Source-linked AI summary
Characterizing Open-Source Video Games from a Software Engineering Perspective
Henrique Mandelli Canella, Yohan Duarte, Cristiano Politowski, Vinicius Durelli, Andre Takeshi Endo
TL;DR
Game-specific engineering has limited comparable code-level evidence, while open-source repositories expose artifacts for empirical study. This paper analyzes 491 itch.io games and finds concentrated genres, largely dormant repositories, common Godot and Unity use, and little automated testing.
Problem
Comparable code-level evidence on game-specific engineering is limited because closed repositories restrict access to implementation details and development histories.
Source: §1 · p. 1Method
The study characterizes 491 open-source games by analyzing repository metadata, implementation and assets, testing artifacts, and relationships among these dimensions.
Source: §1 · pp. 1–2Results
Open-source games concentrate in Puzzle, Action, and Platformer genres, repositories are predominantly dormant, Godot and Unity are common, and only 36 projects include automated tests.
Source: §1 · p. 2Takeaways & Limitations
The dataset records activity, engine use, asset counts, and testing signals, providing a basis for follow-up studies and tooling.
Source: §1 · p. 2Because the findings derive from open-source games on itch.io, they may not generalize to closed-source games or other platforms.
Source: §4 · p. 4
Abstract
from arXiv · showhide
Video games hold a prominent role in the entertainment industry, which constitutes a multibillion-dollar global market. Within this ecosystem, independent studios, small teams, and solo developers operate with constrained budgets while making their games available through platforms such as itch.io, where many projects are also available as open-source. Where even if the project is fully released, other developers can still have access to the development process through the source code, helping other developers to develop their own games. These open-source games also constitute a valuable corpus for empirical software engineering research, offering transparent access to development artifacts that are typically inaccessible in commercial projects. This study characterizes open-source video games from a Software Engineering standpoint, focusing on their genres, popularity, repository activity, implementation characteristics, and testing practices. To this end, we extracted metadata from 491 video games from itch.io, published as open-source projects in GitHub. The results indicate that open-source games predominantly fall within limited genres (Puzzle, Action, and Platformer), that player ratings, when available, tend to be high, bringing a perception that the community receives well those kind of games, that repositories are largely dormant or minimally maintained, that the majority of these projects adopt either Godot or Unity as their game engine, and that documentation of testing-related practices is seldom included. The results offer insights into the development of open-source video games and may contribute to advancing software engineering practices and tools designed for small teams and independent game developers.
1 Introduction
The paper addresses limited code-level evidence about game-specific engineering by characterizing 491 open-source games across genres, ratings, repository activity, implementation, and testing. It reports concentrated genres, mostly dormant repositories, frequent Godot and Unity use, and limited automated testing.
1 Introduction
- Game development spans iterative work across design, code, assets, and live operations, while teams manage performance budgets, platform compliance, and frequent content updates.The section situates this work in an industry described as generating US$93 billion in 2023.Source: §1 · p. 1
- Open-source repositories expose code, issues, pull requests, build scripts, tests, and release notes, enabling reproducible analysis of implementation choices and project evolution.Prior work is described as limited by closed repositories that obscure implementation details and histories.Source: §1 · p. 1
- The study characterizes 491 open-source games across genres and ratings, repository activity, implementation, testing, and relationships among these dimensions.Its dataset records activity, engine use, asset counts, and testing and documentation signals, with entries manually checked for relevance and completeness.Source: §1 · pp. 1–2
- The research questions examine genre distributions and ratings, repository activity, languages and tools, project size, testing practices, and associations between test files, activity, and perceived quality.The questions are motivated by gaps in quantitative mining evidence on testing adoption and broader characterizations linking genres, activity, and quality practices.Source: §1 · pp. 1–2
- Open-source games concentrate in Puzzle, Action, and Platformer genres, while repositories are predominantly dormant, Godot and Unity are common, and automated testing is limited.Player ratings are generally high when available; only 36 projects include automated tests, and testing has no significant correlation with user-perceived quality.Source: §1 · p. 2
2 Related Work
Prior research examines game engineering through reviews, practitioner studies, and mining of development artifacts. The paper positions its contribution as a broader characterization of open-source games that links genres, repository activity, implementation, and testing.
2 Related Work
- Reviews synthesize game-engineering research, including development practices, testing, and automation; Engström et al. identify a gap between software engineering and creative perspectives.Berg Marklund et al. describe planning difficulties tied to subjective requirements such as fun and player experience.Source: §2 · p. 2
- Practitioner surveys and interviews find that game development prioritizes creativity, manual testing, user experience, and fun over technical reliability.The section also covers research on emotional labor, success factors, and practices under time pressure.Source: §2 · p. 2
- Mining studies examine bugs, reproducibility, compatibility, and playability, but often do not link broader characteristics such as genres and repository activity.The paper positions its work as extending this evidence with a multi-dimensional characterization and quantified relationships, including testing and activity.Source: §2 · p. 2
3 Study Setup
The study builds a curated corpus from itch.io projects linked to GitHub repositories, then combines repository metrics, source inspection, asset counts, and reviews of testing artifacts. Manual validation and filtering produce a final dataset of 491 projects.
3 Study Setup
- The initial dataset from Di Sipio et al. contains 10,548 itch.io projects with ratings, genres, development tools, tags, and GitHub links.The study uses this dataset as its starting point for identifying open-source game repositories.Source: §3 · p. 2
- Researchers collected repository age, forks, stars, issue counts, issue close times, and contributor counts through the GitHub API.These metrics were merged with the initial project metadata.Source: §3 · pp. 2–3
- The study flowchart reports that the study started with 10,548 itch.io projects [6], which were filtered through validation and quality checks, leaving a final corpus of 491 projects.Source: Figure 1 · p. 3
- Manual screening checked repositories with one to three commits for versioned game source and retained substantive initial imports, while excluding empty, archive-only, and non-game repositories.Nine repositories met at least one exclusion condition; the final quality-filtered dataset contains 491 projects.Source: §3 · p. 3
- Engine detection scans repository files and directories for technology-specific signatures, with ambiguous assignments marked as mixed for review.The signature catalogue includes patterns such as Unity project files, Godot's project.godot, GameMaker's .yyp, and Construct's .c3p.Source: §3 · p. 3
- The study counts assets by file extension and identifies automated tests through keyword screening followed by manual inspection of source files and configuration.Testing keyword screening flagged 111 projects, of which 36 were verified as having automated tests; documentation analysis filtered Markdown files and used a local LLM to extract and categorize testing guidance, with manual verification.Source: §3 · pp. 3–4
4 Threats to Validity
The findings are based on open-source games from itch.io, so they do not establish how results generalize to closed-source games or other platforms.
4 Threats to Validity
- The itch.io open-source sample limits generalization to closed-source games and other platforms.The authors note possible metadata incompleteness and describe manual validation of automated measures and researcher review of flagged repositories.Source: §4 · p. 4
5 Results Analysis
Across 491 open-source games, genres and implementation choices are concentrated in several common patterns, while repository activity and documented testing are often limited. Repositories with test files show higher commit activity, but test presence is not associated with higher or lower player ratings in this sample.
5.1 RQ1: What is the distribution of genres among open-source games?
- Puzzle (103), Action (97), and Platformer (69) are the most common genres, and most projects carry a single genre label.Of the games, 58.9% lack ratings; when ratings are available, genre means are generally high, though some high means come from only two titles.Source: §5.1 · p. 4, §5.1.1 · p. 4, Table 1 · p. 4
5.2 RQ2: How active are the GitHub repositories of open-source video game projects in terms of development and community engagement?
- Repository activity is highly skewed: 79 projects have five or fewer commits, while a small subset accounts for high development and community activity.The mean commit count is 333.38, and more than 300 projects have no recorded issues.Source: §5.2 · p. 5
5.3 RQ3: What languages, engines/frameworks, and build/CI tools do open-source games use, and how large/complex are they?
- Godot and Unity are the most frequently detected engines, while project assets are predominantly images and developers use varied toolchains.Automated detection identifies an engine in 260 projects; images average 152.39 files per project, compared with 6.82 3D assets.Source: §5.3 · p. 6
5.4 RQ4: To what extent do open-source game projects document or adopt testing-related practices?
- Only 36 projects (7.33%) include automated tests, and documentation of testing practices, bug reporting, troubleshooting, and contributor testing guidance is uncommon.Forty projects document testing practices (8.15%), while 11 reference troubleshooting activities (2.24%).Source: §5.4 · p. 6
5.5 RQ5: To what extent is the presence of test files associated with repository activity and user-perceived quality in open-source game projects?
- With age, stars, and forks held constant, repositories with tests have a 3.34× higher expected commit rate, but test presence is not significantly associated with user ratings.The commit-rate association has a 95% CI of [1.98, 5.64] and p<0.001; the rating coefficient is −0.09 (95% CI [−0.39, 0.21]; p=0.545).Source: §5.5 · pp. 6–7
6 Discussion
The discussion frames the corpus as small-team “Kei Games,” where tests are rare but associated with higher repository activity, while player ratings do not track test presence. It connects sparse testing and contributor guidance to practical recommendations, while noting limits from metadata and sampling.
Who builds these games
- The corpus’s focused genres, 2D-heavy assets, and Godot/Unity use fit the “Kei Games” profile of individual developers or small teams pursuing focused scopes and short feedback loops.The dataset joins genres, repository activity, engines, assets, and testing signals for replication without requiring source distribution; sampling favors public and smaller projects, while metadata has gaps such as missing ratings and incomplete engine tags.Source: §6 · pp. 7–8
Testing and sustainability
- Repositories with automated tests have a 3.34× higher expected commit rate, but the association does not establish that tests cause sustained maintenance.Only 36 projects (7.33%) include automated tests; the authors suggest tests may proxy engineering discipline or co-evolve with ongoing work.Source: §6 · p. 7
Why tests do not move ratings
- Test presence is not associated with player ratings; sparse ratings, ceiling effects, and player attention to content and feel limit what ratings reveal about backend quality.The discussion notes that 58.9% of projects lack ratings and mean ratings exceed 4.0; metadata alone cannot fully answer whether players care about software quality.Source: §6 · p. 7
Documentation and contributor pathways
- Testing guidance is uncommon: 8.15% document testing and 4.28% provide testing guidance, prompting suggestions for concise contributor instructions and bug-reporting support.The discussion proposes CONTRIBUTING.md reproduction steps, issue templates, a short test-running guide, and a crash/diagnostics checklist.Source: §6 · p. 7
Implications for practice
- For practice, the authors recommend a small test core, headless scene-load smoke tests, replayed input traces, and GitHub Actions checks on pull requests.They argue these measures can raise confidence, reduce bug-report back-and-forth, and make outside contributions safer to merge.Source: §6 · p. 7
Implications for research
- For research, they propose engine-specific runners, asset-aware regression checks, privacy-preserving crash bundles, engine-smell catalogues, and longitudinal studies of test introduction.Proposed outcomes include project survival, release cadence, issue-closing time, and regression rates.Source: §6 · p. 7
7 Concluding Remarks
The study’s 491-game analysis finds focused genres, reliance on Godot and Unity, image-heavy 2D assets, and rare automated testing. Tests correlate with repository activity but not player-perceived quality; the authors caution that itch.io sampling and sparse ratings limit commercial parallels.
7 Concluding Remarks
- Godot (24.4%) and Unity (17.1%) lead among engines, while automated tests appear in only 7.33% of projects and correlate with a 3.34× commit rate but not perceived quality.The analysis also reports focused genres such as Puzzle, Action, and Platformer, and 2D-centric, image-heavy assets.Source: §7 · p. 8
- The findings provide code-level data that confirm survey evidence about open-source game development patterns.Source: §7 · p. 8
- Projects with test files show higher repository activity, but test presence does not appear to yield measurable improvements in player-perceived quality.Source: §7 · p. 8
- Itch.io sampling bias and rating sparsity may understate commercial parallels, motivating broader platform coverage and qualitative studies of sustainable, popular projects.Source: §7 · p. 8