Punch cards are not in any way a more basic form of computation, they're just an inefficient data entry mechanism. You could program JavaScript or Python on punch cards if you were so inclined
I mean, it's a holistic approach. I'd you're going to teach computers, you have to start out with vacuum tubes, before you get to transistors, before you get to integrated circuits.
Why start with vacuum tubes? Shouldn't you start with a Babbage Difference Engine? And how much school time do you think should be devoted to computing history?
However much you're devoting to computing. You can't know a thing without knowing its history, which you generally start at the beginning. Of course, this is taking about college courses; I'm not convinced primary school students should ever touch a computer.
I'll be honest with you: in 1997 I said to a friend of mine who was going to college for computer science that the internet was a fad, and they should switch to something useful instead, and I've just become more and more convinced of that over the decades. I wish I'd never started on it.
You may have just unwittingly made the case for an education more thoroughly grounded in the history of the field: en.wikipedia.org/wiki/IBM_CPC
(IBM had shipped hundreds of these when the number of "real computers" on the planet was in the dozens at most.)
Nor are slide rules. There are manners of *using* a slide rule that are forms of computation, but someone needs to actually *do* something about them; failing that, the slide rule just sits there.
(And the cards were the *main memory* of the CPC, not just, as you seemed to believe, an I/O medium.)
I think you misunderstand what I'm saying. Punch cards are not a form of computation because they're just a way that you write stuff down, whether that's IO or inside the machine is completely irrelevant
You can write a program for an old mainframe computer in ASCII and send it to a simulator and you'll have the relevant parts of the experience of a programmer in that era