Also, his fascination with magic sinewaves. I don't know that he ever had a concrete application in mind. I don't recall any circuits to generate timing. Presumably just counters and gates -- or an MCU and timer -- which, quantization completely wrecks the timing required to null all those harmonics.
As it turns out, people do use those techniques today; they're somewhat more mundane (and perform not nearly so much better than plain old PWM or S-D, to displace either one), and way,
way more computationally intensive than perhaps he ever envisioned. Yup, upside to all the CPU power we have these days: you can solve the trig polynomials in real time, say every few cycles* or so, making actual dynamic control possible -- the pulses and timings are all completely dependent on output waveform
and amplitude, so it's a rather useless thing to try and precalculate a couple of operating points, when what you really want is something you can vary continuously (and perhaps not always have the best results, but to have it clean most of the time greatly saves on harmonic energy and therefore filtering and EMI).
*Cycles of whatever waveform you're synthesizing, that is. Not like, CPU or timer clock cycles...
Also an aversion to solar power, at least back in the 90s and 00s. Specifically that it would never be a net energy producer, but here we are, looking it up I see a 2012 powerpoint showing the industry either on track for, or achieving, break-even (more power produced than consumed), and another 2018 article showing we've also cleared total production (net energy producer including development and capital costs), and also that wind has an unusually quick payback time (less than a year!).

I haven't read tinaja.com recently enough to know if that position has been updated, I assume it has.
Just some peculiar contours to a smart and complex individual.

Tim