Artificial Specificity
Me to Claude: I think there’s a less aggressive way to increase the selector’s specificity without using the !important keyword.
Claude to Me: Oh yes, you are absolutely correct! The !important keyword is overly aggressive for this situation and there are alternative approaches that will still allow me to increase specificity without sacrificing maintainability.
For example, [blah, blah, blah].
[…much…much…later…]
Perfect! I have increased the selector’s specificity from 0,0,1,0 to 0,0,2,0. Here’s how it’s calculated, [blah, blah, blah].
Me: Wasn’t I the one who reminded you to take specificity into consideration?
Claude: Haha, you make a really good point! The detailed explanation was over the top considering the suggestion came from you in the first place. My bad!
Me: [Ends session.]
And that’s a “small” detail. They’re like cockroaches: you spot one of them and suddenly they’re everywhere.
Technical debt, it’s a helluva drug.
8 Comments
I think they are pushing because of the money. They need to start showing that it’s worth the investment. That is the only thing that makes sense to me.
I hate all this “AI” hype.
I’ve tested all the popular models; all of them make the same mistakes.
They can’t even make correct changes to small (~100 lines) files…
They always make basic mistakes from popular code instead of prioritizing documentation.
And they always explain things I’ve pointed out, just like you showed in this post.
Not to mention the sycophancy…
Likes