When the anti-spam war started in mid 90s, it mainly relied on detecting common words present in spam and manually blacklisting spam IP addresses. Historically, there has been much controversy around blacklists because of the subjectivity of getting placed on the list and the difficulty of being removed. Whitelists have been developed that maintain lists of legitimate senders. A problem with whitelists and blacklists is that they left a sizable set of senders in the middle of the spectrum that were not classified. This is because there was not enough credible information or feedback to make a binary decision about the sender. As spammers started using fast changing botnets, randomizing and obfuscating content in their messages, those technologies quickly became ineffective. Recently we begin to see more adversarial intelligence and data mining from spammers. Spammers continue to accumulate vast data stores on vulnerable systems and individuals. To address this problem, reputation systems for email became a topic of interest in anti-spam research around 2003.