Security considerations in the open source software ecosystem

Abstract

Open source software plays an important role in the software supply chain, allowing stakeholders to utilize open source components as building blocks in their software, tooling, and infrastructure. But relying on the open source ecosystem introduces unique challenges, both in terms of security and trust, as well as in terms of supply chain reliability. In this dissertation, I investigate approaches, considerations, and encountered challenges of stakeholders in the context of security, privacy, and trustworthiness of the open source software supply chain. Overall, my research aims to empower and support software experts with the knowledge and resources necessary to achieve a more secure and trustworthy open source software ecosystem. In the first part of this dissertation, I describe a research study investigating the security and trust practices in open source projects by interviewing 27 owners, maintainers, and contributors from a diverse set of projects to explore their behind-the-scenes processes, guidance and policies, incident handling, and encountered challenges, finding that participants’ projects are highly diverse in terms of their deployed security measures and trust processes, as well as their underlying motivations. More on the consumer side of the open source software supply chain, I investigated the use of open source components in industry projects by interviewing 25 software developers, architects, and engineers to understand their projects’ processes, decisions, and considerations in the context of external open source code, finding that open source components play an important role in many of the industry projects, and that most projects have some form of company policy or best practice for including external code. On the side of end-user focused software, I present a study investigating the use of software obfuscation in Android applications, which is a recommended practice to protect against plagiarism and repackaging. The study leveraged a multi-pronged approach including a large-scale measurement, a developer survey, and a programming experiment, finding that only 24.92% of apps are obfuscated by their developer, that developers do not fear theft of their own apps, and have difficulties obfuscating their own apps. Lastly, to involve end users themselves, I describe a survey with 200 users of cloud office suites to investigate their security and privacy perceptions and expectations, with findings suggesting that users are generally aware of basic security implications, but lack technical knowledge for envisioning some threat models. The key findings of this dissertation include that open source projects have highly diverse security measures, trust processes, and underlying motivations. That the projects’ security and trust needs are likely best met in ways that consider their individual strengths, limitations, and project stage, especially for smaller projects with limited access to resources. That open source components play an important role in industry projects, and that those projects often have some form of company policy or best practice for including external code, but developers wish for more resources to better audit included components. This dissertation emphasizes the importance of collaboration and shared responsibility in building and maintaining the open source software ecosystem, with developers, maintainers, end users, researchers, and other stakeholders alike ensuring that the ecosystem remains a secure, trustworthy, and healthy resource for everyone to rely on.