But if it ends up in the spam folder there's at least a chance that it'll be trained out of the spam folder by people marking it as "not spam", and it provides a workaround for the people who really need it to work _now_.
I wonder if the correct solution is "send an informative pseudobounced-as-spam message to sender AND put it in the spam folder", for borderline spam (which passes the other checks). Downside is it provides a deliverability oracle for spammers, but since gmail accounts are free, that's not too hard to establish anyway.
I believe that the majority of the users are trained to ignore g-mail spam since it's been working good enough for most people. Unless we condition the customer to always view the spam folder it's asking them to do something on top of already checking their e-mail.
My mother would just simply ignore the spam and often times spam catches majority of the phishing e-mails too. So it's a double edge sword educating and conditioning the users to review the spam folder. Why is it the end user's job to determine what is spam and what is legitimate ?
> Why is it the end user's job to determine what is spam and what is legitimate?
Who else could determine that? I could hire you to send me emails about Viagra sales. They would not be unsolicited, because I specifically asked you to. They wouldn't be commercials, either, so they wouldn't be Spam. An automatic filter can't determine if it's spam or not, it can only take an educated guess.