Re: Probably naive question - SHA1 + MD5 combination

Christian Siebert <iBBiS@xxxxxx> wrote:
Multiple hash functions isn't a good idea.

What if we can find a set of hash functions that can be proven to be
independent? Can't we combine them to create a stronger hash function?

The only "at least as strong as" construction I know of is

g(m) = h_1(m) || ... || h_n(m) .

If you find a collision in this hash function, you've also found a
collision for all the constituent hash functions.

'C(M) = (H(M) + R(M) + S(M) + T(M)) mod 2^160'.

This construction may be stronger or weaker than the constituent hash
functions on their own. You would expect it to be stronger, but for a
quadrupling of computation cost, couldn't you do better (add more rounds
to SHA-1, for instance)?

If yes, how difficult would it be to prove
independence of hash functions?

I have no idea, since "independence of hash functions" isn't a
well-defined concept for me.

--
Kristian Gjøsteen
.