Timothy Taylor summarizes a fascinating new paper on the cat-and-mouse game between e-mail spammers and service providers:
For example, when many people label a message as “spam,” then it helps the anti-spam software to look for those words or URLs repeated in other messages, so that those messages can be filtered out. But then spammer responded with creative misspellings (like “VIagrA”) to trick the anti-spam filter, and used many different URLs that would all take the unwary to the same sales page.
In addition, the spammers use software to mark messages as “not spam,” thus trying to offset those who label them as spam. Rao and Reiley write: “In four months of 2009 Yahoo! Mail data, our Yahoo! colleagues found that (suspiciously) 63 percent of all “not spam” votes were cast by users who never cast a single “spam” vote.”
Talyor also kindly links a free download of the paper, published in the Journal of Economic Pespectives, which he edits.