I'm a very heavy email user, and after a failed attempt to use the shiny new OS X "Mail.app", have reverted to Mozilla, curently the 1.3beta. Both have remarkably effective spam filters. Despite Barry Shein's pessimism in today's Slashdot, I think we may be winning.
I'm a pretty good guinea pig for this, since my email address appears on the front of the XML specification and is splashed all over Usenet and tons of mailing lists; I get hundreds of spams per day. Both the Mozilla and Mail.app spam filters are amazingly, remarkably, effective. The false-negative rate (spam that gets through) declines to almost-unnoticeable, and the more serious false-positives (real email that gets junked) have gone to exactly zero in recent days. It won't be long before I start neglecting my occasional surveys of the "Junk" folder.
It's weird to be away from the computer for a couple hours and see
unread in my
at work inbox and
in my personal inbox, and know that I actually have 8, count 'em, 8
emails that I ought to look at.
(I'm excluding mailing-list traffic that already gets filtered out of the
I think this highlights one of the real social costs of spam; pre-filter, I didn't know at a glance how big my unread queue really was, but now I do. This is a quality-of-life improvement.
I suspect that both Mail.app and Mozilla are using something based on Paul Graham's work and follow-up, although neither tool (for understandable reasons) advertises its approach. For those who care about this and haven't read Graham's work, it's worth a look, and gives real hope that this is one class of filter that the spam scumbags won't be able to work around.
If Mozilla and Mail.app have this in place, it can't be long before Outlook, Gnus, and the rest of the mail readers out there have it too. Paul Graham argues convincingly that at this point, maybe the economic incentive to spam will have dried up.