Turn back the hands of time. I wrote machine code on a Netronics(sic) Elf II. It was my introduction to microprocessors. I thought it was a very very slick chip. Sixteen registers all the same, anyone could be the stack pointer or the program counter. Regarding interrupts, I seem to remember it had very low overhead and subroutine calls and returns were just a few bytes long. In order to leverage the versatility of the register set, you had to designate registers for purposes - PC, stack pointer, and one other use I cannot remember - maybe the return address? There was one instruction which set/switched the PC counter and stack pointer. There was a recommended call and return procedure. Slicker than snot on a door knob. I seem to remember using register A for passing values.
The video was primitive but I managed a character set and repeated a few lines to give the letters a bit of charm. The characters were 3 x 5, but I repeated the second and fourth lines three times. The letter E - and it looked about this big on the screen, too.

And... because I could only fit 8 characters on the screen, I had the lines scroll like a marquee across the screen. I remember adjusting video line start addresses during vertical retrace. The video generator was just one chip and poked the DMA channel for bits as it needed them while tracing a horizontal line on the screen. Non-video processing was done during retraces.
And I made some maze games.
And I burned EEPROMs.
And, the programs were stored on paper tape.
Sheeze, I'm a dinosaur.
But the coolest things was this. This was the first popular CMOS CPU. CMOS is noise immune. When I hooked a primitive ADC up to the cpu board, I thought my scope was going on the fritz. I just couldn't get a clean display w/out 60Hz noise. The scope appeared to have an ungrounded probe. I finally discovered that the 7812(?) regulator on the CPU board was shot -- shot as in power IN was shorted to power OUT, and the regulator had an open ground. So, as it turned out, my scope probes and scope were fine, and the computer board was getting unfiltered and unregulated power from a bridge rectifier - just 120Hz camel humps all the way.
I thought this could not be true. But, I replaced the regulator and powered the system back up. All worked just as before, except this time there was 12 volts on the positive rail instead of AC ripple. I took me about two hours to figure this out.
The RCA manual was very good at getting me started. It also extolled the virtues of CMOS as a low power and noise immune technology. I certainly believed that after my experience. RCA marketed the 1802 to automotive applications because if wasn't fazed by ignition systems.
Interesting note, Alan, about military uses. I worked on radiation hardened processors for space platforms and they used CMOS parts with LARGE gates to avoid single-event-upsets. Actually, IBM made the parts on site at Manassas Virginia. I got to look into the silicon fabrication labs, but never got a walk inside.
The rest is history I suppose. IBM got out of military programs. NTSC video is gone. Eight characters across a screen is laughable. But it was a time of great learning -when you could flip bits and see the results. I commend the1802 and a hex keypad to anyone who wants to understand.
This is a great forum =Steve.