I would assume some limiting factors here are: - the speed of end to end tests - the flakiness of end to end tests - the coupling to the user interface - the f

I would assume some limiting factors here are: - the speed of end to end tests - the flakiness of end to end tests - the coupling to the user interface - the fact that specifications for practical applications can be very complex, with know way of knowing the spec is accurate That being said, I think this approaches the platonic ideal of how to test applications. We may get better at synthesizing programs from specifications - the hardware field does it with FPGAs. I have been really into TLA+ since diving into it, it seems to provide such an elegant model of computation. From what I’ve read, TLA+ was created to address limitations of LTL for large, practical applications. So that will also be interesting to see how it plays out. Kudos for trying though, I’m interested in seeing what comes out of this path.