagram Stories allow video clips of 20 and 15 s, respectively, to be posted, but not long-form videos.Publishers who use Facebook and Instagram Stories for disseminating news videos, therefore, need to produce content that meets the specifications and logics of these platforms.Furthermore, the digital materiality of Meta's platforms is continuously reprogrammed (Poell, Nieborg, and Van Dijck 2019), which impacts how publishers organise their activities in response to the technological changes (Nielsen and Ganter 2022).This study examines how the publisher's configuration of activities interplays with the evolving digital materiality of Facebook and Instagram.
Reflexivity refers to the process by which actors collectively interpret their activities and the materiality of their tools (Ahva 2019).Actors' interpretations consist of feelings, perceptions, thoughts and motivations, and continuously evolve over time (Nicolini 2012).This study considers the reflexivity of the publisher's staff by paying attention to their interpretations of their activities in relation to the digital materiality of Facebook and Instagram.Importantly, analysing the staff's reflexivity provides insights into why the publisher configured activities in certain ways vis-à-vis both Meta platforms.
Guided by practice theory's three elements and the platform configuration framework, this study seeks to investigate how a publisher continually configured the innovation of its practices in relation to the evolving digital materiality of two Meta platforms between 2015 and 2021 by asking the following research question: How and why did the publisher practise platform configuration in relation to the evolving digital materiality of Facebook and Instagram over time?

Method and Material
This study adopts a case study approach, a method that is optimal for examining contemporary phenomena in their real-life context (Yin 2018).The Singaporean publisher in this study was chosen as a theoretically informed case, typifying the empirical phenomenon of interest.The publisher is a for-profit news organisation that positions itself as a national news provider that employs about 300 staff.It has been experiencing falling print advertising revenues and circulation since the early 2010s and had transformed itself from a print to a "digital-first" publisher that prioritises digital content and its online business.This included working with platforms to produce and distribute content, interact with readers and exploit monetisation potential.Notably, the publisher's operations and strategies were impacted by Meta's introduction of changes to Facebook and Instagram.

Fieldwork
The data for this study was collected between the end of 2015 and mid-2021 using a mixmethod qualitative approach.This included semi-structured, in-depth interviews with the publisher's staff, newsroom observation and the tracking of key changes in the platforms.
The interviews with the newsroom staff were conducted at four different points in time: from the end of 2015 to mid-2016 (dated as 2016), end of 2018 to early 2019 (dated as 2019), in early 2020 and in the first half of 2021.A total of 35 staff members were interviewed (Table 1) and some individuals were interviewed multiple times.They comprised of editorial staff and non-editorial staff whose work directly engaged with platforms to provide multiple viewpoints on the publisher's innovation of practices.The editorial staff entailed both digital and print journalists and editors, while the non-editorial staff were technologists, business managers and those who work at the intersections of commercial and technological functions, such as product managers.All interviews were recorded and transcribed using digital transcription software.This study's author listened to the full recordings and repeatedly verified them against the transcripts to ensure accuracy.
The interview data is complemented with 20 h of non-participant newsroom observation conducted over consecutive days in early 2020 to ascertain how the staff interacted with the digital materiality of Facebook and Instagram.Observations included six editorial meetings, a meeting between digital journalists and technologists, and observing two digital journalists and a social media editor at their desks.While observing the staff at their workstations, this author asked questions based on his observations, which provided insights into the staffs' activities and reflections in relation to the digital materiality of the platforms.Each observation session lasted between 60 and 90 min and was done at different times of the day.Field notes of these observations were also taken.Twenty hours of observation may be a relatively small number for a newsroom study but are essential for providing first-hand insights into the practices of the newsroom staff and triangulating the interview data.Furthermore, this study's author conscientiously questioned the interviewees to establish if the observations made were indeed routine occurrences.
All interviews and observations were conducted in-person except in 2021, when interviews were done via video conferencing due to Covid-19 restrictions.Interview and observation guides were prepared prior to the fieldwork and all data was obtained under conditions of anonymity with the names of the publication and interviewees omitted to respect the privacy of those being studied.Interviewee numbers are used to present Another aspect of the data collection that merits elaboration is the tracking of the evolving digital materiality of Facebook and Instagram.To identify key changes in the platforms' digital materiality, this author referred to a trio of secondary data sources: academic literature, industry reports and Meta's websites.This was periodically documented in an Excel spreadsheet and digital note-taking software Evernote.The sources that were used in the tracking of the changes of both platforms are discussed and cited at the beginning of each section in the findings.Overall, this process allowed for a record of changes in the two Meta platforms that impacted publishers' activities globally.Interviewees were then asked questions about the publisher's activities with respect to the platforms' changes, and the data-both observations and interviews-was analysed with the changes in the platforms' digital materiality.

Data Analysis
A constant comparative approach, where data is assessed as it is being collected to iteratively make comparisons during every stage of analysis (Glaser and Strauss 1967), was adopted.The data was analysed in four stages.In the first stage, the data that was collected from 2015 to 2021 was combined into one dataset.This study's author read the transcripts and field notes multiple times and coded the data by identifying instances when the publisher added, removed and/or modified specific activities.Stage two entailed searching for emergent patterns over time to ascertain how the publisher's configuration of activities interrelated with Facebook's and Instagram's key changes.Stages three and four comprised an analysis of the staff's reflexivity towards the publisher's configuration of activities vis-á-vis the platforms' alterations.This was followed by a search for patterns of change or stasis in their reflexivity over time.

Findings
This section presents three key findings that address this study's research question (RQ): How and why the publisher practised platform configuration in relation to the evolving digital materiality of Facebook and Instagram over time?The following subsections respond to the RQ by presenting three salient findings that have been selected from a rich tapestry of results, each as an example of the publisher's practice of platform configuration.Each section begins by describing specific changes in the platforms' digital materiality relevant to each finding.Thereafter, it explains how the publisher configured its activities with respect to the platforms' digital materiality by adding and modifying (increase), modifying (decrease) and/or removing specific activities.This is followed by an analysis of the staff's reflexivity, which sheds light on why it had configured practices in certain ways.

Adding and Modifying (Increase): Embracing Platform Videos
Since 2014, Facebook and Instagram have been prioritising videos.Both platforms have been algorithmically elevating video content on users' news feeds and improving video functionality, such as by introducing new video formats, increasing capacity for hosting higher-resolution content and boosting video loading speeds (Bhattacharjee 2019;Meta 2022).Between 2014 and 2016, both platforms introduced live-streaming videos and elevated their visibility in users' feeds over other types of content (Meese and Hurcombe 2020;Nielsen and Ganter 2022).The platforms also added Stories-video clips that disappear after 24 h.In 2018, Instagram launched IGTV, a vertical format for videos of up to 60 min designed for mobile phones (Systrom 2018).In 2020, it introduced Reels: 90-second looping videos (Instagram 2020).
The publisher configured its activities in relation to these changes by progressively adding activities related to the production and distribution of news videos for Facebook and Instagram.First, the publisher gradually channelled more resources to produce news videos on both platforms.It started a three-staff video unit in 2014, which increased to six personnel by 2016.A digital editor (S2, 2016) said that one of the objectives of this unit was to enable the publisher to "flood" Facebook with video content.The publisher also trained and tasked more journalists to shoot videos using mobile devices for social media.In 2016, the publisher dedicated staff to use platform-provided analytics tools (Crowdtangle, Instagram Analytics and Reel Insights) to closely monitor user-generated videos that were gaining popularity on Meta's platforms.These videos were used for story ideation and were selectively included in the publisher's video content that was also distributed on the platforms.In 2019, the publisher added a TV studio to its newsroom to produce talk shows that were livestreamed on Facebook.
Second, the publisher produced videos in various formats to suit the specifications and logics of both platforms.In 2016, the publisher started producing videos for Instagram Stories, the only feature on Instagram where hyperlinks to articles could be included.Then, only Instagram accounts with more than 10,000 followers were offered this feature.The publisher had met this criterion and leveraged it to direct readers from Instagram to its website (social media editor, S2, 2018).These Stories were also published on Facebook.The publisher also started using Highlights, a digital archive on Instagram, to extend the lifespan of its Stories beyond 24 h.In 2019, reporters trained in broadcast journalism were tasked with producing live videos for Facebook and Instagram.In 2020, the publisher began producing at least one two-to three-minute clip per day to promote its articles on Facebook.It also edited videos to suit IGTV, where longer videos appear in users' feeds as one-minute highlights.A title and summary that described the IGTV video were placed within the first 10 s of each clip to capture viewers' attention (social media editor, S35, 2021).
Third, the publisher was more purposive in producing videos for Facebook and Instagram.Different types of videos and blurbs were prepared for each platform.This purposiveness was observed during a meeting in 2020 when the editors extensively discussed how parts of a multimedia story should be edited differently for each platform.The publisher was also more discerning when distributing long-form videos (usually 15 min or more) on Facebook.A social media editor (S31, 2021) said analytics data indicated that short videos gained better engagement on Facebook and generated more clickthrough to the publisher's website than long videos.Between 2018 and 2021, the publisher produced and distributed longer videos via IGTV and Reels.In 2019, the publisher started using Instagram's opinion poll and quiz features in Stories and Reels to encourage interaction between audiences and its content.This shift towards being more purposive in its video strategy contrasts with 2016, when the publisher disseminated similar videos across multiple platforms.
With respect to reflexivity, the staff acknowledged the benefits of creating videos to leverage the platforms' algorithmic preference for video content, but also expressed caution over allowing the platforms' algorithms to increasingly dictate news decisions.Regarding the benefits, according to a social media editor (S2), increasing video production enabled the publisher to meet audiences' growing demand for video content on the platforms, while a journalist (S15, 2019) felt that producing more videos for Instagram was necessary to "be in tune with younger readers" as "that is where they are, and what they want".A digital editor (S9, 2020) also stressed that social media videos were generating "good engagement and clickthrough, which is great for page views".Other journalists perceived the publisher's efforts to produce more live-streamed videos for the platforms as opportunities to gain recognition for their work from the public and their co-workers (S2, 2020).However, the staff also expressed exasperation with the constant changes in the platforms' video algorithms.A journalist (S27, 2020) said, "Every two to three years, the editors will tell us that the (video) algorithms have changed, and now, we need to do videos differently", before adding that it felt like the editors were unsure of what they were doing.Another editor (S25, 2020) cautioned against becoming overly reliant on the platforms' algorithms and emphasised the need to maintain a diverse content strategy that was not solely dependent on producing videos for the platforms.
Ultimately, the publisher's platform configuration saw it gradually add and modify (increase) video production and distribution activities to capitalise on Facebook's and Instagram's algorithmic preference for videos.These activities were oriented towards "building platform presence" on both platforms, despite the staff's recognition of their encroaching influence on the publisher's news work.

Modifying (Decrease): Reducing Distribution on Facebook
In tandem with the platforms' initiative to prioritise videos, Facebook introduced several algorithmic changes to users' News Feeds that impacted the discoverability of news articles.In 2014, Facebook implemented an algorithm to downrank stories that users clicked on but did not read with the aim of reducing clickbait articles on the platform (El-Arini and Tang 2014).This change increased the visibility of major publishers' content on Facebook at the cost of less prominent ones (Nielsen and Ganter 2022).Between 2015 and 2018, the platform implemented several changes to prioritise content from friends and family, and less content from "branded pages" (Eulenstein and Scissors 2015), including publishers.A notable shift came in 2018 when the platform introduced its "meaningful social interaction (MSI)" algorithm that severely limited the amount of exposure for content from publishers and businesses (Caplan and boyd 2018;Meese and Hurcombe 2020).
Over time, the publisher configured its activities to reduce its dependence on Facebook in light of the platform's algorithmic changes, especially post-2018 when the publisher's Facebook referral traffic decreased by almost half after Facebook implemented the MSI algorithm, according to a social media editor (S2, 2019).One of the publisher's activities was to become more selective in the amount and type of content it distributed on Facebook.In 2016, the publisher's approach was to post almost all its content on Facebook, averaging six posts per hour.By contrast, in 2019, the publisher lowered the post rate to an average of three an hour, having become cognisant that indiscriminately posting content would "harm its algorithm" (social media editor, S35, 2021) on users' Facebook feeds.Furthermore, the publisher was more discriminating about the type of content it distributed on Facebook, choosing to prioritise video and other types of content that had a higher chance of performing well based on audience data (digital editor, S19, 2020).
Another activity involved the publisher seeking alternative news distribution channels besides Facebook.Since 2018, the publisher had been developing its proprietary news dissemination channels, which included its website, microsites, apps and email newsletters.It also innovated activities when distributing news via other non-proprietary channels, such as improving its Search Engine Optimisation (SEO) strategy.According to a digital editor (S8, 2020), the publisher focused more on SEO to "get more traffic from Google … which helped to mitigate the fallout from Facebook referrals".The publisher also distributed news via Telegram and WhatsApp groups (the latter ceased in 2019 when Meta discontinued bulk messaging), private Facebook groups where content is not affected by Facebook's News Feed algorithms, and news aggregators such as Google and Apple News.
In this instance, the reflexivity of the staff shifted from enthusiasm about using Facebook to garner visibility for the publisher's brand and content to frustration over Facebook's algorithmic changes that decreased the prominence of its content on the platform.In 2016, the staff had enthused about using Facebook for distributing news.A digital editor (S8) described Facebook as playing a "significant" role in "increasing the reach" of the publication's content, while a social media editor (S2) said the platform helped the publisher "grow and maintain [its] share of voice as a news provider in the online space".By contrast, in 2019, the same two staff expressed scepticism towards Facebook, mirroring concerns that other newsroom staff had about being exposed to Facebook's changes and becoming over-reliant on the platform for news distribution and referral traffic.The social media editor (S2), who had described Facebook positively in 2016, now stressed that the sudden drop in Facebook referral traffic as a result of the MSI algorithm made the publisher aware of how "unreliable" Facebook was and the importance of "growing readers and income streams outside" the platform, while the digital editor (S8) commented that "[the platforms] say they want to help us grow our audiences, but it feels more like we're being held hostage".
Overall, the publisher's platform configuration in relation to Facebook's algorithmic changes was marked by a modification (decrease) of activities related to using the platform for distributing news.The publisher's activities and the reflexivity of its staff indicate their aim to reduce reliance on Facebook, limit exposure to the platform's algorithmic changes and gain greater control over its news dissemination-approaches that lean towards platform counterbalancing.

Removing: Discontinuing Instant Articles
In 2015, Facebook introduced IA.Facebook incentivised publishers to use IA by promoting its offer of better reach, engagement and revenue (Brown 2018), and assumed that publishers would adopt IA and keep their content within Facebook in exchange for these incentives (Meese and Hurcombe 2020).Facebook claimed that IA was optimised for mobile news consumption as it loaded articles, including multimedia and interactive content, ten times faster than standard mobile web pages (Meta n.d.).IA also allowed publishers to earn advertising income either through direct ad sales or a 70-30 revenue-split between publishers and Facebook for ads sold via the platform company's advertising network.Later, Facebook progressively introduced changes to IA, allowing publishers to integrate the platform with their content management systems, control the hyperlinks at the bottom of each article and set up subscription paywalls within IA.Facebook, however, dictated the rules that governed the type of content, advertisements and hyperlinks that could be included in IA articles (Nielsen and Ganter 2022;Van Dijck, Poell, and De Waal 2018).
In relation to using IA, the publisher configured its activities by initially adopting the format to distribute news but eventually discontinued its use.In 2016, the publisher was one of the first newsrooms in Asia approached by Facebook to adopt IA when it was launched in the region (digital editor, S8, 2016).Then, the publisher was distributing an average of five stories via IA daily.A social media editor (S1, 2016) explained that the publisher was required to format its content to meet IA's technical specification.As it was a time-consuming activity, the publisher assigned one person to do this work.Things later changed, with the publisher first reducing the number of IA articles to one or two a day in 2019 because it "did not want to devote too much resources there" (S8, digital editor, 2020), before stopping its use completely in 2020.
With respect to reflexivity, the staff had initially perceived the publisher's use of IA to be beneficial, but eventually expressed disappointment with it.In 2016, a digital editor (S3) lauded IA's fast-loading speeds and opined that it could improve readers' experience of the publisher's multimedia content on Facebook, while a social media editor (S1) labelled the publisher's use of IA as an "experiment" that showcased how it "was keeping up with the times".Over time, however, the staff highlighted several concerns about IA.Firstly, it provided limited advertising revenue and subscriptions.According to a commercial manager (S17, 2020), IA fell short in providing advertising revenue as Facebook restricted the number and placement of ads in each article.Secondly, a digital editor (S9, 2020) said the publisher could not control readers' experiences of their content and motivate them to subscribe because IA kept audiences within Facebook instead of driving them to the publisher's website.Thirdly, Facebook offered inadequate audience data.Both an audience engagement editor (S22, 2020) and a digital editor (S8, 2020) contend that Facebook was not forthcoming in sharing data about readers who consumed the publisher's content via IA.Fourth, the resources required to use the platform outweighed the rewards.A social media editor (S32, 2020) described the process of having to format content to meet IA's technical requirements as "too cumbersome", while a senior editor (S21, 2020) opined that the amount of work required did not match the commercial returns.
Ultimately, the publisher configured its activities in relation to the use of IA by initially adding, then modifying (decrease), and eventually, removing.The findings suggest that the publisher was initially positive about experimenting with IA, but its experience of losing control over monetisation, advertising and audience data made it circumspect in committing resources to the platform.Over time, the publisher changed its approach from using IA to build a platform presence on Facebook to platform counterbalancing when it discontinued its use.

Discussion and Conclusion
Guided by practice theory (Ahva 2017;2019) and the platform configuration framework (Chua and Westlund 2022), this six-year longitudinal study aimed to examine how and why a Singaporean publisher practised platform configuration in relation to the evolving digital materiality of Facebook and Instagram.In light of this study's aim, the following discusses two overarching conclusions singled out from the analysis of the findings.
One key conclusion is that the publisher was increasingly conforming to the shifts in the digital materiality of Facebook and Instagram that favoured audio-visual content.In terms of how, this study's findings indicate that the publisher had progressively innovated its practices to enhance video production and distribution on both platforms.Furthermore, even when the publisher modified activities to decrease the amount of content it distributed on Facebook after the platform implemented the MSI algorithm, it selectively chose to prioritise video production and promotion on the platform.In terms of why, despite an awareness of the potential pitfalls of becoming over-reliant on the platforms' video algorithms in making editorial decisions, the publishers' staff were pragmatic in recognising the value of embracing the platforms' shifts and adjusted their activities to "build platform presence" (Steensen and Westlund 2021) on both Facebook and Instagram.The staff's activities were based on their reflexivity that the platforms would reward them with improved audience engagement metrics, higher ranking and exposure of its video content, and increased traffic to the publisher's website.Ultimately, this case study has shown that the publisher had over time transformed its editorial operations from being text-and photo-based to one that emphasised video production for platforms.Especially for print-based publishers, this transformation requires significant levels of adjustments and investments in new tools, skills, personnel and narrative approaches (Tandoc and Maitra 2018).In relation to publishers as cultural producers, this demonstrates the immense infrastructural influence of platforms to not only shape the type of content they produce (in this case, videos that are tailored to suit the formats of Facebook and Instagram, such as Stories, Reels and IGTV), but also how they innovate their practices and organise their resources (Zelizer 2019).This study's findings resonate with research that underscores the profound loss of independence and the pressure faced by publishers to comply with the audio-visual algorithmic logic of platforms (Al-Rawi, Al-Musalli, and Fakida 2021; Kalogeropoulos and Nielsen 2018;Nielsen and Ganter 2018;2022;Poell, Nieborg, and Duffy 2022a;Tandoc and Maitra 2018).Moreover, this "turn towards videos" plays into ongoing developments among publishers globally who, in order to maintain their relevance on platforms such as Snap-Chat, TikTok, Twitter, Weibo and WeChat, have been adapting their content and publishing more videos to suit the platforms' preferences (e.g., Nielsen and Cherubini 2022;Vázquez-Herrero, Direito-Rebollal, and López-García 2019;2022;Zhang 2019).
Another conclusion of this study is that despite being circumscribed by the changes in Facebook's and Instagram's digital materiality, the publisher was able to innovate its practices to regain and/or maintain autonomy from the platforms.In relation to how, the publisher engaged in the innovation of specific practices aimed at platform counterbalancing (Chua and Westlund 2019;2022), such as decreasing the amount of content and becoming more purposive about the types of content it published on both platforms and discontinuing the use of IA.The publisher also innovated its own proprietary capabilities, such as its news distribution and communication with audiences, and developing its in-house analytics and data infrastructures.Pertaining to why, an analysis of the staff's reflexivity reveals that although the publisher had on the one hand embraced the logics of Facebook and Instagram, it was on the other hand gradually implementing strategic countermeasures to reclaim and/or preserve control of its audiences, data and revenues from the platforms.In light of three elements of practice theory (activities, digital materiality and reflexivity) and platform configuration, this study's findings suggest that the publisher's alteration of activities was not merely a knee-jerk response to changes in the digital materiality of the platforms but based on the ongoing reflexivity of the publisher's staff.The publisher's simultaneous embracement of and resistance towards platforms sheds light on its ability to selectively pursue platform-related innovations that are aligned with its strategic priorities, and/or counterbalance against the platforms' influence.These findings harmonise with scholars who argue against perceiving publishers as passive players in a platformdominated digital media environment (Westlund, Krumsvik, and Lewis 2021), highlighting how news organisations' agency affects their responses towards platforms (e.g., Sehl, Cornia, and Nielsen 2021;Walters 2021).Appositely, Poell, Nieborg, and Duffy (2022b) emphasises the importance of avoiding one-sided interpretations of platform dominance and scrutinise the "spaces of negotiation" to evaluate the "continuous strategic manoeuvring" (2) by both publishers and platforms.Nonetheless, it is vital to clarify that publishers are by no means on a level playing field with platforms and the asymmetric relationship remains skewed in favour of the latter.
One limitation of this study is that although it employs a mixed-method qualitative approach, this study prioritises interviews.Nevertheless, the rich findings yielded in this study were triangulated using both the interviews and the data collected from the 20 h of newsroom observation.Furthermore, knowledge from this case study contributes to broader debates surrounding the evolving dynamics between publishers and platforms.It paves the way for future longitudinal and cross-sectional research to compare how different types of newsrooms (e.g., legacy media organisations, digital news-startups, etc.) across different countries, market contexts and business models (e.g., for-profit, public service, etc.) practice platform configuration in relation to various platforms and changes in their digital materiality.