The Playful Side of Data Analysis

In his book Now you see it, Few presents a list of personal traits that a data analyst should exhibit (Few, 2009, p19). Some of Few’s traits reflect the common, serious tone of data analysis: being skeptical, methodical, analytical, etc. However, Few begins his list with five traits that seem separate from the other serious, traits. I argue these traits form the “playful” side to data analysis:

Playful Analyst Traits

  • Interested
  • Curious
  • Self-motivated
  • Open-minded and Flexible
  • Imaginative

Serious Analyst Traits

  • Skeptical
  • Aware of what’s worthwhile
  • Methodical
  • Capable of spotting patterns
  • Analytical
  • Synthetical

While play, or being playful, is a notoriously difficult concept to define I am using the term in relation to two definitions that present possible ‘properties of play.’ First, Caillois in Man, Play and Games lays out his properties of play in a sociological pursuit to describe how culture is represented through play and games (Caillois, 2001, p9-10). Second, in the book Play, Brown presents a similar list of play properties but structures his arguments from a clinical perspective, having studied how play is exhibited by both animals and humans (Brown and Vaughan, 2009, p17). The image below compares Caillois and Brown’s respective property lists identifying their similarities and differences.

Both researchers agree that play is free or voluntary, meaning players are not obligated to participate in a playful activity. Both researchers view play as having an unproductive side but one of Brown’s major arguments is that play is vital for learning and living a healthy, happy life (Brown and Vaughan, 2009, p48-51). Caillois’ list goes on to state that play is uncertain and players create new sets of rules to govern how play commences. Brown on the other hand acknowledges the uncertain, improvisational potential found in play and states that rules are not necessary. There are also loose connections between the Caillois property of make-believe and Brown’s properties of ‘diminished consciousness of self’ and ‘freedom from time’. Each point to the alternate reality play creates. Caillois acknowledges a second, make-believe reality created by play and Brown acknowledges what the player experiences, a loss of self and time. Caillois’ additionally states that play is separated from the real world, which seems to point to how play functions as a practice. Brown chooses to instead highlight a player’s desire to continue playing once started, a property that speaks again to the player’s experience.

Each of Few’s data analyst traits I label as playful can be matched with the aforementioned properties of play laid out by Caillios and Brown. Traits such as being interested and curious coincide with the inherent attraction property of play. An analyst that freely expresses interest in a data set is more likely to be able to play with that data. Self-motivated, open minded analysts set their own goals or rules creating a situation where they improvise how they analyze a data set. Analysts may even feel motivated to continue their analysis even after they have gain their initial insights from their data. Finally, analysts must be imaginative, creating make-believe scenarios and new ways to illustrate the hidden patterns in a data set.

Where Few’s analyst traits and the properties of play do not concur is how play must be unproductive and separated. Analysts need a playful and a serious side to their personality according to Few’s listed traits - one connects to the other. Imagining a new visual orientation for a dataset can be proceeded by further analytical and methodical approaches. These actions certainly do not need to be separated, in regards to the ‘separate’ play property Caillois mentions and happens as a natural part of the analysis process. Therefore, data analysis needs to provide both a space for analytic endeavors and a space for play too. Data analysis is both an analytical and a playful activity, an activity that is both serious and playful.

The phenomenon of moving data analysis beyond serious analytic systems has already begun within some data analysis communities, at least at the fringe. Pousman, Stasko and Mateas’ (2007) research reports a growing trend in information visualization, a type of visual data analysis, to provide a wide array of audiences with systems that do not solely focus on serious analysis. They call these types of systems “Casual Information Visualization” which they define as “the use of computer mediated tools to depict personally meaningful information in visual ways that support everyday users in both everyday work and non-work situations” (Pousman, Stasko and Mateas, 2007, p1149). They separate Casual infovis into three categories to correspond to the type of systems they covered. Ambient InfoVis are systems found in “peripheral locations and provide abstract depictions of data.” Social InfoVis systems visualize social networks and allow its users to interact with their social data. Last, Artistic InfoVis are systems with the “goal of challenging preconceptions of data and representation.”

Casual infovis systems can be seen as less productive, offer a wider variety of improvised data representations, are used on a more voluntary (i.e. casual) basis and seem purposeless to other users. Each of these descriptions is also a property used to characterize play by Caillois and Brown. Ambient Infovis systems, for example, provide displays that may include raw data streams which are less productive to interpret. Whereas a Social Infovis system may visualize data in an analytically productive way but only present data related to a single user. One user may find no purpose in interpreting another person’s data but would be inclined to analyze their own data even if only casually. Finally, Artistic Infovis systems improvise, or create their own rules for, the form and representations of data that are visualized. These are meant to express an almost make-believe view of what a data set can look like if a typical, serious data representation is discarded.

When one compares the types of systems found within casual infovis to the properties of play, casual infovis can be described as an example of how play is being introduced into infovis. Seemingly unproductive, uncertain, and voluntary systems match with how play is structured and experienced. However, the three categories of casual infovis do not denote play as a specific quality found within those infovis systems. Ambient, social and artistic information visualizations do not necessarily need to be playful. Thus, I have argue that an additional category should be added to casual Infovis, playful infovis, one that promotes play through computer-mediated visualization (Medler and Magerko, 2011). In this dissertation, I use these past arguments to create a new domain that exists within game analytics. The new domain is play analytics and it is a domain describing the ways in which play is promoted or supported through various forms of data analysis and visualization.

Brown, S. and Vaughan, C. 2009. Play: How It Shapes the Brain, Opens the Imagination, and Invigorates the Soul. New York: Penguin Group.

Caillois, Roger. 2001. Man, Play and Games. Translated by Meyer Barash. Illinois: University of Illinois Press. [1961].

Few, S. 2009. Now You See It: Simple Visualization Techniques for Quantitative Analysis. Analytics Press.

Medler, B. and Magerko, B. 2011. Analytics of Play: Using Information Visualization and Gameplay Practices for Visualizing Video Game Data. Parsons Journal for Information Mapping, 3(1).

Pousman, Z., Stasko, J., and Mateas, M. 2007. Casual Information Visualization: Depictions of Data in Everyday Life. IEEE Transactions on Visualization and Computer Graphics, 13(6), p.1145-1152.

Infographic: Monument Valley

Of Bullets and Angels: Why Mirror’s Edge is Modernist and Assassin’s Creed is Postmodernist