I think teaching people about stacks can be a good thing. But throwing an H-P and a TI calculator may be faster. Then thinking about the design advantages can also be useful.
BUT, I think traditional education about automata, finite, push-down, the Turing tape, may be over done. Not when RAMs and PRAMs exist in theory.
I see Moore on occasion, but Ragsdale more often.
Well: In FP Brookss' introduction to automatic data processing book, he discusses the IBM mod 30 internal...
True, I have my own bias. The Systron Donner type was the providence of my now dead friend Bill. He Godfathered Shaw, Packard, Crutchfield (I only met him for the first time last Nov.), and the others when no one else would. I don't have enough analog experience except to say that this is a community which bears some watching (not just dynamic systems). I am not clear that some of the models from guys like Richardson (weather) in their small analog form are right, or whether specific bistable optical materials are perfect for future machines. But I do know we have to keep an eye in some of these areas if we expect newer faster machines and algorithms (in particular).
And its likely the commercial customers would likely kill some of this stuff.