Bayes scores

[Kevin Miller]

I was reviewing my configs and noticed this in spam.assassin.prefs.conf:

# Bump up SpamAssassin scores on the high and low end
# score BAYES_00 -15.0
# score BAYES_05 -5.0
score BAYES_50 2.5
score BAYES_60 2.75
score BAYES_70 3.0
score BAYES_80 4.0
score BAYES_90 4.5
score BAYES_95 4.75
score BAYES_99 6.0

Given that spamassassin is constantly being tuned and updated, is is prudent to leave those uncommented or would one be better served by going with the spamassassin defaults? I'm sure Julian had a good reason to tweak them some years back, but given the ever shifting spam landscape I'm not sure if it's as applicable as it once was...

?...Kevin
--
Kevin Miller
Network/email Administrator, CBJ MIS Dept.
155 South Seward Street
Juneau, Alaska 99801
Phone: [phone], Fax: [phone]
Registered Linux User No: 307357

[Michael.Bradley]

+1

Great question and I've been wondering the same.

MB

-----Original Message-----
From: mailscanner-bounces at lists.mailscanner.info [mailto:mailscanner-bounces at lists.mailscanner.info] On Behalf Of Kevin Miller
Sent: Monday, April 28, 2014 1:22 PM
To: MailScanner List (mailscanner at lists.mailscanner.info)
Subject: Bayes scores

I was reviewing my configs and noticed this in spam.assassin.prefs.conf:

# Bump up SpamAssassin scores on the high and low end # score BAYES_00 -15.0 # score BAYES_05 -5.0 score BAYES_50 2.5 score BAYES_60 2.75 score BAYES_70 3.0 score BAYES_80 4.0 score BAYES_90 4.5 score BAYES_95 4.75 score BAYES_99 6.0

Given that spamassassin is constantly being tuned and updated, is is prudent to leave those uncommented or would one be better served by going with the spamassassin defaults? I'm sure Julian had a good reason to tweak them some years back, but given the ever shifting spam landscape I'm not sure if it's as applicable as it once was...

?...Kevin
--
Kevin Miller
Network/email Administrator, CBJ MIS Dept.
155 South Seward Street
Juneau, Alaska 99801
Phone: [phone], Fax: [phone] Registered Linux User No: 307357


--
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner

Before posting, read http://wiki.mailscanner.info/posting

Support MailScanner development - buy the book off the website!

Confidentiality Notice: The information contained in this email and any attachments is intended only for the recipient[s] listed above and may be privileged and confidential. Any dissemination, copying, or use of or reliance upon such information by or to anyone other than the recipient[s] listed above is prohibited. If you have received this message in error, please notify the sender immediately at the email address above and destroy any and all copies of this message.

[Jerry Benton]

Those are commented out by default in the prefs file. You should only
define them to override spamassassin defaults. See
http://spamassassin.apache.org/tests_3_0_x.html

Jerry Benton
www.mailborder.com

Post by Michael.Bradley
+1
Great question and I've been wondering the same.
MB
-----Original Message-----
mailscanner-bounces at lists.mailscanner.info] On Behalf Of Kevin Miller
Sent: Monday, April 28, 2014 1:22 PM
To: MailScanner List (mailscanner at lists.mailscanner.info)
Subject: Bayes scores
# Bump up SpamAssassin scores on the high and low end # score BAYES_00
-15.0 # score BAYES_05 -5.0 score BAYES_50 2.5 score BAYES_60 2.75 score
BAYES_70 3.0 score BAYES_80 4.0 score BAYES_90 4.5 score BAYES_95 4.75
score BAYES_99 6.0
Given that spamassassin is constantly being tuned and updated, is is
prudent to leave those uncommented or would one be better served by going
with the spamassassin defaults? I'm sure Julian had a good reason to tweak
them some years back, but given the ever shifting spam landscape I'm not
sure if it's as applicable as it once was...
...Kevin
--
Kevin Miller
Network/email Administrator, CBJ MIS Dept.
155 South Seward Street
Juneau, Alaska 99801
Phone: [phone], Fax: [phone] Registered Linux User No: 307357
--
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner
Before posting, read http://wiki.mailscanner.info/posting
Support MailScanner development - buy the book off the website!
Confidentiality Notice: The information contained in this email and any
attachments is intended only for the recipient[s] listed above and may be
privileged and confidential. Any dissemination, copying, or use of or
reliance upon such information by or to anyone other than the recipient[s]
listed above is prohibited. If you have received this message in error,
please notify the sender immediately at the email address above and destroy
any and all copies of this message.
--
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner
Before posting, read http://wiki.mailscanner.info/posting
Support MailScanner development - buy the book off the website!

[Kai Schaetzl]

I would certainly *not* change the defaults of the middle Bayes values as
given by SA, especially not Bayes_50. Only the values at both ends of the
scale. Moderately. (e.g. I set _99 to 5. That's good enough. If you have
other rules that constantly "tune down" the Bayes ruling too much for
getting caught as spam then you should look at the other rules and why
they do it and fix them (e.g. set to 0) as they don't seem to work very
well in your environment (or in general).
The FP risk is much too high for tuning middle values. And if Bayes runs
smoothly you get > 90 for spam.

Note, that latest SA introduced a new _999 that *adds* to the _99 score.
(At least I think it does, there was some discussion about this on the SA
list and I'm not sure about the final outcome.) It adds that little
"extra" that you may need to overcome the down-ruling mentioned above.


Kai