Anti-Voice accounts on Twitter ‘abusive with impunity’ by using false identities

The most active Twitter accounts opposing the Indigenous Voice to parliament are people using false identities who are stoking an increasingly divisive online campaign, rather than computerised “bots”, an analysis reveals.

Racial abuse is occurring on both sides but there is overwhelmingly more hate speech in tweets supporting the No side, which have twice the amount of offensive language, a Queensland University of Technology analysis found. Accounts supporting the No side are considerably more active, having sent a total of 34,395 tweets (61 per cent) compared to 20,244 tweets on the Yes side (36 per cent).

The debate over the Voice has led to a proliferation of new Twitter accounts.Credit: Sydney Morning Herald/The Age

Over the past week, a number of high-profile Australians including ABC presenter Virginia Trioli and ACTU boss Sally McManus have raised the alarm about a significant uptick in racist abuse and peddling of conspiracy theories about the Voice on Twitter.

There is growing concern among some Twitter users that the debate is becoming over run by computerised “bots”, which appear to be legitimate accounts but are controlled by software and are programmed to spread disinformation.

“It is more likely that problematic and/or suspicious-behaving accounts participating in the #VoiceToParliament debate on Twitter are human-controlled,” said Queensland University of Technology senior lecturer Dr Timothy Graham.

“We found evidence that the No campaign has a much higher concentration of recently created pseudonymous accounts that do not appear to have an identifiable ‘real’ person behind them.

“However, we found little evidence that either side are deploying bots and no evidence that there is any difference in the number of [bot-like] accounts on both sides.”

Graham conducted an analysis of 56,742 tweets between May 1 and May 25 sent by 11,592 unique Twitter accounts.

It found the top 10 accounts promoting the No case – which sent 2609 tweets over this period – appear to be pseudonymous profiles which don’t contain any identifying information, real profile pictures or original photos.

Four out of the top 10 Yes accounts are pseudonymous, while the other six are authentic.

The analysis used a detection model known as Botometer and found little evidence that either side were using bots to boost their campaign efforts.

There are likely about six bots tweeting using the #VoiceToParliament hashtag, with three on each side, the analysis showed.

‘Something different’ starting to happen on Twitter

At some point between October and December, Australians will vote on the referendum to enshrine in the constitution a Voice to parliament, which will make representations to the parliament and the government on issues that impact Aboriginal and Torres Strait Islander people.

Australia’s chief union boss Sally McManus last week said “something different” was starting to happen on Twitter, with hundreds of accounts she had never seen before spreading racist content about the referendum.

“This was different to the ‘normal’ trolling,” the secretary of the Australian Council of Trade Unions said. “It is clearly organised, on a large scale, racist and designed to sow division.”

The No side on Twitter is now more active, sending a total of 34,395 tweets (61 per cent), compared to 20,244 tweets for the Yes side (36 per cent), according to Graham’s analysis.

Out of the top 20 most active accounts, only three are Yes supporters.

There has also been a big influx of new accounts on the No side since Prime Minister Anthony Albanese announced the referendum last year, with 30 per cent of accounts against the Voice created since 2022 compared to 9 per cent on the Yes side.

The analysis used a program called HateSonar to find that 327 tweets on the No side amounted to hate speech while 63 tweets supporting the Voice also qualified. The No side also had twice the amount of offensive language in its tweets.

Leading No campaigners including Warren Mundine and Senator Jacinta Nampijinpa Price have been subject to racial abuse on Twitter, including with the racial slur of “coconut” .

There are concerns that the toxic debate online could lead to violence.

Warren Mundine and Senator Jacinta Nampijinpa Price.Credit: Alex Ellinghausen

The nation’s top spy, Mike Burgess, last week said his agency had recently undertaken security assessments on the referendum, which he labelled a “significant event” which may incite “spontaneous violence”.

The director-general of ASIO said there was no intelligence that other countries were planning to interfere in the referendum, but warned the agency was “on the lookout” for such a risk.

‘Playbook that comes straight from the pandemic’

Like McManus, ABC radio host Virginia Trioli said her Twitter account had recently been flooded with the “most awful racist shit and inflammatory Voice disinformation” since posting in support of her indigenous colleague Stan Grant.

“Very few of these accounts actually follow me, so what evil Elon algo draws this stuff to my account?” Trioli said last week, referring to Twitter’s owner Elon Musk’s changes to the social media platform.

Labor MPs privately said they were experiencing something similar lately on both Twitter and Facebook, with their posts being deluged by No Voice messaging which included conspiracy theories and racist abuse.

Albert Zhang, an analyst with the Australian Strategic Policy Institute’s International Cyber Policy Centre, said his think tank hadn’t yet observed any mass-coordinated inauthentic accounts such as “bots”, but warned that would probably change in the lead-up to the referendum.

“Based on the language used by accounts in the tweets targeting public figures online, it appears that most of the accounts are probably from Australia and a small amount are probably from the US,” Zhang said.

“This reflects some existing racist views in Australia and the additional anonymity on social media allows users to be more abusive with impunity.”

Instead of bots, Zhang said he suspected the online abuse and trolling was being influenced by prominent Twitter users such as Josephine Cashman, who has stated that the Voice is a United Nations and globalist conspiracy.

Cashman, an Aboriginal lawyer and former One Nation candidate who has more than 15,000 followers on Twitter, posted nine replies to Trioli’s two tweets last week which included erroneous accusations against the ABC journalist.

Zhang said the proliferation of conspiracy theories was being worsened by the fact that anyone on Twitter can now pay to amplify their posts after Musk removed blue checks for verified accounts.

“This is partly fuelling distrust in public figures and is creating an environment where misinformation and hate speech thrive,” he said.

Ed Coper, director of communications agency Populares, said he believed there was a coordinated campaign on Twitter to manipulate the narrative even if it did not involve bots.

The digital communications expert said this was being done by copy and pasting the same comments repeatedly so that the algorithm starts boosting the message.

“There’s very similar messages all over Twitter,” he said. “That is a playbook that comes straight from the pandemic, and prior to that around Donald Trump’s election in 2016.

“Either they are genuine accounts, or they might be what we call ‘ sock puppet accounts’ – one person might be operating a handful of different accounts just to give their voice a bit more spread further and wider, and it’s very effective.”

Coper said platforms such as Twitter need to do a much better job of ensuring every account is genuine, and that users aren’t deliberately manipulating the algorithm to spread conspiracy theories.

“The media can play a role as well, in terms of reporting what is the genuine mainstream consensus versus what is getting a lot of noise and activity – there’s always an imbalance there.”

Cut through the noise of federal politics with news, views and expert analysis from Jacqueline Maley. Subscribers can sign up to our weekly Inside Politics newsletter here.