|Comments on Wednesday 21 August 2002:|
|A rather nice statistical spam filter, which gave me other ideas. Once you've trained the filter to recognise spam, why not try training it to recognise other things? Filter your non-spam email into 'interesting' or 'boring' categories. Ask it to tell you what gender the writer is. Getting to look at the statistics later would be interesting, even if it was unable to make correct judgements.
|Raven, you seem to be into spam filters? I had an idea for one but I don't know if the idea is viable, or if someone has already implemented it.|
The idea being to have an extra e-mail address and to never leave one a-mail-address on a page without the other. This way, if a spam-email-sniffer-bot comes by, it should register both addresses, and thus, future spams would target both addresses. The simple filtering mechanism would be to throw away any mails that it received at both addresses.
I know that this approach cannot be used one the genie is out of the bottle, but for a fresh, untainted e-mail-address it might work? Or what do you think?
Just an idea ... don't plan to implement it but I am curious for comments, though.
Identifying duplicates would potentially be a pain too, as they don't always come in at the same time. When I get four-times-duplicates of spam, they tend to be spread over the course of about an hour and half.
|Most of the spam I get seems to come via email lists. I can't remember the last time I got spam that was actually directed at my personal address rather than at a mailing list I'm subscribed to. And I'm not particularly protective of my personal address. I stick it on web pages, in sign up forms, whatever. From this I assume that spammers tend to target email list addresses in preference to individual addresses - makes sense of course, send one email and get somebody else's server to turn it into a thousand. Now, it's actually not that hard to filter out spam once its reached you. SpamAssasin http://spamassasin.org seems to do quite a good job for me. There are a number of other useful filtering programs available. The thing is that before you can filter the mail, you have to download it to your local machine and that costs money and bandwidth that could be better spent downloading amusing .swf files. What's really needed is a genetic filter to be applied at birth to seperate the spammers from normal citizens and make them go work in coal mines with amusingly small pick axes. And they should have to sing the Hi Ho song. This is the only way to reduce the cost of spam for the average citizen. (Yes, average. You think Bill Gates or Lunix Torvalds gets spam? Hah!)|
|There are some spam programs out nowadays that basically do dictionary-spamming; get a domain name, then apply username after username to it, and see if anything comes up. So just because you're conscious of where your email goes, there's no guarantee it won't be found anyway.|
Plus, someone whom you've sent email to might wind up with a virus, and have your address harvested from their inbox/address book. No real way to win.