From: dlv@bwalk.dm.com (Dr.Dimitri Vulis KOTM)
Date: Sun, 26 Jan 1997 19:30:27 -0800 (PST)
To: cypherpunks@toad.com
Subject: Re: Fighting the cybercensor
In-Reply-To: <199701270207.UAA00452@manifold.algebra.com>
Message-ID: <TF561D82w165w@bwalk.dm.com>
MIME-Version: 1.0
Content-Type: text/plain


ichudov@algebra.com (Igor Chudov @ home) writes:
>
> Jim, why don't you stop bullshitting and write a real assassination
> bot. [as a beta, it can be a mailbombing bot] This bot would:

I think this would be a very good demo project, but mailbombing may not
be the best choice...

> 1) Accept bets as combinations of
> 	a) Some amount of cybercash

Be careful about violating Chaumian patents... How about using funny
money in the prototype? Or, look at some of the micropayment schemes.

> 	b) A string that identifies an event that should happen
> 	   such as "domain X is mailbombed"

Not a good idea (see below).

> 	c) [optional] date of that event (no date means that you always
>            lose)

I think the date must always be specified, and the event must occur
(or not occur) on or before that date.

> 	d) Return address (possibly a nym address) to send
> 	   all cash from UNSUCCESSSFUL bets for the event in
> 	   question.
> 	e) [optional] time limit after which the cash will be refunded.
>
> Note that for simplicity, the bot should identify the event as
> a unique string, without any understanding of any semantics of that
> string.

I think we should think about the kinds of events that a 'bot can verify.

> 2) Store these bets in a database.
>
> 3) Have a trusted party (someone really honest, like myself) report to
> the bot the signed strings that, in the opinion of the trusted party,
> are "true".

Why not start with a less destructive event... For example,
"on or before <date> a Usenet article will appear in newsgroup X
saying Y".  That's something the 'bot can verify and anyone with
access to dejanews and the like can confirm. Eliminating the need
for a trusted human is always desirable.

> 4) Upon receipt of such event notifications, the bot will find all bets
> and forward them to the better whose date prediction was the closest.
> If several betters predicted the same date, the money is split between
> them in proportion to the amount of moneys submitted.

Have you ever dealt with a bookie?

I think there need to be two distinct operations:

1. A user can create a new kind of event. For a fixed fee F, one can enter
a new event into the table of events that can be bet on. (In a more
generalized system, the creator might also specify the third party that
determines whether or not the event took place.)

A human bookie decides which events can be bet on (based mostly on the
tradition and supply/demand). Here we let users bet on anything they
want as long as they're willing to may be bookiebot for keeping track
of who bet how much $ that the event will or will not happen.

In fact, when creating an event, the user must immediately bet an amount
>$F and the house enters an opposing bet for $F (or slightly less).

2. A user can bet $B that an existing event will/will not happen.
Bookiebot accepts the $B and promises to pay back an amount that's a
function of $B and the current amounts bet so far on yes and no
(or escrows the winning with a 3rd party).

I'll let Jim et al figure out how to compute the odds when there are
offsetting bets for the same event at two different times.
E.g. E1:"Saddam Hussen will die before April 1" and E2:"ditto June 1"; if the
first one occurs, then the second one occurs too. If a lot of money is
bet on E1, it should somehow affect E2 odds too. Also the bookiebot
should never lose money no matter what the outcome; all the winnings
should come from the losers' bets.

> Examples of use: Suppose I do not like The Right Reverend Colin James III,
> cjames@cec-services.com. I have a lot of money, but do not know how to
> mailbomb. I set some nym address as my return address (for refunds if
> CJ3 is not mailbombed within half a year).
>
> I place a bet with $1000 worth of money and phrase "domain cec-services.com
> disabled". The date would be open which means that I will always be the
> loser. I also post a message (anonymously) saying that anyone who wants
> to mailbomb TRRCJ3 can be rewarded through your assassination bot.

Here's an improved scenario. Say I pay the bot $10 to create the event
"a homophobic article will appear in soc.motss by April 1".
Then I bet $1000 that the event will NOT occur to skew the odds.

> Someone with more knowledge of computers, a T1 link and no money will
> be lured, submit a bet for, say, Feb 1, and on the 1st will start fierce
> mailbombing of cec-services.com. The return address will, of course, be
> a nym.

Someone looking to make a quick buck browses through the list of events
and odds in the bookiebot and sees the very skewed odds for a homophobic
article on soc.motss. He bets a small amount that the article will appear,
so he gets really good odds. Then he posts an article that's recognized
as the event, and collects a winning much larger than his bet.

But if this doesn't happen, I get almost all of my $1000 investment back.
In either event the bookiebot made a small profit for its owner.

---

Dr.Dimitri Vulis KOTM
Brighton Beach Boardwalk BBS, Forest Hills, N.Y.: +1-718-261-2013, 14.4Kbps