3 ms·

"Most systems are not designed for cryptographic agility" but even if they were, that's yet another vector for attack. If your software is configurable in some

by delinka 13y ago

"Most systems are not designed for cryptographic agility" but even if they were, that's yet another vector for attack. If your software is configurable in some way around cryptography (new algorithms in a shared library, new configuration files describing padding or negotiation methods), the the attacker just has to reconfigure your device. When you distribute those changes, they have to be signed; with a configuration currently supported by the app; which probably involves the weak or broken algorithm you're trying to replace. The Problem® as I see it is one of computational power. Everyone carries mobile devices. Everyone expects their mobile devices to be secure. But everyone appears to be more interested in the perceived speed of their devices. How do we make most algorithms more secure? Larger keys, or some other form of increased work for attackers. That slows down smartphones and now people are having a Bad User Experience™ and well, we can't have that. It seems like every time someone makes security more user-friendly, they necessarily introduce weaknesses into the entire system. I gave up on trying to solve these kinds of problems long ago when I realized I was terrible at it.

gizmo686 13y ago
For the most part, the concerns over breaking crypto are related to breaking public key cryptography. Becuase of its relative slowness, public key crypto is generally used for key exchange before the devices start using symetric cryptography. Symetric crypto has its own concerns because we have no theoretical foundation for believing it is secure other than the fact that we have yet to break it. At least RSA is backed by factoring, which we have spent much longer trying to break (and also haven't proven to be secure).
- damarquis 13y ago
  Nitpick: RSA isn't known to be backed by factoring. More precisely, there is no proof that if textbook-RSA encrypted messages could be decrypted quickly that integer factorization could also be done quickly. (Of course if you can factor integers you can decrypt RSA messages quickly). http://en.wikipedia.org/wiki/RSA_problem http://en.wikipedia.org/wiki/RSA_problem There are encryption schemes where decyption of an encrypted message is provably equivalent to factoring. http://en.wikipedia.org/wiki/Rabin_encryption http://en.wikipedia.org/wiki/Rabin_encryption

by delinka 13y ago

gizmo686 13y ago

For the most part, the concerns over breaking crypto are related to breaking public key cryptography. Becuase of its relative slowness, public key crypto is generally used for key exchange before the devices start using symetric cryptography. Symetric crypto has its own concerns because we have no theoretical foundation for believing it is secure other than the fact that we have yet to break it. At least RSA is backed by factoring, which we have spent much longer trying to break (and also haven't proven to be secure).

damarquis 13y ago
Nitpick: RSA isn't known to be backed by factoring. More precisely, there is no proof that if textbook-RSA encrypted messages could be decrypted quickly that integer factorization could also be done quickly. (Of course if you can factor integers you can decrypt RSA messages quickly). http://en.wikipedia.org/wiki/RSA_problem http://en.wikipedia.org/wiki/RSA_problem There are encryption schemes where decyption of an encrypted message is provably equivalent to factoring. http://en.wikipedia.org/wiki/Rabin_encryption http://en.wikipedia.org/wiki/Rabin_encryption