The Twitter That Was: Reflections on Ten Years of #auspol

Paper presented at the Social Media & Society conference, London, 18 July 2024.

1. The Twitter That Was: Reflections on Ten Years of #auspol Axel Bruns and Anand Badola Digital Media Research Centre Queensland University of Technology Brisbane, Australia [email protected] @snurb_dot_info | @[email protected] | @snurb.bsky.social

2. Image: Midjourney (prompt: The Twitter That Was: Reflections on Ten Years of #auspol )

3. http://votecompass.com/2013/07/25/are-you-among-australias-most-influential-political-tweeters-votecompass-maps-the-auspol-twittersphere/

4. Introducing #auspol Hashtag. Community? Amongst the biggest and most persistent hashtags in Australia December 2013 to July 2023: 57m tweets, 1.25m unique accounts Strong core of highly active accounts: 44.6m tweets (78%) by top 1% most active accounts  Indicative of genuine community structures? Dataset: Continuous data collection (TCAT, TweetQuery , Academic API 🪦) Focus on top 1% of accounts by tweet count (12,452 unique accounts) (https://www.smh.com.au/technology/tony-abbott-dismisses-social-media-as-electronic-graffiti-again-20150126-12yg26.html)

5. Patterns over Time

6. (Images: https://www.abc.net.au/news/2024-01-29/tony-abbott-eats-an-onion/103394716 ; https://www.gq.com.au/success/career/our-exclusive-interview-with-malcolm-turnbull/news-story/abdba2bbf5fc5e182ec720b5938ef8d2 ; https://independentaustralia.net/politics/politics-display/scott-morrisons-not-so-good-year-of-2020,14632 ; https://www.theguardian.com/australia-news/article/2024/jul/11/anthony-albanese-appeals-to-western-sydney-amid-muslim-voting-campaigns-on-gaza-war ) Morrison’s appointment as PM to start of 2019 election campaign: Organised agitation against Morrison, later moving from #auspol to #ausvotes? federal election 2016 federal election 2019 federal election 2022 enXittification 🤬 Voice referendum campaign?

7. Community?

8. Finding Community What makes participants central to a community? Engagement: addressing and responding to other participants through @mentions Balance between number of sent @mention and received @mention tweets Balance between number of accounts @mentioned and accounts @mentioning Consistency: maintaining such balance even while posting a lot of @mentions Operationalisation: Engagement – Sent-Received Indices: Tweets: (# sent @mentions - # received @mentions) / (# sent @mentions + # received @mentions) Accounts: (# @mentioned accounts - # @mentioning accounts) / (# @mentioned accounts + # @mentioning accounts) +1 (@mentioning without any response) 0 (balanced) -1 (not responding to @mentions) Posting volume: Consistency – Balanced Mention Tweets: # total @mentions × (1 - | Sent-Received Index for Tweets |)

9. Finding Community #auspol patterns: Sent-Received Indices for tweets and accounts largely move together Key groups emerge: Institutional accounts: many @mentions received, but very little reciprocal engagement Peripheral participants: many attempts to engage, but few responses received In-group cliques: very actively @mentioning, but selective engagement with a subset of accounts Community core: broadly balanced engagement (slightly more sent than received) Lots of @mentions sent, few responses received Lots of @mentions received, few responses sent Many outgoing @mentions, but responding to few accounts Many accounts @mentioned, but engaging with few responses News, journalists, politicians Peripheral participants In-group clique Rare – spammers? Community core

10. Finding Community #auspol patterns: Sent-Received Indices for tweets and accounts largely move together Key groups emerge: Institutional accounts: many @mentions received, but very little reciprocal engagement Peripheral participants: many attempts to engage, but few responses received In-group cliques: very actively @mentioning, but selective engagement with a subset of accounts Community core: broadly balanced engagement (slightly more sent than received) Evolution over time (time step: quarter years)

11. Finding Community #auspol patterns: Sent-Received Indices for tweets and accounts largely move together Key groups emerge: Institutional accounts: many @mentions received, but very little reciprocal engagement Peripheral participants: many attempts to engage, but few responses received In-group cliques: very actively @mentioning, but selective engagement with a subset of accounts Community core: broadly balanced engagement (slightly more sent than received) Trajectories of individual accounts (20 highly active accounts in community core)

12. Practices

13. A new approach to social media data: Quantifying specific aspects of individual participant activities, then identifying and interpreting similar patterns at a group level. In practice: vector embedding of activities at account level, then pairwise comparison of activity vectors across accounts. * With particular thanks to Kateryna Kasianenko. Beyond Network Analysis: Practice Mapping* Image: Midjourney Bruns, A., Kasianenko, K., Padinjaredath Suresh, V., Dehghan, E., & Vodden, L. (2024). Untangling the Furball: A Practice Mapping Approach to the Analysis of Multimodal Interactions in Social Networks. arXiv : 2407.05956. http://arxiv.org/abs/2407.05956

14. Account-to-account interactions (relative to interactive affordances available on any given social media platform) Account’s post content (topics, sentiment, hashtags, named entities, etc.) Account’s use of sources (URLs, domains, embedded videos and images, etc.) Account’s profile information (name, description, etc.) Manually and computationally coded information about the account and its posts … Potential Patterns to Operationalise in Practice Mapping Image: Midjourney Bruns, A., Kasianenko, K., Padinjaredath Suresh, V., Dehghan, E., & Vodden, L. (2024). Untangling the Furball: A Practice Mapping Approach to the Analysis of Multimodal Interactions in Social Networks. arXiv : 2407.05956. http://arxiv.org/abs/2407.05956

15. Nodes: accounts in the top 1% of #auspol posters Node size: strength of similarity with other accounts Node colour: Louvain modularity cluster detection Edge weights: interaction similarity (retweets + @mentions) + hashtag similarity (minimum 1.33) Top @mentions: TurnbullMalcolm ScottMorrisonMP TonyAbbottMHR billshortenmp Top @mentions: ScottMorrisonMP TurnbullMalcolm Top @mentions: ScottMorrisonMP AlboMP

16. An Evolving Community Gradual evolution: Shifts in focus as Prime Ministers come and go But substantial change as Scott Morrison becomes PM Significant influx of new accounts from late 2018 onwards (until Morrison loses the 2022 election) Not predominantly COVID-related: biggest sign-up during Dec. 2019 / Jan. 2020 bushfire crisis widespread global adoption PM Morrison election election bushfires

17. Influx & Renewal Influx of new accounts: More @mentioning than @mentioned at first Gradually developing more balanced engagement patterns Renewal of community: Replacing / replenishing accounts in community core Q2/2014 Q2/2017 Q2/2020 Q2/2023

18. Preliminary Observations

19. Observations and Outlook Community evolution: Strongly committed and very active core of participants Genuine community – highly interconnected and interactive Gradual churn in community core Especially pronounced as PM Morrison comes to power, abruptly ending after 2022 election – indicative of orchestrated opposition campaign? Or genuine reaction to Morrison’s style of politics? Next steps: Further analysis of evolution in themes and topics over the years Practice mapping analysis for selected periods during the ten-year timespan Changes in toxicity and other indicators of communicative dysfunction Information sourcing and sharing practices (reliable URL data available from Aug. 2016)

20. Thank you! Image: Midjourney (prompt: #auspol)

21. This research is supported by the Australian Research Council through the Australian Laureate Fellowship project Determining the Dynamics of Partisanship and Polarisation in Online Public Debate and the Australian Future Fellowship Understanding Intermedia Information Flows in the Australian Online Public Sphere . Acknowledgments