Presidential hopeful Donald Trump’s Twitter support was over-exaggerated due to significant bot activity on the site four times as great as that of his rival Hillary Clinton, according to new research.
Oxford University’s Philip Howard made the claims in a new report analysing tweets made on the night of the first presidential debate and the succeeding three days.
Focusing on accounts which only posted messages with hashtags in favor of one candidate, he then went on to ascertain if any of these had tweeted more than 50 times each day across the period in question – indicating they were bots.
A third (33%) of pro-Trump tweets were deduced to have been posted by bots, versus 22% of tweets in favor of Clinton. That amounts to over half a million (576,178) for Trump and a little over 136,000 for the Democrat.
The study also found that around a quarter (23%) of accounts tweeting either neutral hashtags or ones which switched sides were powered by bots.
Imperva vice president, Tim Matthews, warned that web admins and pollsters must take more care when appraising raw traffic numbers.
“In 2015, humans were responsible for the majority (51.5%) of web traffic, good bot traffic decreased to 19.5% from previous years, but ‘bad bot’ traffic accounted for practically a third, 30% of all online traffic,” he explained.
“Bad bots are the malicious intruders that swarm the internet and, in the case of the Trump tweets, generate automated, non-human social media posts.”
Trumps is said to have more support on Twitter even without the bots. However, the heavy use of automated social media posts promoting the candidate could be seen by some as an attempt in some quarters to manipulate the voting public.
It comes amid accusations that potentially state-sponsored Russian actors have been trying to interfere with the electoral system in order to undermine the result of next month’s presidential election.