Algorithms and Gatekeeping in the Media Coverage of Armed Conflicts: the Syrian Civil War Case

In the digital age, social media, and consequently the algorithms that rule them, have acquired an increasingly central role in disseminating news. As a result, the selection of news for an audience – i.e. the gatekeeping process – is no longer only in the hands of newsworkers but rather sees a number of new actors, including algorithms. In light of these developments, this research aims to investigate how journalists’ work might have changed regarding conflict news’ selection and prioritization. Particularly, the focus will be on the coverage of the Syrian civil war; a first reason for this choice is that it has been defined as “the first social media war” and thus represents a fitting case study to analyze the role played by social media algorithms in conflict news coverage. Secondly, it represents one of the most significant conflicts of the last two decades because of its duration (2011-2024), the peculiarity of its news coverage, and its geopolitical impact on the MENA region and beyond, which turned the Syrian conflict into a ‘world problem’ (Pantti 2016 in Cozma and Kozman 2018, pp. 188).

Evolution of gatekeeping theory

This study is intended to fit within the research framework of contemporary gatekeeping. Therefore, clarifications on the evolution of gatekeeping theory need to be made. Building on Lewin’s (1947) original gatekeeping metaphor and White’s (1950) application of the gatekeeping concept on the journalistic field, Shoemaker & Vos (2009, p.1) define gatekeeping as “the process of culling and crafting countless bits of information into the limited number of messages that reach people each day”. Taking this definition as a starting point, our research specifically focuses on the implications of the rise of algorithms on the gatekeeping role in the current digital era. In our research the notion of “gatekeeping role” is understood as defined by Vos (2016 in Idem 2020, p. 93), namely as “a normative role whereby certain actors in the information environment see it as their duty or responsibility to pass along some information and not other forms or kinds of information”.

However, in today’s networked world, gatekeeping has gone from lying in the hands of journalists and editors to being a “shared process of news diffusion, in which multiple gatekeepers, selection mechanisms and platforms interact” (Thorson and Wells 2016 in Wallace 2018). Underlining the interactive nature of online networks, Shaw (2012) identifies two gatekeeping mechanisms: centralized and decentralized. While the former maintains the traditional one-directional news flow, the latter “consists of numerous, microlevel interactions between individuals engaged in a particular collective endeavour” (Shaw 2012, p. 350). The concept of decentralized gatekeeping developed by Shaw (2012) is particularly relevant to gatekeeping research as it points to how users’ collective interactions contribute to the news selection processes on social media. Nevertheless, Wallace (2018) argues that, whereas much research has been carried out on individual social media users’ increasing role in journalism, few studies have focused on algorithms’ gatekeeping power.

Algorithms as gatekeepers

Just and Latzer (2016, p. 2) refer to algorithms as “problem-solving mechanisms” which perform an “automated assignment of relevance to certain selected pieces of information”, and which nowadays have a central role in the construction of social reality (Just and Latzer 2016 in Wallace 2018). Among the many different kinds of algorithms existing today, only those that perform news selection and dissemination can be considered “social media gatekeepers” for their audience (Vos 2015 in Wallace 2018). These algorithms are understood by Wallace (2018) as “agents in the information flow” and identified as one of the four types of digital gatekeepers in his typology. However, Wallace (2018, p. 281) states that, since algorithms’ sources and criteria are predefined, “there is a risk that unconventional information or previously unknown sources are under-represented or not presented at all”, potentially creating filter bubbles that lead to “ghettoizing citizens into bundles based on narrow preferences and predilections rather than drawing them into a community” (Tandoc and Thomas 2015, p. 247).

In addition, Jeon and Nasr Esfahani (2013) have demonstrated how some kinds of algorithmic gatekeepers influence the nature of online news outlets’ content with the purpose of maximizing their performances on their digital spaces. Likewise, studies by Vu (2014) and Tandoc (2017) demonstrate how metrics systems “can have an impact on news selection and editorial decision making, which can influence news diversity” (Cools et al. 2021). Based on these studies, Cools, Van Gorp and Opgenhaffenn (2021) investigate how algorithmic news recommenders (ANRs) are deployed in newsrooms and point to the emergence of new dynamics of decision-making and autonomy in the processes of gatekeeping, agenda-setting and newsgathering.

User generated content (UGC) and competing narratives in the Syrian conflict

Throughout its course, the Syrian conflict has received extensive media coverage, and just as much attention from the academic community of Journalism Studies. However, to date, the focus of scholarship has mainly been on the emergence of so-called citizen journalism and thus the role of citizens and activists in the coverage of the war due to the absence of international reporters inside the country. Although this research aims to investigate the role played by algorithmic selection rather than the one of individual social media users – accepting Wallace’s (2018) critique – studies such as those by Wall and El Zahed (2015) and Murrell (2018) on citizen journalism have nonetheless represented relevant literature both in providing insight into the journalistic context in Syria and in highlighting the increasingly crucial role of online networks in journalistic practice nowadays.

Furthermore, studies by Cozma and Kozman (2018), Zhang and Luther (2020) and Boyd-Barrett (2022) have provided comparative analyses of the competing narratives on the conflict by Western and non-Western news outlets, leading the authors to ask questions on possible disparities in the algorithmic selection.

RESEARCH QUESTIONS AND SUBQUESTIONS

This paper asks the following research question to examine journalists’ perspectives on how the rise of algorithms has changed conflict news selection processes on social media and thus the gatekeeping role traditionally played by journalists:

RQ1: How do journalists and editorial teams see their gatekeeping role in selecting and prioritizing news on the Syrian civil war on social media in light of the rise of algorithms?

Moreover, we have identified two subquestions:

RQ2: Which social media do journalists and editors consider most influential for their coverage of the Syrian war?

RQ3: Do algorithms select news differently for Western and non-Western outlets?

METHODOLOGY

For the purposes of this study, we have selected the following methods: content analysis and semi-structured interviews.

According to Riffe, Lacy and Fico (2005, p. 37), content analysis allows us to understand both the “manifest” and “latent” messages in order to gain a better grasp of communication practices and processes. This method is based on the systematic classification of content into categories defined by precise rules, namely the process of coding: it involves selecting representative samples of content, then training coders to apply the classification rules correctly, and finally measuring their consistency (Riffe et al. 2005). The data collected with this method serves to identify patterns or important relationships between content characteristics (Riffe et al. 2005).

For this research, we have chosen quantitative content analysis since our aim is to determine the existence and frequency of concepts, ideas, words or other elements in the publications selected by the algorithms on social media, rather than to focus on in-depth analysis of the discourse, as a qualitative content analysis would imply. This allows us to examine the dynamics of the gatekeeping role played by journalists involved in the coverage of the Syrian war, and to analyze how this role manifests itself in stories on social media, once filtered by algorithms. Using triangulation, we then offer this data to journalists as part of our second method: interviews.

Semi-structured interviews are a qualitative research method involving guided but flexible conversations with participants: they are organized around predefined themes or questions, while allowing for the exploration of unexpected topics that might emerge in the course of the discussion. This format balances the coherence between interviews with the adaptability needed to delve deeper into issues of perception (Berger 2013).

The addition of interviews to our research is part of a triangulation approach, aimed at enriching and contextualizing the data collected. By presenting the results to the main stakeholders – journalists and editors – we seek to gather their perceptions and potential changes in their gatekeeping role.

This research will analyze online news content of three international news outlets: Al Jazeera English (AJE), BBC and The New York Times (NYT). The selection of AJE is based on its pioneering role in the online dissemination of events during the Syrian conflict, marking a shift in digital war reporting (Seib 2013). Additionally, as a Quatar-based news outlet, it provides a non-Western perspective on the war, contributing to a broader analysis of media narratives about the Middle East (Seib 2008, Lynch 2005). The BBC was selected for its status as an international news broadcaster and its global reach in conflict coverage. With a weekly audience estimated at 468 million people (British Broadcasting Corporation 2023), it serves as a significant source for examining news selection dynamics in wartime contexts. Finally, The New York Times was chosen for its prominent role in Western media coverage of international conflicts, and particularly for its detailed reporting on the Syrian war, including investigative journalism and multimedia analyses (Aday et al. 2012).

This approach goes beyond the simple presentation of results: it links the empirical observations from a quantitative analysis to the concrete experiences and reflections of the professionals involved. By combining these different perspectives, our study will offer a more complete and nuanced understanding of the issue under analysis, while strengthening its robustness and relevance and giving “the clearest sense of what is occurring” (Pew Research Center 2014).

APPROACH AND EXECUTION

For this research, we adopted a dual methodological approach, combining quantitative content analysis and semi-structured interviews in order to obtain a comprehensive and nuanced view of the role of algorithms in journalists’ coverage of the war in Syria.

The analysis period runs from 2011 to 2015, covering the first key years of the Syrian conflict, in particular the consolidation of power by Bashar al-Assad and the major events that took place during this period. This temporal choice makes it possible to follow on the publications, links, articles and videos shared by the Facebook and Twitter official accounts of the three selected media (AJE, BBC and NYT).

To collect the data, we will use Twitter’s Academic Search API and Meta Content Library and IPA to extract publications about the war in Syria from the verified Twitter accounts of the three media under analysis. These two tools provide comprehensive access to public data from META and Twitter, allowing users to search and filter posts and explore multimedia content from the chosen outlets.

The keywords chosen to select the data will include: ‘war in Syria’, ‘Syrian conflict’, ‘civil war in Syria’, ‘Assad’, ‘Aleppo’, ‘Raqqa’. In order to guarantee the relevance and reliability of the data collected, we will first carry out a pilot phase. A small selection of publications will be analyzed to test and refine the coding system before rolling it out to the entire corpus. We will then apply a systematic random sampling method, making it possible to select representative samples of data for each year and for each of the media under study.

The content will be analyzed using a quantitative coding system. Analysis categories will include:

• Gatekeeping: this variable identifies who is/are the actor(s) involved in the selection of the content under analysis.
• Content framing: this variable looks for possible common discursive patterns and characteristics in the selected sample.
• Impact of algorithms on news prioritization: this variable identifies how algorithms potentially influence what kind of news stories are prioritized and disseminated.
• Audience engagement: this variable involves how an audience interacts with news content, actively and emotionally (e.g. likes, shares, comments).

Once the queries for both platforms are executed, the data will be saved and categorized by the media outlet, ensuring that each media’s dataset is easily distinguishable and sortable. The data will be segmented on the basis of time period (2011-2015) and media outlet. This will allow us to analyze trends across both time and media coverage, facilitating comparisons of how each outlet framed and delivered content during the Syrian conflict. Once the data is organized and segmented, it will be ready for analysis and comparisons with ATLAS.ti.

To complement the content analysis, semi-structured interviews will be conducted to gain qualitative insights onto the role of algorithms in journalistic work. We will select a total of twelve journalists and editors working for AJE, BBC and NYT and directly involved in the coverage of the Syrian war.

Before carrying out all the interviews, a pilot phase will be carried out with two participants in order to assess the clarity and relevance of the questions. The interview guide will be adjusted accordingly. The interviews will be recorded and transcribed in full to facilitate analysis.

Finally, we will triangulate the data by combining the results of the quantitative content analysis with the qualitative information from the interviews. This integration will make it possible to cross-reference the empirical observations with the perceptions of the professionals, thereby providing a more complete and nuanced understanding of the issue under study. In order to do this, we will use ATLAS.ti to conduct data analysis and management.

EXPECTED RESULTS

Based on previous studies, we expect algorithms to behave as gatekeepers and thus to actively influence the process of news selection. Through quantitative content analysis, we will be able to draw conclusions about the impact of algorithms in selecting and prioritizing information during the war in Syria. More specifically, we will identify any recurring patterns in the framing of content, the diversity of sources, and variations in media coverage between Western and non-Western media. This approach will also enable us to analyze differences in algorithmic selection mechanisms according to the types of content presented to different audiences. These findings will provide valuable empirical data to answer our research question (RQ1) concerning the evolution of gatekeeping in the context of conflict, under joint influence of journalists and algorithms. By carrying out interviews with editors and journalists from the three news outlets under study (AJE, BBC and NYT), we will be able to collect the perceptions, opinions and feelings towards algorithms of the people directly involved in the gatekeeping process for the coverage of the Syrian conflict. Since the interviewees’ personal working experiences are most likely different from one another – depending on factors such as their personal sensitivity, their role in the newsroom, the news outlet they work for, the country their newsroom is based in – we expect to be presented with many diverse perspectives on the matter.

In regard to our first subquestion (RQ2), we expect the interviewees to identify Twitter as the most influential social media for their coverage of the Syrian war since, especially in the time period under analysis, it has represented the main platform used by journalists for breaking news, updates, and direct communication with news sources. However, we do not exclude the possibility that interviewees indicate Facebook as an equally important social media, mostly from the perspective of audience engagement and dissemination of visual stories.

Lastly, we assume that the codebook and the interviews will show algorithmic biases in favor of Western outlets (BBC and NYT) over non-Western ones (AJE) (RQ3). The main reason for this argument is that Al Jazeera English tends to offer an alternative viewpoint to Western dominant narratives: since audience engagement metrics follow the preferences of the majority of social media users, they might prioritize Western narratives over non-Western ones despite similar newsworthiness.

In conclusion, this research aims to investigate how algorithms might have influenced journalists’ traditional gatekeeping role during the Syrian civil war, focusing on the period from 2011 to 2015. By analyzing social media content from Al Jazeera English, BBC and The New York Times, alongside carrying out qualitative interviews with journalists and editors, we expect to reveal recurring patterns in content framing, media diversity, and algorithmic biases, with a potential prioritization of Western perspectives over non-Western ones. However, the study faces limitations, including its reliance on a specific set of media outlets and a constrained time frame, which may limit broader generalizations. Additionally, focusing primarily on Twitter, now rebranded as X, could pose challenges in data collection and analysis due to platform changes. Despite these constraints, this study seeks to offer valuable contributions to understanding the evolving dynamics of gatekeeping in the digital era and the influence of algorithms on global conflict coverage.