I think turing incomplete languages are the future, because every property you care about of a program that you will write is reasonably efficient to reason abo

I think turing incomplete languages are the future, because every property you care about of a program that you will write is reasonably efficient to reason about. Otherwise your teammates aren't going to be very happy that you're writing it that way. This defines some subset of computable functions that has some real nice properties. Eventually I expect us to gain more and more understanding of where that boundary lies, defining humans-give-a-shit-about-these-programs-space. Over time, I expect our tools will slowly gain the ability to automatically check the things that we care about and restrict us from doing it in ways that get too hairy. I expect that that future is a very long way off.