Corsi,Fenwick and Gramsci: How bloggers and advanced analytics are changing the National Hockey League

Methodology

Similar to Norman et al. (2019), I have been a member of an online, fan-driven community for a number of years, beginning my own blogging career in 2008 as a graduate student. I remain an active, albeit limited participant in a small subset of blogs for a specific Canadian market. As Norman et al. (2019) describe in their series of auto-ethnographic blogger vignettes, I try to balance my own fandom with a critical perspective of the game and am highly sensitive to the privileges afforded to me as a white male Canadian employed by a university that has provided me with the access and resources needed to perform this research.

The methodology used is a mixture of content analysis, netnography, ethnography and virtual ethnography (Kozinets, 2010; Norman, 2014). I conducted 15 semi-structured interviews with people I have identified as primary innovators within the hockey analytics community – the ‘organic intellectuals’ – in order to learn more about their backgrounds, their processes and their expectations with regard to the future of analytics and their own contributions to that future. These interviews were conducted using the active interview style (Holstein and Gubrium, 1995), in which the respondents’ stories shape the direction of each individual interview, allowing their stories to flow freely and allowing the respondents more agency in the interview process. As each participant’s work is highly visible online, all but one participant waived their right to anonymity and allowed me to disclose either their real names or their regular online pseudonyms. The respondents were exclusively male, and varied in age and race. Nine respondents are Canadian and six are American, although two of the Canadians are now based in the USA. Although there are many women working on hockey analytics online, those I contacted for interviews respectfully declined to participate. Four participants were recruited with help from my pre-existing network of blogger colleagues and online connections, whereas the remaining 11 were recruited using a combination of snowball sampling and direct messaging on Twitter. Although I conducted 15 interviews, responses from only 11 participants are used in this article.

Norman’s (2014) methodological approach, virtual ethnography, was instructive to my own. This method takes the conventional ethnographic immersion in culture, participant observation and data collection and applies it in an online environment. However, unlike Norman’s (2014) study, I am not focused on a single fan community or blog but, rather, a network of bloggers, Twitter users and other content creators from a variety of fan bases throughout Canada and the USA. The analytics community, which is an extension of the hockey ‘blogosphere’ (Norman et al., 2019), is not organized in a traditional sense, but is rather a loosely connected set of individuals who post on blogs and other online forums and produce content that they share freely across blogs and on Twitter. The result is an ongoing conversation between analytics content creators, other Twitter users, and blog and forum commentators about the nature of analytics, how it applies to specific players and teams and why it matters, despite the fact that the creators and commentators are often from different locations and have a vested interest in the success of different teams. The interview data will fill in existing gaps in knowledge and provide context and thick description (Geertz, 1973), in addition to a richness of the data itself.

Theory

For Gramsci, the concept of intellectuals, the knowledge they can produce and the relationship between intellectuals and knowledge is critical; indeed, his focus is on the entire process of knowledge production (Crehan, 2016). As Gramsci states:

Every social group, coming into existence on the original terrain of an essential function in the world of economic production, creates together with itself, organically, one or more strata of intellectuals which give it homogeneity and an awareness of its own function not only in the economic but also in the social and political fields (Gramsci, 1971).

Crehan (2016) suggests that even the term ‘intellectual’ should be defined very broadly and that in order to understand the distinction Gramsci makes between traditional and organic intellectuals, the ‘organic’ element must be emphasized more than the ‘intellectual’ side. She posits that organic intellectuals, when compared with traditional types such as doctors, lawyers and members of the established hegemonic structure, are not even necessarily a ‘particular kind of intellectual. They are the form in which the knowledge generated out of the lived experience of a social group with the potential to become hegemonic achieves coherence and authority’ (Crehan, 2016). Put another way, any given social group will create its own intellectuals (Crehan, 2016).

Alan Bairner argues that organic intellectuals are tied to ‘specific periods and to specific social groupings that operate within those periods’ (Bairner, (2009a) and that they provide essential intellectual leadership that is not always available from traditional sources. Gramsci (1971: 6) writes:

It can be observed that the ‘organic’ intellectuals, which every new class creates alongside itself and elaborates in the course of its development, are for the most part ‘specializations’ of partial aspects of the primitive activity of the new social type, which the new class has brought into prominence.

Given that the majority of the innovation in the analytics community is being driven from a grassroots, community-oriented set of networks, and is concerned with both knowledge production and distribution, I contend that the analytics community is an example of an organic intellectual formation, albeit one lacking in the political utility or subaltern revolutionary ethos that Gramsci’s work suggests.

As Bairner notes, much of the Gramscian-indebted sociology of sport research tends to have the Marxian revolutionary context removed from it; yet, a narrower reading of Gramsci’s work allows for a more culturally oriented emphasis, such as the one I apply here (Bairner, 2009b). Crucially, however, as Bairner suggests, the labour of fans and fan communities does indeed fit in with both Gramsci’s theory of hegemony and the organic framework:

Hegemony theory . . . suggests that activities such as football became contested sites upon which the working class could seek to establish its own authority and values. This approach persists in debates about the contemporary world of football with fanzines and unofficial and even semi-official supporters’ organizations being considered as potentially counterhegemonic and fans’ representatives being portrayed, by implication, as organic intellectuals in the Gramscian sense (Armstrong and Giulianotti, 1997; Brown, 1998; Crawford, 2004, all cited in Bairner, 2009b).

Fans are now creating new scholarship, analysis and ideas around hockey, bridging the gap between passive consumers of the game and active scholars of the traditional intellectual makeup, helping to transform a social group’s incoherent individual experiences and ideas into a coherent set of narratives (Crehan, 2016; Gramsci, 1971).

A brief history of ‘Hockey Men’ and the NHL’s analytics turn

Coaches, scouts and managers tend to be ex-hockey players and, as such, it is increasingly difficult for new voices to establish themselves in the league in the face of this apparent old boys club, often referred to as the ‘Hockey Men’ (Bourne, 2018). Although not a universal experience, many of the established coaches and general managers (GMs) in the league follow this pattern, and are often fired and re-hired, re-circulating throughout the league for decades in various roles. They tend to join the televised hockey media as broadcasters or studio analysts when between coaching and management jobs. Prominent examples include Brian Burke, the former general manager of the Vancouver Canucks, Toronto Maple Leafs, Anaheim Ducks and Calgary Flames who currently works for Sportsnet, the Canadian national broadcast rights holder, and Don Cherry, a former player and coach whose prominent position on Canada’s flagship Hockey Night in Canada programme, has shaped much of the national conversation around the sport for decades (Gillet et al., 1996; Dallaire and Denis, 2000; Whitson and Gruneau, 2006). These ex-coaches and ex-managers often work alongside other former players, creating and reifying a sort of private coded language about the sport that only ‘Hockey Men’ and the most dedicated fans seem to understand. ¹ In essence, hockey culture is somewhat stagnant because that is the way it has always been; yet, professional hockey is undergoing a cultural shift beneath the surface.

The content created online has changed the way the game is understood, and the change is coming not from the ‘Hockey Men’, who traditionally dominate the sport, but, rather, from the fans. As participant Kent W describes:

You had this collection of people with a variety of backgrounds coming together freely to share the insights and ideas that they were all working through and independently discovering. It allowed it to grow organically upon itself. Why is it a bunch of amateurs in their basements coming up with this stuff and not the professionals who played the game? The amateurs in the basements have so many advantages over the pros: free information exchange. We don’t have the pressures of getting something right every time or else we’ll lose our jobs. We don’t have the pressure of seasons and having to turn things around over a season. We can paint broad-brush trends and don’t have to fight with a team or a coach over what something means. We have total freedom to create and test new theories, get them wrong, revise them usually across all of these disciplines and with all of these intelligent people with different backgrounds.

Although some involved in the development of these measurements are, to all intents and purposes, actual experts in mathematical and data science, they still represent an outsider’s point of view. Some have taught themselves how to use complex computing languages such as Python and R in order to write programs that automatically scrape data from the NHL’s website and place the information into scripts and code they have produced to analyse the data and generate real-time updates, charts and ongoing analysis. Others, such as respondent Campbell W, have begun to develop machine-learning programs that can predict some of the randomness of the play and scrape visual data, frame by frame, from television broadcasts. They operate outside the control of the league, working from the bottom up, and the resistance they face is tied to the traditional power brokers of the league, who are attempting to maintain control over both the way the game is played and the way it is consumed by its fans. In essence, this battle is similar to what Millington and Millington (2015) suggest Bill James and other sabermetricians sought to do with their evolutionary baseball statistics, that is, replace ‘traditional, subjective and flawed’ metrics with more accurate assessment tools in a concerted effort to uncover probable truths (Millington and Millington, 2015).

Similar to Jamie Cleland’s (2014, 2015) data, which were derived from online forums and message boards for fans, the development of these measurements continues to emerge in conversations held primarily on blogs and on Twitter, although some private sector and academic inquiry is also beginning to take place. According to respondent Garret H, hockey analytics broadly tries to answer three questions:

How good or not good is this player relative to everyone else; why are they good or bad as they are; third question is the large trivia-style question – sometimes some stats are straight-up trivia. They don’t necessarily tell you how good or not good something is, just interesting trivia.

According to several participants, the earliest iteration of advanced stats in hockey as they are currently understood emerged in 2006 from a now defunct blog known as Irreverent Oilers Fans, and were focused primarily on the first of Garret’s three questions. The first three of the ‘fancy stats’ as they were colloquially called online at the time were ‘Corsi’, ‘Fenwick’ and ‘PDO’. ²

Over the course of the next several years, these bloggers began to exert more of an influence over the fans of the game, starting with the Edmonton Oilers and Calgary Flames blogs and expanding throughout Canada and into the USA. This led to a rapid proliferation of hockey and statistical analysis websites, blogs and Twitter accounts across both countries that focused both on individual teams and the entire league, including Behind the Net, Corsica, War on Ice, Puckalytics and PuckIQ. Garret H describes his own start with analytics:

While I was going to school, it was right around when the [Winnipeg] Jets were purchased from the [Atlanta] Thrashers group. I was following hockey, I used to play hockey and all of a sudden I wanted to find out more. I started Googling, but nobody followed the Thrashers. Twitter was just starting and the blogosphere back then was basically just Alberta and like Gabe’s blog (Behind the Net) and some other stuff here and there. When I rolled in to Gabriel Desjardin’s work, that’s how I ended up in the hockey world . . . It was a very organic thing. I started doing some blogging while I was in school and those things started growing very quickly. And my little dirty secret is that I actually never graduated because this hockey stuff started to pick up.

These sites spawned even more dialogue on other social media platforms beyond individual blogs and the communities that formed around them. Indeed, there is an entire subset of users on Twitter devoted to developing, refining and, most importantly, publicly sharing the data in a variety of ways, from dialogue to advanced visualizations, charts and graphs.

The summer of 2014 became known as the ‘summer of analytics’, as several teams decided they would hire prominent bloggers to develop analytics departments. In some cases, these new members of staff even joined the ranks of upper management. The league itself followed this trend, revamping the NHL.com website to include several of the measurements that had started online, but renaming them in an attempt to make it appear to the casual fan that the league was creating something new (Vollman, 2016).

This is where the story begins to take shape: first, the vast majority of the people hired by teams were not experts or professional data scientists; second, the teams were not entirely sure what to do with the people they hired, which resulted in many being let go within two years; third, reactions to this wave of new stats and new hires were decidedly mixed, because established hockey people both in the league and in the media as well as fans who identified with the old guard felt threatened by the wave of new voices, ideas and information suddenly taking over their sport (Lile, 2016; Proteau, 2016; Richardson, 2014). Finally, as more of the brightest bloggers were being lured to teams (Bourne, 2018; Richardson, 2014), a new set of bloggers emerged to take their place, taking the raw, descriptive analytics and developing more in-depth predictive measurements modelled on baseball’s sabermetrics and wins above replacement (WAR) models.

Analytics and social media

Hall et al. (2003) frame the ways in which journalists report public perceptions using social media platforms such as Twitter. Drawing from their own interview set, they argue that:

There is instant feedback from Twitter and it allows for a conversation. It wasn’t too long ago that we worked in a vacuum – the only feedback, if anything, would be the letters page or the odd bit of correspondence. Now it’s constant. There’s also a form of direct contact with players and clubs. Most of all, Twitter is a source of debate, stimulation, argument and banter, (p. 174).

In the context of hockey analytics and how it has evolved, the same logic can be applied. Given that hockey analytics was essentially born on the Internet, spaces such as blogs and Twitter feeds have been used explicitly to move the conversation forward. As Sunil A stated:

The structure of the blog is what drives the discussion. It’s never ending. Every blog entry is not a finished document. It’s the comment section where you get your spin-off discussions. That may happen on another message board or Twitter or somewhere, and if you try to capture it all at once, that is a massive amount of information to process.

When the more community/electronic tribe approach of the fans’ independent discussions is contrasted with the ways in which teams and traditional journalists use social media, the differences are subtle but present (Norman, 2014). Teams often have their own websites and social media feeds, which act as hubs for teams’ approved editorial content, often uncritical and designed explicitly as commercial activity Hall et al. (2003) More often than not, this gives the appearance that clubs are behaving as reporters and are assuming control of the message. It is notable that official club Twitter feeds employ appropriate journalistic techniques, are often written by journalists directly employed by the teams and focus on the content historically provided by a third-party news agency, but they would hardly be used to criticize the club producing them. It is suspected by many of the analytics creators that the desire for clubs to control their own narratives is at least in part why clubs and mainstream media personalities (known colloquially as the ‘MSM’) are resistant to the point of hostility towards using many of the new measurements. As one respondent suggested, ‘Any time you have an investigation where the results threaten control, there’s really no avoiding it.’ Yet, despite the increasingly active presence of clubs on Twitter, rarely do they come into direct contact or engagement with the analytics community.

Seraj (2012) argues that, ‘online communities create value and become irreplaceable for their participants’. This unique form of fan engagement presents value to its users through content quality and platform interactivity that provide entertainment as well as an opportunity to form relationships. It becomes a self-governed culture that develops trust and respect for the community site and among members. This very culture is what fosters the dialogue and the desire to collaborate and continue to produce content. A few of the leaders in the field have begun to publish books, for example, Rob Vollman with Hockey Abstract (2016, 2018). In addition, scholarship is beginning to appear in peer-reviewed data science and business journals (Chan et al., 2012; Fry and Ohlmann, 2012), but the bulk of the content is still created online.

The people producing this content seek to determine different ways of understanding the game as well as the ability to predict outcomes and models for future individual and team-based success. These processes seem to be divided into a few different styles. Basic descriptive models such as Corsi, Fenwick and PDO are easily tracked and counted manually. More complicated models, such as xGF (expected goals scored), and relative models such as CorsiRel, use linear and multiple regression techniques in order to chart scoring chance and possession metrics relative to both team-mates and the entire league in short-, medium- and long-term sample sizes. In some instances, the data can be automatically scraped from the NHL’s own website immediately following games and published online within minutes (Anderson, 2016). In others, the regression models take data points from the entire season or across several seasons to produce more longitudinal results for players and teams. Some analysts will then create simple, easy-to-understand visualizations like Micah M’s heat maps, which are charts used to isolate player tendencies on the ice over a chosen period of time. See Figure 1 as an example.

OPEN IN VIEWER

Figure 1.

Heat map visualization measuring shot attempts by distance and location for a single player over time.

Micah M describes the use of visualizations as the most effective way to simplify the data for the average fan, stating:

If you instead have a visual perspective, you’ll open yourself up to people who want or can take in information in that form, so the audiences change, the conversations change. The first group of people tends to be resistant to nomenclature changes. The visualizations can open conversations, and I think that is really instructive.

Having visualizations allows the data to be consumed in an easy-to-understand way, ensuring that it is presented effectively and in a digestible format so that even the average fan without any prior statistical knowledge can understand the significance of the metric.

It should be noted that many of these models have not been tested or subjected to peer review and the lack of confidence intervals renders a lot of them more descriptive than truly predictive, despite the number of respondents who suggested otherwise. Respondent Darcy M, however, was quick to claim that almost all of the first waves of analytics were meant to be descriptive; the newer models are much more predictive, for example, the expected goals (xGF) metric presented in Figure 2.

OPEN IN VIEWER

Figure 2.

A breakdown of regular season actual goals scored vs expected goals scored.

Facts vs narratives and the role of ‘expert’ opinions

Scott Stinson, a contributor to the hockey Twitter community, writes: ‘as much as there is acceptance now of the value of data, often what that data says is something that pushes back against deeply held beliefs’ (Stinson, 2017). ‘Deeply held beliefs’ and ‘conventional wisdom’ are terms that are broadly reminiscent of ‘common sense’, a Gramscian concept used extensively by Brent McDonald (2013). In this formulation, Gramsci (1971) suggests that common sense is a sort of folklore of philosophy that reinforces the ideologies of dominant groups. It is supported by myths, half-truths and truths that combine to form a ‘taken-for-granted’ yet rigid understanding of the social world (McDonald, 2013). Gramsci (1971: 419) claims:

Common sense . . . takes countless different forms. Its most fundamental characteristic is that it is a conception, which, even in the brain of one individual is fragmentary, incoherent, and [inconsistent], in conformity with the social and cultural position of those masses whose philosophy it is.

Yet, this concept is not inherently problematic. Given that any community’s intellectuals draw from knowledge and experiences mediated by the ‘narratives available to us’ (Crehan, 2016), all groups will have their own versions of common sense that will be shared among their members and that are different from those of circles further away.

The challenge in mediating this problem is the divergence between analytics ‘experts’ on the one side and the establishment ‘Hockey Men’ on the other. One anonymous respondent characterized the divide perfectly:

Mainstream media (MSM) is bad and NHL people are dinosaurs; we are the rebel alliance, they are the (evil) empire. There is only one way to know things and that is through numbers and, most importantly, we will shit all over anyone who disagrees.

Even those analytics experts who played the game growing up, such as respondents Cole A, Kent W and Campbell W, suggested that their interest in data often ran against the perspectives of the coaches and other established ‘Hockey Men’ with whom they interacted throughout their playing careers. As Cole A notes from his own experience:

I used to play on a team with some of the dinosaur types, who would say that Corsi isn’t really a thing. I like those guys and I always appreciate those guys who would stick up for me and make me feel, like, less threatened. I think I’d have some of those inherent biases being an insider to the whole machine of junior hockey and college hockey. I think it’s a blessing and curse, how I got here.

Other bloggers interviewed have suggested that the reason they became interested in using data-driven analysis was to cut through faulty premises and narratives that were often prevalent in mainstream hockey coverage. Sunil A describes his reasons for getting involved in these conversations:

I was growing really tired of the false narratives, things that were so subjective, the types of criticisms of players. It wasn’t smart enough analysis – I wanted something more in depth . . . So I decided that I’m going to use the blog that I started for grad school and expand my thoughts a little. So I started finding wherever data was available, Excel files and plugging them into MS Access, really simple things. But everything I wanted to analyse always came with questions that came up while I was watching. I remember watching one game and Shawn Horcoff was the best player on the ice, but the commentators and talking heads between periods kept talking about his +/− [the number of goals scored with a player on the ice vs the number of goals against with that same player] and how bad of a player he was and how bad a season he was having, all this criticism. It didn’t sound right. So I thought ‘why don’t I look for the evidence, because I’m pretty sure he’s a good player.’ That’s how I really got started – bad information that was being presented to me on broadcasts, in the paper, and then I also wanted to support my own analysis with data, because I know it solidifies your argument and try to build on whatever I was working on. But nothing was ever a finished article – I would go and revisit it every few months, just to see.

As Millington and Millington (2015) suggest, the intention behind using data in sport is to uncover probable truths; Sunil’s perspective demonstrates that this is exactly the point of his project. Indeed, Sunil’s attempt to break down faulty premises by using data, links his work to the history of analytics usage across global sport both in the recent past such as English football club Manchester City’s fan engagement with analytics and the analytics turn in baseball in the 1970s (see Furnas and Lezra, 2012, cited in Millington and Millington, 2015).

‘Hockey Men’ in the mainstream media

There are several prominent MSM sportswriters and commentators in Canada who have gone on record as rejecting the fan-driven analytics. They tend to rail against people who are doing a significant amount of detailed research and legwork to try and move the conversation about the game forward. Instead, traditional hockey journalists tend to write more traditional hockey stories that focus on ‘grit’, ‘heart’ and ‘leadership’, intangible concepts that cannot be precisely measured. This is the very sort of taken-for-granted understanding of the game that fits in with McDonald’s (2013) interpretation of Gramsci’s common sense model.

As they consider themselves to also be ‘Hockey Men’, these journalists effectively deploy their insider’s position to dismiss anything that may challenge the traditional view of the sport, often appealing to an authoritative rhetorical fallacy to justify their points of view (Bourne, 2018; Pucktistic, 2017; Yost, 2014). Scott Stinson stated in a 2017 blog post: ‘rarely does a week go by when someone doesn’t take a shot at analytics, or when some kind of development with a team isn’t seen as an indictment – or exoneration – of its reliance on data’ (Stinson, 2017). However, as respondent Rob V described, mainstream sportswriters have been criticized for not doing much of the legwork themselves, rejecting these new ideas without necessarily investigating their potential or demonstrating a willingness to listen when explanations are offered.

Prominent examples in Canada include Mark Spector, David Staples and Derek Van Diest in Edmonton and Damien Cox and Steve Simmons in Toronto, as well as Don Cherry and Nick Kypreos from Sportsnet, the national NHL broadcaster in Canada. Pierre McGuire, ex-head coach and assistant GM of the Hartford Whalers and current hockey analyst for NBC sports in the US and TSN in Canada was once quoted as saying ‘analytics doesn’t measure heart, courage or desire’ (Stinson, 2017). Van Diest has routinely critiqued the very legitimacy of most of the new stats, whereas Spector has been known to use personal and ad hominem attacks aimed directly at the bloggers (Yost, 2014). A prime example of this type of reporting is found in an article written by Spector in 2017, in which he discusses the impact of a deal the Edmonton Oilers made, acquiring a defenceman in exchange for a forward. Spector (2017) writes:

The stats geeks still hate the [Taylor] Hall trade. Conduct a poll of 200 ‘Hockey Men’, and it might be unanimous: Edmonton got what it needed in that deal, and giving up Hall was well worth it. Hall’s Oilers would curl up in a ball when met with this game. [Adam] Larsson’s prevailed. It’s not so simple – no trade is – but GM Peter Chiarelli really nailed this one, acquiring a hard, sometimes dirty, physical player who perfectly embodies everything the old Oilers lacked.

Such an argument suggests that the archetypal ‘Hockey Man’ is a sort of omniscient expert of the game, one who gains his knowledge from years of experience watching the sport without using a data-driven approach, and that those who attempt an empirical analysis do not actually understand the way the game is actually played. Indeed, it could be argued that this resistance to new ideas is a form of gatekeeping that stifles innovation and new thought in a sport that some argue desperately needs it, including current NHL players such as the aforementioned Taylor Hall, who won the Most Valuable Player Award in 2018 (Hall, 2019; Pucktistic, 2017).

In many cases, the journalists tend to mirror the perspective of other established ‘Hockey Men’ who are still active, such as coaches, scouts and managers (Bourne, 2018). The aforementioned Spector works in Edmonton in Canada, covering the Oilers’ games on a regional sports television network and writing regular columns online. It is of little wonder that his perspective mirrors that of the Oilers’ coaching staff. Former head coach Todd McLellan has often derided analytics with such notable quotes as, ‘the best analytics is a set of eyeballs’ and ‘my players all understand his importance. So analytics that if you want’, the latter remark made in defence of a player viewed more sceptically in the analytics community than by mainstream hockey media members (Henderson, 2017). Other teams, such as the Florida Panthers and Arizona Coyotes, have fired coaches who used more data-driven approaches, in favour of more established ‘Hockey Men’ (Pucktistic, 2017). ³ Given that many who cover the sport or have front office roles with teams often played at the highest levels, they trust in the fact that they know the game better than anyone else (Bourne, 2018).

Respondents Rob V and Cole A suggested that the best way to challenge the resistance is to simply get better at selling the ideas, almost as if the responsibility is on the content creators to teach the concepts to the journalists. In turn, it becomes the responsibility of the MSM to actually take the time to listen. In Rob V’s case, this has been a moderately successful tactic because several prominent mainstream hockey journalists have endorsed his regularly updated Hockey Abstract book, leading to his recent hire by the Los Angeles Kings as an analytics consultant. However, despite his moderate success in reaching mainstream hockey journalists, even those that endorse his work still rely primarily on more common sense-style narratives in their coverage, a sentiment echoed by Micah M:

[S]ome people prefer to inflame the rhetoric, even as it mitigates. . . There will be an elaborate face-saving ritual by all concerned. I think some statisticians will start to mollify some of their critics and concerns as they start to gain relevance . . . Many different journalists do primarily parrot the responses that are fed to them by teams in order to maintain access.

Clickbait journalism and ‘hot takes’ as they are referred to online, is the current norm in the journalism industry, one crippled by a dwindling subscriber base and an overreliance on advertising to generate revenue (Himma-Kadakas, 2017), an issue Cole A feels is a major part of the problem in creating legitimacy for hockey analytics: As Cole A suggests:

I’m not a media person at all. I think media and hockey is like an entertainment product. Nobody’s curing cancer. So I understand why hockey media personalities eschew the data. But there’s also a growing market for it . . . It’s tough to keep up and sell the general public on it. If you put 100 analytics people in a room at a conference, it would be tough to sell some of those people on your ideas, let alone the entire public who have no idea what you’re talking about. So I understand the desire to keep [hockey stats] simple and accessible. That’s how people are going to consume the NHL. You want that engagement. It’s evolving.

Although sport media still remains incredibly popular, the writers’ rooms at many newspapers across North America are contracting. In Canada, the merger of the Postmedia and Sun Media groups has led to a combined staff and more articles shared across platforms and networks, resulting in a smaller pool of articles, and journalists competing for a diminishing number of readers and clicks (CBC News, 2016). Sunil A states:

There’s so much misinformation out there. I don’t believe most of what these guys write. They don’t just sit there and come up with something and write about. There’s obviously someone from the team or a player agent or player to get something out there and then the masses have to chew it up like it’s gospel. Then we talk about it– I take a step back and realize that so much of it is garbage. I want factual information; I want to read an article and not feel like I lost a brain cell. Just to get that independent analysis is so important.

Given the current media landscape and the rise of ‘fake news’ and ‘alternative facts’ within it (Farkas and Schou, 2018), the scepticism of many of the bloggers with regard to the MSM is, at least, somewhat understandable. The bloggers are generating massive amounts of data, presenting them freely across a multitude of sources and, yet, the near constant rejection of evidence fits into a broader trend towards the increased erosion of evidence-based discourse in the news (Farkas and Schou, 2018).

Farkas and Schou argue that the prevalence of fake news has become something akin to a floating signifier, whereas the erosion of evidence-based inquiry ‘has become a deeply political concept used to delegitimise political opponents and construct hegemony’ (Laclau, 2005, cited in Farkas and Schou, 2018: 300). Laclau (2014) states:

There are no facts without signification, and there is no signification without practical engagements that require norms governing our behaviour. So there are not two orders – the normative and the descriptive – but normative/descriptive complexes in which facts and values interpenetrate each other in an inextricable way, (p. 128).

Laclau’s framework is valuable here given that it offers insight into the way in which facts and values are intimately connected and, more importantly, into how values have an impact on the framing (or lack thereof) of any fact. Rob V suggests that many of the critics of hockey analytics do not even believe that there is any value at all in performing the analysis of the sport in the first place:

In hockey, sometimes you’ll run into that, where you’ve explained something and it’s not that they disagree with your interpretation of the results or that they disagree with the way you gathered the data or how you got there, but they don’t even see value in even measuring something. Why would you even count how many of those events occurred, why would you even base an opinion on a measurement. That’s literally where they sort of get stuck.

Micah M echoes this sentiment, taking the analysis even deeper into the realm of hegemony and control:

In-group of hockey people adopting processes – that particular strain of populist politics, particularly in the USA, has strengthened the need and desire for evidence-based inquiry on the left. I don’t think it’s a coincidence that some of the angles that these amateur hockey statisticians have taken. I don’t think the resistance or rejection of evidence between the two groups has to do with data methods, but rather I think the data methods arise because of the opposition. Any time you have an investigation where the results threaten control, there’s really no avoiding it. Nobody will tolerate giving up things they control or feel like they control.

Similar to the way in which Farkas and Schou (2018) characterize the way the political Right has weaponized the term ‘fake news’, it could be argued that the ‘Hockey Men’ and mainstream journalists often delegitimize analytics in order to ‘invalidate their position within the field of power, deconstruct their public authority and re-hegemonize their power’ (Farkas and Schou, 2018). Micah M shares this perspective, asserting that:

[M]any different journalists do primarily parrot the responses that are fed to them by teams in order to maintain access. . . . I don’t know if it’s important for groups like this who are doing analytic work, I don’t think they need any league recognition to operate.

This perspective demonstrates that the work being done online can survive outside the confines and constraints of the league.

In hockey, the differences in approach between the MSM analysis and a more data-driven one are summed up effectively by Micah M:

When mainstream commentators, or really anyone with a platform but especially mainstream journalists say something very silly, it provides me with an avenue for research that I know will be interesting. One of the pet peeves that I have is people who describe players as ‘perimeter players’ and this comes up constantly. I couldn’t tell you who the first person was to fall under that banner unnecessarily – or not even unnecessarily – but I felt unjustifiably where I didn’t know if it were true or not what was being said. So I started making heat map charts to show ‘where does this player take their shots from’ and ‘where does the team take their shots from when they’re on the ice’. . . . I started taking these things and using them as avenues of inquiry that would drive new visualizations and new work. Not only do I come up with something I think is actually good, but also because of its genesis, I know it’s going to be interesting to a number of people – it’s already part of the conversation. So, I don’t always and need not engage directly and I rarely respond to journalists, because I find that most journalists have nothing to say to me and have nothing to do but just yell.

What Micah suggests is that the resistance to data is part of a much broader resistance to bursting a particular narrative bubble. This resistance demonstrates that many MSM writers are afraid of losing their sense of absolute authority over their particular domain as a result, including the freedom to criticize certain players based on their perceived expert understanding of the game despite evidence that proves otherwise.

Ironically, this rejection of evidence has occurred despite a much more robust, nearly 24-hour news cycle in which sports journalists are also using blogs and Twitter to break stories before their articles appear on traditional platforms Hall et al. (2003) Micah M. speculates that the aversion to the data is concerned with the numbers themselves: ‘people will engage with pictures in ways they won’t engage with numbers or tables, especially numbers. There’s an aversion about numbers.’

The aversion to data in general and a numbers-focused analysis in particular, connects with the prevailing wisdom that ex-players often make the best broadcasters in hockey. The anti-intellectual bias associated with sport culture is highly pronounced in hockey circles, and the resistance to evidence is part of a larger ‘jocks vs nerds’ narrative from an earlier era within the broader history of hockey analytics (Bourne, 2018). It is also part of the emergent trend towards accepting narratives at face value without examining all the available evidence, a trend consistent with the rise of ‘fake news’ and ‘alternative facts’ in the North American media landscape.

The analytics camp is attempting to overthrow one particular hegemonic system of thought in favour of another (Farkas and Schou, 2018). Although respondent Dom L has a journalism background, he discusses the disconnection found in hockey coverage very bluntly, stating that his role as an analyst is:

Just about providing a different voice. I don’t want to say it’s ‘fact-based’ but in mainstream reporting, they talk about how they feel. There’s always going to be a market for that. What they feel, what they see, what other ‘authoritative’ figures see. If that doesn’t line up with the actual results, someone should also be talking about [those results] and why [they might be] different . . . I look for the story behind the numbers and the facts rather than making a story out of numbers and facts.

Because the league has invested so heavily in the production, control and distribution of blogs and Twitter-driven analytics, they and their creators could be viewed as a threat to maintaining legitimacy and control. Many bloggers feel that those who seek to criticize analytics use a single example of a just one statistical category in their argument and fail to necessarily comprehend the concept behind the analytics movement, namely, that all the categories together help to colour in some of the grey areas of the game.

Conclusion

As I have argued, the analytics camp is locked in its own ‘war of position’ (Gramsci, 1971, cf. Bairner, 2009b) against the hegemony of traditional hockey fans, coaches, management and, especially, members of the mainstream hockey media. The people who are driving innovation in this field are not exclusively professional data scientists but, rather, intellectually curious fans of the game. Some of the people involved in this new wave of hockey analytics, such as the founders of PuckIQ, are incredibly resistant to the idea of working for teams and have determined that their work should remain open source and freely accessible to all who want to use it. Ganesh M, whose work as ‘Oilers Nerd Alert’ has drawn the attention of the Edmonton Oilers’ management, has not considered working for the franchise because of his personal commitment to provide data that are open and freely available. The bottom-up, almost subaltern approach the bloggers have primarily taken is consistent with Gramsci’s organic intellectuals framework. As Kate Crehan (2016) argues,

Gramsci was especially concerned with the role organic intellectuals play in a subaltern group’s rise to hegemony, and to him it was clear that such intellectuals could not simply be traditional intellectuals with progressive aims; they would represent a new kind of knowledge producer (2016, p. 36).

The ‘Hockey Men’ in both the league and the MSM, and the online analytics camp each hold a distinct worldview that ‘does not translate unaltered across hegemonic projects’ (Farkas and Schou, 2018, p. 308). What we see here is a struggle similar to the battle with regard to what constitutes real vs ‘fake’ news in an increasingly partisan news media and political sphere; in this case, it is a struggle to determine who maintains the power to shape the narrative of the sport and either preserve or rise towards a hegemonic position. Jhally’s (1989) contention that sport may be ‘relatively free of contestation’ might no longer apply. Although it is clear that this ideological and epistemological battle for control of the sport is a reflection of a broader debate between qualitative and quantitative approaches to knowledge production, media and politics, it will be interesting to watch this struggle continue to unfold online, in the media and on the ice.