Card Playing Protocol?

[hughes@ah.com (Eric Hughes)]

   only notices that she can now be part of the World Wide Duplicate
   Bridge Tournament that she heard about on All Things Considered.

Duplicate games won't work on the net because the assumption is that
the players have no advance knowledge of the cards of the other
players.  Even if the same hand is dealt simultaneously to multiple
virtual tables., the differences in order of play will reveal cards
early for some players.

The hole is the sharing of information between players.

Duplicate could still be supported with physically based, but
distributed, rooms of play, using the Internet for logistical support.

   Is there a flavor of effort I forgot?

There is a non-crypto issue of how one finds playing partners without
a central server.  An IRC channel seems to have the right properties:
real-time, centrality of name, distributed information paths.  IRC
might be able to be hacked into directly.

The code to find of playing partners should integrate digital
signatures for identity, in order to make possible long scale
tournament play.

Mutual agreement should be required for the formation of a group.
Automatic agreement can always be implemented in client software.
There is likely an interesting protocol here for the negotiation of
group formation without revealing preferences that are not manifested
in the creation of a group.

I would strongly suggest the separation of the communications, user
presentation, and decision parts of the client software.  Folks should
be able to pick the presentation of the cards that they want: table
layout, card backs, etc.  Decision in current card games is currently
all by user input; the user sees the cards, decides what to do, and
clicks.  People will want to try out card playing algorithms, and you
might as well leave a hook in for them.

Eric