public key size

Hi, everyone! :)

On Wednesday, August 06, 2003 16:28, Lance James wrote:
> Now I know what was throwing me off and why I was insistant on 4096 bit
> primes.
>
> Looking through the list of primes here, please take a look at 4096 and
> next to it it's session key equivalency. If we are to be using 256 bit
> session keys for AES, I would think we want to exchange it with the same
> equivalency. Next to each prime size is the size of the prime, and the
> session key equivalency.
>
> Let's adhere, and just resort to 128AES/2048ELG.
>
> Lance
[snip!]
> 		2048, 160, //160
> 		FFFFFFFF FFFFFFFF C90FDAA2 2168C234 C4C6628B 80DC1CD1

	OK. You'll have to forgive my ignorance in this area, but I want to 
understand this. :)
	First of all, the symmetric key size equivalent to 2048 bit ElG listed below 
if 160 bits, not 128. Granted, AES can't make a 160 bit key (the next highest 
size about 128 bits is 192 bits), but 128 bits is still weaker.
	Secondly, why should we use a smaller AES key in the first place? If the 
answer is that using anything larger would mean the ElG key is weaker than 
the symmetric key it protects, then I have one objection. ElGamal has been 
around much longer than AES, and it has still not been broken. It is most 
likely AES will become vulnerable before ElGamal does. AES has already shown 
small weakness [1] (though it's currently not enough to make it vulnerable). 
So 160 bits might be enough _now_, but that might not be enough later.
	So why not use a 256 bit AES key? (If the answer is size, I'd like to remind 
everyone that we're using a 32 byte block for an IV in the Payload structure 
when the size of the IV only needs to be 16 bytes. Coincidentally, a 256 bit 
AES key is 32 bytes, and a 128 bit key is 16 bytes. I don't think the answer 
is size.)

	Anyway, that's my puzzling. What am I missing here? nop? thecrypto? Does 
someone want to educate this poor hacker? :)

	-- Hezekiah

[1] http://www.counterpane.com/crypto-gram-0209.html#1
	This is a link to a CryptoGram article in which Bruce Schneier mentions the 
attack. More links concerning the attack can be found at the bottom of the 
article.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQA/McyGeHiZTMH32ioRAuTXAJ97OgRAmB6kp1rLNFyT08lr5GSrVQCdGTpe
dDrtWj15/cvTyNsWKZtxnD0=
=e0Da
-----END PGP SIGNATURE-----