New Scientist, 27 October 2010 by Jim Giles
Spammers and astroturfers beware: computer scientists have a new method for catching attempts to artificially inflate the popularity of people and ideas through Twitter. The researchers have already identified two groups that are using suspicious strategies to promote Republican causes in the run-up to next week's midterm elections.
Political activists sometimes operate multiple Twitter accounts that appear independent but actually carry the same or very similar messages. To outsiders, it seems as if many different Twitter users are saying the same thing, when in reality it is a single activist putting out their message.
Ploys like that may soon be detected automatically. Filippo Menczer and colleagues at Indiana University in Bloomington collected suspicious tweets [see truthy.indiana.edu] by looking for unusual activity, such as sudden bursts of similar messages. After investigating the patterns of tweets generated by and between Twitter accounts they found several ways in which some accounts were being used in potentially deceptive ways.
One involved a pair of anonymous and apparently automated accounts named @PeaceKaren_25 and @HopeMarie_25 - both now suspended by Twitter - which mainly relayed messages from Republican politicians. The former generated over 10,000 tweets since June; the latter retweeted them, but produced no messages of its own, says Menczer.