[Discuss] Most common (or Most important) privacy leaks

[kentborg at borg.org (Kent Borg)]

On 02/18/2015 02:35 PM, Richard Pieri wrote:
> The article you recall probably based it's assertion on brute force 
> attacks. Mathematically, a brute force attack against 9 characters 
> will take longer than it would against 8 characters but that's a very 
> narrow-minded approach. There are other ways to attack passwords like 
> known plaint text, dictionaries, rainbow tables and differential 
> cryptanalysis.

Passwords are different from encryption keys. Completely different.

A password is something you whisper through the little opening in the 
door of the speakeasy. After a couple failed attempts the guy on the 
other side will lose patience and tell you to get lost. (Maybe send out 
a heavy to make it more clear.)

Okay, maybe you don't whisper the password through a door, maybe you 
send it off to some computerized doorkeeper that gets to consider 
whether it be good or not, gets to delay before answering, gets to count 
the number of failed attempts. But it is the same idea.

A 4-digit PIN is a GREAT password--if the number of failed attempts is 
limited. (ATM cards are the prime example here. They do not get brute 
forced, even at only 4-digits. 4-digit PINs make great passwords for 
ATMs, really!)

A 4-digit PIN is a TERRIBLE encryption key--if you are up against more 
than an 8-year-old.

Encrypted data can be duplicated across hundreds of CPUs or worse, and 
billions of attempts can be made in a second against your key, for 
cheap. Very different from the password.

The two are very different.

I don't trust the systems I log into, they might be cracked or be 
crooked, so I don't recycle passwords. I also don't trust that they 
rate-limit guesses very well (Apple?), so I generate passwords that have 
more than 4-digits worth of entropy, but I don't get carried away (I 
tend to a minimum of 32-bits of entropy but not much more). I also don't 
trust that j-random-site is not silently truncating my password, so I 
frequently put a few randomly chosen hex characters at the beginning. 
But then I keep my passwords secret and quit worrying. Passwords don't 
have to be *that* strong.

But that is for passwords. For encryption keys, one has to go further, 
possibly to extremes, but only for encryption keys. They are different. 
I try to keep the number of encryption keys I deal with to a minimum, 
because good ones are too hard to type without error.

That's one of the reasons I don't do ssh keys, it requires I encrypt my 
keys; doing that crappy kind of defeats the whole thing and doing that 
well is hard. Passwords are much easier to remember and type.

-kb, the Kent who has no reason for why he should trust Lastpass with 
anything of importance at all.