Not that I seek it out, but whenever I come across a Kent Beck post, I feel exactly the same. Authoritatively Pointing out the obvious with metaphors while having zero practical advice.

Know when to leave your existing design alone as "good enough", even if it's not perfect. Know when to improve the existing design. Know when your existing design cannot be improved to meet the need, and therefore you need a new one.

How do you know each of these "whens"? There's nothing there that I can see. But there is some value in knowing that those are the three options that you have to decide between.

I am not Kent Beck, and all I can tell you is "you'll know when you see it".

It comes with experience. Software is more art than science.