Re: [Evolution] Evolution and gmail

[Pete Biggs <pete biggs org uk>]

On Thu, 2022-08-18 at 04:10 +0200, Ángel wrote:
> On 2022-08-16 at 23:24 +0100, Patrick O'Callaghan wrote:
> > Spam filtering is under the control of the receiver. It's trivial to
> > tell Gmail that a message is not spam, and it will learn that for
> > future reference. You might also look at *why* your mail is being
> > classified as spam. Could it be that some people have marked it as
> > such?
>
> Actually, I would consider this a weak point of gmail. As a receiver,
> you don't know why a message is considered spam (compare that with a
> system like SpamAssassin, where you can view the scoring of the
> different modules), and even Google itself would often be unable to
> determine why the ML engine considered it bad.

This is getting way of topic, but ...

My work uses Microsoft O365.  We have similar issues with bad spam
filtering at times, including it binning internal mailing list mails.
The mail admins have tried to get some insight into the spam
classification process so that we can get some idea how to avoid it,
but MS won't talk about it because it's proprietary confidential
information. At first glance that's a "typical big brother" attitude,
but they say that if the spam classification algorithms are known,
spammers will just craft emails to get around them, which I can
understand.  I wonder if Gmail has the same sort of thinking on it.

BTW, we think the reason the mailing list mails were being spam binned
was because we have a very very large userbase (big university, lots of
staff and students), with some compulsory lists for communication. All
it takes is for people to blindly click on "I don't want to see this
sort of mail any more" thinking it gets them off the list, whereas it
just increases the spam score and eventually - actually quickly - gets
our mailing list servers globally sunk.

P.