I would be willing to wade through a number of additional spam emails to avoid losing important ones but of course this is Google so there is no user facing dial to adjust the sensitivity. Users just have to trust that Google's generalized approach is well calibrated for them.
I now skim my spam filter regularly because of this, but not everyone realizes they should do this.
Same, but I feel like it's almost kinda socially acceptable now. Happens to everyone and it's not something to get upset about... "oh, it wasn't me, it just went to spam." Like Gmail managed to alter our public norms instead of ensuring a zero false-positive rate :)
It sounds like a lot of people here check their spam folder regularly, which is good, but I don't know how widespread that is. I remember Gmail early on deemphasizing the spam folder since, in their view, the filtering was so good people didn't need to check it.
In the spam folder I just scroll down the list as fast as I can skim, and then close the tab if there's nothing.
If they were all just in my inbox I'd have to manually mark each one (and risk accidentally doing it to a real message while going through hundreds of messages).
There's no perfect solution here. Just personal preferences. I think I'd prefer a clean inbox with a 1% false positive rate vs having to manually flag a bunch of missed spam all the time.
I disagree with you. I use Postfix with rspamd plugged into it for my personal email account. I get way more spam to my gmail than I do to my personal account, and I sign up to everything with my personal account.
rspam also dkim signs my emails when I send them etc, verifies SPF/DKIM/DMARC on recipet etc.
Now to counter that - I am a TINY mail server - Probably 100 emails a day tops.
"I run my own mail server and get better spam results than Gmail"
Using Postfix+Rspamd gave me good insight into SPF, DKIM and DMARC and how to use them effectively.
rspamd is very, very impressive. I guess most of the hard work I've put into it is adding some of the not-turned-on-by-default things, like Pyzor and Razor. Also adding some other RBLs that weren't included by default (I spent a lot of time personally researching them and only picking ones that I believed to be of high value) The other big thing that I think is important is the RBL whitelists - DNSWL.org and HostKarma have a whitelist as well.
About one a week I spend 10-15 minutes looking at the logs of what it's accepted/rejected during the week to see if I can spot any obvious mistakes - it's pretty rare. If I do spot something I make config changes to address it. That said there's been months before where I haven't done this and none of the users of my platform have complained about spam (or missing email)
rspamd really is that amazing. I don't understand why more people don't scream it's praises from the rooftops.
I wouldn't be surprised if Gmail spam is higher-effort (like those individual SMS spam apps that politicians use) but higher-breakthrough.
Before I decided to leave it due to its horrendous false positive rate, gmail was driving like half of notification emails from my servers and mailing lists to spam, despite me never marking them as such. I was regularly missing important things.
It's much better with just regular client side bogofilter and some training on my personal mail/spam archive. And I do zero server side filtering, it's just all content based.
I don't care about capabilities, I just want near 0 false positive rate on the kind of email I receive (and not some common model), even at cost of some false negatives, and Gmail doesn't deliver there at all. And I don't want any arbitrary 5xx rejections for my senders, since I know how annoying that is on the sender side. Gmail will not guarantee that.
I think I have the opposite preference to you: false positives are OK to me if that means less spam gets through. In fact I've seen many of those notifications in my gmail spam and thought to myself, "Huh, you know, maybe I don't need those that badly after all... I'll just let gmail keep it there."
The overwhelming majority of my human contacts use other channels anyway (some chat app, or SMS), not email. I might get like ten real emails from humans in a year, and even then 90% of them are from people already in my contact list (and so bypass spam).
Phone calls are similar these days. Google Fi/Android also applies a similarly strict spam filter to incoming calls, and marks and blocks a lot of them as spam. I check once in a while, but overall I just don't really mind. If someone really needs to reach me they'll find a way... if they don't try, it's a good filter for how important their message really is anyway, lol.
I really wish Github had a DM feature =/ It feels so weird these days to email someone out of the blue.
And if you use non-GMail email providers, you would know they do fine. Not perfect, and of course it differs among providers, countries and accounts, but it's generally fine.
No it isn't.