aalto1 untyped-item.component.html
Continuous experimentation engine for video games: A constructive research study
Loading...
URL
Journal Title
Journal ISSN
Volume Title
School of Science |
Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
Department
Major/Subject
Mcode
Degree programme
Language
en
Pages
69
Series
Abstract
This thesis explores how continuous experimentation (CE), a practice widely adopted in modern software development, can be systematically applied in the context of video game development as video games evolve into continuously updated services. The main goal is not only to adjust game parameters, but also to optimize the experimentation process itself by exploring the parameter space strategically to find the optimal gameplay settings rather than experimenting with only a limited number of predefined game variants. This goal is accomplished by designing, implementing, and evaluating a modular CE Engine. The engine applies iterative parameter optimization using multi-armed bandit algorithms and combines objective performance metrics with subjective player feedback via a hybrid reward function. A proof-of-concept was created using a Flappy Bird-style game, where five gameplay parameters were modified during player sessions. A pilot user study with 10 participants and 150 gameplay sessions validated the engine's technical feasibility and offered insights into how players react to adaptive difficulty. The results show that the engine follows key principles of CE, including hypothesis-driven development, learning systematically from data, and rapid feedback loops. This research offers a practical contribution with a tool for game developers who want data-driven methods for parameter tuning. It also offers a theoretical contribution by providing insights into adjusting continuous experimentation frameworks for real-time software. The modular design allows for adding more optimization algorithms while requiring little integration effort. Although this implementation mainly focused on session-based game parameter updates and small-scale validation, the foundations from this thesis provide a basis for future research into live parameter adaptation and large-scale experimentation in game development.
Description
Supervisor
Fagerholm, FabianThesis advisor
Baumeister, HubertFagerholm, Fabian