According to a recent article in a network rag I read, expect email spam to dissapear within the next two years. There are many techniques used to try to filter spam: keyword searching, black and white lists, hashes/signatures, heuristitcs, Reverse DNS lookups, header analysis, image scanning... but one in particular is gonna shake out this "industry". Bayesian filtering is ideally suited to run on the client machine, and tailer itself to the individual user's email, and shut out almost all of the offending crap. The best thing about it is its adaptive nature, so that spammers cannot just figure out what gets thru, and take advantage of it... if they did, the bayesian software would adapt to their new strategy, and start filtering it.
Now that the technique to do this is becoming known, expect products in the next year or two to take advantage of it.
Mozilla (www.mozilla.org) is a web browser and e-mail client which includes Bayesian filtering. It's free, just download it.
The filtering improves with time, but I think I have probably approached its peak after about 3 weeks of use. It's about 95% effective in detecting spam and sending it to the spam folder automatically.
But there is a downside, of that 95% it thinks is spam perhaps 3% is not spam. So in order to not delete e-mails you want you have to review what it thinks is spam. You can do this quickly because it puts it all together for easy review, but its still a problem! It still wastes a lot of my time.
The only solution I think can be 100% effective yet consume almost no user time is TMDA (www.tmda.org). It's a 'white list' solution. Unfortunately my ISP doesn't yet offer it.