View Single Post
Old 13 Jan 2019, 12:28 AM   #14
Cornerstone of the Community
Join Date: Mar 2002
Location: Florida
Posts: 541
Originally Posted by n5bb View Post
Thatís unfortunate, Edward. Have you whitelisted the From addresses in your address book? If so, check the X-Spam-known-sender header to see if authentication detection is somehow failing.
Hi Bill,

I automatically whitelist my address book. IOW, my Sieve code that tests spam scores is guarded by

if not header :contains ["X-Spam-known-sender"] "yes" {
But there are limits to this approach.
  • I use a desktop email client (The Bat!), and I've not been able to persuade FM and Ritlabs to work together to get them to sync address books. (I'll admit, it's been a year or so since I last nudged them.) As a result, FM's knowledge of my address book lags.
  • I don't want to have to enter every contact into my address book just to circumvent a wildly inaccurate spam test.
  • I certainly don't want to have to enter new contacts before even receiving the first email from them.

I did get a response from Yassar Ali on Jan 4 -- forgot to come back here and update, so thanks for the reminder -- saying "it appears the reason for these false positives has now been mitigated". Of course that's the same thing he told Lesslame in mid December. Since them, I've been keeping an eagle eye on what drops into junk mail, and I've had no more false positives, and I have received email from the personal correspondent who was being blocked.

I'm really more upset that FM is introducing new tests that essentially override all other tests, instead of starting the new tests with small scores and gradually ramping them up. A few months ago it was ME_PHISHING_URL at 10, now ME_VADE_SPAM at 5. Basically they are repeatedly saying oh, we have this new test that we immediately trust more than all the tests (including the Bayesian ones) that have been in place for years.

paleolith is offline   Reply With Quote