Michael Rowe

Open source software

note5 min read

The value of open source lies in the control its makers give up

Open source software publishes its source code under a licence that lets anyone inspect it, change it, and pass it on. The practical consequence for an institution is that the software can outlive its vendor, be adapted to local needs, and be checked rather than trusted.

Open source software

One-sentence definition: Open source software is software distributed with its source code under a licence that grants the rights to study, modify, and redistribute it (Open Source Initiative, 2007).

Most software is closed: you get the compiled1 program and a licence to run it, and the code that makes it work stays with the company. Open source inverts that: the code is public, and the licence — GPL, MIT, Apache, and a handful of others — says what you may do with it. Raymond’s The cathedral and the bazaar (2000) is the classic account of how a community working in the open can produce better software than a firm working in private; Stallman’s essays (2002) make the older, more political case that users have a right to control the software they depend on.

The infrastructure behind almost everything runs on open source: Linux on servers, Python and R for research, Git for version control, the compilers and libraries behind commercial products. Moodle, a virtual learning environment, is open source, and so is Zotero, the reference manager. This site is built with Quartz, an open source static site generator,2 and edited in Obsidian, which isn’t.

Exit, adaptation, and audit

  • The exit is real. When a vendor is bought, raises prices, or discontinues a product, an institution running closed software has to migrate. With open source, the code stays available; someone else can maintain it, or you can.
  • You can change it. A learning platform that nearly does what a programme needs can be made to do it, and the change can be shared. Moodle’s plugin directory works this way: an extension one institution writes for its own teaching is there for others to install. Avila et al. (2016) built an e-portfolio for undergraduate medical education on WordPress because commercial systems were costly and didn’t fit the institution’s needs. Closed software can be configured; it can’t be fixed.
  • You can look. An assessment tool that makes decisions about students can be audited when its code is open. That matters more as software starts to include AI components, and it’s why the argument over what “open source AI” should mean (Wiley, 2024) is worth following: a model released with weights3 but without training data is open in a narrower sense than the term used to carry.

What open doesn’t give you

Open source isn’t free of cost. Someone has to host, maintain, secure, and update it, and if that’s an already-stretched IT team the closed product with a support contract can be the cheaper choice. Open code doesn’t guarantee a living project either. Belshaw (2022) offers a useful typology of how open source communities grow and stall; the healthy ones have governance as well as code, and governance is the part institutions underestimate when they adopt open source software.

Openness of code also says nothing about openness of data: a closed source tool can still let you export everything, and an open source one can still lock your material in a bespoke format, so when you choose a tool it pays to ask about the code and the data separately.

Sources

Footnotes

  1. Compiled — turned from the human-readable source code a programmer writes into machine instructions a computer runs directly. A compiled program works, but reading it to see how it works is close to impossible, which is why having the source matters. Wikipedia ↩

  2. Static site generator — a program that turns a folder of text files into a finished website in one go, so there’s no database or server-side software to run. Quartz takes the Markdown notes this site is written in and produces the pages you’re reading. Wikipedia ↩

  3. Weights — the billions of numbers a language model learns during training, which together encode what it does. Releasing the weights lets anyone run the model; without the training data, nobody can see or reproduce how it came to behave as it does. Wikipedia ↩

Continue reading

  • Open access licensing

    Open licences, mostly Creative Commons, state in advance what others may do with a work — read it, share it, adapt it, build on it — so nobody has to ask. The Creative Commons licence on teaching materials decides whether a colleague can adapt a resource for their own programme or has to rebuild it from scratch.

  • Context sovereignty

    What is context sovereignty? The idea that people should own and author the context they bring to AI (their values, commitments, and ways of making sense of things) so that AI adapts to their thinking. How it differs from data sovereignty and model memory, its three principles, and what it means for students and educators.

  • GitHub

    GitHub is a website that hosts git repositories online and adds the tools people need to work on them together, including issues for discussion, pull requests for proposing and reviewing changes, and automation that can publish a project as a website. It was built for software developers, but anything kept as text files can live there, including teaching materials and scholarship.