I really like that Apple and Microsoft are pushing for sandbox models in their OSes.
Of course there are things like HP-UX safes, Tru64 and Solaris trusted zones.
None of them are POSIX.
Besides the average user would not even know where to start.
You can use VMs for some of that, but that's the limit on sharing (and that's if you trust your hypervisor to be a separation kernel thing; reasonable for many people. Not for others.) Docker/containers isn't enough. Users aren't enough. Processes aren't enough.
Edit: Okay, think about it this way - if you're the perfect sysadmin who doesn't run anything bad, how do you go about protecting your non-so-perfect users? How do you protect the server you're tasked to maintain?
Thanks.
Afterwards we can discuss UNIX security models.
Desktop OSes are generally pretty bad. Unsandboxed Windows and OS X apps can do an awful lot with your computer.