Along the lines of of your examples, I would add distributed systems.
Still, cryptography is much, much worse than all these examples for a simple reason: with a sufficient amount of testing for functionality you can convince yourself that you've gotten things like concurrency correct, or at least correct enough that it won't be a problem most of the time. You can't do that with security - instead of testing for functionality you have to test for attack resilience, but whereas you know what functionality to test for, you can't test for attacks you don't know about. Furthermore, you can't be correct merely most of the time - you have to be correct all of the time, because if you slip up just once, your adversary will exploit you.
Still, cryptography is much, much worse than all these examples for a simple reason: with a sufficient amount of testing for functionality you can convince yourself that you've gotten things like concurrency correct, or at least correct enough that it won't be a problem most of the time. You can't do that with security - instead of testing for functionality you have to test for attack resilience, but whereas you know what functionality to test for, you can't test for attacks you don't know about. Furthermore, you can't be correct merely most of the time - you have to be correct all of the time, because if you slip up just once, your adversary will exploit you.
I really don't think there's any parallel.