The open source AI fight is just the 1980s software fight on repeat

David Siegel, co-founder of Two Sigma, spent two years arguing with Richard Stallman about open source software in the 1980s. The debate then was whether keeping code proprietary or sharing it openly would produce better outcomes. Stallman won that argument. Now Siegel is watching the same fight happen with AI models. The parallels are striking. In the 1980s, software companies wanted to lock down their code. They argued that proprietary control was necessary for quality, for safety, for competitive advantage. Stallman and the free software movement said the opposite: transparency produces better software because more people can inspect it, fix it, fork it. Decades later, open source powers most of the internet. The AI debate uses identical language. Closed model companies say regulation requires keeping weights proprietary. They claim open models are dangerous because anyone can fine-tune them for harm. Siegel points out this is the same fear-based argument that lost in software. Open source did not collapse the industry. It created Linux, Apache, PostgreSQL, the tools that run critical infrastructure today. The difference this time is the stakes feel higher. A bug in a text editor is annoying. A model that generates convincing misinformation or automates phishing at scale is a societal problem. But Siegel argues the solution is still transparency, not lockdown. If a model has a bias or vulnerability, you want thousands of researchers able to find and fix it, not a single vendor deciding what gets patched. What I find interesting is how short institutional memory is. The open source wars were forty years ago, not ancient history. The people who lived through them are still working. Yet the conversation around AI models acts like this is a novel dilemma. It is not. We already ran this experiment with software. Open won.

---

Source: I argued with the father of open source for 2 years Now the AI fight is the same