I disagree that "digital is about discrete steps". There are no advantages to discrete steps, it's just an unavoidable, practical consequence of being about numbers. A good hint of that is that we strive to reduce the step size as much as practical (24-bit audio, 32- and 64-bit float processing). Sampling is about discrete
time, digital is about storing them as numbers. Discrete
steps are about neither, just about resolution.
"...they don't have to be numbers, and numbers are incidental."—Oh? As a DSP guy, I find that pretty statement pretty hard to defend—care to try? I don't care if you want to call them symbols or whatever, in the end you've only made up a new numbering system. How are you going to do a gain change if they aren't numbers? Filter? FFT?
You'll need to do math, with numbers.
This just comes down to what words you like to use, but we're talking about audio in this thread, not the sense of touch or taste. As such, I don't think it does anyone any good to call something like AES-EBU "analog". It's a digital signal in the continuous time domain—more simply, "digital". Analog may imply "a smooth continuum of signal", but a continuous signal doesn't necessarily imply an analog signal.
It won't make me mad if you want to call AES-EBU "analog", I'm just saying I don't think it helps anyone. We all know what it means when someone says "analog audio". It's something that, on an oscilloscope, looks like what we expect the speaker to do when it's playing back that audio.