write their own papers on that topic
As long as that topic is what they would like to research, I don't see a problem. And if it's not, then they chose a wrong lab/advisor.
But this leaves them a very narrow set of skills
I developed a pretty wide range of skills as a phd student. From developing device models in Spice, to Tensorflow. In fact, I wish I focused on a subset of those skills.
A grad student is an apprentice. This article is about a less taken path for a master. As a former PhD grad student, I can say most grad students are not in a position to conduct independent research. Most get attached to some funding obtained by someone more experienced than them, and then spend most of their time getting up to speed until they are able to write their own papers on that topic. But this leaves them a very narrow set of skills. This is why most academics go through a postdoc.