this post was submitted on 13 Sep 2026
68 points (94.7% liked)
Free and Open Source Software
23168 readers
6 users here now
If it's free and open source and it's also software, it can be discussed here. Subcommunity of Technology.
This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.
founded 4 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
A really important part of FOSS is testing and bug fixing. Often the bulk of code is done by a small number of people but what makes their software truly useful is the testing and fixing contributions. Unless LLMs could deliver bug-free code, the FOSS projects that have community and userbase would always be more reliable and therefore useful than bespoke LLM-written software. I wouldn't write my own grep with some custom feature. I'd use GNU grep and augment it with another tool, or perhaps get an LLM to add the feature I need to it.
What made me think along those lines is that someone I know vibe-coded their own personal/private XMPP client recently. I'm sure it has lots of problems, but it does what they need and has a nice UI that looks how they wanted. In their case, it doesn't need to be bug-free - just bug-free on their machine, for the tasks that they do with it.
But for some software, I definitely agree about the reliability and edge-cases / environments. Knowing 1000s of people have tested my backup software and ironed out the bugs is a big plus. I can't imagine many people would want to use a vibe-coded one.
Thinking about Blender, IIRC, their filetype is effectively a memory dump of the process. So, if I worked in an animation studio and wanted a custom feature, I'd definitely write a plugin instead of making a bespoke tool. But then again, if I could export to
.ofxor another common format, and have everything import perfectly, I'd be more tempted to go for a bespoke tool again. It's interesting to think through the scenarios.That's related to the proverb "writing code was never the main part of software engineering." Also related to the difference in difficulty of writing the first 90% and the last 10% of useful software.