So, hash tables don't really work like this. There's no reason for the "FOOBAR" password to be stored close to "FOOBAZ", if you pick a well behaved hash functio

So, hash tables don't really work like this. There's no reason for the "FOOBAR" password to be stored close to "FOOBAZ", if you pick a well behaved hash function. You might get to know how many passwords are on the same hash bucket, but that's it. Furthermore, the authentication will be accessed over a network. The network latency (and it's variance) is several orders of magnitude larger than the possible variations. It would be seriously hard to measure something like this with the resolution needed (nanoseconds). Even if you're taking samples and doing an average based guess, you'd need an absurd number of samples (so a bruteforce attack on passwords would be more feasible?) Edit: Actually, even if by coincidence you did get FOOBAZ on the same bucket as FOOBAR, you wouldn't be able to easily distinguish from a single comparison with 5 equal characters from multiple comparisons with smaller number of equal characters. Also, if you're getting that many collisions to the point that this is a problem, you've sized your hash table incorrectly.