Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site ucbvax.BERKELEY.EDU
Path: utzoo!watmath!clyde!burl!ulysses!ucbvax!athena.mit.edu!jh
From: jh@ATHENA.MIT.EDU
Newsgroups: net.micro.atari
Subject: Re: news being bounced off arpanet! (sort of long)
Message-ID: <8510261755.AA04337@MIT-ARTEMIS>
Date: Sat, 26-Oct-85 13:55:07 EST
Article-I.D.: MIT-ARTE.8510261755.AA04337
Posted: Sat Oct 26 13:55:07 1985
Date-Received: Mon, 28-Oct-85 03:15:37 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 81

The reason we can't follow your suggestion is simple.  A mailing list
like ours is a simple reassignment of names; my name and the names of
the other arpa readers get put in a recipient field, and you are still
the sender (Which is why I can reply to you now.).  This is fine for
small groups, as no one has to go through the hassle of moderating the
list.  Any mail you send is exactly like sending n messages, one to
each person on the list.  You would get a message from the appropriate
mail demon if one of those messages were unsuccessful in getting to its
recipient, and judging from your complaints, this is indeed what has
been happening (Believe me, from a list this size, be glad there are
only eight irresponsible individuals who didn't tell anyone when their
mail addresses changed!).  This format was really designed only to
work for a small group.  I am on and run several such groups, one for
my living group and its alumni, one for an experimental program I'm
in at school, etc. each with less than 100 people on it.  It would be
a real pain to readdress each message 80 times by hand in order to get
to those same people, though.  Using this format for a large list,
especially one spread out over the country, is going to run into some
trouble of the sort you just described.  That is why there are
digests.  Digests are simply all the messages sent to the list put in
a single (long) mail message, so that the recipient can "undigestify"
them and read them.  A typical digest will go out two to ten times a week.
The benefit is that the person who "digestifies" (digests?) the
messages is the one who resends them to the list, and is thus the one
who receives all the mailbounce.  The problem is that someone has to
do a LOT of work (this list would require a new digest every 2-4
days).
	This arpa list is of the first kind, which gets us messages
faster, but has obvious drawbacks.  For one, we now have NO moderator,
as Mark Crispin stopped some while ago, and no one has stepped forward
to take his place.  This means that there are no changes being made to
the file containing our names.  I have been trying to desubscribe for
a month now, without success.  It would be quite possible (and quite
nice) if someone wrote yet a third kind of mailing list program which
would do as you suggested -- automatically send out a message FROM
info-atari-request TO (everyone on the list) each time it received a
message TO info-atari; not unlike an automatic single-message digest.
That way, mailbounc would go to info-atari-request, and the moderator
could handle it there.  This is indeed a part of a program called pmd
(personal mail demon) which is written in C and runs on any system
which looks for a .forward file before putting mail in someone's
mailbox (e.g. any system running sendmail).  A friend of mine wrote
pmd, and I will send copies to anyone who wants one.
	Unfortunately, since we don't have a moderator, this
won't happen for a while, so all of us (including usenet recipients)
will have to suffer a while longer.  This brings up your last
suggestion.  DON'T START A MAILER WAR!  What will happen is this:  you
send a message into the net.  3 machines work for a few days trying to
send your message, and send back one refusal message per day (all the
others copies of your message get to us).  That's nine messages you
get back.  You resend those messages to the "offending machines"
(which are only doing their jobs).  Each machine works for three more
days trying to send the messages...you get 27 messages back.  Your
phone bill will blow up before the machines will.  If you send to
postmaster at those machines and complain, they will tell you not to
expect mail to be delivered to nonexistent addresses, and they HAVE to
return it (arpanet requires it).  The source of the problem is not at
what you call the "offending machines", but on our list and the way it
is (isn't) being run.  The only solution is to FIND A MODERATOR!  I
can't do it, because I don't have the kind of storage capacity
necessary to hold the archives, and frankly, I want to get off this
list. 
	An arpa-arpa mailer war can bring (and has brought) down the
Arpanet in a disastrous way.  An arpa-usenet mailer war will kill the
usenet recipient with phone bills long before the arpa machine dies
from overload.  If you bounce your bad mail to the machine the list is
on, most likely it will slow things down there (as well as load up the
list, which you receive) and the people using that machine will
complain and the list will go away.	
	There are two immediate solutions: 1. suffer (i.e. accept the
mailbounce as the cost of posting.) or 2. don't post.  I don't like it
(think: I get 8-15 messages a day that I don't want, whether or not I
post to the list!), but aside from writing to postmaster@score and
telling him to completely kill the list (BAD IDEA!), we're stuck.

--jh--

e-mail:                                  post:
jh@mit-athena.MIT.EDU (preferred)        Joe Harrington
jh%oz@mit-mc.MIT.EDU                     69 Chestnut Street
                                         Cambridge, Massachusetts 02139