I hadn't thought about this before, but someone told me that the renewal emails I send out for The Perl Review get a high enough SpamAssassin score, about 3.5ish, to be blocked and probably never seen.
That's no good. Since we're about to start our third year of print issues, it's time for most subscribers to renew. If you haven't seen your renewal message, you can still renew online without it, but I'd also like to know if a message titled "The Perl Review: Time to renew" showed up in your spam folder so I can fix that for next time.
I checked my code and found it didn't have the parts i thought I had put in last time, including names in the To and From fields. I also got a report that the email might be missing Message-ID. If anyone would like to help me investigate this, please send your raw message with headers to me (it's the reply address on the email).
This is actually a big deal in the magazine industry as people tyr to figure out how to get around bulk mailings and giving money to the post office.
What SpamAssassin tests does the mail match?
This is actually a big deal in the magazine industry as people tyr to figure out how to get around bulk mailings and giving money to the post office.
SpamAssassin isn't that mysterious. It's easy to find out what the tests match on, and therefore you can change your mails so that they don't match on those tests.
Re:What SpamAssassin tests?
link on 2006-07-28T09:27:08
X-Spam-Status: No, score=3.335 tagged_above=-999 required=5 tests=[DNS_FROM_RFC_ABUSE=0.479, MSGID_FROM_MTA_ID=0.927, NO_REAL_NAME=0.55, SPF_NEUTRAL=1.379]
This claims to be from Copperfasten Mail Firewall but they seem to be SpamAssassin rules. As was mentioned message id and to/from names are missing.Re:What SpamAssassin tests?
jmason on 2006-08-24T10:25:15
Those do look a *lot* like SA rules, which is odd -- the copperfasten web site doesn't mention SA! As an SA developer, I find this interesting...
Anyway, it's expected that occasional non-spam will score about 3.3 points -- and a newsletter sent to many subscribers would be quite typical for that score range.
The SA default threshold is 5, which is quite a bit higher than 3.3; anyone using a threshold that low needs to expect to spend some time monitoring for false positives, and rescuing and whitelisting them. Personally, I would not recommend it, but it's up to them!Re:What SpamAssassin tests?
link on 2006-08-25T18:43:04
Between
http://www.dcu.ie/csd/spam/technical_changes.shtml and http://www.dcu.ie/csd/spam/copperfasten.shtml
it sounds like whatever they are selling includes spam assassin.
Re:no prob
brian_d_foy on 2006-07-28T15:39:43
I've seen some headers now that people have sent me their SA scores, and when I get back from traveling I'll set up SA on one of my machines to investigate further.
It looks like SPF_NEUTRAL gets a pretty high score, so I'll have to think about that.Re:no prob
jmason on 2006-07-28T16:23:43
good point, I hadn't noticed that. interesting.
it has got a pretty high correlation with spam; 95% of the SPF_NEUTRAL hits fire on spam these days (3.2% of spam, only 0.16% of ham). still, it's nonintuitive.
Re:Notifications
brian_d_foy on 2006-07-28T15:54:17
It's just the people who need to renew. In gerenal, if you wonder where you are in your subscription, send me email. I am working on a way to put this online without divulging personal details, but that's not fully finished.
Your subscription, however, ends with the issue after the next one, so you're okay right now.:) Re:Notifications
sigzero on 2006-07-28T16:46:54
Cool, thanks a bunch.Re:Notifications
shild on 2006-07-28T17:00:58
If we renew early, it justs adds to year to your account anyways, right? So just renew early.:-)
