RE: Egonomics

[Vernon Schryver <vjs at calcite.rhyolite.com>]

> From: "Andrew M. Benhase" <abenhase at gte.net>

> I think that that simplest answer is to have this
> list moderated by an IETF official. What could be
> an easier answer than that?
> ...

At the cost of violating my rule of trying to not send more than 1 msg/day
nor more than 4/week to any single list, and I'll offer some obvious
answers.  Please pick one you dislike least:

#1. Moderated mailing lists have the same fatal problem as moderated
   newsgroups.  Moderating involves lots of labor for no reward except
   grief from Special People who are above the rules.  Moderators quickly
   burn out.  When the moderators quit, the forum dies completely.
   Counterexamples such as the "Risks Digests" prove the rule.

#2. There are too many special individuals who would feel personally
   censored by any human moderation (in all senses of 'censor' and
   'moderation'), who would raise a big stink, and possible cause legal
   hassles for a standards organization.

#3. When you start paying the salaries of IETF officials, you can start
   directing their work.

#4. Little moderation is needed.  As I recall, there have been only the
   following types of inappropriate traffic in the last month or so:

      A. never ending TLD/where-is-Jon/DNS/IPv8 messages from a very few.
      B. advertisements, such as for the commercial instructor's test.
      C. the content-free Juniper press release and other non-technical
	(eg. mine) or bogus statements about ATM in that thread (e.g. that
	ATM should replace IP), which should have been sent to something
	like the c.d.cell-relay newsgroup, and the milding interesting
	EPD vs. PPD words that should have been in the ION list or c.d.c-r.
      D. messages about inappropriate messages (including this of mine).

    I think only type A has been proposed for filtering or moderation,
    and such moderation could be implemented with familiar anti-spam
    mechanisms applied to a dozen addresses, far easier than having an
    IETF official receive, delay, read, delay, and forward every message.

 ....


A human moderator for the list is almost certainly too expensive, no matter
how you compute expenses.  Here are some cheap, computerized alternatives
to deal with type A (first 2 repeated):

  - automatically rate limit each author.

  - automatic RED.

  - make the ietf at ietf.org reflector automatically add to each message text
    something like:

    Previous msgs since Feb 1997 by this author have been rated Wrong 72%
    vs. Right 5% by responding subscribers.  See http://xxx/www.ietf.org
    to rate this msg or check ratings of other authors.  Unsubscribe by
    sending to majordomo at ietf.org.

     (Key rating database by Message-ID: and From: lines, and only ~3 bits
     of data/key, besides pointer to original msg.  Only one rating per
     subscriber per msg.  Only subscribers and those able read headers to
     get the keys allowed to play.  Don't merely count ratings, but record
     them all, allow people to change or delete them, and tally them only
     when generating the text for a new msg.  Permanently sanction and
     expel from the IETF anyone who tries to cheat, and report them to
     the FBI for "cracking."  About 300 lines of code would do it.)

  - NoCEM's

  - tell people to use procmail, self-discipline, or other mechanisms on
     their own mailboxes if they don't like hearing from kooks.  Tell them
     that dealing with kooks is one of the costs of doing and even merely
     watching politics.


Vernon Schryver    vjs at rhyolite.com