[Bug 5636] New: Confusion of rules and scores for confirmed opt-in and opt-in status of email

[bu...@bugzilla.spamassassin.org]

http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5636

           Summary: Confusion of rules and scores for confirmed opt-in and
                    opt-in status of email
           Product: Spamassassin
           Version: unspecified
          Platform: All
               URL: http://www.isipp.com/iadbcodes.php
        OS/Version: other
            Status: NEW
          Severity: normal
          Priority: P5
         Component: Rules
        AssignedTo: dev@spamassassin.apache.org
        ReportedBy: amitchell@isipp.com
                CC: amitchell@isipp.com


Hey guys!

We're looking over the rules and scores for the latest release, and thrilled to
see that SA is taking fuller advantage of our unique code structure (designed
with the help of Craig Hughes specifically for SA to take advantage of, I might
add! :-) )... 

BUT (there's always a but, isn't there?), we're very confused about a couple of
things and are looking for some clarification:

1.  Where the opt-in level of the email is "IADB: Opt-in used more than 50% of
the time  RCVD_IN_IADB_OPTIN_GT50", there is a score of -0.499, but where opt-in
is used *all* the time ("IADB: All mailing list mail is opt-in 
RCVD_IN_IADB_OPTIN"), and even where it is all opt-in and *confirmed opt-in*
more than half of the time ("IADB: Confirmed opt-in used more than 50% of the
time  RCVD_IN_IADB_DOPTIN_GT50") the score is +1 (i.e. a spam score of +1).

Huh? :-)

So if they use opt-in sometimes (but *not* all the time, and not confirmed
opt-in) they get their spam score reduced (by .49) but if they are using opt-in
*all* the time, and even *confirmed* opt-in their spam score increases by 1?  

Help?

2.  Similarly, "IADB: Confirmed opt-in used less than 50% of the time 
RCVD_IN_IADB_DOPTIN_LT50" means that their email is all entirely opt-in, but not
all confirmed opt-in, we're confused as to why this *increases* their spam score
by 1 (yes, we push everyone to confirmed opt-in, but all of the other SA scores
for opt-in *reduce* the spam score - here, fully opt-in and even some *confirmed
opt-in* (just not 50% or more) is increasing the spam score.

3.  Where a sender's email is entirely confirmed opt-in ("IADB: All mailing list
mail is confirmed opt-in  RCVD_IN_IADB_DOPTIN") the score is -4 (thank you!),
but the *exact* same mail under the Habeas check ("Habeas Accredited Confirmed
Opt-In or Better  HABEAS_ACCREDITED_COI") gets a -8.  Can we pretty please have
the same score for the exact same email standard?

4.  "IADB: Sends no material under Michigan's CPR  RCVD_IN_IADB_MI_CPR_MAT" gets
a spam score of 1 - this means is that they send no alcohol, tobacco, weapon, or
adult themed materials  - we're confused as to why that would cause them to
increase their spam score?

5.  "IADB: One-to-one/transactional email only  RCVD_IN_IADB_OOO" also gets a
spam score of 1 - yet by its very definition, one-to-one email is very unlikely
to be spam - this is transactional email and corporate communications, so we are
confused as to why this takes a spam hit?

Thank you *so* much for any light you can shed on this, and any assistance you
can provide!

Anne



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug 5636] Confusion of rules and scores for confirmed opt-in and opt-in status of email

[bu...@bugzilla.spamassassin.org]

http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5636





------- Additional Comments From mkettler_sa@verizon.net  2007-08-31 21:46 -------
Please specify exact versions. I'm assuming you mean 3.2.3, which is the
latest officially released version of SA.

> 1.  Where the opt-in level of the email is "IADB: Opt-in used more than 50% of
> the time  RCVD_IN_IADB_OPTIN_GT50", there is a score of -0.499, but where opt-in
> is used *all* the time ("IADB: All mailing list mail is opt-in 
> RCVD_IN_IADB_OPTIN"), and even where it is all opt-in and *confirmed opt-in*
> more than half of the time ("IADB: Confirmed opt-in used more than 50% of the
> time  RCVD_IN_IADB_DOPTIN_GT50") the score is +1 (i.e. a spam score of +1).

RCVD_IN_IADB_OPTIN should have a score of 0, not +1.

Check your 50_scores.cf:

score RCVD_IN_IADB_OPTIN 0 # n=1 n=2 n=3

> 2.  Similarly, "IADB: Confirmed opt-in used less than 50% of the time 
> RCVD_IN_IADB_DOPTIN_LT50"
>   
again should be 0.
score RCVD_IN_IADB_OPTIN_LT50 0 # n=1 n=2 n=3

 > 3.  Where a sender's email is entirely confirmed opt-in ("IADB: All mailing list
> mail is confirmed opt-in  RCVD_IN_IADB_DOPTIN") the score is -4 (thank you!),
> but the *exact* same mail under the Habeas check ("Habeas Accredited Confirmed
> Opt-In or Better  HABEAS_ACCREDITED_COI") gets a -8.  Can we pretty please have
> the same score for the exact same email standard?
>   

Seems reasonable to me, IADB's S/O appears to be slightly better than
HABEAS as well. However, not my department to implement.


> 4.  "IADB: Sends no material under Michigan's CPR  RCVD_IN_IADB_MI_CPR_MAT" gets
> a spam score of 1 -

Ditto:
score RCVD_IN_IADB_MI_CPR_MAT 0 # n=1 n=2 n=3

> 5.  "IADB: One-to-one/transactional email only  RCVD_IN_IADB_OOO" 

Once more, a zero-score rule:
score RCVD_IN_IADB_OOO 0 # n=1 n=2 n=3

> Thank you *so* much for any light you can shed on this, and any assistance you
> can provide!
>   

Are you sure your install isn't hosed up somehow, or are you using a
version other than 3.2.3?






------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug 5636] Confusion of rules and scores for confirmed opt-in and opt-in status of email

[bu...@bugzilla.spamassassin.org]

http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5636





------- Additional Comments From amitchell@isipp.com  2007-08-31 22:31 -------
Hi Matt!

I appreciate the clarification - thanks! 

I was actually going by what is posted at:

http://spamassassin.apache.org/tests_3_2_x.html

So I guess the lesson here is to go by what is bundled into the release, yes?

Thanks!

Anne




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.