Even better. After they hash it they are sending the hash with the first 38 chars trimmed off and (presumably) getting a result set of only the hashes (or maybe

Even better. After they hash it they are sending the hash with the first 38 chars trimmed off and (presumably) getting a result set of only the hashes (or maybe one hash... not sure) that match the ending. Then comparing locally the full hash to the result set. I'd say that was a pretty smart way to do this. Edit: the author explains it a little more here: http://news.ycombinator.com/item?id=4075767 http://news.ycombinator.com/item?id=4075767