I think that's the wrong way to go, because often you don't need to implement a complete slice of a type of functionality to enable complete and valuable transactions.
I have put some notes on the pages for Iteration 1 and Iteration 2 where I propose that we demote some Use Cases from Iteration 1 to Iteration 2, and that we promote one of the Machine Translation Use Cases from Iteration 2 to Iteration 1.
I think this will make Iteration 1 more manageable, and will allow us to experiment with Machine Translation earlier in the process.
This will probably end up making Iteration 2 a bit too fat, and we may have to demote some Use Cases down to Iteration 3, but we can deal with that in time.
If nobody objects to my proposed modifications, I will update the Iteration plan accordingly on Thursday.