On Mon, 5 Feb 1996 14:32:29 -0800 (PST) Diane Barlow Close
<close@lunch.engr.sgi.com> said:
>> LISTSERV's spam detector catches spams even when the list is totally
>> open and anyone is allowed to post (and of [snip]
>
>Sounds exactly like Majordomo's new "taboo-headers" and "taboo-body".
Sorry, but it's completely different. Keyword based filters may be very
effective at catching recurring spams, like the magazine subscription
club, but they are of limited usefulness with brand new spams, like the
nude video conferencing one we got recently. Plus, if you filter too many
keywords, you end up blocking perfectly legitimate postings. And
filtering out entire domains just because one of the accounts was used
for spamming, it's outright unfair to the other innocent users.
LISTSERV's spam detector, which has been around for nearly a year, is not
keyword based. It just detects messages that are posted to large number
of lists in a short period of time, and blocks them. It doesn't matter
whether the message contains the usual catch phrases, which I'm afraid to
type here since my message might get filtered, even though it isn't a
spam. In fact, the software isn't passing judgment on the *contents* of
your message (and exposing you to lawsuits), it's just decreeing that any
message posted to thousands of mailing lists is bound to be out of topic
for a large fraction of these lists. The definition of "spam" is *not*
"advertisement", but "message posted to a bunch of lists that have
nothing to do with the subject matter". You can't build a spam filter
with a pattern matching engine - all you can create is an advertisement
filter, and not all advertisements are inappropriate. Conversely, not all
spams are ads (aren't you tired of being invited to conferences in exotic
resorts all over the world?), and it's easy to reword an ad so it doesn't
sound like an ad any longer.
Eric
Follow-Ups:
|
|
