This particular line is wrong in my opinion:
> the tool (or system) is too complicated (or annoying) to use from scratch
Unless you're doing trivial things, any tool or system will require some setup (which people call "ceremony").
Tools and systems can be easy to use from scratch only if they are either super-specialized or they impose significant constraints on what you will be doing with that tool and how you will be doing it.
Such tools are usually very tightly couple to a specific job/environment/task and are hard to keep around, keep updated and to evolve.
Make and similar tools instead are generic and can be adapted. The fact that you can reuse the previous work done is actually a feature. You can dive as deep as you want or need. They're very widely used so it's not an issue to keep updated. They're so widespread you can find people already familiar with those tools. Learning such tools is a great investment because you can keep using them over and over across project and companies.
Some of those tools are either timeless (gnu make) or have a very long life (more than a decade, which very long for this industry).
Anecdotal example: I learned a bit of apache ant while in high school because my laptop at the time (a netbook with 1GB ram and an atom processor) could not run NetBeans decently, so I had to learn a bit of apache ant and resort to writing and maintaining my own build.xml file. Fast forward 14 years and I see a build.xml file in the $FAANG codebase I was working on. That learning did pay off beautifully many years later.
The article is shortsighted, if anything it's promoting a shallow way of working. You are supposed to learn about the tools you use.