This sounds like the "fail fast" philosophy in Erlang. Crashing is the standard behavior for any unexpected data, but the process model makes it straightforward to manage those crashes.