Schooling the Scholar, Poaching the Fan: Fannish Intellectual Production and Digital Humanities Methods

July 18, 2013, 08:30 | Short Paper, Embassy Regents F

Scholars have often maintained a critical distance between their own forms of knowledge making and those of fans, a distance that we find worth interrogating in the setting of digital humanities. After all, the kind of mapping and charting of vast amounts of cultural data that digital humanists are beginning to do seems closely tied to fan practices of collective textual analysis (and production). We argue against a general academic hesitation to seriously incorporate or interrogate fannish intellectual production as compatible with academic work, and we see digital humanities as a promising site for cooperation between fans and academics. Indeed, digital humanities appear indebted to the intellectual productions of fan communities in an age of media convergence.

Specifically, this paper engages with fan material in a way that acknowledges how it may inform and work alongside academic work in television studies. Fan communities’ collective work analyzing and producing digital work around television shows, including mapping characters and their affective relationships, has heavily influenced our own television analysis method, which involves graphing and analyzing social networks of television characters. This method also draws on the academic work of Franco Moretti and other digital humanities scholars who have demonstrated the usefulness of social network analysis of various texts, but who have not discussed the relationships between their analysis and that of fans. We use digital graphing tools to investigate affective relationships, as defined primarily by fans, between television characters across lines of racial or sexual “difference” in ensemble character-driven dramas. Our data sets have in some cases been gleaned from fan forums, from which we have constructed affective social network graphs using tools like Gephi, ManyEyes, and Mathematica. Analysis of these graphs, we found, illuminates trends and “underlying structures” that might otherwise be difficult to notice (Moretti).

To explore the usefulness of graphing social networks, we demonstrated the absence of queer characters and relationships in the television series Lost, a show with a sizable multi-ethnic cast that draws heavily on the sexual, affective, and familial tension of its characters for dramatic effect. First establishing the overwhelming preponderance of heterosexual pairings, we reorganized the nodes so that the most frequently appearing characters gravitate towards the center of the field, and the less frequently appearing characters radiate outwards. As we might expect, characters who appear most frequently have multiple lines of (heterosexual) relationships, while less frequently appearing characters are likelier to be unattached or to have only one relationship over the course of the series. The suggestion of an undergirding heterosexual matrix, visible here, is borne out in the conclusion of the show. In a "flash-sideways," the storyline in a parallel timeline where Oceanic Airlines Flight 815 never crashes, characters recall the moments on the island when they make contact with their principal love interests. This forces a heteronormative pairing as an organizing force (possibly queering the pair of John Locke and Ben Linus, who recall their past lives but without the benefit of a significant other). The graph also makes legible the outliers, exceptions, and outsiders that relate to the frequently appearing characters in ways that support the structuring heterosexual matrix (See Figure 1). One interesting cluster that the graph brings to our attention, for instance, is the set of characters who are positioned as paternal figures to specific nodes.

Fig. 1:
Affective and Parental Relations in Lost (Nodal size indicates relative frequency of character’s appearance). Constructed using Gephi.

The lines of paternity as a special category that our graphing highlights, and the general heteronormative drift of the show, also invites us to consider the separateness of maternity as a category with a unique valence. Apart from the running theme of pregnancy dramatized by repeated attempts to capture or rescue the visibly pregnant character, maternity we uncover is an organizing type distinct from paternity. While the characters Kate and Jack both co-parent a child, only Kate’s status as a mother symbolically revokes her candidacy as protector of the island in the finale.

Along similar lines, social network graphs of Friday Night Lights reveal the extent of the characters’ racial segregation in the early seasons. [1] We chose this show to analyze because race is a central theme as characters negotiate racial tensions in a small Texas town. In the case of Friday Night Lights, segregation is apparent through conventional modes of analysis, but a graph makes it more starkly visible. This graph of a Friday Night Lights episode, for example, shows a notable degree of segregation between black and white characters. The mix of such characters on the television screen hides the fact that there are very few interracial conversations, which this graph depicting all interactions in the episode maps clearly. Here we see that the majority of white characters are only connected to other white characters, and, importantly, that the white characters tend to be more strongly socially connected than black characters—they are given more social power (See Figure 2). This graph then provides in one image a sense of how racial interaction plays out across a whole episode.

Fig. 2:
Interactions in Friday Night Lights Season 2 Episode 8. Constructed using ManyEyes.

One way we have used social network graphs to heighten visibility of character segregation (or commonality) is through “deformance,” a playful textual reimagining that in some ways resembles remix culture. Deformance is described by Lisa Samuels and Jerome McGann in reference to poetry analysis, as a kind of “reading backwards,” a reconsideration of a text by undoing it, upsetting its order, revealing its gaps (30). In the context of this project, deforming graphs has entailed removing certain networked nodes—and thus removing certain characters from the plot—to see what new connections come to the fore, as well as to reveal the role of those removed nodes in the networked system. If a network falls apart when one node is removed, that often speaks more to the significance of that character than does looking at the graph before he or she is removed from the network. In our analysis of Friday Night Lights, for instance, deforming social networks created from fan-generated relationship information (or from our own fannish data generation) revealed that interactions across race hinged on just one or two key characters. Deforming the episode graph above, for example, by removing the central black character Smash and all those characters who interact exclusively with him, revealed the extent to which he functioned as a “bridge” node connecting two otherwise largely distinct communities of characters (See Figure 3).

Fig. 3:
Interactions in Friday Night Lights Season 2 Episode 8, with Smash and characters who speak only to him removed. Constructed using ManyEyes.

As this work demonstrates, we are not interested in fan labor solely as sources for data, but also as rich and challenging sites for methodological exchanges. The intellectual production of fans has been acknowledged and celebrated; indeed Henry Jenkins’s now-classic text on fan cultures, Textual Poachers, highlights the intelligence of audiences and the seriousness of their responses, which engage in familiar practices of literary criticism. Yet to Jenkins, while fans do work that is critical and interpretive, their criticism “is playful, speculative, subjective” and directed to the fan community (284). We see our methodology as a gesture toward the possibility of digital humanities to engage this playful approach seriously, and to consider the importance of the relationships between fans and other interpretive communities.

With academics and fans less strictly monitoring their boundaries in this era of convergence, and with scholars using digital techniques that may formerly have been dismissed as too “playful” or not seriously analytical, a reconsideration of the fan/scholar relationship and possibilities of exchange (or poaching) is called for. Visualization and mapping are widely circulated by fans broaching similar themes to those we explored. In toggling between an episodic mode of analysis and a mode that exceeds the episode, we rely on the community and labor of fans, appropriating their knowledge as the basis of our inquiry. What this suggests is that the digital humanities can anticipate not just a new kind of appreciation of the fan but an acknowledgement of the rich, intellectually productive, and rigorous strategies of knowledge making that new scholarship exploits in its interpenetrative incursions into the terrain of fandom.


Jenkins, H. (1992). Textual Poachers: Television Fans & Participatory Culture. New York: Routledge.
Jenson, J. (1992). Fandom as Pathology: The Consequences of Characterization. In Lewis, L. A. (ed), The Adoring Audience: Fan Culture And Popular Media. London: Routledge, 9-30.
Moretti, F. (2011). Network Theory, Plot Analysis. Stanford Literary Lab Pamphlet #2. http://litlab.stanford.edu/LiteraryLabPamphlet2.pdf.
Moody, J. (2001). Race, School Integration, and Friendship Segregation in America. American Journal of Sociology, 107(3): 679–716.
Samuels, L., and J. McGann (1999). Deformance and Interpretation, New Literary History, 30(1): 25-56.


1. Our interest in using social network graphs specifically to look at racial interaction was partly inspired by James Moody’s 2001 study of integration in high schools, in which he used social network graphs to make visible certain trends of friendship formation between high school students.