Re: [md5] string giving specific hash

"Nathan" <ntspam2@xxxxxxxxxxxx> writes:

I'm too lazy to do any checking, but I don't think this should work
any better than just testing random strings. A cryptographic hash
is supposed to be very sensitive to even small changes, so there's
no connection between the hashes of strings that differ in one byte.
I would think a genetic algorithm would make sense on a problem
where it's possible to make some kind of gradual progress.

I believe that the point is that *if* there are any inherent weaknesses
in MD5 (small changes => related changes) then a genetic algorithms'
approach may have some success in locating and exploiting them.
If all is not random, a GA may have some greater success than randomness.

I don't know if MD5, or even SHA1, have been *proven* to have no such
weaknesses.

--
Chris.
.

Quantcast