Multics required unusual hardware that was expensive at the time. The hardware was abandoned by GE and taken over by Honeywell. Honeywell, which made thermostats then and makes thermostats now, was never a major player in computing. The big advantage of UNIX was that it runs on vanilla hardware.<p>UNIX is all wrong for microservices. Interprocess communication barely existed at first, and it's still mediocre. QNX did this right, with a true message-passing architecture, and a message-oriented network protocol. (Reliable, any-length message, not just raw UDP packets.) QNX continues to power many real-time systems, passing messages around.
> See, the brilliance of Unix didn't stop at functional orthogonality, they also used C: A high-level programming language, as compared to machine-specific assembly code. Once Unix started getting ported to new hardware architectures, the two-dimensional implementation matrix becomes three-dimensional.<p>Yeah, like plenty of other OSes since the late 1950's.<p>Multics was also written in an high level language, PL/I.<p>If anything, I am looking forward to the cloud platforms to replace UNIX.<p>It doesn't matter what AWS runs on, as long as my language runtime, or Kubernetes runs there.<p>UNIX, hypervisor, Linux, Windows, bare metal,...., I just don't care.
> Write programs to handle text streams, because that is a universal interface.<p>I wish they'd invented a simple structured data model, like JSON, and exchanged that rather than plain text.
> "Code the Perimeter" is the key insight of Kevin Greer's fabulous 2016 analysis of why Unix beat Multics<p>Brian Kernighan's "UNIX: A History and a Memoir" delves into why multics failed and unix was created and opinions on why unix succeeded ( something they really didn't anticipate ). It's a great book, broader than just unix - bell labs history, the people involved, computer history, etc.