Hi Dave,
Oh, I think it's not just an "opinion". Specially the SNFIPREP sample seems
to reduce the weight for 45% of all emails - when we all know what 98% are
spam. But, let's look at the facts, and then you can correct me where I got
them wrong so that either one of us can learn from their mistake.
>> I don't believe it triples the score I think the max would be 2 tests
based on the same information <<
1. A black or truncate will trigger
a) The SNF test (either Truncate or "IP Rules")
b) The SNFIP test (either Truncate of Black)
c) The SNFIPreputation test (because Truncate or Black will NOT have a
Good or Neutral reputation)
Is it therefore my believe that the score is tripled - or where am I
thinking wrong?
Now - this may actually a somewhat desirable outcome (because the mail will
be blocked) - I do agree with you on that. But, it can result in undesirable
outcome once customers attempt to slightly adjust the default settings
without realizing that some tests are triplicated.
>> What would be helpful is making a suggestion to what settings you use
based on your results <<
2. Fair enough, my suggestion would obviously have to be to comment
out the two redundant SNF rules and let SNFIP handle the IP scoring part -
and increase those weights if you don't like to be "just" 10 (remember, if
the "content" SNF tests find a match, THAT score will also be added!)
SNFIPBLACK SNFIP x 5 20 0
SNFIPTRUNCATE SNFIP x 6 30 0
# SNFTRUNCATE SNF x 20 10 0
# SNIFFER-IP-RULES SNF x 63 10 0
This way, your users can SEE that the SNF options exist and how they would
be coded, but would be realize that they have to research the implications
first before removing the comment.
3. I suspect that the biggest problem is the SNFIPREP test - but I'm
waiting for Pete to give some input. The way I understand your email from
Thursday about your algorithm, it potentially assigns 10% of all emails a
score of 0, it potentially assigns 45% of ALL emails a score of -5. And it
adds a weight of 10 to the remaining 45% of all emails - which also seems
rather arbitrary. (Disclaimer: What we don't know is the "distribution"
curve, is it a "bell" curve, where the majority fall into the range of -0.05
to +0.05 and very few fall in the + or - side. Or is the distribution some
logarithmic curve, that has very few on the "good" side, a moderate
frequency in the middle and the increases sharply the further it gets on the
"bad" side.
Now, maybe you analyzed the distribution curve before you developed this
sample?
Until all these crucial questions are resolved (I wouldn't want to reduce
the weight for a totally unknown percentage of all the spam!) I would
comment it out for sure:
# IPREPUTATION SNFIPREP x 0 10 -5
But, I have a really good suggestion on how to make this entire test more
usable:
The whole point of the reputation "scale" (between -1 and +1) is to allow
Sniffer customers a "graduated" response - not just 2 values for 90% of the
number scale. My suggestion would be to slightly enhance your formula by
multiplying the reputation value with the assigned weight (after shifting
it for the base score). THEN I think this test would be useful, because it
would actually produce a sliding scale of weights based on the reputation
scale. In other words:
(( Abs(Reputation Value) * 10 ) - Base Value) * [Pos or
Neg]WeightFactor = Final Weight
For this line:
# IPREPUTATION SNFIPREP x 0 2 -1
it would results in weights between +20 and -10 - which is in line what the
reputation scale was intended to provide:
Reputation 0.0: ( ( 0.0 * 10 ) - 0 ) * 2 = 0
Reputation 0.3: ( ( 0.3 * 10 ) - 0 ) * 2 = 6
Reputation 1.0: ( ( 1.0 * 10 ) - 0 ) * 2 = 20
Reputation -0.3: ( ( 0.1 * 10 ) - 0 ) * -1 = -3
Reputation -1.0: ( ( 1.0 * 10 ) - 0 ) * -1 = -10
Best Regards,
Andy
From: supp...@declude.com [mailto:supp...@declude.com] On Behalf Of David
Barker
Sent: Saturday, May 01, 2010 10:11 AM
To: declude.junkmail@declude.com
Subject: RE: [Declude.JunkMail] Sniffer IP vs. Sniffer IP Reputation vs.
Sniffer Truncate
My quick response.
The "out of the box Declude" Customer CAN use the samples given. The extra
scoring ensures that bad IP's are eliminated as spam. It would be the same
as placing an extra high score on a specific test. Pete's notes suggest:
63 - Black
Systems should usually quarantine or reject messages produced by this IP.
20 - Truncate
Systems should usually refuse connections from this IP.
Which means for the majority of our customers an exaggerated score on these
message is fine (I will have to check on Monday but I don't believe it
triples the score I think the max would be 2 tests based on the same
information) Unfortunately a large portion of our customers today do not
understand or even care about the details. The beauty of Declude is that
you are welcome to score tests however you feel appropriate for your email
server.
I do agree with you that it could be made more clear, but to advise the list
NOT to use the current declude settings is your opinion. What would be
helpful is making a suggestion to what settings you use based on your
results.
David
_____
From: "Andy Schmidt" <andy_schm...@hm-software.com>
Sent: Friday, April 30, 2010 9:26 PM
To: declude.junkmail@declude.com
Subject: RE: [Declude.JunkMail] Sniffer IP vs. Sniffer IP Reputation vs.
Sniffer Truncate
Thanks Pete - that confirms what I feared.
Declude's own sample should NOT be used "as is" because it duplicates the IP
results (at minimum)
>> The SNFIPREP test gives you a variable weight based on the IP reputation
in GBUdb. This allows you to get some weighting positively or negatively
based on the reputation even when that reputation is not in one of the
defined GBUdb envelopes. <<
Yes - according to Dave's explanation earlier today, Declude will get a
decimal number between -1 and +1. Their Sample/Default configuration treats
"0" as normal, treats anything negative as "GOOD" (and subtracts 5 points)
and anything positive as "BAD" (and adds 10 points).
So - even though Sniffer returns information on a vary graduated scale,
Declude then returns 3 discrete numbers. In fact, 0 is only returned for 10%
of the range - 90% of the range returns either "-5" or "10".
>> I presume that even when SNFIP does return Caution, Black, or Truncate
that SNFIPREP continues to work and in that case will provide some shading
to those values... so, if you will, more or less Black, etc.<<
Based on Dave's explanation, "Caution", "Black" and "Truncate" would
certainly always return a value > 0. Consequently, "10" would ALWAYS be
added to the weight for those 3 reputations.
Their default example basically TRIPLES the "10" weight that is assigned in
many cases (once for SNFIP, once for SNFIPREP, and once for SNF).
Let's see if Dave's chips in - but it certainly seems to me that Declude's
Sniffer sample/default config should NOT be used (because it doesn't do what
an "innocent" user might expect). It's not at all clear that after all
their Sniffer rules, 30 would be added to the weight in several cases.
-----Original Message-----
From: supp...@declude.com [mailto:supp...@declude.com] On Behalf Of Pete
McNeil
Sent: Friday, April 30, 2010 7:07 PM
To: declude.junkmail@declude.com
Subject: Re: [Declude.JunkMail] Sniffer IP vs. Sniffer IP Reputation vs.
Sniffer Truncate
On 4/30/2010 5:16 PM, Andy Schmidt wrote:
> Hi Pete,
>
> I'm look over Decludes recommended Sniffer configuration and trying to
> understand how much overlap there is between these options:
>
> IPREPUTATION SNFIPREP x 0 10 -5
>
> SNFIPCAUTION SNFIP x 4 5 0
> SNFIPBLACK SNFIP x 5 10
> 0
> SNFIPTRUNCATE SNFIP x 6 10 0
>
> SNFTRUNCATE SNF x 20 10
> 0
> SNIFFER-IP-RULES SNF x 63 10
> 0
>
> Looking at the Sniffer documentation IP test result codes
>
http://www.armresearch.com/support/articles/software/snfClient/resultCodes.j
> sp
> it seems that the SNFIP tests for "4", "5" and "6" (SNFIPCAUTION,
> SNFIPBLACK, SNFIPTRUNCATE) might coincide with 40, 63 and 20.
>
I am not intimately familiar with Declude's configuration and SNF
integration --- not like I used to be anyway (soooo many platforms now).
I _think_ these tests work like this:
The SNFIPREP test gives you a variable weight based on the IP reputation
in GBUdb. This allows you to get some weighting positively or negatively
based on the reputation even when that reputation is not in one of the
defined GBUdb envelopes. It's a subtle nudge in the right direction.
The SNFIP test gives you a hard result code based only on the IP
reputation when that reputation is within one of the envelopes defined
for GBUdb. So if the IP reputation is in the Caution, Black, or Truncate
range then that test will fire.
Presumably all of the IP tests happen before SNF scans the message --
because they can -- I don't know that they do, but I know that IP
reputations can be queried before and separately from a scan. (Scans
MUST happen in order for GBUdb to build up reputation data however).
Finally the SNF test responds to the normal blended result codes that
SNFClient would return.
So result code 20 is Truncate- meaning that the IP reputation was so bad
that SNF stopped the scan and returned the result code.
Result code 63 is Black which could mean that an SNF IP rule fired (rare
these days) or that no pattern matched but the IP was in the Black range
in GBUdb so GBUdb took over and forced the result code from 0 (no
pattern found) to 63 (Black).
Other result codes are also possible:
http://www.armresearch.com/support/articles/software/snfClient/resultCodes.j
sp#msgScan
David -- if I got any of this wrong please correct me.
> However, Declude ALSO tests for your Rule Group Result Codes "20" and "63"
> which are documented here:
> http://www.armresearch.com/support/articles/software/snfServer/core.jsp
>
> 1. It seems to me, as if their SNFTRUNCATE is the same as their
> SNFIPTRUNCATE, and their SNIFFER-IP-RULES is the same as their SNFIPBLACK
--
> effectively artificially inflating (doubling) the weights for these tests?
>
Yes -- if you have them configured that way. Some of the results are
predictable.
If SNFIP is Black or Caution then you are virutally guaranteed to get a
Black or Caution result from SNF -- Unless SNF matches a pattern in
which case you will get a pattern result code from the SNF test.
If SNFIP is Truncate then SNF should also return Truncate.
The weights you assign to these should be set accordingly.
> 2. How do those Caution/Black/Truncate exit codes relate to SNFIPREP.
> There, any reputation> 0 (up to 1) is given an extra weight of 10. But
> doesn't SNFIPREP report from the same reputation data as the SNFIP (and
> possibly even group result codes 20 and 63)? In other words, are those IP
> addresses that generate a reputation factor of> 0 ALSO reported as
> Caution/Black or Truncate - if so, we'd now TRIPLE count that score.
>
That's not quite true...
I presume the SNFIPREP test uses a sliding numeric value that combines
the probability factor and the confidence factor for the IP. This is not
the same thing as Caution, Black, and Truncate.
The SNFIPREP result is a sliding value that will work even when the
reputation is not in the (White) Caution, Black, and Truncate ranges.
When an IP's reputation is in one of those ranges then the appropriate
result from SNFIP will either be returned or not (On or Off).
Now-- I presume that even when SNFIP does return Caution, Black, or
Truncate that SNFIPREP continues to work and in that case will provide
some shading to those values... so, if you will, more or less Black, etc.
I don't think that I would necessarily use all of these together --
though it is possible to do so. It seems to be that it might become very
complicated since there is some overlap.
That said -- I do think that some of these tests can be combined
successfully without too much confusion... it's just a matter of knowing
how they interact. Hopefully my description is helpful (and my
assumptions are correct).
Best,
_M
--
President
MicroNeil Research Corporation
www.microneil.com
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to imail...@declude.com, 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 imail...@declude.com, and
type "unsubscribe Declude.JunkMail". The archives can be found
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to imail...@declude.com, 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 imail...@declude.com, and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
