The bots exaggerated support for the Republican, it suggests, but Trump would still have won a better range of supportive tweets whether or not that they had not.
The authors warn such software has the capacity to “manipulate public opinion” and “muddy political problems”.
The report has nevertheless to be peer-reviewed.
And one critic noted that it absolutely was not possible to be completely certain which accounts were real and which were “net robots”.
The investigation was led by Prof Philip Howard, from the University of Oxford, and is part of a wider project exploring “computational propaganda”.
It coated tweets posted on 26 September, the day of the controversy, and the 3 days afterwards, and relied on widespread hashtags linked to the event.
Examples of pro-Trump hashtags tracked by the study:
Examples of pro-Clinton hashtags tracked by the study:
First, the researchers identified accounts that exclusively posted messages containing hashtags associated with one candidate but not the opposite.
These accounted for concerning one.eight million professional-Trump tweets and 613,000 pro-Clinton posts.
The researchers then analysed which of these had been posted by bots. They identified an account as such if it had tweeted a minimum of fifty times daily across the amount, that means no less than two hundred tweets over the four days.
The results urged that seventeen of such professional-Trump tweets had been posted by bots and twenty two.three% of such pro-Clinton ones.
In total, that represented a total of 576,178 tweets benefiting the Republican nominee and 136,639 in support of the Democratic one.
“On the balance of chances, if you pulled out a heavily automated account the chances are four to at least one that you may realize it is a bot tweeting in favour of Trump,” said Prof Howard.
There is no suggestion, however, that bots were generated by either of the official Presidential campaign groups.
“We have a tendency to don’t seem to be trying at the supply, who is operating on the bots or to what end, just the metrics of the data,” said Prof Howard.
Wanting wider – to accounts that tweeted neutral hashtags or a combine of various sorts – the study urged that 23% of all the tweets were driven by bots.
One machine learning expert cautions that the standards used to identify the bots would possibly have been too imprecise to have sifted out all the human-based activity.
“Real people will write a script and use an algorithm to tweet frequently with specific responses, or humans will tweet content that looks almost similar to a series of bots flooding a political hashtag”, comments Caroline Sanders, an ex-IBM researcher who currently works for Buzzfeed.
“Conjointly, political commentators or folks eagerly engaged in the political discussion may conjointly tweet this many times.”
So, is it potential that Trump supporters may merely are additional enthusiastic than Clinton’s and have done a better job at leveraging social media to their advantage?
Prof Howard said that it’s unlikely to be the sole explanation.
“Most of the significant automation and tweets happened overnight and shared similar hashtags and data,” he explains.
“They show behavior that’s not human and usually don’t have comments [about alternative problems but] the particular topic in question.”
He adds that the fifty-tweets-a-day rule was borne out by analysis of posts made during a past Venezuelan election and the Brexit vote.
In both cases, his team double-checked a sample of accounts that had been flagged as bots and confirmed they displayed different characteristics of being inhuman.
“From our knowledge most real Twitter users do not get on my feet to 50 times daily,” he said. “Therefore, on balance, that benchmark has worked well.”
Bots tackle numerous guises but have some give-away signs.
They usually do not feature a profile image, and when they do it’s often shared among multiple accounts – thus watch out for duplicates.
Bots also tend to follow many additional accounts than than they are followed by in turn – an indication that they are doing not have real friends or work colleagues.
They typically have little to say aside from the subject of conversation they need been created to post concerning, and may tweet prolifically while not apparent recourse to sleep.
Conjointly be careful for accounts that reply to your messages in less time than was humanly potential to read what you wrote.
A final giveaway is that if scrutiny of the bot’s account reveals it has sent the same response to you to dozens of others too.
In addition to being additional numerous, Prof Howard additionally points to the professional-Trump tweets being additional effective, whether or not generated by bots or not.
They were a lot of probably to feature multiple hashtags and links to relevant internet addresses to stock up the obtainable one hundred forty-characters, he explains, that in turn helps keep tags alive and bolsters Trump’s message.
“Someone eager to follow [one amongst the hashtags] will see a heap additional content and more cross posting,” said the professor.
It’s unclear what result the bots actually have on voter behavior, but Prof Howard believes that negative messages are more probably to own a bearing than positive ones.
“Within the 2008 and 2012 US elections, bots were used to make politicians look more standard with voters,” he said.: “These days it’s about engaging along with your support base and constantly feeding them info, and bound hashtags that can keep their level of interest high.”
One silver lining from the study is that humans are still the dominant force on Twitter and for the foremost half they get out posts from other folks.
“It’s affirming that social media platforms can still present spaces where some individuals can have political conversations with their networks to genuinely discuss their views, ” concludes Prof Howard.
“However when nearly 1 / 4 of Twitter activity turns out to be automated it will compound the read that politicians are out to govern public opinion.”