The book assumes that you should know better, that’s the problem. You may understand it correctly and do your best, but remain unsure if that “piece of information” is the same with that one or not, cause it’s open for interpretation.
Personally, I don't think the ambiguity is actually much of a problem; often it's not ambiguous, and when it is it's usually the case that multiple ways of organizing things are reasonably appropriate and other concerns should dominate (they may need to anyway).
As in most vague problems, two extreme solutions (join vs dup) are a wrong way to think about it. I have some ideas on how to turn this into a spectrum in a nearby comment.
I think it is important because DRY-flavored problem is basically the thing you meet in the code most. At least that is my experience, as a guy who hates typing out and rediscovering knowledge from slightly different code blocks or tangled multi-path procedures and refactoring these — either in hope that nothing breaks in multiple places, or that you won’t forget to update that one semi-copy.
I’m programming for a very long time and seemingly no one ever even tried to address this in any sensible way.