To start my research on Project Seesaw, I first had to compile a list of potentially useful research papers and articles.
I looked for 30 relevant sources deciding their relevancy only by reading their titles.
Then, I read the abstract section of each paper and pointed the paper a relevancy score:
0 for irrelevant,
1 for slightly relevant,
2 for relevant, and
2+ for crucially relevant
|1||Automated tweaking of levels for casual creation of mobile games||Click Here||1|
|2||Evaluating Competitive Game Balance with Restricted Play||Click Here||2+|
|3||Real-time challenge balance in an RTS game using rtNEAT||Click Here||2|
|4||Understanding Game Balance with Quantitative Methods||Click Here||0 (duplicate #2)|
|5||Comparing Effects of Dynamic Difficulty Adjustment Systems on Video Game Experience||Click Here||1 (similar to #3)|
|6||Strategic Pattern Discovery in RTS-games for E-Sport with Sequential Pattern Mining||Click Here||2|
|7||Method for dynamically adjusting an interactive application such as a videogame based on continuing assessments of user capability||Click Here||0|
|8||On the Impact of Software Patching on Gameplay for the League of Legends Computer Game||Click Here||2+ (end result perspective)|
|9||Investigating the Impact of Game Features on Champion Usage in League of Legends||Click Here||2+ (end result perspective)|
|10||Dynamic difficulty adjustment on MOBA games||Click Here||1|
|11||Game balancing with ecosystem mechanism||Click Here||2 (criticizes current dynamic balancing tools)|
|12||Using coevolution to understand and validate game balance in continuous games||Click Here||2|
|13||Automatic computer game balancing: a reinforcement learning approach||Click Here||1 (dynamic difficulty ajustment)|
|14||Target assistance for subtly balancing competitive play||Click Here||1 (dynamic difficulty ajustment)|
|15||Establishing competitive domination cycles for peer-to-peer game combat||Click Here||2+ (more theoretical)|
|16||A Competitive Markov Approach to the Optimal Combat Strategies of On-Line Action Role-Playing Game Using Evolutionary Algorithms||Click Here||2+ (coevolutionary + on-line)|
|17||An Argument for Game Balance: Improving Student Engagement by Matching Difficulty Level with Learner Readiness||Click Here||0|
|18||Geometric Analysis of Maps in Real-Time Strategy Games: Measuring Map Quality in a Competitive Setting||Click Here||2+ (map balance perspective)|
|19||Resolving Simultaneity Bias: Using Features to Estimate Causal Effects in Competitive Games||Click Here||1|
|20||Automatic Playtesting for Game Parameter Tuning via Active Learning||Click Here||2 (simple, explains the fundamental idea of automated playtesting for game balance)|
|21||Game Theoretic and Machine Learning Techniques for Balancing Games||Click Here||2+ (thesis, easy to read)|
|22||Debugging Game History Section 21. Game Balance||Click Here||2 (good terminology explanation)|
|23||Mathematical balance metrics in competitive multiplayer games||Click Here||1 (references other papers already in the list)|
|24||Orthogonal analysis of StarCraft II for game balance||Click Here||2|
|25||Multimodality and the Competitive Metagame: Exploring Issues of Balance in Multimodal Game Environments||Click Here||2+ (matches my research goal: competitive and, online)|
|26||Dynamic Difficulty Adjustment in Computer Games Through Real-Time Anxiety-Based Affective Feedback||Click Here||0|
|27||The effectiveness of adaptive difficulty adjustments on students’ motivation and learning in an educational computer game||Click Here||0|
|28||Dungeons & Replicants: Automated Game Balancing via Deep Player Behavior Modeling||Click Here||2|
|29||Rarity and Power: Balance in Collectible Object Games||Click Here||2+ (non-videogame perspective|
|30||Game team balancing by using particle swarm optimization||Click Here||1 (similar to other papers)|
|31||Solving the Balance Problem of On-Line Role-Playing Games Using Evolutionary Algorithms||Click Here||1 (similar to other papers)|
Reading the 5 most relevant papers according to their relevancy score
2: Evaluating Competitive Game Balance with Restricted Play
Balancing is difficult and time-consuming. Expensive development tuning cycles for subtle adjustments where each tweak involves substantial playtesting and designer intuition. Small tweaks can have unexpected consequences, so it is hard to test how tuning affects the gameplay. Current AI technologies cannot keep up with the diversity of viable strategies in complex games. Suggests reducing complex gameplay to restricted measures that are then used to test parts of the game.
8: On the Impact of Software Patching on Gameplay for the League of Legends Computer Game
LoL has been patched over 160 times since its release in 2009 with an average of about 1.5 patch files per month. LoL devs have access to millions of games, over 450,000 between Seasons 4-6. Analyzes how Riot patches the game and how these patches effect game length, champion ban/pick rate, item usage rate etc.
15: Establishing Competitive Domination Cycles for Peer-to-peer Game Combat
The paper investigates if domination loops can be determined using agents in a rock, scissors, paper style approach. The paper concludes that domination loops do exists in the context of single unit combat scenarios. Their models and methods are capable of finding frequent examples of units in a cyclic dominance relationship.
16: A Competitive Markov Approach to the Optimal Combat Strategies of On-Line Action Role-Playing Game Using Evolutionary Algorithms
Divides combat strategies into three distinct classes, Strategy of Motion, Strategy of Attacking Occasion, and Strategy of Using Skill. They analyze such strategies of a basic game model in which the combat is modeled by the discrete competitive Markov decision process. They focus on how SUS (Strategy of Using Skill) pairs can be balanced using coevolution and CCEA.
20: Automatic Playtesting for Game Parameter Tuning via Active Learning
Playtesting for finetuning a game is expensive. The paper shows that active learning techniques can formalize and automate a subset of playtesting goals. They argue that parameter tuning can be automated to reduce the cost of playtesting to achieve a design goal by intelligently picking designs to test.