John Stanley stanley at skyking.OCE.ORST.EDU
Mon Feb 27 13:54:49 EST 1995

In article <D4MDKF.L16 at>,
Richard E. Depew <red at> wrote:
>>Moderating the newsgroups would mean changing them so that only
>>preapproved messages would be distributed.
>     Not if you permit the use of retroactive moderation as an
>alternative to preapproval moderation.
>     Retroactive moderation is moderation through the removal of
>inappropriate articles *after* they have been posted.  

Since Richard hasn't bothered to explain fully what "retroactive
moderation" is or how it works, and certainly hasn't told you why it
would work to prevent spamming (there is a good reason he hasn't -- it
won't), allow me.

"Retroactive Moderation" is a normal, unmoderated newsgroup in which a
few people have decided to pick someone who will remove any posting
which is unwanted. The unwanted posting is first distributed throughout
the world just like any other posting. Then, after the "moderator"
decides that he doesn't want this posting to appear, a removal message
is issued to remove it. This "unwanted" is defined by the one who is
chosen, and can mean anything from grossly off-charter to poorly
formatted (yes, "lines longer than 72 characters" can be cause for
removing articles under RM).

What Richard calls "retroactive moderation" is really retroactive
censorship (RC for short). There is no moderation involved -- the
newsgroup is not moderated. Unless Richard has cobbled together a
personal newsreader that understands a non-standard coding for
newsgroups, there are no news systems in the world that know the
difference between unmoderated and RC, which makes it hard to talk
about RC as an existing system (as Richard always does).

(Whether RC is noted by news systems is moot -- there are currently no
RC newsgroups. It isn't being used anywhere. What Richard is desperately
seeking is a test-platform. He failed to tell you that.)

>The mechanism by which retroactive
>moderation achieves an improved signal/noise ration is by discouraging
>authors from posting off-charter articles in the first place.

This is the sole basis for how RC will protect a newsgroup. Once you
understand how RC works, you can see through the facade.

The newsgroup is not protected "in the first place". Nobody is
discouraged from posting "in the first place". Anyone who posts
any article will have his article distributed throughout the world. It
will appear in news spools worldwide, and will be shown to users as
would any other article. People who come to the newsgroup looking for
help will not know that the group has a censor. They will post what
they will. And what they will will be transported for them.

Only AFTER the article arrives at the censor's site, and after he has
seen it, will a removal be issued. This removal will take time to
propogate, which means the "unwanted" article will stay in place even
longer (this time delay has been presented as a "feature" by Richard).

Because the article is distributed just like any other, and because of
the delay between its appearance and its removal, people will be able to
reply to this article. Some may choose to flame the poster, some may
think the article is on-charter and reply normally. What is important to
note is that the noise will not be lessened.

Of course, if the offense is the first for the poster, then the article
will NOT be removed. It will remain in the news spools of systems
worldwide. You get zero (0) protection from this article.

Now, I have been careful to use the terms "remove" and "removal
message" instead of the more usual news vernacular "cancel". That's
because RC is not technically a "cancel", although early versions of it
were. RC now uses a header called "supersedes". This performs the same
removal function as does a cancel, but it has the bad side-effect of
leaving another article in the newsgroup itself. (Cancel messages are
not presented to readers of a newsgroup, in most cases. Superseding
articles are. The act of admonishing or rejecting an article creates
more noise in the group itself -- as much noise as the article being
rejected. This side-effect is not a bug, it's a feature, according to

You should also be aware that the header that RC uses to control what
appears is not defined anywhere in the standards. It is an extension
which has not yet been accepted as a standard. Systems are free to
ignore this header, which leaves them with no protection. (Moderation,
however, is in the standards, and will protect systems.)

Now, this should be enough to prove that you will receive no protection
from spammers by using RC. Their postings will appear, and they will
have achieved their goal -- people will see their article. But, if you
need more proof, it should be sufficient to point out that there are
ways to prevent the censor from seeing the article, and if he doesn't
see it, he can't remove it. There are also ways of foiling the removal.

>     Numbers 2, 3, and 4 are addressed somewhat differently by
>retroactive moderation.  The moderators duties' are little more
>difficult than reading the group and admonishing or removing the
>occasional off-charter posting (functions which can be done from
>within trn with the help of the proof-of-concept scripts.)  

If you note carefully, the censor's duties, in the case of an
admonishment, are no different than what any common reader of the
newsgroup has available to himself. Any reader can admonish what he
considers an off-charter article, and there has never been a lack of
readers who will do this. Electing one person to do it will not stop
anyone else from admonishing, and there is nothing to stop them from
doing it in the newsgroup. It is foolish to believe that USENET readers
will happily admonish in every group except your censored ones.
Remember, there is nothing to mark the group as censored.

In other words, if you think that most of your problem is ignorant
people posting things they wouldn't once they know the charter, then you
do not need a censor. If you want to stop spamming, then electing a
censor won't help.

>Thus it is volunteers
>from among the readers and posters in the newsgroup community who set
>the standards for how that newsgroup will interpret its charter.

This is not how the RC proposal works, and Richard is disengenuous to
make this statement. It is not volunteers; there is an election based on
the pretense of "newsgroup democracy". Someone wins an election.
Presumably, if they don't interpret the charter the way the "newsgroup"
wants, they can be replaced, but Richard has not, in the year or so he
has been stumping for this plan, provided a single method of removing a
censor who doesn't want to stop.

>     Moderation is the "wave of the future", but so is

This is why Richard wants you to centralize your decision making in one
place -- the censor. The censor will be choosing for you what you should
read, even after your system has spent the resources transporting it.
Even after you have decided that the posting is on-charter and have
spent your time responding to it. For the audacity of deciding for
yourself, you may very well wind up being chastised, or worse, having 
your article removed.

>     Yes, the "central control" registration and moderation procedure
>would require considerable administrative effort... 

No, not really. There is already software that manages mailing lists.
There are already auto-moderation systems in place. Even if there
weren't, it would not be difficult to write an auto-moderation program
that uses the names in a dummy mailing list as the list of authorized
posters. Those who want to register pretend they are subscribing to a
mailing list. Not very hard.

>It will be as robust as, but less noisy than, an unmoderated group.

Since RC does not prevent anyone from posting, this claim is patently false.

>Retroactive moderation is also suitable for groups that are gated to
>mailing lists.  

Non-sequitor. Moderated newsgroups are more suitable for gating.

>The admonishment and rejection messages should be
>filtered out of the mailing list stream.  

And so RC is not suitable for gating to mailing lists. You have to
worry about which articles pass through the gate. For moderated groups,
you pass every article that appears in the newsgroup. Very simple.

>Those who get the mailing list
>will derive their benefits from the decreased rate of off-charter
>posting that retroactive moderation will gradually achieve.

This claim is also patently false. Unless you program your gateway to
delay every article for several days (just to see if the censor removes
it), every article that is posted -- even the unwanted ones -- will
appear in mail. And you cannot cancel mail like you do news -- once it
is mailed, it is mailed, and it will show up in everyone's mailbox.
There is even less benefit to mailing list users.

I sympathize with your desire to reduce the effects of spammers and
off-charter postings. Unfortunately, RC will not achieve anything but
making Richard feel good that someone is finally using his plan. It will
not prevent off-charter or spam postings, it will not improve your
signal/noise ratio. In fact, it will decrease your S/N, since each
off-charter posting will have a superseding article showing up a day or
two later. If you have a problem with 50 off-charter articles showing up
per day, imagine what it will be like with those 50, plus 50 more
"REJECT:" messages showing up each day. "Double your pleasure, double
your fun..."

Your best solution is auto-moderation. Use a mailing list management
tool to generate the list of approved posters. It won't be that hard.
Don't let Richard scare you into being his guinea pig.

More information about the Bioforum mailing list