If you have only the former, you are a technician, who understands what to do and how but without a deep understanding of why. (And there are plenty of programs that turn out such people in the software field, typically through two-year degrees in community colleges.)
If you have only the latter, you are a theorist with a deep sophisticated understanding of hard problems in your field, but without the skills to build much of anything right now. You known the why, but not the how. But if that's really what you want, try a degree in math or maybe physics.
Technical undergraduate degrees are supposed to bridge this gap by teaching both the how and the why, producing people who both understand the subtleties and can do real work right now. That's really useful, which is why these graduates are so sought-after. A university that neglects either side of this span is producing people who are a whole lot less useful, and that's what Gelertner is pushing back against in his article.
This falls apart at graduation + n years, unless you accept that up-to-the-minute skills must be maintained and regathered as the minutes progress.
If it's possible for an educated mind to continue to keep up-to-date, then one must accept that an educated mind, freshly graduated with a sophisticated understanding of the field can, and probably should, pick up the up-to-the-minute skills as part of the on-the-job training any new graduate acquires.
There's always a ramp-up in any new role, especially so for someone who's never worked professionally before. Given a choice, I'd rather aim for an education providing the depth, and the up-to-the-minute ramp-up happening on the job rather than the other way around, because I've never seen a developer's day job provide any significant theoretical depth beyond what is required to get the job done. If you don't learn how compilers and operating systems and low-level memory management and design patterns and concurrency and big-O notation work before you hit the workforce, you're not very likely to pick all those up at work.
It should be food for thought as we discuss this article that the author is the son of a famous computer scientist who is a professor of computer science at Yale. I think the nature of a university education and what a university education is for was probably dinner-table conversation at his home as he was growing up.
That's why other engineering disciplines have CPD and it takes several years to get to full chartered status.