Still - it looks like Clarke's first law could easily apply: "When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong."

A practical application of this is at http://www.cypherpunks.ca/otr/ (with a plugin for a few common AIM application, most usefully for pidgin née gaim).

This one has an implementation called the "Socialist Millionaires Problem", which sounds the same, although I recall it being used only to tell if two secret values are the same on both side, thus augmenting the key exchange protocol with man-in-the-middle detection capabilities, provided the parties has shared knowledge about something (and something reasonably private).