Interesting idea, I like the "as much certainty as you want" with the probability.
Why can't this be parallelized? Sure you have to work on one key at a time because they are sequential, but 1000 machines can be cracking that one key. Every key cracked would be distributed and the cluster would start in on the next.
It's a little more complicated, but I'm not seeing how it's really any less parallelizeable.
I think, but am not certain, that with such a small keyspace the IO cost would dominate if you tried to parallelize it. (Assuming each decryption attempt is rather quick).
Why can't this be parallelized? Sure you have to work on one key at a time because they are sequential, but 1000 machines can be cracking that one key. Every key cracked would be distributed and the cluster would start in on the next.
It's a little more complicated, but I'm not seeing how it's really any less parallelizeable.