imho. you have to look at those statements from the eyes of the early 1990ties: back then it was clear, that microkernels are the future*) ... but: just not yet ;))

why? because cpus in microcomputers where still pretty slow and the performance-penalty for using a "full blown" microkernel a la [cmu|gnu] mach was not justifiable ... again: at least at this point in time.

which imho. had 3 outcomes during the 90ties etc.:

1. the development of more lightweight microkernels - idk. for example L4 from i think the technical university of dresden/germany is the first which comes to my mind

2. the development of "hybrids" to mitigate the performance-penalty - a la darwin/the base of mac os x: a combination of cmu mach and freebsd hardware-support ... why the older cmu mach? to avoid the GNU licence ;)

3. the persistence of somewhat modularized monolithic kernels a la linux (and *BSD)

*) afaik even early versions of windows nt, which was developed during that time, tried to incorporate some aspects of microkernels, which where removed over later generations :))

Well that argumentation didn't age well. Kind if highlights that saying of, "Those who can, do. Those who can't teach" Research and experimentation are important, but this just highlights the problem of acidemia thinking they have applied knowledge of the real world.