I wish one day that it gets open sourced (but I'd settle for a non-open read-only license) because it's unlike anything else.
Seems that Chris Hinsley (inventor of TAOS) is working on a modern day version of TAOS. 
A big thread about his current project happened last week: https://news.ycombinator.com/item?id=15466124.
Missed last weeks thread about ChrysaLisp, will read.
TAOS is/was a "mixed-processor parallel operating system from Tao Systems Ltd of Belsize Park, London" says http://www.cbronline.com/news/tantric_has_taos_developers_ki... . It was for iAPX-86 machines.
I don't believe it's related to Taos, "the operating system for the DEC SRC Firefly multiprocessor workstation" mentioned in https://homes.cs.washington.edu/~bershad/Papers/p37-bershad.... and http://hpl.americas.hp.net/techreports/Compaq-DEC/SRC-RR-63.... .
FWIW, this Usenet post is cited in several patents: https://www.google.com.au/patents/US7836292 .
Submitter? Why is this link interesting?
In its more mature form, as Intent/Elate, it is _the_ single most radical OS there's ever been, pretty much. It makes Inferno look conservative & staid, Plan 9 no more than a tweaked Linux distro, and Minix 3 a tweak of NetBSD.
It's also been on HN before:
The first two links to http://www.dickpountain.co.uk/home/computing/byte-articles/t... , which was informative, and I enjoyed david-given's HN comments in the second submission.
Do note that both of those links were submitted by the same person on the same day, so that's a duplicate submission.
The third link noted (1995) in the submission title; useful information missing from this one. It links to the same Usenet posting.
The last link you gave is this discussion thread, so I don't think it's quite fair to include it as a "on HN before" reference.
Perhaps someone more knowledgeable about the topic could update https://en.wikipedia.org/wiki/List_of_operating_systems and/or https://en.wikipedia.org/wiki/Tao_Group ? That's where I expect to find mention of "TAOS operating System"; a name which simply redirects to Tao Group.
I mean, now that minix uses the netbsd userland I'm pretty sure this really is true.
I always though the hard part was in the kernel design and not the userland.
Otherwise it would have been the gnu os and not gnu/linux.
And if it was not for a book license there was a good chance it would have been gnu/minix. But meh I'm not in that universe.
And to your point: I'm not sure which of kernel or userland is harder. Userland is the part I actually interact with most of the time, so I tend to think of that as the main part of the OS. But it may be fair to say that both halves count.
There was no obvious way of running a program which was not designed for this architecture on the OS, except for making one massive object, which then lost all the other benefits.
Pity it didn't go further. :-(
> 386 x86 ARM MIPS PowerPC MCore ColdFire Transputer ShBoom and a V, um, V840? Or something?
So, quite a few architectures.