Hey, saw your post regarding Google Hash. You're right that it is dictionary based. The idea was spawned amongst colleagues when we were considering the idea of buying a Google applicance to index hashes. I argued that you'd still need to have enough storage to hold the original data, or that you'd need to create it dynamically. The PoC was merely to see if it was possible. We never did get that appliance :(. Your implementation looks quite cute, I wonder how long before Google takes the bait.