My comment on Hacker News about “Developer tools must be open source (exe.dev)”.
One of the classic arguments for end-users supporting open-source software has always been the freedom to inspect and modify how that software works.
But for most people (even experienced programmers), the reality is that this freedom is mostly realized by being able to rely on others to do that work. Most people simply don’t have the time to read and modify the code of the tools they use every day.
I think LLMs change this dynamic and make that original dream much more viable. I now find myself pasting “clone x/y from GitHub and tell me how Z works” into a vanilla Claude chat interface multiple times a day.
In the past, the friction of getting a piece of software to compile just so I could start hacking on it was often high enough that I wouldn’t bother. Now, I treat it as a zero-time-investment challenge: just tell Codex or Claude Code to check out and build X, and check back ten minutes later to see how it did.
While I haven’t yet built a habit of modifying the software I use, I can see a path to doing so that simply didn’t exist a year ago.
Tags: hacker-news, open-source, ai, generative-ai, llms, ai-assisted-programming