This is one of the most interesting CS ideas I've found in the past month, here
I have started to understand rationally why dynamic languages like Python may
generally be not significantly more bug prone than tightly statically typed
languages like ML or D:
Interesting idea, especially when you start considering how software
might survive errors in dependent systems, I.E. hardware, networking,
or software components outside the control of the compiler.
It sounds like the Erlang approach is far more likely to survive
spurious bitflips,