Here's an explosive paragraph from Anon's excellent essay (https://docs.google.com/document/d/1E_ToOMG2l1XThx6YnyXEajXaf6H1k2yjq8XkAF0ScB4/mobilebasic?pli=1) that deserves to get some attention at Miami:
"In addition to solving some of the previously mentioned problems with proof of work, NXT also offers a number of potential advantages. While it is not possible to predict who will author the block 1440 blocks from now, it may be possible to predict, one, two or maybe even ten blocks into the future. What this means is that, if a merchant trusts the next six forgers in line to author the next 6 blocks (perhaps because they have a long standing reputation for honesty), then it is equivalent to the merchant as if this transaction has 6 instantaneous confirmations. What we are talking about here is the possibility of a decentralized crypto with instant transactions. No more waiting around for confirmations, as with Bitcoin. This is the first Holy Grail of NXT. Furthermore, you could, in theory, rather than broadcasting your transactions to the whole network, send them directly to the next block author. He could, in theory, clear your transactions with other market actors before recording them in the blockchain. So as an example, imagine that a shopper is paying Walmart 10 dollars and Walmart is paying a contractor 10 dollars during the same 1 minute period. Instead of recording both transactions in the blockchain, the block author could simply record it as the shopper having paid the contractor directly. If that contractor wants to then spend 10 dollars on a new pair of shoes in the same block, then the Walmart shopper’s 10 dollars could be paid directly to the shoe salesman. This could potentially allow NXT to match or even surpass Visa’s transaction load. This is the second Holy Grail."