Re: MD5 To Be Considered Harmful Someday

On Tue, 7 Dec 2004, Tim wrote:

> Due to the slowness of public key, most digital signatures are
> performed on a digest of the original document.

I think it's time to change the way we do digital signatures to
compute several hashes of the original document using different
algorithms (SHA1, MD5, ...) and sign the concatenation of the resulting
digests. Does any standard signing software do this yet?

Also, it's probably a good idea to alter in some small and
unpredictable way anything that you're asked to sign. (Add some
spaces, add text like "Signed by David F. Skoll", etc.) This makes it
impossible to precompute two versions that hash to the same value,
though it's still not very good protection if it becomes easy to find
MD5 collisions.

Regards,

David.
Received on Dec 09 2004