[WBEL-users] Re: SpamAssassin without Razor

[Mike Staver]

Paul M. Bucalo wrote:
> On Mon, 2004-10-25 at 11:48, Mike Staver wrote:
> 
>>Yes, I've been using spamassassin without issue for over 2 years now. 
>>However as I said, apparently somebody figured out how to get around it. 
>>  Here is some header examples of spam that slipped by it:
>>
>>X-Spam-Checker-Version: SpamAssassin 2.64 (2004-01-11) on mail.fimble.com
>>X-Spam-Level: **
>>X-Spam-Status: No, hits=2.9 required=3.5 
>>tests=HTML_40_50,HTML_IMAGE_ONLY_04,
>>	HTML_MESSAGE,MIME_HTML_ONLY,PRIORITY_NO_NAME autolearn=no version=2.64
>>
>>X-Spam-Checker-Version: SpamAssassin 2.64 (2004-01-11) on mail.fimble.com
>>X-Spam-Level: ***
>>X-Spam-Status: No, hits=3.1 required=3.5 
>>tests=HTML_40_50,HTML_IMAGE_ONLY_06,
>>	HTML_MESSAGE,MIME_HTML_ONLY,PRIORITY_NO_NAME autolearn=no version=2.64
>>
>>X-Spam-Checker-Version: SpamAssassin 2.64 (2004-01-11) on mail.fimble.com
>>X-Spam-Level: ***
>>X-Spam-Status: No, hits=3.4 required=3.5 tests=BIZ_TLD,HTML_60_70,
>>	HTML_FONTCOLOR_RED,HTML_IMAGE_RATIO_14,HTML_MESSAGE,HTML_WEB_BUGS,
>>	MIME_HTML_NO_CHARSET,MIME_HTML_ONLY,UPPERCASE_25_50,WHY_WAIT
>>	autolearn=no version=2.64
>>
>>X-Spam-Checker-Version: SpamAssassin 2.64 (2004-01-11) on mail.fimble.com
>>X-Spam-Level:
>>X-Spam-Status: No, hits=0.6 required=3.5 tests=HTML_40_50,HTML_MESSAGE,
>>	MIME_HTML_ONLY autolearn=no version=2.64
> 
> 
> Hi, Mike,
> 
> I use POPFiles, not Spamassassin, so I wish to apologize right up front
> for my ignorance of SA. I noticed something in your headers that leads
> me to ask you this question:
> 
> Shouldn't 'autolearn=' be set to 'yes' instead of 'no'? Isn't this where
> you let the 'AI' aspect of SA learn to do its thing?
> 
> Sorry again for my ignorance. Just had me wondering why you had this
> switch set to 'no'.

Yes, you can set SA to autolearn and I have done that in the past - but 
I also had situations where I was getting false positives in version 
2.55, so that false positive would then be "learned" as spam, and then 
there was a process you had to go through to "unlearn" it. Even with a 
threshold of only 3.5, I haven't had a false positive in about 6 months 
with version 2.64, so I should probably consider turning this feature 
back on.  Thanks for the input Paul, it's appreciated as always.
-- 

                                 -Mike Staver
                                  staver@fimble.com
                                  mstaver@globaltaxnetwork.com