N&MRC 2025 Seminar Series | Dr Timothy Graham (QUT) Beyond traditional bot detection: Using sequential coordination analysis to identify automation practices on social media
n 2025, the N&MRC is hosting a seminar series focusing on misinformation and resilience. Leading Australian researchers from the communication, education, social, and psychological sciences will explore the impact of false information on society, the role humans, media and technology play in spreading it, and strategies for identifying and combatting it. The series convenor is Prof. Mathieu O’Neil. Our seminar - Beyond traditional bot detection: Using sequential coordination analysis to identify automation practices on social media will be presented by Dr Timothy Graham from QUT.Detecting automated ‘bot’ accounts in online political discussions remains a challenging research problem. Classical supervised machine learning approaches suffer from high false positive rates as humans exhibit bot-like behaviours while sophisticated bots mimic human patterns. The advent of ChatGPT and scheduling tools has further blurred human-machine boundaries, making traditional feature- and content-based detection unreliable. This has prompted researchers to explore coordinated network analysis, which identifies clusters of accounts performing similar actions within short time windows (e.g., 60 seconds). However these methods capture both authentic grassroots organising and inauthentic manipulation. Reporting on research conducted with Rio (Guangnan) Zhu, this paper introduces sequential coordination analysis: we propose “syntenic coordination” – drawing from genetics where synteny describes conserved sequential arrangements – to identify accounts consistently posting in identical or highly similar order. This targets automation’s operational signatures: software looping through account lists, coordinated manual operations across devices, systematic execution of predetermined sequences, etc. Our method identifies rigid sequential patterns characteristic of automated processes but atypical of organic human behaviour by capturing the longest common substrings among users’ action sequences. Testing on Voice of Parliament referendum data collected from X, we surface a network of 17 accounts exhibiting near-perfect sequential coordination, 11 of which are not captured through time window-based methods. These results contradict concerns about bot prevalence on X during the referendum, but raise other opportunities for research into other automation practices. By focusing on sequence rather than temporal synchronisation, this approach reveals operational fingerprints of automated systems in the aggregate, providing a more reliable foundation for detecting, analysing, and deliberating over what constitutes normatively problematic behaviour in online political communication.Presented by: Dr Timothy Graham, who is an Associate Professor of computational communication at the Queensland University of Queensland (QUT). His research expertise is in online networks and platforms, with a particular interest in propaganda and social influence, knowledge production, and algorithmic curation. Tim previously held an Australian Research Council DECRA Fellow, for his project, “Combatting Coordinated Inauthentic Behaviour on Social Media” (2022-2025). He is currently Chief Investigator of the QUT Digital Media Research Centre and Associate Investigator of the ARC Centre of Excellence in Automated Decision-Making and Society (ADM+S). He is CI on a new ARC Discovery Project, “Understanding and Combatting ‘Dark Political Communication’” (2024-2027). He has authored over forty peer-reviewed journal articles and book chapters, featured in thousands of news articles in leading outlets including The New York Times, The Guardian, ABC News, The Washington Post, and BBC World News, and has developed and maintains open-source software for the collection and analysis of data from the web and social media. Future seminars:Wednesday 8 October: Dr Tanya Notley, WSUWednesday 12 November: Dr Momoko Fujita, UC
