I firmly believe that there will be no real breakthrough until we invent (or accept) something to abolish complexity instead of hiding it under multiple abstraction layers so that we could more of it unnoticed.
Here's an example. Although I'm not a guru in Lisp and certainly not in Forth, I like early Forth and even earlier Lisp systems for being self-sufficient. Because your early Lisp/Forth IDE doesn't run on top of an operating system, it *is* an operating system. You have less layers and tighter integration, less space to plant bugs or backdoors.

Here's what I'm talking about: https://www.exemark.com/FORTH/eForthOverviewv5.pdf
They described an implementation on top of an early DOS, but the same could easily have been done over bare metal.