[Previous entry: "RIP: A Musical Journey"] [Main Index] [Next entry: "Happy One Year House Anniversay"]

06/17/2002 Entry: "Anti-Spam Statistics"

A co-worker pointed out that I have not bloggedi n a while. So today's boring geeky topic is some analysis of my Anti-Spam efforts.

Unhappy with how I was filtering (with filter) mail off of the shell server I have access to, I went on a hunt for new anti-spam software. I actually started this even before we left the old house. I learned how to use Procmail. Not hard, but not easy. Eventually, I found a nice, highly configurable set of Procmail recipies called JunkFilter. One of my testing/usage constraints is that I did not want to change any of the echoes.com addresses I have been using. And since I've had the domain for 6 years there's been a lot of search engines and people used to finding me. So I found a free shell server and bounced all of my mail there while testing. Once I got used to JunkFilter, I installed it on my shell server and started mirroring the configuration. I soon found out that my ISP was using SpamAssassin by defaut for each mail message being delivered to their shell users like me.

So now I'm using both... and SpamAssassin is doing better!

One major problem I have is that there is an ISP called echoes.net. So some of the less savvy users there sometimes use .com instead of .net. Trust me I get a lot of interesting mail. I've collected a fair number of these email addresses and any mail to those addresses are killed instantly. Some get through, but if it becomes a problem. Some of them I can't get rid of. I get some mail from popularmechanic.com and MSN Gaming Network that I can't kill and can't unsubscribe. Outlook catches them, but that's another blog.

After 19 days and 4291 emails delivered to me....

08.25% caught by JunkFilter
36.43% is sent to echoes.net email addresses that I kill
01.26% is using Big5 character set and I kill it
01.72% has invalid time zone information... and I kill that too
00.35% has no "To:" header... killed!
51.99% actually gets delivered to my mailbox... and I would guess that 30% of that mail is actually caught my my filters in Outlook.

Not great actually... but it's better than none of it getting filtered!