Slashdot Mirror
How Would Crypto Back Doors Work?
frantzdb writes "We've been hearing about adding crypto back doors for the govement to snoop on us, but how would they work? Would there be one key that could be cracked opening up all such traffic? Also, how would/does the government know wether a bitstream is random bits, or encrypted data?"
They could never work.
The simple reason is that as long as there is an algorithm that cannot be penetrated, either by force or by escrow, that algorithm can hide data. On this, at least, the cat is out of the bag.
One of the more likely scenarios which could possibly keep criminals away from data while allowing governments to have access would be an agreement worldwide on a data-encryption standard that included key-escrow. Likely this would be implemented with a large database of registered keys rather than a "skeleton key" approach simply because the "skeleton key" would be a ridiculously easy target. Of course, this whole scenario cannot work for catching dissidents and criminals, and therefore cannot serve the purpose of fighting terrorists.
The reason is that under any reasonable key-escrow scheme a government would be required to show evidence before using the person's key to find the data. This works fine for average citizens who only use the mandated encryption standard, but, Surprise! When the government uses the key of terrorist Tim to decode his messages, they find that not only did he use the mandated scheme, but he also encrypted his data with his own scheme, which, of course, is unbreakable with current technology. Terrorist Tim wins in two ways here, not only did his data remain secure, but he also managed to waste a large amount of the government's time and resources.
The fact that this is even being proposed shows the ignorance of technology rampant in Congress. I live in NH, maybe I'll write a letter to Senator Gregg.
"He's more machine now than man, twisted and evil."
Making crypto 'safe' with a back door effectively makes it useless. Why would anyone in their right mind use a cryptographic algorithm knowing that a perfect stranger has a 'backdoor pass' to their information? The whole point of crypto is to only allow the intended recipient to view the secret information.
This idea would weaken any cipher that this idea is applied to. Why? Simple. Key recovery in a datastream you haven't ever seen before depends basically on one of 2 things: Brute force, and a little ingenuity. If you know that the cipher has a 'universal backdoor' then each stream encrypted with the cipher will be that much easier to crack -- because the streams will have to be somewhat similar.
What happens when the wrong people get the 'back door' key? You don't think that someone dangerous is going to somehow either recover the key manually, or steal it? Think again. A 'back door' key (or set of keys) of this scope would be too good to pass up. Why bother attempting to recover a key that unlocks one stream, when you can unlock a whole set of streams?
The cat's already out of the bag Why would somebody who really wants to keep information secret use a cipher that didn't keep it secret -- especially when there are so many good ciphers (RC4, Twofish, etc.) that don't have a backdoor? In short -- this is a braindead thought process that will lead the U.S. straight into another disaster.
If you're talking about public key cryptography or some form of key exchange protocol (such as what happens with PGP, SSL, and the like), then, yes, there'll be more than one key that can decrypt the message. PGP already allows you to encrypt a message to more than one recipient; a simple solution would be to require all software to always encrypt to Uncle Sam's key in addition to the intended recipients.
The other solution is to weaken the encryption algorithm in some way. There are very subtle approaches, but the simplest is to limit the length of the key. A 40-bit key takes half as long to crack with brute force as a 41-bit key, and a 42-bit key takes twice as long again (all else being equal). If you have an application that uses 128-bit keys, it could be ``dumbed down'' to a 40-bit key by forcing all keys to start with 88 zeroes (or some other known pattern).
How to get people to use such software when there's a wealth of reliable strong cryptographic software readily available is left as an exercise to the reader.
Most encrypted streams have header information to make identifaction easy for the recipient. If you've ever gotten PGP-signed or -encrypted email, you've seen ``BEGIN PGP MESSAGE'' or some such at the top.
You could, of course, remove all such identification. If the encryption method is strong, what remains is provably indistinguishable from pure noise. If the recipient adds the identifaction back--if she puts ``BEGIN PGP MESSAGE'' before the bits--the result can be fed to the decryption proces without trouble.
But how many people send random bitstreams to each other? Somebody doing so would stand out like a sore thumb against the usual traffic of ASCII.
The most commonly accepted solution is steganography, the art of hiding secrets in plain sight. ``All the twenty clever kings'' could mean ``attack'' if you were to just look at the first letter of every word. Common modern methods of steganography include encoding the message in the low-order bits of a JPEG, but the field is still young and many techniques a bit crude. If ``they'' are already looking at you, ``they'' will have a good chance of finding the message.
As always, Bruce Scnhier's Applied Cryptography is a wonderful resource.
b&
All but God can prove this sentence true.