🚀 go-pugleaf

In article <1992Dec22.222639.26141@cbfsb.cb.att.com> osan@cbnewsb.cb.att.com (Mr. X) writes:
>Would multiple encryptions add to the level of security of, say, a PK system
>like RSA or a knapsack scheme?  What I mean is to encrypt the clear text and

I will talk only about knapsack, as I am currently having literature around
which covers that facette :-)
It seems that, although first rated computationally infaisible (before
Shamir did show a way to break specifc single-iterated knapsacks), Ernie
Brickwell broke a knapsack-system of 40 iterations and hundred weights
in the cargo-vector. (Taking about 1 hour of cray-1 CPU) 
E.F. Brickwell  "Breaking iterated knapsacks" Procedings of Crypto '84
(vol 12, pp. 342-358)
I do not know how much knapsack could be 'strengthened' by using independant
keys for each iteration, but I assume knapsack is _broken_.

>then encrypt the encrypted text, probably with a different key.  If encryption

With different keys. Using the same key more than once could cause trouble.
(I guess the resulting operation would not be weaker than an only-once
encryption, but does not actually need to be stronger!)
Somebody with more theoretical knowledge, please tell us why this is so ;-)

But I _guess!_ that using different keys, and the encryption mechanism NOT
being a group (wasn't there some posting about multiple DES - encryptions
which talked about the possible problems which did not rise in DES?) you
actually create a wider keyspace.

>offer the equivalent of encryption with very large keys (say, >10000 digits).

At which computational costs ? I guess the method is good and slow .-)

Don't take me for an expert, I am only doing this for fun, and speculating
here.
Friendly greetings & a happy new year,

	/gec



p.s. there are no stupid questions. only stupid answers :-)

-- 
Instruments register only through things they're designed to register.
Space still contains infinite unknowns.

Germano Caronni    caronni@nessie.cs.id.ethz.ch