On Fri, Jul 16, 2021 at 05:12:38PM +1000, raf <post...@raf.org> wrote:
> On Wed, Jul 14, 2021 at 02:51:25PM +1000, raf <post...@raf.org> wrote: > > > On Tue, Jul 13, 2021 at 10:35:15PM -0400, Bill Cole > > <postfixlists-070...@billmail.scconsult.com> wrote: > > > > > On 2021-07-13 at 21:18:46 UTC-0400 (Wed, 14 Jul 2021 11:18:46 +1000) > > > raf <post...@raf.org> > > > is rumored to have said: > > > > > > > I'm beginning to think that DKIM headers might be > > > > getting added just to improve spam detection scores. > > > > Perhaps I'm getting too cynical. :-) > > > > > > That would not be very effective. > > > > > > For example: in Apache SpamAssassin, the presence of a valid DKIM > > > signature > > > has a net zero score. If it is valid and aligns with both the envelope > > > sender and the From header address, it can net only -0.2 in a scoring > > > system > > > with a standard spam threshold of 5.0. That's not quite a meaningless > > > benefit, but it is not substantial. > > It seems that there are corporate mail services that > operate differently (well, at least one). > > According to this: > > https://postmarkapp.com/blog/proof-dkim-and-senderid-improve-delivery > > DKIM can mean the difference between mail being put in > the Junk folder or the Inbox (when there's a PDF > attachment). It might be to help services like > mailchimp. I've received mail From: gmail.com with a > mailchimpapp.net DKIM signing domain (so really sent by > mailchimp). And mail-tester.com, which purports to tell you how "deliverable" your mail is, scores mail on a not-spammy-scale from 1 to 10. One whole point (10%!) is given for DKIM signing. That's sad. I like the fact that the mere presence of a DKIM header makes spamassassin treat an email as ever so slightly more spammy (+0.1), unless it's valid (-0.1), in which case it has no bearing on spam status, unless it also matches the from address (-0.1), but even then it's not important. cheers, raf
