Hacker News Reader

Comment by jrockway | Hacker News Reader

Good full-disk encryption systems use key strengthening, making even short passphrases reasonably secure. Testing one passphrase on my machine, for example, takes about 1 second of CPU time. 26^8 seconds is a long time (6,000 years).

The other alternative is to brute-force the entire keyspace, which puts you back in 2^128 (or 2^512) territory but with the option of trying billions of keys a second. A few Universe lifetimes, and my .emacs will be all yours.

Oh, but I use a long password anyway. The key, IMO, is to avoid using a password you have ever used for anything else. You might not have to testify against yourself, but your employer or mail provider will be happy to cough up your password for the authorities. So don't give it to them.

View on HN

A few Universe lifetimes, and my .emacs will be all yours.

Got it: http://github.com/jrockway/dotfiles/blob/master/emacs

Well there are 2 advantages that lets you reduce that time dramatically.

First you can use a dictionary - then munch it (with Jack or something) to create a large dictionary within the keyspace. It's surprising how much that covers.

Then you can parallel the process; I know a few commercial companies (ourselves included) with a large computer capacity for these purposes. I'd be surprised if the govt. didn't have some form of similar access.

If you can reduce the search space by a factor of 1,000 and run your cracker on 1,000 machines it makes a dramatic difference.

OK, let's assume you can try 1 billion (10^9) keys per second. Now add numbers to your keyspace (remember, no special characters or capital letters yet!), and make your password 12 characters instead of 8. Not much harder for you.

But now it takes 150 years to brute-force the keyspace.

I doubt anyone is going to devote 150 years of computer time to someone going through customs.

No clearly not. Which is why I said "you'd be surprised how many people use an 8 to 10 char password sometimes with numbers". :-)

Your password strengthening won't help.

The forensics guys will clone your drive, then feed huge fuzzed dictionary attacks using the same algorithm until they find something that works, and they'll do it with machines a lot faster than your laptop.

They don't have to brute-force the keyspace. They only have to brute-force your password-space.

I doubt this. Do you have a link to a paper covering this technique, or a conviction that used this technique?

Well you have information about the person including his name, perhaps hobbies, passwords from Google, Yahoo or other sites, which will provide them if they are fed some "terrorist threat, brown alert" mumbo-jumbo. In rest NSA and Choicepoint and help you on that front.

Then you just start building a customized keyspace from that information. Include some common substitutions: letters for numbers, numbers for letters. If you have other passwords, look for general patterns. Some people use long passphrases, but they have a pattern or algorithm on how they build all their passphrases! There you just might need a human expert to take a look.

In the end, yes, it is very difficult but it is still a lot better than plain brute-force.

But yeah, I doubt Customs have the equipment or knowledge to do this. That is something that NSA would do. They would have access to the best tools. I can imagine there might be some kind of very constrained cooperation between them. "You send us a disk clone + personal info of the owner, and we'll try to send you back the cracked passphrase. You don't ask as questions how we got it, and you don't even mention that we got involved at all."

Reply on news.ycombinator.com