On Nov 3, 9:16pm, era eriksson wrote:
Subject: Re: Best anti-spam-filter tests?
...
As an aside, it would be interesting to put up a "test bench" for
various spam filters where the offerings that are out there could be
objectively compared with real-world spam and the results published.
Ah, that would be very nice. I'd be glad to contribute the reams of
spam I have already caught with my existing filters.
I don't have the expertise to write the "test bench" though. I'd be
glad to submit my filters (composed of many of this list's recommended
filters) for public evaluation.
And the key point in the answers is that if you're not testing it with
real-world messages, is there even a point in trying it? Why restrict
your imagination to only the perversions you yourself can think up?
Thanks era, I value your input.
I guess you're right -- I only intended on catching the spam that
comes to me -- and so I wasn't thinking that I was coming up with
various perversions; but, in catching the spam that does come thru,
I do have to think up various "perversions" in order to catch more
than just a repeat of that one message.
In that respect, my methods (of generating permuations of spam that
arrived in my mailbox) are only testing my ingenuity, or lack thereof.
Still, any test method that actually causes me to receive _more_ spam
than I already do, is unacceptable -- (unless there is a perfect spam
filter, in which case it wouldn't matter how much spam I received.)
And why not feed the message to procmail on standard input, if indeed
it is Procmail you want to test?
Actually, I do -- I was just trying to learn "better" ways. My current
methods work fine. But, there is _always_ a better way; I was looking
for that better method. (Nobody is volunteering it, apparently.)
Thanks, I value your input tremendously,
John Gianni
Cadence Design Systems, Inc.
Makers of software to design chips and boards