I think one distinction often missed is if the programming language can share memory between threads natively (without copying) and if the language has stable n

I think one distinction often missed is if the programming language can share memory between threads natively (without copying) and if the language has stable nio & concurrency libraries to do this today on a live service. The only bottleneck we have is CPU, and because of litography/memory bandwidth it will never go away for transistor processors. I just called this "joint parallel" for my completely async. non-blocking HTTP server/client; meaning it can have many cores work "simultaneously" on the same memory, which is really rare in the async. world.