Hi Dave,

Yep... we use SURBL tests...  We also have several types of in-house tests
to quickly adjust to things like the German wave.  Very little slips through
(less than 0.1%), just curious if others have noticed that DNSBLs and RHSBLs
have become next to useless...

Darin.


----- Original Message ----- 
From: "Dave Marchette" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Friday, June 03, 2005 11:48 PM
Subject: RE: [Declude.JunkMail] Blacklist effectiveness


Darin,

If you have not yet, you might consider adding SURBL testing as well.
Darrell (http://www.invariantsystems.com)  has a product, invURIBL, that is
competent at interfacing SURBL to Declude(which in reality should and may at
some point in time do this natively) as an ext. test.  SURBL looks at the
target link of the spam, and compares it to numerous blacklists(including
name server bl).

Drawbacks:
1  Processor intensive(testing showed a 15% increase in proc usage)
2  Difficult to fine tune.  'Out of the box' this product returns a weight
that is a factor of several configurable tests that run inside INV.  You
have to fine tune each, then observe the end result.  There is likely an
easier way to tune this but I have not yet delved too far in.

Upside:
1  As effective as Sniffer, and utilizes a different mechanism for
identification.  Low false positives.
2  Cheap


Sniffer is _amazing_.  However, we were discouraged after it took 8 hours to
get a Sniffer rulebase for the last wave of German spam.  So, we started
testing SURBL to give Sniffer some help.

Side note:  The very instant we initialized testing, we started seeing a
significant increase in picture spam (just a gif file, nothing else, not
even a link - therefore undetectable to SURBL)   We attribute this to the
fact that we did not sufficiently cloak the test name in the headers and
body, and the mass mailers determined by way of 'mailbox full' bounces from
the test domain, that we were utilizing SURBL.

Dave


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of Scott Fisher
Sent: Friday, June 03, 2005 7:11 PM
To: [email protected]
Subject: Re: [Declude.JunkMail] Blacklist effectiveness


I've posted my spamtest effectiveness from Feb 2004 forward at
http://it.farmprogress.com/declude/declude.htm

----- Original Message ----- 
From: Darin Cox
To: [email protected]
Sent: Friday, June 03, 2005 8:33 AM
Subject: [Declude.JunkMail] Blacklist effectiveness


Anyone else noticing over the past few months that DNSBLs and RHSBLs have
almost completely lost their effectiveness?

We're seeing only a few (e.g. SBL, MXGATE, MAILPOLICE) that catch more than
5% of incoming spam, and they top out at less than 6%.

If it weren't for Sniffer and the specialized tests in Declude we'd be
buried.

Just curious as to what others are seeing...

Darin.
---
This E-mail came from the Declude.JunkMail mailing list.  To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail".  The archives can be found
at http://www.mail-archive.com.

---
This E-mail came from the Declude.JunkMail mailing list.  To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail".  The archives can be found
at http://www.mail-archive.com.

Reply via email to