Using probabilistic model checking to balance games

Kavanagh, William (2021) Using probabilistic model checking to balance games. PhD thesis, University of Glasgow.

Abstract

In this thesis, we consider problem areas in game development and use probabilistic model checking to address them. In particular, we address the problem of multiplayer game balancing and introduce an approach called Chained Strategy Generation (CSG). This technique uses model checking to generate synthetic player data representing a game-playing community moving between effective strategies. The results of CSG mimic the metagame, an ever-evolving state of play describing the players’ collective understanding of what strategies are effective. We expand upon CSG with optimality networks, a visualisation that compares game material and can be used to show that a game exhibits certain qualities necessary for balance.

We demonstrate our approach using a purpose-built mobile game (RPGLite). We initially balanced RPGLite using our technique and collected data from real world players via the mobile app. The application and its development are described in detail. The gathered data is then used to show that the model checking did lead to a well-balanced game. We compare the analysis performed from model checking to the gameplay data and refine the baseline qualities of a balanced game which model checking can be used to guarantee.

We show how the collected data via the mobile app can be used in conjunction with the prior model checking to calculate action-costs – the difference between the value of the action chosen and the best action available. We use action-costs to evaluate player skill and to consider other factors of the game.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QA Mathematics > QA76 Computer software
• Colleges/Schools: College of Science and Engineering > School of Computing Science
• Supervisor's Name: Miller, Professor Alice and Norman, Dr. Gethin
• Date of Award: 2021
• Depositing User: Theses Team
• Unique ID: glathesis:2021-82618
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 10 Jan 2022 15:55
• Last Modified: 08 Apr 2022 16:59
• Thesis DOI: 10.5525/gla.thesis.82618
• URI: https://theses.gla.ac.uk/id/eprint/82618

Related URLs

• Article DOI
• Enlighten Publications Record
• Enlighten Publications Record
• Enlighten Publications Record
• Article DOI
• Enlighten Publications Record
• Data DOI