(a) cf https://news.ycombinator.com/item?id=26953443 for one aspect. It didn't do much good to have massively parallelisable logic programs without any equally massively parallel hardware on which to run them. I ran jobs on top500 machines back then, and even there coordination overheads would kill anything too fine-grained.
(b) it took a lot longer for Moore's law to end for traditional ISAs than we had been fearing, back in the 1980s. Like many things in tech, maybe Fifth Generation had just been Too Early.
(a) is the nub of the question - the spiel from Japan at the time was that they had massively parallel hardware in the works that was on the cusp of delivery...
At the time I was tinkering about with a bunch of transputer boards, modding Minux, thinking about capabilities, and mangling Occam (the language) into something less awful (IMHO).
I was more curious about how this decade (roughly) long chapter in Japan's IT history didn't rate a mention in the authors article (aside from "Japan plowed ahead with hardware in the 80's" (IIRC) throw aways)
(b) it took a lot longer for Moore's law to end for traditional ISAs than we had been fearing, back in the 1980s. Like many things in tech, maybe Fifth Generation had just been Too Early.