On Fri, Jan 24, 2020 at 1:27 PM Gregory Heytings via mailop < mailop@mailop.org> wrote:
> > Brandon Long: > > > > > sender in addressbook is definitely a whitelisting signal, as is > > replying to a message the user sent or on the same thread. They used to > > be much stronger whitelisting signals than they are now, but were abused > > by spammers, so it's not a guarantee. > > > > I stand corrected on those points. I'm not inside Google (alas ;-)), so > the only thing I could do is by experimenting things, and from my > experiments I concluded that these things do not make a significant > difference. Obviously you know better than me what actually happens. > > Still, this does not solve the OP problem: how to make sure that > "first-time" emails arrive in the inbox of his (or his wife's) recipients. > I still believe that this is what happens with legitimate emails sent by a > correctly configured server. > There is no way to guarantee that a first-time email arrives in the inbox. If there was, the spammers would all use it. The best you can do is "attach" your email to some existing source of reputation. Unfortunately, running your own mail server for 20 years sending <10 messages a month to Gmail isn't an existing source of reputation. Where you're hosting your mail server is... and it's usually bad. The most common thing is to use the smtp-relay server provided by your hosting provider. They won't be perfect, but they're probably better than the IP space of their hosting. > Having your authenticated mail marked as not spam by the user is still > > the strongest signal you can use, though sometimes it may take doing it > > on 2-3 messages... or maybe more if you previously marked it as spam. > > Okay, but this is definitely not how 99% users use their spam folder: they > simply never look at it, and if they do they do from time to time there is > about 90% chance that they will not see the few false positives in the > list. > We're aware of the challenge, I went into this in the last thread. Obviously, the better our spam checking is, the less effort people use to validate it. Is there perfect false-positive/false-negative ratios where enough people see the signals but don't think your anti-spam system is terrible? Dunno, ask us in 10 more years and we'll see how we're doing then. This was for a classroom, however, so there's a very clear mechanism by which an out-of-band communication can occur to look into the spam label and fix it... presumably also obvious by the fact that the person knew it went to spam for everyone at Gmail. Brandon
_______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
