I wrote about a spam message that a friend of mine received and asked me about because she was suspicious. It's got a different line of making money which is much more seductive than the usual suspects and probably much more effective. They've a few methods to realistically guess the email addresses of people who may have used online dating services. Then they send out an email that is personal enough (and is missing all kinds of spam clues that automatic filters rely upon) that many recipients feel that just maybe it is legit. You have to actually pay money to a dating service to respond though. Much easier to get people that have joined one dating service to join another than selling a viagra (especially when the recipients are girls!).

http://twp.zrox.net/blog/archives/000027.html

I've had lots of people from people who were a little bit suspicious and found my page and felt betrayed; I've left comments open and let them rant a bit. I'd guess that there were a lot of people who were not suspicious enough, got duped, and would never admit it. I posted in mid-November and still tend to see over twenty hits a day about it.

The real twist, and why I posted in the first place, is that I see no way for automatic methods to determine it is spam without a person at some point inspecting the message, recognizing it as spam, then then identifying specific characteristics for blacklisting. Of course that is not fully automatic.

Travis: Look into Bayesian filtering. It's not a silver bullet, but it basically learns what is and isn't spam by example.

http://spambayes.sf.net/ is a good filter which has an Outlook plugin and so on.

Personally I run bogofilter (a similar project, though with different tweaks to the algorithms; I'd switch to spambayes if the database could be migrated, though), and I only get maybe one false negative per week, and a false positive maybe once a month and it's typically been for a message I was expecting and had a feeling it would show up in the spam folder.

A learning AI isn't as good as a human reader, but it's a lot more time-efficient.

I get called those sorts of things all the time, so even filtering might not work. Looks like I am in trouble!

I actually wrote the Spam Rant the other day but stored it (b/c I had other things to get out there first) but when I saw this, I thought I trackback to it - seemed too perfect not to :)

you know Candy and I have had this discussion many times--can a woman be an "asshole?" it seems almost always applied to males. it seems that if one wishes to insult a female there are harsher and more degrading terms than "asshole..."

Just wanted to let you know that when I accidently came across your site from some search device, I discovered it to be quite interesting. Thanks, Jim

I'd offer to buy you coffee like the kid who heard you sing (one of your many writes), but I live in Dallas, TX and the drive is a bit much for me:)

Take care, Jim

Why don´t you use java for the comment links so that no PR will given away?

The one I still remember as getting past my mental filters had subject '$376.25 has been charged to your account'. I thought that was a pretty clever way of bypassing my calm and rational consideration of whether I should open it...

Like you say, though: only once.