Dave Aitel wrote: >On Mon, 2002-10-14 at 14:40, Dan Kaminsky wrote: > > > >>> >>> >>> >>> >>For remotely computed data / hashes, you can't -- thus the folly of >>trusting MD5 hashes on critical files downloaded off of untrusted >>servers. If somebody can modify the tarball, they can probably modify >>the hash too. >> >> > >Well, not always, if there is a semi-trusted third party or two - see >http://www.immunitysec.com/hashdb.html for one implementation of this >sort of thing. > > > Cool stuff there! Maybe host the DB over DNS or something trivial. hash.filename.immunitysec.com :-) Incidentally, Bitzi was/is trying to do something like your stuff for arbitrary data -- they didn't care what(P2P), they just hosted the translation between hash to content. Genuinely cool crypto, using Merkle's old Hash Tree concept. The great thing about hash trees is that you don't need the entire file to find out you're being fed bad data. I believe Bitzi opened their code, too: www.bitzi.com. --Dan
This archive was generated by hypermail 2b30 : Mon Oct 14 2002 - 13:08:22 PDT