From: Gregory Seidman <gsslist+ruby@...>
Date: 2008-04-07T22:49:51+09:00
Subject: Re: Is it possible to stop SPAM?

On Mon, Apr 07, 2008 at 10:15:09PM +0900, Karl von Laudermann wrote:
> On Apr 7, 8:37 am, Gregory Seidman <gsslist+r...@anthropohedron.net>
> wrote:
> > Spam is a global problem. If it were easy to stop, it would already be
> > gone. You are using gmail, however, which has pretty decent spam
> > filters. Use them.
> 
> I find the attitude of "filter it yourself" to be dismissive and
> unhelpful. No, you can't stop spam from being sent, but there do
> exist  tools to detect and filter spam, such as the one you pointed
> out that gmail has. If a spam filter could be applied to the mailing
> list so as to prevent the spam messages from being sent out to
> everyone in the first place, then that would be preferable, as it
> would remove the burden from each individual participant. IOW, having
> everyone filter it themselves seems to violate the DRY principle. ;-)

The problem with a global filter that is forced on everyone is false
positives. Either a human being (e.g. a moderator) needs to check for false
positives from time to time, or the filter has to be sufficiently
permissive to make sure all errors are one-sided (i.e. only false
negatives). The moderators of this list are largely hands off, and have
better things to do with their time. That leaves a permissive filter...
which may already be in place! How would you know, since you don't see what
is filtered out?

Even if a global filter is in place and a moderator checks for false
positives, consider the situation where someone starts a really relevant
thread... in Korean. Looks like spam, so it gets filtered out. But wait,
it's a false positive, it's relevant... but the moderator can't read Korean
so it's out. Of course, it really is worthless to everyone who doesn't read
Korean, but the poor Korean user doesn't even see his/her posts to the list
at all, and doesn't know what's going on. He/she doesn't even get a helpful
response from one of the few other people on the list who can read and
write Korean explaining that this is an English-only list and that there is
probably a Korean one out there. Substitute for Korean any other language
the moderator doesn't read.

Another example: are job postings spam? To some people, yes. To others, no.
The global filter better not filter them out. But if I don't want to see
them, it's up to me to do the filtering on my end.

> Also, not everyone is using the mailing list. Some of us read and post
> to the comp.lang.ruby newsgroup. Google Groups (which I use) doesn't
> have a spam filter AFAIK, and neither do most news readers with which
> I am familiar. (Killfiles aren't quite the same thing; you have to
> know in advance what you're filtering.)

Just as your mail client should be capable of spam filtering, so should
your news reader. Thunderbird may not be an ideal news reader, but I
believe it supports spam filtering. Google Groups allows you to report
messages as spam and applies a (fairly permissive, for reasons mentioned
above) filter.

The point is that there is a small subset of spam that can be 100%
identified as spam without resulting in false positives that cause problems
for some segment of the readers, and for all you know that level of
filtering is already in place. Everyone has to filter at the edges because
each person has a different idea of spam.

--Greg