Numbers have become incredibly important in the world of social media; it is seen as authority and popularity, but can one buy popularity and authority? According to Professor Marco Camisani Calzolari, CEO of DigitalEvaluations, many brands seem to be willing to pay for bots, Twitter followers and Facebook likes just to be seen as popular.
DigitalEvaluations have developed an analytics software with Professor Calzolari, the brain behind the algorithm, to distinguish human and bots on Twitter and Facebook. Professor Calzolari is confident about the accuracy of the software; hence he carried a study on high-profile individuals' Twitter accounts. He says:
"While it is impossible to distinguish a fake user from a real one with utter certainty, the algorithm that I have developed is able to identify 'human' and 'bot' (or inactive) behaviours on existing profiles with reasonable accuracy. Based on this method and with direct access to a Twitter account, its fake or inactive followers can be removed."
According to the study, up to 29.9 percent of President Barack Obama's 17.82 million Twitter followers and 21.9 percent of Mitt Romney's 814,000 followers are likely to be inactive.
Professor Calzolari said "the bot profiles -- in other words, the profiles which are most likely generated by robots or inactive -- would be in absolute terms 5,326,883 for the current White House tenant and 178,237 for his opponent. The human users -- users almost certainly real -- would be 8,020,826 (or 45.01 percent of the total) for Barack Obama and 405,357 (or 49.75 percent) for Mitt Romney."
The researcher decided not to consider uncertain accounts, those profiles which cannot be confidently assigned as human or robots. The study has found 11.56 percent uncertain profiles for the U.S. president and 13.64 percent for Mitt Romney. Moreover, protected accounts following the U.S. president (13.54 percent) and Mitt Romney (14.73 percent) cannot be examined by the program as they do not share their activities publically; thus they are not assigned as human or bots and excluded from the overall analysis.