Re: ecash lottery (Was: ecash casino)

[Bill Stewart <stewarts@ix.netcom.com>]

Ian wrote:
>>The idea behind my proposal was that any participant can determine if he has
>>won.  The winning number (a hash of which was published beforehand, as in
>>a "bit commitment" scheme) is announced.  Anyone who picked a number
>>that, say, matched in the last digit, wins $2.  If you matched the
>>last two digits, you win $10, etc.  The house (without sacrificing
>>reputation) can't arrange who the big payoffs will go to, nor can it
>>(as I think Tim suggested) give out lots of small prizes and no big ones.

>There are two main kinds of lotteries - systems where many people have a 
>moderate chance of winning a fixed amount of money, and lotto-systems where
the winners split a jackpot that's based on current and/or previous players' 
losses\\\\\\ ticket purchases.  The former kind, such as numbers games,
works because the expected payoff is less than the ticket price -
you can make it honest through techniques like bit committment,
but as Tim points out, the kind of people who care about things like
bit committment know enough math to know these games are money-losers.
Perhaps you can attract some people with a flashy user interface,
but it's not like really being in Vegas with booze and babes and blinkenlights
and incessant noise and stale smoke and Siegfried and Roy's tiger act.

On the other hand, lotto systems are too easy to cheat with anonymous payees.
Sure, you can make sure that any genuine winners get paid, 
but you can also throw in the occasional ringer to share the prize,
and nobody can tell unless you do it often enough that people start checking
the probabilities.  Bit committment doesn't prevent this kind of cheating;
it just verifies that you knew the number in advance (oops :-).

You need some sort of externally verifiable source of random numbers;
an MD5 hash of the Wall Street Journal stock price web page from
a few hours after the last bet and digital timestamping or something
might be good enough.  (Of course, a single unified stock price page
becomes even less meaningful with large volumes of on-line anonymous private
stock trading.)

(The pre-digital version of this scam was to hire some poor old grandma from the
South Bronx as the shill, since nobody'd accuse Granny of conspiring with
the mob.
It wasn't really enough then, and now Granny may be using anonymous digicash
for her payoffs...)

Another way to cheat progressive lotto systems is to not report all
your ticket sales each week, since anonymity makes it excessively hard
to verify.  Committing to the list of all the numbers bet almost works,
but you could still skim a few percent by reporting smaller sales of
popularly chosen numbers, with the risk that you've occasionally got to
pay all six real winners 1/5 of the jackpot if you're unlucky.
You can't get around this by using large enough numbers that there's
a small probability of duplication, because that also reduces the 
probability of a win too far, and because some folks are going to pick
7777777...777 regardless of how many bits long their bet is.

Government lotteries are surprisingly free from this sort of problem.
Not because governments are excessively honest, of course :-)
Employee pay isn't generally dependent on profitability, 
there's enough public scrutiny that it's harder for employees to cheat
for their own gain (rather than their lotto organizations'),
and (perhaps most important) they're already taking such an appallingly
high cut of the winnings that there's no need for regular cheating -
it's one of the efficiencies possible in a highly unnatural monopoly.

But if all else fails, you might be able to talk Scott Adams into giving
you the World Wide Web franchise for Dogbert's Value-Priced Lottery Tickets..

#--
#				Thanks;  Bill
# Bill Stewart, Freelance Information Architect, stewarts@ix.netcom.com
# Phone +1-510-247-0663 Pager/Voicemail 1-408-787-1281