I visit Snopes every other day or so, checking the 'Additions' page to see what new (stupid|deranged|utterly nonsensical) thing people are emailing around to each other or passing off as fact in casual conversation.
I've noticed several times that my stepmother (whose family is comprised of forty or fifty spamaholics) sends me one or two of the Snopes items a few days after Snopes points to it. It then occurred to me that a component could be built to a spam filter where Snopes entries would be scraped for keywords, which could be weighted based on the date it landed on Snopes, and used to identify spam.
Once I get the hang of scraping pages (which is something I've been meaning to play with for a long damn time now), I plan to take a shot at this.
As a bonus, there are some patterns that the webmaster of Snopes has identified in topical spam revival (for instance, the series of war-related spam and horror-related spam that followed the WTC collapse), and if these were more closely identified, weight patterns could be shifted in anticipation.
