[cryptography] Password non-similarity?

John Levine johnl at iecc.com
Sat Dec 31 22:21:16 EST 2011

>> I finally realized, that's so when the organization gets pwn3d, you
>> won't have used the stolen passwords anywhere else.  Or maybe they
>> imagine that if your password is stolen somewhere else, you won't have
>> changed all the passwords at the same time.
>Really?  So you're proposing *cross*site* non-reuse?  How does that work? 
>If you make me change passwords, and many sites do that, what incentive 
>is there to do anything other than use the same password [or a trivial 
>mod] for each?

I didn't say this was a particularly good rationale, just that the
idea is that your password won't be exactly the same as the one they
used other places, because their password rules are so stringent.

>> There's also the backup tape that fell off a truck issue, ...

>but I don't understand again: if that happens, then presumably the IT 
>folk *know* and _then_ you can make everyone change their passwords [at 
>least for a reason].

How would they know if the tape fell off the truck?  When it gets to
the offsite vault, do you really think they carefully count the number
of tapes in each incoming box and compare it to some manifest?  And if
they don't match, is the count or the manifest more likely to be
wrong?  Again, I don't think this is a particularly compelling
argument, but backup media do get lost from time to time, and people
often don't notice until they look for it and can't find it.


More information about the cryptography mailing list