Invisible Election: Social media bots proving low-key in 2017

Automated social media accounts are proving more subtle during the 2017 General Election compared to other campaigns.

"Bots" are Twitter accounts set up to tweet automatically according to algorithms, while "botnets" are large, linked networks of bots which tweet similar messages simultaneously.

Research between Sky News and the Centre for Analysis of Social Media (CASM) uncovered previously unknown botnets tweeting during the 2017 General Election.

But the work also found very few Twitter accounts had tweeted at "super-human speeds".

High-volume tweeters - whether human or bot - also tended to espouse more left-wing messages, according to the research.

Image:Bots have had a low-key impact on the 2017 vote

Josh Smith, a researcher at CASM, told Sky News: "The botnets we picked up were tweeting the same thing at the same time - that's very stupid.

"Bots which tweet hundreds of thousands of times a day are very easy to pick up. They can be very effective at shutting down debate… but we have not seen that kind of activity in this election.

"What we're interested in is botnets that have the ability to be extremely clever in the way they behave.

"If you're trying to be a real person… you need to employ quite sophisticated tactics to convince you are indeed just concerned about [Labour leader] Jeremy Corbyn or politics in Britain."

Pinpointing who controls bots is tricky.

Philip Howard, professor of internet studies at the Oxford Internet Institute, said: "It's tough to say where the bots come from.

"We know, depending on the country, between 10 to 20% of all the social media traffic over Twitter is driven by these highly automated accounts.

"We know some of them originate in Russia, we know others originate in the US, and there's plenty of home-grown bots here in the UK."

Image:Coordinated human accounts and small-scale bots boosted Marine Le Pen

The low-key impact of bots during the 2017 General Election contrasts with a less subtle, more sledge-hammer approach seen in other countries.

In 2014, the Mexican government used around 75,000 automatic accounts to tweet spam material and drown hashtags used to organise protests with a huge volume of useless posts.

And when Russian opposition politician Boris Nemtsov was assassinated in 2015, fake stories began circulating that his killers were Ukrainian.

Those messages were supercharged by 2,900 bots all tweeting the same messages.

Mr Howard said: "We call it computational propaganda.

"It's the combination for a computer programme which gets a message out to thousands and thousands of users with some junk news or some fake facts that have been made up that try to poison conversations."

But less obvious bots may be more effective. During the recent French presidential elections, coordinated human accounts - combined with small-scale bots - helped boost the National Front's social media impact.

Mr Smith told Sky News: "One of our worries is what is being called automated activity is quite often just people who are excited about a cause.

"And this has especially been true for [Donald] Trump and [Marine] Le Pen - it's the side of people who think they have something to gain in an election, on the cusp of historic victory."