I ran into this post a few years ago and have thought about this often.
https://www.freestompboxes.org/viewtopic.php?p=306736#p306736
This looks similar to the Crowther Prunes and Custard. The First op-amp looks almost like a normal clipping stage. The second one has a crazy confusing maze of diodes and resistors.
This image and the explanation starts to reveal what is going on.
With SW1 in position 3, the output of IC la will drive the four pairs of diodes, D4-11, to produce four waveforms clipped at ± 0.5, 1,1.5 and 2V. These are then fed to alternate inputs of a difference amplifier (IC lb). So as each diode begins to conduct, the gain of the circuit reverses polarity.
With SW1 in position 1, only the lower pair of diodes is driven so the circuit produces ‘ordinary’ fuzz.
For the intermediate effect. R19 is used to attenuate the signal reaching the upper diodes.
With the switch in position 2 (center) we get classic soft clipping. With the switch in position 3, as each pair of diodes begin to conduct the voltage changes at the inverting and non-inverting pins of IC1b cause the output to be inverted. In position 1, the effect is moderated to fuzz effect.
The input stage of the original looked too complicated. It didn’t look like it was adding much to the circuit. I went out on a limb guessing that it is just there as a buffer/booster. I tried simplifying this to a single LPB-1/Common Emitter stage with a gain control.
I’m putting this here as a not tested in any way shape or form. Just an idea that deserves some attention.
I only wonder if anyone had ever built the original from the article in 1988? Maybe they did and it wasn’t worth writing about. The name Hyper Fuzz isn’t helping us since Boss made a Hyper Fuzz, searches seem to home in on that.