It’s the last day of the conference, the day where often new members make their first INFORMS presentations. Many of us know someone whose training is not in ORMS per se, but whom we’d enjoy seeing at the conference because of their related work. I decided to interview someone whom I have encouraged to attend a future INFORMS conference, Dr. Wayne Mackey.
Wayne is a Neuroscientist at NYU (and received his PhD from there), and is the Founder and CEO of Statespace. Statespace’s mission is to bring standardized gamer performance analytics, training, and scouting to Esports. Statespace’s first product, Aim Lab, is a training tool that utilizes neuroscience and artificial intelligence to become a smart personal trainer for both professional and casual first-person shooter players. He was kind enough to answer a few questions of interest about the present and future of video game analytics, and how he approaches analytics from his neuroscience background. I’ve also included some sample screens from the Aim Lab product to give you an idea of the analytical output, for those interested.
How does your background as a scientist affect how you look at analytics problems? Have you worked closely with mathematicians or engineers in analytics, and what differences do you see in perspectives?
“As a scientist, I am very hypothesis-driven. That is, before I even start looking at data I try to have some things that I want to test specifically. Rather than just throw the proverbial machine learning (ML) kitchen sink at a set of data, I think deeply about what I want to get out of it and why, and let that drive what methods I use to collect or analyze data. With all of the cool ML and stats toolboxes out there, just about anyone can download a package and some data, and run some technically sophisticated methods of analysis at some data and get something to come out the other end. While it’s amazing that it is so easy to do so, that can also be harmful without the proper theoretical framework around the problem you are trying to solve.
I’ve worked with mathematicians, engineers, physicists, neuroscientists, and just about anyone else with a strong computational background interested in ML and data science. Honestly, I haven’t noticed any practical differences in perspectives based on domain, but more so just differences based on personality and unique thought processes for problem solving in general…and I think that variability is a very valuable strength and much-needed.”
How have you handled the challenge of presenting detailed performance results to an audience of gamers who may not understand statistics well? What changes have you had to make to better accommodate their feedback needs?
“This is actually something we are constantly optimizing and learning from. Our first approach was to appease the hardcore players who would be data-hungry, and assumed that our product would attract those types of the players the most. We were very wrong. We quickly found that dumping data on players was a terrible idea, and that many of our data visualizations were confusing and lacked context. That wasn’t true for all players, but definitely for the majority. So we took that valuable feedback and learned from it, rolling out new ways of conveying critical inferences about their performance data.
Our recent approach has been to split the data across a few tabbed screens. So when receiving feedback, the initial screen is as simple and straightforward as possible. We actually take important things to take away from the data and codify that directly into actionable text. For instance, instead of just showing a player with data or a visualization that they are undershooting targets, or less accurate at one portion of the screen compared to another, we literally just say it. For those who want to dig a bit deeper and get their hands dirty with data, we have advanced tabs where all of that information is just a click away. So far we have found this is a great happy medium for both types of players.”
Video game analytics are still in their infancy in some ways. What statistics do you see becoming more important over time? Is it easy to find past databases of gamer performance in top competition? How are gamers responding to the concept of analytics in improving gameplay?
“Gamers, as well as coaches and team owners at the professional level are craving analytics. There are constant comparisons and parallels being drawn between traditional stick-and-ball sports and Esports. Although Esports has already surpassed several traditional sports in viewership, it lags behind in core things like data analytics. I don’t know that there is a single statistic that will become more important over time, but I do think that individual fundamental skill and ability data is severely lacking and something that will be vital in Esports just like it is in traditional sports. In traditional sports, you measure how fast someone is, how strong someone is, how high they can jump. These are fundamental skills based solely on your own performance, are measured outside of the game being played (such as the NBA or NFL combine), and are free from confounds of the skill level of your teammates or opponents.
What the analogues of these fundamental skills are in Esports, and how to measure them in a standardized way have evaded the industry as a whole. This is something we as a company are changing. This type of data is profoundly valuable in traditional sports – from scouting to training and storytelling. And we believe this data will be just as important in Esports.”
My thanks to Wayne for his time. I searched the INFORMS program and noted we had three presentations that mentioned video games this year. I’ve encouraged him to submit a presentation to INFORMS next year, and hope to see more work in the emerging area of video game analytics. I also encourage others to join me in inviting their friends and colleagues in related fields to make an INFORMS presentation in Seattle next year!