‘Majority of bots on Twitter cannot be discovered’
Most bots active on Twitter are not detected by the social network. This is shown by experimental research conducted by the Federal University of Minas Gerais in Brazil.
The scientists investigated how so-called social bots can infiltrate the social network without being discovered, writes MIT Technology Review, among others, on Friday. They created 120 bots on Twitter and kept them active for a month. Of these, 38 were caught by Twitter, while the other bots managed to evade control. Of the latter group, some gained an influential position within the Twitter community.
The researchers tried to make the bots look as real as possible. For example, the profiles were immediately assigned a gender and a few followers, to arouse less suspicion. The bots generated messages by retweeting tweets from Twitter users or by combining words about a particular topic into a single sentence. Before that, programs used rules that were predefined.
The bots weren’t always active. For example, some posted within a few hours, while others published something every two hours. There was no posting between ten o’clock in the evening and nine o’clock in the morning, because at that time the ‘real’ Twitter users were also asleep.
The research yielded different results. Twenty percent of all social bots gained more followers than regular Twitter users. In addition, some bots were able to achieve a higher social status than known human Twitter users within a month. Finally, the bots that generated tweets from certain words were found to outperform, suggesting the scientists say that Twitter users are not good at distinguishing between messages posted by humans and those posted by robots.
According to the researchers, social bots can lead to various undesirable consequences. For example, it appears difficult for Twitter to act against the bots, which would encourage spam on the social network. In addition, the scientists warn that information from Twitter cannot be used haphazardly in research into sentiments about, for example, a brand or an event such as the Project X party in Haren, because those tweets may have been generated in part by bots. Communication science in particular uses Twitter to analyze sentiments online. The scientists want to investigate the extent to which social bots can influence public opinion in follow-up research.