- 264comments
- 251comments
- 44comments
- 5comments
- 272comments
- 68comments
- 54comments
- 78comments
- 58comments
- 12comments
- 62comments
- 246comments
- 5comments
- 183comments
- 15comments
- 1comments
- 179comments
- 26comments
- 19comments
- 55comments
- 201comments
- 4comments
- 91comments
- 5comments
- 41comments
- 16comments
- 186comments
- 3comments
- 152comments
- —discuss
https://imgur.com/a/nzkMNDS
The message from Facebook in the screenshots is about as technically good as you can get AFAIK. It passes SPF (hard fail record), DKIM, and DMARC (reject policy). Is there a way I can tell GMail / GSuite that I don't want it to ever SPAM filter email that passes all those checks?
I understand SPAM filtering is hard, but I don't think there should be any question about messages from high value domains like facebookmail.com.
Related, a description of the problem, but not a solution:
* Google Is Eating Our Mail
https://news.ycombinator.com/item?id=19756125
My personal mail server has all the industry-standard SPF, DKIM, and DMARC records. But Google still randomly eats my mail, and the system gives no explanation, sometimes it was my kernel patches submitted to the Linux maintainers, yikes! Weeks of time wasted.
The filtering is schizophrenic. Sometimes I think it's because normal users are incorrectly flagging stuff as spam, but I can't understand the one I posted. I don't think anyone is signing up for CT Log notifications and then flagging them as spam. It must be a poor quality algorithm somewhere that's flagging it.
It's understandable to flag mails from reputation domains as spams if there's no SPF and DMARC, but there is. It's totally unreasonable.
The unfortunate result of machine learning? I guess Google's spam filter has lots of pattern-matching and heuristics that don't follow "the rules of E-mails" exactly. Anecdotes in the comments I linked says Google's spam filter has became more aggressive in recent years, I suspect it must be ML.
I hope other HN readers can give a more informative answer to your post, as a sender, I'm having the same problem, it's two sides of the same coin...
I just read through the comment section of this article and, as someone who is planning to host my own email, it is really demotivating me.
Do you (or anyone) happen to know how the big players (AWS SES, SendGrid, Mandrill etc) work around this issue? I understand they offer dedicated IP address, but I suppose a considerable amount of their emails goes through the "shared pool".
Maybe a high volume of sends actually helps?
I believe they are simply "too big to fail". If you start blocking AWS, lots of people will have problems with Google. As a result, mail sending services are unlikely to be flagged.
The good thing is, I've hosted emails on my personal server for 4 years or so, and I don't have any issue so far, as long as it's not Google, but it's probably not so assuring to you because of this exact reason...
Funnily enough, cert issuance notifications (not Let's Encrypt) and infos about upcoming maintenance from my hosting provider are pretty much the only things that end up in my GMail spam folder, too. I have serious doubt that many people flag "here's your cert" emails as spam, and you have to sign up for the maintenance emails, so I suspect it's mostly about "this email looks similar to many other emails".
I've marked them "not spam" many times, but they ignore that completely.
This can also happen with shared email providers or email servers on shared IPs. If someone else is sending spam from that IP, you can get canned too