Re:Encryption mostly overrated (Score 1)

It isn't difficult to imagine a dictionary attack working against a password of the form "concerned$citizen". There are roughly 1,000,000 words in the english language http://hypertextbook.com/facts/2001/JohnnyLing.sht ml. It wouldn't be difficult to transform each word by doing every permutation of common substitutions, e.g. s=>$ o=>0 e=>3 etc. I'm spectualating but for purposes of calculcations lets say that on average each word will have 20 such permutations. That gives us a total of roughly 20,000,000 possible words. In addition one could take it a step further and try all combinations of two such words, 20,000,000*20,000,000 = 4e14. In addition we may want to allow for arbitrary characters to be inserted between such words, 4e14 * 73 = 2.9e16, where 73 is 52 leters + 10 numbers + 10 punctuation + 1 blank.

Using your assumption of 2,000,000 hashes per second per machine. It would take 2.9e16 hashes / 2,000,000 hashs/(sec*node) / 60 sec/min / 60 min/hour = 4e6 node*hours

With a 1024 node cluster it would take 4e6 node*hours / 1024 nodes = 3906 hours = 163 days
To do it in 90 days would take 4e6 node*hours / 90 days / 24 hours/day = 1852 nodes.

I think these numbers demonstrate that it is definately within the realm of possibility for a well funded government agency to crack such passwords.