The counter argument is bunk, you change the comments when you change the code. It’s not hard. If you forget, them a reviewer will notice. If not, then you aren’t focusing enough attention on the finer details of a project, to your detriment.
Personally, I find that a lot of good, accurate comments and documentation helps me to trust a code base more because I can see that the devs cared and focused on the finer details rather than just tossing code over a fence at me.
Can't you just say that you used Claude and be done with it? I mean, I don't really care but in general I would refrain from posting "Show HN" with stuff that has been vibed, or state that you did since in that case it's even more interesting from my point of view, like you could describe how you kept everything under control during generation, etc...
I mean, no biggie and I don't understand the reaction but hey, I'm old so...
I put Overview section from the Readme into an AI content detector and it says 92% AI. Some comment blocks inside codebase are rated as 100% AI generated.
That said, totally fair read on the comments. Curious if they helped/landed the way I intended. or if a multi-part blog series would’ve worked better :)
Thanks for the link, very interesting data structure.
I'm wondering is it really worth dumping a general knowledge articles into code comments? To me it feels like the wrong place. Would just the wikipedia link be enough here?
I also notice a lot of comments like this
// IsEnd checks if this is the EOF sentinel
//
// Example usage:
//
// if pos.IsEnd() {
// // We've reached the end, stop searching
// }
func (p *Position) IsEnd() bool {
return p.Offset == EOF
}
Is it really necessary to have a text description for a code like "a == b"? It would be really annoying to update comment section on every code change.
This is one of the typical issues when AI creates "code comments", because it always describes "What" is happening. A good comment should answer the question "Why" instead.
For the linked skip list module, a good comment could say why skip list was chosen over b-tree or other data structure and which trade offs were made. AI will never know that.
I wonder if I should really explain or if that would provide a list of things to sanitize before publishing stuff.
If someone has ever written any code is well aware of what can be done in a weekend and especially that no one doing something "in a weekend" will ever add all those useless comments everywhere, literally a few thousand lines of comments. That takes more time than writing code. Comments in Claude style. Other claude-isms all around.
It's ok to vibe things, but just say so, no shame.
And yes, after 5 minutes of looking around I had enough evidence to "prove it". Any moderately competent engineer could.
