Esports Research: A Literature Review

Business

The business literature dates the birth of esports back to the rise of the competitive scene of early 1980’s arcades (Borowy & Jin, 2013). Its growth is attributed to the value of the experience economy for consumers, the popularity of video games, the social recognition of video game players, and advances in technology (Borowy & Jin, 2013; Seo, 2013). The identification of these factors has helped in exploring motivations for esports consumption, understanding the networks and organizations surrounding the players, and designing effective marketing techniques (Hamari & Sjöblom, 2016; Lee & Schoenstedt, 2011; Seo & Jang, 2014; Weiss, 2011). This research is most often done in a naturalistic setting using surveys, interviews, and case studies. For example, Hamari and Sjöblom (2016) apply the Motivation Scale for Sports Consumption (MSSC) to measure motivation for esports consumption. MSSC has been used to measure motivation for traditional media and sports generally, so their results in a new context can be compared to previous work in more widely studied areas (Hamari & Sjöblom, 2016).

Esports has evolved into a complex ecosystem of consumers, players, organizations, and other stakeholders, where players and consumers are the most common subjects of study for business researchers. Seo and Jung (2014) conceptualize esports consumption as an “assemblage of consumption practices, where consumers actualise and sustain the eSports phenomenon through their engagement with the interconnected nexuses of playing, watching and governing of eSports” (p. 637).

The convergence of Western and Asian esports cultures (Seo, 2013) has led to a focus on the internationality of esports. Parshakov and Zavertiaeva (2015), for example, examine the relationship between a country’s tradition of playing esports, its country-level characteristics (e.g., Hofstede’s cultural dimensions), and the performance of players from that country by using tournament prize data. Other studies, however, have narrowed their research by analyzing both players and consumers in specific countries. The literature has expanded over the years from China being the only region analyzed (Szablewicz, 2011; Zang et al., 2007) to including research about communities in regions like South America (Menasce, 2017) and Europe (Lokhman, Karashchuk, & Kornilova, 2018; Stein & Scholz, 2016).

Sports Science

A group of sports science researchers interested in the implications of competitive video gaming are categorizing esports within the frame of traditional sports. Most publications from sports science are agenda setting—by using the standard of traditional sports, they are evaluating the potential of esports to be considered sports. Early discussion around cybersport (Hemphill, 2005) defined characteristics by which competitive computer games can be considered sports, namely how the immersion and interactivity of computer games can emulate and require skilled physicality. As work reconciling esports with traditional sports continued, Jonasson and Thiborg (2010) inserted esports in Guttmann’s (2004) model of modern sports (Figure 2). This discussion continues in sports science, as illustrated by Hallmann and Giel’s (2018) summary of previous work providing the following criteria for esports to be categorized as sports: physical activity, recreation, competitive elements, organizational structure, and social acceptance of esports.

OPEN IN VIEWER

Figure 2.

Guttmann’s (2004) model of modern sports.

Empirical studies of esports in sports science are mostly case studies utilizing qualitative methods. Rambusch, Jakobsson, and Pargman (2007), for example, conducted interviews with players at World Cyber Games (WCG) and discussed important elements shaping and influencing gameplay in Counter-Strike (Valve Corporation, 2000) on four analytical levels: (1) player actions during play, (2) interactions within and between teams, (3) players and fans on the Internet, and (4) the Counter-Strike gaming scene. These empirical studies tend to focus more on how players engage in esports competitions and less on whether participation in esports can be considered sporting.

Cognitive Science

Research in cognitive science and psychology has focused on player performance and cognitive and behavioral differences between novices and experts. Until recently, this work had relied on naturalistic observations to better understand the cognitive processes required for competitive play. The earliest research explored how competitive players understand the games they play and the contexts they play them in (Ash, 2012; Rambusch, 2011), and through these explorations, a trend developed of studying what sets the elite players apart. Huang, Yan, Cheung, Nagappan, and Zimmermann (2017), for example, collected data on habit formation in StarCraft II (Blizzard Entertainment, 2010) players. Expert StarCraft II (SC2) players, according to Huang et al. (2017), develop habits in consistent ways, yet those habits can be unique to individual players. In other words, the same methods for developing good in-game habits at a high skill level are used by many players, but which habits they develop differ by player.

Experimental work on cognition in esports is less common but growing, as noted in W. D. Gray’s (2017) call for action games to be “an experimental paradigm for Cognitive Science” as a context in which to explain complex human behavior. P. B. Gray, Vuong, Zava, and McHale’s (2018) experiment studying hormone levels during League of Legends (LoL; Riot Games, 2009) play exemplifies the foundation building required for competitive video games to become a central experimental paradigm for cognitive science. In the 26 subjects who played against human players, P. B. Gray et al. (2018) found no significant difference in testosterone, cortisol, dehydroepiandrosterone (DHEA), androstenedione, or aldosterone levels when compared to the control group of 17 subjects who played against the computer. In both groups, aldosterone levels decreased during play, and as games against other people lengthened, testosterone, DHEA, and androstenedione levels increased. In short, these are null results. The authors, however, identify the consistent cortisol and decreasing aldosterone levels as indicators that the context of their study—“an informal, familiar location playing against known competitors” (P. B. Gray et al., 2018)—facilitated less competitive, more relaxed play. Gray and colleagues suggest future work uses a more competitive venue, highlighting the nascency of esports research.

Informatics

Esports research in informatics collects from a wide variety of data sources including game telemetry and user-generated play data (El-Nasr, Drachen, & Canossa, 2013), physiological data (Nagel, 2017), and text mining (Olshefski, 2015) in combination with observations to analyze in-game performance, team dynamics and formation, and interactions between players. Esports’ technology-mediated nature gives researchers this ability to collect massive amounts of data at various levels of analysis. For example, Low-Kam, Raissi, Kaytoue, and Pei (2013) collected player inputs from replays of 90,678 professional SC2 matches and developed a machine learning algorithm to detect unexpected strategies in those data in the interest of informing models of player behavior.

Work that focuses on team performance tends to rely more heavily on mixed-methods approaches. S. J. Kim, Keegan, Park, and Oh (2016) demonstrate work that is representative of the team expertise research generally. They collected LoL game data from the Riot application programming interface (API) to access performance metrics, like kills, deaths, assists, wins, losses, and rankings, in addition to which champions each team selected in which order. The researchers also conducted interviews and focus groups with participants to provide context for their quantitative findings. Both the gameplay data and the participants’ responses supported the hypotheses that (1) team performance is correlated with how familiar team members are with their assigned role, (2) team performance is correlated with the extent to which team members’ roles complement each other, and (3) individual performance is correlated with a player’s likelihood of selecting a role that compliments their team over a role that they are more proficient with.

Some informatics work around team dynamics in esports focuses solely on social interactions between players, largely ignoring in-game performance. Freeman and Wohn (2017b) developed an interview study to understand how esports players give and receive social support among themselves. The study of team performance, approached from these varied perspectives, exemplifies of informatics’ multidisciplinary nature.

Law

Law in esports is rooted in concern for how concepts of copyright and intellectual property are applied to virtual worlds. Law papers are primarily analyses of how a certain moral concept or right is affected or applied in the esports space. By tracing existing cases and compiling tests, authors discuss how legal concepts may influence or shape the governance of esports. Writing on how the virtual–analog division in esports creates a new opportunity for shaping Internet law, Burk (2013) asserted that “copyright is likely to be the lynchpin in any dispute.” Other legal concepts such as the right to publicity or rights derived from necessary associations in the space, gambling law, or international accommodations for sports will all be relevant as well, but governance or legal thought associated with esports must address the ownership of the game and the debate of where creativity and ownership lie. These have manifested as cases debating who has the right to use player-created avatars, who owns mods to a game, who may broadcast tournaments of gameplay. Historically, ownership—and responsibility—of games and associated products have been granted to companies, with player contributions classified as adaptations or derivative works (Burk, 2013). While this leaves companies with some control over player-generated content, for instance, it also makes them at least partially responsible for the gambling, doping, and cheating that go on around esports. For this reason, classification of esports as sport or computer game is a frequent point of discussion in esports law because sports enjoy special accommodations in federal and international laws. As Holden, Kaburakis, and Rodenberg (2017) noted, the legitimacy convenience granted by classifying esports as sports would, similarly, come with oversight; there seems to be tacit agreement that esports is en route to classification and regulation similar to traditional sports, although formal litigation models have only been suggested, not established in any state at time of writing.

Media Studies

Research in media studies has focused on relationships between esports, sports, and media; the definition and delimitation of esports; the methodologies used to study esports; and the practice of live streaming gameplay. Themes such as the roles of physical exertion, spectatorship, historical precedence, and interaction are analyzed in an effort to classify esports. In these analyses, comparisons are often drawn between esports and chess, poker, and other traditional sports (e.g., Holt, 2016). The existence of esports across digital and physical spaces is possible, according to Hutchins (2008), because of media, communication, and information flow. T. L. Taylor (2012) also points out that “esports has encoded in its very nature a deep rooting in both technology and media” (T. L. Taylor, 2012).

Media studies researchers examine the esports community through the phenomenon of live streaming in the interest of exploring how the community is formed and how it interacts with streamers (e.g., Burroughs & Rama, 2015; Devia-Allen, 2017; Hamilton, Garretson, & Kerne, 2014; Kaytoue, Silva, Cerf, Meira, & Raïssi, 2012). Based primarily on qualitative data such as interviews, observations, and content analysis, these descriptive papers examine the experience of being in various roles in the esports ecosystem. Before 2012, research took place around events celebrated in physical spaces, like the WCG (e.g., Hutchins, 2008). For example, Cheung and Huang (2011) created a taxonomy of spectators and described spectatorship not as simply watching the game but as actively engaging with the community. The spectators described here are watching either from a physical venue or on dedicated video-sharing channels. Since then, technological advances, specifically live streaming and platforms like Twitch, generated a new media phenomenon which has been the primary focus of research for the last several years (Burroughs & Rama, 2015; Kaytoue et al., 2012). The merged space between the in-person and online worlds has become a focus of the field, as seen in T. L. Taylor’s (2012), Whalen’s (2013), and Burroughs and Rama’s (2015) works discussing the distinctions and lack thereof between what is virtual and what is real in the context of streaming.

Sociology

Much of the work in the sociology of esports has explored questions around live esports events and the interactions between audience and gameplay (Gommesen, 2012; T. L. Taylor & Witkowski, 2010). T. L. Taylor and Witkowski (2010) demonstrate the kind of generative exploration possible in a live esports context. In their own words, live events “give us an opportunity to explore the ways games can be both contained within larger cultural activities and yet can also cycle back and shape how people think about their leisure time and identity more generally.” In this vein of viewing games and esports as both a vehicle for and result of cultural change, they highlighted the growing numbers of women in competitive gaming spaces. Sociologists have focused more on gender and identity than on any other single esports research topic for its salience and social relevance to the esports community. The first of these papers on gender explored the presence of women in competitive video games (Bryce, & Rutter, 2002; N. Taylor, Jenson, & de Castell, 2009), but the field has recently adopted a focus on the perceptions of gender discourse in esports communities, as shown by J. Kim’s (2017) discussion of LoL players’ perceptions of differences between genders in video game competition and of solutions for narrowing the gender gap in esports.

Articles in this line of inquiry have approached gender by juxtaposing the roles of women and the portrayal of masculinity in esports. For instance, researchers have asked about “invisible women” in competitive gaming (Bryce & Rutter, 2002), what roles women tend to take on in esports communities (N. Taylor et al., 2009), and attitudes of professional players toward women, gender identity, and hypermasculine culture (Ruvalcaba, Shulze, Kim, Berzenski, & Otten, 2018; N. Taylor, 2009, 2011; Witkowski, 2013; Zolides, 2015). Although these questions lead researchers to conclude that gender inequality is currently a facet of esports (J. Kim, 2017), they also see esports as an opportunity to encourage a culture of diversity (N. Taylor et al., 2009).

Contested Definitions

Across disciplines, esports have been defined as competitive gaming, computer-mediated sport, or interactive spectatorship (Freeman & Guo, 2017a), with varying degrees of emphasis on physicality, computer mediation, institutional infrastructure, and spectatorship. Defining esports is a nontrivial debate that underlies scholars’ framing of their research.

Competitive gaming is a widely accepted description of esports. Regardless of whether esports are classified as sports, their status as video games rooted in principles of game design is not contested. Proponents of this definition of esports do not consider the broadcast popularity and level of institutional infrastructure surrounding a game title to be factors in classifying a game as an esport (Holt, 2016). Competitions can be held between amateurs and professionals, in a garage between friends or in a stadium between world-class teams. For esports titles that afford team play, scholars often focus on the formation and practices of teams (e.g., Freeman & Wohn, 2017c) and how technology mediates team communication and gaming. Zhang, Wu, and Li (2007), for example, define esports as “a sport of wisdom between people with hi-tech software and hardware as sports equipment….” More recently, Hamari and Sjöblom (2016) describe esports as “a form of sports where the primary aspects of the sport are facilitated by electronic systems; the input of players and teams as well as the output of the eSports system are mediated by human-computer interfaces” (p. 213). In these definitions, communication within a team or between competing teams is possible or enhanced because of the mediating technology.

Other definitions trivialize the role of technology, instead focusing on similarities with traditional sports. For example, one of the most cited definitions of esports is Wagner’s (2006): Esport is “an area of sport activities in which people develop and train mental or physical abilities in the use of information and communication technologies.” While Wagner (2006) himself equated esports to competitive computer gaming, in “On the Scientific Relevance of eSports” his definition of sports is inclusive of competitive gaming. In categorizing esports as sport, the physical motion of traditional sports is given less weight than the cultural significance and formal support of the activity. Support for this view of sports is growing as governing bodies emerge to guide developments in esports spaces and as “the development of a nascent professional infrastructure [includes] features familiar from the world of physical sports of entertainment, including tournaments, leagues, fans, teams, team owners, player contracts, sponsors, and the like” (T. L. Taylor, 2012; via Burk, 2013). Similarly, systems of prize and payment within a “professionalized context” (T. L. Taylor, 2012; via Brock, 2017) perpetuate the parallels between traditional sports and esports. These similarities make sports terminology useful for both players and researchers in talking about practices and issues in esports.

Comparisons to traditional sports are also a useful measure of the scope of spectatorship. This perspective on esports emphasizes the union of media and competitive gaming by focusing on the community and technology of spectatorship that surrounds esports (Kaytoue et al., 2012). The most precise definition is given by N. Taylor (2016):

E-sports involves the enactment of video games as spectator-driven sport, carried out through promotional activities; broadcasting infrastructures; the socioeconomic organization of teams, tournaments, and leagues; and the embodied performances of players themselves. (Borowy & Jin, 2013; Harper, 2013; T. L. Taylor, 2012; N. Taylor, 2016; Witkowski, 2012a)

In short, esports are often defined as games, as sports, or as mass entertainment. These definitions are not mutually exclusive but stem from different frameworks for understanding esports. They also each have unique practical implications, like leading to the regulation of esports under the same laws that govern traditional sports.

Nature of Expertise

Researchers in cognitive science, informatics, and sociology each study expertise in esports through different levels of analysis. Cognitive scientists are looking at the performance of individual players, who may or may not be nested in teams, and aim to identify constructs previously explored outside of esports that correlate with higher skill levels in competitive play. Huang and colleagues (2017) summarized in the esports within the disciplines: Cognitive Science subsection exemplify this approach relying on practice, routine, and habit formation literature from contexts as disparate as construction work and chess to understand expert SC2 players’ in-game behavior.

In informatics, more work has been published on expert teams than on individual performance. Informatics researchers often implement gameplay data analysis in search of optimal team compositions (Sapienza, Goyal, & Ferrara, 2018) and patterns in team play that correlate with in-game performance (Leavitt, Keegan, & Clark, 2016). Pobiedina, Neidhardt, Moreno, Grad-Gyenge, and Werthner (2013) and Nascimento, Melo, da Costa, and Marinho (2017), for example, both use the team-level data available from game publishers’ APIs to test hypotheses about what makes a strong team. They identify factors like role distribution, role selection, player experience, and teammate familiarity (Pobiedina, Neidhardt, Moreno, Grad-Gyenge, & Werthner, 2013) and statistics like ward placements and neutral minion kills (Nascimento et al., 2017) as factors that correlate with how often a team wins.

Sociologists look at broader social structures to examine what goes into expertise beyond in-game performance. Witkowski (2012a), in her doctoral dissertation, includes a section on what it takes for players to make it to professional teams. While honing in-game mechanics is fundamental to professional play, she emphasizes the necessity for players to develop social, business, and tech savvy. From interviews and observations of these elite players, the author details the out-of-game work and knowledge necessary to be considered an expert:

Players spoke of the self-promotional work outside of the game that aided in their upward mobility: Recording game footage and minimally overlaying it with tactical voiceovers then placed on YouTube, winning a duelling tournament organized by a prominent blogger, participating on Arena specific forums as a knowledgeable poster, making extravagantly produced gaming movies for Warcraftmovies.com (using a video editing program and adding music, effects, and text), or attaining a spot as a regular blogger on one of the major Arena community pages are just some examples of the extracurricular busy work that experts have engaged in during their rise to the top. (Witkowski, 2012a, p. 71)

While cognitive science focuses on what expertise is for a competitive player, informatics on what it is for a team, and sociology on what it means outside the game, these differing views of what constitutes expertise are not in conflict. To encompass what an expert is in an esports community, we must account for each of these perspectives.

The State of Nascent Research

For now, much of esports research is naturalistic observation of ecosystems without intervention. As the field develops, we expect more experimentation to add to this body of exploratory and descriptive work. Surveys are common, but for many, the representativeness of their samples is difficult to assess. They are often limited in their scope to specific regions, age groups, or games in an effort to produce knowledge useful in a business context.

An advantage of this wealth of exploratory knowledge is that it has already produced research questions for experimental designs. For instance, is the relationship Leavitt, Keegan, and Clark (2016) found between ping usage and player performance in LoL causal? What about the correlation they found between pings and deaths? Designing experiments that might help reconcile findings that we can only make inferences about will produce much work of interest to both the esports community and researchers’ respective disciplines.

The amount of work using API data also continues to grow. The wealth of data we can capture without intervention allows for naturalistic data collection on a scale that would not otherwise be possible. Reliance on these APIs, though, means that the metrics we use to predict performance, for example, must be narrowly defined and restricted by the data available from an API. Many APIs offer massive amounts of game data but do not supply a few key statistics, like player position. Researchers have thus far either restricted their models to data that are available through an API or created their own tools for gathering more data locally, restricting their sample size to the number of players they can bring into a lab. Game publishers’ fears of releasing too much data are well founded. Many have had issues with cheaters using game data to create tools that give them an unfair advantage. Finding solutions for industry to enable researchers to access richer data sets, though, would permit much less restricted models of player and team performance.