Re: Card Playing Protocol?

hughes@ah.com wrote some interesting stuff:
>Duplicate games won't work

Damn!  People are paying attention.  It was an off-hand remark.  Any
bells and whistles along those lines are certainly banned from any
early version.

>There is a non-crypto issue of how one finds playing partners without
>a central server.

My mind wandered to that very point this very morning.  The simplist
way to find players is the same we currently find email addresses: the
hard way.  Type in the addresses of the other players.  (Assuming the
software is already running on those nodes, those players would not
have to retype the other addresses, accepting the invitation to play
would be more like a single "click".)

I think anything more elaborate along these lines is a candidate for
banning from 1.0.  (One problem is that the "I'm looking for a
game."-problem is at least as big and interesting as building a deck
of cards.)

>I would strongly suggest the separation of the communications, user
>presentation, and decision parts of the client software.

And that is one of the wonderful sort of engineering problems I love:
keeping the different parts clear of each other's private parts yet
still considerate of their desires and needs.

>client software

My instinct is for a peer-to-peer design.  Yes, they will serve each
other cards, etc., but I would like to avoid the user confusion of
having two different sorts of software needed.  (At a comms protocol
level there might always be a single server per game--I don't know
yet--but I would like to hide that sort of stuff from users.)


-kb, the Kent who will be driving to Pasadena early in the morning,
but not to watch soccer.


--
Kent Borg                                                  +1 (617) 776-6899
kentborg@world.std.com                                
kentborg@aol.com                                      
          Proud to claim 35:00 hours of TV viewing so far in 1994!

(At a comms protocol
   level there might always be a single server per game--I don't know
   yet--but I would like to hide that sort of stuff from users.)

There's no need for a central server per game, even running on one of
the player's own machines.  What is possible with crypto is completely
flat distribution of the simulation.  The difference is profound.  I
would suggest that all who don't understand this meditate upon coin
flipping protocols, the simplest flatly distributed simulation--here,
of a random number generator.

Eric

On Sun, 17 Jul 1994, Kent Borg wrote:

> Damn!  People are paying attention.  It was an off-hand remark.  Any 
> bells and whistles along those lines are certainly banned from any 
> early version. 
>
> My mind wandered to that very point this very morning.  The simplist 
> way to find players is the same we currently find email addresses: the 
> hard way.  Type in the addresses of the other players. (Assuming the 
> software is already running on those nodes, those players would not 
> have to retype the other addresses, accepting the invitation to play 
> would be more like a single "click".) 
> 
> I think anything more elaborate along these lines is a candidate for 
> banning from 1.0.  (One problem is that the "I'm looking for a 
> game."-problem is at least as big and interesting as building a deck 
> of cards.) 
> 

The only problem is if a government spy is listening on this 
conversation, he is going to learn how to play this game to and learn how 
to intercept the messages and therefore learn how to decode the messages...
 
Aaron