Since 2016, there’s been a big focus on the influence of bots (i.e. computer controlled accounts) on social media. Experts who’ve studied them say they’re influencing Australia’s elections by promoting topics and hashtags to drive debate. And recently, some have alleged that bots might be behind a anti-Dan Andrews hashtag — but it appears the evidence doesn’t support that claim.
Over the weekend, one of Australia’s leading experts on social media bots and misinformation ran an experiment on the #DanLiedPeopleDied hashtag. What he found suggests that most accounts using the hashtag seem legit.
QUT senior lecturer Tim Graham used a ‘state-of-the-art bot detection’ model to examine more than 5,200 tweets sent by more than 2,000 users using the hashtag.
With all the #COVID19Vic buzz today, I see many folks calling one another "bots".
So I ran a formal bot analysis of a currently trending hashtag "Dan|Lied|People|Died", and here's what I found.
Spoiler: not bots*
A brief thread:
— Tim Graham “Lil Pharma” (@timothyjgraham) September 26, 2020
According to his analysis, only two accounts had a 50 per cent chance of being a bot.
And in analysis provided to Gizmodo Australia, Graham found similar results for other anti-Dan Andrews and anti-Scott Morrison hashtags:
- #SackDanAndrews: 2374 tweets collected, 3 accounts identified as bots
- #LiarFromTheShire: 2478 tweets, 9 bot accounts
- #ScottyTheSimp: 1852 tweets, 2 bot accounts
Graham cautions that his analysis is not peer-reviewed, but said evidence points to bots having only a minor presence in this online debate about Australian politics.
“More work is needed to fully understand what’s really going on in our polarised information space, but to say it’s all about ‘bots’ is misleading,” he said.
Why does it matter who’s behind anti-Dan Andrews and anti-Scott Morrison hashtags?
Graham said he’s seen an increase in people accusing accounts of being bots. He worries that some of the focus on bots means other issues aren’t given enough attention.
“Bots are an issue but I think trolling is a much larger concern, particularly when there’s a large portion of anonymous accounts, often recently created, piling onto a polarised issue or hashtag,” he said.
He takes issue with people accusing accounts of being bots to discredit or delegitimise other people who they disagree with.
“In this way, it’s unhelpful and may only further undermine and erode democratic trust and civil political deliberation,” Graham said.
What Graham’s analysis shows is that it’s likely some people are jumping to conclusions about why other accounts are promoting arguments and ideas they don’t agree with. Graham hopes that by quantifying the small scale of bots, people will focus on what’s really happening.
“Basically, we shouldn’t rush to label people we don’t agree with as bots,” he said.