Yes this is exactly my point. The Mac was a semi-expedient branching point in an effort to get an idea to market, but that's not really the sum of it. Choices were clearly made for personal and political reasons, and it cost Apple 10 years later when it had no answer for Windows NT or Unix.
And I don't think it was about "worse is better" -- they shipped the org chart, really, forked a new team under Jobs to make a "like Lisa but cheap" but it wasn't just "but cheap", it was 100% incompatible, and sacrificed on basic engineering fundamentals.
It also makes no sense to me. The Lisa hardware was expensive, but I think LisaOS could have been made to run on less expensive hardware by jettisoning features, and then picked up again later. Instead because of personalities and org chart they went and made a completely incompatible other-thing that looked like LisaOS without being it, duplicating effort and creating internal ill will, and short circuiting potential futures.
Anyways, Jobs profited it off it twice. Ego satisfaction with shipping the Mac, and killing off the Lisa -- his grudge/nemesis. And then again when Apple was forced to come to him 10 years later and buy NeXTstep because of what Jobs had done in 84.
Larry Tesler is spinning in his grave somewhere.
Another weirdness was that for the first couple of years of the Mac's existence you had to have a Lisa if you wanted to write code for it. The Mac had so little RAM that it couldn't run a Pascal compiler. For this reason, when I bought a Mac in 1984 I also bought a Lisa with a huge 5MB (!) hard drive.
You bring up a great point though: Whatever happened to LisaOS? Did anybody archive the source anywhere or did it completely vanish?
https://info.computerhistory.org/apple-lisa-code
https://github.com/azumanga/apple-lisa
Just two years ago Apple released it under a very strict non-commercial use "be careful how you look at this" license, to the CHM:
https://computerhistory.org/press-releases/chm-makes-apple-l...
I briefly looked at it, it's a pile of Object Pascal and M68k asssembly. I haven't looked to see if anybody has managed to make it compile in any kind of available-today compiler yet.