Re: anti-spam?

Mark Grant (
Thu, 26 Jun 1997 22:34:37 +0000

On Tue, 24 Jun 1997, Alexander 'Sasha' Chislenko wrote:

> Could such an "anti-spam spam" help generate enough "negative feedback"
> for the spammers to stop asking for snail-mail replies? That may make
> things much more difficult for them.

I don't see that it would have much effect, since the spammers don't pay
for it... sure, they have to sort out the junk, but they can hire a few
minimum wage workers for that. You'll pay far more to send them the
snail-mail than they'll pay to sort it.

Personally, as much as I dislike spam I'm no longer too concerned about
it. I now have some pretty sophisticated mail-filtering software on my
account which has at least a 90% success rate, and the few messages which
get through *might* not be 'real' spam (i.e. they're sent directly to me
and have at least *some* relation to the things I have on my Web site).

So from my point of view it's merely forced me to develop some software
which I really should have done ages ago, and which in future I can
probably extend to delete mailing list messages which I'm probably not
going to be interested in. As more and more people start to use this kind
of software spam will slowly disappear, or they'll have to give us an
incentive to read it, such as including an ecash payment with the message.
That's the only way to really stop the spammers; develop technology so
that spamming simply doesn't work.

In case anyone's interested, I'm basically using the following procedure:

Always pass PGP-encrypted or PGP-signed mail (since I can
eventually update that to pass or block based on PGP
Always pass mail from my friends and mailing lists (of course
the header could be faked, but few spammers would go
that far; and I haven't seen enough spam posted to the
lists to worry about scanning for it)
Autodelete anything sent from known spam sites
Autodelete anything with some magic words in the header
(e.g. known spammers IP addresses)
Count the number of magic words (e.g. 'MLM', 'Make Money') in the
message subject and body, and increase the count if the
message isn't sent directly to me or in various other
circumstances. If the count is over one limit then it's
saved to a spam folder which I clear out every few weeks,
and if it's over a higher limit the message is just deleted.

So far I've had no false positives for deletion and <1% false positives
for the spam folder, and the extra overhead is only a fraction of a second
per message; since the processing is all done on the ISP I also don't have
to worry about the cost of downloading the extra mail to process it
locally (e.g. last year most of my Net access was through Internet Cafes).
I'd really like to make the content analysis more sophisticated, but I'm
not sure where to look for information on text analysis. Any suggestions?


|Mark Grant M.A., U.L.C. EMAIL: |