From: rjc@gnu.ai.mit.edu (Ray)
Date: Sun, 3 Apr 94 04:07:56 PDT
To: cypherpunks@toad.com
Subject: DEATH TO THE
Message-ID: <9404031107.AA14433@geech.gnu.ai.mit.edu>
MIME-Version: 1.0
Content-Type: text/plain



   re: Hal's suggestion

   I don't think that banning multiple recipients solves the whole
problem which is the spamming of the remailer network by cybervandals
like Detweiler. What needs to be done is to create some system that
"charges" remailer requests to the sender which then limits the
size or frequency of their messages based on their ability to pay.
For a play money system, the issue is, how do you distribute the
play money accounts without letting spammers open up as many accounts
as they like? (e.g. if Detweiler spends his one account dry with
his cyberterrorism, how do we prevent him from opening up an unlimited
number of accounts without trying to attach accounts to a user's
e-mail address?)
   By making remailers charge users, and even other remailers on the
network, we can also prevent rogue remailers from allowing Detweiler
attacks through the network (since they'd have to pay for his
traffic)

   Here's a nutty idea I had one night which would accomplish these
goals and also encourage more people to run remailers:

o each remailer upon being setup distributes a large but finite amount of
digicash certificates. These certificates are one use only. The
initial distribution and price system can be altered to tailor the
usage of the remailer to the owner's wishes.

o after the initial distribution, the remailer distributes
a somewhat smaller amount of digicash each renewal period (probably
once a week) This smaller amount is sufficient for casual use, but
not abuse

o if you want more than what a remailer is willing to give you, you
must trade service for it. That is, you must set up your own remailer
and issue some cash to the other remailer which can be redeemed for
chaining service. Example: Bob was unsatisified with his $10 of
Ann's remailer coupons, so he set up his own remailer and issued
$100 worth of service to Ann, in return, she gave him $50 of service
for her remailer. (obviously, her remailer is more popular and is more
in demand so his is only worth 1/2 of hers. Ann could use bob's
coupons to either chain requests through him, or she could sell them
off to other remailer operators.)

  The hard part is making sure that Bob doesn't cheat, offer to
run a remailer, issue $X worth of credit to Ann, and then just
file her remail requests to /dev/null  I'm hopeful that a "consumer reports"
like organization will pop up which periodically tests remailers to
make sure they aren't cheating. (besides, the remailer network itself
should do such testing with fake traffic)

  Call it "the free market of remailers"

Other issues: How to distribute coupons/stamps/dollars? First come
first serve? Popularity/Reputation? Reputation market? We want to
prevent people from collecting digicash certificates multiple times
during distributions, but at the same time, we don't want to use
their real e-mail address. So in a system where users can create as
many identities as they wish, how can we achieve a fair distribution?

  The only thing I can think of is to distribute cash to users based
on their reputation or trust level. If a Detweiler is found abusing
remailers, operators simply stop giving him cash for their remailers.
Remailers which sanction Detweilers end up paying for them in the
long run. (either because he runs up the costs for the remailer to use
other remailers on the network, or because the operator gets too
many complaints from systems which it directly delivered Detweiler
mail to)

  Anyway, limiting multiple recipients will stop geometric growth,
but it won't prevent Detweiler from hacking up a script to send 
a hundred thousand remail requests posting to every newsgroup and
mailing list out there. Remailers would also have to limit the amount
of remailer requests allowed per day, but this would still allow
Detweiler to spam up the system by preventing anyone else from using
it (by running out the global quota) Some sort of authentication is
needly, IMHO.

-Ray












-- Ray Cromwell        |    Engineering is the implementation of science;   --
-- rjc@gnu.ai.mit.edu  |       politics is the implementation of faith.     --