What Open Source Software Is and Why It Matters
Open source software is software whose source code is made publicly available under a licence that permits anyone to view, use, modify, and distribute it. The source code availability distinguishes open source from proprietary software, whose source code is kept secret and whose use is governed by licences that restrict what users can do with it. The open source licence — of which there are dozens of variants ranging from the permissive MIT and Apache licences (which allow nearly any use including incorporation into proprietary products) to the copyleft GPL licences (which require that derivatives also be released under the same open source terms) — defines the specific rights and obligations that apply to users, contributors, and distributors of the software.
The open source software scope that most surprises people who associate open source primarily with developer tools: the majority of the infrastructure that the modern internet runs on is open source. The Linux operating system runs on approximately 96% of the world’s top one million web servers; the Apache and Nginx web servers that serve the majority of web traffic are open source; the MySQL and PostgreSQL databases that store the majority of web application data are open source; the Kubernetes container orchestration platform that manages most cloud-native workloads is open source; the TensorFlow and PyTorch frameworks that most AI and machine learning work is built on are open source. The technology industry has bet its infrastructure on open source software, and the compounding contribution of thousands of developers to these shared foundations has produced software quality and capability that no individual company could have built.
How the Open Source Economy Works
The open source economic model that most people who use open source without contributing to it do not fully appreciate: the open source software they use for free is produced by developers whose time is paid for by a combination of volunteer contribution, employer contribution (companies that allow employees to contribute to open source as part of their job because the company benefits from the improvements), and commercial support models (companies that build businesses around providing enterprise support, professional services, hosted versions, or commercial features on top of open source foundations — the models used by Red Hat, Elastic, HashiCorp, and many others).
The open source sustainability challenge that affects many important projects: the free rider problem, where organisations that use open source software extensively and profit from its quality do not contribute financial support or development effort to the project. The Linux kernel and other foundational projects with strong corporate sponsor bases have largely solved this problem; the smaller utility libraries and tools that much commercial software depends on often have a single maintainer who is effectively providing a public service without compensation, creating a fragility that the Log4Shell vulnerability exposed dramatically in 2021. The open source funding initiatives (Open Collective, GitHub Sponsors, Tidelift) are attempting to address this sustainability gap, but the problem remains a significant risk in the open source ecosystem.
How Businesses Use Open Source
The open source usage patterns that most businesses follow without necessarily describing their activities as open source strategy: using open source databases, web servers, container platforms, and programming language runtimes as the infrastructure layer of their technology stack (where the open source option is often technically superior to and dramatically cheaper than proprietary alternatives), using open source libraries and frameworks within their applications (which is nearly universal in modern software development — a typical web application may include hundreds of open source dependencies), and deploying open source AI models (which has become increasingly relevant as foundation models like Llama, Mistral, and Falcon have become capable enough for many production AI applications).
The open source licence compliance consideration that most businesses adopt without implementing: the actual management of open source licence obligations. The organisation that uses open source software is typically subject to the licence terms that govern that software — which may require attribution (crediting the open source project in the application), may require disclosure of modifications (under copyleft licences that require sharing modifications), or may prohibit use in certain contexts. The software composition analysis (SCA) tools that scan codebases for open source dependencies and identify their licences and known vulnerabilities have become standard in mature software development organisations, but many smaller organisations are not systematically managing their open source licence obligations.
Contributing to Open Source: The Career and Competitive Advantage
The open source contribution benefits for individual developers that most motivate contribution beyond the altruistic motivation: the portfolio evidence that public contributions to recognised open source projects provide for employers evaluating technical skills (the GitHub profile with meaningful contributions to well-known projects is a credible demonstration of capability that resumes and certificates cannot match), the learning opportunity that contributing to production-quality codebases maintained by experienced developers provides (the code review feedback from open source maintainers is often the most detailed and highest-quality technical feedback that a developer receives), and the professional network that open source communities provide (which spans companies, countries, and career stages in ways that individual employer networks do not).
The open source contribution benefit for companies that most convincingly justifies allocating engineering time to contribution: the influence over the direction of software that company operations depend on. The company that contributes to the open source database, container platform, or networking library that its infrastructure runs on can advocate for and contribute the features, fixes, and improvements that most benefit its specific use case — rather than depending on the priorities of a proprietary vendor whose roadmap may not align with the company’s needs. This influence over shared infrastructure is particularly valuable for infrastructure-scale problems where the open source alternative is technically competitive with proprietary options.
The Open Source AI Revolution
The open source AI development that has most changed the competitive landscape for AI capabilities: the release of capable open-weight language models (models whose trained parameters are publicly available, allowing organisations to run the model on their own infrastructure rather than through an API) by Meta (the Llama model family), Mistral AI, and others. The availability of these models has enabled organisations to run capable AI locally — without the data privacy concerns of sending information to a third-party API, without the per-token cost of commercial API usage, and with the ability to fine-tune the model on specific domain data to improve performance for specific use cases.
The open source versus proprietary AI comparison that most accurately reflects the current capability gap: for general-purpose tasks, the most capable proprietary models (GPT-4o, Claude 3.5 Sonnet, Gemini Ultra) still outperform the most capable open-weight models on most benchmarks, but the gap has narrowed significantly and continues to close with each successive generation of open-weight releases. For specific domain tasks where fine-tuning is applied and where the proprietary model’s general capability advantage is less relevant, open-weight models fine-tuned on domain-specific data can match or exceed proprietary model performance. The open-weight AI ecosystem is maturing rapidly, and the organisations that develop the expertise to deploy and fine-tune open-weight models are building capabilities that their competitors who depend exclusively on proprietary model APIs will not easily match.
