Taneja begins with the explosive growth in the gaming industry over the last decade, where, according to his statistics, the video game consumer community ballooned to nearly 10 times its previous size - from around 200 million users (2003) to approximately two billion (2013). At Entertainment Arts, said Taneja, this translates to intense amounts of data transfer - more than one terabyte (TB) of data per day for popular games like "Battlefield," and 50 TB of data overall per month. All this data represents 2.5 billion user-initiated game sessions each month. (Read more about big data in (Big) Data's Big Future.)
The nature of gaming, said Taneja, also has changed to a paradigm where games are enjoyed "in an always-on, always-connected, multi-device, social manner." At the same time, said Taneja, game companies are developing their own, more interactive methods for evaluating game play, and changing their models from monitoring descriptive data to making predictions about game play in real-time.
To help audiences understand how Entertainment Arts has worked toward accomplishing this objective, Taneja showed a model of a legacy architecture involving simple data warehouses that had data latency of up to three days. Taneja then compared this to a more current plan involving visual deep game play insights driven by data taxonomy and new data handling methods that move game data quickly. Here, a model that includes Hadoop and the use of data-culling algorithms accomplishes a latency that can be measured in hours, rather than days.
Companies like Entertainment Arts, said Taneja, also try to capture behavioral data about players, such as monitoring player activities around virtual goods, social interactions, in-game branding or merchandising and other key elements. One goal, he said, is to look for a game player’s total behavioral identity - not the identities that players enter in avatars or other self-expressing user activities, but how they behave within the game, in general. The implications here are important. Capturing this kind of data may very well help video game companies respond to player preferences and actions in real-time. However, in an age where any less than evident digital monitoring can be seen as a step toward more comprehensive "uberveillance," this kind of data mining is almost always a sensitive topic. Here, Taneja presents some sensible objectives behind this kind of monitoring. In-game "snooping," he says, can be a driver of game player accommodation and choice. Of course, it also can be another manifestation of how intensive personal data collection spooks consumers, just as it does in the world of e-commerce or social media.
This video provides a relevant look at a major digital industry, as well as a revealing portrait of how video game companies are currently approaching the digital frontier of business IT.