More on The Elevator Problem: I'll confess that we're working on a new private key encryption product. Free and clear of existing claims on intellectual property, I'd like a protocol that answers the elevator problem: how can Alice and Bob agree on a private key via a compromised channel? I don't want them to meet under a street lamp and whisper in each other's ears _before_ trying to establish a secret on a crowded elevator. I'm willing to accept that the basic question is undoable without stepping on DH et al., I just don't want to give up until all parties are convinced it can't be done. As I stated in my original post, we're willing to consider less-than-perfect implimentations if we can quantify mathematically the risk involved. In some scenarios, Alice and Bob might not expect any Eve's in the crowded elevator or value the content at such a price that they're willing to live with less than 100%. As long as we can document the risk, the end user can make the decision as to whether the risk is worth it. Something that I neglected to mention before is that Alice and Bob do have a unique ID within this system, not that they necessarily know each others. If one assumes that they do, they then share a secret: the knowledge of each other's unique ID. Another segment of our attention is being spent daydreaming about the potential applications of a secondary channel in the cleartext portion of a crypto "package" or "bundle". The comment about secondary channels that I believe Tim made regarding the inmates and the warden is something we discuss in our spare time. Then again the wheels of justice doth turn and patents ain't forever... Jerry Whiting jwhiting@azalea.com