For example, are some architectures not susceptible to Meltdown like attacks because they are designed properly or because we haven't found a way to yet?
You can't escape the JVM, but you can escape a Virtual Machine.
The mind of man, nor his trivial algorithmic automations are deception free.
Yet are discrete and codified insulation of intrinsic function beyond laws of what may be? I say no.
The problem is that we ourselves are self deceivers. We do only what is necessary for intended outcome. It takes ten years of reflection for our progressive learning to catch on. Despite how smart we think of ourselves.
Idempotent logic is one example of potentially “immune” logic, yet the devil is always in details of implementation as so many layers exist between client interface and serviceable results.
Wrapping layer by layer, where nothing really trusts or takes anything for granted must be intrinsic design for the hardest core.