strong crypto => increase in rubber-hose attacks?
I remember seeing a news report a while back that said the number of violent car thefts has increased because more people are using sophisticated car alarms and/or car tracking devices. According to the news report, instead of breaking into cars when the owners were away, car thieves would wait until the owner shows up and deactivates the car alarm. The thieves then pounce on the owner, killing or severely injuring him/her, and take the car. Regardless of the actually magnitude of this problem (the media would call increase of 1 a major new trend) this did get me to thinking. Assume you use strong crypto to protect your secrets. Assume a lot of people start using crypto to protect their secrets. Assume there are people who want to discover these secrets. Might we some day see an increase in the number physical attacks as bad guys resort to rubber-hose methods to get at the keys that protect the secrets? Don't get me wrong, I'm still in favor of using strong crypto. I'm just wondering about some of the social implications. Jim_Miller@suite.com
Jim Miller writes :
Assume you use strong crypto to protect your secrets.
Assume a lot of people start using crypto to protect their secrets.
Assume there are people who want to discover these secrets.
Might we some day see an increase in the number physical attacks as bad guys resort to rubber-hose methods to get at the keys that protect the secrets?
I think this phenomenon is more or less inevitable, unless serious thought is given to a way to prevent it. Let's take a simple example and progress to a more complex scenario: If I want your money, I could steal your ATM card and try to deduce the PIN number (tough), or I could wait in the bushes with a .44 until you use the ATM and either steal the money you get out (easy) or convince you to tell me the PIN number (harder, although a .44 is remarkably persuasive). However, it's easy for you to lock me out of your accounts by changing the PIN number the next day - to get continued access to your account, I'd have to get the PIN number and then kill you (begging your pardon, of course). Even if I did all that, all I'd have is a bank account. Hardly worth it. If I want your *life* (metaphorically speaking; your network connections, your digicash, your 'reputation capital', etc.) and all I have to do to get it is beat your PGP pass phrase out of you and kill you afterwards, you're in much more danger. I could lie in wait, get your pass phrase, (ahem) remove the evidence, and step into your net.shoes the next day. Bottom line: As the value protected by our encryption systems increases, we must devote more effort to the solution of problems like the thug with the .44 (or the jealous co-worker; insert favorite bogey-man here) who wants our password. Key revocation certificates are nice, for example, assuming you are able to issue one - 'dead men revoke no keys,' however. Duress codes seem like a better deal; even though the enemy may kill you after you give him a code that (seemingly) works, your 'estate' would be protected. What we *really* need is a hat trick that makes strong-arm tactics useless - any ideas? When the tactic of beating a pass phrase out of a citizen becomes as stupid as killing for a PIN number, we'll know we've succeeded. <Shudder> Damn! My coffee's gone cold. I'm off to get a refill- talking about killing people has given me a definite chill. Not my favorite topic. ........................................................................ Philippe D. Nave, Jr. | The person who does not use message encryption pdn@dwroll.dw.att.com | will soon be at the mercy of those who DO... Denver, Colorado USA | PGP public key: by arrangement.
It seems like it would be relatively simple to program in a sort of dead-man switch at the time of creation of the secret key. As with other double-pass systems, use of the second phrase works once, but either/both sends a silent alarm and changes the passphrase to get the secret key. If you want to get fancy, you might even program in a script that, on activation of the 2nd "duress" phrase would run around re-encrypting everything with a second private key. As you can imagine, there are increasing levels of personal security you might employ. For example, using the duress phrase might be set up to change the pass-phrase to something *you* don't know but which is known by a trusted other party (wife, mother, agent/lawyer, etc.). Knowing this phrase doesn't help them since that phrase can't access your secret until *after* you've given the duress phrase and the software has disabled your normal access phrase. Depends how paranoid you want to be and how valuable your data is, I guess... --Alan Wexelblat, Reality Hacker, Author, and Cyberspace Bard Media Lab - Advanced Human Interface Group wex@media.mit.edu Voice: 617-258-9168, Pager: 617-945-1842 PUBLIC KEY available by request "To pleasure!" "To passion!" "To paradise!" "To pain!" "Tonight!"
Philippe Nave:
As the value protected by our encryption systems increases, we must devote more effort to the solution of problems like the thug...
Distributed pass phrases have been mentioned, as have distress codes and dead-man switches. Along with distributed pass phrases I'd add distributed keys with M-out-of-N voting, for valuable shared data (eg corporate data, family bank account). Other options include: * steganography and foiling traffic analysis: don't let the thugs know you have valuable encrypted data in the first place. * multiple keys with multiple (easy to remember!) pass phrases. We use different keys for our house, car, safes, etc. as well as different passwords for different accounts, and this will likely be also true for our encryption strategies. For example, it's a good idea to encrypt files on one's disk with a different key than one uses to receive e-mail. The ability to choose high-entropy but easy to rememember pass phrases is an interesting art. One strategy is to pick short pass phrases with high entropy (eg by requiring mixed cases, letters+numbers, etc.) PGP allows one to pick a long phrase, easier to remember, with less entropy per character but more total entropy. I despise the current situation where I have to choose a bunch of difficult to remember alphanumeric codes for different kinds of accounts, ATM access, etc. Nick Szabo szabo@netcom.com
participants (4)
-
Alan (Miburi-san) Wexelblat -
jim＠bilbo.suite.com -
Philippe Nave -
szabo＠netcom.com