I think the basic idea is that the hash has a fairly uniform probability distribution, so knowing the prefix means you can estimate its location in a sorted lis

I think the basic idea is that the hash has a fairly uniform probability distribution, so knowing the prefix means you can estimate its location in a sorted list. For instance if we were talking about n random sequences of digits then if you want to look for a number starting with 42 then you can start looking at the the 0.42n element and it is likely already very close to a match.