Socio-technical ecosystems for scientific software development

该奖项是根据2009年美国复苏和再投资法案（公法111-5）资助的。本研究以使用开放科学网格的多个虚拟组织为样本，考察了协作科学研究中软件开发、共同创造和共享的实践。从仪器的配置和控制，到统计分析、仿真和可视化，几乎每一个产生科学成果的工作流程都涉及到软件。创建和维护软件是包括科学和工程虚拟组织在内的科学实验室中的一项重要活动。成功取决于克服协作的挑战和分布式软件开发的复杂协调问题。一个关键的挑战是，虽然科学是合作的，但它不是无私的；科学家在整个职业生涯中都面临着竞争压力，保持对他们编写的软件的持续控制可能与他们的职业成功有关。因此，像传统的开放源码项目这样的通用软件开发模型可能不太适合科学软件，因为从开放软件生态系统（如Eclipse）中提取的替代模型支持同样是竞争对手的团队之间的协作。如果对协调和激励挑战的复杂问题没有足够的理解，科学资助机构就不能在适当的政策上提供清晰的指导，以促进和维持协作科学中有效的软件开发。这项研究扩展了分布式软件开发中体系结构一致性的理论，并假设了技术体系结构在建立协调、合作和竞争发生的框架方面的核心作用。特别地，我们假设了一个架构的启用角色，它适当地支持协作，特别是当激励随着时间的推移而变化时。为了检验现有的实践和这些假设，研究小组将在三年内通过访谈、开放科学网格的软件使用数据、软件源代码和开发存储库收集数据。研究结果将有助于理解科学家如何创建和分享他们的软件。这样的理解将有助于改善科学研究中的合作实践，并帮助科学资助机构通过资助科学研究中的软件开发的政策、指导方针和教育来支持科学家。

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). This research examines practices of software development, co-creation and sharing in collaborative scientific research based on a sample of multiple virtual organizations using the Open Science Grid. Almost every workflow that generates scientific results involves software, from configuration and control of instruments, to statistical analysis, simulation and visualization. Creating and maintaining software is a significant activity in scientific laboratories, including science and engineering virtual organizations. Success depends upon overcoming both the challenges of collaboration and the complex coordination problems of distributed software development.A key challenge is that while science is collaborative, it is not selfless; scientists face competitive pressures throughout their careers and maintaining ongoing control over the software they write may be related to their career success. Consequently, common software development models like traditional open source projects may be less appropriate for scientific software as alternative models drawn from open software ecosystems, such as Eclipse, that support collaboration between groups that are also competitors. Without adequate understanding of the complex issues of both coordination and incentive challenges, scientific funding agencies cannot provide clear guidance on appropriate policy to promote and sustain effective software development in collaborative science.This research extends theories of architectural alignment in distributed software development and hypothesizes a central role for technical architectures in establishing the framework within which coordination, cooperation, and competition take place. In particular we hypothesize an enabling role for architectures that are appropriately aligned to support collaboration, particularly as incentives change over time. To examine existing practices and these hypotheses the research team will collect data through interviews, the Open Science Grid's software usage data, software source code and development repositories over three years. The results will build understanding of how scientists create and share their software. Such understanding will help to improve collaboration practices in scientific research as well as helping science-funding bodies support scientists through policy, guidelines and education on software development in funded scientific research.