Luis Villa: "What ride?": where libre software could go after LLMs
A recent poll reveals stark differences in coding practices across social platforms, with 65.1% of Mastodon users writing more than 20% of code by hand compared to just 8.4% on Twitter.
Developer Armin Ronacher conducted a revealing poll across three social platforms, asking users if they still write more than 20% of code by hand. The results showed dramatic variation: 8.4% on Twitter, 26.3% on Bluesky, and 65.1% on Mastodon. This discrepancy reflects deeper divides within the free and open-source software (FOSS) community as large language models (LLMs) transform development practices.
True user freedom
The most significant opportunity lies in making Freedom 1—the freedom to study and modify software—truly practical for most users. For forty years, this freedom has existed in principle but required technical expertise to implement. Now, with AI assistance, users can customize complex software like GNOME and social media clients like phanpy without deep coding knowledge, realizing the long-standing promise of user empowerment.
Best practices become reality
Many FOSS best practices previously considered too resource-intensive for volunteers—such as test-driven development, accessibility testing, and formal methods—are becoming feasible with AI assistance. This technological shift allows for higher-quality software development that was previously impractical due to time and resource constraints.
Including non-coders
AI is changing collaboration dynamics by enabling participation from non-coders including designers, writers, and domain experts. However, there's concern that open-source projects may lose their structural advantage of incorporating user feedback if proprietary vendors develop faster tools for synthesizing user input into code.
If we don't get there first, with open processes, we lose one of the few structural advantages open collaboration has left.
Countering proprietary giants
The FOSS community benefits from having clear antagonists, and today's tech giants represent more concentrated, capital-intensive threats than historical opponents like Microsoft. The author suggests engaging with proprietary tools strategically, as the community has done before with projects like Firefox on Windows, to advance open-source goals.
Replacing unsafe languages
AI makes it feasible to rewrite decades of legacy C and C++ code in safer modern languages, addressing long-standing security vulnerabilities. Once considered impossible even over millennia, this technical debt can now be systematically addressed, improving user security and creating opportunities for a new generation of programmers to work with more modern codebases.
Related
