A huge percentage of election-related talk on Twitter isn’t coming from humans. It’s coming from bots.
And the majority of those bots are gunning for Trump to win the race, according to a paper released yesterday from Oxford University’s Project on Computational Propaganda.
During the third presidential debate, Twitter bots sharing pro-Trump-related content outnumbered pro-Clinton bots by 7 to 1. And in the span between the first and second debates, more than a third of pro-Trump tweets were generated by bots, compared with a fifth for pro-Clinton tweets.
A Twitter bot is an account that’s automated by software to act on its own. Bots can retweet, like and reply to tweets. They can also follow accounts and tweet themselves.
Twitter accounts that have extremely high levels of automation, meaning they tweeted over 200 times during the data collection period (Oct. 19-22) with a debate-related hashtag or candidate mention, accounted for nearly 25 percent of Twitter traffic surrounding the final debate, the researchers found.
The problem with the outpouring of automated engagement on Twitter — as any social media manager will tell you — is that campaigns often measure success (and decide where and how to invest in further outreach) by counting these retweets, likes, replies and mentions.
And it’s awfully hard to tell how many retweets and likes are from real supporters. A proliferation of bots can give candidates and issues unwarranted clout. Our own Walt Mossberg thinks Twitter’s bot problem is one of the company’s biggest issues.
Throughout the race, Trump has discounted the value of polls. They’re rigged, he says. Instead, his campaign implores Americans to reference how viral he is on social media and the size of his rallies.
The third debate came on the heels of the leaked tape of Trump bragging about sexually assaulting women, which went viral. And Trump’s uptick in automated Twitter fandom during the debate may have been intended to counteract the lingering outrage against the candidate on social media.
Increasingly, journalists use Twitter to report stories and demonstrate public interest by embedding tweets in articles, reading them on the radio or on TV. It’s a fantastic way to bring audience voices into a political discussion, though more voices doesn’t always make for a better conversation. Keeping in mind that much of the engagement numbers aren’t from real people is also a sobering reminder that virality is no demonstration of genuineness.
Donald Trump likes to boast that he’s more popular than Hillary Clinton on social media. He does, after all, have 12.9 million Twitter followers, while Clinton lags behind with a mere 10.1 million.
But it’s hard to say how much those numbers mean if many of them represent robots.
This article originally appeared on Recode.net.