I didn't really think you were, but I wanted to make sure _everyone_
realizes this.
We should really try to cut this general discussion off short, and
see if someone wants to run with some of the ideas offline and come up
with a proposal without cluttering the list unduely.
} >... If we exempt articles posted to
} >*.answers groups, they'll just start forging approvals. We have plenty
} >of experience with this. Dating all the way back to Canter and Siegel.
} Interesting. I'd point out that forged moderator approvals are
} even more of a fully-justified reason for cancels than spam (they can
} legitimately be done even without a cyberspam or other indicator). However,
} if moderators don't want to get into this problem, I can understand their
} reluctance.
It just adds an additional and substantial complication into the mess.
Worse, you'll see inexpert despammers start trying to cancel them too,
and they won't know about adding Approved: lines to their cancels - the
moderators will start seeing lots of cancels in their mailbox.
I just wanted to make the point that if you make FAQs immune to
despamming, then they'll start spamming "FAQ: FREE CHILD PORN!!!" -
spammers will have a substantial financial incentive to call
their spam "a FAQ". We might even start seeing some spammers trying
to register their adverts as "official" FAQs, and if the administrators
try to block this, they'll start getting lawsuit threats.
One of the most important reasons I "get away" with doing what I do
is because I can prove to anyone's satisfaction that what I do is
(a) perfectly independent of content, (b) plays no favourites and
(c) desired by the vast majority of sites and users.
Letting FAQs through would be playing favourites. And provide the
persistent spammer several new avenues of attack.
} >I'm currently chatting with the Worldnet Usenet developers and I've
} >asked whether they still forward >5 and simply block new postings >5.
} >If there are other questions you'd like me to ask, please let me know.
}
} Thanks. If it's the latter,
It's the latter. Worldnet's development team is interested in finding
a compromise. This sounds good, though I'll warn you that this is
only _one_ ISP - full cooperation from Worldnet doesn't help you
with IBM or Netcom or the hundreds of other sites already doing it.
Such as the one I learned about today which is one of the world's
most important and respected backbone sites - it has G7 now if I
remember my email correctly.
I have a possible solution, though, this will be a bit of an hassle for
many sites. Due to the nature of the way feeds are specified, you can
have multiple feeds of different sets of newsgroups to the same
destination. By specifying the main list of groups with G10, and
another feed description containing only news.answers without G10, you
have registered FAQs propagating without the cross-posting restriction.
But you've just doubled your newsfeeds file, and may encounter system
limits.
[Yes, this is subject to attack, but controllable.]
One possible assist would be to make certain sites "*.answers"-only
sites, and distribute them freely.
} > - Administrators are now starting to run "poison-pill" patches.
} > This allows them to dump any article on the floor that is
} > cross-posted to a "poison group". For example, if you
} > cross-post the alt.binaries.pictures.erotica.d faq to
} > alt.answers and elsewhere, we silently and quite happily dump
} > it on the floor. This is partially in response to the
} > massively-crossposting sex spammers, and partially in response
} > to the "hidden warez" group problem (which is scaring the
} > crap out of some sites).
} Sounds like this does need modification, as in the case you mention.
I think we need to be careful about saying things tantamount to "we know
better, the sites _need_ to run the way we think they should". Frankly,
if I poison-pill alt.binaries.pictures.erotica.children, I want _all_
articles cross-posted to it solidly dead, including _any_ FAQs posted
there. I have users who'll get up in arms simply because they see
a group we don't carry appears in a newsgroups line. And some of
these users are senior managers.
} I will be interested to see a court case analogous to the Prodigy one in which
} a copyright holder sues an ISP on the basis of not shutting off a warez
} group fast enough... when the ISP was doing so on a basis other than
} complaints.
I'll counter your Prodigy court case with a Compuserve court case ;-)
It always boils down to "reasonable efforts". The warez situation is
sufficiently well-known now that turning a blind eye to it or blandly
responding only after the fact has become risky, yet, they won't expect
you to be prescient. It's well-known because anyone running a
Usenet site _knows_ what warez are, and often have the existance of
a new hidden warez site when "alt.fan.lightbulbs" receives 500Mb
of postings in a day and blows your carefully designed disk layout.
} >In the end, system owners always get to decide what happens on their
} >systems. This is called "property rights". This is what most spammers
} >fail to appreciate, and what differentiates them from the rest of the
} >world. It's not really any of your business how long an article of
} >yours resides on my machine, whether it ever resides on my machine, or
} >whether I pass it on to anyone.
} As a libertarian, I am certainly aware of the concept of property
} rights. I'd say that "not really any of your business" basically means,
} at least to the degree that I agree with it, that I should not be able to
} _force_ you to change those characteristics. It doesn't mean that I can't
} be aware of them; it doesn't mean that I can't change my actions (which are
} _my_ business so long as I don't use force against you or others) in response.
Absolutely. I have nothing against about exploring the situation, and
coming up with workarounds. I'm just making damn sure that everyone realizes
that the "I won't give an inch", "I'm going to post every three days
if I have to", "you must run your system the way I think" is more than
ineffective - it's counter-productive. If you change your actions to
consume yet more resources, sites will respond simply by restricting
even more - you accelerate the decline.
} I'd also point out that for FAQs on certain topics (e.g., for gay adolescents
} on dealing with parents) the ultimate FAQ consumer may not have much choice
} on what ISP someone else (e.g., the kid's parents) chooses and its policies.
This of course is one of the downsides, especially when parents gravitate
to "safe" ISPs that are poison-pilling "naughty" groups as a way to
sell to the "family-oriented crowd". The only solution to that is
carefully designed up posted pointers to web sites. This happens
regardless of cross-posting.
} Ultimately, given that most ISPs (including those whose decisions I'd
} be responding to) are neither governmental entities nor significantly
} governmentally controlled in such matters (governmental restrictions won't be
} on expiration times; they're on carrying a group at all), and my FAQ is not
} of the variety I mentioned, I ultimately have to agree with you.
Oh gosh ;-)
} >So they just crosspost in groups of 5, where the fifth is always the same
} >moderated alt group.
} (Re: alteration of 5+ limits to not include moderatd groups)
} I can see the problem here, namely of getting automated recognition of
} multiposts (as opposed to crossposts). I've taken a look at text-based
} steaography and digital fingerprinting before, and I can see the difficulties.
No kidding. But, believe it or not, multiposts ("classic spam", aka EMP)
is rather easier to deal with than massive cross-posts ("spam" used
in a looser generic sense, aka ECP, aka "velveeta"). I've probably
cancelled over 500,000 spammed articles over the past two years.
And this is with trivial heuristics. I've been on the lookout for
good text matching, but none of the leads, none of the several people
with enough time on their hands to experiment, have demonstrated
anything useful.
Massive EMP is much easier to detect, yet hits fewer groups than massive
ECP.
} >} The latter idea would be a good one, yes; such messages aren't
} >}causing as much problems.
} >Only from a flame troll perspective. Not from a spamming perspective -
} >if EROSNET put a followup-to in each of their 40 group cross-posts, it
} >wouldn't do a darn thing to the spam level.
} Let's step back a second here. The essential purpose of limits on
} crossposting is to increase the signal relative to the noise. (Flame trolls
} consume resources just as much as spam does, and I see no reason to be more
} against the latter than the former especially looking at things in a
} content-independent (but receiver-_wishes_-_dependent_) manner.) In this
} context, such a modification - to, say, 9 groups max for 1-group or poster
} followups vs 5 groups max for multi-group followups - could make sense. I can
} see the argument that spammers will just use the 9 groups... but that's a lot
} less than the 40 you mention, and they could just as well adapt to a rigid
} 5-group max. Of course, you've got (lots) more experience with this than I
} have; as with all of this, I'm perfectly willing to admit to problems with
} my ideas.
You're missing the primary rationale here. Flame trolls are not why
they're implementing cross-posting restrictions. EROSNET-style
spamming, which is parhaps the worst thing we're seeing now, is tens of
thousands of articles posted to 60 groups each. _That's_ why Worldnet
and IBM are implementing cross-posting controls. Consider flame troll
control simply as an accidental side benefit.
You have to realize that the EROSNET style spammers are equivalent to
weeks or months worth of news.answers cross-postings _every_ day, and
are completely obliterating the usefulness of hundreds (if not thousands)
of groups at any given moment. Just wander into a alt.binaries.pictures
.erotica group some day, with a graphical browser like netscape. You'll
be appalled at the quantity (and quality ;-) of junk - and that's probably
after taking several thousand cancels per day of this crap into account.
One thing that would be useful is to see a consensus on whether we're
trying to preserve Expires: control in all groups, or just in the
*.answers groups, or just news.answers.
-- For more information on spam, including countermeasures and resources, see the Internet Spam Boycott, at <URL:http://www.vix.com/spam/>.Chris Lewis: _Una confibula non sat est_
[
Usenet Hypertext FAQ Archive |
Search Mail Archive |
Authors |
Usenet
]
[
1993 |
1994 |
1995 |
1996 |
1997
]
© Copyright The Landfield Group, 1997
All rights reserved