What have we learned in 3 decades? Not much.

The guys who did the Multics penetration tests for the Air Force have
re-released it, with commentary on what 30 years has changed (and more
importantly, not changed). Most depressing quote:

    Thus, systems that are weaker than Multics are consid-
    ered for use in environments in excess of what even Mul-
    tics could deliver without restructuring around a security
    kernel. There really seem to be only four possible con-
    clusions from this: either (1) today's systems are really
    much more secure than we claim; (2) today's potential
    attackers are much less capable or motivated; (3) the in-
    formation being processed is much less valuable; or (4)
    people are unwilling or unable to recognize the compel-
    ling need to employ much better technical solutions.