一句话：首次大规模实证刻画公共生成艺术代码仓库的实践者、技术模式、常用工具与可复现性。
生成艺术家数十年来以源代码为创作媒介，大量作品代码公开在开放仓库中，但此前缺乏对其技术属性的实证研究。本文提出一项基于 Software Heritage Archive 的大规模实证分析，系统收集公共生成艺术仓库，刻画实践者群体的多样性、常见技术模式与艺术工具生态。为评估可复现性，作者选取代表性项目尝试构建与执行，并对常见失败原因进行分类。该研究首次在规模上对生成艺术代码进行定量刻画，为软件艺术品的文档化、保存与文化遗产维护策略提供了实证依据。
Generative artists have been using source code as a creative medium for decades, and a large amount of this code is publicly available in open repositories. However, there is no empirical study characterizing the technical properties of the generative art code. In this work, we propose a large-scale empirical analysis of public generative art repositories sourced from the Software Heritage Archive. We characterize the diversity of practitioners, technical patterns, and commonly used artistic tools. In addition, we assess the reproducibility of generative artworks by attempting to build and execute a representative sample of projects and by categorizing common failure causes. Our study provides the first quantitative characterization of generative art code at scale and offers empirical evidence to support documentation and preservation strategies for software-based artworks, contributing to our cultural and software heritage.