Card Playing Protocol

Oh, GREAT!
 
Tim says (roughly): "Go for it, too bad you are doomed to lose
interest shortly."
 
Geeze, I hate people who make generalizations which are, well, likely
to be borne out yet one more time.  (I *hate* that!)
 
<Mutter grumble hrumph errrrrrah, phooy!>
 
So I am either supposed to put my tail between my legs now, or take
this as a challenge to "Follow through this time.", or let it soon die
quietly and hope Tim takes mercy and doesn't rub my nose in it.
 
Grrr.  I *hate* reality.
 
 
So here is where I am:
 
1) I am wondering whether a "digital deck of cards" is a good choice.
 
2) If it is, I am wondering how the protocol would roughly be framed
   (Fundamental card operations, etc.), with an eye towards what the
   cryptographic protocols can offer.
 
3) Then, if things make sense, appear tractable, and (drum roll) I
   haven't gone onto fresher blue-sky ideas, I figure out how to start
   building the damn thing.
 
4) And if I ever get to building it I will start first with the little
   pieces (the cryptographic fragments) which might be useful
   individually when I lose interest in building the larger beast.
 
I assume that I will have to do real work at each of these
stages--though I welcome any help.  Both now when the talk is still
cheap and later when the bits hit the disk.
 
 
So far I am at step #1, nudging towards portions of step #2.  I refuse
to be shamed about abandoning step #3 until I have at least embarked
on it.  (Then you can make fun of me.)  Just producing a complete
RFC-quality protocol would be something not to be sniffed at.  In
fact, I am prepared to stop there and *still* feel smug.  (So there!)
 
As for getting people to want to use this digital deck of cards, I
rely on my passion for good user interface design combined with the
continuing popularity of card games.  (And people's continued interest
in playing games with other people rather than just computers.)
 
 
So I am currently at step #.  Is the Card Playing Protocol a good
choice for being:
 
1) cryptographically interesting
2) tractable
3) "harmless"
4) appealing to users?
 
Comments?  (You too Tim.)
 
And Tim, don't worry about my eyes becoming glazed over with images of
Donald Trump.  I don't like The Donald.  Gambling is boring.
(Besides, generalized transactions are far more appealing to a
megalomaniacal fool like me.  How CPP applies remains for me to
understand...)
 
 
-kb, the Kent who is going to be Cometing tomorrow, handy annual open
house at JPL this weekend, etc.
 
 
--
Kent Borg                                                  +1 (617) 776-6899
kentborg@world.std.com
kentborg@aol.com
          Proud to claim 32:00 hours of TV viewing so far in 1994!

Kent Borg writes:

> Tim says (roughly): "Go for it, too bad you are doomed to lose
> interest shortly."
>  
> Geeze, I hate people who make generalizations which are, well, likely
> to be borne out yet one more time.  (I *hate* that!)

No, I think it's a fine project, certainly more useful in the long run
than another PGP shell. But also more complicated, if done right.
(Done right = reusable building blocks for the various needed primitives.)

> So I am either supposed to put my tail between my legs now, or take
> this as a challenge to "Follow through this time.", or let it soon die
> quietly and hope Tim takes mercy and doesn't rub my nose in it.
>  
> Grrr.  I *hate* reality.

Glad you are taking my comments in the spirit in which they were
given. There are some pretty good reasons many of the ideas excitedly
discussed here never reach fruition:

1. No time. Most people have full-time jobs doing other things.

2. No funding sources to _force_ people to complete things they've
already been paid for.

3. No group of co-workers to chat with, to reignite interest, to exert
peer pressure to finish. It's just _so easy_ to let a project kind of
s-l-i-d-e  a-w-a-y...


> So here is where I am:
>  
> 1) I am wondering whether a "digital deck of cards" is a good choice.

Read up on the "playing cards by telephone" papers of the early to
mid-80s. Maybe implementing just one of the sets of ideas would give
your further insights.

> 2) If it is, I am wondering how the protocol would roughly be framed
>    (Fundamental card operations, etc.), with an eye towards what the
>    cryptographic protocols can offer.

That's the central issue. 

> 3) Then, if things make sense, appear tractable, and (drum roll) I
>    haven't gone onto fresher blue-sky ideas, I figure out how to start
>    building the damn thing.
>  
> 4) And if I ever get to building it I will start first with the little
>    pieces (the cryptographic fragments) which might be useful
>    individually when I lose interest in building the larger beast.
>  
> I assume that I will have to do real work at each of these
> stages--though I welcome any help.  Both now when the talk is still
> cheap and later when the bits hit the disk.

Lots of work. Remember, the mathematicians and computer people who did
these papers did not bother to build them into computer code, though
some of them surely could have if it were easy. (Chaum's people built
a running simulation--and crypto simulation is what we're talking
about here--of digital cash, but the version I saw was unusable by
other programs. That is, it was a "user at the console" sort of thing,
not a tool or class library or even a function call.)

What's lacking in crypto is a reasonable "framework" for these
concepts and functions to live it.


> As for getting people to want to use this digital deck of cards, I
> rely on my passion for good user interface design combined with the
> continuing popularity of card games.  (And people's continued interest
> in playing games with other people rather than just computers.)

Good user interface is probably the wrong thing to be thinking about
now, if the goal is wide use. Think "client-server" (or choose your
own paradigm). The building blocks are more important than a snazzy
Windows or Mac interface.

> So I am currently at step #.  Is the Card Playing Protocol a good
> choice for being:
>  
> 1) cryptographically interesting

Yes,

> 2) tractable

Unknown.

> 3) "harmless"

Not a real issue.

> 4) appealing to users?

For researchers, it would be interesting to have the set of
abstractions reified into running code. This is a longstanding
interest of many of us, and was one of the motivations two years ago
to form the Cypherpunks group. Eric and I figured it was high time to
take the various theoretical abstractions and implement them in code;
we hoped that a bunch of people would generate "Pretty Good Digital
Money," "Pretty Good DC-Nets," etc. So far, it's been slow. (And some
actual deployments, such as Digital Money, have faltered for other
reasons. Kent should look at MM and why it isn't in wider use and try
to learn some lessons for a gambling scheme.)

> Comments?  (You too Tim.)

See above.

--Tim May


-- 
..........................................................................
Timothy C. May         | Crypto Anarchy: encryption, digital money,  
tcmay@netcom.com       | anonymous networks, digital pseudonyms, zero
408-688-5409           | knowledge, reputations, information markets, 
W.A.S.T.E.: Aptos, CA  | black markets, collapse of governments.
Higher Power: 2^859433 | Public Key: PGP and MailSafe available.
"National borders are just speed bumps on the information superhighway."

1) I am wondering whether a "digital deck of cards" is a good choice.

Premature abstraction is a severe problem if it happens to you.  Read
some of the literature to get an idea of the techniques before you
pick an abstraction.  Your remarks about knowledge models for an
abstraction proposal of "a table with stacks of cards" seem on target.

Most card games require a random permutation, mutually trusted to be
random, which can be revealed one card at a time.  That permutation
need not be generated in advance.  Games like Magic--The Gathering in
which each player shuffles their own deck, are easier to implement and
only require bit committment.

The revealing of cards cannot be global, since at the beginning each
player sees only their own cards.  The revealing of cards should
require that the cooperation of each player that sees the cards, and
possibly some others.

Time to read crypto.

Eric