Re: META: traffic-limited list idea

Lee Daniel Crocker (lcrocker@calweb.com)
Fri, 11 Apr 1997 16:52:53 -0700 (PDT)


You're probably right that a simpler non-crypto honor-system first
implementation would help prove the idea and find new bugs. Here's
a specific proposal that I could implement in a few weeks, given
the cooperation of the server owners:

The existing lists stay as they are. A new list, say "exi-paid" or
something like that, is created. Everything posted to the new list,
paid or otherwise, is copied to the old list so the existing archive
becomes the complete open archive. Some people may still want to
subscribe to that, or else just occasionally read it on the Web to
find things to spend their tokens on.

Tokens are simply 5- or 6-digit numbers (depending on traffic). To
pay for a posting, the number is placed in the text of the message
in square brackets: [73567] anywhere in the post. Shouldn't be a
problem for anyone with any mailer to do that.

The robot that recieves postings to exi-paid looks for the token,
matches it against its list of outstanding tokens, replaces the text
with [PAID] or something, removes that token from the outstanding
list, and forwards it to the archive and to the paid list.

A robot running daily or as a cron job on lucifer walks through the
subscriber list, and at each name tosses the dice to decide whether
to send that person a token (the probability is adjusted at will to
control traffic). If it decides to send one, it generates a random
token different from any outstanding one, and then sends mail to the
person selected that looks something like:

You have been issued the token [54637] for the discussion list
exi-paid@extropians.org. If you wish to post to that list,
just include this number (along with the brackets) in some line
of your post. It can only be sent once, and will be retired
when it is used. If you want to give this token to someone
else, just forward them this message or tell them the number.
If you want this token to be reissued randomly, reply to this
message and change the subject line to the word "REISSUE".
See (some URL) for further details.

Note that even non-members can post if they get a token from
someone. Those who are broke can post to the unpaid list a message
with a title beginning with "BEG: <subject>", trying to convince
others to fund them.

The token mint logs every issuance and every redemption, and maybe
produces statistics we can use to tune the process. We may want to
expire very old ones, and perhaps notify people who were issued them
before we do.

-- 
Lee Daniel Crocker <lee@piclab.com>  <http://www.piclab.com/lcrocker.html>
"All inventions or works of authorship original to me, herein and past,
are placed irrevocably in the public domain, and may be used or modified
for any purpose, without permission, attribution, or notification."--LDC

> I like this idea. It is worth trying; the main problem is getting > someone to implement it. So one should strive for a plausible > *simple* design. > > I don't think the quoting issue is a big problem; authors who don't > quote enough won't be understood, and authors don't like that. > Similarly while a per page or line charge might be best, a per post > charge seems workable and much simpler; overly long posts will not get > read. > > A non crypto first implementation also seems fine. Just go down the > subscriber list at some rate sending out random ASCII strings of > length 30, which people can forward to whomever they like. The big > problem is not people stealing tokens from peeking at messages; it is > people creating new addresses to get more tokens. So I suggest tokens > only go to subscribers who have posted to the list within the last > month (or three months?), with the first post per such period being > free. > > Having tokens expire after some period might be more robust to dealing > with unknown accumlations of tokens. Then it would be fine to have a > target number of posts per week, and use a one or two week moving > average of the recent posting rate to adjust the rate of token > distribution. > > The token for a new post should probably go in a special header line. >