Re: [exim] Gmail's new 'suspicious sender' flag

On Wed, 2011-07-06 at 19:34 +0300, Mihamina Rakotomandimby wrote:
>
> There are two topics:
> - SPAM detection and tagging
> - SPAM blocking
>
> > If you check at the MTA, then you
> > can arrange this (false positives and false negatives
> > notwithstanding).
>
> If you _check_ at MTA level, you can help the user to make the right
> decision, with a clever scoring method.
>
> Then after your check, if you block at MTA level, I think it's bad.

If the mail isn't going to be read, it should be rejected. That's the
only way that *genuine* senders are going to know that their mail isn't
received, in the case of false positives.

False positives *do* happen. To design a system that makes them *silent*
failures is just wrong.

You *have* to reject at SMTP time, if you want the system to still be
considered reliable.

> > If you leave it to the end user, it's too late to reject the email.
>
> If the user uses some filtering tool at MUA level _and_ the scoring is
> good quality, the final user wont see any SPAM.
>
> > As far as the spammer is concerned, the email is delivered.
>
> I dont agree.

s/spammer/sender/. And I don't see how there is any rational scope for
disagreement.

--
dwmw2