I doubt that outside of special cases this would actually provide significant benefits. Suitable languages already exist (see Erlang), but the fact of the matt

I doubt that outside of special cases this would actually provide significant benefits. Suitable languages already exist (see Erlang), but the fact of the matter is that a vast array of problems don't benefit from parallelism or aren't parallelisable at all. Not to mention that a good chunk of the benefits would be negated by the increase in coordination and synchronisation between parallel tasks.