From: ichudov@algebra.com
Date: Mon, 3 Feb 1997 12:11:48 -0800 (PST)
To: Peter Hendrickson <pdh@best.com>
Subject: Re: Dissolving Choke Points
Message-ID: <199702032011.MAA01076@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


i do have unix (linux) and stuff, but i can't take a lot of subscribers
-- maybe 200-300 or so.

i actually wrote a proposal for a mailing list without a central control
point, with several advantages being impossibility of control, absense
of a single point of failure, and cryptographic verification of honesty 
of moderators.

if there is any interest, i will post it here.

igor

Peter Hendrickson wrote:
> 
> I was amused by the similarities of USG crypto policy and the
> moderation of the cypherpunks list.  Recently a number of new
> regulations were announced to go into effect by a certain date.
> Requests for comments were made after the policy was announced.  Many
> of us thought that was pretty sneaky.  I feel the same way about the
> moderation plan.
> 
> Moderation has been a failure.  I'm pretty good at filtering and I
> can sadly report that there is very little signal out there.
> 
> There are lessons to be learned here.  One is that censorship does
> not promote a stimulating and creative dialogue.  The cypherpunks list
> right now is about as interesting as hanging out by the 7-11.
> 
> Another lesson is the danger of choke points.  We can see how
> tempting it is for people to exercise their control.  Even John Gilmore
> was unable to restrain himself from involuntary social engineering
> experiments.  Who would we have considered to be more trustworthy?
> 
> Toad.com is a choke point, not just in terms of moderation but in
> terms of the rate at which it can distribute messages.  Let's
> replace it.
> 
> What we want are many machines carrying the cypherpunks list.  A
> message posted to any machine goes to all of the others.  Each
> machine sends messages to its subscribers only once.  Some of
> these machines should be across borders.
> 
> The mail loop and multiple posting problems are solved by observing
> the message IDs.
> 
> Fast implementation: use moderated mailing list software.  Put a
> filter in the .forward file of the "moderator" account which looks
> at the message ID and forwards the message if it hasn't been seen
> already.  The mailing list machines all subscribe each other.
> 
> I've been looking for a stable machine with a good net connection to
> do this.  I haven't found one.  However, if we have many machines
> sharing the load, the stability of any one unit is not as important
> because the list will survive multiple "hits".  Only the subscribers
> on one machine will be affected by having their messages delayed.
> This greatly reduces the work and responsibility for any one list
> operator.  (As John will attest, keeping a machine running 24 hours a
> day, 7 days a week, rain or shine, is a lot of work.)
> 
> Also, with multiple machines, each unit handles a small amount of the
> load.  This makes more machines available and has less impact on
> people's net connections.
> 
> Last I checked, there were about 1200 addresses on the mailing list at
> toad.com.  All we need are about 10 machines to take 120 subscribers
> each.  (This is a completely manageable load.)
> 
> Do you have a Unix machine on the Net?  Does it have sendmail and
> Perl?  Then you have all that it takes to participate.  Send me
> mail and I'll help you set it up.
> 
> Peter Hendrickson
> ph@netcom.com
> 
> P.S. I like and respect John and Sandy and I've learned a lot from
> both of them.  While basically well-intentioned, they just made a
> mistake in this instance.
> 
> 



	- Igor.