It gives me a pre-screen in which I can preview text and sender, before allowing the traffic to hit my email accounts. It also allows for setting up a blacklist and a friends list. Oh, and my favorite? Bouncing the spam back to whence it came. At least that's what I hope the bounce is accomplishing; tying up spammers' servers.
Honestly, it's unlikely to be bouncing the email back to the spammer's servers. AFAIK, most spammers spoof the return address, so that replies will go to some hapless ISP somewhere. All it's really doing is: (a) using up your outbound bandwidth and your ISP's outbound bandwidth, which is unlikely to be a huge problem for you, but in the aggregate, it sucks for your ISP; and (b) taxing the spam filter and/or mail server at the ISP that the spammer spoofed.
Better choice is just to turn off the bounce feature, and dev/null the spam.
Is Thunderbird's a bayesian spam filter? Because if so, I'd bet you could just manually mark about a third or so, and then run the filter over the rest of the inbox. It would make a few mistakes but those (I imagine) wouldn't be too hard to sort out.
Oh, Thunderbird's filtering is Bayesian, but Mozilla developers think of their products as "experimental", regardless of how long they've been around. As a result, sometime around the run-up to Thunderbird 0.7, it seeems something changed in the spam training backend that causes Thunderbird to lose all prior training. I had to retrain the thing a few weeks ago and even after some 4000 spam messages thrown at it (thank goodness for an unused 8-year-old POP account I can always count on for 300 new pieces of spam a day for training purposes), it's only now starting to catch about a third of the spam.
There really is no excuse for this. If you're going to encourage millions of people to use a product and brag about how much better it is than the competition, then you have to make releases (even "technology previews") reasonably compatible even if it means someone has to spend an extra day or two writing an extra routine to convert existing spam training data when improvements bring changes to the backend.
s.m. is right. When it lost my entire Inbox, I figured it was a fine time to upgrade (from .5 to .7), and that's when I lost my filter settings. Annoying.
Dude, I love fubzy baby camels!
Posted by baa | Link to this comment | 07-12-04 4:37 PM
For dinner, probably.
Posted by ogged | Link to this comment | 07-12-04 4:38 PM
I take them for walks in the park, and buy them cotton candy.
Posted by baa | Link to this comment | 07-12-04 5:56 PM
I really like http://www.mailwasher.net/
It gives me a pre-screen in which I can preview text and sender, before allowing the traffic to hit my email accounts. It also allows for setting up a blacklist and a friends list. Oh, and my favorite? Bouncing the spam back to whence it came. At least that's what I hope the bounce is accomplishing; tying up spammers' servers.
Posted by rick pietz | Link to this comment | 07-12-04 8:57 PM
Rick -
Honestly, it's unlikely to be bouncing the email back to the spammer's servers. AFAIK, most spammers spoof the return address, so that replies will go to some hapless ISP somewhere. All it's really doing is: (a) using up your outbound bandwidth and your ISP's outbound bandwidth, which is unlikely to be a huge problem for you, but in the aggregate, it sucks for your ISP; and (b) taxing the spam filter and/or mail server at the ISP that the spammer spoofed.
Better choice is just to turn off the bounce feature, and dev/null the spam.
Posted by paperwight | Link to this comment | 07-12-04 9:43 PM
Is Thunderbird's a bayesian spam filter? Because if so, I'd bet you could just manually mark about a third or so, and then run the filter over the rest of the inbox. It would make a few mistakes but those (I imagine) wouldn't be too hard to sort out.
Posted by ben wolfson | Link to this comment | 07-12-04 10:25 PM
Oh, Thunderbird's filtering is Bayesian, but Mozilla developers think of their products as "experimental", regardless of how long they've been around. As a result, sometime around the run-up to Thunderbird 0.7, it seeems something changed in the spam training backend that causes Thunderbird to lose all prior training. I had to retrain the thing a few weeks ago and even after some 4000 spam messages thrown at it (thank goodness for an unused 8-year-old POP account I can always count on for 300 new pieces of spam a day for training purposes), it's only now starting to catch about a third of the spam.
There really is no excuse for this. If you're going to encourage millions of people to use a product and brag about how much better it is than the competition, then you have to make releases (even "technology previews") reasonably compatible even if it means someone has to spend an extra day or two writing an extra routine to convert existing spam training data when improvements bring changes to the backend.
Posted by s.m. koppelman | Link to this comment | 07-13-04 6:36 AM
s.m. is right. When it lost my entire Inbox, I figured it was a fine time to upgrade (from .5 to .7), and that's when I lost my filter settings. Annoying.
Posted by ogged | Link to this comment | 07-13-04 8:50 AM