> My guess is that Claude is trained to bias towards making minimal edits to solve problems.
I don't have the same feeling. I find that claude tends to produce wayyyyy too much code to solve a problem, compared to other LLMs.
That has been my impression too, it takes particular guidance to get it to write concise code without too much architecture airmanship.
That has been my impression too, it takes particular guidance to get it to write concise code without too much architecture airmanship.