Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site aoa.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!bbnccv!bbncca!aoa!mbr
From: mbr@aoa.UUCP (Mark Rosenthal)
Newsgroups: net.news,net.news.notes
Subject: Prohibition of postings to multiple groups - a bad idea
Message-ID: <312@aoa.UUCP>
Date: Thu, 3-Oct-85 11:08:25 EDT
Article-I.D.: aoa.312
Posted: Thu Oct  3 11:08:25 1985
Date-Received: Sat, 5-Oct-85 06:34:06 EDT
References: <10381@ucbvax.ARPA> <3274@nsc.UUCP> <698@tpvax.fluke.UUCP> <3166@nsc.UUCP> <126@sdencore.UUCP>
Reply-To: mbr@aoa.UUCP (Mark Rosenthal)
Organization: Adaptive Optics Assoc., Cambridge, Mass. USA
Lines: 54
Xref: watmath net.news:4014 net.news.notes:29
Summary: 

In article <126@sdencore.UUCP>, which was posted to both net.news and
net.news.notes mark@sdencore.UUCP (Mark DiVecchio) writes:
>
> One simple step, which has to have been suggested before, is prohibit
> posting the same message to multiple newsgroups.

I've seen this suggested before, but I've never understood how people think
it is going to help the situation.  It would simply encourage posters who
want to get around the restriction to post via multiple invocations of
postnews/Pnews, rather than a single invocation.  Then we all get to read the
garbage twice!

There are already enough duplicate articles that readnews/vnews/rn can't
screen out because the sender posted them with multiple invocations of
postnews/Pnews.  I usually attribute this to the poster's inexperience, and
send them mail explaining why it is preferable to post to multiple groups in
a single invocation.  However the malicious poster will always be able to do
this in order to force an article to appear in multiple newsgroups.  Software
to screen out such malicious postings would be virtually impossible to write.

The simplest software solution to screen out multiple-invocation postings
would be to compare every article to every other article.  Doing this at the
sending site requires all other sites to upgrade their software in order for
the scheme to work.  Experience clearly demonstrates that this is not going to
happen.  If you do the compare at the receiving site, you have a task which
will chew up an enormous number of CPU cycles.  You might reduce this by
comparing only articles from the same sender.  You do store articles in a
data-base indexed by sender, don't you? :-)  This works assuming the poster
has only one login id.

But the real problem is that of writing an article comparison routine which 
could filter out trivially changed articles.

    A slight change in wording in a single sentence would cause the comparison
	to fail.
    Even slight changes in wording in a single sentence might cause the
	comparison to fail.
    The slightest change in the wording of even a single sentence could cause
	the comparison to fail.

I don't think we have AI techniques advanced enough to be able to handle the
problem.  Certainly not in a form that could be installed on most machines on
the net!

Since it is unfeasable to produce a technical solution to the problem of
multiple-invocation postings, we must depend on the courtesy of the poster.
Anything which presents such an obvious barrier to the poster's wishes is
likely to lower the poster's courtesy level.  Like, from -5 to -10 on a scale
from +1 to +10.  :-) / 2

-- 

	Mark of the Valley of Roses
	...!{decvax,linus,ima,ihnp4}!bbncca!aoa!mbr


Brought to you by Super Global Mega Corp .com