Senate debates

Thursday, 15 November 2018


Social Media

7:14 pm

Photo of Concetta Fierravanti-WellsConcetta Fierravanti-Wells (NSW, Liberal Party) Share this | | Hansard source

Tonight I again wish to examine how political influence campaigns are being run using multiple Twitter accounts. I recently informed the Senate about the activities of Sleeping Giants Oz, an anonymous, politically motivated Twitter campaign, imported from the US, whose heavy reliance on unverifiable Twitter accounts makes its actual size deceptive. The Joint Standing Committee on Electoral Matters is currently looking at cybermanipulation of elections, including considering the extent to which social media bots may have targeted Australian voters and political discourse in the past; the likely sources of social media manipulation within Australia and internationally; and ways to address the spread of deliberately false news online during elections.

A submission to JSCEM from Digital Industry Group Inc, which includes representatives from Facebook, Twitter and Google, concludes:

Fortunately, the experience of DIGI members and the use of their platforms in Australia, to date there is no evidence to suggest that election manipulation has been a widespread problem in Australia as it has been in the U.S.

Similarly a submission from Twitter says:

During the 2016 election, we were not made aware of any activity related to the suppression or interference with the exercise of voting rights in Australia.

These reassurances seem at odds with a recent report in The Australian that Twitter accounts linked to the Internet Research Agency, the infamous Russian troll factory, have spread politically charged posts about Australian politics, including the 2016 federal election and last year's same-sex marriage survey.

However, tonight I wish to outline to the Senate how the Australian Labor Party is benefiting from another influence campaign also being conducted via the Twitter sphere. This campaign employs a calculated and malicious strategy of spreading misinformation and political spam via a large web of mainly anonymous but also automated Twitter accounts. These accounts post similar-to-identical pro-Labor, pro-union, anti-coalition content. They primarily engage by retweeting posts from like-minded accounts, creating an echo chamber of reinforcing noise. Twitter is full of anonymous accounts that often exist only to push partisan and frequently toxic debate by interests groups, including fake news. Twitter permits automated retweets and it is easy to make a Twitter bot that will automatically 'favourite' and/or retweet tweets that contain particular words or hashtags.

Many of the accounts to which I refer have tweeted or retweeted hundreds of thousands of times and continue to do so hundreds of times a day, cranking out pro-Labor, anti-coalition messaging on an industrial scale. Often they admit a union connection or Labor viewpoint, together with an eclectic mix of other interests which collectively cover the entire gamut of left-wing concerns. Some accounts run lies and smears against the coalition or needle coalition candidates and parliamentarians while promoting Labor initiatives or running interference for Labor. They are frequently a vehicle for unfounded and defamatory allegations, low-grade research or catalogues of alleged coalition misdeeds which wouldn't be publishable by or rate any interest from the mainstream media. Some recycle media stories which boost Labor or are unflattering to the coalition. For example, @virgotweet, which mainly retweets 80 times a day, recycles old news about alleged coalition scandals and presents it as if new. They typically follow or are followed by a mix of Labor figures and also engage with Twitter feeds of other left-wing organisations. The aim is to discredit the coalition, to promote allies and to distort public opinion by massively amplifying messages which feed into like-minded networks and engage both anonymous and real Twitter users.

These accounts often show signs of direct user engagement via unique tweets and topical comments, which is indicative of their close maintenance and operation. A key account called @Talaolp tweets rather than retweets an unremitting torrent of Labor propaganda. It claims it is:

… sharing information about the Liberal Governments, State and Federal, their deception, lies and misinformation to the Australian Public.

Based in Western Australia, @Talaolp has tweeted 230,000 times in the last five years. That's about 125 times a day. Some of its anti-coalition material is scurrilous and intended simply to smear. It typically posts to three other accounts: 'Sir Clyde of Nob' @Nobby15, 'Big Al' @banas51 and 'Mari R' @randlight.

Sir Clyde of Nob, supposedly a retired IT specialist also based in Western Australia, has tweeted 790,000 times over the last nine years, an average of 240 times a day. It mostly retweets, but every seventh to eighth engagement, on average, is a personal tweet or comment on a post, showing frequent personal intervention. It retweets TALAOLP extensively and boasts about its Twitter reach, in a recent week receiving over 1,600 mentions, 1,500 likes, almost 400 retweets and 230 replies. Big Al, who describes himself as a 'lefty' and a 'hard worker', has retweeted over 200,000 times in the last four years, an average of 135 a day, namely retweeting a broad fare of left-wing commentary. Mari R, who says she wants Bill for PM, has retweeted almost 450,000 times over the last seven years, an average of 175 times a day.

Another such account is MSM Watchdog, supposedly dedicated to 'Exposing unconscionable attacks on the poor'. This account has tweeted 447,000 times over the last five years, an average of 240 a day, predominantly retweets of predictable anti-coalition and pro-Labor material. But MSM Watchdog was stung into life by my recent speech on Sleeping Giants Oz, claiming that the Liberal Party hates social media because 'they are hopeless at it.' If being good at it means flooding the twittersphere with propaganda up to 100,000 times a year, I'll take that as a compliment. MSM Watchdog retweets far more frequently than Sir Clyde of Nob. Some days it only retweets hundreds of times; other days there are also some personal tweets and comments. Both accounts appear to be operated closely by individual users but are almost totally reliant on retweets as a method of amplification. Who has either the inclination or the resources to, in the main, retweet 240 times a day, year upon year? I suggest that the description of many accounts as being operated by unionists offers a clue.

Another account, 'Old and Cranky', which describes its owner as a 'true believer still looking for the light on the hill'—good luck!—has tweeted 329,000 times in the last four years, an average of 225 a day, of antigovernment messaging. Its last 3,200 engagements are all retweets. Similarly, 'Gold Coast Nurse', which describes its owner as a proud union delegate and member, tweeted 88 times a day in the last five years and has also not tweeted an original thought in its last 3,200 tweets.

What I have described tonight is the Twitter equivalent of a Labor union telephone tree, a Twitter tree, though perhaps a better analogy would be a jungle, and the law of the jungle applies when it comes to its content. An organised union operation backed by Labor volunteers is the most likely source of this influence campaign, but the anonymity of accounts means we can't be sure exactly who they are. These accounts were active during the 2016 election. They're in full swing and, unless checked, will be active during the next election. Twitter's submission to JSCEM claims:

We focus on developing machine learning tools that identify and act on networks of spammy or automated accounts automatically by tracking account behaviour. This lets us tackle attempts to manipulate conversations on Twitter at scale, across languages, and different time zones.

I submit that Twitter is on a steep learning curve and still has a long way to go, and I would suggest it review the activities of the accounts to which I have referred as well as many other high-volume accounts like 'Wowbagger' and 'Fair Dinkum Troublemaker'.

As we approach the next election, we need to be aware that political interest groups as well as potential state actors are trying to amplify their messaging and distort debate, including by disseminating fake news using social media platforms. In relation to state actors, I again note that a US intelligence report assessed that:

Moscow will apply lessons learned from its Putin-ordered campaign aimed at the US presidential election to future influence efforts worldwide, including against US allies and their election processes.

Clearly, there is much to be on guard about as we approach the next election. I will be forwarding this speech to JSCEM as I believe it adds qualitative material.