Well, one of the people arguing with you is ME, one of the pioneers in A/D and D/A conversion (Analog Devices), a designer of signal processing for the first CAT scan systems (Analogic Corporation), a designer of aircraft countermeasure systems (Sanders Assoc.) and, well, I'll stop there. Please don't lecture me on understanding technology.
Maybe yours is bigger than mine, maybe it's not...
Since you are a "recognized expert" in A/D/A conversion, I would expect that you would understand the difference in HOW the analog information is encoded, where the information is contained in the signal and what you can/can't do with the signal. If you come from a digital background, I can see your pre-disposition to a "digital" definition.
Since my background is as an analog design EE, I take the analog definition because the PWM process and theory was in use long before the "digital electronics" existed (going back easily into the early 1900's). The process has been used in generator control, lighting control, motor control, and frankly the theory extends outside of electronics and into the world of graphics and optics (the halftone graphic process being one process of PWM that has existed well before the world of digital electronics was even imagined.
Because the world is not black and white, the definition of PWM (in electronics) that I learned and those of my era practiced is that the signal is generated by purely analog methods (triangle/sawtooth (another analog signal) compared to the desired analog signal) the comparator being another analog process, the resulting pulses are non-linear analog, no more digital than the signals that they were generated from and the pulses themselves mean nothing in particular EXCEPT that the TIME duration of the pulses represent points of the analog signal through time. This can be done with light, water, doesn't matter because it's the process that's both non-linear (switching) and analog.
For an analog example of the PWM process, in a switchmode power supply (another non-linear analog process, though there are digital methods of creating such a supply as well), the output voltage is compared with a reference voltage in the error loop and the duty cycle of the PWM signal of the switcher is adjusted by changing the reference voltage of the switcher's PWM comparator. This is very much an analog process, and the PWM drive signal does nothing more than drive MOSFET gates between saturation and cutoff, the TIME of the on versus off signal can be adjusted by the error loop to maintain regulation (constant output). It's a free running, non-linear analog process that has essentially "infinite" resolution (ie. bitless).
For a light example, a spinning disk in front of a light source with an opening slot(s) or holes that varies with radial location on the spinning disc placed in front of a columnated light source can vary the intensity with the position of the disc relative to the light source. The individual openings are either 0% or 100%, so the question is if this represents digital, or does it represent non-linear?
For extra credit, is a chopper stabilized op amp a digital or analog device?
It depends on perspective, if all you have is a hammer, it's easier to see every screw looking like a nail. As analog guys die out, more and more technology will be claimed by the digital world. Also, digital techniques do evolve to take over analog functions, that's the nature of evolution.
