I don't think you're wrong but I do think putting all the burden and emphasis on the last mile (high-level software libraries and application software) will not

I don't think you're wrong but I do think putting all the burden and emphasis on the last mile (high-level software libraries and application software) will not be enough. Memory safety needs better support from operating systems and hardware, both of which have improved but have been much slower to change than application programming. Systems programming, sitting in the middle of the two, is caught between a rock and a hard place.