INFORMS Open Forum

Hello everyone! 

As an operations researcher who loves and works on video games, I'm excited to be invited by INFORMS to create this post to discuss the applications of operations research in video games.

Video games are complex systems with multiple decision points, making them a great playground for optimization. That's where operations research comes in! As OR experts, we can help game developers balance game mechanics, optimize player engagement, improve monetization strategies, enhance matchmaking algorithms, and streamline game development processes.

In my research, I've explored the use of OR in video games through two recent papers. The paper 'Loot Box Pricing and Design' (Management Science 2020, link) focuses on monetization through a random selling scheme called loot box.  The paper 'Matchmaking Strategies for Maximizing Player Engagement in Video Games' (link) investigates the use of linear programming to optimize matchmaking in a multiplayer online game.

But I also want to hear from you! Have you worked on any video-game-related projects? It may not be a research paper; for example, one student shared with me his final project in a simulation class, where they used simulation to find the optimal matchmaking policy (link).  Furthermore, what are some promising applications of operations research in video games? What ethical considerations do we need to keep in mind?

Looking forward to your thoughts!

------------------------------
Xiao Lei
Assistant Professor
University of Hong Kong
www.xiao-lei.org
------------------------------

Dear Dr. Lei,

Great post: Interesting and relevant to future applications in esports and entertainment more broadly. 

There are some tournament and regular-season scheduling papers that have been in the literature for many years, and I would do a poor job trying to give a good literature review there. Of course, with esports, you probably don't have the same "travel considerations"! So maybe fairness or evenness of a match is a better measure. I have some research in that area, but I really like the papers you mentioned here. This seems to be akin to an online matching or assignment problem. (See http://shj.cs.illinois.edu/sms.htm for some great research by Sheldon Jacobson on the matter.) 

In terms of fairness, I co-authored a paper https://doi.org/10.1093/imaman/dpr024 (Prince, Smith, Geunes, 2013) that looked at different ways of setting up 8- and 16-team knockout brackets, and these might be relevant to esports tournaments if you have player ratings. The idea is that the probability that team i advances to round k should be at least as large as that for team j, regardless of which round we're talking about. The standard way of pairing (for 8 teams) 1-8, 4-5, 2-7, 3-6 achieves this for round 1, but not necessarily for future rounds. 

I'm interested in reading a lot more of the responses you get, and thanks again for the post. I enjoyed learning from what you've already provided.

Sincerely,

Cole

------------------------------
Cole Smith
Syracuse University
------------------------------

Very interesting topic! I wonder if there are any applications of game theory that would also be interesting to consider.  Agree that simulation is another interesting approach, but curious what are your thoughts on the performance requirements to run a sim and evaluate different decisions to feed back to a game maybe happening close to real time.  

------------------------------
Dayana Cope
GM, Decision Science
Delta Air lines
Decatur GA
------------------------------

Over 40 years ago, one of my favorite video games was Rally-X.  https://en.wikipedia.org/wiki/Rally-X  Little did I know it at the time (because I had yet to take my first O.R. course),  as a player, you were solving a prize-collecting traveling salesman problem in order to maximize your score.  Maybe my fascination with that game is why I ended up with a career in optimization.

So aside from O.R. applications that can be applied in the design of games, there are O.R. applications for the players of games as well.

------------------------------
-Irv Lustig
Optimization Principal
Princeton Consultants
------------------------------

I used to work in a videogame development and publishing industry as a PR manager and community manager about 15 years ago when I still lived in Russia. I stayed in touch with my former colleagues for a while after I left.

I did not encounter any OR/MS algorithm applications when I worked there, unless you count linear algebra in 3D visualization and physics equations. But I could see a lot. For MMORPGs, especially before cloud computing, capacity planning was a risk: if the marketing was too successful, players would "overcrowd" the server, leading to performance issues, and the players would stop playing. I don't know how it works now that AWS is around.

Switching from subscriptions to microtransactions brought a lot of interesting challenges. How much advantage should the paid content give a player in the game? What is the right balance between paying and non-paying players? Then, especially when most of your revenue comes from just a few players (which happens more often than the general public likes to think), how do you handle the situations when  a revenue-driving player is cheating or is abusive towards other players?

For the publisher, since the costs are fixed per game developed, and the revenue is proportional to sales, and there is only so many players in the market who have that many days during the Christmas break, selecting the right project portfolio and making decision when to terminate a project that is not going to be worth it are important.

For the game design, in MMORPG there was often a problem that one clan establishes a dominance on a server, creating the situation like in "Make Love, not Warcraft" South Park episode. Since the reward for winning a battle is getting control over the resources that would help you win more battles, this is what to be expected. How do you prevent it?

Ironically, when it comes to game theory, cannot see many applications in the games I used to play or work with, unless it is a behavioral game theory.

------------------------------
Yulia Vorotyntseva
Saint Louis University
St Louis MO
------------------------------

Hi @Xiao Lei ! Alas, I am notoriously TERRIBLE at video games (yes, even Mario Kart). However, I would love to explore this topic on an upcoming episode of the INFORMS podcast, Resoundingly Human. Let me know if you would be interested in an interview!

------------------------------
Ashley Kilgore
Manager, Communications
INFORMS
Catonsville MD
------------------------------