Email Address Harvesting and the Effectiveness of Anti-Spam Filters

Citation
Federal Trade Commission, Division of Marketing Practices, Email Address Harvesting and the Effectiveness of Anti-Spam Filters (Nov. 2005) (full-text).

Overview
This report described the results of an FTC staff study of three aspects of spam in the then-current Internet environment. First, the study explored the current state of email address harvesting &mdash; the automated collection of email addresses from public areas of the Internet. The study found that email addresses posted on websites were at risk of being harvested by spammers, but that addresses posted in chat rooms, message boards, USENET groups and weblogs (blogs) were far less likely to be harvested. Indeed, some chat room operators took proactive measures to prevent the harvesting of email addresses posted by FTC staff.

Second, the study explored the effectiveness of spam filtering by Internet service providers (ISPs). The study showed that the anti-spam filters utilized by two free web-based ISPs effectively blocked the vast majority of spam sent to harvested addresses. The implication of this finding is that ISP spam filtering technologies are substantially reducing the burden of spam on consumers. Nevertheless, spam sent to harvested addresses imposes costs on ISPs receiving the spam.

Third, the study measured the effectiveness of using “masked” email addresses as a possible technique in preventing harvesting. The “masking” of an email address involves altering the appearance of an email address so that it is understandable by a person who sees the address, but less likely to be discernable by automated harvesting software. For example, to mask an unmasked email address such as “johndoe@ftc.gov,” the words “at” and “dot” can be written out, and segments of the email address can be separated by spaces. The masked version of the address would appear as “johndoe at ftc dot gov.” The study found that the “masking” of an email address was very effective in thwarting harvesting.