"A government granted use monopoly in exchange for full disclosure of how the invention works."
This is an interesting point of view, however I'm inclined to disagree on the basis that the async model is representative of how things actually happen; the imperative model is not.
In an asynchronous architecture, the developer is concerned with only the current state and the set of all events that may cause a transition from that state. Whether or not an event is the result of an error condition is irrelevant, one simply invokes a different handler.
I used to consider the asynchronous model much harder to program in, that is until I changed my thought process to work in terms of state machines. Then it actually became much simpler to program in this model since, at the handler level, one has no preconception about what should happen next, only what can happen next and how to handle each of those situations.
>> C++ is pretty nice when it comes to error handling and exceptions as resources are cleaned up on error.
There has been many a discussion about the benefits vs. pitfalls of C++ exceptions. I personally am not a fan since the lack of garbage collection means that there are situations that require lots of tedious and error prone boilerplate code just to ensure that all resources are cleaned up.
I have to admit that perhaps the nicest way I have seen of handling failure is in the concept of Monads in Haskell (and other functional languages). The ability to add the context of failure as a possibility in any computation, without requiring that computation to explicitly consider it is extremely elegant. Add to that the safety of always returning a well defined value that is fully type checked by the compiler and I believe you have a recipe for very effective error handling.