Or the arbitrary separation distances given by Dr. Palm...
Interestingly, it does provide another feature (if not a very useful one for the most part): the output is largely disconnected from mains, meaning it's insensitive to common mode or unbalanced loads. It only activates from differential load (H to N). That might be something of a safety feature, if an incomplete one (as, once the load is engaged, this function disappears).
One of the April issues that's stuck with me, was an Electronics Now proclaiming a revolutionary new slot-type CPU (as the Pentium II was all the rage, at the time), made from gallium arsenide and alloys, which was purported to glow intense red in operation, being made from LEDs and optics, and as we all know, optical waveguides are terribly fast compared to silicon interconnects -- and equally so, GaAs is famous for producing many-GHz RF transistors.
I guess the irony is, the Cray supercomputers (old news by that time) were built with GaAs, or the ones that were, anyway. It would be neat if someone fixed the problem with that (AFAIK, they were entirely NMOS, very power hungry indeed; due to GaAs having quite high electron mobility, but utterly pitiful hole mobility -- CMOS makes no sense whatsoever on it), and perhaps that was one of the revolutionary features developed there.
Nevermind that, as the Slot and x86 architecture were very closely held Intel patents, a 3rd party developing a complete drop-in solution for it all would be honestly the more remarkable piece of work. One of those... practical details easily left forgotten, when reading about something so remarkable...
(Not that it had any practical effect on me -- as, even if I wanted to, there wasn't much a kid like me could have bought back then.
Judging by some of the letters published in the following issue, there were plenty of people, with means, who were more easily swayed.
)
Tim