TWITTER TRIALS AND FACEBOOK JURIES: AN ANALYSIS OF THE AUSTRALIAN SUB JUDICE RULE AND THE REGULATION OF PREJUDICIAL PUBLICITY ON SOCIAL MEDIA DURING HIGH-PROFILE CRIMINAL TRIALS
Rachel Jane Hews LLB(Hons), BComm(Mgmt), GradDipEd(Senior Years), DipPolicing, Cert IV Training and Assessment
Principal Supervisor Associate Professor Nicolas Suzor Associate Supervisor Dr Kylie Pappalardo
Submitted in fulfilment of the requirements for the degree of Doctor of Philosophy (PhD)
School of Law, Faculty of Law Queensland University of Technology 2019 i Keywords
Keywords
Citizen journalists, coding, contempt, content analysis, digital age, digital methods, discourse analysis, disinformation, Facebook, freedom of expression, freedom of speech, freedom of the media, Gable Tostee, Gerard Baden-Clay, high-profile criminal trials, homicide, internet, journalism, juries, juror bias, juror impartiality, juror prejudice, jury trials, media law, media manipulation, media regulation, misinformation, murder, news, open justice, prejudicial publicity, qualitative analysis, quantitative analysis, random sampling, regulation, right to a fair trial, social media, social media analysis, social media platforms, sub judice contempt of court, sub judice rule, Tableau, thematic analysis, TrISMA, Twitter.
Twitter trials and Facebook juries ii
Abstract
High-profile criminal trials have long been a source of public spectacle in the media. Historically, both public interest and the right to free speech fuelled the publication of trial-related information. However, due to concerns the publication of potentially prejudicial information may affect an accused’s right to a fair trial, laws were established in 14th Century England to prevent, in part, the use of words to abuse a party to a suit. The modern evolution of this doctrine, the sub judice rule, aimed to prevent professional journalists and media outlets from publishing information that might prejudice a matter before the courts.
This thesis aims to provide insight into how the sub judice rule operates in practice since the introduction of the internet and social media. While the law was well established in terms of how it regulated the behaviour of publishers before the internet, there are concerns about the effectiveness of the rule in the digital age. By analysing data from Twitter and Facebook during two high-profile murder trials, I investigate the prevalence of prejudicial publicity on social media and examine how professional journalists and non-journalists talk about criminal trials. I also investigate how social media users show prejudice on social media during high-profile criminal trials. More generally, my analysis provides insight into the types of information empanelled jurors might see, should they be exposed to information about their trial on social media.
I found that large proportions of tweets and comments in my samples were potentially prejudicial, including 6% of tweets during the Baden-Clay trial, 17.4% of tweets during the Tostee trial and 62% of comments in my Tostee Facebook sample.
There are no comparable studies against which I can compare these results, but as the iii law is concerned with the publication of any prejudicial information, it is fair to conclude that these results are concerning. I found that professional journalists largely comply with the sub judice rule — they report the news and generally do so using professional, objective, impartial language. In contrast, non-journalists were far more likely to post prejudicial information — they discussed or responded to the news in a way that was more commonly emotive, opinionative and biased. In addition to the potential for individual tweets or comments to be prejudicial, I raise concerns about the cumulative effect of large volumes of tweets or comments working collectively in a way that may prejudice jurors. My samples contained a pro-prosecution bias and there was widespread negative sentiment that may be prejudicial. I also found that some posts lacked accuracy or fairness, and others may not have been made in good faith and without malice, both of which may have the potential to prejudice jurors.
I conclude that the Australian sub judice rule is not well adapted for regulating prejudicial publicity on social media during high-profile criminal trials. Changes to news distribution, consumption and participation, the affordances of social media platforms, and personalised algorithms may mean jurors are exposed to and potentially influenced by prejudicial publicity in new ways. The law has evolved in a way that is suited to regulating prejudicial information published by mainstream publishers.
However, it is not well suited to responding to prejudice caused by multiple prejudicial posts, published by multiple social media users. Future work is needed to understand how jurors might actually be influenced and what approaches to regulation by state or non-state actors are best suited for the age of social media. It may be that social media platforms ought to bear some responsibility for prejudicial posts, but exactly what that would look like in practice is not yet clear.
Twitter trials and Facebook juries iv
Table of Contents
Keywords ...... i Abstract ...... ii Table of Contents ...... iv List of Figures and Tables ...... viii Statement of Original Authorship ...... x Acknowledgements ...... xi 1 Introduction ...... 1 1.1 Overview of research ...... 1 1.2 Research Questions ...... 2 1.3 Methodology ...... 3 1.3.1 Social media ...... 3 1.3.2 Twitter ...... 7 1.3.3 Facebook ...... 9 1.4 Ethics ...... 11 1.5 Significance ...... 12 1.6 Terminology ...... 14 1.6.1 Social media ...... 14 1.6.2 Prejudicial publicity ...... 14 1.6.3 Facebook public pages, posts and comments ...... 15 1.6.4 Twitter ...... 18 1.6.5 Sub judice rule ...... 18 1.7 Publication ...... 19 2 Literature Review ...... 20 2.1 Overview ...... 20 2.2 Public interest and conflicting legal rights ...... 20 2.3 Sub judice rule ...... 25 2.4 Age of social media ...... 30 2.4.1 News consumption and participation ...... 31 2.4.2 Experiencing criminal trials via social media ...... 34 2.4.3 Law not suited to social media ...... 36 2.4.4 Internet jurisdiction and the regulation of social media ...... 37 2.5 Juries ...... 40 2.5.1 Jury exposure to prejudicial media ...... 42 2.5.2 Social media, prejudicial publicity and juror impartiality ...... 46 2.5.3 Efforts to shield jurors from prejudice ...... 48 2.6 Criminal justice ...... 49 2.7 Summary ...... 53 v Table of Contents
3 Research Design ...... 54 3.1 Overview ...... 54 3.2 Data Collection ...... 56 3.2.1 Selecting suitable high-profile criminal trials ...... 56 3.2.2 Publication during trials ...... 57 3.2.3 Twitter data for the murder trial of Gerard Baden-Clay ...... 58 3.2.3.1 Specified date range and set of key terms ...... 58 3.2.3.2 Selecting the data...... 59 3.2.3.3 How does TrISMA collect Twitter data? ...... 60 3.2.3.4 Creating a randomised sample for manual coding ...... 62 3.2.4 Twitter data for the murder trial of Gable Tostee ...... 63 3.2.5 Facebook data for the murder trial of Gable Tostee ...... 64 3.2.5.1 How does TrISMA collect Facebook data? ...... 64 3.2.5.2 Selecting the data...... 65 3.3 Data Analysis ...... 68 3.3.1 Content and discourse analysis ...... 69 3.3.2 Coding for prejudice ...... 73 3.3.3 Coding for user types ...... 82 3.3.4 Coding for themes ...... 83 3.4 Limitations ...... 85 3.4.1 Data selection and tools ...... 85 3.4.2 Tools ...... 86 3.4.3 Twitter and Facebook ...... 88 3.4.4 Social media in general ...... 88 3.4.5 Jurors and witnesses ...... 90 3.5 Summary ...... 90 4 Murder trial of Gerard Baden-Clay – Twitter Analysis ...... 91 4.1 Overview ...... 91 4.2 The trial ...... 92 4.3 Prevalence of prejudicial publicity ...... 94 4.4 Professional journalists versus non-journalists ...... 96 4.4.1 Professional journalists more factual, but non-journalists more emotive ...... 98 4.4.2 Professional journalists still focus on sensational aspects of trial ...... 101 4.4.3 Professional journalists should take care with attribution and compilations ...... 105 4.5 Prejudice on social media ...... 108 4.5.1 Prosecution bias ...... 108 4.5.2 Cumulative negative sentiment ...... 113 4.6 Summary ...... 115 5 Murder trial of Gable Tostee – Twitter Analysis ...... 117 5.1 Overview ...... 117 5.2 The trial ...... 118 5.3 Prevalence of prejudicial publicity ...... 119 5.4 Professional journalists versus non-journalists ...... 123 5.5 Prejudice on social media ...... 127 5.5.1 Prosecution bias ...... 129 5.5.2 Cumulative negative sentiment ...... 131 Twitter trials and Facebook juries vi
5.5.2.1 Audio recording ...... 132 5.5.2.2 Support of Wright ...... 133 5.5.2.3 Post offence conduct ...... 134 5.5.2.4 Trending on Twitter ...... 136 5.5.2.5 Gendered perspectives ...... 137 5.5.3 Lacking fairness and accuracy ...... 139 5.5.3.1 Biased, misleading or withholding material facts ...... 140 5.5.3.2 Issues of law ...... 142 5.5.4 Not made in good faith and without malice ...... 143 5.5.5 Impairing public confidence in the administration of justice ...... 146 5.5.5.1 Public access to evidence and information from the courtroom ...... 149 5.6 Summary ...... 150 6 Murder trial of Gable Tostee – Facebook Analysis ...... 151 6.1 Overview ...... 151 6.2 Prevalence of prejudicial publicity ...... 152 6.3 Professional journalists versus non-journalists ...... 154 6.4 Prejudice on social media...... 156 6.4.1 Prosecution bias ...... 157 6.4.2 Cumulative negative sentiment ...... 158 6.4.3 Lacking fairness and accuracy ...... 161 6.4.3.1 Inaccuracy ...... 161 6.4.3.2 Issues of law ...... 167 6.4.4 Impairing public confidence in the administration of justice ...... 168 6.5 Summary ...... 168 7 Sub judice rule not well adapted for social media ...... 169 7.1 Overview ...... 169 7.2 Summary of results ...... 169 7.2.1 How prevalent is prejudicial publicity on social media during high-profile criminal trials? ...... 169 7.2.2 How do professional journalists and non-journalists talk about high- profile criminal trials on social media? ...... 170 7.2.3 How do social media users show prejudice on social media during high- profile criminal trials ...... 172 7.3 How well adapted is the sub judice rule? ...... 175 7.3.1 Facebook and Twitter ...... 176 7.3.2 Law was well adapted for traditional media ...... 177 7.3.3 Alternatives shift the burden away from the state...... 178 7.4 News on social media is a change in kind ...... 180 7.4.1 Large audiences and affordances of platforms ...... 181 7.4.2 Users (including jurors) have no control over personalised algorithms ...... 183 7.4.3 Social media decentralised and largely unregulated ...... 183 7.5 Social media presents new risks to juror impartiality ...... 186 7.5.1 Is the risk to jurors the same as it always has been? ...... 187 7.5.1.1 Social media is ‘wrapped around our lives’ ...... 188 7.5.1.2 Increased availability of trial-related information ...... 191 7.5.2 Social media can affect how people think and feel ...... 193 7.5.2.1 Emotional contagion caused by news feeds ...... 193 7.5.2.2 Manipulating voters in the 2016 US presidential election campaign ...... 196 7.5.2.3 New opportunities for jury tampering ...... 197 7.5.2.4 Influencing key jurors may be enough to affect verdicts ...... 201 vii Table of Contents
7.5.2.5 Potential for misinformation to influence jurors ...... 205 7.6 Summary ...... 207 8 Conclusions and Future Research ...... 208 8.1 Overview ...... 208 8.2 Regulating prejudice on social media is difficult ...... 208 8.3 Potential future approaches ...... 211 8.4 Social network analysis – network mapping ...... 217 8.5 Future research ...... 218 8.6 Digital methods for legal analysis ...... 223 8.7 Conclusion ...... 224 Bibliography ...... 225
Twitter trials and Facebook juries viii
List of Figures and Tables
Figure 1.1 Facebook visual aid - The Australian public page containing a post with comments and likes...... 17 Figure 1.2 Twitter visual aid – a tweet by The Australian ...... 18 Table 3.1: Research design – overview of methods for data extraction and analysis...... 55 Table 3.2: Tostee Facebook analysis - summary of comments collected...... 68 Figure 3.1 Research design – responding to research questions using quantitative & qualitative analysis...... 69 Table 3.3: Research design - coding scheme for prejudice in tweets and comments...... 74 Figure 4.1: Baden-Clay Twitter analysis (total Twitter dataset) - tweet activity during the trial...... 94 Figure 4.2: Baden-Clay Twitter analysis (random sample) - 6% of tweets contained prejudicial content...... 95 Figure 4.3: Baden-Clay Twitter analysis (random sample) - most tweets were posted by professional journalists...... 97 Table 4.1: Baden-Clay Twitter analysis (random sample) - results after coding for prejudice and user type...... 97 Figure 4.4: Baden-Clay Twitter analysis (random sample) - non-journalists posted more prejudicial tweets than professional journalists...... 98 Figure 5.1: Tostee Twitter analysis (total Twitter dataset) - tweet activity during the trial...... 120 Figure 5.2: Tostee Twitter analysis (random sample) - 17.4% of tweets contained prejudicial content...... 121 Table 5.1: Tostee Twitter analysis (random sample) – results after coding for prejudice...... 122 Figure 5.3: Tostee Twitter analysis (random sample) – most tweets were posted by professional journalists...... 124 Table 5.2: Tostee Twitter analysis (random sample) - results after coding for prejudice (by category) and user type...... 124 Table 5.3: Tostee Twitter analysis (random sample) - results after coding for prejudice (none, low and high) and user type...... 125 Figure 5.4: Tostee Twitter analysis (random sample) - non-journalists posted more prejudicial tweets than professional journalists...... 125 Figure 5.5: Tostee Twitter analysis (random sample) – coded categories indicate a prosecution bias...... 130 Figure 5.6: Tostee Twitter analysis (total sample) – more comments were pro- prosecution than pro-defence...... 131 Figure 5.7: Tostee Twitter analysis - images of Gable Tostee as he exited the lift via the basement ...... 135 ix List of Figures and Tables
Figure 6.1: Tostee Facebook analysis (total sample) - 62% of posts contained prejudicial content...... 153 Table 6.1: Tostee Facebook analysis (total sample) - results after coding for prejudice...... 153 Figure 6.2: Tostee Facebook analysis (total sample) – all comments were posted by non-journalists...... 155 Figure 6.3: Tostee Facebook analysis (total sample) – coded categories indicate a prosecution bias...... 157 Figure 6.4: Tostee Facebook analysis (total sample) – more comments were pro- prosecution than pro-defence...... 158 Table 6.2: Tostee Facebook analysis - thematic analysis of comments with examples...... 160 Figure 7.1: Prevalence of prejudicial publicity across case studies ...... 170 Figure 7.2: Prejudicial tweets / comments posted by non-journalists ...... 171 Figure 7.3: Prejudicial tweets / comments posted by professional journalists ...... 172 Figure 7.4: Tostee Twitter and Facebook pro-prosecution bias ...... 173
Twitter trials and Facebook juries x
Statement of Original Authorship
The work contained in this thesis has not been previously submitted to meet requirements for an award at this or any other higher education institution. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made.
Signature: QUT Verfied Signature
Date: 25 October 2019 xi Acknowledgements
Acknowledgements
I am immensely grateful to my principal supervisor, Associate Professor Nicolas
Suzor, whose intellect and insight has steered my research from conception to completion. Nic is a gifted ‘coach’ who has cultivated my ideas and skills through exceptional instruction and patience. I have been truly privileged to be led on a journey of learning and enrichment over many years by such an outstanding and committed supervisor. Nic—as always, thank you for everything!
I also wish to thank my associate supervisor, Dr Kylie Pappalardo, for her guidance, empathy and fresh perspective. Kylie’s supervision of my research came later in my project and at a time when her insight, expertise and constructive suggestions were particularly important. Kylie—thank you for being an invaluable and supportive advisor through the final stages of my PhD.
To Professor Des Butler, Dr Aljosha Karim Schapels and Professor Afshin
Akhtar-Khavari, thank you for your considerable and incredibly helpful feedback at my final seminar — your advice ensured my final thesis was the best it could be. To
Professor Axel Bruns, thank you for your early input and advice at my confirmation seminar — your guidance was particularly crucial to shaping my use of digital methods. Thank you also to Professor Reece Walters whose early support encouraged me to pursue a research career. I wish to thank Professor Matthew Rimmer who insisted I present my work before I’d even reached confirmation. Despite my hesitation and nerves, this was a significant turning point in extending my understanding and confidence in my own research. I also wish to thank Head of School, Associate
Professor Judith McNamara, whose support in balancing my workload and research Twitter trials and Facebook juries xii has been critical to the completion of my thesis. Finally, I’d like to thank my colleagues and peers in the Law School, Digital Media Research Centre, and School of Justice for their friendship and collegiality.
On a personal note, I wish to thank my family and friends who have encouraged and motivated me, even when I have had doubts. To my mum — your wisdom, strength and logistical support has made everything possible. To my darling daughter
Macy — your unconditional love and faith in me inspires all I do. And to Troy — for your enduring love, support and belief in me, I am truly blessed.
On a final note, I would like to acknowledge receipt of a Research Training
Program Scholarship and a QUT Excellence Top Up Scholarship. These scholarships were significant in allowing me to take time away from full-time work, so I could undertake my research project and, ultimately, pursue an academic career.
1 Introduction
1 Introduction
1.1 Overview of research
In countries where trial by jury is the norm, governments and courts are increasingly concerned about the potential impact of social media on the fairness of criminal trials.1 Information that might influence jurors and prejudice an ongoing trial has the potential to be spread through social media conversations in a way that is not as easily controlled as information disseminated through traditional media. The increasing prevalence of social media accordingly raises important questions about the potential for unregulated prejudicial publicity to influence juries and jeopardise the right to a fair trial.
This thesis investigates whether the protections afforded by the sub judice rule— which prohibits the publication of information that might prejudice pending criminal trials—are fit for the digital age. Before the introduction of the internet, the law in
Australia was well established in terms of how it regulated the behaviour of publishers, but there are concerns about its effectiveness in the age of social media.
To investigate how well adapted the Australian sub judice rule is for regulating prejudicial publicity on social media during high-profile criminal trials, we need to
1 John Fairfax Publications Pty Ltd v District Court (NSW) (2004) 61 NSWLR 344; News Digital Media Pty Ltd v Mokbel (2010) 30 VR 248, 267; R v K (2003) 59 NSWLR 431, 448; Gilbert v The Queen (2000) 201 CLR 414; R v Glennon (1992) 173 CLR 592, 603; Murphy v R (1989) 167 CLR 94, 99; Hinch v Attorney General (Vic) (1987) 164 CLR 15, 74; Lodhi v R [2007] NSWCAA 360, [153]; Justice M Warren, ‘Open Justice in the Technological Age’ (2013) 40(1) Monash University Law Review 45, 46; Justice P McLellan, ‘Looking inside the Jury Room. Speech Delivered at the Law Society of New South Wales Young Lawyers 2011 Annual Criminal Law Seminar’ (5 March 2011); James Spigelman, ‘The Internet and the Right to a Fair Trial’ (2005) 29(6) Criminal Law Journal 331; Virginia Bell, ‘How to Preserve the Integrity of Jury Trials in a Mass Media Age’ (2005) 7(3) Judicial Review 311.
Twitter trials and Facebook juries 2 better understand the prevalence of prejudicial publicity on social media, how professional journalists and non-journalists talk about criminal trials, and how social media users show prejudice on social media during high-profile criminal trials. To answer these questions I engaged in a detailed examination of two high-profile criminal trials in Brisbane—the murder trials of Gerard Baden Clay and Gable Tostee.
Through my analysis I gained insight into the types of information empanelled jurors might see, should they be exposed to information about their trial on social media. This also made it possible to better understand the effectiveness of existing media law.
Overall, I concluded that the sub judice rule is largely effective for professional journalists, but less so for non-journalists. I found there was a prosecution bias in tweets and comments, and this may lead to a broader general sentiment that is weighted against the accused. I raise concerns about tweets or comments that are legally compliant—but that contain low-level negative sentiment—and the potential for their collective weight to cause prejudice. Finally, I note that the sub judice rule is not suited to regulating material that is massively distributed and communicated collectively by large numbers of users on social media.
1.2 Research Questions
This thesis considers the following research question: How well adapted is the
Australian sub judice rule for regulating prejudicial publicity on social media during high-profile criminal trials? To respond to this research question, I consider the following sub-questions:
a. How prevalent is prejudicial publicity on social media during high-profile
criminal trials?
3 Introduction
b. How do professional journalists and non-journalists talk about high-profile
criminal trials on social media?
c. How do social media users show prejudice on social media during high-
profile criminal trials?
1.3 Methodology
I used digital methods to collect, analyse, and visualise social media data from the criminal trials of Gerard Baden-Clay and Gable Tostee. I used content and discourse analysis to analyse the data and respond to my research questions. My socio- legal approach brings an empirical perspective to this important area of media law. I have set out below my rationale for undertaking social media research, with a focus on
Twitter and Facebook. My methodology allows me to investigate user action and participation2 — specifically, how social media users view, post, share content, collaborate and discuss information about high profile trials. I provide a comprehensive overview of my methods in Chapter 3.
1.3.1 Social media The integration of social media into daily lives is indisputable.3 Social media platforms have seen a surge in adoption rates in recent years and they have become platforms ‘for rich expression and exchange for a highly diverse user base’.4 This increase in adoption rates has had the effect of producing vast quantities of user generated content that is valuable for research.
2 G Veletsianos, ‘Higher Education Scholars’ Participation and Practices on Twitter’ (2012) 28(4) Journal of Computer Assisted Learning 336, 339. Jean E Burgess and John A Banks, ‘Social Media’ in The Media and Communications in Australia [4th Edition.] (Allen & Unwin, 2014) 285. 3 Lori McCay-Peet and Anabel Quan-Haase, ‘What Is Social Media and What Questions Can Social Media Research Help Us Answer?’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd) 13, 13–14. 4 P Jurgens, ‘Communities of Communication: Making Sense of the “Social” in Social Media’ (2012) 30(3–4) Journal of Technology in Human Services 186, 186.
Twitter trials and Facebook juries 4
Social media research makes it possible to bring new perspectives to respond to complex problems that have otherwise been considered using more traditional methods.5 Data derived from user posts, likes, friends or follows, provides a rich source of information that can be used to answer social science research questions across a broad range of disciplines.6 It provides great opportunities for scholars as it is characterised by ‘high velocity, large volume, diverse variety, exhaustivity in scope, fine-grained resolution, relational in nature, and flexibility in its approach’.7 Social media scholarship is frequently concerned with big data,8 but increasingly it is also concerned with smaller scale qualitative approaches, such as the approach I use in this study.9 McCay-Peet and Quan-Haase have provided a useful summary of the benefits of social media research:
Because of its proliferation in society as well as its unique technological
affordances, social media provides new avenues for researchers across
multiple disciplines, including health sciences, sociology, and political
science, to collect rich, vast, and networked data, recruit diverse groups of
participants and perform complex analyses.10
When choosing social media platforms to study, researchers often focus on proprietary platforms — platforms that preserve proprietary control to ‘a single
5 DO Otieno and VB Matoke, ‘Social Media as Tool for Conducting Academic Research’ (2014) 4(1) International Journal of Advanced Research in Computer Science and Software Engineering 962. 6 A Gopaldas, ‘Marketplace Sentiments’ (2014) 41(4) Journal of Consumer Research 995; Lee Rainie et al, ‘Social Media and Political Engagement’ [2012] Pew Research, Washington DC; NJ Reavley and PD Pilkington, ‘Use of Twitter to Monitor Attitudes toward Depression and Schizophrenia: An Exploratory Study’ (2014) 2 PeerJ 647. 7 R Kitchin, ‘Big Data, New Epistemologies and Paradigm Shifts’ (2014) 1(1) Big Data & Society 1. 8 Ibid. 9 McCay-Peet and Quan-Haase (n 3) 14. 10 Ibid.
5 Introduction provider that solely controls its technology’11 — such as Twitter or Facebook.12 These two platforms have been particularly popular sources of data for social media research for two reasons: Facebook due its vast uptake and user base, and Twitter due to the
‘transformative effect [it has had] on how information and news diffuse throughout society’.13 Twitter has not been as widely adopted by the public as Facebook, but as traditional media often makes references to Twitter activity in news broadcasts, it has
‘become an important part of public discourse’.14
It is for these reasons I have chosen to collect Twitter and Facebook data for analysis in this study. As well as being valuable platforms on their own, my choice to investigate both Twitter and Facebook makes it possible to understand how ‘users engage with features that are similar across platforms’ and to develop theories that may be applicable more broadly across social media.15 An examples of a feature that is similar across platforms is the hashtag. Hashtags are a type of bookmarking and signalling used by the collective public to curate posts on social media,16 such as
#badenclay or #tostee. They provide a type of ‘social filtering’ of information online that users can include in their posts and that others can use as keywords for locating posts on particular topics of interest.17 The fact that news is curated through the use of hashtags means, unlike traditional media, it is easy to find news and commentary that has been posted about a high-profile event using a simple search function.18 Other
11 Thomas R Eisenmann, ‘Managing Proprietary and Shared Platforms’ (2008) 50(4) California Management Review. 12 McCay-Peet and Quan-Haase (n 3) 15. 13 McCay-Peet and Quan-Haase (n 3). 14 Ibid. 15 Ibid 15. 16 Axel Bruns and Jean E Burgess, ‘The Use of Twitter Hashtags in the Formation of Ad Hoc Publics’ in Proceedings of the 6th European Consortium for Political Research (ECPR) General Conference 2011 (2011). 17 McCay-Peet and Quan-Haase (n 3). 18 Michele Zappavigna, ‘Ambient Affiliation: A Linguistic Perspective on Twitter’ (2011) 13(5) New Media & Society 788 (‘Ambient Affiliation’).
Twitter trials and Facebook juries 6 examples include the affordances of each platform — such as ‘like’, follow, or share functions — that allow social media users to ‘create, share, interact with, and mobilize content’ and build communities.19
The underlying structures of Twitter and Facebook, however, are different and this is important for my general data analysis. Twitter supports ‘one-way flows of information’ — it allows users to follow another account without reciprocation.20
Facebook, however, only allows linkages when both users consent to being connected.21 This changes the way users connect with each other, the way users may access news stories, and the way they may participate online or communicate with each other.
Finally, it is important to understand what sets social media apart from traditional media,22 as it is these differences that affect how the law operates in the digital age compared to the pre-internet era. Bruns suggests: ‘All media are social, but only a particular subset of all media are fundamentally defined by their sociality, and thus distinguished, for example, from the mainstream media of print, radio, and television’.23 It is the social aspect that makes social media so different from traditional media24 — an aspect never anticipated when the sub judice rule was established. It is also the digital aspect that makes social media different from traditional media, as social elements and affordances are so deeply embedded into technology, such as
19 Lori McCay-Peet and Anabel Quan-Haase, ‘A Model of Social Media Engagement: User Profiles, Gratifications and Experiences’ in Heather L O’Brien and Luke Sloan (eds), Why Engagement Matters: Cross-Disciplinary Perspectives and Innovations on User Engagement with Digital Media (Springer-Verlag, 2016) 199. 20 McCay-Peet and Quan-Haase (n 3). 21 Ibid. 22 B Hogan and Anabel Quan-Haase, ‘Persistence and Change in Social Media’ (2010) 30(5) Bulletin of Science, Technology & Society 309. 23 Axel Bruns, ‘Making Sense of Society through Social Media’ (2015) 1(1) Social Media + Society 1. 24 Burgess and Banks (n 2).
7 Introduction smartphones.25 This has completely changed the way members of the community can engage with news stories: while they were once asynchronous consumers of news media, they are now active participants who can synchronously respond to news stories or even publish their own.
1.3.2 Twitter Twitter is a type of microblogging service that focuses ‘on short updates that are pushed out to anyone subscribed to receive [them]’.26 One of its functions is to serve as an information network, relied on by users for sharing information including news.27
It is recognised as a ‘particularly important global platform for public (as opposed to private or semi-private) communication’28 and this makes it especially well-suited as a network for covering breaking news and live events.29 As mentioned previously,
Facebook’s network is created by reciprocal ‘friend’ relationships which create
‘smaller-scale, stronger-tie networks’.30 However, Twitter users can follow any globally public account without the need for those accounts to follow back, resulting in ‘larger-scale, weaker-tie networks’.31 This communicative structure makes Twitter particularly ‘responsive to breaking news events … facilitating the broad dissemination of emerging information within very short timeframes’.32 With 320
25 McCay-Peet and Quan-Haase (n 3) 15. 26 T Grahl, ‘The 6 Types of Social Media’ (2013)
Twitter trials and Facebook juries 8 million active monthly users worldwide,33 Twitter is a powerful platform for ‘ongoing discussion — and instant evaluation — of newsworthy events’.34 It provides
‘interaction, participation and connectivity’ for journalists and audiences alike, 35 and has become one of the most prominent platforms in changing how everyday people access, discuss, and share news stories.36
Twitter’s volume of users, underlying communicative structure, and the fact that posts from 95 per cent of all accounts are globally public,37 make it valuable for my research. It provides a particularly useful lens through which to examine the dissemination of trial related news and respond to my research questions. In Australia, there are approximately 3.72 million unique Twitter accounts, representing approximately 15 per cent of the general population.38 A 2016 Sensis report suggests this figure may be as high as 19 percent.39 It is not only those with a Twitter account who have the potential to be exposed to prejudicial publicity on the platform; due to the public nature of Twitter tweets will also be visible even to those who are ‘logged- out visitors to the site’.40
33 Axel Bruns and Katrin Weller, ‘Twitter As a First Draft of the Present – And the Challenges of Preserving It for the Future’ in Proceedings of the 8th ACM Conference on Web Science (Paper presented at 8th International ACM Web Science Conference 2016, Hannover, Germany, 22–25 May 2016) 183, 183
9 Introduction
1.3.3 Facebook Facebook is a web-based social networking site that allows users to ‘construct a public or semi-public profile within a bounded system, articulate a list of other users with whom they share a connection, and view and traverse their list of connections and those made by others within the system’.41 Worldwide, ‘more than one billion people use Facebook each day to connect and interact with friends, colleagues, family, and others’.42
Throughout the last decade, Australians — as individuals, businesses, corporations and government — have increasingly embraced Facebook as an important platform for daily communication and for distributing information of interest. It has also become the primary source of news for many users. An Australian media survey conducted in 2018 reported that 85% of respondents (and 95% of 14-33 years olds or ‘millennials’)43 were active on some form of social media44 and that
Facebook was the platform on which they spent most time each week.45 Thirty-seven percent of respondents reported most frequently using ‘digital sources, such as online newspapers and news sites, social media sites, and news aggregators’ as their primary source of news.46 And 17% of all respondents (28% for Millennials) reported relying on social media as their primary news source.47 Interestingly, 64% of respondents
41 D Boyd and NB Ellison, ‘Social Network Sites: Definition, History, and Scholarship’ (2007) 13 Journal of Computer-Mediated Communication 210, 211. 42 J Vitak, ‘Facebook as a Research Tool in the Social and Computer Sciences’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd, 2016) 627, 627. 43 Leora Nevezie and Jessica Eade, Media Consumer Survey 2018 - Australian Media and Digital Preferences - Seventh Edition (Deloitte, 2018) 1, 6. In this survey, the term ‘millennial’ refers to any person aged 14 – 33 years in 2016, at the time of the survey. 44 Ibid 12. 45 Ibid. 46 Niki Alcorn and Lucy Buchanan, Media Consumer Survey 2017 - Australian Media and Digital Preferences - 6th Edition (Deloitte, 2017) 1, 25. 47 Nevezie and Eade (n 43) 17.
Twitter trials and Facebook juries 10 agreed that ‘nothing would entice them to pay for the news’,48 perhaps suggesting our reliance on free news is likely to continue. That Millennials are even more active in their social media use and reliance on it for news, may mean the risk of exposure to prejudicial publicity on social media will continue to increase in future as this generation ages.
As a primary news source, Facebook is quite different from traditional media.
First, the affordances of the platform allow users to comment on each news article or on other users’ comments. Second, when news articles appear in a user’s Facebook newsfeed, at least some comments are already visible together with the news article headline and link. The remaining comments are accessible by clicking to view them.
Finally, there is no way for a social media user to control what information they may see — including potentially prejudicial publicity. It is the platform algorithm that will dictate which news articles and comments appear in which users’ newsfeeds, and there is no option for users (jurors) to filter out information about criminal trials. Jurors using social media may inadvertently encounter news articles about a trial and be exposed to potentially prejudicial publicity contained in related comments.
Facebook’s vast userbase, its use as a primary source of news, and its algorithmic approach to determining which users see what information, make it an important platform in the daily lives of our communities and, as a result, it presents a significant opportunity to investigate how users participate or engage with trial-related news articles.
48 Ibid 10.
11 Introduction
1.4 Ethics
Following submission of a high-risk human research ethics application to
Queensland University of Technology, I received ethics approval for my research
(Approval #1400000861). I contributed to the preparation of the original National
Ethics Application Form — together with Associate Professor Nicolas Suzor — which received approval on 2 February 2015, and the subsequent variation which received approval on 19 October 2015. There have been more recent variations to this ethics application, but they relate to the work of other listed investigators and do not relate to my particular research.
In the preparation of the application, I considered ethical questions relating to data collection, aggregation, reporting or publishing of data, and issues around privacy, accuracy and accountability. 49 The social media data I use is publicly available, but I needed to address questions of consent in my application as ‘not all users of social media sites are comfortable with (or aware of) the use of the data for analysis’.50 I also needed to consider what repercussions there might be for the Twitter and Facebook users in my dataset. If they were identified as having breached the law of sub judice contempt, or some other law or moral code of conduct, there may be consequences ranging from ‘imprisonment [to] stigma, ridicule, and harm to reputation … [by] making tweets and … [other] text searchable, even if anonymized.’51 When tweets or comments (text) are searchable, individual users might be re-identified using data from
49 J Grimmelmann, ‘The Law and Ethics of Experiments on Social Media Users’ (2015) 13(2) Colorado Technology Law Journal 221; V Goel, ‘As Data Overflows Online, Researchers Grapple with Ethics’, The New York Times (online at 12 August 2014)
Twitter trials and Facebook juries 12 my thesis or publications.52 Re-identification can occur, for example, by copying and pasting a tweet or comment into Google in an effort to identify the original author, owner, or publisher.
To protect the privacy of individuals in my dataset, I have anonymised individual user details by removing identifying information, other than whether they identified as a professional journalist or non-journalist. Importantly, I have chosen to paraphrase the content of tweets or comments where publication posed a risk of re-identification.
I made an exception, however, for a small number of tweets posted by professional journalists, where including the specific tweet was important for demonstrating my results or findings. In those instances I included tweets in their original form if they were both significant to the discussion and posted by a high profile professional journalist (or media organisation) that would be aware of the public nature of their tweets.
1.5 Significance
This thesis builds on existing research that looks at how exposure to prejudicial publicity in the media can influence juries.53 My research responds to the need to understand with some empirical precision the themes and the actors who are responsible for posting prejudicial content about criminal trials on social media. In doing so, I am able to draw conclusions about the prevalence of prejudicial information on social media, how users engage with that content, and also to provide insight into what prejudice actually looks like in a social media context. My approach takes into
52 Anabel Quan-Haase and Lori McCay-Peet, ‘Building Interdisciplinary Social Media Research Teams: Motivations, Challenges, and Policy Frameworks’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd, 2016) 40, 44. 53 Kerstin Braun, ‘Yesterday Is History, Tomorrow Is a Mystery-the Fate of the Australian Jury System in the Age of Social Media Dependency’ (2017) 40 UNSWLJ 1634, 1635.
13 Introduction account the structure and affordances of social media platforms and uses this to better understand how users experience conversations about criminal trials. By understanding user experiences, it is possible to assess how the sub judice rule is working in practice and whether it is fit for the digital age.
As research trends continue to acknowledge and seek out the benefits of interdisciplinary work, it will be important for scholars to have exposure to and some understanding of emerging internet based methods. The methods I use lend real insight into existing legal debates, insight that is difficult to achieve through other means. The use of digital methods in the discipline of law is relatively uncommon,54 but it is important we develop new ways of answering legal questions, particularly in the internet era. By using a combination of digital and social science methods, my research allows us to understand the voices of everyday social media users and to use this insight to inform our responses to emerging legal problems. My use of content analysis, including thematic analysis, tells us more than how users might access prejudicial publicity — it also provides insight into how ordinary people are talking, engaging with, and experiencing the judicial system. This approach can be applied to many other problems. This project can be used as an example of how to use mixed digital and social methods to inform more traditional legal analysis. These methods may be applied to any number of legal and criminal justice contexts, including the many problems arising out of new communications technologies.
54 McCay-Peet and Quan-Haase (n 3). Social media research tends to be more common in disciplines such as marketing, communications, politics, computer science, economics, health, business and education.
Twitter trials and Facebook juries 14
1.6 Terminology
Many of the terms I use in this thesis can be defined in different ways, depending on discipline and context. The rapid evolution of social media throughout the last decade also means there have been shifts in the way we understand different social media platforms and their affordances. Below are the definitions or explanations of terms as they apply to this research. At the end of this section, I have included two visual aids to assist with understanding my explanations of Twitter and Facebook data
(Figures 1.1 and 1.2).
1.6.1 Social media Social media has been variously defined. For this study, social media ‘are web- based services that allow individuals, communities, and organizations to collaborate, connect, interact, and build community by enabling them to create, co-create, modify, share, and engage with user-generated content that is easily accessible’.55 My research focusses on Twitter and Facebook as these have become common sources of news, including news about criminal trials. When I refer to social media conversations, I am referring to content that is ‘created, organised and distributed by [users]’,56 specifically tweets (Twitter) and comments (in response to news articles (posts) on Facebook public pages).
1.6.2 Prejudicial publicity There is no single definition of prejudicial publicity, however, it can be understood by reference to legal cases where publications have been found to be prejudicial. Prejudice can be explained as an antipathy to others ‘based upon a faulty and inflexible generalization’ and may refer to bias resulting from factors such as
55 Ibid 17. 56 The Hon Thomas F Bathurst, ‘Social media: The end of civilization?’ The Warrane Lecture, University of New South Wales, Sydney, 21 November 2012, p7.
15 Introduction membership of a group,57 the attractiveness or demeanour of defendants, witnesses, judges or legal representatives,58 or negative media coverage.59 It may be factual prejudice such as information about an accused’s prior convictions or personal character, or emotional prejudice such as that generated from graphic media reports of crimes, injuries and suffering.60 The broad categories of publications that have been found to be contemptuous by the courts in criminal trials include those containing material asserting guilt61 or innocence,62 information about confessions63 or convictions,64 and content that criticises or disparages the accused.65 Publicity refers to ‘notice or attention given to someone or something by the media’.66 In this study, references to prejudicial publicity refer to tweets, comments or content that have been posted on social media and that could be described as prejudicial.
1.6.3 Facebook public pages, posts and comments This thesis focuses on publicly accessible pages created by a business or organisation (or celebrity, cause or brand). Media organisations and professional journalists post publications, such as news articles about criminal trials, on their
57 Gordon W 1897-1967 Allport, The Nature of Prejudice (Doubleday, Abridged., 1958) 9. 58 Ronald Mazzella and Alan Feingold, ‘The Effects of Physical Attractiveness, Race, Socioeconomic Status, and Gender of Defendants and Victims on Judgments of Mock Jurors: A Meta-Analysis1’ (1994) 24(15) Journal of Applied Social Psychology 1315 (‘The Effects of Physical Attractiveness, Race, Socioeconomic Status, and Gender of Defendants and Victims on Judgments of Mock Jurors’). 59 James RP Ogloff and Neil Vidmar, ‘The Impact of Pretrial Publicity on Jurors: A Study to Compare the Relative Effects of Television and Print Media in a Child Sex Abuse Case’ (1994) 18(5) Law and Human Behavior 507 (‘The Impact of Pretrial Publicity on Jurors’). 60 Ibid 510–511. 61 Attorney-General (NSW) v Radio 2UE Sydney Pty Ltd and John Laws [1998] NSWSC 28; Attorney- General (Qld) v WIN Television (Qld) Pty Ltd [2003] QSC 157. 62 Director of Public Prosecutions v Wran (1987) 86 FLR 92; R v Castro, Onslow & Whalley’s Case [1873] LR 9 QB 219; R v Pearce (1992) 7 WAR 395. 63 Attorney-General (NSW) v John Fairfax and Sons Ltd (1980) 1 NSWLR 362; Attorney-General (NSW) v TCN Channel Nine Pty Ltd (1990) 20 NSWLR 368, 380. 64 Maxwell v Director of Public Prosecutions [1935] AC 309, 317; Attorney General (NSW) v Willesee (1980) 2 NSWLR 143. 65 Director of Public Prosecutions v Francis (2006) 95 SASR 302; R v Saxon, Hadfield and Western Mail Ltd [1984] WAR 283; R v Australian Broadcasting Corporation [1983] Tas R 161. 66Oxford Dictionaries Language matters, publicity (16 November 2015)
Twitter trials and Facebook juries 16
Facebook page. Facebook users choose to like a public page in order to follow it.
Users are prompted to comment on articles (posts) or other users’ comments and they can also like or share the article (post). Facebook pages are accessible to the public without the need for pre-approval by the account owner, however, to comment on a post users will need to have their own Facebook account. A Facebook page should be distinguished from a Facebook profile which is set up by an individual. In their
Facebook profile, users share information about themselves and may upload photos and videos to share with friends. Facebook profiles are generally considered to be a private space67 accessible to friends and family who become friends with permission of the user.
A post is an item placed on a Facebook public page by the owner (or authorised user) of that page. For this research, Facebook posts refer to news articles about criminal trials posted on the Facebook pages of media organisations or professional journalists. For example, The Australian (newspaper) places news items (posts) on its
The Australian News/Media Facebook page.68
Facebook users can interact with posts to public pages in a number of different ways. Users can respond to posts by adding a comment directly on the post as it appears on the organisation’s page. Comments might respond to the news article (post) or to other users’ comments. A user can also choose to share individual posts to their own profile, allowing others to add comments that are only visible to those who can see the user’s post. This research only examines posts to public pages and comments that users have made directly to a news article posted by a page – comments that are visible to
67 Should they choose, a user may set their Facebook profile settings to public, in which case the profile will be visible to anyone, regardless of whether the viewer has their own Facebook account or is logged into the network. 68 See https://www.facebook.com/theaustralian?fref=ts.
17 Introduction the world. It does not include comments made to smaller audiences when a news article is shared to a user’s profile. This is an important limitation to protect Facebook users’ privacy and comply with ethics restrictions. However, this limitation also means my analysis misses what happens when Facebook users share public posts with their friends. It may be that a discussion with a much different character happens there, but
I will not have access to this.
Figure 1.1 illustrates how a news article (post) and associated comments appear in Facebook users’ newsfeeds.
Facebook (public pages) Facebook post (including public page for image) The Australian
Total likes at time of viewing
Users can like, comment or share
Further comments User comments (many of which will also have additional replies)
User replies to comments
Figure 1.1 Facebook visual aid - The Australian public page containing a post with comments and likes.69
69 Names and profile images of users have been shaded to ensure privacy. Excerpt selected from https://www.facebook.com/theaustralian/ 23 November 2015.
Twitter trials and Facebook juries 18
1.6.4 Twitter Twitter is an American online news and social networking service where users
post and interact with messages or ‘tweets’. At the time of conducting this research,
tweets were restricted to 140 characters, however, this limit was increased to 280
characters in late 2017. Figure 1.2 illustrates how a tweet, in this case containing a link
to a news article, appears in Twitter users’ feeds.
Twitter user (in this case the user is The Australian).
url link to full report in The Australian
Users can reply, retweet or like (heart) the tweet. When this tweet was copied there were 5 comments13 retweets and 7 likes.
Figure 1.2 Twitter visual aid – a tweet by The Australian
1.6.5 Sub judice rule The sub judice rule — also known as sub judice contempt of court or contempt
by publication — prohibits the publication of prejudicial material while a criminal (or
civil) matter is before the courts.70 It prescribes that it is a contempt to publish material
70 Sub judice contempt prohibits the publication of material that has a ‘real and definite tendency … to prejudice … particular legal proceedings’: see, New South Wales Law Reform Commission, Contempt Discussion Paper, above n 6, 106 [4.3]. It need not be shown that a juror has actually been influenced by published material, only that there is a sufficient risk of influence to juror impartiality: see Bell v Stewart (1920) 28 CLR 419, 432 (Isaacs and Rich JJ); A-G (NSW) v John Fairfax & Sons Ltd (1980) 1 NSWLR 362, 368 (The Court); R v West Australian Newspapers Ltd; Ex parte DPP (WA) (1996) 16 WAR 518, 531 (The Court).
19 Introduction relating to pending (current or forthcoming) legal proceedings if that material has, ‘as a matter of practical reality, a tendency to interfere with the due course of justice’.71
The rule began as a means of preserving juror impartiality, so as to safeguard jury trials. By making it unlawful to publish prejudicial publicity, the rule aimed to prevent juror exposure to prejudice and, in turn, possible juror influence. I provide further detail about the sub judice rule in Chapter 2.
1.7 Publication
In 2017, I published an article based on my analysis of the murder trial of Gerard
Baden-Clay:
Rachel Hews and Nicolas Suzor, ‘‘Scum of the earth’: An analysis of
prejudicial information on Twitter during the Baden-Clay murder trial’ (2017)
40(4) UNSW Law Journal 1604.
Portions of this thesis have been published, including most of Chapter 4 and parts of Chapters 1 and 3. I jointly-authored this paper with my principal supervisor,
Associate Professor Nicolas Suzor. I was the lead author and had overall intellectual carriage of the research. I was responsible for background research, conception and design, interpretation of data, legal analysis, and drafting most of the paper. I was also the corresponding author which included managing and addressing the majority of revisions and editing of the document until publication. Associate Professor Nicolas
Suzor’s contribution focussed primarily on general supervision, supporting data collection, and drafting those aspects of the article relating specifically to internet governance and media regulation. These substantive components are peripheral to the main research questions in this thesis.
71 Ex parte Auld; Re Consolidated Press Ltd (1936) 36 SR NSW 596, 597.
Twitter trials and Facebook juries 20
2 Literature Review
2.1 Overview
In this chapter, I draw on both legal and criminal justice literature to provide the broad context for my research. I explore the long-standing history of public interest in criminal trials, and the conflicting notions of freedom of speech and a right to a fair trial. I consider the complexities of the sub judice rule and the challenges it faces in the age of social media. Finally, I contextualise my research within the broader bodies of literature concerning juries and criminal justice.
2.2 Public interest and conflicting legal rights
A combination of public interest in criminal trials and conflicting legal rights makes the enforcement of the sub judice rule quite difficult in practice.72 The public’s desire to know details about trials73 — combined with the principle of open justice74
— mean that criminal trials are a type of news and there is a demand for the publication of as much information about them as possible. The media represent ‘the eyes and the ears of the general public’.75 They provide the means through which members of the public who cannot attend court may be kept informed.76 The principle of free speech
72 Law Reform Commission of Western Australia, Report on Review of the Law of Contempt (No Project No 93, June 2003) 17–23; Law Reform Commission of Western Australia, Discussion Paper on Contempt by Publication (No Project 93 (II), March 2002) 1–3, 6–16. 73 Lawrence M Friedman, The Big Trial: Law as Public Spectacle (University Press of Kansas, 2015) 137–159. 74 Roxanne Burd and Jacqueline Horan, ‘Protecting the Right to a Fair Trial in the 21st Century - Has Trial by Jury Been Caught in the World Wide Web?’ (2012) 36 Criminal Law Journal 103, 173; Law Reform Commission of Western Australia, ‘Discussion Paper on Contempt by Publication’ (n 72) 14– 16. 75 Attorney-General v Guardian Newspapers (No 2) (1990) 1 AC 109, per Donaldson MR. 76 R v Davis (1995) 57 FCR 512; Ex parte Terrill; Re Consolidated Press Ltd (1937) 37 SR NSW 255, 257–258; Australian Law Reform Commission, Prejudicial Publicity and the Courts - Reference on Contempt of Court, Tribunals and Commissions (1986) 96.
21 Literature Review or expression — or freedom of the media — supports media organisations and professional journalists in the publication of fair and accurate reporting about trials.77
Within this context, the defendant’s right to a fair trial78 requires that media organisations and professional journalists must refrain from publishing any information that might prejudice the trial.79
That members of the public are interested in criminal trials is not a new phenomenon. High-profile criminal trials have long been a source of public spectacle,80 arguably propagated in mainstream media as a form of public entertainment.81 A broad public interest in crimes has seen media outlets respond by publishing significant information about criminal trials, especially ‘sensational … or extraordinary’82 details, including those relating to violence or killings, convictions and confessions, guilt or innocence, and suitable punishments. This type of publicity raises questions, however, about the balance between the freedom of journalists to report on matters of public interest and the right to a fair trial for the accused.83
Freedom of the media relies, broadly, on the principle of free speech (or expression), which is commonly accepted to be a ‘basic human right’.84 There are compelling arguments for ensuring a system of open justice, where court proceedings
77 Dagenais v Canadian Broadcasting Corp (1994) 120 DLR 12, 40–41; New South Wales Law Reform Commission, Contempt by Publication, Report 100 Summary (2003) 12–13. 78 R v Macfarlane; Ex parte O’Flanagan (1923) 32 CLR 518, 541. 79 The sub judice rule has traditionally applied to media organisations and professional journalists, but in the age of social media it applies to any person who publishes information about a trial, including by posting on social media. For further discussion of ‘publication and the internet’, see Des Butler and Sharon Rodrick, Australian Media Law (Thomson Reuters (Professional) Australia Limited, 4th ed, 2012) 315–318. 80 Michel Foucault, Discipline & Punish: The Birth of the Prison (Random House, 1975) 48. 81 Lawrence M Friedman, The Big Trial: Law as Public Spectacle (University Press of Kansas, 2015) 137–159 (‘The Big Trial’). 82 Ibid. 83 David Eady and ATH Smith, Arlidge, Eady & Smith on Contempt (Sweet & Maxwell, Thomson Reuters, 4th ed, 2011) 96. 84 Commonwealth of Australia Constitution Act xii.
Twitter trials and Facebook juries 22 are transparent and open to the public85 in a way that supports the ‘fundamental importance that justice should not only be done, but should manifestly and undoubtedly be seen to be done.’86 The benefits of the media (or others) publishing information about trials are that it may:
maximise the chances of individuals with relevant information hearing
about a case and coming forward with new information;
prevent perjury by placing witnesses under public scrutiny;
prevent state and / or court wrongdoing by placing the criminal justice
process under public scrutiny;
reduce crime through the public expression of disapproval for crime; and
promote the public discussion of important issues.87
When media publications include prejudicial information, however, they can undermine the right to a fair trial — a ‘right’ to which every accused is entitled.88 The
‘right to a fair trial’ is not explicitly set out in Australian legislation at a national level, although it is enshrined at a State level in both Victoria89 and the Australian Capital
Territory.90 It is also set out in Article 14 of the International Covenant on Civil and
Political Rights91 and Article 6 of the Convention for the Protection of Human Rights
85 Jacqueline Horan, Juries in the 21st Century (Federation Press, 2012) 173; Burd and Horan (n 74) 103. 86 Joseph Jaconelli, Open Justice a Critique of the Public Trial (Oxford University Press, 2002) 31; R v Sussex Justices, ex parte McCarthy (1924) 1 KB 256, 259. 87 Dagenais v Canadian Broadcasting Corp (n 77) 40–41. 88 Hinch v Attorney General (Vic) (n 1) 58–59. 89 Charter of Human Rights and Responsibilities Act 2006 (Vic) s 24. In this Act, s 24 is titled ‘Fair Hearing’. Section 24(1) provides that: ‘A person charged with a criminal offence or a party to a civil proceeding has the right to have the charge or proceeding decided by a competent, independent and impartial court or tribunal after a fair and public hearing’. 90 Human Rights Act 2004 s 21. In this Act, s 21 is titled ‘Fair Trial’. Section 21(1) provides that: ‘Everyone has the right to have criminal charges, and rights and obligations recognised by law, decided by a competent, independent and impartial court or tribunal after a fair and public hearing’. 91 International Covenant on Civil and Political Rights 1966 (adopted and opened for signature 16 December 1966,) 2200A (XXI) (entered into force 23 March 1976).
23 Literature Review and Fundamental Freedoms.92 While the right to a fair trial is not explicitly provided for nationally, or in Queensland, that it exists is fundamental to the criminal justice system:
That such a right exists as a personal right seems to me so deeply rooted in
our system of law and so elementary as to need no authority to support it. It is
a right which inheres in every system of law that makes any pretension to
civilization.93
The aspects of a right to a fair trial that are of most importance for this research include that: ‘everyone shall be entitled to a fair and public hearing by a competent, independent and impartial tribunal established by law [emphasis added]’;94 and
‘everyone charged with a criminal offence shall have the right to be presumed innocent until proved guilty according to law’.95 Both of these entitlements — the right to an impartial tribunal (or jury) and to the presumption of innocence — have the potential to be compromised if members of the jury are exposed to and influenced by prejudicial publications,96 including those posted by everyday users on social media.
Despite the fact that the right to a fair trial is not explicitly enshrined in legislation Australia-wide, the right to a jury trial is expressly provided for by statute.
At a Commonwealth level, the right to a ‘trial by jury’ is set out in s 80 of the
Australian Constitution.97 It is important to note, however, that this constitutional right
92 Convention for the Protection of Human Rights and Fundamental Freedoms (opened for signature 4 November 1950) entered into force 3 September 1953 in accordance with article 66. 93 R v Macfarlane; Ex parte O’Flanagan (n 78) 541; Robert Marsden Hope, ‘A Constitutional Right to a Fair Trial? Implications for the Reform of the Australian Criminal Justice System’ (1996) 24 F L Rev 173. 94 International Covenant on Civil and Political Rights 1966 (n 91) Article 14. 95 Convention for the Protection of Human Rights and Fundamental Freedoms (n 92) Article 6. 96 Peter Gregory, Court Reporting in Australia (Cambridge University Press, 2005) 43. 97 Commonwealth of Australia Constitution Act (n 84) s 80.
Twitter trials and Facebook juries 24 only applies to persons charged with indictable Commonwealth offences.98 For state offences, a similar entitlement is enshrined in legislation in Queensland99 and all other states of Australia.100
The conflicting rights and principles surrounding free speech and fair trials, have traditionally placed pressure on media organisations and professional journalists to carefully discriminate between what they can and cannot publish. Within the boundaries of the sub judice rule, the media is permitted to publish factual information about ongoing criminal matters.101 After a defendant has been charged, a report about the matter must be limited to the ‘bare facts of the crime’ and must not include information that might identify the accused (where identity is in issue) or cause a jury to be prejudiced against them.102 ‘Bare facts’ include ‘extrinsic ascertained facts to which any eyewitness could bear testimony, such as the finding of a body and its condition, the place in which it is found, the persons by whom it was found, the arrest of a person accused, and so on’.103 Information such as that about an accused’s criminal history, witness accounts, or the use of weighted terminology such as ‘notorious prisoner’ or ‘underworld hitman’ are to be avoided.104 Once a trial commences, any reports of judicial proceedings must be fair and accurate,105 and made in good faith.106
98 Clifford L Pannam, ‘Trial by Jury and Section 80 of the Australian Constitution’ (1968) 6(1) The Sydney Law Review 1. 99 Criminal Code Act 1899 (Qld) s 604. 100 Criminal Procedure Act 1986 (NSW) s 131; Juries Act 1927 (SA) s 6; Criminal Procedure Act 2004 (WA) Div 6; Juries Act 1962 (NT) s 6; JM Bennett, ‘The Establishment of Jury Trial in New South Wales’ (1961) 13 Sydney Law Review 463; Criminal Code Act 1924 (Tas) s 361; Crimes Act 1900 (ACT) s 281. 101 Anthony Gray, ‘Contempt and the Australian Constitution – Part I’ (2017) 27(3) Journal of Judicial Administration 1, 10. 102 Pearson and Polden, above n 26, 138. 103 Packer v Peacock (1912) 13 CLR 577, 588 (The Court). 104 Gregory (n 96) 47. 105 Brook v Evans (1860) 29 LJ Ch 616; New South Wales Law Reform Commission (n 77) 12–13; Australian Law Reform Commission (n 76) 97. 106 Ex parte Terrill; Re Consolidated Press Ltd (1937) 37 SR (NSW) 255, 257–9 (Jordan CJ); A-G (NSW) v John Fairfax & Sons Ltd (1985) 6 NSWLR 695, 714 (McHugh JA).
25 Literature Review
The test to determine whether a published account of judicial proceedings is fair and accurate is that the report must be ‘one which a person of ordinary intelligence using reasonable care might reasonably regard as giving a fair summary of the proceedings’.107 A report will not be made in good faith if it ‘was published because of a belief as to its publicity value’ or ‘its value in attracting sales’, particularly where it is in ‘complete and serious disregard of its consequences … on the trial of the accused’.108
Traditionally, these parameters around publishing trial-related information were well understood by journalists and could be applied in a relatively straightforward way.109 This does not mean sub judice contempt was non-existent in traditional media, but it was uncommon as it was well regulated. However, in the age of social media — when any person can be a publisher — the sub judice rule is very difficult to enforce.110
2.3 Sub judice rule
The law prohibiting contempt by words dates back to 14th Century England, when it was designed, in part, to prevent the use of words to abuse a party to a suit.111
By 1742, it was the view of the courts that there was nothing ‘of more pernicious consequence, than to prejudice the minds of the public against persons concerned as parties in causes, before the cause is finally heard’.112 These notions form the basis of the sub judice rule, the principle purpose of which is to ‘prevent interference with the administration of justice’.113 The modern evolution of the doctrine, in the pre-internet
107 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259; Attorney-General (Qld) v WIN Television (Qld) Pty Ltd (n 61) [9]. 108 R v Scott and Downland Publications Ltd [1972] VR 663, 675. 109 Horan (n 85) 176–178. 110 Mhiripiri A Nhamo and Chari Tendai, Media Law, Ethics, and Policy in the Digital Age (IGI Global, 2017) xix; Bell (n 1) 311. 111 Eady and Smith (n 83) 3. 112 The St James’s Evening Post Case [1742] 2 Atk at 469. 113 Attorney General v Times Newspapers Ltd (1973) 3 ER 54, at 60.
Twitter trials and Facebook juries 26 era, was aimed at preventing professional journalists and media outlets from publishing information that might prejudice a matter that is sub judice,114 that is while it is ‘under judicial consideration’.115 In particular, it sought to protect parties from
‘scurrilous abuse’ and to punish those responsible for ‘abusing parties’ or ‘prejudicing mankind’ against them.116
One of the primary objectives of the sub judice rule is to protect ‘criminal trials from being contaminated by irrelevant and inadmissible material about the defendant’,117 while the matter is before the courts.118 More specifically, it prohibits the publication of material that has ‘a real tendency to prejudice or interfere with, or which is intended to prejudice or interfere with, particular pending criminal … proceedings’ (emphasis added).119 It has similarly been described as prohibiting publications that have ‘as a matter of practical reality, a tendency to interfere with the due course of justice’.120 When applying the sub judice rule, it is a matter for the courts to balance the competing rights or principles outlined earlier:
114 New South Wales Law Reform Commission, Contempt by Publication - Discussion Paper 43 (2000) 8–10; New South Wales Law Reform Commission (n 77) 8–9. 115 Gregory (n 96) 44. 116 Attorney General v Times Newspapers Ltd (n 115) at 60. 117 Horan (n 85) 176. 118 Sub judice contempt prohibits the publication of material that has a ‘real and definite tendency … to prejudice … particular legal proceedings’: see, New South Wales Law Reform Commission, Contempt Discussion Paper, above n 6, 106 [4.3]. It need not be shown that a juror has actually been influenced by published material, only that there is a sufficient risk of influence to juror impartiality: see Bell v Stewart (1920) 28 CLR 419, 432 (Isaacs and Rich JJ); A-G (NSW) v John Fairfax & Sons Ltd (1980) 1 NSWLR 362, 368 (The Court); R v West Australian Newspapers Ltd; Ex parte DPP (WA) (1996) 16 WAR 518, 531 (The Court). Sub judice contempt prohibits the publication of material that has ‘a real tendency to prejudice or interfere with, or which is intended to prejudice or interfere with, particular pending criminal … proceedings’ (emphasis added). See R v Payne (1896) 1 QB 577, 580; R v Daily Mirror (1927) 1 KB 845, 848. It need not be shown that a juror has actually been influenced by published material, only that there is a sufficient risk of influence to juror impartiality. See Bell v Stewart (1920) 28 CLR 419, 432; Attorney-General (NSW) v John Fairfax and Sons Ltd (1980) 1 NSWLR 362, 368; R v West Australian Newspapers Ltd; Ex parte Director of Public Prosecutions (WA) (1996) 16 WAR 518. 119 R v Payne (1896) 1 QB 577, 580; R v Daily Mirror (1927) 1 KB 845, 848. 120 Ex parte Auld; Re Consolidated Press Ltd (n 71) 597.
27 Literature Review
There can be no doubt as to the vital importance of freedom of speech in a free
society … However, it is not the only public interest in issue … The matter
comes down to a balancing of public interests, that is, on the one hand, the
public interest in the integrity of the criminal law and in preserving public
confidence in the fairness of criminal trials … and, on the other, freedom of
speech and discussion of matters of public interest. In this regard it is to be
stressed that the law of contempt does not exist to protect the private interests
of an accused person, but to protect the public interest in the due
administration of justice.121
When a balance must be struck between the administration of justice and freedom of speech, the common law has traditionally favoured the former.122 After all, the purpose of the law is ‘not to prevent publication of … [prejudicial] material but to postpone it’.123
Before the internet, the sub judice rule was largely effective in regulating prejudicial media coverage of trials in the traditional media — including newspapers, radio and television — when ‘authorship was local and easily identifiable’.124 The severe penalties for breaching the rule — including large fines for editors, significant penalties for publishers and news companies, and even imprisonment in serious cases125 — appear to have served as an incentive for the traditional media to comply with the law. As media organisations incorporated systematic controls into the
121 R v West Australian Newspapers Ltd; Ex parte Director of Public Prosecutions (WA) (1996) 16 WAR 518, 538. 122 Law Reform Commission of Western Australia, ‘Discussion Paper on Contempt by Publication’ (n 72) 2–3; Jonathan M Gill, ‘Sub Judice Contempt Regarding Publications: An Evaluation and Proposals for Reform with Particular Attention to Australia’s International Law Obligations / Jonathan M. Gill’ (1992)
Twitter trials and Facebook juries 28 publication process, including editorial management and extensive fact-checking,126 the high profile publication of contemptuous material in mainstream media appeared to be fairly uncommon and this may be reflected by the relatively small number of prosecutions for contemptuous publications. These legal obligations were also incorporated into journalistic training and codes of practice, establishing them as ‘best practice’ in the industry.127
In the past, the sub judice rule was applied to influential people and firms — typically journalists, editors, producers and proprietors of media organisations — that were responsible for prominent publications or those that had significant impact.128 By imposing liability on those responsible for publications, the law protected the trial process from prejudice,129 while also preserving the principle of free speech
(media).130 This had the effect of protecting the defendant’s right to a fair trial, but also maintaining the principle of open justice by permitting the publication of trial-related information that is of interest to the public.131 The fact that social media platforms do not exercise the same degree of editorial control over user posts that traditional media exercises over journalists makes it difficult to prevent prejudicial publicity from being
126 Anne Wallace et al, ‘Courts and Social Media: Opportunities and Challenges?’ (2013) 40(7) Brief 36. 127 For example, the Media, Entertainment and Arts Alliance hosts a journalist training course, “Media Law and You”, which includes media law training on contempt of court. See http://www.alliance.org.au/media-law-and-you. See also Jane B Singer, ‘Who Are These Guys? The Online Challenge to the Notion of Journalistic Professionalism’ (2003) 4(2) Journalism 139, 150 (‘Who Are These Guys?’). 128 New South Wales Law Reform Commission (n 77) 8; New South Wales Law Reform Commission (n 116) 9. 129 New South Wales Law Reform Commission (n 77) 8; Horan (n 85) 174. 130 Anthony Whealy, ‘Contempt: Some Contemporary Thoughts’ (2008) 8(4) Judicial Review 441, 448. 131 R v Karakaya (2005) 5 Cr App R [24]-[25].
29 Literature Review published on the internet. As social media users are not constrained in the same way, their posts may be more likely to be prejudicial.132
While media institutions have typically been deterred from publishing prejudicial material by the ‘threat’ of being prosecuted for sub judice contempt, in recent years some judges have become less confident in the effectiveness of the law.133
As a result, some courts are turning to pre-emptive measures such as suppression orders (also known as non-publication orders)134 — which prohibit the publication of particular information.135 Not only are courts less confident in the effectiveness of sub judice contempt, but prosecutions against the media for contempt have significantly declined.136 As a result, courts and parties are turning to alternatives not only including suppression orders but also to applications for orders to: adjourn proceedings to a later time when the publicity has faded;137 admonish or otherwise direct the jury to disregard prejudicial publicity;138 change the venue to a different location where members of the jury pool may be less likely to have been exposed to prejudicial publicity;139 question individual persons who have been selected to serve on a trial;140 order a permanent
132Tom Griffith, Katie Kossian and Tania Kowalczuk, ‘Twitter, Suppression and the Courts’ (2012) 15(3) Internet Law Bulletin 42. 133 Jason Bosland, ‘Restraining “Extraneous” Prejudicial Publicity: Victoria and New South Wales Compared’ (2018) 41(4) University of New South Wales Law Journal 1263, 1266 (‘Restraining “Extraneous” Prejudicial Publicity’). 134 Legislation New South Wales Government Attorney General’s Department Policy and Criminal Law Review Division, Report on Access to Court Information (June 2008) 38
Twitter trials and Facebook juries 30 stay of proceedings where it is deemed impossible for the defendant to ever receive a fair trial;141 discharge the jury; or to quash convictions on appeal.142 If a defendant is aware of prejudicial publicity at the commencement of the trial, they may also apply for the trial to be heard by a judge alone, without a jury.143 However, in the age of social media, the effects of some of these alternatives are also limited. For example, adjourning proceedings until a later time may not be as effective in ensuring a fair trial as it was once due to ‘the immediate accessibility of information on the internet with an efficiency that overrides the practical obscurity of the past’.144
Of all these alternatives, suppression orders provide a particularly useful means of comparison with the sub judice rule, as they seek to ‘restrict publicity that may prejudice a fair trial or the administration of justice in general’.145 They also face a number of similar challenges in terms of enforcement in the age of social media.146
2.4 Age of social media
There are a number of challenges for the sub judice rule in the age of social media.147 The way users consume and participate in conversations about news is
141 The successful application in R v Liddy [2010] SADC 80 may be compared with the unsuccessful application in Dupas v The Queen (2010) 241 CLR 237. 142 New South Wales Law Reform Commission, Contempt by Publication, Report 100 Summary (2003) [2.69]–[2.87] (‘Contempt Report Summary’); Michael Chesterman, ‘OJ and the Dingo: How Media Publicity Relating to Criminal Cases Tried by Jury Is Dealt with in Australia and America’ (1997) 45 American Journal of Comparative Law 109, 122–3, 130. 143 Criminal Code Act 1899 (Qld) ss 614–615E. The right to apply for a no jury order was added to the Criminal Code Act 1899 (Qld) in 2008: Criminal Code and Jury and Another Act Amendment Act 2008 (Qld). See also, Supreme Court Act 1933 (ACT) s 68B; Criminal Procedure Act 1986 (NSW) ss 132–132A; Juries Act 1927 (SA) s 7; Criminal Procedure Act 2004 (WA) pt 4 div 7. See also Rebecca McEwen, John Eldridge and David Caruso, ‘Differential or Deferential to Media? The Effect of Prejudicial Publicity on Judge or Jury’ (2018) 22(2) The International Journal of Evidence & Proof 124 (‘Differential or Deferential to Media?’). 144 John Fairfax Publications Pty Ltd v District Court (NSW) (n 1) 360. 145 New South Wales Law Reform Commission (n 116) 31 [91]. 146 Mark Pearson and Caroline Graham, ‘Suppression Orders: Reskilling Journalists and the Judiciary’ (2010) 32(1) Australian Journalism Review 97, 99. 147 Elizabeth Greene and Jodie O’Leary, ‘Ensuring a Fair Trial for an Accused in a Digital Era: Lessons for Australia’ in Patrick Keyzer, Jane Johnston and Mark Pearson (eds), The Courts and the Media - Challenges in the Era of Digital and Social Media (Halstead Press, 2007) 101, 101–103.
31 Literature Review completely different from the age of traditional media. This includes the consumption of news about criminal trials and online participation in conversations about them.
These changes make the law difficult to enforce, particularly against everyday social media users. Also contributing to these difficulties is the fact that social media companies, like Twitter and Facebook, are become increasingly powerful, while at the same time the influence of traditional media organisations is diminishing.
2.4.1 News consumption and participation Online news consumption and participation in the age of social media has significantly changed the way people gather and share news content,148 and the frequency with which they access news.149 Greater access to the internet150 and the growth of social media151 mean that news consumption is no longer limited to that distributed in the morning newspaper or on the evening news. Below is a valuable and comprehensive explanation as to how new media is different from traditional media, and this is important when considering the effectiveness of the sub judice rule in the age of social media:
What sets this new media apart from traditional media is that it is more
accessible by users, allows for more interactivity, and allows for creative
participation and feedback. The new media is accessible by computers,
148 Axel Bruns and Tim Highfield, ‘Blogs, Twitter, and Breaking News: The Produsage of Citizen Journalism’ in In Rebecca Ann Lind (Ed), Produsing Theory in a Digital World: The Intersection of Audiences and Production (Peter Lang, 2012) 15. 149 56% of respondents reported accessing news more than once per day and this figure is increasing over time. See Sora Park et al, Digital News Report: Australia 2018 (News and Media Research Centre, University of Canberra, 2018) 1, 51
Twitter trials and Facebook juries 32
laptops, tablets, smartphones, and smart televisions in a wide range of areas,
including at home, at work, and while in transit. The key element of new media
is that connection and conversation are created by the interest and efforts of
participants. This stands in contrast to traditional media that are defined as
being created by single producers in one form with the intention that they be
passively consumed by audiences. New media use also varies in that the users
have much more control of the media experience. They can select from print,
audio, or visual elements, depending on their individual preferences, and can
contribute to the media as well in their choice of print comments, audio
selections, or video uploads. New media also includes non-traditional
mediums of individual communication, including text messaging, blogs, video
games, and online newspapers. All of these forms allow users to direct their
own experiences by selecting what they read, hear, and see and by enabling
them to actively contribute.152
The majority of Australians now use digital devices to access news153 — most commonly, a smartphone154 — and, increasingly, their primary source of news is from digital sources (such as online newspapers and news sites, social media sites, and news aggregators).155 For the first time, in 2018, ‘the number of participants using digital
152 ‘New Media Analysis’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) 1091
33 Literature Review platforms (websites, apps, social media, blogs) as a news source surpassed those accessing traditional platforms (TV, radio and print)’.156 Many consumers have shifted away from traditional news sources altogether, instead relying on social media platforms, such as Facebook, YouTube and Twitter,157 as their ‘one main source of news’.158 Of particular interest to my research is that the most popular social media platform for news is Facebook — used by 41% of all respondents as a source of news
— and the third most popular is Twitter — used by 10% of all respondents.159 The use of social media as a source of news was particularly high in the youth category (71%), with 38% nominating social media as their main source of news.160 When asked about the next three to five years, many believed they would ‘gain most value from ‘fast’ real-time, accurate news’ and would ‘get the most value from using social media to keep up-to-date on breaking news’.161
As social media platforms — including Facebook and Twitter — continue to grow as both a source and the main source of news for consumers, and as consumers continue to access news more frequently, it may be there is an increasing risk of exposure to potentially prejudicial publicity about criminal trials online.
156 Park et al (n 152) 60. 157 Ibid 12. 158 17% of respondents reported using social media as their “main source of news” (17%) and 29% reported using online news as their “main source of news”. See Park et al (n 152) 51. 159 Ibid 55. 160About 71% of 18 to 24-year olds (millennial age-group) use social media for news, and about 38% of this age group nominate social media as their main source of news compared to older participants who nominated TV news. See Ibid 60. 16146% of all respondents believed that in the next three to five years, they would “gain most value from ‘fast’ real-time, accurate news” and 35% that they would “get the most value from using social media to keep up-to-date on breaking news”. See Alcorn and Buchanan (n 46) 26–27.
Twitter trials and Facebook juries 34
2.4.2 Experiencing criminal trials via social media Social media has established itself ‘at the heart of everything [consumers] do’,162 meaning there have been significant shifts in the way consumers access and participate in news,163 including news about criminal trials. Today, everyday people experience news and court processes synchronously, in real-time, often collaboratively, and they have greater access to details and evidence from trials. It is possible this online information is more influential than that published in traditional media.
The fact that consumers can access news about criminal trials synchronously, in real-time, means the relationship between audiences and influential mainstream publications has become two-way in a much more pronounced way than, say, letters to the editor. For high-profile trials, users can engage in conversations about the trial with larger audiences. Conversations that may once have taken place in person with just a few people may now be visible to thousands. Users’ social media posts — including those that are prejudicial — can form part of highly visible conversations with a potentially global public and they can spread rapidly.
Today, everyday people also have greater access to details and evidence from trials. The immediacy of networks like Twitter provides a window into criminal courtrooms previously only accessible to those physically present.164 Social media
162 Niki Alcorn et al, Media Consumer Survey 2015: Australia Media and Digital Preferences - 4th Edition (Deloitte, 2015) 2, 7. 163While 54% of consumers still discuss news face-to-face, many are active sharers and participants of online news discussion, with 30% talking online to friends and colleagues about a news story, 21% sharing a news story via social networks, 19% commenting on a news story on social networks, 16% rating, liking or favouriting a news story and 16% sharing a news story via email. See Watkins et al (n 157) 18, 35; Dominic L Lasorsa, Seth C Lewis and Avery E Holton, ‘Normalizing Twitter: Journalism Practice in an Emerging Communication Space’ (2012) 13(1) Journalism studies 19, 21– 23 (‘Normalizing Twitter’); Alfred Hermida, ‘Twittering the News: The Emergence of Ambient Journalism’ (2010) 4(3) Journalism practice 297, 299–301 (‘Twittering the News’). 164 In Queensland, accredited media in a courtroom are permitted to use electronic devices to report on proceedings in real-time using text-based communication (including social media), provided they do not interrupt the court: see Supreme Court of Queensland, Practice Direction No 8 – Electronic Devices in Courtrooms, 17 February 2014, [8]–[9].
35 Literature Review provides a way for people involved in news events to ‘share information directly from the scene’,165 and real-time citizen journalism gets picked up by professionals and concerned citizens alike. It also enables users to disseminate a greater level of detail than generally seen in mainstream media, where time and space restrictions would make such extensive reporting impracticable. In criminal cases of high public interest, users will often transcribe and tweet large sections of oral testimony from witnesses, counsels’ arguments, and instructions from judges. This means viewers can now track developments in near real-time, comment on the story (or on other people’s comments), share additional information, and evaluate what is known.166
Information consumed on social media may also be more influential than that published in traditional media. The fact that social media users can actively share and participate in online news discussions167 may strengthen and transform the influence of published material in ways that were not possible before the internet. Information posted by people they know — family or friends — may also influence users in new ways. In contrast to traditional media publications, there are also concerns that social media coverage of criminal proceedings may tend to be emotionally-orientated, at times indicating guilt before trial, and may include prejudicial photographs, footage
165 Folker Hanusch and Axel Bruns, ‘Journalistic Branding on Twitter: A Representative Study of Australian Journalists’ Profile Descriptions’ (2017) 5 Digital Journalism 26, 27. 166 Axel Bruns, ‘Working the Story: News Curation in Social Media as a Second Wave of Citizen Journalism’ in Chris Atton (ed), The Routledge Companion to Alternative and Community Media (Routledge, 2015) 379; Axel Bruns and Tim Highfield, ‘Blogs, Twitter, and Breaking News: The Produsage of Citizen Journalism’ in Rebecca Ann Lind (ed), Produsing Theory in a Digital World: The Intersection of Audiences and Production in Contemporary Theory (Peter Lang, 2012) 15. 167While 38.9 per cent of consumers still discuss news face-to-face, many are active sharers and participants of online news discussion, with 14.6 per cent talking online to friends and colleagues about a news story, 21.5 per cent sharing a news story via social networks, 16.6 per cent commenting on a news story on social networks, 12.5 per cent rating, liking or favouriting a news story and 11.2 per cent sharing a news story via email: see Jerry Watkins et al, ‘Digital News Report: Australia 2016’ (Report, News & Media Research Centre, University of Canberra, Reuters Institute for the Study of Journalism, University of Oxford, 2016) 11–12.
Twitter trials and Facebook juries 36 and emotive statements designed to increase levels of sentiment and sympathy in the community.168
2.4.3 Law not suited to social media At law, the sub judice rule applies to social media,169 but it does not translate well to the online environment.170 This makes it difficult to regulate information about high-profile criminal trials posted on social media by everyday users as independent publishers. The large number of decentralised users (corporate and individual), together with the fact that users may maintain anonymity through the use of pseudonyms,171 means it is can be difficult to identify those who have published prejudicial material.172 Facebook and Twitter will comply with discovery orders, including those to identify users, but this approach is not generally pursued in the context of sub judice contempt.173 It can also be difficult to identify the time or place of publication, and users’ geographical locations may cross jurisdictional boundaries where Australian laws do not apply. Social media is also continually evolving and this is very different from the type of media the law has traditionally regulated:
New media represents a continuous flow of information, which is very
different from the traditional media process of defined points of creation and
publishing practices that provide clear start and end points. New media
continually evolves as older or unpopular forms become discontinued or
changed as users add commentary or links to additional information … [There
168 Isaac Frawley Buckley, ‘Pre-Trial Publicity, Social Media and the “Fair Trial”: Protecting Impartiality in the Queensland Criminal Justice System’ (2013) 33 The Queensland Lawyer 38, 42. 169 R v Hinch [2013] VSC 520.: For the purposes of sub judice contempt, Tweets and Facebooks comments will constitute a ‘publication’ at law. 170 Horan (n 85) 177. 171 Tom Griffith, Katie Kossian and Tania Kowalczuk, ‘Twitter, Suppression and the Courts’ (2012) 15(3) Internet Law Bulletin 42. 172 Wallace et al (n 128) 37. 173 In Chapter 7, I discuss how this approach might work in practice.
37 Literature Review
is] an ever-changing feed of comments, pictures, and news from a few or very
many contributors from a wide range of age and location.174
Because it is so difficult to enforce restrictions on published content, the law has not regularly been enforced against ordinary consumers of media who might make prejudicial comments to relatively small audiences.175 In practice, this means social media users are not constrained in the same way as mainstream publishers.176 It is also the case that many social media users may not even be aware of the sub judice rule, so its existence is unlikely to provide any deterrence.177 Due to these various factors, social media users may be more likely than their more legally-aware traditional media counterparts to post information that is unfair or lacks accuracy,178 or that is prejudicial or contemptuous.179
2.4.4 Internet jurisdiction and the regulation of social media Changes in the communications landscape, caused particularly by the internet, mean that the role of institutional media organisations in ‘promoting and protecting the principle of open justice ... [and] safeguarding the public interest in transparency in the courts’ is declining.180 In the age of social media, the influence of traditional media organisations is diminishing, while social media companies, like Twitter and
Facebook, become increasingly powerful.181 These companies are wielding greater
174 ‘New Media Analysis’ (n 155) 1092. 175 Horan (n 85) 177. 176 Mark Pearson, ‘Why the Public Isn’t Allowed to Know Specifics about the George Pell Case’, The Courier; Ballarat, Vic. (Ballarat, Vic., Australia, Ballarat, Vic., 27 March 2018) 0. 177 New South Wales Law Reform Commission (n 116) 147 [4.88]. 178 Jane Johnston et al, Juries and Social Media: A Report Prepared for the Victorian Department of Justice (Australia, Standing Council on Law and Justice, 2013) 1, 4. 179 Griffith, Kossian and Kowalczuk (n 174). 180 Jason John Bosland and Judith Townend, ‘Open Justice, Transparency and the Media: Representing the Public Interest in the Physical and Virtual Courtroom’ (2018) 23(4) Communications Law 183, 183,190 (‘Open Justice, Transparency and the Media’). 181 ‘Are Tech Companies Becoming the Primary Legislators in International Cyberspace? - Lawfare’ 28 March 2019
Twitter trials and Facebook juries 38 control over news distribution than ever before, as they seize control over who
‘publishes what to whom’ and ‘how that publication is monetized’.182 On the one hand traditional publishers are losing control of news distribution,183 while on the other it feels like ‘Facebook is eating the internet’184 or even ‘the world’.185 Social media platforms set the rules for how we participate online — they ‘make decisions, in their rules and in their technical design that shape the kinds of content that people can post and the kinds of content made visible’.186 The power to make these decisions has a significant effect on how people communicate and connect in their daily lives — a power that is exercised by just a small number of major corporations.187 This small number of corporations govern in a ‘lawless’ way that does not meet acceptable standards of legitimacy — they are not currently required to meet any particular standard of accountability, predictability, or fairness.188
That it is difficult to apply the sub judice rule to these increasingly powerful social media companies must be cause for concern. Not only is the law itself not well suited to the online environment, but these massive companies are incorporated outside
Australia and immunised from legal penalty by US law.189 Even if the law could be applied effectively,190 it can be difficult to determine whether it is only users who post
182 Emily Bell, ‘Facebook Is Eating the World’ on Columbia Journalism Review (7 March 2016)
39 Literature Review prejudicial content online who should be pursued for contempt, or whether liability should also extend to internet service providers, internet intermediaries, or other key points of influence.191 More work is needed though to understand what approaches to regulation by state or non-state actors are best suited for the digital age. It may be that social media platforms ought to bear some responsibility for prejudicial posts, but exactly what that should look like in practice is still an open question.
It is ultimately a matter of jurisdiction that makes it so difficult to hold any of these private companies accountable at law.192 The traditional notion of linking laws to territoriality193 — where laws are predicated on ‘the existence of traditional borders that define jurisdictional boundaries’194 — is not well suited to the online environment.195 While local trials may be affected by publications that can be read anywhere on social media, the private companies that provide the framework for these publications are located and incorporated in countries all over the world. To demonstrate the extraordinary challenge of internet governance, it is useful to consider the complexity of different laws and legal systems that apply to the simple act of posting something on social media:
In most instances … you will have to take account of the law of the state you
are in at the time you make the posting ... You may also need to consider the
law of the state in which you are habitually residing (and/or domicile) and the
judgements have successfully been applied to Google: see, eg, Google Inc v Duffy [2017] SASCFC 130; Trkulja v Google Inc [No 5] [2012] VSC 533. 191 Horan (n 85) 177; Eady and Smith (n 83) vi; Bell (n 1). 192 Dan Jerker B Svantesson, Solving the Internet Jurisdiction Puzzle (Oxford University Press, 2017). 193 Stephan Wilske and Teresa Schiller, ‘International Jurisdiction in Cyberspace: Which States May Regulate the Internet’ (1997) 50 Fed. Comm. LJ 117, 129 (‘International Jurisdiction in Cyberspace’). 194 Kevin A Meehan, ‘The Continuing Conundrum of International Internet Jurisdiction Note’ (2008) 31 Boston College International and Comparative Law Review 345, 347; Dan Jerker B Svantesson, Private International Law and the Internet (Wolters Kluwer Law & Business, 3rd ed, 2016). 195 Svantesson (n 195) 107; Joel R Reidenberg, ‘Technology and Internet Jurisdiction’ (2005) 153(6) University of Pennsylvania Law Review 1951, 1951.
Twitter trials and Facebook juries 40
law of your state of citizenship. Then you will probably also need to consider
US law, as most major social media platforms are based in the US … If your
posting relates to another person, you may also need to consider the laws of
that person’s location, residence, domicile, and citizenship … [or the location]
where content is downloaded or read … you also need to take account of the
laws of all the states in which re-posted versions of your posting may be
downloaded or read.196
These types of complexities, combined with inconsistencies and a lack of interoperability between different domestic and international legal systems frequently render our courts powerless.197 There is a need to achieve an increased level of ‘legal interoperability among multiple jurisdictions’ as this will lead to ‘better laws that foster the development of fundamental values and rights, such as information privacy and freedom of expression’.198
2.5 Juries
To better understand how prejudicial conversations on social media may affect jury decision-making, it is helpful to consider the jury literature. In particular, I draw on literature that considers juror exposure to prejudicial media, prejudicial publicity on social media and juror impartiality, and efforts that have been made to shield jurors from prejudice. My research does not allow me to draw conclusions about juror behaviour, or whether jurors are actually influenced by prejudicial publicity on social media, but it does provide another perspective that complements the existing literature.
196 Svantesson (n 191) 106; Dan Svantesson, ‘A New Jurisprudential Framework for Jurisdiction: Beyond the Harvard Draft’ (2015) 109 AJIL Unbound 69, 72–73. 197 Svantesson (n 195) 113, 115. 198 J Palfrey, Interop the Promise and Perils of Highly Interconnected Systems (Basic Books, 2012) 178.
41 Literature Review
First, it is important to understand what is expected from jurors. Juror impartiality is an underlying tenet of the ‘right to a fair trial’199 — a defendant’s right to a verdict based only upon the evidence produced before a jury in court.200 At law, juries are required to undertake their role competently, independently and impartially.201 Impartiality is the attribute most at risk if jurors are exposed to and influenced by prejudicial publicity, or other information about a trial that might influence their decision-making, other than the evidence presented in court.202 A juror will not be impartial when ‘prior to the trial or during its currency, [s/he] comes into the possession of special knowledge about the crime or the accused which has the potential to prompt the formation of a belief in guilt or innocence that may not adequately be laid aside’.203 Jurors need not be oblivious to the trial on which they are empanelled, but if they do have any pre-existing knowledge, or knowledge gathered during the trial, they must be able to put it aside and ‘perform their function in a fair- minded and unbiased manner’.204 In 2010, the High Court of Australia held that juror prejudice could be laid aside through the use of proper jury directions.205 This decision was made at a time when social media was in its infancy and it is not clear whether today’s social media ‘saturation’ may have an effect on the High Court’s position. It is important to note those comments were made in the context of an application for a
199 Dietrich v R (1992) 177 CLR 292, 326; The 'right' to a fair trial is not enshrined in Australia although internationally it is recognised by Art 14 of the International Covenant on Civil and Political Rights, adopted and opened for signature 16 December 1966, 2200A (XXI) (entered into force 23 March 1976) and Art 6 of the Convention for the Protection of Human Rights and Fundamental Freedoms, opened for signature 4 November 1950, CETS 005 (entered into force 3 September 1953). It is also recognised at a state level in Victoria: Charter of Human Rights and Responsibilities Act (n 89) s 24. 200 R v Karakaya (n 133) [24]-[25]; Australian Law Reform Commission (n 76) 2–3. 201 Article 14, International Covenant on Civil and Political Rights (ICCPR) (1966) 999 UNTS 171; 6 ILM 368; [1980] ATS 23. 202 Gregory (n 96) 176–177; R v Macfarlane; Ex parte O’Flanagan (n 78) 541–2. 203 Australian Law Reform Commission (n 76) 3; R v Macfarlane; Ex parte O’Flanagan (n 78) 541–2. 204 Australian Law Reform Commission (n 76) 7; R v Macfarlane; Ex parte O’Flanagan (n 78) 541–2. 205 Dupas v The Queen [2010] HCA 20; R v Glennon (n 1).
Twitter trials and Facebook juries 42 permanent stay of proceedings due to extensive pre-trial publicity. The court held that a permanent stay of proceedings was not necessary, due to the fact that memories of media would fade over time and there would come a time when a fair trial may be possible. This position may be different to the ‘during-trial’ publications that are the focus of this work.
2.5.1 Jury exposure to prejudicial media The effects of online prejudicial publicity upon jurors have been examined during the last decade through studies of juror misconduct,206 juror use of social media,207 and juror psychology in decision-making.208 This juror-centred approach has provided much insight into the broad challenges surrounding jurors and social media.
Studies of particular interest for my research include those that show some jurors will consult information in the media during trials, and that negative media coverage can be determinative of verdicts. It also includes studies that show juror bias tends to increase with the quantity of prejudice they are exposed to and that, despite judicial instructions, many jurors are confused about whether they can discuss trials with family and friends or look at trial related information online. Some jurors also admit
206 Johnston et al (n 181) 9; Wallace et al (n 128) 37; Patrick Keyzer et al, ‘The Courts and Social Media: What Do Judges and Court Workers Think?’ (2013) 25(6) Judicial Officers Bulletin 47, 49 (‘The Courts and Social Media’); Lorana Bartels and Jessica Lee, ‘Jurors Using Social Media in Our Courts: Challenges and Responses’ (2013) 23 Journal of Judicial Administration 35; Paula Hannaford-Agor, David B Rottman and Nicole L Waters, ‘Juror and Jury Use of New Media: A Baseline Exploration’ [2011] National Center for State Courts Center for Jury Studies, United States 1; Amy J St Eve and Michael A Zuckerman, ‘Ensuring an Impartial Jury in the Age of Social Media’ (2012) 202(2) Duke L. & Tech. Rev. 1. 207 Johnston et al (n 181) 9; Wallace et al (n 128) 37; Keyzer et al (n 209) 49; Bartels and Lee (n 209); Hannaford-Agor, Rottman and Waters (n 209). 208 Hannaford-Agor, Rottman and Waters (n 209); David Tait, ‘Deliberating about Terrorism: Prejudice and Jury Verdicts in a Mock Terrorism Trial’ (2011) 44(3) Australian & New Zealand Journal of Criminology 387 (‘Deliberating about Terrorism’); Cheryl Thomas, Are Juries Fair? (Ministry of Justice Research Series, 2010)
43 Literature Review they want to go online and, as a result, complying with judicial instructions not to do so is difficult.
It has long been recognised that juror exposure to prejudicial media has the potential to interfere with the administration of justice by affecting juror decision- making.209 Regardless of judicial instructions or deliberations, juror bias has been shown to be caused by both factual publicity — that contains incriminating information about the defendant — and emotional publicity — that contains information likely to arouse negative emotions (but that does not contain explicitly incriminating information).210 Studies over the last four decades have shown jurors refer to media reports in their deliberations,211 and jurors exposed to prejudicial publicity are more likely to reach a guilty verdict.212 Even jurors who regard themselves as unbiased after being exposed to prejudicial content are more likely to reach a guilty verdict than those who have not been exposed to prejudice.213 Of those jurors who do regard themselves as biased after being exposed to prejudicial content, it is concerning to note that some choose not to disqualify themselves from service, even when given the opportunity to do so.214
In Australia, it has been shown there can be a link between the sentiment expressed in media coverage about criminal trials and jury verdicts. A New South
Wales study published in 2001 is one of the most comprehensive projects of its kind.
Researchers used a multiple-case study approach — including interviewing and
209 Australian Law Reform Commission (n 76) 2-31 (Susceptibility of Juries). 210 Geoffrey P Kramer, Norbert L Kerr and John S Carroll, ‘Pretrial Publicity, Judicial Remedies, and Jury Bias’ (1990) 14(5) Law and Human Behavior 409, 414, 426. 211 Ibid 413. 212 S Sue, RE Smith and G Pedroza, ‘Authoritarianism, Pretrial Publicity, and Awareness of Bias in Simulated Jurors’ (1975) 37(3) Psychological Reports 1299, 1299. 213 Ibid. 214 Ibid.
Twitter trials and Facebook juries 44 surveying jurors — to examine 41 criminal trials where there was significant media coverage.215 Researchers used a combination of professional assessors (who considered the evidence and reached a verdict which could be compared with the verdict of jurors) and interviews with jurors to determine whether jury verdicts were open on the evidence or appeared likely to have been influenced by publicity that was not replicated in evidence.216 They found that negative media coverage was likely to have been determinative of the verdict in 8% of trials (3 trials), possibly determinative of the verdict in another 18% of trials (7 trials), and one verdict (an acquittal) was considered ‘unsafe’ due to the likely influence of prejudicial publicity.217 A more recent Australian study also suggests that jurors’ negative attitudes and emotions
(prejudice) — including those bred from negative media coverage — contributed to guilty verdicts.218
It is not only the nature or quality of prejudice in media coverage — such as highly prejudicial content, as opposed to low-level prejudice — but also the quantity of prejudice that has the potential to affect jurors.219 Quantity can refer to how much prejudice is seen in one medium — such as the number of tweets or comments seen on social media — or how much prejudice is seen across multiple mediums — such as on social media plus on television or radio or newspapers. Jurors have demonstrated bias following exposure to prejudicial publicity in even just one medium — either print or television media — but those subjected to more than one medium — print and
215 Michael Chesterman et al, Managing Prejudicial Publicity: An Empirical Study of Criminal Jury Trials in New South Wales (Published by the Law and Justice Foundation of N.S.W. on behalf of the University of N.S.W. and the Justice Research Centre, 2001) 40 (‘Managing Prejudicial Publicity’). 216 Ibid 101. 217 Ibid 127–128. 218 Tait (n 211) 390–391. 219 Australian Law Reform Commission (n 76) 11; W Wilcox, ‘The Press, the Jury and the Behavioural Sciences’ [1968] Journalism Monographs 1, 34.
45 Literature Review television media — show the greatest bias.220 This is similar to the finding that jurors exposed to greater quantities of negative media coverage are significantly more likely to prejudge a defendant as guilty, compared to those exposed to lower quantities of media coverage.221
In addition to the effects of quality or quantity, jurors may be exposed to prejudice because of their confusion around what constitutes appropriate juror behaviour and their desire to go online. It has been found that, despite clear instructions not to undertake independent research about a case, significant numbers of jurors believe they can engage in some types of research (15%) and many do not understand whether they can do so (20%).222 Twelve percent of jurors admit to actively looking for internet reports about trials on which they are serving, with as many as 26% of jurors in high-profile cases seeing reports on the internet.223 Similarly, despite clear judicial directions not to post information about a case on social media or communicate with family or friends, some jurors believe they can do so, while others are unsure.224
A further complication for the criminal justice system is that many jurors want to use the internet during trials, for purposes such as obtaining definitions of legal terms or for finding information about the case, the parties, the lawyers, the judge or the witnesses.225 Fourteen percent of jurors in this study claimed they could not refrain
220 James RP Ogloff and Neil Vidmar, ‘The Impact of Pretrial Publicity on Jurors: A Study to Compare the Relative Effects of Television and Print Media in a Child Sex Abuse Case’ (1994) 18(5) Law and Human Behavior 507, 520 (‘The Impact of Pretrial Publicity on Jurors’). 221 Christina A Studebaker et al, ‘Assessing Pretrial Publicity Effects: Integrating Content Analytic Results’ (2000) 24(3) Law and Human Behavior 317, 323–324, 328, 334 (‘Assessing Pretrial Publicity Effects’); Wilcox (n 222) 21; Australian Law Reform Commission (n 76) 11. 222 Hannaford-Agor, Rottman and Waters (n 209) 6. 223 Thomas (n 211) 43. 224 Hannaford-Agor, Rottman and Waters (n 209) 6. 225 Jurors have been shown to want to use the internet during trials for purposes such as obtaining definitions of legal terms (44%) and finding information about the case (26%), the parties involved (23%), the lawyers (20%), the judge (19%) and the witnesses (18%). See Ibid.
Twitter trials and Facebook juries 46 from all internet usage for the duration of a trial, even if instructed to do so by the judge.226 That social media has created a one-stop shop for jurors who want to research trial details online, comment about or discuss the trial online, or contact persons involved in the case via social media must be cause for concern.227
2.5.2 Social media, prejudicial publicity and juror impartiality A number of studies of jurors and social media in recent years have raised concerns over the potential for prejudicial publicity on social media to cause juror bias and, therefore, to jeopardise the right to a fair trial.228 In a 2013 Australian study, participants — including judges, magistrates, tribunal members, court workers, court public information officers and academics working in the field of judicial administration — were asked to identify the top two most significant challenges that social media poses for the courts. Their top two responses were: (i) juror misuse of social media (and digital media) leading to aborted trials; and (ii) sub judice issues or breach of suppression orders (by tweets, Facebook or other social media).229 Jurors can freely access prejudicial information on social media, they may personally know those who have posted the content, and they may also be exposed to an accumulation of prejudicial content — all of which provide the potential for influence.
226 Even after clear judicial directions to the contrary, some jurors believe they can post information about the trial on social media or communication with family or friends (7%), while others are unsure (7%). A number of jurors (14%) claim they could not refrain from all internet usage for the duration of a trial, even if instructed to do so by the Judge. See Ibid. 227 ‘Social Media and Its Impact on Trials’, LawCPD.com.au (19 February 2014)
47 Literature Review
Information on social media is publically and easily accessible. If prejudicial information about high-profile criminal trials is posted online, it is likely a person who searches for it will find it. However, it is difficult to know whether a person (or juror) who does not search for it, might still be exposed to — and possibly influenced by — online information about criminal trials.230 Even those who are not social media users may be susceptible to the influence of social media, given social media publications
(including those by ‘ordinary’ citizens) have been shown to play a role in agenda- setting for mainstream media.231
The shift towards online news consumption has also introduced a different relationship between jurors and the publishers of trial related news. On social media, information about criminal trials — including potentially prejudicial material — may be generated by a person’s friends, colleagues, acquaintances or other influencers, and this may have the effect of being more persuasive than traditional news, that was generally written and published by those outside an individual’s acquaintance. The way users engage with news has also changed. The ability of users to create, share and respond to content, adds an interactive dimension which may strengthen and transform the potential influence of published material upon jurors in ways that were not possible in traditional media.232 This new type of news presentation may also be less neutral.
Social media coverage of criminal proceedings tends to be emotionally-orientated,233 at times indicating guilt before trial, and may include prejudicial photographs, footage
230 McEwen, Eldridge and Caruso (n 146) 126. 231 Sanja Milivojevic and Alyce McGovern, ‘The Death of Jill Meagher: Crime and Punishment on Social Media’ (2014) 3(3) International Journal for Crime, Justice and Social Democracy 22, 28. 232 Bartels and Lee (n 209) 35. 233 Buckley (n 171) 42.
Twitter trials and Facebook juries 48 and emotive statements designed to increase levels of sentiment and sympathy in the community.234
Of further concern is that large numbers of social media posts may have a collective or cumulative effect that reinforces prejudice or negative sentiment. This collective or compounding effect may mean that negative sentiment in social media more strongly reinforces prejudice than individual comments in mainstream media.235
The sub judice rule has typically been enforced against individual media organisations or professional journalists, but it is not well-suited for regulating collective prejudice amassed across large numbers of publications posted by different users.
2.5.3 Efforts to shield jurors from prejudice Although the risks of prejudicial publicity on social media are not well understood, some jurisdictions have nevertheless responded by introducing preventative measures. During the 2012 murder investigation of Gillian Meagher,236 the Victoria Police service tried to educate the public about the risks of posting prejudicial information online. They did this by posting a message on the police
Facebook page asking users to ‘refrain from posting anything on social media which could jeopardise or endanger the presumption of innocence, as this has a very high potential to interfere with the administration of justice’ and that it is ‘inappropriate to post speculation or comments about matters before the courts’.237
A more extreme approach was taken in Florida in 2013, where a decision was made to sequester jurors on a murder trial for the entire duration of proceedings (one
234 Buckley (n 171). 235 Johnston et al (n 181) 5; Horan (n 85) 177. 236 R v Bayley [2013] VSC 313. 237 ‘Fears of Trial by Social Media in Meagher Case’
49 Literature Review month) in response to concerns about juror exposure to prejudicial publicity and outside influences.238 For the duration of this month, including weekdays and weekends, jurors only had limited access to family, friends and media publications.
Australian jurors are no longer commonly sequestered,239 not even during deliberations. Instead judges provide clear directions to jurors to ignore prejudicial publicity.240 During the summing-up in the Baden-Clay murder trial, for example,
Justice Byrne repeatedly reiterated the need for jurors to reach their verdict only upon the evidence presented in court, stating they must ensure no external influences
(including social media reports) play a part in deliberations or determining what was important.241 It is difficult to know how effective these efforts to shield jurors are in practice: ‘expecting jurors to refrain from using social networks because of a jury direction may be naive in light of society’s general social media use’.242 It can also be quite difficult for jurors to avoid prejudicial publicity during a trial.243 Further empirical work on juror social media use is needed before effective safeguards can be established.244
2.6 Criminal justice
Regardless of whether jurors are exposed to online prejudice by actively — albeit inappropriately — searching online or by inadvertent exposure, juror bias comes
238 Wallace et al (n 128) 36. 239 ‘The contemporary view is that it is not necessary for the jury to be sequestered with exposure to the media thereby eliminated.’ See Skaf v R [2008] NSWCCA 303, [28]. 240‘Apart from occasions when a trial is delayed to distance it from publicity, trial judges always tell jurors to confine their consideration to the evidence in the proceedings. If a trial has previously attracted or is likely to attract publicity, Judges are careful to remind jurors that they are to ignore any publicity’. See Ibid. 241 R v Baden-Clay (2014) QSC 154, 4, 7. 242 Braun (n 53) 1661. 243 McEwen, Eldridge and Caruso (n 146) 127. 244 Braun (n 53) 1662.
Twitter trials and Facebook juries 50 at a cost to the criminal justice system.245 The effects — or potential effects — of juror exposure to prejudicial publicity must be taken seriously by courts and this takes a financial and emotional toll on society, offenders and victims. There have been a number of cases in recent years where the potential for jurors to be influenced by information on social media or online has had costly outcomes for those involved and the community.
In a recent murder trial in New South Wales, a photograph and caption posted on both Instagram and Twitter resulted in the jury being discharged and trial abandoned.246 To protect potential jurors in a retrial, Justice Bellew issued a suppression order to ensure details of the Instagram and Twitter posts remained secret as, in his view, ‘they may have an adverse effect on the right of one of the accused to a fair trial’.247 He also expressed concern at the considerable public expense and wasted time caused by having to vacate the trial.248
In another recent case, the New South Wales Court of Criminal Appeal considered an application for leave to appeal convictions on the basis of issues surrounding a juror bringing newspaper clippings about the trial into the jury room.249
Concerns were raised that a juror (or jurors) may have taken the further step of making inquiries on the internet and that this could have revealed prejudicial information about the accused.250 While the court determined there was insufficient evidence to demonstrate jurors had made online inquiries, it was satisfied the risk of prejudice was
245 Elizabeth Byrne, ‘Is Facebook Ruining the Chance of a Fair Trial?’, ABC News (Text, 20 May 2017)
51 Literature Review ever present and pointed out that had jurors made online inquiries, ‘there was no doubt that a substantial miscarriage could and probably would occur’.251
Also in New South Wales, the Supreme Court considered an application for a judge-alone trial based on the substantial volume of media publicity about the accused published both online and in mainstream media.252 The court accepted the publicity was extensive, strong in content and prejudicial to the accused, especially where references were made to the defendant’s criminal history.253 However, as the passage of time from the period of significant prejudicial publicity to the commencement of the trial was some 14 months, the court was not minded to order a trial by judge- alone.254 The court did, however, consider the issues raised were significant enough to warrant it taking steps to avoid any ‘public trigger which may lead to members of the public (including the jury panel) engaging in electronic searching concerning the
Applicant’.255 Accordingly, the court made non-publication orders to ensure there would be no publication of the listing of the Applicant’s trial or of the name of the
Applicant until after the completion of the trial.256
In 2016, a murder trial jury needed to be discharged after a Sydney journalist published prejudicial information that had not been heard by the jury.257 She was subsequently convicted of sub judice contempt and given a two-year good behaviour
251 Ibid [14]-[21]. 252 R v Shaun Stewart McNeil [2015] NSWSC 357, [13]. 253 Ibid [72]. 254 Ibid [77], [86]. 255 Ibid [106]. 256 Ibid [111]-[112]. 257 ‘Journalist Has Murder Jury Discharged’ (22 August 2016)
Twitter trials and Facebook juries 52 bond.258 Justice John Dixon said aborting the trial had caused ‘great inconvenience and stress to the accused, the victim’s family and witnesses’.259 In a recent Perth murder trial, a juror was discharged after he conducted independent research into the case and, in doing so, disregarded the Judge’s clear instructions not to research.260 In one of my own case-studies, the Tostee murder trial, a juror identified herself on Instagram and, in response, the defence filed an application for a mistrial which, if successful, would have resulted in the trial being aborted. I discuss this further in Chapters 5 and 6.
These cases demonstrate the range of costly outcomes that stem from media and other publications about criminal trials. There are also many international examples where efforts to overcome interference with the administration of justice have resulted in trials being abandoned and juries discharged,261 or appeals launched and verdicts overturned.262 In the United States, from 1999 to 2010, at least 90 verdicts were challenged due to juror misconduct and behaviour related to the internet.263 As recently as November 2015, a Queens (New York City) juror sitting on an armed robbery trial was found in contempt of court for her ‘gossipy’ discussion of the trial on Facebook.264
She was fined $1000 and a mistrial was declared at significant expense to the criminal justice system. Similarly, in the United Kingdom from 2005 to 2012 there were 18
258 Joanna Crothers, ‘Yahoo7 Fined $300k over Article That Caused Murder Trial to Be Aborted’, ABC News (Text, 17 February 2017) 7
53 Literature Review appeals related to juror misconduct and the internet or social media.265 Another United
Kingdom study (2008) provides details of child cruelty, manslaughter, murder, and sexual abuse cases where juries had to be discharged or trials abandoned because of issues associated with jurors and the internet.266 It is in these types of often violent high-profile proceedings where, arguably, prejudicial publicity in the media may have the most serious implications. A defendant in a high-profile criminal trial may face a potentially long period of imprisonment if found guilty and, in this sense, has much to lose if a jury is biased. Victims also suffer if a verdict is tainted and, when the jury is discharged, a trial abandoned or a verdict overturned, they may face the solemn task of having to sit through a trial a second time.
2.7 Summary
It is clear the context in which the sub judice rule operates is a complex one, one made even more complicated by the internet and social media. The demand for news about criminal trials, the conflicting rights of publishers and defendants, the public’s broader interest in and right to fair and open justice, the complexities of jury trials, and the unique challenges of social media, mean there are no simple solutions when it comes to applying the sub judice rule. It is within this broad and complex context that my research is situated.
265 United Kingdom Law Commission, Contempt of Court: Consultation Paper (2012) 62. 266 Thomas (n 211) 6.
Twitter trials and Facebook juries 54
3 Research Design
3.1 Overview
In this chapter, I outline the methods I used to respond to my research question and sub-questions. In terms of data collection, I explain how I selected suitable high- profile criminal trials, the search parameters and process I used to collect data from
Twitter and Facebook using TrISMA,267 and how I created randomised samples for manual coding. I then discuss the quantitative and qualitative methods I used for data analysis, including content and discourse analysis.268 Finally, I explain the limitations of my research, particularly in terms of data collection. An overview of my research methods for data collection and analysis is set out in Table 3.1.
It is worth noting that my methods developed and evolved over time. My case study of the Baden-Clay trial was particularly significant in terms of developing my approach and it formed the basis of my publication in the University of New South
Wales Law Journal.269 By the time I commenced my second case study — the criminal trial of Gable Tostee — I was able to extend my qualitative analysis. Also, as TrISMA was tracking Facebook in addition to Twitter by this time, I was able to undertake a
267 This research is supported by infrastructure provided through the Australian Research Council – funded Tracking Infrastructure for Social Media Analysis (TrISMA – LIEF LE140100148). Bruns, Axel, Burgess, Jean, Banks, John, Tjondronegoro, Dion, Dreiling, Alexander, Hartley, John, Leaver, Tama, Aly, Anne, Highfield, Tim, Wilken, Rowan, Rennie, Ellie, Lusher, Dean, Allen, Matthew, Marshall, David, Demetrious, Kristin, and Sadkowsky, Troy. (2016). TrISMA: Tracking Infrastructure for Social Media Analysis, http://trisma.org/. I use TrISMA to collect data from Twitter and Facebook. I explain how it works in more detail later in this chapter. 268 I also used the visual analytics tool Tableau: see https://www.tableau.com/. 269 Rachel Hews and Nicolas Suzor, ‘“Scum of the Earth”: An Analysis of Prejudicial Twitter Conversations during the Baden-Clay Murder Trial’ (2017) 40(4) University of New South Wales Law Journal 1604.
55 Research Design
Facebook analysis. As my approach is more extensive for my second case study, I have separated it into two chapters: Twitter Analysis and Facebook Analysis.
RESEARCH QUESTION AND SUB-QUESTIONS How well adapted is the Australian sub judice rule for regulating prejudicial publicity on social media during high-profile criminal trials?
a. How prevalent is prejudicial publicity on social media during high-profile criminal trials?
b. How do professional journalists and non-journalists talk about high-profile criminal trials on social media?
c. How do social media users show prejudice on social media during high-profile criminal trials? DATASET CASE USED FOR METHODS TOOLS STUDY ANALYSIS Data Collection Data Collection original dataset collected using TrISMA Murder Twitter dataset keywords and date parameters DMI-TCAT trial of random sample with confidence Gerard original dataset level 95%, confidence interval Data Analysis 33,067 tweets Baden- ±1%, total population = original manual coding for prejudice - random sample Clay dataset close reading & excel 7,427 tweets spreadsheet prejudicial Twitter Data Analysis manual coding to identify dataset quantitative and qualitative professional journalists Analysis 446 tweets analysis (content and discourse data analysis and visualisation analysis) using Tableau Data Collection Data Collection Murder Twitter original dataset collected using TrISMA trial of dataset keywords and date parameters Gable random sample with confidence original dataset level 95%, confidence interval Data Analysis Tostee 8,934 tweets ±1%, total population = original manual coding for prejudice - random sample dataset close reading & excel Twitter 4,629 tweets spreadsheet Analysis prejudicial Data Analysis manual coding to identify professional journalists dataset quantitative and qualitative 807 tweets analysis data analysis and visualisation (content and discourse analysis) using Tableau Data Collection TrISMA Facebook Data Collection Murder original news items identified trial of dataset original dataset collected using keywords and date parameters via Factiva Gable original dataset 1,575 Tostee Data Analysis Data Analysis comments from 14 posts quantitative and qualitative manual coding for prejudice - Facebook (ie news analysis close reading & excel Analysis articles) (content and discourse analysis) spreadsheet manual coding to identify professional journalists Table 3.1: Research design – overview of methods for data extraction and analysis.
Twitter trials and Facebook juries 56
3.2 Data Collection
In this section, I explain how I went about choosing suitable high-profile criminal trials as case studies, what search parameters and processes I used to collect the data, and how I created random samples for manual coding.
3.2.1 Selecting suitable high-profile criminal trials There is no single definition of a high-profile criminal trial, so I relied on a combination of different descriptions from other studies when choosing my case- studies. They have been described as trials that ‘attract a moderate or high level of specific publicity’,270 ‘highly publicized criminal cases’,271 or ‘[a] highly emotional … trial that attracted nationwide media coverage’.272 In some studies, trial duration has been factored in to describe them as trials ‘of at least two weeks duration and … [that receive] ‘substantia’ pre- and during-trial media coverage’273 and, similarly, as trials
‘lasting two weeks or more with substantial pre-trial and in-trial media coverage’.274
I did not set out specifically to investigate murder trials, however, by their nature they tend to attract large volumes of media coverage. Based on the various descriptions above, I chose the murder trial of Gerard Baden-Clay as my first case-study. It was clearly a high-profile trial as it lasted five weeks, generated intense public interest, and attracted significant nationwide media attention. For my second case-study, I chose
270 Chesterman et al (n 218) 17. 271 Dorothy J Imrich, Charles Mullin and Daniel Linz, ‘Measuring the Extent of Prejudicial Pretrial Publicity in Major American Newspapers: A Content Analysis’ (1995) 45(3) Journal of Communication 94, 97 (‘Measuring the Extent of Prejudicial Pretrial Publicity in Major American Newspapers’); Gary Moran and Brian L Cutler, ‘The Prejudicial Impact of Pretrial Publicity1’ (1991) 21(5) Journal of Applied Social Psychology 345, 355. 272 Marjorie O Dabbs, ‘Jury Traumatization in High Profile Criminal Trials: A Case for Crisis Debriefing’ (1992) 16 Law and Psychology Review 201, 201 (‘Jury Traumatization in High Profile Criminal Trials’). 273 Hannaford-Agor, Rottman and Waters (n 209) 3. It should be noted, this publication does not further elaborate on the definition of the word “substantial”, despite having placed it in inverted commas. 274 Thomas (n 211) 40–41.
57 Research Design the murder trial of Gable Tostee, as it generated significant media attention in both
Australia and internationally. Although it was a slightly shorter trial — at 10 days — it nevertheless appeared to fall within those definitions of a high-profile criminal trial that did not refer specifically to trial duration.
The timing and location of these trials was also relevant. There was a limit to the number of high-profile criminal trials that took place during the data collection phase of my research and I factored this into my selection. That these trials took place in
Brisbane was also important, as I was able to attend parts of the trials to ensure I had a comprehensive understanding of the facts and evidence.
3.2.2 Publication during trials I chose to analyse tweets and comments posted during the trials of Baden-Clay and Tostee, rather than pre-trial publicity. 275 Changes in the way news about criminal trials is distributed and consumed, has led to concerns that jurors are accessing social media during trials276 and, possibly, while deliberating.277 The during trial period is significant as at this point in criminal proceedings jurors have been empanelled and the trial has commenced. This is the stage when jurors may be at greatest risk of influence up until they deliver their verdict.278 It is also during the trial period that
275 This study focusses on during trial data – a period during which the jury is open to influence up until the verdict: see New South Wales Law Reform Commission, Contempt Report Summary, above n 12, 26, [75]. Many studies examine pre-trial prejudice, however, this is not the focus of my research. 276 Jacqueline Horan, Juries in the 21st Century (Federation Press, 2012) 187–94; Jane Johnston et al, ‘Juries and Social Media: A Report Prepared for the Victorian Department of Justice’ (Report, 2013) 1, 9–13; Cheryl Thomas, ‘Are Juries Fair?’ (Report, Research Series No 1/10, Ministry of Justice, February 2010) 44
Twitter trials and Facebook juries 58 prejudicial publicity may result in a trial being aborted and this takes place at significant expense to the parties and the criminal justice system.279
3.2.3 Twitter data for the murder trial of Gerard Baden-Clay Before collecting Twitter data, I needed to establish the parameters of what I was looking for. My aim was to extract as many tweets as possible that were posted during the murder trial of Gerard Baden-Clay. To do this, I needed to identify a combination of trial dates and trial-related key words to use as search parameters. I also needed to create a random sample of tweets for manual coding.
3.2.3.1 Specified date range and set of key terms To create my original dataset, I specified the following date range: from 9:00 am on 9 June 2014 — the morning the jury was empanelled — to 11:51 am on 15 July
2014 — one minute before the verdict was delivered. To develop a set of keywords, I considered the key terms and hashtags that had been used on Twitter when discussing the case in the months leading up to the trial. These generally reflected the names of the accused and victim. After considering a range of terms — including ‘Gerard
Baden-Clay’, ‘Allison Baden-Clay’, ‘Baden Clay’, ‘RIP Allison’ and hashtags
‘#badenclay’, ‘#baden-clay’, and ‘gbc’ — I settled on the terms ‘baden-clay’ and the hashtags ‘#badenclay’ and ‘#gbc’.280 I needed to use both a keyword and hashtag term
expired: see Delbert-Evans v Davies (1945) 2 ER 167; Ex parte Attorney-General; Re Truth & Sportsman Ltd [1961] SR NSW 484. However, once a verdict has been handed down and the jury is no longer involved in the trial process, the potential effect of prejudicial publications is considered to be slight. There are divided opinions as to whether judges themselves are likely to be influenced by prejudicial media. Some are of the opinion that judges are not open to influence: see Attorney-General (UK) v British Broadcasting Corp [1981] AC 303, 342–343; Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (1982) 152 CLR 25, 102; Attorney- General v Times Newspapers Ltd [1974] AC 273, 301; R v Duffy; Ex parte Nash (1960) 2 QB 188, 198; Bell v Stewart (1920) 28 CLR 419, 425–426. Others concede it is possible for judges to be influenced, see: Attorney-General (UK) v British Broadcasting Corp 335; Bell v Stewart 433; Kerr v O’Sullivan [1955] SASR 204, 209–210; Civil Aviation Authority v Australian Broadcasting Corporation (1995) 39 NSWLR 540, 550. 279 Joanne Baker, Adrian Allen and Don Weatherburn, ‘Hung Juries and Aborted Trials: An Analysis of Their Prevalence, Predictors and Effects’ (2002) 66 Crime and Justice Bulletin 1, 2–3. 280 Bruns and Burgess (n 34).
59 Research Design for Baden-Clay, as they are treated differently during the search process. While the search term ‘baden-clay’ would filter tweets containing ‘baden-clay’ anywhere in the tweet — including in a hashtag — the search term ‘#badenclay’ would only filter tweets containing the term, including the hashtag. To accommodate discrepancies about the use of spaces, no space, or a hyphen between ‘baden’ and ‘clay’, I also included some minor variations such as ‘badenclay’ and ‘baden clay’.
3.2.3.2 Selecting the data To create my dataset, I input my specified date range and set of keywords into two different computerised tools. I extracted data directly from the Twitter Application
Programming Interface (API)281 using the Digital Methods Initiative – Twitter Capture and Analysis Toolset (DMI-TCAT)282 and I supplemented this data using TrISMA.
TrISMA refers to the Tracking Infrastructure for Social Media Analysis project which is an ARC-funded LIEF project that develops comprehensive infrastructure for the large-scale, continuous tracking of public social media activities by Australian users.
At the time, TrISMA was capturing the public tweets of 2.8 million Twitter accounts that had been identified as Australian.283 TrISMA was in its early stages of development during the Baden-Clay trial, and as my methods were also in the early stages of developmental, it made sense to use both DMI-TCAT and TrISMA to ensure
I collected as comprehensive a dataset as possible. By the time I collected data for the
281 APIs are data interfaces offered by many Web platforms. Their main purpose is to provide software developers an unambiguous, data-only version of a site’s content for use in their own software.’ See Weller et al, above n 94, 65. 282 Erik Borra and Bernhard Rieder, ‘Programmed Method: Developing a Toolset for Capturing and Analyzing Tweets’ (2014) 66 Aslib Journal of Information Management 262. 283 The Australian Twitter userbase was first collected in 2013 and resulted in the tracking (by TrISMA) of just under 2.8 million Australian accounts. This is the userbase relied upon in this research. In early 2016, the userbase was re-collected and TrISMA is currently tracking over 4 million accounts identified as Australian.
Twitter trials and Facebook juries 60 murder trial of Gable Tostee, TrISMA was more advanced and I could confidently use it alone.
After running DMI-TCAT and TrISMA, my total dataset for the trial period contained 33,067 tweets.284 The DMI-TCAT data is generally a superset of the
TrISMA data; DMI-TCAT should collect all tweets with matching keywords, while
TrISMA only collects matching tweets from Australian accounts, as detailed below.
However, in addition to tweets containing keywords, TrISMA was able to extract tweets that formed part of conversations between Australian accounts where only some of the tweets directly mentioned one of the keywords. This means that in some cases,
TrISMA captured tweets that did not contain a keyword, where they were in response to a tweet that did contain a keyword. This also applied vice-versa. Approximately
1,300 tweets were additional tweets captured by using TrISMA in this way. Collecting these additional tweets was useful as they formed part of relevant conversations about the trial and had the potential to provide insight.
3.2.3.3 How does TrISMA collect Twitter data? TrISMA285 currently provides infrastructure to track data from both Twitter and
Facebook. It is important to note that all data collected from Twitter is publically available — that is, it is available to anyone regardless of whether they are logged into the network. Any social media that is ‘private’ — such as content posted by locked
Twitter accounts or direct messages — is specifically excluded.
284 Individual tweets are identified by unique tweet identification numbers. 285I note much of the information used to describe TrISMA in this section (3.2.3.3), including in footnotes, has been copied from a QUT internal document titled TrISMA Social Media Data Resources (version 0.1, 27 July 2015) and prepared by Professor Axel Bruns. The document has been distributed to members of the QUT Digital Media Research Centre to ensure TrISMA is accurately and consistently explained by members of the Centre. In many instances, I have used verbatim words, phrases and sentences—or compilations of words, phrases and sentences—from different sections of the document. I have not attempted to identify these words, phrases, sentences or compilations as quotes, as this would be confusing for the reader.
61 Research Design
TrISMA gathers tweets from all identified Australian Twitter accounts, on a continuous basis, and stores them in a database available to accredited TrISMA researchers.286 This is also referred to as the Australian Twitter Collection. The collection draws on a database of Australian Twitter accounts. This database is generated by filtering the global Twitter userbase for accounts which match any of the following criteria: account profile timezone is set to an Australian timezone; account location field refers to a recognisably Australian location; or account description field refers to an Australian location or contains other Australia-specific terms.287 The
Australian Twitter Collection itself tracks the public tweets from each of these accounts on a continuous basis. Because of the considerable effort involved, the full list of Australian accounts established through this process is updated only periodically. As at September 2013 — and at the time of the Baden-Clay trial — the list of Australian Twitter accounts being tracked contained approximately 2.8 million accounts.288 In early 2016, the userbase was re-collected and TrISMA is currently tracking over 4 million accounts identified as Australian. The dataset gathered since
August 2015 is complete, except for gathering outages advertised on the TrISMA
Website. It should be noted that historical Twitter data may be incomplete.289
286I am an accredited TrISMA user. 287 Location terms being tested for include the 45 largest Australian cities (covering more than 80% of the Australian population), the names and abbreviations of states and territories, “Australia” and variations, and terms such as “down under” and “Oz”. Most of the common false positives (e.g. Perth, Scotland; Brisbane, California) have been removed automatically. A very small number of placenames (e.g. Orange; ACT) could not be tested for because of their frequent occurrence in other contexts. 288 At the time of collecting my data, the list of Twitter accounts being tracked dated to September 2013. It contained 2.827 million accounts matching any of the selection criteria. This included: 1.563 million accounts matching one of the eight Australian timezones; 1.866 million accounts matching one of the location filters; and 414,000 accounts matching one of the description filters. It should be noted that accounts may match one or more of the three filter criteria; thus, the sum of matches across the three filters is greater than the total number of 2.8 million accounts. Facilities for flagging erroneously included non-Australian accounts for removal from the collection, and for suggesting additional Australian accounts to be manually added, will be made available at a later stage. 289 Historical Twitter data from before August 2015 is subject to the limitations of the Twitter API. The API delivers up to 3,200 past tweets per account; this may cover months or years for most
Twitter trials and Facebook juries 62
3.2.3.4 Creating a randomised sample for manual coding While computer-aided analysis is making it increasingly possible to code for almost any content in a text based dataset, identifying prejudicial publicity was something I needed to code for manually. As my original dataset of 33,067 tweets was too large, I created a random sample.290 By using a random sample it was possible to recreate activity patterns in the data, such that highly active users were better represented than less active users, and highly retweeted tweets were better represented than tweets not retweeted. This strategy makes the tweets of influential users — particularly professional journalists and media organisations with large established audiences — more prominent, reflecting their increased visibility as they might be perceived by average users, or by non-users searching for information on Twitter. For the purposes of understanding prejudice in social media conversations, it was important that these activity patterns were recreated.
So that I could undertake some basic quantitative analysis, I selected a random sample size that was statistically generalisable across the original dataset (ie the total population).291 From my original dataset of 33,067 tweets, I chose a confidence level of 95% and a confidence interval of one (that is, a margin of error of ±1%), to calculate
accounts, but only weeks or days for some of the most active accounts. For tweets dated prior to August 2015, it should therefore be assumed that the available record is increasingly incomplete for earlier periods: data for the first half of 2015 may be close to complete, while data for 2007 would only cover those accounts which were already active during that time, but have not tweeted more than 3,200 times between then and August 2015. Data prior to August 2015 should be used with considerable caution, therefore. 290 ‘Random Sampling’ in Encyclopedia of Survey Research Methods (Sage Publications, Inc., 2008) 683
63 Research Design a random sample size of 7,427 tweets (or 22.5 per cent of the original dataset).292 This meant I could be 95% certain that the statistics I generated from my random sample would apply to the original sample, within a margin of error of ±1 per cent.293 Based on this calculation, I randomly selected 7,427 tweets from my original dataset to become my random sample.294
3.2.4 Twitter data for the murder trial of Gable Tostee My search strategy for collecting Twitter data for the murder trial of Gable
Tostee was the same as for my earlier case-study — using a specified date range and keywords — except that I was able to collect all my data using TrISMA, so I did not need to use DMI-TCAT. As for the Baden-Clay trial, I focussed on the during-trial period. By applying the same strategy I could make useful comparisons between my
Twitter analyses from both trials.
To create my original dataset, I specified the following date range: from 9:00am on 10 October 2016 — the day the jury was empanelled and Day 1 of the trial — to
3:20pm on 20 October 2016 — one minute before the verdict was delivered. I chose the keyword ‘tostee’ as it appeared to be used consistently — both on its own and as a
292 This sample size has been calculated using the Sample Size Calculator located at http://www.surveysystem.com/sscalc.htm. 293 For example, if 6% of tweets in my random sample contained prejudicial information, I could be 95% certain that between 5-7 percent of tweets in the original dataset contained prejudicial information. 294To select these tweets I used a sampling strategy known as simple random sampling without replacement, such that tweets from my original dataset were selected by chance, each tweet had an equal chance of selection, and a tweet could only be included once in the random sample: see ‘Random Sampling’ (n 293) 683; ‘Random Sampling’ in The SAGE Encyclopedia of Qualitative Research Methods (SAGE Publications, Inc., 2008) 726
Twitter trials and Facebook juries 64 hashtag #tostee. Using TrISMA295 I accessed the Australian Twitter Collection296 and collected tweets containing the keyword ‘tostee’ that were posted during the trial period. This resulted in an original dataset containing 8,934 tweets. For coding purposes, I created a random sample297 containing 4,629 tweets (52% of the original sample). As my original sample is smaller for this case study, I needed to manually code a much larger proportion of the original dataset in order for the findings to be statistically generalisable.
3.2.5 Facebook data for the murder trial of Gable Tostee As mentioned previously, by the time the Tostee trial took place in late 2016,
TrISMA was tracking Facebook data as well as Twitter data. This provided the opportunity to analyse conversations about the trial on a second social media platform.
My aim was to identify any trial-related news articles posted by media organisations on their Facebook public pages during the trial and then collect all comments that responded to those articles. Those comments would then form a dataset — similar to my Twitter datasets — that I could use for analysis.
3.2.5.1 How does TrISMA collect Facebook data? Before explaining my process for collection, it is worth noting how TrISMA collects Facebook data, as it tracks the platform in a different way to Twitter. Facebook
295 Bruns, Axel, Burgess, Jean, Banks, John, Tjondronegoro, Dion, Dreiling, Alexander, Hartley, John, Leaver, Tama, Aly, Anne, Highfield, Tim, Wilken, Rowan, Rennie, Ellie, Lusher, Dean, Allen, Matthew, Marshall, David, Demetrious, Kristin, and Sadkowsky, Troy. (2016). TrISMA: Tracking Infrastructure for Social Media Analysis, http://trisma.org/. 296 The Australian Twitter Collection was last updated in early 2016. It is a collection of the full Australian Twitter userbase and includes just over 4 million accounts. It includes accounts that have set a distinct Australian timezone, or reference one of Australia’s 30 largest cities, an Australian state or territory, or Australia itself in their location or description information, on their profiles. False positives (e.g. Perth, Scotland) have been excluded wherever possible. While the user accounts will be included in the collection, the actual tweets by ‘protected’ accounts will not be included, as they would only be visible to approved followers of those accounts. 297 ‘Random Sampling’ (n 293) 683; ‘Random Sampling’ (n 297) 726; ‘Simple Random Sample’ (n 297).
65 Research Design provides only a very limited API for tracking public communication on the platform and this means the data available through TrISMA is more limited than for Twitter.
As for the data collected on Twitter, all data collected from Facebook is located on public pages and is available to anyone without needing to log into the network. Any social media that is ‘private’ — such as content posted on Facebook profiles or walls
— is specifically excluded from this data extraction.298
As there is no public register of Facebook pages, or any reliable approach to identifying all Australian pages on the site, TrISMA’s Facebook data gathering infrastructure operates on an on-demand basis.299 This means accredited TrISMA users must place a request for a Facebook public page to be tracked, before data capture will commence. Once data capture commences, TrISMA will continuously capture any new posts, as well as likes, comments, and shares of these posts, and likes of and replies to comments. Some limited historical data may be gathered where possible.300
3.2.5.2 Selecting the data To identify which news organisations had published articles about the trial, I undertook a keyword search (‘tostee’) for the trial period on the Factiva database and on the internet.301 From this search, I identified nine prominent news organisations that
298 The semi-private nature of Facebook’s communicative spaces, and the observer-specific algorithmic presentation of user activity on the platform, mean that for ethical, privacy, and technical reasons TrISMA will not gather data from personal user profiles, groups, or other private or semi- private spaces on the platform. As a result, TrISMA will only gather data from public pages on Facebook. 299 All accredited TrISMA users may add pages to be continuously tracked, and the data gathered through this process are in turn made available to all TrISMA users, independent of which user initiated the tracking. While there are currently no automatic restrictions built into the system, users are requested to focus on tracking Facebook pages of Australian relevance through the TrISMA platform. 300 Comprehensive gathering of all historical posts on a page (going back to when the page was first created) is not currently available as a standard feature, but may be requested on a case-by-case basis. 301 Factiva is a global news database of ‘nearly 33,000 premium sources, including licensed publications, influential websites, blogs, images and videos’: see https://www.dowjones.com/products/factiva/.
Twitter trials and Facebook juries 66 appeared to have published two or more articles during the 10 day trial: ABC News;
Brisbane Times; Daily Telegraph; news.com.au; The Courier Mail; Gold Coast
Bulletin; The Australian; Herald Sun; and nine.com.au (previously Ninemsn). I chose
‘two or more articles’ as my parameter, as this allowed me to focus on those media organisation that were most active in publishing news about the trial. I then used these nine media organisations as the basis for collecting data using TrISMA.
Using TrISMA, I undertook a search of each of the nine media organisation’s
Facebook public pages I had previously identified. I filtered the search by trial dates
— from 10 to 20 October 2016 — initially to locate all news articles (on any topic) the organisation had posted on Facebook during the trial period. By then reading through the headlines, I was able to condense the list of articles to those that related to the
Tostee trial. I excluded any articles posted after the verdict was delivered. Through this process, I was able to locate articles posted on Facebook by four of the nine media organisations: ABC News, Brisbane Times, Daily Telegraph, and news.com.au.
The TrISMA results for four of the five, The Australian, The Courier Mail,
Herald Sun and nine.com.au, suggested there were no trial-related articles posted during the trial period. A search of Factiva suggests each of these news organisations did publish articles either in hardcopy or on their websites during the trial period, but these do not appear to have been shared to Facebook. A separate online search of
Facebook confirmed that these media organisations did not share their publications to
Facebook during the trial period, although they did share some publications before and after the trial. Given cases remain sub judice until any appeal periods have expired,302 it is possible these organisations chose not to publish on Facebook in order to prevent
302 Attorney-General (Qld) v WIN Television (Qld) Pty Ltd (n 61) 157.
67 Research Design any potential liability — but there is no evidence to confirm this. The fifth media outlet,
The Gold Coast Bulletin, for technical reasons had not been tracked by TrISMA during my timeframe of interest.
Despite these limitations, from the four major news organisations that were accessible I found 14 trial-related news articles (posts). I note this sample size is quite small and so my results are not necessarily representative of a broad cross-section of the media. However, these publications do provide reasonable coverage and representation of the newspapers in Brisbane where the trial took place and, therefore, where the jury resides. These publications also represent the three of the most prominent media corporations in Australia including Murdoch’s NewsCorp Australia
(Daily Times and news.com.au), what was then Fairfax Media Group which is now known as Nine Entertainment Co (Brisbane Times), and the public, independent
Australian Broadcasting Corporation (ABC News).303
From these, I extracted a total of 1,575 comments that responded to those news articles. The breakdown of posts and comments is set out in Table 3.2. As this dataset was smaller than my Twitter samples, there was no need to generate a random sample for manual coding. Instead I manually coded all 1,575 comments.
303 Tim Dwyer, ‘FactCheck: Is Australia’s Level of Media Ownership Concentration One of the Highest in the World?’, The Conversation 12 December 2016
Twitter trials and Facebook juries 68
Date of News Number of Post (article) organisation comments collected 11-Oct-16 Brisbane Times 8 12-Oct-16 ABC News 239 12-Oct-16 Brisbane Times 102 12-Oct-16 Brisbane Times 69 12-Oct-16 Brisbane Times 38 13-Oct-16 Brisbane Times 2 14-Oct-16 ABC News 439 14-Oct-16 Brisbane Times 111 17-Oct-16 Brisbane Times 46 17-Oct-16 Daily Telegraph 8 17-Oct-16 news.com.au 229 18-Oct-16 Brisbane Times 30 18-Oct-16 Brisbane Times 131 19-Oct-16 Brisbane Times 123
Total comments extracted 1575
Table 3.2: Tostee Facebook analysis - summary of comments collected.
3.3 Data Analysis
It is difficult to know how serious are concerns that jurors may be exposed to — and potentially influenced by — prejudicial publicity on social media during high- profile criminal trials. There is no existing data to reveal how prevalent prejudicial publicity is on social media and it is only through content analysis that it is possible to answer this question. It is also not clear how professional journalists and non- journalists talk about trials, and whether there are differences in the way they comply with the sub judice rule. Further, it is important to better understand how social media users show prejudice on social media as means of understanding how they experience criminal trials.
In order to answer these questions, I use a combination of quantitative and qualitative analysis, including content and discourse analysis. In this section, I explain
69 Research Design how I coded for prejudicial publicity and user types. I also explain how I identified themes (thematic analysis) to make meaning from tweets and comments. Figure 3.1 demonstrates how I used my research methods to answer my research sub-questions and, ultimately, to respond to my overall research question.
Figure 3.1 Research design – responding to research questions using quantitative & qualitative analysis.
3.3.1 Content and discourse analysis Content analysis is particularly popular in media and communications studies304 where it is has been used to analyse newspapers or advertisements and, more recently, with electronic forms of media and the internet.305 It can be applied to social media online platforms306 to examine the metrics of tweets or comments, the language used, tweet or comment content, conversational structures and the diffusion of
304 Jeremy Hunsinger, Lisbeth Klastrup and Matthew Allen, International Handbook of Internet Research (Springer, Springer Verlag, Springer Netherlands, 1st ed, 2010) 233–249; ‘Content Analysis, Definition Of’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017)
Twitter trials and Facebook juries 70 information.307 This type of content analysis is useful for describing a phenomenon, particularly when ‘existing theory or research literature on a phenomenon is limited’.308 To apply this approach, each tweet or comment is regarded as a single unit of meaning,309 or a sampling unit, ‘sent by a unique user at a particular moment’.310
From a quantitative perspective, content analysis is a ‘systematic, quantitative process of analysing communication messages by determining the frequency of message characteristics’311 — this is sometimes referred to as ‘quantitative analysis of qualitative data’.312 From a qualitative perspective, content analysis is a method for
‘systematically describing the meaning of qualitative data’ by ‘assigning successive parts of the material to the categories of a coding frame’.313 It is a process of categorising data into different themes in order to gain ‘a sense of how different parts of the material compare and relate to each other’.314 This makes it possible to ‘show trends in the content of media messages’ rather than focussing on understanding single
(media) messages.315
307 Weller et al (n 284) 98. 308 Hsiu-Fang Hsieh and Sarah E Shannon, ‘Three Approaches to Qualitative Content Analysis’ (2005) 15(9) Qualitative Health Research 1277, 1279. 309 Weller et al (n 284) 100–101. 310 Ibid. 311 ‘Content Analysis: Advantages and Disadvantages’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) 239
71 Research Design
I used quantitative analysis to count the frequency of different types of prejudicial content in tweets and comments. I also used it to count the frequency of tweets and comments posted by professional journalists, compared with non- journalists. I used qualitative analysis to identify themes and interpret the underlying text in my datasets (thematic analysis).316 This approach helped me to reduce my large volume of tweets and comments into a more manageable selection of themes that
‘focus[ed] on selected aspects of meaning’ related to my research questions.317 I was then able to interpret these themes by ‘seeking commonalities, relationships, overarching patterns, theoretical constructs, or explanatory principles’.318
My analysis extends beyond content analysis, however, and also includes discourse analysis. Discourse analysis is a broad term used to describe the social study of language as used in talk, text and other forms of communication or, for this research, in tweets and comments.319 I used discourse analysis to identify the main users tweeting about the trial, what types of language they used and the consequences320 of this as demonstrated through tweet content, conversational structures and indicators of the potential diffusion of information.321 I looked at the substance of conversations as well as the style and apparent purpose of users who tweeted.322 Critical discourse analysis relies on a variety of approaches and techniques to investigate ‘how discourse works in the social world … [including how it] interject[s] in institutional, social, or
316 ‘Thematic Analysis’ in Encyclopedia of Case Study Research (SAGE Publications, Inc., 2010) 925–926
Twitter trials and Facebook juries 72 political controversies’.323 It pays attention to ‘the text of the discourse itself’, ‘the content and processes of production of the text’, and social structures surrounding the text.324 It is useful for considering tensions in values and how they affect the way people talk.325
Using this approach I am able to better understand how social media users use language in the online social settings of Twitter and Facebook. Their tweets and comments provide ‘naturally-occurring conversation and dialogue, or talk that is not influenced by a researcher’326 — a more valuable source of insight than communication where users know they are the subject of research.327 As for individual journals or online blogs, tweets and comments serve as written accounts of interaction between social media users — a type of micro-level discourse.328 My analysis also extends to macro-level discourse to consider the broader social patterns and practices of social media users in my dataset.329 This type of discourse analysis is commonly used for media texts such as news from television, newspapers, magazines, internet
323 ‘Discourse Analysis’ (n 322) 525. 324 David Byrne, ‘What Is Discourse Analysis?’ in Data Analysis and Interpretation (2017)
73 Research Design news and online media.330 It allows me to investigate common or enduring patterns of talk and text across contexts in my datasets.331 This makes it possible not only to identify what the themes are in my data, but also to consider how those themes are framed in the media, how different users talk about criminal trials, and how they show prejudice on social media.
3.3.2 Coding for prejudice I used quantitative analysis to investigate the prevalence of prejudicial publicity on social media during high-profile criminal trials.332 By referring to the legal principles underlying sub judice contempt cases,333 I developed a coding scheme to apply to tweets and comments.334 By applying this approach, I was able to count the frequency of different types of prejudicial content.335
There is no single definition at law as to what constitutes a prejudicial publication but, based on the history of sub judice cases, a range of accepted categories of prejudice exist. Prejudicial information includes statements as to guilt,336 information about confessions337 or prior convictions,338 statements as to innocence,339
330 LeGreco (n 328) 77. 331 Ibid 70. 332 Hsieh and Shannon (n 311) 1278. 333 Ibid 1277. 334 Ibid 1279. 335 Norman K Denzin and Yvonna S Lincoln, The Sage Handbook of Qualitative Research (Sage, 4th ed., 2011) 596, 599. 336 Attorney-General (NSW) v Radio 2UE Sydney Pty Ltd and John Laws (n 61); Attorney-General (Qld) v WIN Television (Qld) Pty Ltd (n 61). Statements of guilt include ‘material that asserts, suggests or creates the impression that an accused person committed the crime with which he or she has been charged’. See Des Butler and Sharon Rodrick, Australian Media Law (Thomson Reuters (Professional) Australia Limited, 4th ed, 2012) 332. 337Unless the confession has been admitted into evidence, in which case it may lawfully be published. See Attorney-General (NSW) v John Fairfax and Sons Ltd (n 63); Attorney-General (NSW) v TCN Channel Nine Pty Ltd (n 63) 380. 338 Maxwell v Director of Public Prosecutions (n 64) 317; Attorney General (NSW) v Willesee (n 64). 339 Director of Public Prosecutions v Wran (1987) 86 FLR 92. Statements asserting the accused is innocent can also be prejudicial, as sub judice contempt is concerned not only with the risk of the innocent being convicted, but also with the risk of the guilty being acquitted. See R v Castro, Onslow & Whalley’s Case (n 62); R v Pearce (n 62).
Twitter trials and Facebook juries 74 content that criticises or disparages the accused,340 or creates sympathy for the accused or a victim,341 or claims that prejudge an issue of fact or law that is awaiting judgment, other than guilt or innocence.342 I used these categories as the basis for my coding scheme and they are set out in Table 3.3 below.
Table 3.3: Research design - coding scheme for prejudice in tweets and comments.
In addition to coding the content of tweets or comments as prejudicial or non- prejudicial, I further categorised them as containing either high or low level prejudice.
Although the sub judice rule is concerned with prejudice that is both against or in favour of an accused, the courts tend to be more concerned with the potential for the innocent to be convicted, rather than for the guilty to go free.343 For this reason, I assigned those coding categories that more firmly asserted guilt as high level prejudice
340 Director of Public Prosecutions v Francis (n 65); R v Saxon, Hadfield and Western Mail Ltd (n 65). This includes comments that are generally adverse or denigrate the accused or that excite feelings of hostility or antipathy against an accused. See Director of Public Prosecutions v Francis (n 65). This may also include a photograph or film footage. See R v Australian Broadcasting Corporation (n 65). 341 Davis v Baillie [1946] VLR 486; R v Truth Newspapers (unreported, Vic SC, Phillips J, 16 December 1993). 342 Attorney-General v Times Newspapers Ltd (n 281); Civil Aviation Authority v Australian Broadcasting Corporation (n 281); Re Coroner’s Court of Western Australia; Ex parte Porteous [2002] WASCA 144 (4 June 2002). 343 Imrich, Mullin and Linz (n 274) 102. It should be noted that although this study was undertaken in the US, the coding methods and decisions used are very similar to mine and are consistent with Australian laws. See also, James Bradley Thayer, ‘The Presumption of Innocence in Criminal Cases’ (1897) 6(4) The Yale Law Journal 185, 186–187, for an overview of the longstanding legal principle of it being preferable to "[let] the guilty go, than … [to convict] the innocent" or, as declared by William Blackstone, "It is better that ten guilty persons should escape than that one innocent person should suffer".
75 Research Design
— these included statements as to guilt, confessions, or prior convictions.344 I coded the remaining categories as low level. It is important to note that tweets or comments coded as either high or low could potentially be contemptuous – this categorisation is not meant to suggest that low level prejudice is not serious or legally problematic.
However, as prosecutions for sub judice contempt are relatively rare, I am working on the presumption that the court is more likely to take action where there is a more severe likely impact. Each category is essentially equally capable of being prejudicial under the doctrine, but they are not all equally likely to result in either serious harm to a trial or to a conviction under sub judice contempt. This categorisation is meant as a convenient way of grouping categories of prejudice based on my observations of media attention. In the few trials where potentially prejudicial social media attention has been relevant, there does seem to be some difference between the different types of categories. For example, in the murder trial of co-accuseds Roger Rogerson and Glen
McNamara, an individual post — posted on both Twitter and Instagram via accounts purportedly owned by a barrister involved in the trial — which implied one co-accused was guilty and the other innocent, was sufficient to result in the court aborting the trial on the grounds of prejudice.345 In the Tostee case, a juror posted images of her coffee with comments suggesting she was deliberating on a ‘nasty trial’ in the Supreme Court of Queensland.346 This was sufficient to prompt the defence to apply for a mistrial,
344 New South Wales Law Reform Commission (n 116) 48–51. 345 Louise Hall, ‘How Barrister Charles Waterstreet Caused Rogerson, McNamara Trial to Be Aborted’, The Sydney Morning Herald (15 June 2016)
Twitter trials and Facebook juries 76 although the court did not grant it in this case. Both of these examples involve the social media accounts of persons involved in the trial, so they are not directly comparable with my samples of persons who were presumably not involved in the trial. Nor are these two examples sufficient to draw conclusions about how the courts approach these cases in general. However, they do provide some sense of context and the possibility that tweets or comments containing information asserting guilt may be seen as treated more seriously by the courts than tweets or comments that are more benign, albeit still potentially contemptuous. Table 3.3 above demonstrates this allocation of high, low and no prejudice (none).
For the Baden-Clay trial, I coded tweet content based only on these three allocations: high, low or none. As my research is based on emerging digital methods which are not commonly used in a legal context, part of my research included the actual creation and development of my methods. As I progressed, I continued to learn and build on my previous approach. By the time I coded my Tostee sample I realised that a greater granularity in coding would provide additional insights that could not be achieved through the high, low or none scheme. This was prompted in part by my qualitative analysis in the Baden-Clay trial which suggested there was a prosecution bias in the sample. I realised that by increasing the granularity of my coding scheme for my Tostee sample, I would be able to expand my qualitative analysis of any prosecution bias to also include a quantitative component. In this exploratory work, improving my methodological approach was an iterative process that built on my previous approaches. For this reason, my methods evolved for the Tostee trial such that I coded the content of tweets and comments using all eleven categories (as listed in Table 3.3 above). By using all eleven categories, I was able to undertake a deeper analysis of how people talk about criminal trials, and how they show prejudice on
77 Research Design social media. I was also able to collapse the Tostee data back into the three categories of high, low and none to make direct comparisons with my Baden-Clay analysis.
To ensure I could undertake basic statistical analysis using percentage calculations347 — such as percentages of overall prejudice, different categories of prejudice, and user types — I ensured that each tweet or comment was allocated to only one coding category. If two categories were relevant, I chose that category the content was ‘most like’. I note at the time of coding, tweets were restricted in size to
140 characters and Facebook comments to 8,000 characters.348
In some instances, there were complexities in coding and these varied between social media platforms and case studies. For the Baden-Clay trial, challenges related to differences in how tweet content may be perceived pre-trial compared to during trial. I also needed to decide how to code tweets that contained attribution to the maker of any claims presented. I explain these challenges in further detail below:
Prejudicial pre-trial but not during trial: A small number of tweets in my
dataset contained content that may have been prejudicial pre-trial, but that were
not necessarily prejudicial during the trial. For example, the tweet ‘Baden-Clay is
an adulterous liar’ would be coded as low level prejudice pre-trial, because it
criticises or disparages the accused.349 However, the accused gave evidence during
his trial that he had lied to his wife and been unfaithful,350 and this makes it less
347 Norman K Denzin and Yvonna S Lincoln, The Sage Handbook of Qualitative Research (Sage, 4th ed, 2011) 596, 599. 348 Josh Constine|September 21 and 2011, ‘Facebook Increases Character Limit on Posts to 5000, Rolls Out Floating Navigation Bar and More Amid Unrest’
Twitter trials and Facebook juries 78
clear whether the tweet is prejudicial. Witness testimonies do not become accepted
‘facts’ until they are accepted by the court — usually when an accused is found
guilty351 — however, self-incriminating claims made by an accused may generally
be presumed to be true. In this context, tweet content that is true will not be
prejudicial, even if it is weighted against the accused.352 Instead, provided the
tweet is fair and accurate, it will be a lawful report from the courtroom.353 I coded
these types of tweets as none.354
Attribution: Sometimes users included a reference to the maker of the claims that
were set out in their tweets: ‘Prosecutor: you killed your wife Mr Baden-Clay’.
This attribution indicates it was the prosecutor who made the claim ‘you killed
your wife Mr Baden-Clay’.355 Without attribution, the tweet would be coded as
high level prejudice (a statement as to guilt). With attribution, however, it is less
clear whether the tweet is prejudicial. It is lawful to present reports from court
even if they are prejudicial, provided the fact they were made during court
proceedings is mentioned.356 As above, provided the tweet is fair, accurate,
351 DPP v Pell (Sentence) [2019] VCC 260; ‘Sentencing Remarks in DPP v George Pell - County Court of Victoria’
79 Research Design
contemporaneous, and made in good faith, it will be lawful.357 As tweets
containing attribution to the maker of the statement appear to be lawful, I coded
them as none. Other versions of attribution included: prosecution, crown, defence,
judge, Justice Byrne (the Judge’s name), Cash (the Prosecutor’s name), and Holt
(Defence Counsel’s name).358 I also took this approach for more generic examples
of attribution including ‘court told’ and ‘jury hears’.359 I did not extend this
approach, however, to tweets that attempted attribution by only including the word
‘trial’ or using quotation marks (single or double) with no attribution.360 The word
‘trial’ and the use of quotation marks were too general to distinguish between
lawful reports from the court room or the opinions of the user tweeting.
Drawing distinctions: I acknowledge the distinctions I have made in coding this
way are not straightforward. It is difficult to accept readers of tweets containing
attribution would always understand the distinction being made between a legally
compliant report and an opinion being expressed by the user. It is also difficult to
know whether the mere addition of attribution is likely to change the potentially
prejudicial effect of any particular tweet. I do not know whether a tweet such as
‘he killed her’ is actually likely to influence a juror any more than a similar tweet
with attribution such as ‘Prosecution: you killed her’. Nonetheless, this approach
357 Ex parte Terrill; Re Consolidated Press Ltd (1937) 37 SR (NSW) 255, 257–9 (Jordan CJ); A-G (NSW) v John Fairfax & Sons Ltd (1985) 6 NSWLR 695, 714 (McHugh JA). 358 For example: ‘Prosecution: Wright had no choice but to try to escape’; ‘Crown: Warriena is terrified – she wasn’t permitted to leave’; ‘Defence says this is just a date gone horribly wrong’; ‘Judge: the prosecution argued Wright feared Tostee’; ‘Justice Byrne says Wright appeared to have assaulted Tostee’; and ‘Cash: Tostee intended to do grievous bodily harm’; ‘Locking the door behind Wright was a de-escalation, Holt’. 359 For example: ‘Wright hit Tostee with a telescope, court told’; and ‘Jury hears Wright’s final cries before fatal fall’. 360 For example: ‘NZ tourist died trying to escape from Tostee: trial’ [use of word trial as attribution]; ‘Tostee as good as pushed Wright over the balcony’ [single inverted commas used in tweet]; or “Wright was too scared to go back into the apartment” [double inverted commas used in tweet].
Twitter trials and Facebook juries 80
allowed me to be conservative in my coding and ensured I did not inaccurately
inflate the number of tweets coded as prejudicial.
There were also complexities in coding in the Tostee trial. These related to coding tweets or Facebook comments that created sympathy for the accused or victim, that blamed the victim for her death, and that contained statements claiming the accused was guilty of criminal offences other than those with which he had been charged. I explain these in further detail below:
Emotion and bare fact reports: Some tweets or comments that contained
emotional content appeared to be opinionative, while others appeared to be fair
and accurate reports. Emotive or opinionative examples include: ‘The tostee case
is awful, that poor girl, such a victim, may she rip and justice prevail’ or ‘Wright
shouldn’t have thrown things at him, plus he told her to leave’. As these may
create sympathy for the accused or victim, I coded them as low level prejudice.
Examples of fair and accurate reports include: ‘Wright’s mum pays tearful
tribute’, ‘Victim’s mum sobbing as recording is played’, or ‘Wright’s mum
surrounded by supporters’. As these are fair and accurate reports of what was
taking place in court, I coded them as none. Again, I do not know whether the
opinionative tweets in these examples are actually likely to influence a juror more
than the fair and accurate reports, but this approach allowed me to be conservative
in my coding. I note that both types of tweets would, however, have the potential
to form part of cumulative negative sentiment and I discuss this further in later
chapters.
Victim blaming: There were a number of tweets or comments during the Tostee
trial that engaged in victim blaming — that is, they blamed Wright for her own
death: ‘this is a cautionary tale for females – don’t go home with someone you
81 Research Design
just met on Tinder’, ‘it was her choice to climb over the balcony’, and ‘she could
have left at any time’. By blaming Wright, these tweets and comments tended to
support the accused — if Wright’s death was her own fault, then Tostee could not
be held responsible. Based on this context, I chose to code these types of tweets
as creating sympathy for the accused (low).
Potential other criminal offences: Sometimes users posted tweets or comments
claiming that on the night of Wright’s death, the accused had committed criminal
offences other than those with which he was charged (murder and manslaughter).
These included kidnapping, deprivation of liberty, domestic violence, assault, and
rape. In determining how to code these, I distinguished offences with which
Tostee could have been charged from those with which he was actually charged.
Using this distinction, I chose to restrict my coding of statements as to guilt (high)
only to those relating to murder or manslaughter. As these were claims about guilt
for other potential offences arising from the facts in the case, I coded them as
‘claims that prejudge an issue of fact or law that is awaiting judgement, other than
guilt or innocence’ (low). As the Queensland State Coroner is currently reviewing
Wright’s death, if the matter goes to inquest Tostee may face further charges such
as those raised by social media users.361 In this sense, they are matters awaiting
judgement that do not relate to the guilt or innocence of those offences with which
Tostee has already been charged. While this category may not be perfect, this
choice ensured I did not artificially elevate my figures around high level prejudice,
361 ‘Gable Tostee Could Face an Inquest into the Death of Warriena Wright | Newshub’ 13 January 2019
Twitter trials and Facebook juries 82
while also ensuring I could still take into account the potentially prejudicial nature
of these claims.
3.3.3 Coding for user types I used both quantitative and qualitative analysis to investigate how professional journalists and non-journalists talk about high-profile criminal trials on social media.
By separating tweets posted by accounts that were identifiable as professional journalists, from tweets posted by accounts of non-journalists, I was able to consider whether they each talk about trials in different ways. I made this distinction so I could understand whether media professionals — who are required both by their institutions and at law to operate within the constraints of the sub judice rule — converse about trials in a different way to non-journalists, who are not accountable to an employer or who may not be aware of the law.
I defined professional journalists as those users ‘with experience as reporters, editors, and producers for print and broad-cast media … [as well as] professionals working for online publications that still adhere to most mainstream journalistic practices’.362 I included individual users, as well as accounts of mainstream news sources, outlets and organisations.363 I did not include social media advisors, digital marketers or bloggers, unless they appeared to otherwise meet the definition of professional journalists. I also did not include users who identified themselves as a journalist, if they did not also provide a further description, or a link to a media institution or publication, or some other information that identified them as a
362 Andrew M Lindner, Emma Connell and Erin Meyer, ‘Professional Journalists in “citizen” Journalism’ (2015) 18(5) Information, Communication & Society 553, 555. 363 Ibid 554.
83 Research Design professional journalist.364 I did include freelance, student or amateur journalists who included a link to a media institution in a way that suggested they were reporting for that organisation as, although they will be personally responsible for the content of their tweets, their identified employers may additionally be liable for unlawful posts.365
Where user accounts were ‘no longer available’ on Twitter,366 I allocated them to the non-journalists’ category. This ensured I did not artificially inflate my findings about prejudicial information posted by professional journalists. All other users falling outside the professional journalists’ category were allocated to the non-journalists’ category.
To define users as professional journalists or non-journalists, I examined their
Twitter profile (including their screen name), their ‘bio’ or user description, and their background images. This included any information about their work or profession, their personal details, their location, any links to their institutional or personal website, as well as a photo or avatar.367 I specifically looked for key terms or information relating to the definition of professional journalists,368 such as: ‘professional journalist’, ‘journalist’, ‘news’, ‘crime reporter’ and the names of media outlets and news organisations. This process was sufficient to determine whether or not each user was a professional journalist.
3.3.4 Coding for themes I used qualitative content and discourse analysis to investigate how social media users show prejudice on social media during high-profile criminal trials. Using
364 There were a small number of users who identified as a journalist, however, these accounts appeared to be the personal accounts of journalism students who were not posting on behalf of any professional organisation. 365 There were approximately half a dozen users who fell into this category. 366 There were also approximately half a dozen accounts that were ‘no longer available’ on Twitter. 367Hanusch and Bruns, above n 45, 2, 4. 368 ‘Content Analysis: Advantages and Disadvantages’ (n 314) 239.
Twitter trials and Facebook juries 84 thematic analysis, I immersed myself in the tweets and comments so that themes and insights could emerge or flow from the data.369 I did this by carefully reading every tweet and comment, and looking for ‘recurrent themes, topics, or relationships’ to code or label for later analysis and theory building.370 This allowed me to derive coding categories (themes) directly from the texts I was reading,371 rather than using preconceived categories.372 I initially identified themes deductively, on the basis of my research questions and the quantitative outcomes from the content analysis I had already undertaken.373 However, my approach became more inductive as I worked through my data in an iterative way. As new patterns and insights emerged, I revised my themes and codes.374
When deriving themes, I looked not only at the messages themselves in tweets and comments, but also at what could be inferred from the messages.375 This allowed me to understand more than just the presence of prejudice. I did this by taking into account factors such as who was talking, who they were talking to, how they were talking, what they were saying, how they framed concepts, and what sort of language they were using. I took particular note of whether they were using negative or emotive language and what this meant in the context of what was taking place in court at the time the tweet or comment was posted. This also made it possible to better understand
369 Nancy L Kondracki, Nancy S Wellman and Daniel R Amundson, ‘Content Analysis: Review of Methods and Their Applications in Nutrition Education’ (2002) 34(4) Journal of Nutrition Education and Behavior 224 (‘Content Analysis’). 370 ‘Thematic Analysis’ (n 319) 926. 371 Hsieh and Shannon (n 311) 1279. 372 Kondracki, Wellman and Amundson (n 372). 373 ‘Thematic Analysis’ (n 319) 926. 374 Philipp Mayring, ‘Qualitative Content Analysis’ (2000) 1(2) Forum: Qualitative Social Research [12]
85 Research Design and interpret how users show prejudice — how they express it and whether they notice it, respond to it, and amplify (or limit) it in any way.
3.4 Limitations
In this section, I set out the limitations of my research. These limitations relate to how I selected data and the tools I used, Twitter and Facebook, social media in general, and jurors and witnesses.
3.4.1 Data selection and tools My approach to data selection means that some tweets related to my case studies may have been excluded and those generated by bots may have been included. As I used a keyword search strategy to select data, tweets and comments that do not contain the keyword will not be included in my sample. As we do not fully understand the algorithmic curation process on Facebook and Twitter, it is possible my samples are incomplete. My samples may, however, contain tweets or comments generated by bots. I explain each of these in further detail below.
Keyword search: I used keywords and hashtags to establish datasets of the most
visible tweets and comments relating to both trials. However, my samples will not
contain tweets and comments that do not include the keywords or hashtags,376
even if they relate to the trials. Some users will respond to comments about the
trial without including a keyword or hashtag. Some tweets in my sample will be
in response to tweets that do not contain a keyword or hashtag. As none of these
tweets will be included, my data samples may be incomplete.
Private companies and algorithms: As social media data is curated and
controlled by private companies through the use of algorithms, about which we
376 Weller et al (n 284) 100.
Twitter trials and Facebook juries 86
have a limited understanding, it is possible that the information captured does not
represent all the data that exists.
Bots: Tweets or comments may be generated by bots (an internet or web robot
software application that runs automated tasks over the internet). Information
from bots is sometimes identifiable due to its random or unrelated content, the
quantity of posts, and the timing and location of posts. While this information may
not be relevant, I did not explicitly exclude it, as it is conceivable that a bot may
be designed in such a way that it produces and reproduces significant quantities
of prejudicial publicity. In this way, it could become a key point of influence in a
network.
3.4.2 Tools Due to the data collection tools I used, my Twitter sample for the Tostee trial is limited to the Australian Twittersphere. This means that I did not capture tweets by users located outside Australia. My DMI-TCAT sample for the Baden-Clay trial did not capture Day 1 of the trial, so I commenced analysis from the morning of Day 2 when the sample became most complete. I explain each of these limitations in further detail below.
Australian Twitter data (Tostee): My Tostee dataset is limited to the Australian
Twittersphere and so will not have captured tweets about the trials from accounts
that have not been identified as Australian. This will mean tweets posted by any
user whose location is recorded as outside Australia would not have been
captured, even if they are located in Australia. Similarly, some users might have
recorded their location as Australia, even if they are not in fact located in
Australia. In this case, their tweets would have been captured. The same limitation
does not apply to Facebook where all comments, regardless of user location, will
87 Research Design
have been captured as they are linked to the specific news articles I identified in
my sample. This same limitation also does not apply to the Baden-Clay data, due
to the inclusion of DMI-TCAT data which extends beyond the Australian
Twittersphere.
Twitter data (Baden-Clay): My DMI-TCAT dataset did not contain data for Day
1 of the trial (as it was not captured) and my TrISMA data for Day 1 was
incomplete. It has only been possible to extract comprehensive data samples using
TrISMA since August 2015. When extracting data from before this time-frame,
we are limited by the API of the platform. In this study, I sought to back-capture
tweets from June/July 2014 (during the trial), however, the API would only
provide up to 3,200 tweets per Twitter account. This meant that for those accounts
where users tweeted fewer than 3,200 tweets from June/July 2014 to August 2015,
I will have captured all their tweets. However, for those highly-active users who
tweeted many thousands of tweets during that timeframe, I will have been
restricted to capturing only their most recent 3,200 tweets and these likely fell
after the trial dates. This means I may have missed many relevant tweets from
highly-active users. However, these highly-active users will have been included
in the DMI-TCAT dataset, meaning this limitation will be minimised from Day 2
of the trial onwards. On this basis, I excluded Day 1 from coding and analysis. I
selected Day 2 as the starting point, specifically 9:39:01 am,377 11 June 2014, as
this is the time at which the dataset became most complete and these limitations
were minimised.
377 This is the time at which the DMI-TCAT dataset commenced.
Twitter trials and Facebook juries 88
3.4.3 Twitter and Facebook There are limitations in how the results and findings from my Twitter and
Facebook analyses can be used. While they represent the results and findings for my two case studies, they will not necessarily be representative of prejudicial publicity in all criminal trials across Queensland or Australia. They are indicative, however, of a concerning problem and suggest there is sufficient evidence to warrant further investigation and future research. I have set out further details about these limitations below.
Twitter population: Twitter is not representative of the entire population in terms
of quantity or demographic and, therefore, may not be generalisable across the
population. As mentioned previously, at the time of data collection TrISMA was
tracking approximately 2.8 million Australian accounts, from a total Australian
population of almost 24 million.378 Some users may have more than one Twitter
account. Twitter is not representative of the population demographically, as the
disadvantaged and the elderly tend to be under-represented.379 Further, users who
are active users on Twitter may be more visible (and therefore more of a focus),
than those who read but rarely post.380 As this research focusses on the most
visible users, this second issue is less problematic.
3.4.4 Social media in general My research is useful for providing some broad insight into the prevalence and nature of prejudicial publicity on social media in general, however, my analyses do
378 This figure is based on the projected population of 23,937,101 as at 23 November 2015 at 14:50:10 (Canberra time), as generated by the Australian Bureau of Statistics’ population clock. See http://www.abs.gov.au/ausstats/abs%40.nsf/94713ad445ff1425ca25682000192af2/1647509ef7e25faac a2568a900154b63?OpenDocument. 379 Scott Golder and Michael Macy, ‘Social Science with Social Media’ (2012) 40(1) ASA Footnotes 7, 7. 380 Weller et al (n 284) 64–65.
89 Research Design not take into account the fact that some users in my sample may have more than one
Facebook or Twitter account. Nor does it take into account that some users may appear in both my Twitter and Facebook samples. My work does not account for conversations about trials that take place on other social media platforms. Nor does it consider any links contained in tweets or comments that may direct users to other sites where prejudicial publicity may exist. These are explained below.
Unique users or single users with multiple accounts: My research views each
Twitter and Facebook user as a unique user, however, it is possible one person
may have multiple accounts. If a single user with multiple accounts was to post
multiple prejudicial tweets or comments from multiple, seemingly distinct
accounts, in the absence of other identifying information it would be difficult to
establish that those accounts belong to a single user.
Multi-platform users: My research studies the use of Twitter and Facebook by
users on each platform, but does not account for the fact that some users may have
used both platforms in tandem (multi-platform).381
Other social media: My datasets will miss trial-related discussions and
conversations on other social media sites (such as, for example, Instagram,
WeChat or Snapchat), as well as those taking place either offline or online outside
of social media (such as directly on a media organisation’s website). My methods
do not allow me to extract data from a media organisation’s website, however, this
would be a valuable focus for future research.
381 Anabel Quan-Haase and AL Young, ‘Uses and Gratifications of Social Media: A Comparison of Facebook and Instant Messaging’ (2010) 30(5) Bulletin of Science, Technology and Society 350.
Twitter trials and Facebook juries 90
Links: My research does not identify how users link to web resources and non-
social media resources, or how they link to social media on other networks where
prejudicial publicity might exist.
3.4.5 Jurors and witnesses My research provides insight into the types of information empanelled jurors might see, should they be exposed to information about their trial on social media.
However, it is not possible to draw conclusions from my research about whether jurors are actually likely to be influenced by prejudicial publicity on social media. This is because it is difficult to determine whether they have actually been exposed to prejudice and, if they have, whether that exposure is actually likely to influence their deliberations. As the sub judice rule aims not only to prevent juror exposure to prejudicial publications but also to prevent witnesses in trials from being exposed to prejudice,382 my findings may also apply to witnesses.383 For the purposes of scope, however, I have limited my research to the context of jurors.
3.5 Summary
By combining digital methods for data collection and analysis, as well as applying quantitative and qualitative analysis methods, I am able to respond to my sub- questions in a way that provides new insights that cannot be found using traditional legal research methods alone. These insights allow me to assess how well adapted the
Australian sub judice rule is for regulating prejudicial publicity on social media during high-profile criminal trials.
382 Vine Products Ltd v Green [1966] Ch 484, 495 (Buckley J); Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (n 281); Civil Aviation Authority v Australian Broadcasting Corporation (n 281). 383 New South Wales Law Reform Commission, above n 115, 7, 12, Ibid 51-52 [2.50].
91 Murder trial of Gerard Baden-Clay – Twitter Analysis
4 Murder trial of Gerard Baden-Clay – Twitter Analysis
4.1 Overview
In this Chapter, I present my analysis of tweets during the high-profile murder trial of Gerard Baden-Clay.384 I use a combination of content and discourse analysis to understand the prevalence of prejudicial publicity in the dataset, to compare how professional journalists and non-journalists talk about the trial, and to investigate how social media users show prejudice on social media during high-profile criminal trials.
From this analysis, I established that 6% of tweets in my dataset contained prejudicial information. Professional journalists were more factual and neutral in the way they talked about the trial, whereas non-journalists were more opinionated and emotive. Although they were more factual, professional journalists still focussed on the more sensational aspects of the trial and, on occasions, they could have been more careful with attributing tweet content and with the way they compiled information.
Both professional-journalists and non-journalists tended to focus on the prosecution narrative in their tweets, while largely ignoring the defence narrative. I raise concerns about the potential for cumulative negative sentiment to cause prejudice.
Overall, the sub judice rule appears to be largely effective in regulating the behaviour of professional journalists, but less so for non-journalists. The law is limited, however, to preventing prejudice in individual publications that are published by individual publishers. As a result, it cannot address the potentially prejudicial effects
384 R v Baden-Clay [2014] QSC 154.
Twitter trials and Facebook juries 92 of multiple publications (tweets), posted by multiple social media users, that collectively demonstrate a prosecution bias or cumulative negative sentiment.
4.2 The trial
Few criminal trials in Brisbane’s history have attracted as much media attention as the 2014 murder trial of Gerard Baden-Clay. On 15 July 2014, Baden-Clay was convicted of the murder of his wife, following a lengthy police investigation and high- profile trial. The trial attracted such large crowds that the Supreme Court needed to introduce a ticketing system and open overflow courtrooms in order to accommodate those in the long queues.385 This extraordinary interest in the case and the prolific media coverage prompted the trial judge — Justice John Byrne — to invoke a rarely used power to question prospective jurors about their attitudes towards the defendant.386 This power had long existed, but the courts only first accepted an application for its use in 2013 at the commencement of another high-profile trial.387
At the end of the four week Baden-Clay trial, the jury found him guilty of murder. However, on 8 December 2015, the Queensland Court of Appeal unanimously decided to downgrade the murder conviction to manslaughter.388 The Court of Appeal found it was reasonably open for the jury to find Baden-Clay had killed his wife, but
385 The ticketing system ensured those members of the community wanting to observe the trial were permitted into the courtroom on a first-come, first-served basis. Those remaining in the queues were ushered into overflow courtrooms where they could view the trial in real-time on large screens: David Murray, ‘Baden-Clay Murder Case Has Tickets to Get in, Security to Stop Fights and Crowds Flocking to Courts of Law Building in Brisbane’, The Courier Mail (online), 28 June 2014
93 Murder trial of Gerard Baden-Clay – Twitter Analysis not that he had intended to do so. This decision was followed by extraordinary public outcry. In response to the Court of Appeal decision, the Queensland Director of Public
Prosecutions lodged an appeal to the High Court of Australia seeking a reinstatement of the murder conviction. On 26 July 2016, the appeal was heard by the High Court and on 31 August 2016 it handed down its decision to allow the appeal and reinstate the original verdict of murder.389
During the trial itself, the extraordinary publicity surrounding this case could not be ignored by the Court and counsel. Criticism of the media’s ‘ill-informed comment[s]’390 and concerns about sensationalised coverage prompted both the trial judge and defence counsel to remind jurors that the trial was ‘not a soap-opera’ and to ignore media headlines about the guilt of the accused. While it is not possible to determine with certainty whether media reports during the trial played a role in the jury’s verdict,391 some commentators have suggested that the media coverage may have been influential. When the Queensland Law Reporter published its summary of the Court of Appeal decision, the editor (an experienced barrister and Queen’s
Counsel) expressed the view that ‘it would be far from unreasonable to assume that the amount of pre-trial publicity which this matter received had no small impact on the verdict such that the correction by the Court of Appeal was both necessary and inevitable’.392 Although media attention did not actually form the basis of the first or second appeals, this case raises clear questions about the potential for prejudicial publicity in the media — including on social media — to influence jurors.
389 R v Baden-Clay (2016) 258 CLR 308. 390 Roger Derrington, ‘Editors’ Selection of New Cases: R v Baden-Clay [2015] QCA 265’ [2015] (49) QLR Queensland Law Reporter 10. 391 I do not contend the accused’s right to a fair trial was compromised in this case. My conclusions are only general and made for the purpose of comparison with future case studies. 392 Derrington, above n 20, 10.
Twitter trials and Facebook juries 94
4.3 Prevalence of prejudicial publicity
To understand the prevalence of prejudicial publicity in my dataset, I first created a visualisation of overall twitter activity during the four-week trial. This allowed me to contextualise my analysis and understand what was taking place in court during the peaks and troughs of tweet activity. I then coded my random sample for prejudicial publicity and user type.
I used Tableau393 to create a visualisation of my total Twitter dataset (33,067 tweets). As set out in Figure 4.1, this visualisation represents the number of tweets posted on each day of the trial, up until one minute before the verdict was delivered.
Figure 4.1: Baden-Clay Twitter analysis (total Twitter dataset) - tweet activity during the trial.
Tableau allowed me to pinpoint the underlying data for peaks and troughs in
Twitter activity. By reading the underlying tweets — as well as by both attending part of the trial in person and reading media reports about the trial — I was able to summarise what was taking place in court during key points in my dataset. I found that activity peaked when key witnesses gave evidence, when the accused took the stand,
393 Tableau is a visual analytics tool: see https://www.tableau.com/.
95 Murder trial of Gerard Baden-Clay – Twitter Analysis during closing arguments, and in the lead up to the verdict. I saw least activity on those days when court was not sitting and on weekends.
To determine the prevalence of prejudicial content in my dataset, I coded my random sample of tweets using my coding scheme set out in Chapter 3. After coding tweets as containing low-level, high-level or no (none) prejudicial publicity, I found that 6% of tweets in my random sample contained potentially prejudicial content. As set out in Figure 4.2, this comprised 4.9% of tweets containing low-level prejudice and
1.1% of tweets containing high-level prejudice. Based on my confidence level and confidence interval, this meant I could be 95% confident that 5-7% of my total twitter dataset contained potentially prejudicial publicity.
PREJUDICIAL CONTENT High 1.1% Low 4.9% None Low High None 94.0%
Figure 4.2: Baden-Clay Twitter analysis (random sample) - 6% of tweets contained prejudicial content.
The majority of tweets in my random sample were non-prejudicial, but 6% were problematic. It is difficult to know how concerned we should be about this result. It has been argued that ‘broad, asynchronous, lightweight and always-on communication systems such as Twitter are enabling citizens to maintain a mental model of news and
Twitter trials and Facebook juries 96 events around them’.394 It is difficult to know how this ‘ambient journalism’395 may affect jurors if their ‘mental model’ contains prejudicial information about an accused.
There are no comparable studies of trials with which I can compare this finding.
Those undertaken previously are concerned with traditional news coverage
(newspapers, radio and television) and focus on the frequency of crime reporting, for example, rather than prejudicial publicity.396 Alternatively, I could have compared my
Twitter data with mainstream news reports that were published during the trial.
However, as traditional media continues to be subject to editorial controls and as professional journalists appear to largely comply with the sub judice rule,397 there appeared to be limited value to this comparison.
Overall, as 6% of tweets containing prejudice is non-negligible, it appears to be significant enough to warrant further investigation. In my next chapter, I compare this result with those in my second case-study.
4.4 Professional journalists versus non-journalists
To understand how professional journalists and non-journalists talk about trials,
I coded my random sample based on whether the user (publisher) who posted each tweet was a professional journalist or non-journalist. As set out in Figure 4.3, I found
394 Hermida (n 166) 301. 395 Ibid. 396 Imrich, Mullin and Linz (n 274) 96; GE Antunes and PA Hurley, ‘The Representation of Criminal Events in Houston’s Two Daily Papers’ (1978) 55 Journalism Quarterly 756; D Humphries, ‘Serious Crime News Coverage and Ideology: A Content Analysis of Crime Coverage in a Metropolitan Paper’ (1981) 27 Crime and Delinquency 191; M Ryan and D Owen, ‘A Content Analysis of Metropolitan Newspaper Coverage of Social Issues’ (1976) 53 Journalism Quarterly 634; TF Simon, F Fico and S Lacy, ‘Covering Conflict and Controversy: Measuring Balance, Fairness, Defamation’ (1989) 66 Journalism Quarterly 427; JR Dominick, ‘Crime and Law Enforcement in the Mass Media’ in C Winick (ed), Deviance and Mass Media (Sage, 1978). 397 Hews and Suzor (n 272) 1623.
97 Murder trial of Gerard Baden-Clay – Twitter Analysis that approximately two-thirds of tweets in the sample were posted by professional journalists (64.8%) and the remaining third were posted by non-journalists (35.2%).
USER TYPE non- journalists professional journalists 35.2% non-journalists professional journalists 64.8%
Figure 4.3: Baden-Clay Twitter analysis (random sample) - most tweets were posted by professional journalists.
To compare how professional journalists and non-journalists tweeted about the trial, I combined my results from coding for prejudice and coding for user type. This is set out in Table 4.1.
Professional Total Random Non-journalists Journalists Sample 85.3% 98.7% 94.0% None 2230 tweets 4751 tweets 6981 tweets 12.0% 1.0% 4.9% Low 313 tweets 48 tweets 361 tweets 2.7% 0.3% 1.1% High 71 tweets 14 tweets 85 tweets 100.0% 100.0% 100.0% Total 2614 tweets 4813 tweets 7427 tweets Table 4.1: Baden-Clay Twitter analysis (random sample) - results after coding for prejudice and user type.
Twitter trials and Facebook juries 98
I found that 14.7% of tweets posted by non-journalists in my random sample contained prejudicial content (comprising 12% low-level prejudice and 2.7% high- level). This compared with only 1.3% of tweets posted by professional journalists
(comprising 1% low-level prejudice and 0.3% high-level). This data is represented in
Figure 4.4. By generalising across the total Twitter sample, I could be 95% confident that 13.7-15.7% of tweets posted by non-journalists contained prejudicial content, compared with only 0.3-2.3% of tweets posted by professional journalists.
Non-journalists Professional Journalists 1.3%
14.7%
85.3% 98.7%
Prejudicial Non-prejudicial Prejudicial Non-prejudicial
Figure 4.4: Baden-Clay Twitter analysis (random sample) - non-journalists posted more prejudicial tweets than professional journalists.
These results suggested that non-journalists posted over 11 times (11.3) more prejudicial tweets than professional journalists. By analysing these tweets more closely
I was able to compare how professional journalists and non-journalists talked about the trial.
4.4.1 Professional journalists more factual, but non-journalists more emotive I observed that professional journalists were more factual and that non- journalists were more emotive. Likely as a result of the way mainstream media organisations have incorporated the law into their normal workflows, tweets posted by professional journalists mostly contained legally compliant news reports and rarely included prejudicial information. Professional norms mean that even on Twitter, journalists are expected to be objective and keep their personal opinions to
99 Murder trial of Gerard Baden-Clay – Twitter Analysis themselves.398 In my sample they maintained this professionalism — they appeared to be careful with their language choices so as to present factual information about the trial in objective or neutral terms, and avoided the inclusion of opinionated or offensive terminology. Overall, they appeared to behave ethically by using balance, fairness and impartiality when presenting facts.399
As identified earlier, only a very small percentage of tweets by professional journalists in my sample (0.3–2.3%) appeared to meet the technical requirements of prejudice. This suggests they ‘did their job’ — by ‘gathering (reporting) and processing (writing) accurate and important information so it [could be] disseminated to a wider audience’.400 Although they were using a participatory form of communication, they ‘held fast to their traditional functions’.401 That professional journalists are continuing to behave professionally in the age of social media is admirable, particularly given that online culture runs ‘counter to the notion of professionalism in that it specifically devalues expertise by placing a premium on newness and controversy rather than on public service’.402
By contrast non-journalists in my sample generally did not report the news — they talked about and responded to it. The shift towards ‘voluntary engagement’ means
398 Lasorsa, Lewis and Holton (n 166) 23; Michael Schudson and Chris Anderson, ‘Objectivity, Professionalism, and Truth Seeking in Journalism’ in The Handbook of Journalism Studies (Routledge, 2009) 108; Bill Kovach and Tom Rosenstiel, The Elements of Journalism: What Newspeople Should Know and the Public Should Expect (Three Rivers Press (CA), 2014) (‘The Elements of Journalism’). 399 John Pavlik, Gary Morgan and Bruce Henderson, ‘Information Technology: Implications for the Future of Journalism and Mass Communication Education’ [2001] Journalism and Mass Communication Education: 2001 and Beyond 16. 400 Singer (n 129) 144; Brian S Brooks et al, News Reporting and Writing (St Martin’s Press, 7th ed, 2001); Christopher Scanlon, Reporting and Writing: Basics of the 21st Century (Harcourt College Publishers, 2000); Melvin Mencher, News Reporting and Writing (McGraw-Hill College, 8th ed, 1999). 401 Lasorsa, Lewis and Holton (n 166) 20. 402 Bill Kovach and Tom Rosenstiel, The Elements of Journalism: What Newspeople Should Know and the Public Should Expect (Crown Publishers, 2001).
Twitter trials and Facebook juries 100 that non-journalists are able to play a role in constructing meaning around news,403 such that they become part of the news experience through interaction404 — a reality that was not possible in traditional media. Twitter enables non-journalists to be a type of citizen journalist who need not have recourse to institutional norms and this allows them to undermine the gatekeeping function of journalists.405 In my case study, the discourse of non-journalist ‘news participants’ lacked the careful language of professional journalists and was, instead, conversational, informal, and sometimes included coarse or offensive vocabulary. This type of discourse is not restricted to trials — the use of swearing, impulsivity, and high-arousal emotions (anger and frustration) are common in shaping discursive participation online.406
Compared with professional journalists, non-journalists were more inclined to tweet statements about the guilt of the accused, to be critical or disparaging, and to respond to news about the trial in highly emotionally charged ways. Examples of tweets by non-journalists that strongly implied guilt or criticism included claims that the accused was ‘scum of the earth’, ‘guilty as hell’, ‘full of shit’, ‘a lying killer’ or
‘murdering scum’.407 Other tweets implied the accused was guilty by calling for a conviction, claiming there was plenty of evidence to convict him, hoping the accused would be made to suffer in prison, and by demanding the jury reach a guilty verdict.
403 Mark Deuze, ‘What Is Multimedia Journalism?’ (2004) 5(2) Journalism studies 139, 147. 404 Jo Bardoel and Mark Deuze, ‘“Network Journalism”: Converging Competencies of Old and New Media Professionals.’ (2001) 23(2) Australian journalism review 91, 94 (‘“Network Journalism”’). 405 Hermida (n 166) 300. 406 K Hazel Kwon and Daegon Cho, ‘Swearing Effects on Citizen-to-Citizen Commenting Online: A Large-Scale Exploration of Political versus Nonpolitical Online News Sites’ (2017) 35(1) Social Science Computer Review 84, 87–89 (‘Swearing Effects on Citizen-to-Citizen Commenting Online’); Ofir Turel and Hamed Qahri-Saremi, ‘Explaining Unplanned Online Media Behaviors: Dual System Theory Models of Impulsive Use and Swearing on Social Networking Sites’ (2018) 20(8) New Media & Society 3050, 3052 (‘Explaining Unplanned Online Media Behaviors’). 407 The user accounts of many tweets are able to be re-identified by posting the tweet text into an internet browser such as Google. Due to ethical considerations, I have generally removed the text of tweets from my discussion and relied on a summary instead.
101 Murder trial of Gerard Baden-Clay – Twitter Analysis
In addition to problematic language choices, non-journalists rarely provided the attribution that would be required to negate these types of prejudicial content.
The considerable difference in the percentage of prejudicial tweets posted by non-journalists compared with professional journalists reflects not only journalistic training and codes of conduct,408 but also the remarkably different way that journalists and non-journalists tweet about ongoing trials. Given the low proportion of apparently prejudicial tweets by professional journalists, the sub judice rule appears to be largely effective in shaping the way they report on and communicate about the news.
However, it appears to be less effective for non-journalists, who may not be familiar with the law409 or who may lack the skills or desire to moderate the content of their tweets. These factors — combined with different styles of discourse and issues of attribution — meant tweets by non-journalists were much more likely to meet the technical requirements of prejudice.
4.4.2 Professional journalists still focus on sensational aspects of trial Although professional journalists were significantly less likely to post prejudicial tweets than non-journalists in my sample, they still tended to focus on the more sensational aspects of the trial. This focus may have the potential to influence jurors, particularly when sensational key messages are reduced to 140 character tweets.
The immediacy with which details from the courtroom were distributed to the public may also have an effect.
Even where tweet content was attributed, I found evidence of a strong tendency for media reports by professional journalists to focus on the more sensational aspects of the case and the highly emotive claims of the prosecution. It may be that there is
408 Singer (n 129) 141, 145. 409 Johnston et al, above n 2, 4.Johnston et al (n 181) [2.3].
Twitter trials and Facebook juries 102 nothing peculiar about tweets in this regard, as there was also some focus on the sensational aspects of stories in traditional media.410 It has always been in the interests of the media to ‘make their presentation of any news event of optimal interest to maximal numbers of consumers’,411 although recent research suggests there is a growing pressure upon journalists towards publishing sensational news.412 This pressure can be difficult to balance with the principle of fair and accurate reporting. It may also be tempting to push the boundaries of the law given prosecutions for sub judice contempt are relatively rare.413
There appears to be a fine line between a fair and accurate report and one that is distorted or exaggerated, particularly when information is presented as a 140 character narrative. Two examples of legally compliant but potentially prejudicial tweets include: ‘Prosecution: GBC efficient and effective killer’ and ‘Prosecution: whatever method was used to kill Allison it was two things, efficient and effective’. Regardless of attribution these tweets appear to assert guilt or criticise the accused, but it was true the prosecution made these claims and the principle of open justice means it is lawful to fairly and accurately report what is stated in court.
However, it is possible that a juror who hears the prosecution make a claim in court may be further influenced if the claim is also amplified by repeated publication in the media, as repetition contributes to how a person (juror) remembers what they have seen or heard.414 This may especially be the case where that repetition hones in
410 Jeff Brand and Mark Pearson, ‘The Newsroom vs. the Lounge Room: Journalists’ and Audiences’ Views on news. [Paper Presented after Peer Review to the Communications Research Forum (2001: Canberra).]’ (2001) 23(2) Australian Journalism Review 63, 76 (‘The Newsroom vs. the Lounge Room’). 411 Australian Law Reform Commission (n 76) 12. 412 Folker Hanusch, ‘Transformative Times: Australian Journalists’ Perceptions of Changes in Their Work’ (2015) 155(1) Media International Australia 38, 44 (‘Transformative Times’). 413 Bosland and Townend (n 183) 1266. 414 Wilcox (n 222) 34.
103 Murder trial of Gerard Baden-Clay – Twitter Analysis on the most prejudicial aspect of the claim — efficient and effective killer — without including the context that was evident in the courtroom. A person (juror) may forget the specifics of the media they have seen, but they are likely to retain a positive or negative impression which is ‘quite likely to result in a vague impression of guilt or innocence related to an accused person’.415 Technically, the journalists in the examples above are reporting the ‘fact’ that the prosecution made a claim. But it is somewhat difficult to accept that readers of these claims, presented as facts, would necessarily be able to quickly distinguish the content as an argument made by the prosecution from a statement of fact. It may be that either the attribution or the timing of the types of tweets — coming after the jury has already heard the allegations from the prosecution
— may mean that they are not likely to prejudice the trial, but this assumption seems potentially suspect.
For professional journalists, Twitter (and social media generally) offers a novel approach to distributing information from the courtroom.416 It is used by professional journalists not only to report news but also for self-promotion and branding,417 as well as amplifying and advertising tweets and stories of their own, as well as those from colleagues, associated broadcasters, and mainstream networks. The economy of
Twitter encourages users to craft tweets that are likely to be retweeted and there is a clear reputational incentive for users to write interesting tweets, but sometimes these may go too far and be sensational. When linking to news articles, there is a direct financial incentive for journalists418 and news organisations to use headlines and tweets that are likely to catch attention, draw clicks, and be retweeted in large volumes.
415 Australian Law Reform Commission (n 76) 13–14. 416 Hermida (n 166); Lasorsa, Lewis and Holton (n 166). 417 Hanusch and Bruns (n 35). 418 Singer (n 129) 154.
Twitter trials and Facebook juries 104
I see in my data that the most retweeted tweets are also those containing the most sensational information, evidence or claims as made by the prosecution or witnesses.
In most cases, though, these are carefully crafted to avoid being prejudicial at law.
New challenges are also created by the immediacy and speed with which tweets
(or other social media posts) are generated and by their ability to convey information in near real-time.419 In my sample, a handful of journalists were so detailed in their reporting that they tweeted almost continuously. Their moment-by-moment reports saw them each post thousands of tweets throughout the trial, allowing viewers to experience the courtroom proceedings almost live. Verbatim courtroom details were tweeted to potentially large audiences of thousands, or even hundreds-of-thousands, of followers who could instantly respond both by participating in conversations and interacting with the ‘story’, prejudicial or otherwise, as it unfolded. This immediacy and volume of tweets from the courtroom allowed followers to engage with the murder trial at a pace and in detail not generally provoked by traditional media reporting. This also created the potential for jurors to be exposed to or participate in conversations that are quite different from those typically seen in traditional media coverage.
Non-journalists also focussed on the sensational aspects of the story, but this may be less surprising given their purpose in tweeting was generally to engage in commentary about the trial rather and not to report the news. They also engaged in rapid tweeting but, at least in my sample, their tweets were not those from persons present in the court room. They were highly interested members of the public who engaged actively with the news story.420
419 Ibid 152. 420 Deuze (n 406) 147; Bardoel and Deuze (n 407) 94; Hermida (n 166) 300.
105 Murder trial of Gerard Baden-Clay – Twitter Analysis
4.4.3 Professional journalists should take care with attribution and compilations While it is true that professional journalists were responsible for only a small number of prejudicial tweets, there were occasions when they could have been more careful to avoid criminal liability. Some professional journalists could have made better use of attribution,421 while others needed to be more thoughtful in how they compiled information from the courtroom. Due to these concerns, I could not simply conclude that all tweets by professional journalists were benign and unlikely to influence jurors. These concerns are also reinforced by recent examples of professional journalists using social media to post prejudicial information — one where a professional journalist was convicted of sub judice contempt422 and another where a large number of professional journalists may face prison for contempt of court (for breaching a suppression order).423
Primarily, journalists ought to ensure they use attribution to make clear they are reporting on facts rather than expressing an opinion, as attribution may help to distinguish a fair report from one that is unfair. This also ensures the immunity of fair and accurate reporting will apply to what may otherwise be prejudicial statements.424
This is clearly a difficult task given tweets were restricted at that time to 140
421 New South Wales Law Reform Commission (n 116) 289. See also R v Scott and Downland Publications Ltd (n 110) 673 (Menhennitt J). This case may provide further support for the proposition that attribution ought to be included. Justice Menhennitt suggests that for a report to be ‘understood by the ordinary reader to be a report of judicial proceedings’, it may be required to ‘state it is a report of judicial proceedings’ or to explain that ‘what is stated was given as evidence in judicial proceedings’. 422 Crothers (n 261) 7. 423 Adam Cooper, ‘DPP Moves to Jail Dozens of Editors, Journalists over Reports after Pell Verdict’, The Age (26 March 2019)
Twitter trials and Facebook juries 106 characters, but the law still applies regardless of this restriction. Almost all tweets by professional journalists that I coded as prejudicial contained direct reports of claims made in court, but without attribution. These included claims that the accused violently killed the victim and dragged her body to the car.
In these instances, professional journalists may have thought they were tweeting bare-fact reports from the courtroom, but there are fine distinctions in what constitute actual facts in evidence and these add complexities to journalistic reporting. This trial relied heavily on circumstantial evidence — there were no eyewitnesses, no closed circuit television footage and no clear forensic conclusions. This meant that any claims made about how the victim was killed, or about the disposal of her body, could not be accepted as actual facts until the time the jury — the triers of facts — handed down its verdict. Only then — and only if the jury delivered a guilty verdict — could the court accept the facts presented in evidence as true,425 subject to any exceptions the court might identify in its sentencing remarks.
At the time these tweets were posted, these claims were not actual facts — they were only assertions put to the Court by the prosecution or expert witnesses. In repeating these assertions through unattributed tweets, some journalists gave the impression they were making personal claims about the guilt of the accused. For a bare-fact report, journalists would need to have identified the maker of the claims — the bare-fact can only be that the prosecutor made a particular claim, not the truth of the claim itself.
425 For example: see DPP v Pell (Sentence) (n 354); ‘Sentencing Remarks in DPP v George Pell - County Court of Victoria’ (n 354); ‘This Is What a Victorian Judge Told George Pell as He Sent Him to Jail for Abusing Two Choirboys’ (n 354). The judge in Pell’s sentence stated: “You fall to be sentenced on a basis consistent with the jury verdict on your trial. It was common ground at the plea that this effectively means that you are to be sentenced on the basis of the account of the victim J who gave evidence at trial. Your counsel accepts this.”
107 Murder trial of Gerard Baden-Clay – Twitter Analysis
In some instances, journalists in my sample compiled a number of prosecution claims and compressed them into a single tweet, but sequenced the evidence differently to how it was presented in court. For example: ‘Prosecution: Gerard
#badenclay was looking for a way out & driven by love & money to efficiently & effectively kill his wife’. The purpose of this tweet appears to be to summarise the arguments made by the prosecution and witnesses, and this is difficult to do in 140 characters. As a result, this tweet included only an abstract summary of the evidence without the broader context in which each part of the evidence was given, and it is presented as a cohesive sequence even though the evidence was presented in a more sporadic way in court. In this way it has the potential to be misleading and prejudicial, as it creates a plausible and potentially persuasive narrative about the defendant’s motives and guilt. Yet, as this tweet includes attribution, and as the various claims were in fact made in court, it is probably not strictly against the law. For this reason I coded it as non-prejudicial at law, but it may still be prejudicial in practice. It is clearly the role of journalists to create narratives from the information before them, but it may be that the concentration of key themes or messages into 140 characters could create a more distorted or exaggerated view than traditional news reporting. If so, it is possible the presentation of information in this way could actually influence jurors if they have access to it.
It is important for journalists to keep in mind that even a single social media post has the potential to breach the sub judice rule or to cause a trial to be aborted.426
426In 2015, a jury was discharged and trial abandoned because of a single post on Instagram (and a link on Twitter to the same Instagram post) posted from an account appearing to be that of one of the co-accused’s barristers: Louise Hall, ‘How Barrister Charles Waterstreet Caused Rogerson, McNamara Trial to be Aborted’, The Sydney Morning Herald (online) 15 June 2016
Twitter trials and Facebook juries 108
Overall, it is important that they identify the maker of any claims made in their tweets and take care when creating narratives from the information before them. Compiling courtroom evidence and sequencing it differently to how it was presented in court may be misleading and prejudicial even where attribution is included.
It is clear that these same principles also apply to non-journalists, but as the purpose of their tweets is generally not to report the news, they may be less likely to think it necessary to qualify tweet content through attribution. Many may also be unaware of the ‘fair and accurate reporting principle’ and how it applies to them, as non-journalists.
4.5 Prejudice on social media
By reading and analysing tweets I was able to better understand how social media users show prejudice on social media during high-profile criminal trials. Social media users includes all users — both professional journalists and non-journalists — although given the majority of prejudicial tweets were posted by non-journalists, my observations more specifically apply to them. For this first case-study, my discourse analysis was less comprehensive than for my second case study. In this trial, I focussed mostly on how tweets discussed the prosecution or defence argument, or whether they expressed positive or negative sentiment.
4.5.1 Prosecution bias Using this binary style of analysis, I found more tweets focussed on the prosecution narrative than on the defence narrative, and this could be cause for concern if it is a pattern that plays out generally across criminal trials. I note this would be
involved in the trial, but it does make clear how a single post on social media can result in a negative outcome for the trial process and the criminal justice system.
109 Murder trial of Gerard Baden-Clay – Twitter Analysis equally concerning if the bias was pro-defence, but that was not the case in my datasets.
There was a distinct trend for tweets in my sample to accept and reinforce the prosecution’s theory of the case and this is perhaps more worrying than the presence of technically prejudicial information. The defence’s narrative was largely ignored in reporting, by comparison, which may lead to a more general public sentiment that is weighted against the defendant. The theory of the prosecution’s case as expressed in its line of questioning and assertions obviously reflects a guilty narrative, but these are also the more interesting and sensational aspects of the story. This is also the type of information that is made ‘most readily available to the press … through policing authorities and the prosecution’.427 The danger then is that ‘the media will present and re-present the same side of the litigation, not only forming people’s first impressions but reinforcing and revivifying them’.428 On Twitter, it is these assertions that may be disproportionately amplified by first the media and then the general public. The net result, in my sample, shows a social media discourse that is one-sided and biased towards the prosecution.
I identified this bias by looking particularly at the types of content that were amplified through retweets and found that, during the trial, the most retweeted posts were those recounting evidence from the prosecution or discussing prosecution style arguments. Twitter users responded slowly (if at all) to less-sensational evidence, especially where it supported the defendant or had a tendency to weaken the prosecution case. The most highly amplified claims were those discussing evidence about alleged fingernail scratches on the accused’s face and particular species of leaves that were found in the victim’s hair — evidence that proved pivotal in establishing the
427 Australian Law Reform Commission (n 76) 29. 428 Ibid.
Twitter trials and Facebook juries 110 accused’s guilt. Highly amplified tweets also included claims about the accused’s affairs, recordings of his conversations with police (during which it was claimed he had lied), and disparaging descriptions of him such as ‘murdering scum’.429
Questions asked by the prosecution and witness responses were tweeted in greater volumes and were more likely to be amplified, even when clarifications or contrary evidence were later provided. For example, tweets including claims made by a witness (a neighbour) that they heard screams coming from the direction of the
Baden-Clay home on the night the victim went missing were amplified in Twitter conversations. Yet there was minimal amplification when another neighbour testified the screams were those of her teenage daughter who, on the same night, had been frightened by a cobweb. Although jurors are physically present when evidence is presented in court, it is difficult to know whether the amplification or concentration of prosecution claims — in the absence of defence counter-claims — might distort the evidence, or a juror’s recollection of the evidence,430 such that they are influenced in favour of the prosecution.
This amplification of the prosecution’s assertions meant the majority of tweets supported the notion of guilt or were neutral as to guilt. I saw few retweets of claims defending the accused, other than neutral statements such as those reporting that the accused continued to maintain his innocence or that counsel was summing up the case for the defence. In this way the case for the defence was muted, with very few people reporting the accused defending himself and even fewer voicing the possibility he was not-guilty. It could be argued this disproportionate focus on the prosecution narrative
429 The user accounts of many tweets are able to be re-identified by posting the tweet text into an internet browser such as Google. Due to ethical considerations, I have generally removed the text of tweets from my discussion and relied on a summary instead. 430 Australian Law Reform Commission (n 76) 11–31.
111 Murder trial of Gerard Baden-Clay – Twitter Analysis was simply because the evidence pointed towards the guilt of the accused. However, this case was far from straightforward.
It is fair to say the weight of public sentiment was against the defendant in this case. He was a divisive figure who, on his own admission, had lied to his wife for years, engaged in multiple affairs, and borrowed hundreds of thousands of dollars from friends and acquaintances which, the evidence suggests, he could not repay. When his wife was initially missing he presented in the media as a grieving husband seeking his wife’s return, although once he was charged with her murder it became clear this may have been an act. The accused was also considered by many to be arrogant, selfish, and oddly reliant on his genetic connection to his great-grandfather, Lord Baden-
Powell,431 for notoriety and prestige.
The accused chose to take the stand to give evidence in this trial, and this appears to have provided an opportunity for the public to criticise him and to reiterate claims made by the prosecution in cross-examination. This overall negative public sentiment was reflected in Twitter conversations in which members of the public made strong claims about their dislike for him. Many focussed on their belief in his guilt and expressed their anger by attacking him and agitating for a guilty verdict and life-long jail sentence, or even wishing harm would come to him. Naturally, many also expressed their grief for the victim, her three daughters and her family.
Although the weight of public sentiment was against the accused, the same could not be said for the strength of the evidence against him. This was a trial in which the evidence was wholly circumstantial. As mentioned earlier, there were no eye-
431 Lord Baden-Powell was a celebrated member of the British Army in the late 1800s and early 1900s, who rose to the rank of Lieutenant-General. He is particularly well known for his role in founding the scout movement and was the first Chief Scout of The (Boy) Scout Association: see John Fox, ‘Lord Robert Baden-Powell (1857–1941)’ (2013) 43 Prospects 251, 255, 260.
Twitter trials and Facebook juries 112 witnesses or closed-circuit television footage, and even the victim’s cause of death could not be determined conclusively. The uncertainty surrounding the facts was further highlighted by the fact the matter went on appeal to the Court of Appeal and then to the High Court.432 Given the complexities involved in these appeals, it is apparent the case was not clear-cut. This makes it seem unlikely that the tenor of the evidence alone could be responsible for the largely negative sentiment seen on Twitter.
This focus on the prosecution narrative may be caused by general public interest, the unusual circumstances of the case, or sympathy for the victim and her family, but there is also research that suggests that the power of bad events is much stronger than good ones.433 Coined ‘if it bleeds, it leads’ in the media,434 a viewer’s response to bad news is stronger than good.435 It may be that as the prosecution narrative is the bad story — the accused murdered his wife — that the public’s response may be stronger than the defence’s good story. It may be that this has contributed to social media users focussing on the prosecution narrative, rather than the defence.
This pro-prosecution focus and an absence of the defence narrative is a clear cause for concern. Perhaps it is not surprising that some Twitter users themselves raised concerns about negative sentiment in the media by levelling criticism at the
432 R v Baden-Clay (2016) 258 CLR 308. 433 Roy F Baumeister et al, ‘Bad Is Stronger Than Good’ (2001) 5(4) Review of General Psychology 323. 434 Matthew Robert Kerbel, If It Bleeds, It Leads: An Anatomy of Television News (Westview Press, 1st Paperback ed., 2001) (‘If It Bleeds, It Leads’); Lyn Barnes and Elesha Edmonds, ‘If It Bleeds, It Leads? Changing Death Coverage in the New Zealand Herald’ (2015) 21(2) Pacific Journalism Review: Te Koakoa 162 (‘If It Bleeds, It Leads?’); Crystale Purvis Cooper and Debra L Roter, ‘“ If It Bleeds It Leads”? Attributes of TV Health News Stories That Drive Viewer Attention.’ (2000) 115(4) Public Health Reports 331 (‘“ If It Bleeds It Leads”?’); Kestutis Kveraga et al, ‘If It Bleeds, It Leads: Separating Threat from Mere Negativity’ (2014) 10(1) Social cognitive and affective neuroscience 28 (‘If It Bleeds, It Leads’); Ineke Stoop, ‘If It Bleeds, It Leads: The Impact of Media-Reported Events’ [2007] Measuring Attitudes Cross-Nationally. Lessons from the European Social Survey, London, SAGE 95 (‘If It Bleeds, It Leads’). 435 Adam DI Kramer, Jamie E Guillory and Jeffrey T Hancock, ‘Experimental Evidence of Massive- Scale Emotional Contagion through Social Networks’ (2014) 111(24) Proceedings of the National Academy of Sciences 8788, 8790.
113 Murder trial of Gerard Baden-Clay – Twitter Analysis sensational headlines, the hype and the bias. Some users questioned the impact of social media reporting on the trial process, suggesting jury trials are no longer workable or relevant, criticising the ‘media circus’ around the trial and claiming the verdict was disturbing given the potential for jurors to be biased. Certainly this data suggests there is some legitimacy to these various concerns436 and only further research will reveal whether a pro-prosecution bias on social media is replicated for other criminal trials. In my next chapter, I compare these findings with those in my second case-study.
4.5.2 Cumulative negative sentiment The collective weight of low-level prejudicial or negative information across multiple posts on social media may also work to influence juror impartiality. Prejudice caused by cumulative negative sentiment tends to occur in two ways, by contextualising or sequencing, and by repetition or amplification. Many tweets in my sample seem individually innocuous by legal standards, but when they are contextualised by or sequenced with other posts, they may together form a narrative that is prejudicial in effect. Similarly, isolated negative tweets might pass with little effect, but when hundreds or thousands of negative tweets are posted en masse, their repetition or amplification may result in prejudice.
While jurors may forget specific, literal details of what they have seen, impressionistic memory means they may recall a positive of negative impression that is indicative of guilt or innocence related to the accused.437 At law, jurors should not have read tweets during the Baden-Clay trial, but studies suggest at least some jurors
436 This data legitimates concerns in a general sense only; I do not contend the jury was biased in this case. 437 Australian Law Reform Commission (n 76) 13–14.
Twitter trials and Facebook juries 114 are likely to have seen or engaged with media coverage about the trial,438 possibly including Twitter. Jurors would not have needed to read many tweets to recognise the collective sentiment in my dataset that was largely critical of the accused and reflected a broad community belief in his guilt. It is this ‘collective effect of commentary on a case’ that may constitute prejudice.439
Contextualising and sequencing had the potential to reinforce a guilty narrative, as the minutiae of evidence in this trial were conveyed through large numbers of legally compliant tweets. Discussions included claims about: the victim’s depression; the couple’s ‘non-existent sex life’; the accused’s affair and large debts; his mistress’ anger; his desire to ‘wipe the slate clean’; loud screams coming from the couple’s home the night the victim went missing; traces of blood in the car which matched the victim’s DNA; fingernail scratches on the accused’ face and grazes on his chest; and his calls to his insurer after his wife’s death. Presented as actual facts, these are the aspects of the evidence that most people following the trial would have seen on social media. These facts present a compelling guilty narrative — the motives, screams, blood and injuries from a struggle are all consistent with murder.
These are also the types of information that were most likely to be repeated and amplified in my sample, given their pro-prosecution focus, while defence arguments were more likely muted. For this reason, few social media users would have been aware of significant evidentiary developments that weakened the prosecution case.
They may not have known that the screams and the phone call about the insurance claim were ultimately unable to be relied upon by the prosecution,440 or that the traces
438 Chesterman, Chan and Hampton, above n 6, 82–88, 91; Thomas, above n 2, 41. 439 Johnston et al, above n 2, 5. 440 A neighbour’s alternative explanation for the screams made it impossible for the prosecution to argue the screams were those of the victim. Similarly, the accused’s father provided evidence that it
115 Murder trial of Gerard Baden-Clay – Twitter Analysis of blood could not be dated or definitively linked to the victim’s death. My sample also revealed little support for the accused’s claims that scratches on his face were from shaving or that marks on his chest were self-inflicted from scratching.
While individual tweets may seem innocuous, there is a good chance that the light in which the public discusses a trial — including contextualising, sequencing, repetition and amplification — could have a serious prejudicial impact. This may have a significant effect on a social media users’ perceptions of the way a crime was committed and it is not clear how this may affect jurors.
4.6 Summary
In this chapter, I investigated the prevalence of prejudicial publicity in tweets during the murder trial of Gerard Baden-Clay. I established how professional journalists and non-journalists talked about the trial, and analysed how social media users showed prejudice on Twitter. I found that 6% of tweets in my sample contained prejudicial information and had the potential to breach the sub judice rule. I found that professional journalists reported the news and in doing so made careful language choices that were mostly factual, neutral, fair and legally compliant. In contrast, I found that non-journalists discussed the news in a way that was more emotive or opinionated. Their tweets were more likely to use non-professional language, to lack impartiality and to be contain prejudicial information. Despite these differences, I found there were occasions when both professional journalists and non-journalists could be more careful with how they tweeted about the trial. Based on my qualitative analyses, I raised concerns about what appeared to be a pro-prosecution bias in my
was he who insisted the accused contact the insurer, meaning this information could not be used as evidence of post-offence conduct that might indicate an intention to murder.
Twitter trials and Facebook juries 116 sample — if this pattern plays out across criminal trials in general, this may present a significant risk to fairness in the criminal justice system. I also raised concerns about cumulative negative sentiment, where the collective effect of multiple tweets working together may have the potential to influence jurors, regardless of whether each individual tweet is legally compliant. The idea that multiple tweets, posted by multiple users, may together be prejudicial is not a scenario contemplated by the sub judice rule and this may present a problem for the law as it currently exists.
117 Murder trial of Gable Tostee – Twitter Analysis
5 Murder trial of Gable Tostee – Twitter Analysis
5.1 Overview
In this Chapter, I present my analysis of tweets during the high-profile murder trial of Gable Tostee.441 I use a combination of content and discourse analysis to understand the prevalence of prejudicial publicity in the dataset, to compare how professional journalists and non-journalists talk about the trial, and to investigate how social media users show prejudice on social media during high-profile criminal trials.
My discourse analysis is more extensive than my first case-study, as I delved deeper into the themes of tweets and how the discourse was framed during this trial.
From this analysis, I found that 17.4% of tweets in my dataset contained prejudicial publicity. I again established that professional journalists were more likely to comply with the sub judice rule than non-journalists. I found that the law is not well suited to addressing the potentially prejudicial effects of multiple publications
(tweets), posted by multiple social media users who collectively demonstrate a prosecution bias or cumulative negative sentiment. I also found that the law is generally not well suited to regulating tweets that might be unfair or inaccurate, that are not made in bad faith, or those that may impair public confidence in the administration of justice. Similarly, the law is not well suited to addressing the potential for juror influence that may result from discussions about gaps in evidence in the trial, through public access to courtroom evidence like the recording, to
441 R v Gable Tostee (QSC, 10 October 2016).
Twitter trials and Facebook juries 118 discussions that may be coloured by gender, or to the potential effects of the trial trending on Twitter.
5.2 The trial
In 2016, the murder trial of Gable Tostee made news around the world due to the unusual circumstances that led to the death of Warriena Wright on the Gold Coast.
On 7 August 2014, Wright and Tostee connected via the dating app Tinder, met in person and went to Tostee’s apartment. The visit lasted some hours, but towards the end the pair got into a heated argument and Tostee locked Wright on his balcony.
While trying to escape, Wright fell to her death from the 14th floor. During investigations police discovered Tostee had audio-recorded three hours of his and
Wright’s conversations without her permission, including their interactions immediately before and after she was locked on the balcony.
For his role (if any) in Wright’s death, Tostee was charged with murder and, in the alternative, manslaughter.442 The prosecution put to the jury that Wright had climbed the balcony railing because she reasonably feared for her life and had no alternative but to attempt an escape. If the jury accepted this argument, at law Tostee would be held responsible for Wright’s death.443 The defence argued Tostee tried to defuse the situation by locking Wright outside and claimed he could not be held responsible for her decision to try to climb down. The jury deliberated for three days during which they submitted a number of questions to the judge and, at one point, notified the court they were unable to reach a verdict. After being asked to continue to
442 Ibid. 443 Royall v R (1991) 172 CLR 378; R v Grimes and Lee (1894) 15 NSWLR 209.
119 Murder trial of Gable Tostee – Twitter Analysis try to reach a verdict, however, the jury retired for a final day and delivered their decision that Tostee was not-guilty of murder or manslaughter.
This case, dubbed the ‘Tinder trial’ because the accused met Wright via the online dating site, generated intense public interest. In particular, the revelation the accused had secretly recorded hours of the date, and the fact the recording was released to the public during the trial, provoked significant public discussion online. Whether the substantial media attention around the trial — particularly on social media — might have affected juror impartiality is difficult to determine, but this research makes it possible to better understand prejudicial conversations during the trial.
It is worth noting that even though Tostee was acquitted, this does not mean that the sub judice rule is irrelevant — contempt may still be established where publications tend to prejudice an accused even though they are acquitted.444 I also note the
Queensland State Coroner is currently reviewing the death of Warriena Wright in order to determine whether to hold an inquest.445 If the matter goes to inquest, and depending on the outcome, it is possible Tostee could face further charges.446 Prejudicial publicity about his murder trial, as well as any sensational information that might be generated during coronial proceedings,447 may have the potential to prejudice future proceedings.
5.3 Prevalence of prejudicial publicity
As for my previous case study, to understand the prevalence of prejudicial publicity in my dataset I first created a visualisation of overall twitter activity during
444 Attorney-General (NSW) v TCN Channel Nine Pty Ltd (n 63) 382. 445 ‘Gable Tostee Could Face an Inquest into the Death of Warriena Wright | Newshub’ (n 364) 13 January 2019. 446 Gable Tostee Could Face Fresh Charges for Death of Tinder Date Warriena Wright on Gold Coast (12 January 2019) NZ Herald
Twitter trials and Facebook juries 120 the trial. This allowed me to contextualise my analysis and understand what was taking place in court during the peaks and troughs of tweet activity. I then coded my random sample for prejudicial publicity and user type.
I used Tableau to create of visualisation of my total Twitter dataset (8,934 tweets), based on the number of tweets posted on each day of the trial, up until one minute before the verdict was delivered. This is set out in Figure 5.1.
Figure 5.1: Tostee Twitter analysis (total Twitter dataset) - tweet activity during the trial.
By reading the underlying tweets, I was able to summarise what was taking place in court during the peaks and troughs in my dataset. I found that tweet activity peaked during the prosecution and defence openings, and when the jury retired to consider its verdict. There were smaller peaks when key witnesses gave evidence, during closing arguments and when the judge was summing up. The highest level of interest was immediately before the verdict was delivered.
To determine the prevalence of prejudicial content in my dataset, I coded my random sample of tweets using my coding scheme set out in Chapter 3. After coding tweets as containing low-level, high-level or no (none) prejudicial publicity, I found that 17.4% of tweets in my random sample contained potentially prejudicial content.
121 Murder trial of Gable Tostee – Twitter Analysis
As set out in Figure 5.2, this comprised 14.0% of tweets containing low-level prejudice and 3.4% of tweets containing high-level prejudice. This meant I could be 95% confident that 16.4% - 18.4% of my total twitter dataset contained potentially prejudicial publicity.
PREJUDICIAL CONTENT High 3.4% Low 14.0%
None Low
High None 82.6%
Figure 5.2: Tostee Twitter analysis (random sample) - 17.4% of tweets contained prejudicial content.
Unlike my first case study where I only coded the data as low, high and none, for this case study I also coded the data based on categories of prejudice. My coding results are set out in Table 5.1 below. This approach allowed me to make comparisons between my two case-studies and also better understand how social media users talked about the trial.
Twitter trials and Facebook juries 122
Number of tweets % of total Category Types of potentially prejudicial information in category tweets (4,629) total % 1. Statements as to guilt 151 3.3% Total high 2. Information about confessions 0 0.0% 3.4% 3. Information about prior convictions 6 0.1% 4. Statements as to innocence 110 2.4% 5. Content that criticises or disparages the accused 404 8.7% 6. Content that creates sympathy for the accused 38 0.8% Total low 7. Content that creates sympathy for a victim 62 1.3% 14.0% 8. Claims that prejudge an issue of fact or law that is 36 0.8% awaiting judgement (other than guilt or innocence) 9. All other content 3,361 72.6% 10. Prosecution claim in above category (1-7) but with 233 5.1% attribution (Crown Prosecutor Glenn Cash) Total non- prejudicial 11. Defence claim in above category (1-7) but with 82.6% attribution (Defence Counsel Saul Holt QC and solicitor 228 4.9% Nick Dore) TOTAL 4,629 100% 100% Table 5.1: Tostee Twitter analysis (random sample) – results after coding for prejudice.
As for my first case-study, the majority of tweets in my random sample were non-prejudicial, but 17.4% were prejudicial. This figure is almost three times higher than in the previous trial (6%) although, other than by speculating, it is not possible to know why this is the case.
Leaving this question aside, that any prejudicial information in the media may come to the attention of jurors after a trial has commenced is cause for concern.448 The sub judice rule does not require proof that prejudicial publications have actually come to the attention of empanelled jurors, only that there is a ‘real, not a fanciful’ possibility that the publication came to their attention.449 But even if a publication is unlikely to have come to the attention of jurors, the courts have still found publishers guilty of sub judice contempt due to a tendency for offending publications to prejudice the
448 New South Wales Law Reform Commission (n 116) 146 [4.86]. 449 Attorney General (NSW) v Radio 2UE Sydney Pty Ltd (New South Wales Court of Appeal 40236/96, Priestley AJA, Meagher AJA and Powell AJA, 11 March 1998) 7 (Powell JA).
123 Murder trial of Gable Tostee – Twitter Analysis proceedings.450 Some courts have held that whether jurors have seen a publication is irrelevant to liability for contempt451 and found publishers in contempt regardless.452
It is difficult to know whether prejudicial publications like those in my datasets may come to the attention of jurors, but it is likely there is a ‘real, not a fanciful’ possibility that at least some jurors may be exposed to them. Regardless, the law is only concerned with the possibility, not the actuality, so the fact these prejudicial publications exist at all means they are likely to breach the sub judice rule. In the context of ‘ambient journalism’,453 as the prevalence of prejudice in this trial (17.4%) is higher than the first, it may mean there is an even greater potential for citizens
(including the jurors) to see some of these posts and for their ‘mental mode’ to contain prejudicial information about the accused.
5.4 Professional journalists versus non-journalists
To understand how professional journalists and non-journalists talked about the trial, I coded my random sample based on whether the user (publisher) who posted each tweet was a professional journalist or non-journalist. As set out in Figure 5.3, I found that 61.6% of tweets were posted by professional journalists and the remaining
38.4% were posted by non-journalists. This breakdown was very similar to that in my
450 Registrar of the Court of Appeal v John Fairfax Group Pty Ltd (New South Wales Court of Appeal No 40478/92, Mahoney JA, Priestley JA and Sheller JA, 21 April 1993); R v David Syme & Co Ltd [1982] VR 173; R v Pearce (n 62); Attorney General (NSW) v Radio 2UE Sydney Pty Ltd (n 452); Michael Chesterman, ‘Media Prejudice During a Criminal Jury Trial: Stop the Trial, Fine the Media, or Why Not Both’ (1999) 1 UTS L. Rev. 71, 77 (‘Media Prejudice During a Criminal Jury Trial’). 451 R v Nationwide News Pty Ltd (Unreported, Victoria Supreme Court No 6129/97, Gillard J, 22 December 1997). 452 Registrar of the Court of Appeal v John Fairfax Group Pty Ltd (n 453); Registrar, Supreme Court of South Australia v Advertiser Newspaper Ltd (Unreported, Supreme Court of South Australia no 2418/95, Bollen J, 17 May 1996); R v Meissner (Unreported, New South Wales Court of Criminal Appeal No 60170/92, 30 October 1992). 453 Hermida (n 166) 301.
Twitter trials and Facebook juries 124 first case-study, where 64.8% of tweets were posted by professional journalists and
35.2% by non-journalists.
USER TYPE non- journalists professional journalists 38.4% non-journalists professional journalists 61.6%
Figure 5.3: Tostee Twitter analysis (random sample) – most tweets were posted by professional journalists.
As I coded by categories of prejudice for this trial, I combined this coding with my coding by user type. This is set out in Table 5.2.
Non- Professional Types of potentially prejudicial information journalists journalists 83.4% 16.6% 1. Statements as to guilt 126 tweets 25 tweets 0.0% 0.0% 2. Information about confessions 0 tweets 0 tweets 100.0% 0.0% 3. Information about prior convictions 6 tweets 0 tweets 96.4% 3.6% 4. Statements as to innocence 106 tweets 4 tweets 86.4% 13.6% 5. Content that criticises or disparages the accused 349 tweets 55 tweets 65.8% 34.2% 6. Content that creates sympathy for the accused 25 tweets 13 tweets 91.9% 8.1% 7. Content that creates sympathy for a victim 57 tweets 5 tweets 8. Claims that prejudge an issue of fact or law that is 91.7% 8.3% awaiting judgement (other than guilt or innocence) 33 tweets 3 tweets 30.7% 69.3% 9. All other content 1,032 tweets 2,329 tweets 10. Prosecution claim in above category (1-7) but with 12.9% 87.1% attribution (Crown Prosecutor Glenn Cash) 30 tweets 203 tweets 11. Defence claim in above category (1-7) but with attribution 6.1% 93.9% (Defence Counsel Saul Holt QC & solicitor Nick Dore) 14 tweets 214 tweets Table 5.2: Tostee Twitter analysis (random sample) - results after coding for prejudice (by category) and user type.
125 Murder trial of Gable Tostee – Twitter Analysis
So I could directly compare these results with my earlier case study, I also combined my results from coding for prejudice (none, low and high) with user type.
This is set out in Table 5.3.
Professional Non-journalists Total Random Sample Journalists
None 60.5% 96.3% 82.6% 1,076 tweets 2,746 tweets 3,822 tweets
Low 32.1% 2.8% 14.0% 570 tweets 80 tweets 650 tweets
High 7.4% 0.9% 3.4% 132 tweets 14 tweets 157 tweets
Total 100% 100% 100% 1,778 tweets 4813 tweets 4,629 tweets
Table 5.3: Tostee Twitter analysis (random sample) - results after coding for prejudice (none, low and high) and user type.
I found that 39.5% of tweets posted by non-journalists in my random sample contained prejudicial content (comprising 32.1% low-level prejudice and 7.4% high- level). This compared with only 3.7% of tweets posted by professional journalists
(comprising 2.8% low-level prejudice and 0.9% high-level). This data is represented in Figure 5.4. By generalising across the total Twitter sample, I could be 95% confident that 38.5% - 40.5% of tweets posted by non-journalists contained prejudicial content, compared with only 2.7% - 4.7% by professional journalists.
Non-journalists Prejudicial Professional Prejudicial 2.3% 15.2% Journalists
Non- Non-prejudicial prejudicial Prejudicial Prejudicial 97.7% 84.8% Non-prejudicial Non-prejudicial
Figure 5.4: Tostee Twitter analysis (random sample) - non-journalists posted more prejudicial tweets than professional journalists.
Twitter trials and Facebook juries 126
As these results echo my first case study, I again conclude that the sub judice rule appears to be largely effective in regulating the behaviour of professional journalists, but less so for non-journalists. The behaviour of professional journalists and non-journalists in this trial was consistent with those in the Baden-Clay trial.
Professional journalists tweeted in a way that mostly conformed to their traditional function454 of reporting the facts, in objective or neutral terms.455 They generally kept their personal opinions to themselves456 and, mostly, they appeared to present reports fairly and impartially.457
The behaviour of non-journalists was also consistent with my previous findings.
I found that non-journalists did not report the news, but instead commented on it in a way that constructed meaning around it.458 Rather than breaking the news story, like professional journalists, non-journalists were participants who interacted with the news story.459 They used discourse that was informal and conversational, and that included swearing, high-arousal emotions such as anger and frustration,460 and generally lacked the careful language of professional journalists. They were far more likely to post prejudicial information than professional journalists. My quantitative results suggested that non-journalists posted almost 11 times (10.7) more prejudicial tweets than professional journalists — an almost identical result to my first case study.
Non-journalists posted the most tweets in all categories of prejudice (high and low).
454 Lasorsa, Lewis and Holton (n 166) 20. 455 Singer (n 129) 144; Brooks et al (n 403); Scanlon (n 403); Mencher (n 403). 456 Lasorsa, Lewis and Holton (n 166) 23; Schudson and Anderson (n 401); Kovach and Rosenstiel (n 401). 457 Pavlik, Morgan and Henderson (n 402) 16. 458 Deuze (n 406) 147. 459 Bardoel and Deuze (n 407) 94. 460 Kwon and Cho (n 409) 87–89; Turel and Qahri-Saremi (n 409) 3052.
127 Murder trial of Gable Tostee – Twitter Analysis
Although professional journalists largely complied with the law, I again found they tended to focus on the more sensational aspects of the trial, particularly the highly emotive claims of the prosecution. For example, a pathologist’s evidence that Wright’s blood alcohol level at the time of death was 0.156%, was tweeted by a professional journalist as: ‘Wright’s blood alcohol reading at time of death over 3 times legal limit’.
There is, of course, no legal limit to alcohol consumption unless a person is driving.
Wright was very drunk and, perhaps, the presentation of evidence in this way might make it easier for viewers to understand. But it could also be seen as potentially prejudicial as it is not a fair report and is loaded due to preconceptions about drunk driving. In another example, evidence from a pathologist that Wright’s height could not be determined because of severe disruption from the fall, was reworded in tweets by a professional journalist (and others) to describe her body as severely mangled, arguably a more sensational description that is not a fair report.
Although professional journalists were responsible for only a small number of prejudicial tweets, there were again occasions when they could have been more careful to avoid criminal liability. These concerns related primarily to the failure to attribute tweet content and to misleading compilations of information. As these aspects of the way professional journalists tweeted were wholly consistent with my first case study, there is little merit in repeating my previous discussion. The more interesting part of my analysis for this trial is in my qualitative analysis below.
5.5 Prejudice on social media
I used discourse analysis to better understand how social media users showed prejudice on social media. My analysis of social media users included both professional journalists and non-journalists, however, given the majority of prejudicial
Twitter trials and Facebook juries 128 tweets were posted by non-journalists, my observations more specifically apply to them.
In the context of how users show prejudice on social media, it is particularly important that my discourse analysis allows me to draw conclusions about the potential for influence regardless of how any particular tweet was coded. It allows me to recognise that tweets coded as non-prejudicial will not necessarily be non-influential, and that tweets coded as prejudicial will not necessarily be influential. I am able to consider the potential effects of multiple or connected tweets working together in the context of a broader conversation, rather than treating each post as an atomic tweet that is random and unconnected to others.
Before exploring the themes that emerged during my analysis, it is worth noting some important qualifications to the sub judice rule. While my Baden-Clay analysis was focussed more on prejudicial or non-prejudicial content, my Tostee discourse analysis allows me to focus on the nuances of tweets and comments and their apparent context or purpose. This more complex and nuanced analysis (as explained in my coding scheme) allowed me to account for reports of judicial proceedings that must be fair and accurate, and made in good faith and without malice, in a more sophisticated way. Reports about judicial proceedings will be protected from the law of sub judice contempt if they meet certain conditions. First, they must be a report of the proceedings — an account of the evidence, submissions or other public aspects of the proceedings.461 Significantly, a comment about judicial proceedings will not be a report.462 A report need not be a verbatim account of proceedings463 — a summary is
461 Hinch v Attorney General (Vic) (n 1) 25 (Mason CJ); R v Sun Newspapers Pty Ltd (n 355) 287. 462 Packer v Peacock (1912) 13 CLR 577, 587. 463 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259.
129 Murder trial of Gable Tostee – Twitter Analysis acceptable — but it must only report what has taken place in open court and in the presence of the jury.464 It must be fair and accurate,465 reasonably contemporaneous with proceedings,466 and must be published in good faith and without malice.467
In the context of these qualifications, large volumes of tweets were problematic.
I again found a prosecution bias and cumulative negative sentiment caused by the collective effect of multiple tweets published by multiple users. I also found a lack of fairness and accuracy, and some tweets appeared not to have been published in good faith and without malice. Others had the potential to impair or undermine public confidence in the administration of justice.
5.5.1 Prosecution bias As for my first case study, I found more tweets focussed on the prosecution narrative than on the defence narrative, although this was not the case for tweets containing attribution.468 The majority of these were posted by professional journalists
(90%) and they related either to the prosecution or defence in almost equal proportion
(51% prosecution and 49% defence). However, by considering tweets that did not contain attribution — the majority of which were posted by non-journalists (87%) — a prosecution bias became more apparent. Of all tweets in my sample, 6% contained an explicit statement about the guilt or innocence of the accused. Of these, 58% claimed Tostee was guilty and 42% that he was innocent. A pro-prosecution focus was
464 Ibid 258; R v Day & Thomson [1985] VR 261. 465 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259; R v Gray (1865) 10 Cox CC 184; R v Evening News; Ex parte Hobbs (1925) 2 KB 158; R v Kray (1969) 53 Cr App R 412; R v Sun Newspapers Pty Ltd (n 355). 466 Minister for Justice v West Australian Newspapers Ltd [1970] WAR 202, 207; R v Scott and Downland Publications Ltd (n 110); Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (n 281) 132 (Wilson J). 467 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 257. 468 See Table 5.1 earlier in this chapter. Tweets containing attribution in this trial were coded into categories 10 (prosecution claims with attribution) and 11 (defence claims with attribution).
Twitter trials and Facebook juries 130 also evident when I combined those coding categories that are pro-prosecution — those containing statements as to guilt, confessions, prior convictions, criticism of the accused, or sympathy for the victim.469 I also combined those categories that are pro- defence — those containing statements as to innocence or sympathy for the accused.470
In that case, the results were indicative of a marked bias, with 81% of tweets pro- prosecution and only 19% pro-defence. These are set out in Figure 5.5.
GUILTY V NOT- PRO-PROSECUTION GUILTY V PRO-DEFENCE Pro- defence Not- 19% guilty 42% Guilty 58% Pro-prosecution 81%
Figure 5.5: Tostee Twitter analysis (random sample) – coded categories indicate a prosecution bias.
These results represented 14% pro-prosecution in the total random sample, whereas pro-defence categories represented just 3% — these results suggest over four times more tweets supported the prosecution than the defence. These results are set out in Figure 5.6 below.
469 See Table 5.1 – coding categories 1, 2, 3, 5 and 7. 470 See Table 5.1 – coding categories 4 and 6.
131 Murder trial of Gable Tostee – Twitter Analysis
TWITTER RANDOM SAMPLE
Pro- Pro-defence prosecution 3% 14%
Neutral 83%
Figure 5.6: Tostee Twitter analysis (total sample) – more comments were pro-prosecution than pro- defence.
As for my first case study, these results demonstrate a prosecution bias in tweets posted during the trial. At law, a report will be unfair if it is, among other things, biased,471 however, this type of prosecution bias across a large volume of tweets poses two distinct challenges for the law. First, tweets in my sample tended to be comments and not reports,472 and the law has not been shaped to address these effectively.
Second, the principles requiring fair and accurate reporting473 relate only to individual publications and do not extend across a sum of publications that may collectively convey bias. The law was well suited to regulating bias in the traditional media, but it is not well suited to responding to a prosecution bias across large quantities of rapid fire publications on social media.
5.5.2 Cumulative negative sentiment Also as for my earlier case study, I again raise concerns about the potential for the collective weight of low-level prejudicial or negative information across multiple
471 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259. 472 Packer v Peacock (n 465) 587. 473 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259; R v Gray (n 468); R v Evening News; Ex parte Hobbs (n 468); R v Kray (n 468); R v Sun Newspapers Pty Ltd (n 355).
Twitter trials and Facebook juries 132 posts on social media to influence jurors. Social media users appeared to focus on a number of key themes and, while individual tweets were not always prejudicial, the collective commentary had the potential to work against the accused. The themes included tweets about Tostee’s audio recording, collective support of Wright, and concerns about post-offence conduct.
5.5.2.1 Audio recording There was considerable negative sentiment about the three hour audio recording of Tostee and Wright’s date. A number of tweets focussed on the question of why
Tostee made the recording and the potential for Tostee to manipulate his recorded words to deflect culpability. Many users found Tostee’s decision to secretly record his and Wright’s date distasteful, but as he chose not to give evidence in court he was not required to answer questions about it: ‘Listened to Tostee’s audio – chilling stuff – what an asshole’; ‘why did Tostee make a recording on the terrible night in question?’;
‘it’s so creepy – makes me shudder’; ‘Good on Tostee for creepily recording everything, hope it helps convict him’; and ‘why was he recording anyway? A planned alibi?’
Others expressed negative sentiment because they felt that too much weight was placed on Tostee’s recording, given he may have manipulated his words and actions to protect himself against future allegations: ‘I listen to it as a form of insurance for him – everything he says is contrived to create an impression and to steer her’; ‘Tostee clearly had a habit of recording his encounters and was practiced in choosing his words’. ‘I don’t trust Tostee’s recording – he had the advantage of knowing he was recording and manipulating, controlling Wright’; ‘Tostee definitely choosing his words on post-death audio’; and ‘Tostee’s recording is particularly unreliable – she doesn’t know he’s recording’. We do not know whether these claims have merit, but
133 Murder trial of Gable Tostee – Twitter Analysis as Wright was unaware of the recording and highly intoxicated, she was vulnerable to manipulation and oblivious to any need to counter Tostee’s claims or to provide her own justification.
5.5.2.2 Support of Wright There was also substantial negative sentiment surrounding Tostee’s treatment of
Wright, the terror in her voice before she died, and Tostee’s post offence conduct. As
Wright’s death meant she could not provide evidence, many were left questioning what was actually happening during key points on the recording. Some suggested Tostee provoked Wright into throwing rocks at him. Others said Wright’s claim ‘that really hurt my vagina’, and Tostee’s laughter in response, suggested he had assaulted her.
Some felt too much was made of Wright ‘beating up’ Tostee, given the size difference between the two: ‘Wright was small and she’s entitled to defend herself with a weapon when Tostee’s a large man who won’t let her exit safely’.
A considerable number of tweets about Wright’s last moments also contributed to cumulative negative sentiment. Before she was locked on the balcony, she is heard on the recording to scream ‘No, no, no, no’ some 31 times in 46 seconds.474 The sound is chilling and was described by the prosecutor as being indicative of her experiencing
‘terror, bordering on hysteria’.475 It is apparent that Tostee is hurting Wright, but exactly what he is doing is unclear: ‘what was he doing to her?’ The Prosecution hypothesised that Tostee restricted Wright’s breathing for some 45 seconds, possibly
474 Melissa Fyfe, ‘What Happened to Warriena Wright after She Met Gable Tostee?’, The Sydney Morning Herald (17 March 2017). 475 Jorge Branco, ‘Gable Tostee Trial: Tostee Declines to Testify in Murder Trial’, Brisbane Times (online at 14 October 2016)
Twitter trials and Facebook juries 134 by a chokehold,476 but there is no way of knowing for sure. This caused substantial speculation and criticism of Tostee online.
5.5.2.3 Post offence conduct There were a number of tweets about Tostee’s post offence conduct and these seemed also to demonstrate negative sentiment. The fact Tostee chose not to call an ambulance when Wright fell, that he fled the scene rather than render assistance, and that he ordered pizza nearby when he left the building, all resulted in substantial criticism on social media. Some suggested that fleeing or remaining silent were examples of post-offence conduct that demonstrated consciousness of guilt. But when the jury questioned the court, they were told not to consider Tostee’s conduct after the alleged offence as part of their deliberations.477 The Court’s view appeared to be that
Tostee’s behaviour was not indicative of consciousness of guilt and that the potential prejudicial effect of the evidence outweighed its probative value.478
Negative sentiment was also prompted by images of Tostee exiting the building holding an unknown object after Wright’s fall, as depicted in Figure 5.7.
476 Ibid. 477 ‘Gable Tostee Jury Struggle for Unanimous Verdict’, Brisbane Times (18 October 2016)
135 Murder trial of Gable Tostee – Twitter Analysis
Figure 5.7: Tostee Twitter analysis - images of Gable Tostee as he exited the lift via the basement
A lack of clarity in the images made it impossible for the court or experts to identify the item — although it was thought to be a long metal object.479 Speculation that the item might be a weapon prompted significant negative sentiment online. When the jury asked the court about it, they were told there was no evidence about the object beyond the images, that it was irrelevant and that they should not speculate about it.480
It is difficult to know whether jurors easily accepted this response or whether they may instinctively have felt compelled to find a way to fill this evidentiary gap. A
2006 study in the United States considered the effect of judicial instructions to disregard inadmissible evidence — which are similar to directions to disregard prejudicial publicity.481 They found that ‘judicial instructions do not effectively
479 ‘Questions Gable Tostee Jury Wants Answered: What Was the Long Metal Object He Had?’ (online at 17 October 2016)
Twitter trials and Facebook juries 136 eliminate jurors’ use of inadmissible evidence. Juror noncompliance is particularly likely when a reason for rejection of [inadmissible evidence] is not provided … [and] jurors resist giving up evidence that they believe is probative’.482
5.5.2.4 Trending on Twitter Cumulative negative sentiment may also be reinforced when mainstream news is informed or shaped by what is ‘trending’ on social media. When a hashtag experiences a large enough increase in traffic — generally a sudden spike, rather than a slow increase over time — the topic will starting ‘trending’.483 Once it is trending, it is seen even more, and this further perpetuates the ‘trending’. This is similar to the notion of posts ‘going viral’.484 The Tostee trial reached ‘trending’ status: ‘New York
Magazine's second-highest trending story right now is about Gable Tostee’; ‘Tostee -
Know the hourly TOP trends’; ‘#tostee is now trending in Australia, ranking 40’; and
As the public spectacle of criminal trials is reinforced in so many ways, it is difficult to know whether it is possible for jurors to avoid exposure to information on social media about a trial on which they are sitting. The following tweet, posted in jest, captures some of this concern: ‘Result: Instagram says guilty but Facebook says Not Guilty while Twitter declares a mistrial #tostee’. These new types of potential exposure raise questions over whether it is possible to receive a fair jury trial in the 482 Ibid 469–470. 483 ‘How Does Twitter Decide What Is Trending? | ReThink Media’ 13 July 2016 137 Murder trial of Gable Tostee – Twitter Analysis age of the internet: ‘With amount of coverage, one might think he'd argue he couldn't get fair jury trial - which is lesser risk for #Tostee!?’; and ‘Australia relies on courts, not social media to determine who goes to jail. Thank goodness #tostee’. However, the courts continue to rely on the view that fairness is ensured by giving proper judicial directions to jurors and that jurors will comply with these: ‘Jury told Tostee has been portrayed as a ‘villain’ but is ‘a real person and you will trial him properly, we are sure’’; and ‘Holt is warning the jurors to make their decision based only on the evidence, not newspaper reports. Gable Tostee is a real person’. It is not only Twitter users who stand to be exposed to, and potentially influenced by, stories that are ‘trending on Twitter’. The notion of social media cross-pollination means that information presented on one platform, particularly popular or highly amplified information, is easily ‘pushed’ to other platforms such as Facebook, Instagram and others, meaning non-Twitter users (jurors) may still be exposed to these conversations even if they are not a Twitter user. Some mainstream media platforms also highlight news that is ‘trending on Twitter’ or other social media, meaning even those who do not use social media at all (including jurors) may still be exposed. While we assume most information picked up and published by professional media organisations complies with the law, it is still the case that some legally compliant content may have the effect of influencing jurors. 5.5.2.5 Gendered perspectives My data sample does not include the gender of account users, but the nature of some tweets gave the impression they were gendered, where males supported Tostee’s version of events and females supported and sympathised with Wright. For example, Tweets criticising feminist perspectives appeared to be posted by males, those Twitter trials and Facebook juries 138 referring to their ‘bros’ and ‘brahs’485 also appeared to be posted by males, and those claiming more needed to be done to protect women from violence appeared to be posted by women. It is not possible to know whether this may affect jurors and, if so, whether it may have a different effect on jurors depending on their own gender. Those in support of Tostee placed blame on Wright for being aggressive and (arguably) violent, for being intoxicated, for choosing to go the apartment of someone she had just met on Tinder, for choosing to climb over the balcony, and for not choosing to leave earlier when she could have. They supported Tostee’s decision to secretly record the events and justified this approach as one that ‘bros’ ought reasonably take, especially if they had previous experiences with women making false accusations. They did not accept the view there was anything immoral or unethical in this approach, and some said they too had done this or would do this in future to protect themselves. Those in support of Wright expressed anger about Tostee’s treatment of her, including his violence, aggression, insults, and providing her home-brew alcohol when she did not know its strength or contents. They criticised him for putting any person who was so heavily intoxicated on the balcony, took issue with his decision to record their interactions, and were outraged by his conscious decision not to look over the balcony when Wright fell, not to call an ambulance, and not to render assistance. Some tweets were also concerned with the gender of jurors, with one viewer asking for feedback on the gender, age, and demographic of jurors: ‘Any idea of the gender/age demographic of the jury? Wondering if it's a predominantly female jury?’; ‘The #Tostee jury is evenly split - 6 men and 6 women’; and ‘Six men and six women have now been sworn in on the jury panel, with a reserve female juror also selected’. 485 References to brothers or males friends and associates. 139 Murder trial of Gable Tostee – Twitter Analysis The first tweet example suggested this user felt the gender of the jurors was relevant to the outcome of the trial. The idea that the gender of decision-makers is relevant to reaching a verdict has long been one of the factors taken into account by barristers (both Crown and Defence) when excluding particular jurors (challenge or standby) whose number is drawn from the ballot box. There is some evidentiary basis for these choices,486 however, ongoing cultural and legal shifts may mean that discriminating between genders is increasingly unpalatable.487 In any case, as previously mentioned, juror details are highly confidential and it is unlawful to identify a juror,488 even though those present in court can see the jurors, and family and friends of jurors may know who they are. But the idea that users in an online space might actively seek information about jurors is concerning. 5.5.3 Lacking fairness and accuracy Many of the tweets in my sample lacked fairness and accuracy. The test to determine whether a published account of judicial proceedings is fair and accurate is that the report must be ‘one which a person of ordinary intelligence using reasonable care might reasonably regard as giving a fair summary of the proceedings’.489 A report will be unfair if it is biased, if it misrepresents the truth, if it focusses on unfavourable aspects of the proceedings, or if it withholds material facts that would put a different complexion on the facts truly reported.490 A fair report must be uncoloured, as ‘a 486 Blake M McKimmie and Barbara M Masser, ‘Chapter 6 The Effect of Gender in the Courtroom’ in Forensic Psychology: Concepts, Debates and Practice (Routledge, 2nd ed, 2010) 104–105. 487 Andrew LT Choo and Jill Hunter, ‘Gender Discrimination and Juries in the 20th Century: Judging Women Judging Men’ (2018) 22(3) The International Journal of Evidence & Proof 192 (‘Gender Discrimination and Juries in the 20th Century’). 488See eg Jury Act 1995 (Qld) s 70(2). Section 70 includes provisions that a person must not publish to the public jury information, where jury information includes, in part, ‘information identifying or likely to identify a person as, or as having been, a juror in a particular proceeding’. 489 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259; Attorney-General (Qld) v WIN Television (Qld) Pty Ltd (n 61) [9]. 490 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 259. Twitter trials and Facebook juries 140 coloured report would hardly answer the tests of fairness’.491 Most tweets in my sample are comments which, as previously mentioned, are not reports.492 This may be enough, on its own, to suppose that those tweets (comments) containing prejudice are unlawful. I have nonetheless analysed my tweets in the context that they may be lawful, even if they contain prejudice, provided they contain a fair and accurate report. 5.5.3.1 Biased, misleading or withholding material facts Those tweets in my sample that lacked fairness or accuracy included content that was biased, misleading, focussed on matters unfavourable to the accused, or withheld important information. I have set out a number of examples below that collectively illustrate the potential for prejudice when courtroom evidence is tweeted in a way that is unfair or inaccurate. It seems unlikely that a ‘person of ordinary intelligence’ would regard them as fair, or uncoloured, summaries of proceedings. Some tweets were unfair because they implied the defendant made statements that were actually made by the victim. In various tweets, Tostee’s name was placed immediately before each of the statements: ‘I’m going to beat you up for fun’ and ‘I will f***ing destroy your jaw’. On the recording, both statements can be heard to be made by Wright in jest. But the placement of Tostee’s name, without further context, is misleading as it suggests he was making violent threats. As for the Baden-Clay trial, some tweets compressed different aspects of evidence and presented them as if they occurred sequentially. In this way they suggested a causal link, but this was misleading: ‘I’m going to beat you up for fun’: it was self-defence when Tostee locked Wright out on the balcony’. This tweet withholds the fact that there was over an hour between Wright’s ‘threat’, presented as a possible 491 Minister for Justice v West Australian Newspapers Ltd (n 469) 207 (Virtue SPJ and Jackson CJ). 492 Packer v Peacock (n 465) 587. 141 Murder trial of Gable Tostee – Twitter Analysis provocation, and Tostee’s decision to lock her outside. After listening to the recording, it seems clear that there is probably no causal link between these two particular aspects of the evidence, but this would not be clear to a person reading the tweet. A similar example is: ‘#tostee said ‘you're lucky I haven't chucked you off my balcony you god dammed [sic] psycho little b*tch’ before she fell’. The words ‘before she fell’ are problematic, as they imply Tostee’s statement was made only moments before Wright’s fall. In this way, they imply a sequence of events that might suggest, for example, that Tostee threatened then pushed Wright. However, the tweet withholds the fact that there were six minutes between Tostee’s statement and Wright’s fall. On the recording, substantial conversations and interactions can be heard to take place in this time, and it seems unlikely there is any direct causal link between the two. The same tweet is also potentially biased, as it overlooks the nuance of Tostee’s statement which, when heard on the recording, appears to have been said in jest — an unfortunate use of humour given the events that follow. A reader of the tweet also would not know that it was actually Wright who first described herself as a ‘psycho drunk’ about an hour before Tostee used the label. In my sample alone, this prejudicial tweet appears six times and is posted by users with a collective follower base of approximately 109,000 users — this figure would be significantly higher across all tweets about the trial, including those I did not capture. In order to be fair and accurate, reports about judicial proceedings must relate to what has taken place and not to what will take place. The tweet 'No, no, no, no, no!' Tostee downstairs neighbour will give evidence about woman screaming’ is problematic. A reader may not immediately understand that this tweet pre-empts the evidence of a witness who is yet to take the stand — the publication of evidence that Twitter trials and Facebook juries 142 does not yet exist and in a way that is weighted against the defendant is unlikely to be considered fair. The presentation of alternative hypotheses in tweets was also problematic and may be misleading. This was not something we saw in the traditional media. The following tweet interprets actual evidence presented in court and then uses it to present an alternative hypothesis: ‘Tostee allegedly had strangled her beforehand anyway so thrown over already dead?’ While the Prosecution did put to the court that Tostee was trying to strangle Wright in the final moments before locking her on the balcony, there was no suggestion she died at this point. And if the evidence that she tried to escape by climbing the balcony is true — and the Court has accepted it is — then clearly she was still alive. Yet the hypothesis that she had already died, once read, cannot be unread. While we may assume that jurors would know, when reading this tweet, that Wright was still alive on the balcony, it may be case that these types of alternative hypotheses raise questions in the minds of jurors. Arguably, any publication that affects the way a juror processes or thinks about the evidence is concerning. 5.5.3.2 Issues of law The other category of tweets that appeared to be lacking fairness or accuracy were those including opinions about the law in this case. In most instances these tweets were not coded as prejudicial, but they still had the potential to influence jurors. The legal questions in this trial were particularly challenging, as the charge of murder was based on relatively unknown legal principles.493 The Crown case was not that Tostee intended to kill or cause grievous bodily harm, but that Tostee had created such fear in Wright that she felt her only reasonable and rational option was to try to escape. A 493 Royall v R (n 446); R v Grimes and Lee (n 446). 143 Murder trial of Gable Tostee – Twitter Analysis number of tweets included information about technical aspects of the law, but some were inaccurate and had the potential to confuse jurors. It is possible exposure to this information may interfere with jurors’ own knowledge or memory of the law as it was presented in court. Tweets included discussions of how particular evidence ‘fits with the legal intent test’, whether Tostee was ‘being tried for alternative manslaughter’, and whether the accused’s conduct met the ‘substantial cause ‘significant contribution’ test’. Some included cases and legal principles: ‘R v Sherrington [2001] QCA 105. Need not be the sole or major cause R v Pagett (1983)’. Others argued about criminal and civil standards: ‘no that's the civil standard. It's about whether climbing down was reasonable given 'threat'’. And others about how the law operates: ‘Even in rape cases, each count must be separately proved to jury. History of sexual predation comes into sentencing’; and ‘no, the issue for the jury is did he wilfully kill her? If not, then no murder conviction’. In addition to the complex legal information presented in court, it is possible tweets about the law — especially those that are inaccurate — might influence jurors. 5.5.4 Not made in good faith and without malice As my data focussed on reports of judicial proceedings (during trial publications), those reports must not only be fair and accurate, but they must also be made in good faith and without malice. A number of tweets in my sample do not appear to have been made in good faith. They were not necessarily coded as prejudicial, but they may still have the potential to influence jurors. As discussed earlier, for an otherwise prejudicial publication to be justified at law, not only must it be fair and Twitter trials and Facebook juries 144 accurate, but it must also be made in good faith and without malice.494 A report will not be made in good faith if it ‘was published because of a belief as to its publicity value’ or ‘its value in attracting sales’, particularly where it is in ‘complete and serious disregard of its consequences … on the trial of the accused’.495 There were many examples where tweets posted by professional journalists may have been posted for the sake of publicity — as mentioned earlier there is increasing pressure on journalists towards sensational news.496 I note tweets that did not appear to have been made in good faith were often coded as non-prejudicial, because they did not fall into one of the recognised categories of sub judice contempt. However, they may still have the potential to influence a juror. I note the content does not need to fall within one of the recognised categories to be prejudicial, and therefore contemptuous, but I used the categories as a way of undertaking my coding. A number of tweets appeared to be posted for the purpose of entertainment, rather than to convey important information. They used humour, including black humour, and some tweets even encouraged viewers to vote on the accused’s guilt. For professional journalists, the flippancy and levity expressed in some tweets may be likened to conversations about national sport:497 ‘We have movement at the courthouse’; ‘Strap yourselves in folks, for Day 2 of the trial’; ‘ROLLING COVERAGE: it’s day 5 of the trial’; ‘Will Tostee be found guilty? We'll have live coverage of the verdict’; ‘Hear the bombshell audio on Gable Tostee's phone before a tourist fell to her death’; ‘#Tostee’s fate hangs in balance as jury deliberates’; and ‘Don't miss the untold story of Gable Tostee and the Tinder tragedy in tomorrow's 494 Ex parte Terrill; Re Consolidated Press Ltd (n 76) 257. 495 R v Scott and Downland Publications Ltd (n 110) 675. 496 Hanusch (n 415) 44. 497 For discussion regarding the public spectacle of criminal trials: see Friedman (n 81); Foucault (n 80) 48. 145 Murder trial of Gable Tostee – Twitter Analysis @[media institution]’. It seems likely the positive energy and sense of excitement embedded in these tweets was done for the purpose of attracting viewers or clicks from followers, and this may mean they were not made in good faith. The law requiring that reports of proceedings be made ‘in good faith and without malice’ date back to 1937 and 1972. The law then was directed at journalists (and the like) who were solely responsible for publishing and distributing news, including reports about court matters. However, in the age of social media, when any person can be a publisher, any person who reports on proceedings from court may also be subject to the requirement to report “in good faith and without malice”. For non-journalists, as it was not their intention to report the news but instead to comment on it, they likely did not set out to comply with this law. While users may not have set out to harm the accused by way of malice, tweets encouraging betting, polling and voting online are likely not to have been made in good faith: ‘$1.73 not guilty of Murder - a gift’; ‘Putting money on a hung jury’; ‘The jury can't decide, so what do you think? Guilty or not guilty?’; ‘Based on the results of this survey it would be a hung jury’; and ‘While we wait let's have a vote, we the Twitter jury find Tostee … guilty or not guilty?’ One tweet included a link to a website where users could vote on the verdict in a formal poll and watch as responses were tallied in real time. These tweets were coded as non- prejudicial in my sample, as they did not include prejudicial content, but tweets that encourage users to engage in conversations that may be contemptuous are particularly problematic for the law. In a similar vein, it seems unlikely those responsible for humorous Twitter handles (usernames) or hashtags had any regard for their consequences on the trial: Twitter trials and Facebook juries 146 @MyFavMurder; @ToastedShoes;498 @Acquitmenow; @MissMurder84; and #myfavouritemurder. There were also a number of links in tweets to websites such as websleuths499 and aussiecriminals.500 These sites encourage user participation where citizen journalists501 or ‘armchair detectives’ work out ‘whodunnit’. While murder mysteries have long been popular games for those who enjoy investigations and evidentiary analysis, this type of approach for real-life crimes is more concerning as they have the potential to contribute another layer of prejudice to online conversations. 5.5.5 Impairing public confidence in the administration of justice There were a number of tweets in my sample that had the potential to ‘impair or undermine public confidence in the administration of justice’ — also known as scandalising the court.502 These tweets are of interest as there are substantial similarities and overlap between sub judice contempt and the contempt of scandalising the court. While the sub judice rule seeks to protect the administration of justice in the particular case, scandalising the court503 is concerned with preserving public confidence in the administration of justice more broadly. For a publication to be in contempt for scandalising the court,504 it must tend to detract from the authority and influence of judicial determinations … [or be] calculated to impair the confidence of the people in the Court’s judgments because the matter published aims at lowering the authority of the 498 Some tweets included ‘banter’ around Tostee’s name, with a focus on ‘toasted’ sandwiches, melba toasts, cheese toastees, as well as the idea that Tostee was ‘toast’ for what he had (allegedly) done. 499 See www.websleuths.com/forums/. 500 See http://aussiecriminals.com.au/. 501 Martin Hirst, Navigating Social Journalism: Friends, Enemies, Followers and Likes (Routledge, 2019) (‘Navigating Social Journalism’). 502 Attorney General (NSW) v Mundey (1972) 2 NSWLR 887, 911; Solicitor-General v Radio Avon Ltd (1978) 1 NZLR 225; Fitzgibbon v Barker and others (1992) 111 FLR 191, 201; R v Hoser & Kotabi Pty Ltd [2001] VSC 443, [52]; Wade v Gilroy (1986) 10 Fam LR 793, 805. 503 Gallagher v Durack (1983) 152 CLR 238, 243; H Burmester, ‘Scandalising the Judges’ [1985] Melbourne University Law Review 313. 504 Wade v Gilroy (n 505) 798; McGuirk v University of NSW [2009] NSWSC 1058, [338]-[344]. 147 Murder trial of Gable Tostee – Twitter Analysis Court as a whole or that of its Judges and excites misgivings as to the integrity, propriety and impartiality brought to the exercise of the judicial office.505 An offending publication need not be direct specifically at courts or judges, they may also be directed at magistrates506 or jurors.507 The contemnor need not have intended to scandalise the court, they need only to have intended to publish the offending material.508 As for sub judice contempt, there must be a ‘real risk’ that public confidence in the administration of justice will be undermined,509 but there need not be an actual undermining of public confidence. This will be measured by ‘its impact on the reasonable person’.510 There were a number of tweets that had the potential to scandalise the court, particularly those that were critical of the Judge or jury. In reference to the Judge, examples included: ‘Clearly the Judge should have directed the jury to discount Tostee’s self-serving, secret and unsworn recording’; I’d appeal the shit out of the judge’s directions to the jury’; ‘the judge is subliminally encouraging the jury to find not guilty’; ‘is the judge’s real name Judge Tostee? Never seen a judge try to spoon feed the jury a not guilty verdict like this in my life’; and ‘hope the Judge corrects that error of law in his jury directions’. In the context of jurors, published comments will be contemptuous, particularly where they are abusive, if they tend to 505 R v Dunbabin; Ex parte Williams (1935) 53 CLR 434, 442 (Rich J); Gallagher v Durack (n 506) 243; Fitzgibbon v Barker and others (n 505) 201. 506 Attorney-General (Qld) v Lovitt [2003] QSC 279; R v Hoser & Kotabi Pty Ltd (n 505). 507 Re Johnson (1887) 20 QB 68, 74; Prothonotary of the Supreme Court of New South Wales v Katelaris [2008] NSWSC 389. 508 R v Editor of New Statesman; Ex parte Director of Public Prosecutions (1928) 44 TLR 301; Attorney General (NSW) v Mundey (n 505) 911; Solicitor-General v Radio Avon Ltd (n 505) 232–234; Ahnee v Director of Public Prosecutions (1999) 2 AC 294, 307; Attorney-General (Qld) v Lovitt (n 509) [58]; Prothonotary of the Supreme Court of New South Wales v Katelaris (n 510) [30]-[31]. 509 Solicitor-General v Radio Avon Ltd (n 505) 233–234. 510 Bell v Stewart (n 281) 425; R v Hoser & Kotabi Pty Ltd (n 505) [204]-[225]. Twitter trials and Facebook juries 148 deter those jurors in relation to their participation in future trials … deter persons generally from participating as jurors in future trials … [or where they have] a tendency to undermine public confidence in the administration of justice by suggesting that jurors had not properly discharged their duties in the proceedings.511 Tweets discussing the jury tended to express an anti-jury sentiment, primarily relating to three themes: the length of time taken to reach a verdict; the number of questions put to the judge; and criticism of a juror who unlawfully identified herself on Instagram, causing the defence to request a mistrial. Criticisms about the time taken to deliberate were amplified when, on day three of deliberations, the jury advised the court it could not reach a verdict512 and when they put a number of questions to the court: ‘The prick did it. Jury is stupid if they don’t say it was murder’; ‘jury hands the judge another note: 'We cannot reach a verdict because we are too thick to know what we are doing’; ‘Wouldn't be surprised if the next question Jury asks is about Boaty McBoatface. Get with the program jurors’; ‘Juror question #433 - what type of pizza did he order? I'm a vegan’; and ‘#TosteeJuryQuestions: 'Your Honour can I have a Ouija board to contact Wright to ask if #Tostee really killed her?’ The anti-jury sentiment also increased when, shortly before the jury was ready to deliver its verdict, counsel notified the court that a female juror had identified herself in multiple posts on Instagram:513 ‘Judge says particularly disappointing one female juror has identified herself on Instagram as jury member’. This juror’s misconduct delayed proceedings, as the defence sought a mistrial and the court needed to 511 Re Johnson (n 510). 512 ‘Gable Tostee Murder Trial: Jury Can’t Reach Verdict’, The Queensland Times (18 October 2016). 513 Blucher (n 349); Shadwell (n 349). 149 Murder trial of Gable Tostee – Twitter Analysis determine what approach it would take in response. While the court found the comments did not jeopardise a fair trial, these unusual circumstances resulted in substantial backlash on Twitter: ‘What an idiot if true. When I was juror I didn't even speak to my husband about anything’; ‘If it looks like a bimbo and talks like one, chances are. And I have proof #tool #juror’; ‘Idiot #tostee juror outed herself on Instagram ... you are putting justice at risk, so stupid!’; ‘Hits head in disbelief that juror on the #tostee trial has been posting comments to insta’; and ‘juror instagramming during trial - we are officially doomed as a species’. This type of published public criticism that suggests jurors had not properly discharged their duties in the proceedings was not something typically seen in traditional media and it is possible it may deter jurors or others from participating in future trials. Jurors were once sequestered during deliberations,514 but as this is no longer common practice jurors in this trial went home on each of the four days it took to reach a verdict. They may have been exposed to criticism on social media during this time, but we do not know how this may have affected them and whether it placed pressure on jurors to deliberate in a particular way or to reach a particular verdict. 5.5.5.1 Public access to evidence and information from the courtroom Public access to evidence may also affect juror impartiality or impair public confidence in the administration of justice. It was clear from some tweets that a number of users sought first-hand access to evidence, including the recording: ‘So far, I've only found a couple of links to extracts from #Tostee recording - can anyone direct me to longer ones!’; ‘where do I find the recording?’; and ‘does anyone have a transcript of the recording?’. Prior to the internet, it was not possible to listen to recorded evidence 514 ‘The contemporary view is that it is not necessary for the jury to be sequestered with exposure to the media thereby eliminated.’ See Skaf v R (n 242) [28]. Twitter trials and Facebook juries 150 unless you were physically present in court. Yet, today, users apparently want to see, listen to or view evidence themselves. If the public has access to evidence from the courtroom, then jurors will have access as well. While jurors are directed not to access information about the trial outside the courtroom, there have been reports of jurors actually doing this.515 Listening to the recording from this trial is distressing. In court, counsel for the parties determined when and how often extracts from the recording were played to the jury. But outside the courtroom, jurors could have accessed the entire recording and listened to it as often as they chose. They may do so with good intentions. They might want to refresh their memory. But having unfettered access to in-court evidence may taint a juror’s perspective of the evidence, even if they do not think it will. This type of access was not possible in the past and may represent an entirely new range of risks to juror impartiality. 5.6 Summary In this chapter, I identified the prevalence of prejudicial publicity in tweets during the murder trial of Gable Tostee. I analysed how professional journalists and non-journalists talked about the trial, and investigated how social media users showed prejudice on Twitter. I found that 17.4% of tweets in my sample contained prejudicial information and had the potential to breach the sub judice rule. I found that professional journalists were more careful with their language choices and their tweets were mostly legally compliant. In contrast, I found the tweets of non-journalists were 515 Owen Bowcott and legal affairs correspondent, ‘Juror Jailed over Online Research’, The Guardian (online at 23 January 2012) 151 Murder trial of Gable Tostee – Facebook Analysis more likely to be emotive, lack impartiality and contain prejudicial information. Through both qualitative and quantitative analysis I identified a distinct pro- prosecution bias in my sample. Of all tweets that expressed a clear view as to guilt or innocence, 58% expressed the view the accused was guilty. Even more concerning was that of all tweets that expressed a pro-prosecution or pro-defence perspective, 82% were pro-prosecution. I again raised concerns about cumulative negative sentiment, where the collective effect of multiple tweets working together may have the potential to influence jurors, regardless of whether each individual tweet is legally compliant. In this case study, I also identified tweets that had the potential to be prejudicial because they lacked fairness or accuracy, or because they did not appear to be made in good faith and without malice. Finally, I discussed those tweets that may impair public confidence in the administration of justice. 6 Murder trial of Gable Tostee – Facebook Analysis 6.1 Overview In this Chapter, I present my analysis of Facebook comments made by non- journalists in response to trial-related news articles posted on Facebook by professional journalists during the high-profile murder trial of Gable Tostee.516 I use a combination of content and discourse analysis to understand the prevalence of prejudicial publicity in my sample, to compare how professional journalists and non- journalists talk about the trial, and to investigate how social media users show 516 R v Gable Tostee (n 444). Twitter trials and Facebook juries 152 prejudice on social media during high-profile criminal trials. This analysis makes it possible to compare my findings across both Twitter and Facebook. From this analysis, I found that 62% of comments in my Facebook sample contained prejudicial publicity. All comments appeared to have been contributed by non-journalists, rather than professional journalists. I again conclude that the sub judice rule is not well suited to regulating the behaviour of non-journalists. It is also not well adapted for addressing a prosecution bias, cumulative negative sentiment, or comments that are inaccurate, not made in good faith, or that impair public confidence in the administration of justice. Further consideration must be given to determine whether social media organisations and professional media organisations ought to bear some responsibility for prejudicial information about high profile criminal trials published on social media. 6.2 Prevalence of prejudicial publicity To understand the prevalence of prejudicial publicity in my dataset, I coded my total Facebook sample (1,575 comments) using my coding schemed set out in Chapter 3. After coding comments as containing low-level, high-level or no (none) prejudicial publicity, I found that 62% of comments in my sample contained potentially prejudicial content. As set out in Figure 6.1, this comprised 56% of Facebook posts containing low-level prejudice and 6% of posts containing high-level prejudice. 153 Murder trial of Gable Tostee – Facebook Analysis High PREJUDICIAL CONTENT 6% None 38% None Low High Low 56% Figure 6.1: Tostee Facebook analysis (total sample) - 62% of posts contained prejudicial content. As for my Tostee Twitter analysis, I also coded the comments based on categories of prejudice. My coding results are set out in Table 6.2 below. This approach allowed me to make comparisons across both Twitter and Facebook, to see whether there were any differences between platforms in how social media users talked about the trial. I did not further break down these percentages by publication, as the number of comments available for each publication varied greatly and would not have produced any accurate indication of differences between the four news organisations. % of total Coding Types of potentially prejudicial information Number comments level (1,575) 1. Statements as to guilt 96 6.1% High 2. Information about confessions 0 0.0% High 3. Information about prior convictions 3 0.2% High 4. Statements as to innocence 43 2.7% Low 5. Content that criticises or disparages the accused 410 26.0% Low 6. Content that creates sympathy for the accused 294 18.7% Low 7. Content that creates sympathy for a victim 79 5.0% Low 8. Claims that prejudge an issue of fact or law that is awaiting 50 3.2% Low judgement (other than guilt or innocence) 9. All other content 600 38.1% None 10. Claim supporting prosecution (above categories 1-7) but with 0 0% None attribution (Crown Prosecutor Glenn Cash) 11. Claim supporting Defence (above categories 1-7) but with 0 0% None attribution (Defence Counsel Saul Holt QC and solicitor Nick Dore) Table 6.1: Tostee Facebook analysis (total sample) - results after coding for prejudice. Twitter trials and Facebook juries 154 The percentage of prejudicial publicity contained in my Facebook sample (62%) is significantly higher than in my Twitter sample for this trial (17.4%). These results are also higher than my sample in the Baden-Clay murder trial, where 6% of the overall sample contained prejudicial publicity. As all comments in my Facebook sample appear to be contributed by non-journalists, it is not surprising there is a greater percentage of potentially prejudicial content. Whereas on Twitter, where the most prevalent users were professional journalists reporting the news, on Facebook the conversations are not reports of the news but instead discussions about the news. These discussions include the expression of strong opinions, criticisms, judgements, coarse language, and inflammatory claims. This may explain the significant difference in results between the two social media platforms. It may also be that as Facebook posts are longer than Twitter posts, there is the opportunity for users to be more detailed in their speculation or advancing of controversial opinions. Regardless of the cause, the effect of these results may be that users (including jurors) who are exposed to conversations about a trial on Facebook are more likely to encounter prejudicial publicity than those exposed to tweets on Twitter. Regardless, the potential for exposure to prejudice on either platform is concerning. 6.3 Professional journalists versus non-journalists To understand how professional journalists and non-journalists talked about the trial, I investigated whether the user (publisher) who posted each comment was a professional journalist or non-journalist. The TrISMA data did not include user details for Facebook in the same way as for Twitter. However, by understanding how Facebook public pages work, and by reading every comment, I concluded that the comments in my sample were all, almost certainly, posted by non-journalists — see Figure 6.2. I explain this conclusion in further detail below. 155 Murder trial of Gable Tostee – Facebook Analysis professional journalists, 0% USER TYPE non- professional journalists journalists, non-journalists 100% Figure 6.2: Tostee Facebook analysis (total sample) – all comments were posted by non-journalists. On Facebook, it was professional journalists who uploaded the original news articles, but it appeared to be everyday consumers of news who posted comments in response (either to the article or to other comments). As a result, my Facebook dataset contains what are, essentially, private conversations between individuals playing out in a public forum. There were no comments in my sample that included attribution (more commonly used by professional journalist) and comments did not include any links, references, @mentions or other information that would suggest they were posted by professional journalists. They also used language consistent with non-journalists in my previous sample. That 100% of comments in my Facebook sample were posted by non-journalists makes it quite different from my Twitter samples, where approximately only a third of tweets were posted by non-journalists.517 The style of language used in my Facebook sample was particularly opinionated, emotive, and aggressive, rather than professional. Users tended to adopt a winner or loser discourse, in favour of either guilt or innocence — or Wright or Tostee. There 517 Non-journalists posted 38.4% of all tweets in the Tostee trial and 35.2% in the Baden-Clay trial. Twitter trials and Facebook juries 156 was bickering amongst users who were highly critical of each other and engaged in name calling and vicious hate-talk. Those conversations appeared to be fuelled by differences in opinion about the appropriate verdict, but also by users’ perspectives on gender, race, violence against women, and feminism. It appeared that males more frequently supported Tostee, while females more frequently supported Wright. It was females, however, who appeared to be more commonly criticised for expressing their views: ‘f***ing feminazis’. While non-journalists in my Twitter samples also used a similar style of discourse, there appeared to be a greater use of inflammatory, aggressive and coarse language in my Facebook sample. There were no professional bare-fact reports — likely because comments were not posted by professional journalists. Also, as users were not limited to 140 characters in their conversations on Facebook, some wrote long and structured opinions about the evidence, the accused, and the victim. Some of their posts contained detailed and persuasive perspectives about who was in the wrong and the guilt or innocence of the accused. This represents a significant shift in the way news is published. In the past, news about criminal trials would never have been published together with unedited comments from readers, who may or may not know the facts or evidence in the case. News articles were traditionally intended (at least theoretically) to represent a neutral account of what had taken place, but comments by private individuals are unconcerned with neutrality, expressing instead opinions and judgements that often contain bias. 6.4 Prejudice on social media I used discourse analysis to better understand how social media users showed prejudice on social media during high-profile criminal trials. As my sample only included comments by non-journalists, my observations specifically apply to them. As for my Twitter analysis, I found comments in my Facebook sample were problematic 157 Murder trial of Gable Tostee – Facebook Analysis for a number of reasons. These included a prosecution bias, cumulative negative sentiment, a lack of fairness and accuracy, comments that did not appear to have been published in good faith and without malice, and the potential to impair or undermine public confidence in the administration of justice. 6.4.1 Prosecution bias Although comments included discussion both in favour of and against the accused, as for my previous samples there was a clear prosecution bias overall. Of all comments in my Facebook sample, 9% contained an explicit statement about the guilt or innocence of the accused.518 Of these, 69% claimed Tostee was guilty and only 31% that he was innocent. The results were similar when I combined all categories that supported the prosecution or defence. In that case, 64% of comments were pro- prosecution519 and only 36% pro-defence.520 These results are set out in Figure 6.3. GUILTY V NOT-GUILTY PRO-PROSECUTION V PRO-DEFENCE Not- Pro- guilty defenc 31% Guilty e 69% 36% Pro- prosecution 64% Figure 6.3: Tostee Facebook analysis (total sample) – coded categories indicate a prosecution bias. Overall, 37% of my total Facebook sample was pro-prosecution, 21% pro- defence, and 41% neutral. These results are indicative of an overall prosecution bias. 518 See Table 6.1 above. Categories 1 (statements as to guilt – 96 comments) and 3 (statements as to innocence - 43 comments) represent 139 comments from a total of 1,575. 519 See Table 6.1 – coding categories 1, 2, 3, 5 and 7. 520 See Table 6.1 – coding categories 4 and 6. Twitter trials and Facebook juries 158 TOTAL FACEBOOK SAMPLE Pro- Neutral prosecution 42% 37% Pro-defence 21% Figure 6.4: Tostee Facebook analysis (total sample) – more comments were pro-prosecution than pro- defence. 6.4.2 Cumulative negative sentiment I again raise concerns about the potential for the collective weight of low-level prejudicial or negative information across multiple posts on social media to influence jurors. As was the case for Twitter, Facebook users appeared to focus on a number of key themes and, while individual posts were not always prejudicial, the collective commentary had the potential to work against the accused. As discussed previously, my coding data indicated a prosecution bias. This bias is also evident in the themes that emerged from my discourse analysis, where the content appeared generally to be skewed against the accused. I draw this conclusion based on the number and breadth of topics or themes expressing negative sentiment about him, rather than favourable. In Table 6.2, I have set out the recurring themes and examples from my sample based on whether they appear to be weighted against or in favour of Tostee. Overall there are eight categories of common themes that weigh against Tostee and three categories that weight in his favour. I have included some examples of misinformation — which I discuss in more detail later in this chapter — because even though this information was not part of the evidence, it was part of the Facebook conversations and so had the potential to influence jurors. I have not included themes relating to 159 Murder trial of Gable Tostee – Facebook Analysis criticisms of the court, jurors, or neighbour, as they tend not to clearly be in favour of or against Tostee. Overall, there appeared to be far more collective negative sentiment against the accused, rather than in his favour. Negative sentiment was also evident in the Facebook sample as a number of users inserted themselves into the facts and made claims about what they would have done (as either Tostee or Wright) in the circumstances. This had the effect of criticising the parties. For examples, users placing themselves in Wright’s position claimed they would have ‘climbed down’, ‘left the apartment’, ‘called police’ or ‘never gone to his apartment in the first place’. Those placing themselves in Tostee’s position claimed they would have ‘put her out the front door’, ‘recorded her too’, ‘deleted the recording’ or ‘called an ambulance’. As users explained what they would have done in the circumstances, by implication they were also making claims about what Tostee or Wright should have done. This was not evident in either of my Twitter samples or in traditional media in the past. Cumulative negative sentiment is concerning as it is possible jurors may be influenced by the breadth of claims made against an accused, the repetition of those claims, the number of prejudicial comments they are exposed to, or any combination of these. Twitter trials and Facebook juries 160 Tostee - Against In favour TOSTEE IS GUILTY VICTIM BLAMING He must be guilty, because he legally changed his name to Eric Thomas after the incident to It was her fault - she's crazy / a conceal his identity psychopath / suicidal / a self- His choice not to look over the balcony after Wright fell to see whether she was ok is repulsive harmer / weirdo / nutcase and not the choice of an innocent man She was too intoxicated to His choice not to call 000, seek assistance, or render assistance is immoral, not the actions of outsmart Tostee and it backfired an innocent man It was her choice to climb the Instead of calling 000, he calls his lawyer just 38 seconds after Wright falls. He would only do balcony this if he was guilty She could have left at any time His choice to buy and eat pizza after Wright fell are not the actions of an innocent man (post She was violent to Tostee - she offence conduct) threw rocks at him She didn't get down from the TOSTEE SHOULD BE HELD RESPONSIBLE FOR WRIGHT’S DEATH balcony when the neighbour told He should have put Wright out the front door rather than on the balcony her to He should be held responsible for the effects of his home brew vodka on Wright She took the risk of hooking up with someone from Tinder for He should be held responsible for Wright’s death, even if it wasn't murder sex He was so abusive and made her so scared, she had no choice but to try to escape from him, even if it meant risking her life He committed other criminal offences including kidnapping, deprivation of liberty, domestic LEGAL ISSUES violence, assault and rape [Accused not charged with any of these offences] He didn't intend to kill her, so it's not murder TOSTEE IS A BAD PERSON He is a narcissist and sociopath. He is crazy. MISINFORMATION His choice to audio-record the date without Wright's consent is repulsive She knew Tostee was recording He was racist towards Wright them the whole time He has Asperger’s / autism / obsessive compulsive disorder and this is why he lacked empathy He didn't know she'd fallen off towards Wright the balcony when he left the He's gained so much weight since the incident (jokes and criticism) building to get pizza - he thought she was still on the balcony TOSTEE ABUSED OR ASSAULTED WRIGHT He was holding an item that looked like a weapon when he exited the building via the basement He must have been hurting Wright (choking her) when she was screaming ‘no’ - her screams are blood-curdling He should have let Wright leave when she asked. His refusal to give her 'stuff' back is abusive He was abusive to Wright when he continued to cruelly taunt her, such as ‘you're lucky I haven't chucked you off my balcony, you goddamn psycho little bitch’ TOSTEE HAS A CRIMINAL HISTORY He has a criminal history and has served time in jail for drink-driving, high-speed dangerous driving, and evading police He has also been in trouble with the law for making fake identification cards for underage students wanting to get into nightclubs TOSTEE SHOULD BE PUNISHED Threats to Tostee, such as: ‘hope I see u out on the town mate - can’t wait’ Hope he goes to jail SPECULATION WEIGHTED AGAINST TOSTEE Perhaps Tostee raped Wright, based on her comments that he hurt her vagina and his laughter in response He chose to play ‘I believe I can fly’ in the background - perhaps this is symbolic. Perhaps he played this and then forced her over the balcony MISINFORMATION He pushed her over the balcony He held her dangling over balcony to scare her He was trying to push her over the balcony when she is heard screaming ‘no’ on the audio recording He held her at gunpoint or threatened her with a knife through the balcony door - he was holding the gun / knife when he left the building. [Misinformation - although the evidence could not establish the nature of the item Tostee was holding when he left the building] Table 6.2: Tostee Facebook analysis - thematic analysis of comments with examples. 161 Murder trial of Gable Tostee – Facebook Analysis 6.4.3 Lacking fairness and accuracy As for my Twitter analysis, a number of comments in my sample lacked fairness and accuracy. There was a distinct presence of inaccurate information in the Facebook conversations as well as misunderstandings about the law. 6.4.3.1 Inaccuracy In my Facebook sample, there appeared to be an alarming reliance by users on inaccurate information — specifically misstated facts or evidence — to justify their views. As the majority of tweets in my Twitter samples were posted by professional journalists, large portions of those datasets contained bare-fact reports about the trial. However, in my Facebook sample — where comments were all posted by non- journalists — there were far more highly charged criticisms of Tostee or Wright based on inaccurate information. Misstated facts or evidence related to each of Tostee’s role, the neighbour’s role, and Wright’s role in her death. There were also misstatements about the sequence or timing of the events that took place in Tostee’s apartment. While people (jurors) might always speculate about the facts or evidence in a trial, comments in my dataset were not presented as speculative or hypothetical, but were instead presented as accepted facts. To understand examples of misstated facts and evidence, it is first necessary to understand some relevant facts and evidence as they were actually presented in court: Some hours into their date, on the audio-recording we hear Tostee and Wright arguing. It is after midnight and they are both drunk. Wright sounds particularly intoxicated and, arguably, almost incapacitated. She can be heard slurring her words in a way that makes it difficult to understand what she is saying. Tests following her death confirmed her blood alcohol level was extremely high (0.156%). Twitter trials and Facebook juries 162 As Tostee and Wright’s arguments escalate, it sounds like Tostee starts to physical hurt Wright. According to the prosecution case, at this time Tostee was attempting to ‘forcibly and violently’ choke Wright ‘by placing his forearm across her throat’ for up to one minute.521 During the alleged choking, Wright is heard to scream ‘no’ some 31 times in 46 seconds.522 Her screams are blood-curdling and have been described by the prosecution as coming from a person in a state of ‘abject terror’.523 It is accepted both Tostee and Wright were inside the apartment during the alleged choking, but that Tostee then locked Wright on the balcony. Shortly after she was locked out, Wright climbed over the railing in an attempt — it is believed — to escape. There is evidence the neighbour from the floor below briefly saw Wright dangling and said something like ‘you can’t get down that way’ before she fell. It is accepted Tostee remained inside, behind a closed balcony door, for the time Wright was on the balcony. As Wright climbed over the balcony, Tostee is heard to gasp and swear. He is then heard to move around the apartment and soon leave. The audio recording continues. Tostee exits the building via the basement and walks the streets. He can see police and ambulances at the place where Wright must have fallen, but he does not render assistance. He tries, unsuccessfully, to contact his lawyer. He leaves the apartment building and buys a slice of pizza to eat. He speaks to his father by phone and tells 521 ‘Gable Tostee Not Guilty of Murder’ (20 October 2016) 163 Murder trial of Gable Tostee – Facebook Analysis him Wright has fallen from the balcony. Approximately 15 minutes later Tostee’s father picks him up. Their conversations continue to be recorded for some time after. Tostee has confirmed in interviews since the trial that he knew Wright fell and that he made a conscious choice not to look over the balcony for fear of being perceived by potential witnesses to have pushed her. Some discussions about Tostee’s role in Wright’s death were based on misinformation. While Tostee remained inside after locking Wright on the balcony, some claimed he pushed Wright over the balcony or that he held her dangling over the balcony before she fell. Some claimed it was these acts that caused Wright to scream ‘no’ so many times. Others claimed Tostee had raped Wright, drugged her, held her at gun point, threatened her with a knife or otherwise pressured her to climb over the balcony. Some were certain that Tostee did not know Wright had fallen when he left his apartment and bought pizza to calm down. There were claims that Tostee consciously chose to play a ‘symbolic’ song that can be heard on the recording — ‘I believe I can fly’ — before forcing Wright over the balcony. Others claimed changing his name after the event,524 his behaviour that was consistent with Asperger’s, autism, obsessive compulsive disorder, and his criminal history were all grounds for believing him guilty. Some of these claims are true, others are clearly untrue, and others may not be easily discernible as either. There is the potential for jurors who are exposed to these types of claims to become confused about or question the evidence before them. 524 ‘Accused Balcony Killer’s New Name’, NewsComAu (4 February 2015) Twitter trials and Facebook juries 164 There were also discussions about the neighbour’s role in Wright’s death that were based on inaccuracies. The neighbour, who briefly saw Wright dangling from the balcony before she fell, was criticised for ‘looking at [Wright] but not helping her’ and for not ‘grabbing Wright’s legs and pulling her in’. They questioned why Wright did not ‘ask them [the neighbour] for help’, why she ‘refused the help of the neighbour’, and why ‘she didn’t listen to them when they told her not to climb down’, or ‘to climb back up’ or ‘to get down’. Some seemed to believe Wright was ‘sitting on a ledge’ having a lengthy conversation with the neighbour. It seems some users have not recognised how briefly Wright was able to hang-on before she fell and this affected their comments in a way that may have the potential to be prejudicial. Comments about Wright’s role in her own death also contained inaccuracies. Some misattributed motives to Wright, including that she committed suicide or was trying to self-harm: ‘Wright self-destructive, suicidal stupidity’. Others that she wanted to outsmart Tostee, but she was ‘too drunk to pull it off’. Despite clear evidence to the contrary, some claimed Wright went out to the balcony herself and others that she knew she was being recorded: ‘She was aware he was filming and happily participated’. While jurors would have known that these claims did not form part of the evidence presented in court, they also knew there were gaps in the evidence. Comments like these on Facebook have the potential to raise questions in jurors’ minds about alternative scenarios or how to fill evidentiary gaps. There was also misinformation in my Facebook sample about the sequencing and timing of events on the night Wright died. To reach a verdict, jurors require a clear understanding of who said what and when, and who did what and when. It is possible that jurors’ memories of these details may be influenced if they see these details misstated online. There are aspects of the narrative that are particularly confusing but 165 Murder trial of Gable Tostee – Facebook Analysis they are important for jury decision making. These include information about the timing of intimacy and drinking, when Tostee said ‘you’re lucky I haven’t thrown you off the balcony you psycho little b**ch’, when Wright ‘throwing rocks at him’, and what had occurred before the audio started. Users appeared to place these events in different sequences according to those that best suited their perspective on guilt or innocence. This was also important because Tostee’s post-offence conduct was not to be taken into account by the jury, but there was considerable conversation around the sequencing of Tostee leaving his apartment, attempting to contact his lawyer, contacting his father, and eating pizza. We do not know whether these types of conversation — such as Tostee’s immediately attempt to contact his lawyer or his choice to buy pizza — may have on jurors who are attempting to determine his guilt. These types of inaccuracies about facts and evidence would never have been published in the traditional media without committing a sub judice contempt. They are particularly concerning because it was sometimes very difficult to establish whether the contents of a comment were accurate or not. Some were obvious, but the truth was harder to establish for some comments that were near enough to the truth, but different enough to matter. Even as a person who is thoroughly familiar with this case, there were examples where I needed to double-check. Jurors are not meant to ‘go back and check’ facts and evidence by looking online. They must either hand questions to the court — for which they received much criticism in this case525 — or rely on each other’s recollections. Seeking clarification from other 525 As discussed in Chapter 5 (5.5.5 impairing public confidence in the administration of justice), some users in my datasets were very critical of jurors. Twitter trials and Facebook juries 166 jurors may not be straightforward and may not guarantee a correct answer, depending on the knowledge of each juror, the group dynamics, and the personalities of leaders and influencers in the jury. The inaccuracies in my sample were persuasive because they presented perspectives that were possible and even plausible. They were similar to the actual evidence, but also sufficiently different that a juror who accepts them on face value may be swayed in their decision making. A juror cannot ‘unhear’ inaccurate or alternative facts or evidence once they have heard them. We do not know how jurors refresh their memory or reconcile gaps in evidence during the decision making process — it is possible they may, even inadvertently, draw on inaccurate information from outside the courtroom. The courts have found that even judges may be subconsciously affected by information outside of the evidence presented in the courtroom.526 Judges are expected to put it out of their mind — just as they do when they hear prejudicial evidence but declare it inadmissible527 — but they may still be influenced. The law has long presumed that jurors are susceptible to prejudice.528 If highly trained judges can be influenced, it seems difficult to accept that untrained, inexperienced jurors would be resistant to the same influence. The publication of inaccurate information then, must surely be cause for concern. 526 Attorney-General (UK) v British Broadcasting Corp (n 281) 335 per Viscount Dilhorne; R v McInroy (1915) 26 DLR 615, 615; Bell v Stewart (n 281) 433 per Isaacs and Rich JJ; Pennekamp v Florida (1946) 328 US 331, 358; Kerr v O’Sullivan (n 281) 209–210; Civil Aviation Authority v Australian Broadcasting Corporation (n 281) 550; Solicitor-General v Smith (2004) NZLR 2 540, [79]. 527 Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (n 281) 58. 528 Butler and Rodrick (n 79) [6.280]. 167 Murder trial of Gable Tostee – Facebook Analysis 6.4.3.2 Issues of law The other category of comments that appeared to be lacking fairness or accuracy were those including opinions about the law. It can be difficult for non-lawyers to understand the subtle differences between the prosecution’s onus of proof, the defence’s right to silence, the fact a victim need not be shown to be at fault in order to acquit the accused, and that a verdict of ‘not guilty’ does not necessarily mean the accused is innocent.529 Comments in my Facebook sample contained a number of inaccuracies both when discussing these nuances in the law, and when discussing why the accused was charged with murder. In most instances these comments were not coded as prejudicial, because they did not fall within the recognised categories of prejudiced used in my coding, but they still had the potential to influence jurors. Examples of comments containing inaccuracies included those discussing laws that are not relevant in Queensland: ‘castle law – his house, his rules’, ‘fair or equal force’ in Tostee’s actions; and ‘first degree manslaughter’. Others included discussions of offences that were not raised in this case: ‘entrapment … deprivation of liberty … false imprisonment’. As for Twitter, there were also Facebook comments about post-offence conduct — particularly Tostee’s decision to leave the building after Wright had fallen, to call his lawyer, and to buy pizza. The jury was told not to consider post-offence conduct, but it is possible repeated exposure to this type of information may be difficult for a juror to put out of mind when deliberating. This may be particularly so when there are a number of gaps in evidence as there were in this case. 529 It just means there is insufficient evidence for the prosecution to prove its case beyond reasonable doubt: see Woolmington v DPP [1935] AC 462; R v Mullen (1938) 59 CLR 124; Chan Kau v R [1955] AC 206; Brimblecombe v Duncan; Ex parte Duncan (1958) 8 Qd R; Thomas v R (1960) 102 CLR 584; La Fontaine v R (1976) 136 CLR 62; Van Leeuwen v R (1981) 36 ALR 591; Chamberlain v R (No 2) (1984) 153 CLR 521; Hoch v R (1988) 165 CLR 292; R v Falconer (1990) 171 CLR 30. Twitter trials and Facebook juries 168 6.4.4 Impairing public confidence in the administration of justice As for my Twitter sample, in addition to criticising Tostee or Wright, users were highly critical of the judge and the jury. Some expressed anger over their claim the judge ‘told the jury to acquit’ when summing up. Some condemned the jurors for their ‘stupid questions’ to the court and for the length of time they took to reach a verdict. 6.5 Summary In this chapter, I identified the prevalence of prejudicial publicity in comments on the public Facebook pages of selected news organisations during the murder trial of Gable Tostee. I analysed how social media users (all of whom appeared to be non- journalists) talked about the trial and how they showed prejudice on Facebook. I found that 62% of comments in my sample contained prejudicial information and had the potential to breach the sub judice rule. I identified a distinct pro-prosecution bias. Of all comments that included a statement as to guilt or innocence, 67% favoured the view the accused was guilty. Of those comments that expressed a pro-prosecution or pro- defence perspective, 64% were pro-prosecution. I again raised concerns about cumulative negative sentiment, and the potential for comments to be prejudicial because they lacked fairness or accuracy, or because they did not appear to be made in good faith and without malice. Finally, I discussed those Facebook posts that had the potential to impair public confidence in the administration of justice. 169 Sub judice rule not well adapted for social media 7 Sub judice rule not well adapted for social media 7.1 Overview In this chapter, I summarise my results and consider what they mean for the sub judice rule. I explore further why changes in news consumption and participation in the age of social media have the potential to affect jurors in new ways. I discuss how prejudicial publicity on social media may present risks to juror impartiality that did not exist in the traditional media. Through this analysis, I address my overall research question and conclude that the Australian sub judice rule is not well adapted for regulating prejudicial publicity on social media during high-profile criminal trials. 7.2 Summary of results Before considering how well adapted the sub judice rule is for social media, I have set out a summary of the results from my Twitter and Facebook analyses. Specifically, I have summarised those results and findings that I rely on to respond to my research sub-questions. 7.2.1 How prevalent is prejudicial publicity on social media during high- profile criminal trials? The prevalence of prejudicial publicity on social media during the high-profile criminal trials in my sample are set out in Figure 7.1 — 6% of tweets in Baden-Clay trial, 17.4% of tweets in the Tostee trial, and 62% of Facebook comments in the Tostee trial. Twitter trials and Facebook juries 170 Figure 7.1: Prevalence of prejudicial publicity across case studies There are no comparable previous case studies against which I can measure these results, however, as the sub judice rule is concerned with the publication of any prejudicial information, my findings suggest there is cause for concern. It may be that Facebook is even more concerning than Twitter. The law has long recognised that the potential for jurors to be influenced by prejudicial publicity in the media is significant enough that regulation is required. These risks may be even more significant in the age of social media. Although my analysis of two case studies is insufficient to draw generalised conclusions about all criminal trials, it is enough to suggest there may be a broader problem and further investigation is warranted. 7.2.2 How do professional journalists and non-journalists talk about high- profile criminal trials on social media? To investigate how professional journalists and non-journalists talk about high- profile criminal trials on social media, I coded the users in my dataset based on whether they were professional journalists or non-journalists. This allowed me to consider whether professional journalists, who are required by both their institutions and the law to operate within the confines of the sub judice rule, talk about trials differently to non-journalists who are not accountable to an employer and who may not be aware of the law. The fact that journalists reported the news, while non-journalists discussed it appeared to be an important distinction in terms of how they each tweeted or commented about the trial. Professional journalists were more likely to comply with 171 Sub judice rule not well adapted for social media the law — their tweets were factual, objective, professional and impartial. I found non- journalist were less likely to comply with the law — their tweets and comments tended to be more emotive, opinionated and biased. Statistically, non-journalists posted far more prejudicial tweets and comments than professional journalists. In the Baden-Clay trial, 14.7% of tweets contained prejudicial information posted by non-journalists. In the Tostee trial 15.2% of tweets contained prejudicial information posted by non-journalists and all of the 62% of prejudicial comments posted on Facebook appeared to be posted by non-journalists — see Figure 7.2. Figure 7.2: Prejudicial tweets / comments posted by non-journalists By contrast, prejudicial information posted by professional journalists represented just 1.3% of tweets in the Baden-Clay trial, 2.3% in the Tostee trial, and 0% in the Facebook sample — see Figure 7.3. Twitter trials and Facebook juries 172 Figure 7.3: Prejudicial tweets / comments posted by professional journalists These findings suggest the sub judice rule is largely effective in regulating the way professional journalists report and communicate news about criminal trials. However, it is largely ineffective in regulating the way non-journalists communicate about trials. Non-journalists are far more likely to post potentially prejudicial tweets and comments on social media than professional journalists. 7.2.3 How do social media users show prejudice on social media during high- profile criminal trials To investigate how users show prejudice on social media, I analysed individual tweets and comments using my coding scheme based on the identified categories of sub judice contempt. I also considered tweets and comments collectively to identify broad themes of potentially prejudicial publicity. Through qualitative analysis, I identified a prosecution bias in my Baden-Clay sample. I was able to take this a step further in my Tostee datasets using quantitative analysis. By looking only at data that included statements as to guilt or innocence, and then only at data that tended to be either pro-prosecution or pro-defence, I found a distinct bias towards guilt and a pro-prosecution narrative — see Figure 7.4. This bias can also been seen as a proportion of the overall samples. There appears to be a tendency for both professional journalists and non-journalists to focus on the 173 Sub judice rule not well adapted for social media prosecution narrative and to largely ignore the defence narrative by comparison. This has the potential to lead to a more general public sentiment that is weighted against the defendant. It is particularly concerning if this bias is indicative of a general trend for social media conversations to be weighted against the defendant.in criminal trials. Figure 7.4: Tostee Twitter and Facebook pro-prosecution bias I also found that even when tweets or comments are legally compliant, widespread collective or cumulative negative530 sentiment — caused by the collective weight of low-level negative sentiment — may have the effect of causing prejudice. It is possible this collective weight of low-level prejudicial information across multiple posts on social media may work to influence juror impartiality. Many tweets in my sample appeared to be non-prejudicial when considered individually, but when read together their sequencing or the way they contextualised other posts may work together to form a narrative that is prejudicial in effect. Similarly, individual negative tweets that seemed relatively harmless on their own, may work together with hundreds or thousands of other negatives tweets or comments such that their repetition or amplification may result in prejudice. 530 In certain cases, there may be widespread collective or cumulative positive sentiment—in favour of the accused—and this would also have the potential to cause prejudice in favour of the accused. The sub judice rule is concerned with both negative and positive prejudice. Twitter trials and Facebook juries 174 The law is limited, however, to preventing prejudice in individual publications that are published by individual publishers. As a result, it cannot address the potentially prejudicial effects of multiple publications (tweets), posted by multiple social media users, that collectively demonstrate a prosecution bias or cumulative negative sentiment. The categories of sub judice contempt are not closed. Normally the law has been able to adapt so that publications that are prejudicial in effect, but not within one of the recognised categories, will still be prohibited under the doctrine of sub judice contempt. The difficulty here is that none of the tweets read in isolation will cause prejudice, it is only together that their cumulative effect is likely to prejudice and the law cannot deal with that. The law is also not suited to regulating tweets that are fair and accurate reports of judicial proceedings, and are therefore excused from the sub judice rule, but that nonetheless have the potential to influence jurors. Further, I found that tweets that lack fairness or accuracy, as well as those that did not appear to have been made in good faith and without malice, may work either individually or collectively to potentially prejudice jurors. This may particularly be the case for jurors who, either consciously or subconsciously, use online information to refresh their memory or to fill gaps in evidence. Tweets or comments that lacked fairness or accuracy tended to be biased, misleading or withhold material facts. They sometimes included inaccurate information or misinformation, and sometimes included significant errors in their descriptions and discussions of the law. Those that did not appear to be made in good faith generally included humour (including black humour) and tended to treat the trial as a public spectacle and form of entertainment. They sometimes included information presented in a way that may have been crafted to attract views or ‘clicks’, or that encouraged viewers to participate in voting, polls and betting on the verdict. 175 Sub judice rule not well adapted for social media 7.3 How well adapted is the sub judice rule? The Australian sub judice rule is not well adapted for regulating prejudicial publicity on social media during high-profile criminal trials. Post-internet changes to the distribution of news, as well as the way users consume and participate in news, mean that jurors may be exposed to potentially prejudicial publicity in new and different ways to the traditional media era. The affordances of social media platforms allow users to be participants in what they read rather than asynchronous consumers. Conversations are visible to large audiences and they take place in a more public way. The content of users newsfeeds are largely determined by personalised algorithms, in a way that may be tainted by the views of those in the users’ social network. Social media content may also have the potential to affect jurors’ emotions and decision- making. All of these factors have the potential to contribute to prejudice. These changes mean the presentation of news has shifted from a centralised environment to one that is decentralised. This has had the practical effect of shifting news consumers (including jurors) away from the highly regulated traditional media to the largely unregulated social media, not because the law does not apply on social media but because it has not caught up with this new environment. It is clear from my case studies that prejudicial publicity that potentially breaches the sub judice rule exists on social media, however, the law is not generally being applied in a way that would capture these posts. The law was designed to respond to news reports and not to conversations about the news like those taking place on Twitter and Facebook. The law is still working reasonably well for professional journalists, but it is generally not working for non-journalists. It is also clear from both my Twitter and Facebook analyses that the sub judice rule is not well suited to regulating material that is massively disbursed and communicated collectively by large numbers of users Twitter trials and Facebook juries 176 on social media, where no single tweet would justify pursuing the account holder legally. There is the potential for trials to be unfairly prejudiced by adverse media coverage that falls outside the ambit of the law, such as that caused by multiple publications or multiple user accounts or multiple platforms. 7.3.1 Facebook and Twitter The results of my Facebook analysis reinforce my Twitter findings that the sub judice rule is not particularly successful in regulating the way non-journalists post on social media. This study is not large enough for my findings to be generalised across all criminal trials but, when it comes to the potential to influence jurors, my results suggest some social media may be safer than others. Facebook appears to contain a much higher percentage of prejudicial publicity, likely because posts are by non-journalists, and a significant number of comments contain content that is biased, negative, inaccurate, not made in good faith, or that criticises the judge and jury. While particularly offensive comments might occasionally be deleted through Facebook’s content moderation process, comments are not moderated based on factual, legal or evidential accuracy or on their potential to prejudice jurors. For jurors who access social media during a trial, these results suggest they may encounter more prejudicial publicity on Facebook. This does not mean encountering prejudice on Twitter is not concerning — it is — it just has a lower level of prejudice because of the platforms high proportion of professional journalist users.531 This means the majority of tweets likely comply with the sub judice rule, whereas the majority of Facebook comments do not. 531 Tostee trial: 61.6% of tweets were posted by professional journalists in the random sample of 4,629 tweets; Baden-Clay trial: 64.8% of tweets were posted by professional journalists in the random sample of 7,427 tweets. 177 Sub judice rule not well adapted for social media Just as the law does not appear to be well suited to regulating prejudicial publicity on Twitter, nor is it well equipped for addressing prejudicial Facebook comments. The presentation of news articles with embedded opinionated comments did not exist in the traditional media and users in my sample did not appear to be swayed by the law. When, on occasion, a user raised concerns about the content of another user’s comments, they were quickly and aggressively ‘shut-down’: ‘F**k off, I’ll say what I want’. It is difficult to know how the presentation of news and comments together may affect jurors. Social media platforms are specifically designed to encourage user interaction and participation. Professional journalists may post the original news stories, but it is everyday consumers of news who use the affordances of social media platforms to easily comment on what they have seen. These affordances may even encourage the publication of prejudicial information and, if so, consideration must be given to whether platforms themselves ought to bear some responsibility for unlawful material online. If they should, much work is needed determine how these corporations can be held accountable for encouraging users to engage in conversations that may be against the law. 7.3.2 Law was well adapted for traditional media Traditionally, the sub judice rule was well adapted for the media environment in which it operated. Traditional media organisations were stable gatekeepers of the news — they took advantage of their scarce access to expert source material, their exclusive means of media production, and their control over the distribution of news to wide audiences.532 They were an exclusive group of publishers against whom the law could 532 PJ Shoemaker and TP Vos, Gatekeeping Theory (Routledge, 2009); RG Picard et al, ‘Historical Trends and Patterns in Media Economics’ in Hand- Book of Media Management and Economics Twitter trials and Facebook juries 178 be enforced. However, the law is more limited in the age of social media, as large volumes of potentially prejudicial material may fall outside the ambit of the sub judice rule. The law is concerned with the possibility of prejudice, not the actuality,533 so it seems appropriate that it would address the broad risks of prejudice caused by prosecution bias and cumulative negative sentiment. However, as the law must be applied to an individual publisher, it cannot be used to seek legal redress in response to the effects of multiple publications (tweets) posted by multiple users (publishers). This means the sub judice rule cannot address groups of social media users whose posts may collectively be responsible for prejudice.534 There is no way of enforcing existing laws in a way that addresses prejudice arising from multiple publications collectively, and the effect of reducing this threshold is that everyone would be in contempt. 7.3.3 Alternatives shift the burden away from the state Beyond the sub judice rule there are legal alternatives that may address, at least in part, issues of prosecution bias or cumulative negative sentiment. However, these alternatives may not always be suitable and they can be costly to society, offenders and victims.535 For example, before a trial, a defendant who is concerned about prejudicial publicity and the inability to receive a fair trial may make application for a (Lawrence Erlbaum, 2006); Seth C Lewis, ‘The Tension between Professional Control and Open Participation: Journalism and Its Boundaries’ (2012) 15(6) Information, Communication & Society 836, 838 (‘The Tension between Professional Control and Open Participation’). 533 Bell v Stewart (1920) 28 CLR 419, 432 (Isaacs and Rich JJ); A-G (NSW) v John Fairfax and Sons Ltd (1980) 1 NSWLR 362, 368 (The Court); R v West Australian Newspapers Ltd; Ex parte DPP (WA) (1996) 16 WAR 518, 531 (The Court). 534 The actus reus of sub judice contempt has three elements. There must be (i) a publication of material: R v Griffiths; Ex parte Attorney-General [1957] 2 QB 192, 202 (The Court); (ii) which is published whilst the criminal proceeding is sub judice (‘under a judge’ or pending): James v Robinson (1963) 109 CLR 593, 615 (Windeyer J); and (iii) as a matter of practical reality, the publication must have the requisite tendency to interfere with the course of justice: A-G (NSW) v John Fairfax & Sons Ltd (1985) 6 NSWLR 695, 697 (Samuels JA). 535 New South Wales Law Reform Commission, Contempt Discussion Paper, above n 6, 42. 179 Sub judice rule not well adapted for social media no jury order (or trial by judge-alone).536 These types of applications have the advantage of taking into account not only a single prejudicial publication, but instead the cumulative effect of multiple publications containing adverse publicity.537 In a 2015 case, the Court granted a no jury order based on the cumulative effect of a substantial volume of both social and mainstream media publicity that was prejudicial to the accused.538 During a trial, a defendant who is concerned about prejudicial publicity may make application for the jury to be dismissed. For example, it was the effects of multiple publications that caused a 2016 case to be aborted mid-trial because of the ‘cumulative effect of the errors and overstatement of … the crown case’ in media reports.539 Justice Blokland noted that ‘none of the reports referred to … by themselves would … justify discharging the jury’, however, it was their cumulative negative effect that had the ‘potential to seriously prejudice the trial’.540 In another case in 2016, the cumulative negative effect of media publications — that humiliated an accused and were likely to have an enduring impact on her reputation — was relied on as a mitigating factor in sentencing.541 536 Criminal Code Act 1899 (Qld) ss 614–615E. The right to apply for a no jury order was added to the Criminal Code Act 1899 (Qld) in 2008: Criminal Code and Jury and Another Act Amendment Act 2008 (Qld). See also, Supreme Court Act 1933 (ACT) s 68B; Criminal Procedure Act 1986 (NSW) ss 132–132A; Juries Act 1927 (SA) s 7; Criminal Procedure Act 2004 (WA) pt 4 div 7. 537 For examples of high-profile cases where the accused has made an application for a trial by judge alone on the basis of cumulative prejudicial publicity – some successfully and others not – see R v McNeil (2015) 250 A Crim R 12; R v Gittany [2013] NSWSC 1503; R v Patel [2012] QSC 419; R v Fardon [2010] QCA 317; Dupas v The Queen (2010) 241 CLR 237; R v Ferguson; Ex parte A-G (Qld) (2008) 186 A Crim R 483. 538R v McNeil (2015) 250 A Crim R 12, 16 [13], 24–6 [51]–[62], 29 [72] (Johnson J). In the same case, the Court expressed concern at the ease and speed with which members of the public (including the jury panel) could engage in electronic searches to access the range of prejudicial information (23 [45], 33 [106], 33–34 [111]–[112]). 539 Avani Dias, ‘Mulhall Trial Aborted, Judge Blames “potentially Prejudicial” Reporting’, ABC News (Text, 20 April 2016) Twitter trials and Facebook juries 180 Although these alternatives exist, they will come at a cost to the parties involved. For a defendant this means bearing responsibility for both bringing an application and paying the associated legal costs. For victims this may mean facing a second trial and for society, these applications and outcomes can be costly for the courts and the criminal justice system. Significantly, these alternatives have the effect of shifting the burden and costs for prosecuting those responsible for prejudicial publications away from the crown or the state and onto the parties involved in the trial, some of whom may have insufficient means to protect their rights. 7.4 News on social media is a change in kind For the law to evolve in response to new technologies, it is important to understand whether changes in technology are ‘just an improved version of something [we’ve] already seen, or something else entirely’.542 Small changes, such as an improvement to an existing capability will be a change in degree – it ‘[does] not fundamentally reshape our culture’.543 Larger changes — such as technological leaps — will be a change in kind, as they are ‘not just an enhancement of an old technology but something new unto themselves’.544 The law can generally adapt to small changes, but for large changes an entirely new approach might be needed. In the context of this study, it is useful to consider whether the presentation of news on social media is just an extension of what already existed in newspapers, television and radio, or whether it is something else entirely. As publications can now be placed online or on social media (in addition to newspapers, television and radio), 542 Rebecca J Rosen, ‘The Thorny Combination of Old Laws and New Tech’, The Atlantic (10 November 2011) 181 Sub judice rule not well adapted for social media at first it may seem that this new technology is just a change in degree — the creation of a new place where professional journalist can report the news. However, when we consider the participatory nature of social media we can see a fundamental reshaping of our culture. That non-journalists can publish to the public at large, that they can consume news anywhere anytime, that they can interact and engage with news through the affordances of a range of platforms, suggests these changes are something new unto themselves. Non-journalists are no longer asynchronous consumers of news. This argument is even more compelling when we factor in the realities of algorithms and automation,545 and their effects on the distribution of news. That these shifts likely represent a change in kind may help explain why the sub judice rule is so poorly suited to social media, and why a new approach may be needed. In this section, I expand on these features of news on social media that make it so different from the traditional media. These include new types of audiences, new affordances of platforms, and the fact jurors have no control over personalised algorithms. I also discuss the decentralised nature of social media and how, in the context of prejudicial publicity, it remains largely unregulated. 7.4.1 Large audiences and affordances of platforms The internet has had a profound effect on the way people communicate. The emergence of Web 2.0 and social media has made it possible for ordinary individuals to publish information to large audiences in ways never before possible. Any person is able to use social networks, blogs, online forums, and other online tools to write, post, comment, or participate, and publish their contribution to social networks in a 545 Seth C Lewis, Amy Kristin Sanders and Casey Carmody, ‘Libel by Algorithm? Automated Journalism and the Threat of Legal Liability’ (2019) 96(1) Journalism & Mass Communication Quarterly 60 (‘Libel by Algorithm?’). Twitter trials and Facebook juries 182 highly public and widely visible way. The affordances of social media platforms also allow everyday users to be active participants in news and this has added a new dimension to the way users share information about news stories, including stories about high-profile criminal trials. Social media platforms are specifically designed to be social. They are structured to ensure users can instantaneously share or amplify particular posts, for example, by retweeting, instant messaging, emailing, liking, loving, or commenting. Participating in news in this way was simply not possible, or even imagined, in the days of traditional media. This new ‘omnipresence of social media’ has weakened the traditional mechanism for regulating prejudicial publicity.546 There has been a distinct shift away from the centralised news environment of the traditional media era, to the decentralised news environment of social media. Finding a balance between a defendant’s right to a fair trial and the media’s right to free speech has always been challenging, but in the context of social media this balance is becoming ever more elusive. Increasingly, social media platforms are the ‘places’ where large numbers of users ‘congregate’ for the purpose of engaging in everyday conversations, including those about criminal trials. The ‘court of public opinion’ has the potential to play out on social media through public discussions, where prejudicial publicity may be dispersed or amplified by anyone from anywhere. Social media platforms, like Twitter and Facebook, allow users to make comments in response to news items, or post remarks about criminal trials, with minimal moderation. These conversations take place in a much more public way than in the traditional media, and highly prejudicial posts may become or remain visible to large audiences without any regulation and regardless of whether they have the potential to influence jurors. 546 McEwen, Eldridge and Caruso (n 146) 127. 183 Sub judice rule not well adapted for social media 7.4.2 Users (including jurors) have no control over personalised algorithms Given that many consumers rely on social media platforms as their primary source of news,547 it is problematic that information included in social media news feeds is largely determined by the algorithms governing each particular platform. Even if jurors do not deliberately seek out information about a trial, social media is a major source of news and potentially prejudicial publicity may still be presented in users’ personal news feeds. Social media platforms do not provide any option for individual users to filter out information about a trial, meaning a user has limited control over the content they see. The most they can do is ‘mute’ or ‘hide’ people or pages, but this may have unwanted consequences such as disconnecting users from their networks — a consequence that is growing in significance as our lives become increasingly interwoven with social media platforms. Information presented to users (jurors) may also be selectively amplified and coloured by the views of their friends and acquaintances548 in ways that might include bias or a tendency to support a particular verdict. Some researchers suggest this polarisation can have the effect of isolating users from those viewpoints with which they do not agree, such that they only view media with which they do agree,549 while others say there is limited evidence of this.550 If polarisation exists even within some users’ newsfeeds, that this type of exposure might influence jurors is certainly cause for concern. 7.4.3 Social media decentralised and largely unregulated This shift from a centralised to a decentralised news environment has made the sub judice rule difficult to enforce, as prejudicial content on social media is largely 547 Watkins et al (n 157) 51, 55, 60. 548 Michela Del Vicario et al, ‘The Spreading of Misinformation Online’ (2016) 113(3) Proceedings of the National Academy of Sciences 554, 554, 558. 549 ‘New Media Analysis’ (n 155) 1092. 550 Axel Bruns, ‘Echo Chamber? What Echo Chamber? Reviewing the Evidence’ (2017) Twitter trials and Facebook juries 184 unregulated. Traditionally, media regulation operated in a top-down, hierarchical mode. The law was applied to people and firms who were responsible for high profile publications that had significant impact — typically journalists, editors, producers and proprietors of media organisations.551 The sub judice rule was also incorporated into the institutions of mainstream media publishing. Since media institutions would be liable if their journalists breached the law, there was always a strong incentive to exercise control over the conduct of staff.552 As such, media law served as the impetus for organisations to set up extensive compliance processes to ensure any material they published complied with the law.553 It was also embedded within ethical codes of practice, in university journalism subjects,554 and in the normal workflows of mass media organisations. Enforcing the sub judice rule in the traditional media era was relatively straightforward, as the chain of those who were held responsible — and upon whom obligations were imposed — was relatively small.555 In simple terms, the law applied to ‘publishers’ and this meant it only needed to contend with a relatively small and easily identifiable body of media professionals. However, the term ‘publisher’ means 551 New South Wales Law Reform Commission (n 77) 8. 552 Ibid. 553 Mark Pearson and Mark Polden, The Journalist’s Guide to Media Law: A Handbook for Communicators in a Digital World (Allen & Unwin, Fifth edition, 2015) 7–10 (‘The Journalist’s Guide to Media Law’). 554 See, eg, the Media, Entertainment and Arts Alliance Code of Ethics requires, inter alia, that journalists report honestly, fairly and without distortion: Media, Entertainment and Arts Alliance, ‘MEAA Journalist Code of Ethics’ (Code of Ethics, February 1999) 185 Sub judice rule not well adapted for social media something different today. It no longer refers only to those in the profession of media or publishing, but also to any individual social media user who posts or comments online. The sub judice rule was conceived in an era when the notion of a decentralised news media could never have been anticipated. As such, the law has typically been applied against professionals and their organisation, and not against individuals as ordinary consumers of media who might make prejudicial comments to relatively small audiences when discussing a trial.556 There are occasional exceptions where individuals are pursued for contempt, but more often they are for publishing material in breach of non-publication or suppression orders, rather than for sub judice contempt.557 This shift to a decentralised media environment also means there are now potentially millions of ‘publishers’ to whom the law might apply and this presents an overwhelming challenge to law enforcement. There is now a vast number of speakers who publish content online and this democratisation of speech bypasses the traditional gatekeepers of the mass media era.558 The flow of news information was once controlled by publishers, broadcasters, producers, and others who were the traditional targets of regulation,559 but through decentralisation they have lost that control. This loss of control has made the regulation of speech by governments through media law 556 Griffith, Kossian and Kowalczuk (n 174). 557 Michaela Whitbourn, ‘Blogger Shane Dowling Jailed for Contempt for Naming Tim Worner’s Alleged Lovers’, The Sydney Morning Herald (10 August 2017) Twitter trials and Facebook juries 186 extremely difficult. Instead of gatekeeping, social media has seen the emergence of gatewatching, the practice of ‘bloggers and citizen journalists, who predominantly observe and comment on the stories published by mainstream news sites rather than engaging in their own original news reporting’.560 Social media now plays ‘a crucial role … [in disseminating] news and information’ with platforms like Twitter and Facebook serving ‘to further establish gatewatching practices’ such that they have become ‘widespread and influential’.561 Not only has the number of ‘publishers’ increased exponentially, but everyday users posting content on social media may not have the capacity, legal training or knowledge to comply with the law. They may not be able to easily evaluate whether their posts are unlawful or they may lack the desire or time to comply with the law. They may also lack awareness of the law itself and this point alone may mean social media users are more likely to post contemptuous material than their more legally- aware mainstream media counterparts.562 7.5 Social media presents new risks to juror impartiality While there are considerable concerns about the potential for prejudicial publicity on social media to influence jurors, some commentators suggest the claims are overstated and unwarranted. In this section, I discuss the assertion that the risk to jurors is the same as it has always been. I then move on to explain why — irrespective of this assertion — there is still reason to think that jurors may be influenced. 560 Axel Bruns, Gatewatching and News Curation: Journalism, Social Media, and the Public Sphere (Peter Lang, 2018). 561 Ibid. 562 Johnston et al, above n 2, 4. 187 Sub judice rule not well adapted for social media 7.5.1 Is the risk to jurors the same as it always has been? One of the principal counterarguments to concerns that jurors may be influenced by prejudice on social media, is that social media conversations are no different from those that people, or jurors, have always had in person.563 The inference, then, is that jurors are no more likely to be influenced now than they were in the traditional media era. One commentator who takes this view is former judge Frank Vincent AO QC, whose extensive expertise and experience in the Victorian court system cannot be denied. Throughout his career, he has observed that jurors do take their role seriously. He suggests the real issue is that ‘we do not trust jurors’ and contends that juries are in fact trustworthy and can be relied upon to reach an impartial verdict.564 There is clearly merit in this argument — it is likely the majority of jurors take their role very seriously, comply with judicial directions, and aim to fulfil their role impartially. This was also the view of the New Zealand Law Commission that conducted an extensive study in the late 1990’s and found that jurors fulfilled their role conscientiously and did their best to comply with directions.565 Chesterman’s comprehensive Australian study in 2001 also drew similar conclusions.566 However, these conclusions are not straightforward. 563 Andrew Dodd, ‘PODCAST: Pell Trial Reporters, a Judge and a Media Lawyer on Why the Suppression Order Debate Is Far from Over’, The Conversation (21 March 2019) Twitter trials and Facebook juries 188 First, social media platforms — like Twitter and Facebook — did not exist at the time of these studies. Second, research suggests some jurors will not do the right thing — some intentionally and some unintentionally.567 I discuss each of these in further detail below. And third, these studies examined the effects of pre-trial publicity, not during-trial publicity, and the findings rely in part on the conclusion that ‘jurors appear to recall distinctly less pre-trial publicity than is professionally supposed’.568 It may be a different case for during-trial publicity, as it is likely jurors will have a greater recollection of what they see, due to the shorter time between being exposed to the media and then recalling it.569 It is also the case that the findings in this study ‘presuppose the existence of a regime of sub judice restrictions and [that they] would carry little or no validity if that regime were to be wholly or substantially dismantled’.570 The introduction of the internet and social media may have had this very effect of ‘substantially dismantling’ the sub judice rule, particularly due to its reliance on territorial boundaries as discussed in Chapter 2 (Literature Review).571 7.5.1.1 Social media is ‘wrapped around our lives’ Unlike traditional media, social media is ‘wrapped around our lives’. Social media is a relatively new phenomenon — Twitter was only founded in 2006 (San Francisco, California, United States) and Facebook in 2004 (Cambridge, 567 Ibid xv, 109–111; Lorana Bartels, ‘Jurors and Social Media: Is There a Solution?’ The Conversation (online), 31 July 2013 189 Sub judice rule not well adapted for social media Massachusetts, United States). As these types of organisations continue their unprecedented rise to power, the risk of being unknowingly exposed to prejudice, or unknowingly influenced by it, is likely on the rise. Even though social media is still in its infancy, it is already infiltrating our daily lives in ways that were never envisaged. When social media first became accessible in Australia, it was likely only used by a small number of tech-savvy individuals. But, today, social media is becoming increasingly interwoven with all we do, in both our professional and personal lives. It is ever-present. Facebook has been said to have ‘very cleverly figured out how to wrap itself around our lives. It's the family album. It's your messaging to your friends. It's your daily diary. It's your contact list. It's all these things wrapped around your life.’572 This reality is making participation on social media increasingly necessary and this rise in new media has increased the possibility of exposure to incidental news.573 This increasing necessity to use social media — combined with a desire by many to use it574 — is having the effect of funnelling users into pipelines of information over which they have limited or no control. As discussed previously, the information in users’ newsfeeds is largely determined by personalisation algorithms and may 572 ‘Cracking the Code: What Facebook Really Knows about You’, Four Corners (Australian Broadcasting Corporation, 10 April 2017) (Dr Suelette Dreyfus, Information Systems Expert, University of Melbourne) Twitter trials and Facebook juries 190 generally be coloured by a user’s own social networks.575 On Facebook, news feeds are ‘determined via a ranking algorithm that Facebook continually develops and tests in the interest of showing viewers the content they will find most relevant and engaging.’576 These methods are similar to those used by targeted advertising and they drive ‘what 'news' appears in your Facebook feed’ in a way that makes users ‘unlikely to see anything that challenges [their] world view’. 577 This is a complicated reality. If an empanelled juror clicks on (or likes or shares) a trial related story, opinion or expression — even though at law they should not — that choice is factored into the algorithmic selection process. This participation may result in a ‘newsfeed filled predominantly with opinions with which the user agrees’.578 This makes it likely they will see more trial-related post and those posts are likely to align with their world view. Unlike traditional media, where the expectation was that journalists present a balanced view of the news, social media contains entrenched bias and this is difficult to avoid: ‘democracy requires a reliance on shared facts: instead we’re being offered parallel but separate universes’.579 In some cases people seek to ‘avoid conflict with the diverse members of their network’ and this is contributing to the ‘polarising of debate’.580 In this way, it is possible jurors may be increasingly 575 Philip Seargeant and Caroline Tagg, ‘Social Media and the Future of Open Debate: A User- Oriented Approach to Facebook’s Filter Bubble Conundrum’ (2019) 27 Discourse, Context & Media 41 (‘Social Media and the Future of Open Debate’); Tanzeela Jameel, Rukhsana Ali and Kamran Ahmed Malik, ‘Social Media as an Opinion Formulator: A Study on Implications and Recent Developments’ [2019] 2019 2nd International Conference on Computing, Mathematics and Engineering Technologies (iCoMET). 576 Kramer, Guillory and Hancock (n 438) 8788. 577 ‘Cracking the Code: What Facebook Really Knows about You’ (n 575) (unnamed Media analyst). 578 Rodney H Jones and Christoph A Hafner, Understanding Digital Literacies (Routledge, 2012) 126. 579 Eli Pariser, The Filter Bubble: What The Internet Is Hiding From You (Penguin UK, 2011) 5 (‘The Filter Bubble’). 580 Seargeant and Tagg (n 578) 42. 191 Sub judice rule not well adapted for social media exposed to — and potentially influenced by — the types of prejudicial information I found in my case-studies. 7.5.1.2 Increased availability of trial-related information Significantly, it has been shown that jurors commonly refer to media reports in their deliberations581 and, at least in some cases, exposure to prejudicial publicity has been considered responsible for causing jurors to reach ‘unsafe’ verdicts.582 Even jurors who regard themselves as unbiased after being exposed to prejudicial content in the media are more likely to reach a guilty verdict than those who have not been exposed to prejudice.583 The challenge here is that jurors may be biased, even when they think they are not. They may be partial, even when they think they are acting impartially. My research is concerned, at least in part, with the potential for jurors to be unknowingly exposed to prejudice or unknowingly influenced by it. Unconscious bias584 and the inability to disregard prejudice585 have the potential to affect decision making, even by the most upstanding and well-intentioned of jurors. It is also the case that some jurors will knowingly engage in misconduct.586 For those jurors, at least, social media and the internet in general provide an extensive source of information and commentary that was not available in the past. There have always been jurors who might choose to disobey judicial directions and undertake their own research, but the outcome of that research may have been less successful than it is likely to be today. Now, vast quantities of information can be found online 581 Chesterman, Chan and Hampton, above n 6, xv. 582 Ibid 63, 109–11. Verdicts were considered ‘unsafe’ where they appeared unwarranted by the evidence and were aligned with publicity, meaning it seemed likely the jury was influenced by the media. 583Ibid. 584 Arterton (n 570). 585 Fein, McCloskey and Tomlinson (n 570). 586 Bartels (n 570). Twitter trials and Facebook juries 192 immediately — through a simple ‘Google search’ — and in the privacy of one’s own home. This type of research is considerably different from that in the traditional media era. No longer does it require a significant investment in time. Nor does it need to take place openly through conversations with peers or acquaintances, or by seeking information at a library, news organisation, or some other relevant location. Today jurors can discreetly and privately search for information without anyone ever knowing. This may mean that jurors are no longer deterred from misbehaving by the risk of ‘getting caught’. There will be even less deterrence for those who — despite judicial directions — do not understand that searching online, in private, is unlawful.587 The perception that judicial directions are ‘often excessively long and complex, making it doubtful that they are understood by [jurors]’ has served as the impetus for the Law Reform Commissions of Queensland, New South Wales and Victoria to consider how ‘short and readily comprehensible’ directions may be given while also ‘ensuring the fair trial of the accused’.588 Our criminal justice system relies on the assumption that jurors will comply with judicial directions and act only on the evidence presented in court — an assumption on which the ‘common law countries have staked a great deal’.589 As McHugh J observed, if we do not act on this assumption ‘there is no point in having criminal trials’.590 Having said this, juries may still return a verdict that is contrary to law.591 The courts have long accepted that although ‘it is the duty of the Judge … to tell the jury how to do right … they have it in their power to do wrong’.592 Jury deliberations take place in secret, after which they 587Hannaford-Agor, Rottman and Waters (n 211) 6. 588 Virginia Bell AC, ‘Jury Directions: The Struggle for Simplicity and Clarity’ (Banco Court Lecture, Supreme Court of Queensland, 20 September 2018). 589 Gilbert v The Queen (n 1) 425 [31]. 590 Ibid. 591 Gammage v The Queen (1969) 122 CLR 444, 451. 592 R v Shipley (1784) 99 ER 774, 824. 193 Sub judice rule not well adapted for social media deliver a general, inscrutable verdict.593 This secrecy, combined with statutory restrictions (with only some limited exceptions), prevent research into jury decision- making594 and mean we cannot know how actual juries reach their verdicts. We can only use research to draw evidence based conclusions as to how we think they may function, and then do what we can to minimise the risks. There is much work still to be done to understand how ‘individual opinions are transformed into jury verdicts’ and how juries approach group decision making.595 7.5.2 Social media can affect how people think and feel Irrespective of counterarguments, there are good reasons to think jurors may be influenced by prejudicial publicity on social media. Facebook research suggests social media can affect how users think and feel, and this may also be true for jurors. The manipulation of social media during the 2016 US presidential election campaign shows the potential to influence users’ decision-making. This same use of new technologies also highlights the potential for new approaches to jury tampering. 7.5.2.1 Emotional contagion caused by news feeds In 2014, Facebook published the results of one of the largest psychological experiments ever conducted.596 The study aimed to determine whether ‘emotional states can be transferred to others via emotional contagion, leading people to experience the same emotions without their awareness.’597 Researchers manipulated the news feeds of almost 700,000 Facebook users598 to be either more positive or more 593 Bell AC (n 591) 3–4. 594 Ibid 2. 595 Jessica M Salerno and Shari Seidman Diamond, ‘The Promise of a Cognitive Perspective on Jury Deliberation’ (2010) 17(2) Psychonomic Bulletin & Review 174. 596 Kramer, Guillory and Hancock (n 438). 597 Ibid 8788. 598 The actual number of user profiles that formed part of the study was 689,003 (N = 689,003). Twitter trials and Facebook juries 194 negative.599 The results of the study showed that ‘emotional contagion occurs without direct interaction between people (exposure to a friend expressing an emotion is sufficient), and in the complete absence of nonverbal cues’. 600 In simple terms, this study showed that users’ moods were affected by what they saw in their news feeds. The more positive the feed, the happier users seemed to be. The more negative the feed, the more depressed they became. This study is said to prove ‘the power of Facebook to affect what we think and how we feel’.601 If exposure to positive or negative emotion can affect Facebook users in a way that causes them to feel the same emotion, what effect then does exposure to prosecution or defence based claims on social media — many of which are highly emotional — have on jurors? It is possible that jurors may be more inclined to take the perspective that aligns with the emotions or arguments to which they are most exposed. This is a similar idea to the ‘illusory truth effect’ where people will believe something ‘just because [they’ve] been told it’s true a lot’.602 On social media, the perspective that users (jurors) will most likely see ‘a lot’, is one coloured by the heterogeneity of their own social network.603 People are more likely to believe stories that favour their 599 Kramer, Guillory and Hancock (n 438). 600 Ibid 8788. 601 ‘Cracking the Code: What Facebook Really Knows about You’ (n 575). 602 Tegan Taylor, ‘Help, My Nan Won’t Stop Sharing Dodgy Posts on Facebook’, ABC News (Current, 14 September 2018) (Dr Jennifer Beckett, Lecturer Media and Communications, University of Melbourne) 195 Sub judice rule not well adapted for social media own perspective — especially if they have ‘ideologically segregated social media networks’604 — and they will filter out things they do not agree with.605 The idea that persons present in court might be influenced by something other than the evidence in a case, without realising it, is not a new concept at law and nor is it restricted to jurors.606 A recent study of 1.5 million judicial decisions in the US showed that a range of extraneous factors affected judges’ decisions, including the weather, sports game scores — including the ‘NFL effect’ based on whether their city’s NFL team won — and the location of the trial.607 An earlier study also showed that judges can be influenced by extraneous factors including time of day, race, or other psychological, political, and social factors.608 If extraneous factors can affect the decision-making of judges — who are said to be able to resist influence in a more robust way than jurors — what does this mean for jurors? It may be possible that prejudicial publicity has a greater effect even than extraneous factors. Kim, Shih-Hsien Hsu and Homero Gil de Zúñiga, ‘Influence of Social Media Use on Discussion Network Heterogeneity and Civic Engagement: The Moderating Role of Personality Traits’ (2013) 63(3) Journal of Communication 498 (‘Influence of Social Media Use on Discussion Network Heterogeneity and Civic Engagement’). 604 Hunt Allcott and Matthew Gentzkow, ‘Social Media and Fake News in the 2016 Election’ (2017) 31(2) Journal of Economic Perspectives 211. 605 Taylor (n 605) (Dr Jennifer Beckett, Lecturer Media and Communications, University of Melbourne). 606 There are divided opinions as to whether judges themselves are likely to be influenced by prejudicial media. Some are of the opinion that judges are not open to influence: see Attorney-General (UK) v British Broadcasting Corp (n 281) 342–343; Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (n 281) 102; Attorney-General v Times Newspapers Ltd (n 281) 301; R v Duffy; Ex parte Nash (n 281) 198; Bell v Stewart (n 281) 425–426. Others concede it is possible for judges to be influenced, see: Attorney-General (UK) v British Broadcasting Corp (n 281) 335; Bell v Stewart (n 281) 433; Kerr v O’Sullivan (n 281) 209–210; Civil Aviation Authority v Australian Broadcasting Corporation (n 281) 550. 607 Daniel L Chen and Markus Loecher, Mood and the Malleability of Moral Reasoning (SSRN Scholarly Paper No ID 2740485, Social Science Research Network, 4 January 2019) Twitter trials and Facebook juries 196 Further research is needed to really understand whether jurors might be inclined to reach a verdict aligned with the perspective to which they are most exposed, but for reasons of human research ethics a study like Facebook’s could never be replicated by a university alone. As a private company, Facebook was not required to comply with the same standards of human ethics that apply to academic research and this has resulted in significant backlash.609 Users were not provided with an opportunity for informed consent or to opt out of the study — concerns so significant that the research paper is prefaced by an ‘Editorial Expression of Concern and Correction’ prepared by the journal’s Editor-in-Chief.610 7.5.2.2 Manipulating voters in the 2016 US presidential election campaign Manipulating what users see on social media is also said to have played a significant part in the 2016 US presidential election campaign.611 Social media platforms were used as a direct source of news for millions of followers, allowing candidates to bypass the traditional editorial and legal processes of mainstream media.612 That regulation does not exist — or, at least, cannot be enforced — in the same way as the traditional media is problematic. It was a legal requirement for political advertisements that were placed on television during the US election to 609 Galen Panger, ‘Reassessing the Facebook Experiment: Critical Thinking about the Validity of Big Data Research’ (2016) 19(8) Information, Communication & Society 1108 (‘Reassessing the Facebook Experiment’); Gregory S McNeal, ‘Controversy Over Facebook Emotional Manipulation Study Grows As Timeline Becomes More Clear’, Forbes 197 Sub judice rule not well adapted for social media disclose who funded them. But this was not the case for social media. Campaign advertisements — concealed as social media news stories — were distributed in a targeted way to users of particular demographics or who were located in key electorates, in an effort to influence their votes.613 Concealing advertisements as stories is not only seen on social media but also in the manipulation of internet search rankings — the search engine manipulation effect — whereby biased search rankings can be designed to influence consumer choices in a way that is focussed on particular demographic groups and where rankings are masked so that users show no awareness of the manipulation.614 There are real concerns about the lack of transparency in these types of data driven, targeted advertising that immerses users in misinformation aimed at manipulating their opinions.615 Misinformation — sometimes referred to as fake news or disinformation — may look like news, function like news, and be shared like news, but it does not ‘match up with traditional ideas of what news is for and what it should do’.616 7.5.2.3 New opportunities for jury tampering Just as social media provided a new forum for manipulation, hacking or tampering with the 2016 US presidential election campaign,617 so too has social media provided a new opportunity for jury tampering.618 Jury tampering refers to any 613 Allcott and Gentzkow (n 607); ‘Cracking the Code: What Facebook Really Knows about You’ (n 575) (Nik Cubrilovic, IT Security expert). 614 Robert Epstein and Ronald E Robertson, ‘The Search Engine Manipulation Effect (SEME) and Its Possible Impact on the Outcomes of Elections’ (2015) 112(33) Proceedings of the National Academy of Sciences E4512, 4512–4513, 4518–4519. 615 ‘Cracking the Code: What Facebook Really Knows about You’ (n 575) Professor Ramesh Srinivasan, Technology, Politics and Society, UCLA. 616 ‘Cracking the Code: What Facebook Really Knows about You’ (n 575). 617 Roberto J González, ‘Hacking the Citizenry?: Personality Profiling, “Big Data” and the Election of Donald Trump’ (2017) 33(3) Anthropology Today 9 (‘Hacking the Citizenry?’). 618 Jeffrey T Frederick, ‘Facebook and Jury Tampering: A New Threat Posed by Social Networking Sites (SNS) to Jury Integrity’, Jury Research Services (8 August 2011) Twitter trials and Facebook juries 198 ‘misconduct in relation to a jury during the course of a trial involving contact being made with a member or members of the jury with a view to attempting to influence the jury’s verdict by persuasion or threat’.619 Examples of contact may include conversations outside of court, bribes, threats, or initiating contact with jurors via acquaintances. Typically, efforts at jury tampering have been quite overt. In 1999, a trial was aborted when, after they had left the court, the jury was approached by a woman wearing disguise who gave them an envelope containing written information that was prejudicial to one of the accused.620 In 2014, a murder trial was aborted after a juror received a phone call from someone trying to influence her verdict in the case.621 In the US in 2004, it was held that the act of government agents’ glaring at jurors during a criminal prosecution constituted jury tampering, regardless of the agents’ intent.622 In the UK in 2003, there were multiple cases of jury tampering, including threats to jurors, offers of bribes, and a note with money left on a juror’s windscreen.623 In 2017, in the US a man was convicted of jury tampering after he handed out juror rights pamphlets to people entering the courtroom in the hopes of swaying jurors.624 In 2013, 619 Jury Tampering: Encyclopaedic Australian Legal Dictionary (Lexis Advance Pacific Research, 2019). There are offences in all jurisdictions in relation to corrupting or threatening jurors: for example, (NSW) Crimes Act 1900 ss 321, 322 ; (QLD) Criminal Code s 122 ; (SA) Criminal Law Consolidation Act 1935 s 245. 620 R v Richards; R v Bijkerk [1999] NSWCCA 114, [7]. 621 ‘Murder Trial Aborted Due to Jury Tampering’, Chronicle (18 November 2014) 199 Sub judice rule not well adapted for social media in the US a ‘star-struck’ prosecutor breached jury tampering rules when she spoke to Oscar-winning actor Tom Hanks, who happened to be a juror in the trial.625 Unlike these overt attempts at jury tampering, social media poses an opportunity for covert jury tampering and this must be cause for concern.626 Challenges to data privacy play a large role in this. The large quantities of personal data that exist on social media627 can be used for purposes different from what was intended.628 Preventing data being captured is very difficult and, in some instances, even impossible.629 While the jury system in the US is different from ours, it is relevant that new technologies of ‘big data’ collection are being applied to jury selection.630 By creating what amount to ‘personalized dossiers on individual jurors’, parties can aim to weight the jury in their favour. But this comes at the expense of privacy and legitimacy.631 It is not unforeseeable that motivated parties might seek to infiltrate juries in new ways through social media with a view to influencing a verdict. Defendants on trial for serious charges often face long jail sentences and — depending on the nature of the charges — the forfeiture of money and assets as proceeds of crime. These are high- stakes and may provide sufficient motivation for defendants to seek to target jurors via social media in an attempt to influence their decision-making, particularly if the 625 ‘Tom Hanks Jury Appearance Results in Mistrial’, ABC News (Text, 13 September 2013) Twitter trials and Facebook juries 200 manipulation can take place covertly and without anyone realising. This risk is not limited to defendants, of course. Victims or other interested parties may also be sufficiently motivated to ‘take matters into their own hands’ in pursuit of a particular verdict. In some instances motivated parties may themselves possess the technological skills to ‘hack’ social media with the goal of tampering with the jury. Others may need to pay for it. As for the US election, approaches aimed at influencing juries may include targeted advertising — concealed as news stories but with a bias in favour of or against the accused, that are targeted at particular social media users. It may include creating misinformation and saturating existing news stories (such as on Facebook public pages) with prejudicial perspectives and misinformation that has the potential to influence jurors. Attempts may be made to influence jurors’ family and friends — or others in their social networks — with a view to those users passing on information. Many of these approaches can be undertaken using human labour, but they can also be achieved with the creation of social bots — ‘algorithmically driven entities that on the surface appear as legitimate users’.632 Bots are powerful communication tools that can be exploited to ‘manipulate online discussion … [and] change the public perception’.633 This risk may be amplified for long trials — which is often the case with more serious charges — because this will increase the time for exposure to prejudice and potential influence. Ultimately, the availability of private data online means it is likely far easier than it once was to identify and make contact with jurors — overtly or covertly. 632 Alessandro Bessi and Emilio Ferrara, Social Bots Distort the 2016 US Presidential Election Online Discussion (SSRN Scholarly Paper No ID 2982233, Social Science Research Network, 7 November 2016) 201 Sub judice rule not well adapted for social media Finally, with the growing use of newswriting bots634 — ‘automated forms of news production [using] machine-driven modes of news creation and publication based on algorithmic procedures’635 — there is the potential for prejudicial reporting even when it is unintentional. Facebook ran into difficulties in 2016 when an erroneous (fake) and defamatory article was automatically amplified through its algorithmic selection of ‘trending topics’636 — resulting in questions about who should be legally responsible: Facebook or the bot?637 These same risks have the potential to play out in publications about criminal trials, as automated and algorithmic news increasingly becomes the ‘norm’. 7.5.2.4 Influencing key jurors may be enough to affect verdicts In order to tamper with a jury or influence a verdict, it is possible the manipulation of key jurors will be sufficient to influence the overall verdict. It only takes the obstinance of one juror to frustrate the determination of the others to convict or acquit when unanimity is required. A number of factors are said to affect the actual 634 Lewis, Sanders and Carmody (n 548) 60. Seth C Lewis and Oscar Westlund, ‘Big Data and Journalism: Epistemology, Expertise, Economics, and Ethics’ (2015) 3(3) Digital journalism 447 (‘Big Data and Journalism’). 635 Lewis, Sanders and Carmody (n 548) 61. There is a growing body of research on automated journalism: see Jaemin Jung et al, ‘Intrusion of Software Robots into Journalism: The Public’s and Journalists’ Perceptions of News Written by Algorithms and Human Journalists’ (2017) 71 Computers in human behavior 291 (‘Intrusion of Software Robots into Journalism’); Carl-Gustav Linden, ‘Decades of Automation in the Newsroom: Why Are There Still so Many Jobs in Journalism?’ (2017) 5(2) Digital journalism 123 (‘Decades of Automation in the Newsroom’); Neil Thurman, Konstantin Dörr and Jessica Kunert, ‘When Reporters Get Hands-on with Robo-Writing: Professionals Consider Automated Journalism’s Capabilities and Consequences’ (2017) 5(10) Digital journalism 1240 (‘When Reporters Get Hands-on with Robo-Writing’); Mario Haim and Andreas Graefe, ‘Automated News: Better Than Expected?’ (2017) 5(8) Digital journalism 1044 (‘Automated News’); Matt Carlson, ‘The Robotic Reporter: Automated Journalism and the Redefinition of Labor, Compositional Forms, and Journalistic Authority’ (2015) 3(3) Digital journalism 416 (‘The Robotic Reporter’); Christer Clerwall, ‘Enter the Robot Journalist: Users’ Perceptions of Automated Content’ (2014) 8(5) Journalism Practice 519 (‘Enter the Robot Journalist’). 636 Will Oremus, ‘How Facebook’s Trending News Feature Went from Messy to Disastrous.’, Slate (30 August 2016) Twitter trials and Facebook juries 202 workings of jury decision-making and these may vary from trial to trial. Factors including juror demographics, personalities and attitudes, as well as pre-trial publicity, inadmissible evidence, the initial distribution of juror verdict preferences during deliberation, and group dynamics have been shown to affect how juries reach a verdict.638 Group dynamics are significant. The group workings of a jury are an example of an ‘autonomous work group’ in which group members are chosen, essentially at random, to perform a function of great importance, for which they generally have had no direct training. They must work in isolation and secrecy, with little or no outside assistance, using procedures and rationales of their own design to produce one product: what will hopefully be a unanimous decision.639 Working in a group in these circumstances contributes to the risk of ‘groupthink’640 in jury decision-making,641 in which members avoid raising controversial issues due to ‘a psychological drive for consensus at any cost that suppresses dissent and appraisal of alternatives’.642 In-group pressures can cause a ‘deterioration of mental efficiency, reality testing, and moral judgement’643 and this is clearly problematic when jurors are deciding verdicts that affect the lives of both offenders and victims. Given what we know about the potential for social media users to be influenced by what they see — combined with the challenges of reaching a group verdict — it is possible that exposure 638 Dennis J Devine et al, ‘Jury Decision Making: 45 Years of Empirical Research on Deliberating Groups’ (2001) 7(3) Psychology, Public Policy, and Law 622 (‘Jury Decision Making’). 639 ‘Jury Dynamics and Decision Making | Consensus Decision Making | Leadership & Mentoring’, Scribd 163 203 Sub judice rule not well adapted for social media to prejudice on social media may also be an important factor in how juries reach a decision, even if they do not realise it. Finally, the dynamics of group decision making likely plays a role in the potential for jurors to be influenced by prejudicial publicity. When reaching consensus as a group during deliberations, jurors may be affected by both informational and normative influence.644 Informational influence occurs where a juror considers statements made by others and decides they are correct. In contrast, normative influence occurs when jurors are influenced by whichever jurors first express an opinion, ‘simply because people are inclined to agree with the group and endorse emergent group norms, with or without outright pressure to conform’.645 Informational influence has been shown to be stronger than normative influence,646 particularly when making factual decision were the answer can be derived or calculated from the information presented.647 However, normative influence is stronger when jurors need to make judgmental decisions that rely on a jurors’ subjective perceptions of moral blame and harm.648 When jurors have ‘fewer hard facts and informational arguments on which they can rely [they will] appeal to fairness and group values to persuade their fellow group members’.649 In criminal trials it is often the case that fact-finding relies on jurors’ subjective judgment. They are frequently asked to make decisions about 644 Martin F Kaplan and Charles E Miller, ‘Group Decision Making and Normative versus Informational Influence: Effects of Type of Issue and Assigned Decision Rule.’ (1987) 53(2) Journal of Personality and social psychology 306 (‘Group Decision Making and Normative versus Informational Influence’). 645 Brewer and Douglass (n 611) 341. 646 Martin F Kaplan, ‘Discussion Polarization Effects in a Modified Jury Decision Paradigm: Informational Influences’ (1977) 40(3) Sociometry 262 (‘Discussion Polarization Effects in a Modified Jury Decision Paradigm’). 647 Phoebe C Ellsworth, ‘Are Twelve Heads Better than One?’ (1989) 52(4) Law and Contemporary Problems 205; Phoebe C Ellsworth and Alan Reifman, ‘Juror Comprehension and Public Policy: Perceived Problems and Proposed Solutions’ (2000) 6(3) Psychology, Public Policy, and Law 788 (‘Juror Comprehension and Public Policy’). 648 Kaplan and Miller (n 647). 649 Ibid. Twitter trials and Facebook juries 204 facts that no-one can possibility know with precision, unless the defendant chooses to confess. There is no option to ‘calculate’ the truth, only to weigh the evidence and collectively decide whether it meets the threshold of proof of beyond reasonable doubt.650 Sometimes, individual juror bias can be overcome through discussion and deliberations, however, this can also exacerbate bias through group polarizatioin where individual responses shift towards greater extremes following deliberation.651 The exacerbation of bias is more common for biases that are shared by jurors within a jury652 — ‘when multiple jurors hold similar opinions, these opinions are strengthened by deliberation, as jurors realise they are in agreement with each other, yet when jurors hold opposing opnions, verdicts reflect efforts to reach a common middle ground’. 653 Another study showed asynchronous exposure to negatively-biased comments on social media significantly increased the number of guilty verdicts, however, these were overcome through jury discussion.654 The researchers suggest further research is needed where participants actively interact with social media rather than asynchronously view comments.655 On one hand it is good news if jury discussions can overcome bias, but on the other, the fact that exposure to negatively-biased comments significantly increased the number of guilty verdicts in the first place is concerning. From a legal perspective, 650 Woolmington v DPP (n 532); R v Mullen (n 532); Chan Kau v R (n 532); Brimblecombe v Duncan; Ex parte Duncan (n 532); Thomas v R (n 532); La Fontaine v R (n 532); Van Leeuwen v R (n 532); Chamberlain v R (No 2) (n 532); Hoch v R (n 532); R v Falconer (n 532). 651 Brewer and Douglass (n 611) 345; Kaplan and Miller (n 647). 652 Tamara M Haegerich, Jessica Salerno and Bette L Bottoms, ‘Are the Effects of Juvenile Offender Stereotypes Maximized or Minimized by Jury Deliberation?’ (2013) 19(1) Psychology, Public Policy, and Law 81. 653 Brewer and Douglass (n 611). 654 J Taylor, ‘Trial by Social Media: How Do You Find the Jury, Guilty or Not Guilty?’ (2019) 1(2) International Journal of Cyber Research and Education (In Press). 655 Ibid. 205 Sub judice rule not well adapted for social media relying on secretive jury discussions, over which we have no knowledge or control, to reduce bias that may cause significant increases in guilty verdicts, is an approach that lacks the exacting principles which are favoured by law. 7.5.2.5 Potential for misinformation to influence jurors Just as misinformation during elections may influence voters’ opinions,656 the presence of misinformation on social media during the Tostee trial — as discussed in Chapter 7 — raises similar concerns about the potential for jurors to be influenced. The dissemination of knowingly false statements of fact is an example of free speech that might have a harmful effect ‘because falsehoods may mislead the public in their decision-making’ and so may justify interference with the freedom of expression.657 The placement of misinformation on social media can be deliberate — such as in targeted manipulation — or inadvertent — such as everyday people discussing a criminal trial online. Both are problematic. The role of professional media organisations and journalists is to report the news, but social media additionally provides a platform for debate, critique and commentary.658 It is this debate, critique and commentary by everyday users that most commonly contained misinformation and prejudice in my datasets. People do not remember all the evidence they see and hear,659 and it has been shown that jurors most 656 Julian Matthews, ‘How Fake News Gets into Our Minds, and What You Can Do to Resist It’, The Conversation 17 April 2019 Twitter trials and Facebook juries 206 confident in their memory of trial evidence are not those who remember the most.660 Ideally, those jurors who most accurately recall the evidence would control jury deliberations and the ultimate verdict, but it has been shown that it is actually the highly confident jurors — whose memories are less accurate — who are in control.661 Significantly, jurors whose memories are more accurate may be persuaded to change their minds if they are not particularly confident. Those who are least confident in their memory are ‘most likely to cave in to the verdict choice of the more confident jurors’ and this could have the potential to ‘lead to a verdict based on the memories of the less accurate jurors’.662 This is seriously concerning if the memories of these highly confident but less accurate jurors are influenced by prejudice or misinformation, including that found on social media. It may be that only one — or a small number — of jurors acting impartiality is enough to influence the entire jury. As mentioned previously, it is possible judges themselves may be subconsciously affected by information outside of the evidence presented in the courtroom.663 While it is accepted they will not knowingly allow themselves to be influenced by adverse or prejudicial publicity — they are expected to put it out of their mind just as they do when hear prejudicial evidence but declare it inadmissible664 — they may still be affected subconsciously. If judges can be influenced in this way, it Brains, and Societies Reconstruct the Past (Harvard University Press, Reprint edition, 1997) (‘Memory Distortion’). 660 Mary E Pritchard and Janice M Keenan, ‘Memory Monitoring in Mock Jurors’ (1999) 5(2) Journal of Experimental Psychology: Applied 152. 661 Mary E Pritchard and Janice M Keenan, ‘Does Jury Deliberation Really Improve Jurors’ Memories?’ (2002) 16(5) Applied Cognitive Psychology 589, 600. 662 Ibid. 663 Attorney-General (UK) v British Broadcasting Corp (n 281) 335 per Viscount Dilhorne; R v McInroy (n 529) 615; Bell v Stewart (n 281) 433 per Isaacs and Rich JJ; Pennekamp v Florida (n 529) 358; Kerr v O’Sullivan (n 281) 209–210; Civil Aviation Authority v Australian Broadcasting Corporation (n 281) 550; Solicitor-General v Smith (n 529) [79]. 664 Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (n 281) 58. 207 Sub judice rule not well adapted for social media seems difficult to accept that untrained, inexperienced jurors would be resistant to the same influence. These concerns support the legal presumption that jurors are susceptible to prejudice.665 7.6 Summary By identifying the prevalence of prejudicial publicity in tweets, establishing how professional journalists and non-journalists talk about trials, and analysing how social media users in general show prejudice, I was able to consider how well adapted the sub judice rule is in the age of social media. Based on these results and findings I conclude that the Australian sub judice rule is not well adapted for regulating prejudicial publicity on social media during high-profile criminal trials. It appears to be largely effective in regulating the behaviour of professional journalists, but less so for non-journalists. The law is not well adapted for addressing the potentially prejudicial effects of multiple publications (tweets), posted by multiple social media users that collectively demonstrate a prosecution bias or cumulative negative sentiment. It is also not well adapted for responding to the large number of tweets that contain fair and accurate reports of judicial proceedings, but that may still have the potential to influence jurors. Although the law technically applies to social media users, in practice changes in news distribution, consumption and participation have made the law very difficult to enforce. While there has always been a risk that jurors may be exposed to prejudicial publicity, there are real reasons to think that social media has the potential to influence jurors in new ways. 665 Butler and Rodrick (n 79) [6.280]. Twitter trials and Facebook juries 208 8 Conclusions and Future Research 8.1 Overview In this final chapter, I outline the current challenges in moderating content on social media and what this means for prejudicial publicity. I identify potential future approaches and the difficulties each of these presents. There are many alternative legal and non-legal mechanisms that might be suitable for regulating or moderating prejudice online but, ultimately, more research is needed in this area. I discuss the experiences of the Australian courts in enforcing suppression orders in high-profile criminal trials and the difficulties in responding to prejudicial publicity in the age of social media. I discuss my next project in which I will use social network analysis to better understand the flow of prejudicial information in Twitter networks. I also provide an overview of other areas for future research. Finally, I highlight the benefits of using digital methods to better understand how the law operates in practice and to provide new insights for legal analysis and policy reform. 8.2 Regulating prejudice on social media is difficult As the growth and reach of social media continues, courts and governments will need to contend with ongoing issues around the administration of criminal trials.666 The problem of prolific, prejudicial online publicity will not disappear with time and ‘it is likely to become more, not less, crucial to find ways to moderate the effects of prejudicial publicity’.667 When it comes to the sub judice rule, some have gone as far 666 Braun (n 53). 667 McEwen, Eldridge and Caruso (n 146) 125. 209 Conclusions and Future Research as to say it is as ‘dead as a dodo’.668 While this reference refers to the South African trial of Oscar Pistorius, the law and the issues the court faced are very similar to those in my Australian case studies. As discussed throughout this thesis, the sub judice rule does technically apply to social media users,669 but in reality it is rarely applied or enforced. The traditional legal view has been that the potential for media publications to influence jurors is sufficiently significant such that regulation is required to ensure fair criminal trials.670 In theory, this approach suggests social media ought also to be regulated. This is no simple task, however, as platforms like Facebook are arguably becoming impossible to regulate, due to their size, complexity, and powerful algorithms.671 Despite these difficulties, as the consequences of harmful social media content continue to play out,672 there is increasing pressure on law-makers to create laws to regulate content 668 Dikgang Moseneke, ‘The Courtroom as TV Studio: The Case of the Oscar Pistorius Trial’ (2018) 14(4) International Journal of Law in Context 493, 500 (‘The Courtroom as TV Studio’). 669 R v Hinch (n 172). 670 The law also applies to civil trials but these fall outside the scope of this thesis. 671 What Facebook Knows About You (n 630) (8 February 2018). 672 I note the ‘Christchurch massacre’ in March 2019, which was livestreamed on Facebook, has been the catalyst for worldwide discussion about the regulation of social media content and governments are responding by drafting new laws. Twitter trials and Facebook juries 210 posted on social media,673 and on social media platforms to moderate harmful content.674 Regulation has traditionally been undertaken by ‘sovereign nation states with responsibility for governance exercised through state agencies’.675 This type of governance also extended to enforcing media laws against professional journalists and media organisations. However, social media is not an unregulated space — the rules for participation are continuously being created, enforced, and contested by both private and public actors.676 Communication on social media is generally curated by private companies,677 and much of that curation takes place through content moderation. Real people work as content moderators to make decisions about how platforms should deal with user-generated content, particularly posts containing potentially harmful material such as hate speech, nudity, racism, violence and 673 Philippa Smith, ‘The Challenge of Drawing a Line between Objectionable Material and Freedom of Expression Online’, The Conversation (29 March 2019) 211 Conclusions and Future Research trauma.678 This means the founders and leaders of social media platforms have extraordinary power in making decisions about what people see online and what the rules are, in a way that does not provide for accountable or participatory governance.679 They effectively control the flow of information around the world and ‘if you control the information, you control the wealth and you control the power’.680 Social media platforms are not currently addressing issues around prejudicial publicity, but this does not mean they cannot do so. In practice, the application of the sub judice rule does not appear to be working online, so more thought must be given to how we approach social media regulation. First, we must establish whether regulation is desirable and second, if it is, we must identify how to regulate differently so as to accommodate the social media environment. These challenges present an opportunity to shape laws in a way that acknowledges the role of social media in everyday lives, but also ensures we maintain a fair and functional criminal justice system. My research may provide useful insights to inform answers to these questions and for considering possible or appropriate responses. 8.3 Potential future approaches While this thesis does not set out to answer the question of ‘where to from here?’, it is important to make some observations about the benefits and pitfalls of potential 678 Sarah Roberts, ‘Commercial Content Moderation: Digital Laborers’ Dirty Work’ in Safiya Umoja Noble and Brendesha M Tynes (eds), The Intersectional Internet: Race, Sex, Class and Culture Online (Peter Lang Publishing, 2016) Twitter trials and Facebook juries 212 approaches that aim to preserve the principles of sub judice contempt. At one end of the spectrum we might do nothing, thereby maintaining the status quo where the burden of responding to prejudice falls to the parties. Instead of the state responding to prejudice through the application of the sub judice rule, this instead requires the parties to take an alternative approach such as applying for suppression orders, take- down orders, no jury orders (judge-alone trials), adjournments, permanent stays of proceedings, changes of venue, the discharge of juries, or the quashing of convictions on appeal. If it is the prosecution that makes these applications, the burden of responding to prejudice will remain with the state, whose role it is to maintain a fair system of justice. But if it is the defendant who must make application — which will always be the case in applications for no jury orders, adjournments (until the potential effects of prejudice may subside), permanent stays, changes of venue, and appeals against conviction — this burden shifts to a party that may be vulnerable and unable to afford the type of legal representation required for these complex applications. Consideration must also be given to whether professional media organisations ought to bear some responsibility for posting stories on platforms. They know that social media users are encouraged to comment and that some will likely post unlawful content. Unlike letters to the editor in the past, that were subject to editorial checks, professional media organisations know that social media users essentially have ‘free reign’ to comment as they choose, regardless of their knowledge, expertise, or the accuracy of what they say. At least on some level, it is clear media organisations have already grappled with this question. A number of media publishers have long ago removed the option for readers to comment on the news (or other publications) on their sites for fear of the legal liability that comes from readers posting inappropriate 213 Conclusions and Future Research content.681 Reader participation is expensive due to the moderation required to ensure compliance with the law, ethics or other codes of conduct.682 Instead, users are linked through to social media platforms, like Facebook, for open discussion. This may shift liability away from media organisations, but it may also increase Facebook’s responsibility. Theoretically, this same approach to preventing comments could be applied to Facebook pages belonging to media organisations, such that users could not comment on trial-related news articles. In practice though, this may not be particularly effective or desirable. This approach would only prevent users from commenting on public pages, but they could still make the same comments on private pages — given users are more likely to see comments by their friends, this approach may be quite limited in effect. It may also have the undesirable effect of only allowing legitimate journalists to talk about cases of high public interest, while non-journalists would be blocked. There are also further complications in terms of how media organisations function. As they rely on user interaction, including comments, to gauge the popularity of their publications and to inform decisions around what they will and will not publish, preventing these comments would make tailored marketing more difficult. Not only would it be difficult to know what to publish, but news articles may no longer appear in users’ news feeds because of the way personalised algorithms (such as those on Facebook) rely on user interaction (including comments) to determine how and when articles are posted to news feeds. By eliminating user interaction, it may be that the visibility of news articles is reduced significantly and this is problematic for media 681 Neil Thurman, ‘Forums for Citizen Journalists? Adoption of User Generated Content Initiatives by Online News Media’ (2008) 10(1) New media & society 139, 150 (‘Forums for Citizen Journalists?’). 682 Ibid 154. Twitter trials and Facebook juries 214 organisations whose business models rely on the visibility of publications and real- time feedback on how audiences are engaging. In some ways, shutting down discussion may have the practical effect of preserving the principles of sub judice contempt, however, it is also likely to infringe on the principles of open justice. While our legal system acknowledges that published conversations ought not to prejudice trials, it also acknowledges there is an important interest in having conversations about trials and maintaining an open, transparent and accountable criminal justice system. Shutting down all conversation online may be an undue interference with freedom of expression. If it is not reasonably appropriate and adapted to achieving a legitimate end, it may also be an egregious interference with human rights. Others may suggest there are different values at stake and that the defendant’s right to a fair trial ought to take precedence. These complexities demonstrate that this is a live and important issue and one that is not easy to resolve. Another option is to return to the days of sequestering jurors — this is possible but generally unpopular amongst jurors and may serve as a disincentive for people to undertake this civic duty. Alternatively, increased training or education of jurors may assist in alleviated the risks of juror prejudice. Some suggest we ought to abolish the jury system altogether, arguing that it is no longer workable. But it may be there are workable solutions that do not require such a fundamental reshaping of our criminal justice system. States may wish to push the responsibility upon platforms to regulate their users. It is possible that our courts could simply hold Twitter and Facebook liable, as there is increasing pressure on social media platforms to be held accountable for the harmful 215 Conclusions and Future Research effects of user-generated content.683 In the context of contempt, they may provide an attractive target for regulation which avoids some of the challenges of pursuing individual social media users and leverages the centralised power of major social media platforms in order to achieve public regulatory goals.684 This solution is not as simple as it seems because it could impose significant costs upon platforms to regulate the behaviour of their users and create incentives for platforms to be overly conservative in their removal of speech.685 The expectation that online intermediaries are to uphold the law by making complex decisions about the lawfulness of users’ content and conduct is an onerous responsibility for private companies.686 Getting intermediary liability legislation right can be a tricky task.687 It may not be reasonable, appropriate or even possible to expect social media companies to wade through hundreds or thousands of tweets or comments about criminal trials and make complex determinations about the potential of each to prejudice a trial. It also may be undesirable to establish a system of law enforcement that imposes this responsibility upon private companies in such a way that our legal safeguards are situated outside the control of an independent legal system and judiciary.688 683 Kylie Pappalardo and Nicolas Suzor, ‘The Liability of Australian Online Intermediaries’ (2018) 40(4) Sydney Law Review 469, 475; Isobel Asher Hamilton, ‘Facebook Is Dialling up Punishments for Users Who Abuse Live Video after the Christchurch Massacre’, Tech Insider - Business Insider Australia (online at 15 May 2019) Twitter trials and Facebook juries 216 It may be there is a balance to be found — one that holds social media platforms liable, but not in a way that creates vague obligations that would require platforms like Facebook and Twitter to make value judgements about what is and is not prohibited speech. It is possible an administrative organisation like the eSafety Commissioner689 has a role to play here. It may be that this type of body can make decisions or determinations about content that may be contemptuous and then liaise directly with the social media platforms where offending material appears. This option would retain some regulatory authority within our legal system and also reduce the somewhat onerous expectation that social media companies might make judgements about the lawfulness of vast quantities of online content. Further research is needed to determine how best to balance this range of competing values and principles, and the various options such as making platforms liable, creating an oversight agency, or maintaining the status quo. This may include a comparative analysis of the approaches being taken in other countries where similar issues are at play.690 It may also factor in complications with the way prejudicial publicity is dealt with in the US — the system is different to that in Australia, but it is 689 The Australian government has established the role of eSafety Commissioner which aims to ensure the online safety of all Australians. Its remit includes identifying and removing illegal online content. https://www.esafety.gov.au/ 690 ‘Online Platforms to Face New “Duty of Care” in the UK’ 217 Conclusions and Future Research the system with which those overseeing social media platforms are likely most familiar. 8.4 Social network analysis – network mapping Upon completion of my thesis, I will extend my use of digital methods to better understand whether tweets containing prejudicial information are highly visible or not. I intend to use social network analysis to investigate how prejudicial publicity about criminal trials flows in social networks. Although this future work will draw on the same data used in this thesis, for reasons of length and scope this approach has been left for later work. As the methods I use are emerging and somewhat experimental in a legal context, much work is needed to develop a legitimate and robust approach to network mapping in the context of prejudicial publicity on social media. I plan to develop these methods both to draw new insights from my dataset and to establish new approaches to legal analysis. My analysis presented in this thesis has shown that prejudicial publicity exists on social media, but the methods I have used treat the data as atomic tweets that are random and unconnected. In reality, though, social media conversations take place within a network of users. Tweets are sequential in terms of the date and time they are posted, but they are not actually experienced sequentially. Instead, a user will see tweets based on their personal social network. When a user @mentions or follows other accounts, this information is used by the platform algorithms to generate a unique social network for that user. This network will govern which tweets appear in the user’s feed and, therefore, which tweets they may see. By using network mapping I will be able to take a list of sequential tweets and recreate a social network that approximates the experience of users on Twitter. The visualisation of these user connections makes Twitter trials and Facebook juries 218 it possible to read the tweets in a way that is sensitive to the relationships between user accounts. Specifically, I will create network maps of the Twitter comments from each of the Baden-Clay and Tostee murder trials. This will make it possible to visually create an approximate simulation of the user networks in each Twitter dataset and to better understand the content and shape of the Twitter discourse. This means I can investigate the main actors who initiate and participate in prejudicial conversations and gain insight into how this type of information flows across the network. I can also determine whether users who post or respond to prejudicial publicity on social media cluster together in the network or whether they are more evenly distributed. Identifying the most influential accounts is significant from a regulatory perspective. If changes are to be made within the criminal justice system — or in the terms of service of social media platforms — it is these actors that may be the most suitable or attractive targets for regulation. In future work, I will conceptualise how existing media regulation might be extended or adapted. I will examine legal and non-legal responses, consider how each response would aim to ensure fairness in criminal trials, and investigate the particular challenges involved in structuring and implementing each approach. 8.5 Future research Beyond my next project, further research is needed to answer a range of important questions about media regulation and balancing competing legal principles in the age of social media. Of those, I consider the following to be the most pressing: Further work is needed to compare my findings to other criminal trials. The sensational and negative aspects of the trials I examined resonated particularly strongly through social media. Evidence that weakened the prosecution case or supported the defence were generally not reported or amplified in the same way. 219 Conclusions and Future Research Social media users appeared to identify with emotive evidence, but more research is needed to understand whether this finding applies to high-profile criminal trials in general. There is a need to better understand how existing laws are working and how people (jurors) actually access prejudicial publicity. There is a chance that the careful way that journalists construct tweets and headlines could prejudice jurors, regardless of whether they are technically prejudicial at law. Greater attention should be paid to the way that witnesses and counsel present evidence in court, given that the most sensational aspects are likely to be immediately and strongly amplified in social media.691 There is little work about how these types of information are perceived. Without it, it is impossible to say at this stage whether the larger proportion of technically prejudicial tweets by non-journalists is likely to be influencing jurors. These preliminary questions need to be addressed in order to understand what sort of activities the law should actually target. More work is needed to understand how likely jurors are to be exposed to prejudicial publicity on social media. Traditionally, it was common to read newspapers ‘cover to cover’, to watch the television news in the evening and, perhaps, to listen to radio news in the morning or during the day. People were more likely to see all the news on offer in the past, so if something was prejudicial 691 By way of example, during the High Court hearing for this case, counsel for the appellant initially described the accused’s behaviour as ‘cold-blooded’: see Transcript of Proceedings, R v Baden-Clay [2016] HCATrans 166 (26 July 2016) 820, 868. Only a short time into the hearing the Court requested counsel refrain from using the term and the appellant’s arguments were thereafter presented using more legalistic terminology. Despite its brief use, the term ‘cold-blooded’ (an arguably prejudicial term) was used repeatedly in news reports both online and in mainstream media. Perhaps not surprisingly, the court’s request was not reported. This example is not one where potential jurors may have been influenced (as it was an appeal hearing), but it demonstrates nonetheless how potentially prejudicial information may be strongly amplified on social media even if referred to only briefly in court. See, eg, ‘Baden-Clay “calculated, cold-blooded”, Crown Says in Murder Downgrade Appeal’, ABC News (online), 26 July 2016. Twitter trials and Facebook juries 220 then jurors likely were exposed. However, it is possible the vast quantities of news stories and social media content may dilute the effect of prejudice or make it less likely that jurors will be exposed. Perhaps in order to be influenced jurors need to actively engage in news stories by searching for trial related news, following news sites on social media, following friends who follow news and news sites, or by signing up to special interest groups like the Tostee Facebook group. There is a need to further conceptualise how existing media regulation might be extended or adapted. The sub judice rule appears to regulate the speech of users with journalistic training and the institutional oversight processes of mass media industries. Non-journalists are more likely to commit contempt. How exactly the law may respond, though, is not clear. There are difficult contested questions of what responsibilities platforms may have in setting norms for acceptable behaviour or enforcing the law. There are inherent tensions here between a platform user’s right to freedom of expression, the need to prevent juror exposure to prejudice online, and any potential requirements that platforms bear some responsibility for user-generated content.692 There is a need to examine the role played by different institutional actors (including social media platforms and news organisations) in influencing speech about criminal trials, in order to develop recommendations for future effective law reform. Further consideration must also be given to the roles and responsibilities of social media platforms, other intermediaries, and news organisations, in managing and regulating comments made on news articles and in relation to ongoing criminal trials. There is currently significant work being done in this area, 692 Rebecca MacKinnon et al, ‘Fostering Freedom Online: The Role of Internet Intermediaries’ (Report, United Nations Educational, Scientific and Cultural Organization, 2014) 132. 221 Conclusions and Future Research particularly following the Christchurch massacre,693 and further research is needed to determine how regulation and legal changes might be adapted in a way that accommodates prejudicial publicity about criminal trials. A robust study is needed to determine whether patterns in online conversations may correlate with verdicts in criminal trials. There is a possibility that prejudicial content or sentiment online may be an indicator or predictor of jury verdicts. If social media content provides an amplification of the thoughts and feelings of the greater community (from which our jurors are selected), could this actually give us insight into the outcome? And if this is a predictor of verdicts, work is needed to determine whether social media content is a predictor of verdicts because it is something like a representative sample of a cross-section of society. Or is it a predictor because jurors are being influenced by social media? If it is not a predictor of verdicts, information online may at least be capable of being used to provide insight into how a jury is thinking or feeling. This type of data analysis may provide useful insight into the issues juries are grappling with and, if so, this could inform the way counsel shapes its arguments as it attempts to persuade the jury of its case. This insight might be used to support counsel for either side to tailor their closing summaries to better address the opinions expressed in the media. There may be value in considering prejudicial publicity from a feminist or gendered perspective. This could consider whether prejudicial publicity varies depending on whether the accused is male or female. It could also consider 693 Hamilton (n 686). Twitter trials and Facebook juries 222 whether males or females are more likely to be responsible for posting or amplifying prejudicial information. Further work is needed to better understand the distribution of misinformation about criminal trials and whether there is media coverage to counter incorrect information. In my samples, misinformation tended not to be corrected, but more work is needed to understand whether this is a broader problem that plays out across criminal trials in general and to investigate how this may affect juror impartiality. It may be interesting to consider the increasing globalisation of trials and investigate whether many of the prejudicial statements about criminal trial are posted by social media users outside of the trial jurisdiction, such as those in the George Pell case. It may be useful to develop a simple methodology that replicates the type of data analysis I have used in a way that can be reproduced by practitioners. This may provide a new sources of evidence for use by practitioners in support of court applications such as suppression orders, take-down orders, no jury orders (judge- alone trials), adjournments, permanent stays of proceedings, change of venue, discharging juries, and quashing convictions on appeal. This research may contribute to a wider discussion of the adequacy of the jury system and also how a pro-prosecution bias might impact on public sentiment about sentencing. The broader questions about the efficacy of the criminal justice system are live and important topics and further work is needed in this area to establish a clearer sense of how the law ought to function in the digital age. 223 Conclusions and Future Research It would be useful to investigate, probably via interviews, how journalists select and prioritise news content, including the need to highlight the most interesting facts or angles about a news story, while ensuring they stay on the right side of the law. 8.6 Digital methods for legal analysis My thesis sets out an example of how mixed digital and social methods can be used to inform more traditional legal analysis. As research trends continue to acknowledge and seek out the benefits of interdisciplinary work, it will be important for scholars to have exposure to and some understanding of emerging internet based methods. Shifts to new media, including social media, represent a significant change from traditional media and ‘is expected to be the main platform for research in the future’.694 The methods I use lend real insight into existing legal debates, insight that is difficult to achieve through other means. The use of digital methods in the discipline of law is relatively uncommon,695 but it is important we develop new ways of answering legal questions, particularly in the internet era. By using a combination of digital and social science methods, my research allows us to understand the voices of everyday social media users and to use this insight to inform our responses to emerging legal problems. My analysis tells us more than how suitable the sub judice rule is for the age of social media — it also provides insight into how ordinary people are talking, engaging with, and experiencing the judicial system. These methods may be applied 694 ‘New Media Analysis’ (n 155) 1090. 695 McCay-Peet and Quan-Haase (n 3). Social media research tends to be more common in disciplines such as marketing, communications, politics, computer science, economics, health, business and education. Twitter trials and Facebook juries 224 to any number of legal and criminal justice contexts, including the many problems arising out of new communications technologies. 8.7 Conclusion While the Australian sub judice rule is not well adapted for regulating prejudicial publicity on social media during high-profile criminal trials, this is not cause for despair. Australia is not the only country grappling with the legal complexities of the sub judice rule in the age of social media. There is a significant opportunity to shape how we might regulate or manage harmful social media content in future. Ideally, we would reach a global consensus on how to manage these issues, but we are likely a long way from such a widespread level of harmonisation.696 Instead we must accept that the ‘road ahead will instead be travelled by many thousands of small steps’ where we minimise inconsistences and increase interoperability between domestic and international legal systems’.697 Ongoing research is crucial to informing future change and to ensuring fairness in criminal trials for future generations. 696 Svantesson (n 195) 113. 697 Ibid. 225 Bibliography Bibliography Articles / Books / Reports A, Mhiripiri, Nhamo and Chari Tendai, Media Law, Ethics, and Policy in the Digital Age (IGI Global, 2017) Alcorn, Niki et al, Media Consumer Survey 2015: Australia Media and Digital Preferences - 4th Edition (Deloitte, 2015) Alcorn, Niki and Lucy Buchanan, Media Consumer Survey 2017 - Australian Media and Digital Preferences - 6th Edition (Deloitte, 2017) 1 Allcott, Hunt and Matthew Gentzkow, ‘Social Media and Fake News in the 2016 Election’ (2017) 31(2) Journal of Economic Perspectives 211 Allport, Gordon W 1897-1967, The Nature of Prejudice (Doubleday, Abridged., 1958) Alvesson, M and D Karreman, ‘Varieties of Discourse: On the Study of Organizations through Discourse Analysis’ (2000) 53 Human Relations 158 Ammori, Marvin, ‘The New New York Times: Free Speech Lawyering in the Age of Google and Twitter’ (2013) 127 Harv. L. Rev. 2259 (‘The New New York Times’) Andreassen, Cecilie Schou, St\a ale Pallesen and Mark D Griffiths, ‘The Relationship between Addictive Use of Social Media, Narcissism, and Self-Esteem: Findings from a Large National Survey’ (2017) 64 Addictive behaviors 287 (‘The Relationship between Addictive Use of Social Media, Narcissism, and Self-Esteem’) Antunes, GE and PA Hurley, ‘The Representation of Criminal Events in Houston’s Two Daily Papers’ (1978) 55 Journalism Quarterly 756 Arterton, Janet Bond, ‘Unconscious Bias and the Impartial Jury Symposium: Unconscious Discrimination Twenty Years Later: Application and Evolution: Evolution of the ID, the Ego, and Equal Protection’ (2007) 40 Connecticut Law Review 1023 (‘Unconscious Bias and the Impartial Jury Symposium’) Australian Law Reform Commission, Prejudicial Publicity and the Courts - Reference on Contempt of Court, Tribunals and Commissions (1986) Australian Law Reform Commission, New South Wales Law Reform Commission and Victorian Law Reform Commission, Uniform Evidence Law Report (ALRC Report 102) (No ALRC Report 102, NSWLRC Report 112, VLRC Final Report, December 2005) 1 Twitter trials and Facebook juries 226 Bardoel, Jo and Mark Deuze, ‘“Network Journalism”: Converging Competencies of Old and New Media Professionals.’ (2001) 23(2) Australian journalism review 91 (‘“Network Journalism”’) Barnes, Lyn and Elesha Edmonds, ‘If It Bleeds, It Leads? Changing Death Coverage in the New Zealand Herald’ (2015) 21(2) Pacific Journalism Review: Te Koakoa 162 (‘If It Bleeds, It Leads?’) Bartels, Lorana, ‘Jurors and Social Media: Is There a Solution?’ The Conversation (online), 31 July 2013 Bartels, Lorana and Jessica Lee, ‘Jurors Using Social Media in Our Courts: Challenges and Responses’ (2013) 23 Journal of Judicial Administration 35 Baumeister, Roy F et al, ‘Bad Is Stronger Than Good’ (2001) 5(4) Review of General Psychology 323 Bell AC, Virginia, ‘Jury Directions: The Struggle for Simplicity and Clarity’ (Banco Court Lecture, Supreme Court of Queensland, 20 September 2018) Bell, Virginia, ‘How to Preserve the Integrity of Jury Trials in a Mass Media Age’ (2005) 7(3) Judicial Review 311 Beninger, K et al, Research Using Social Media: Users’ Views (NatCentre Social Research, 2014) Bennett, JM, ‘The Establishment of Jury Trial in New South Wales’ (1961) 13 Sydney Law Review 463 Bessi, Alessandro and Emilio Ferrara, Social Bots Distort the 2016 US Presidential Election Online Discussion (SSRN Scholarly Paper No ID 2982233, Social Science Research Network, 7 November 2016) Blackwell, David et al, ‘Extraversion, Neuroticism, Attachment Style and Fear of Missing out as Predictors of Social Media Use and Addiction’ (2017) 116 Personality and Individual Differences 69 Blucher, Alexandra, ‘Gable Tostee Juror’s Instagram Posts a Sign Jury Act Must Roll with the Times, Legal Experts Say’, ABC News (online at 21 October 2016) Bock, Annekatrin, Holger Isermann and Thomas Knieper, ‘Quantitative Content Analysis of the Visual’ in The SAGE Handbook of Visual Research Methods (SAGE Publications Ltd, 2011) 265 227 Bibliography Bosland, Jason, ‘Restraining “extraneous” Prejudicial Publicity: Victoria and New South Wales Compared’ (2018) 41(4) UNIVERSITY OF NEW SOUTH WALES LAW JOURNAL 1263 (‘Restraining “extraneous” Prejudicial Publicity’) Bosland, Jason John and Judith Townend, ‘Open Justice, Transparency and the Media: Representing the Public Interest in the Physical and Virtual Courtroom’ (2018) 23(4) Communications Law 183 (‘Open Justice, Transparency and the Media’) Bowcott, Owen and legal affairs correspondent, ‘Juror Jailed over Online Research’, The Guardian (online at 23 January 2012) Boyd, D and NB Ellison, ‘Social Network Sites: Definition, History, and Scholarship’ (2007) 13 Journal of Computer-Mediated Communication 210 Branco, Jorge, ‘Gable Tostee Trial: Tostee Declines to Testify in Murder Trial’, Brisbane Times (online at 14 October 2016) Brand, Jeff and Mark Pearson, ‘The Newsroom vs. the Lounge Room: Journalists’ and Audiences’ Views on News. [Paper Presented after Peer Review to the Communications Research Forum (2001: Canberra).]’ (2001) 23(2) Australian Journalism Review 63 (‘The Newsroom vs. the Lounge Room’) Braun, Kerstin, ‘Yesterday Is History, Tomorrow Is a Mystery-the Fate of the Australian Jury System in the Age of Social Media Dependency’ (2017) 40 UNSWLJ 1634 Brewer, Neil and Amy Bradfield Douglass, Psychological Science and the Law (Guilford Publications, 2019) Brodsky, Angela, Using Content Analysis of Internet News Article Comments to Understand Climate Change Denial (SAGE Publications Ltd, 2019) Brooks, Brian S et al, News Reporting and Writing (St Martin’s Press, 7th ed, 2001) Brundidge, Jennifer, ‘Encountering “Difference” in the Contemporary Public Sphere: The Contribution of the Internet to the Heterogeneity of Political Discussion Networks’ (2010) 60(4) Journal of Communication 680 (‘Encountering “Difference” in the Contemporary Public Sphere’) Bruns, Axel, ‘Echo Chamber? What Echo Chamber? Reviewing the Evidence’ (2017) Bruns, Axel, ‘Gatekeeping, Gatewatching, Real-Time Feedback: New Challenges for Journalism’ (2011) 7(11) Brazillian Journalism Research 117 Twitter trials and Facebook juries 228 Bruns, Axel, Gatewatching and News Curation: Journalism, Social Media, and the Public Sphere (Peter Lang, 2018) Bruns, Axel, ‘Gatewatching, Not Gatekeeping: Collaborative Online News’ (2003) 107(1) Media International Australia incorporating Culture and Policy 31 (‘Gatewatching, Not Gatekeeping’) Bruns, Axel, ‘Journalists and Twitter: How Australian News Organisations Adapt to a New Medium.’ [2012] Media International Australia incorporating Culture and Policy 97 Bruns, Axel, ‘Making Sense of Society through Social Media’ (2015) 1(1) Social Media + Society 1 Bruns, Axel et al, ‘The Australian Twittersphere in 2016: Mapping the Follower/Followee Network’ [2017] Social Media + Society Bruns, Axel and Jean Burgess, ‘Researching News Discussion on Twitter: New Methodologies’ (2012) 13(5–6) Journalism Studies 801 Bruns, Axel and Jean E Burgess, ‘The Use of Twitter Hashtags in the Formation of Ad Hoc Publics’ in Proceedings of the 6th European Consortium for Political Research (ECPR) General Conference 2011 (2011) Bruns, Axel and Tim Highfield, ‘Blogs, Twitter, and Breaking News: The Produsage of Citizen Journalism’ in In Rebecca Ann Lind (Ed), Produsing Theory in a Digital World: The Intersection of Audiences and Production (Peter Lang, 2012) 15 Bruns, Axel and Yuxian Eugene Liang, ‘Tools and Methods for Capturing Twitter Data during Natural Disasters’ (2012) 17(4) First Monday Bruns, Axel and Katrin Weller, ‘Twitter As a First Draft of the Present: And the Challenges of Preserving It for the Future’ in Proceedings of the 8th ACM Conference on Web Science (ACM, 2016) 183 Buckley, Isaac Frawley, ‘Pre-Trial Publicity, Social Media and the “Fair Trial”: Protecting Impartiality in the Queensland Criminal Justice System’ (2013) 33 The Queensland Lawyer 38 Burd, Roxanne and Jacqueline Horan, ‘Protecting the Right to a Fair Trial in the 21st Century - Has Trial by Jury Been Caught in the World Wide Web?’ (2012) 36 Criminal Law Journal 103 Burgess, Jean E and John A Banks, ‘Social Media’ in The Media and Communications in Australia [4th Edition.] (Allen & Unwin, 2014) 285 Burmester, H, ‘Scandalising the Judges’ [1985] Melbourne University Law Review 313 229 Bibliography Butler, Des and Sharon Rodrick, Australian Media Law (Thomson Reuters (Professional) Australia Limited, 4th ed, 2012) Byrne, David, ‘What Is Discourse Analysis?’ in Data Analysis and Interpretation (2017) Cabral, Jaclyn, ‘Is Generation Y Addicted to Social Media’ (2008) 18 Future of children 125 Carlson, Matt, ‘The Robotic Reporter: Automated Journalism and the Redefinition of Labor, Compositional Forms, and Journalistic Authority’ (2015) 3(3) Digital journalism 416 (‘The Robotic Reporter’) Chen, Daniel L and Markus Loecher, Mood and the Malleability of Moral Reasoning (SSRN Scholarly Paper No ID 2740485, Social Science Research Network, 4 January 2019) Chesterman, Michael et al, Managing Prejudicial Publicity: An Empirical Study of Criminal Jury Trials in New South Wales (Published by the Law and Justice Foundation of N.S.W. on behalf of the University of N.S.W. and the Justice Research Centre, 2001) (‘Managing Prejudicial Publicity’) Chesterman, Michael, ‘Media Prejudice During a Criminal Jury Trial: Stop the Trial, Fine the Media, or Why Not Both’ (1999) 1 UTS L. Rev. 71 (‘Media Prejudice During a Criminal Jury Trial’) Choi, Jihyang and Jae Kook Lee, ‘Investigating the Effects of News Sharing and Political Interest on Social Media Network Heterogeneity’ (2015) 44 Computers in Human Behavior 258 Choo, Andrew LT and Jill Hunter, ‘Gender Discrimination and Juries in the 20th Century: Judging Women Judging Men’ (2018) 22(3) The International Journal of Evidence & Proof 192 (‘Gender Discrimination and Juries in the 20th Century’) Citron, Danielle Keats and Helen Norton, ‘Intermediaries and Hate Speech: Fostering Digital Citizenship for Our Information Age’ (2011) 91 BUL Rev. 1435 (‘Intermediaries and Hate Speech’) Clerwall, Christer, ‘Enter the Robot Journalist: Users’ Perceptions of Automated Content’ (2014) 8(5) Journalism Practice 519 (‘Enter the Robot Journalist’) Conley, J and W O’Barr, Just Words: Law, Language, and Power (Univeresity of Chicago Press, 1998) ‘Content Analysis: Advantages and Disadvantages’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) ‘Content Analysis, Definition Of’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) Twitter trials and Facebook juries 230 Cooper, Crystale Purvis and Debra L Roter, ‘“If It Bleeds It Leads”? Attributes of TV Health News Stories That Drive Viewer Attention.’ (2000) 115(4) Public Health Reports 331 (‘“If It Bleeds It Leads”?’) Dabbs, Marjorie O, ‘Jury Traumatization in High Profile Criminal Trials: A Case for Crisis Debriefing’ (1992) 16 Law and Psychology Review 201 (‘Jury Traumatization in High Profile Criminal Trials’) Danziger, Shai, Jonathan Levav and Liora Avnaim-Pesso, ‘Extraneous Factors in Judicial Decisions’ (2011) 108(17) Proceedings of the National Academy of Sciences 6889 Dechêne, Alice et al, ‘The Truth About the Truth: A Meta-Analytic Review of the Truth Effect’ (2010) 14(2) Personality and Social Psychology Review 238 (‘The Truth About the Truth’) Denzin, Norman K and Yvonna S Lincoln, The Sage Handbook of Qualitative Research (Sage, 4th ed., 2011) Deuze, Mark, ‘What Is Multimedia Journalism?’ (2004) 5(2) Journalism studies 139 Devine, Dennis J et al, ‘Jury Decision Making: 45 Years of Empirical Research on Deliberating Groups’ (2001) 7(3) Psychology, Public Policy, and Law 622 (‘Jury Decision Making’) ‘Discourse Analysis’ in The SAGE Encyclopedia of Educational Research, Measurement, and Evaluation (SAGE Publications, Inc., 2018) Dominick, JR, ‘Crime and Law Enforcement in the Mass Media’ in C Winick (ed), Deviance and Mass Media (Sage, 1978) Eady, David and ATH Smith, Arlidge, Eady & Smith on Contempt (Sweet & Maxwell, Thomson Reuters, 4th ed, 2011) Eisenmann, Thomas R, ‘Managing Proprietary and Shared Platforms’ (2008) 50(4) California Management Review Ellsworth, Phoebe C, ‘Are Twelve Heads Better than One?’ (1989) 52(4) Law and Contemporary Problems 205 Ellsworth, Phoebe C and Alan Reifman, ‘Juror Comprehension and Public Policy: Perceived Problems and Proposed Solutions’ (2000) 6(3) Psychology, Public Policy, and Law 788 (‘Juror Comprehension and Public Policy’) Enli, Gunn, ‘Twitter as Arena for the Authentic Outsider: Exploring the Social Media Campaigns of Trump and Clinton in the 2016 US Presidential Election’ 231 Bibliography (2017) 32(1) European Journal of Communication 50 (‘Twitter as Arena for the Authentic Outsider’) Epstein, Robert and Ronald E Robertson, ‘The Search Engine Manipulation Effect (SEME) and Its Possible Impact on the Outcomes of Elections’ (2015) 112(33) Proceedings of the National Academy of Sciences E4512 Fairclough, N, Critical Discourse Analysis: The Critical Study of Language (Pearson Education, 1995) Fein, Steven, Allison L McCloskey and Thomas M Tomlinson, ‘Can the Jury Disregard That Information? The Use of Suspicion to Reduce the Prejudicial Effects of Pretrial Publicity and Inadmissible Testimony’ (1997) 23(11) Personality and Social Psychology Bulletin 1215 (‘Can the Jury Disregard That Information?’) Ferguson, Andrew Guthrie, ‘The Big Data Jury’ (2015) 91 Notre Dame L. Rev. 935 Foucault, Michel, Discipline & Punish: The Birth of the Prison (Random House, 1975) Foucault, Michel, The Archaeology of Knowledge (Pantheon Books, 1972) Friedman, Lawrence M, The Big Trial: Law as Public Spectacle (University Press of Kansas, 2015) Friedman, Lawrence M, The Big Trial: Law as Public Spectacle (University Press of Kansas, 2015) (‘The Big Trial’) Fyfe, Melissa, ‘What Happened to Warriena Wright after She Met Gable Tostee?’, The Sydney Morning Herald (17 March 2017) ‘Gable Tostee Murder Trial: Jury Can’t Reach Verdict’, The Queensland Times (18 October 2016) Gee, JP, An Introduction to Discourse Analysis: Theory and Method (Routledge, 1999) Gibbs, Samuel, ‘Facebook Apologises for Psychological Experiments on Users’, The Guardian (online at 2 July 2014) Goel, V, ‘As Data Overflows Online, Researchers Grapple with Ethics’, The New York Times (online at 12 August 2014) Golder, Scott and Michael Macy, ‘Social Science with Social Media’ (2012) 40(1) ASA Footnotes 7 Twitter trials and Facebook juries 232 González, Roberto J, ‘Hacking the Citizenry?: Personality Profiling, “Big Data” and the Election of Donald Trump’ (2017) 33(3) Anthropology Today 9 (‘Hacking the Citizenry?’) Gopaldas, A, ‘Marketplace Sentiments’ (2014) 41(4) Journal of Consumer Research 995 Gray, Anthony, ‘Contempt and the Australian Constitution – Part I’ (2017) 27(3) Journal of Judicial Administration 1 Greene, Elizabeth and Jodie O’Leary, ‘Ensuring a Fair Trial for an Accused in a Digital Era: Lessons for Australia’ in Patrick Keyzer, Jane Johnston and Mark Pearson (eds), The Courts and the Media - Challenges in the Era of Digital and Social Media (Halstead Press, 2007) 101 Gregory, Peter, Court Reporting in Australia (Cambridge University Press, 2005) Griffith, Tom, Katie Kossian and Tania Kowalczuk, ‘Twitter, Suppression and the Courts’ (2012) 15(3) Internet Law Bulletin 42 Grimmelmann, J, ‘The Law and Ethics of Experiments on Social Media Users’ (2015) 13(2) Colorado Technology Law Journal 221 Grow, Brian, ‘As Jurors Go Online, U.S. Trials Go off Track’, Reuters (online at 8 December 2010) Haegerich, Tamara M, Jessica Salerno and Bette L Bottoms, ‘Are the Effects of Juvenile Offender Stereotypes Maximized or Minimized by Jury Deliberation?’ (2013) 19(1) Psychology, Public Policy, and Law 81 Haim, Mario and Andreas Graefe, ‘Automated News: Better Than Expected?’ (2017) 5(8) Digital journalism 1044 (‘Automated News’) Hamilton, Isobel Asher, ‘Facebook Is Dialling up Punishments for Users Who Abuse Live Video after the Christchurch Massacre’, Tech Insider - Business Insider Australia (online at 15 May 2019) Hannaford-Agor, Paula, David B Rottman and Nicole L Waters, ‘Juror and Jury Use of New Media: A Baseline Exploration’ [2011] National Center for State Courts Center for Jury Studies, United States 1 Hanusch, Folker, ‘Transformative Times: Australian Journalists’ Perceptions of Changes in Their Work’ (2015) 155(1) Media International Australia 38 (‘Transformative Times’) Hanusch, Folker and Axel Bruns, ‘Journalistic Branding on Twitter. A Representative Study of Australian Journalists’ Profile Descriptions.’ [2016] Digital Journalism 233 Bibliography Hawi, Nazir S and Maya Samaha, ‘The Relations among Social Media Addiction, Self-Esteem, and Life Satisfaction in University Students’ (2017) 35(5) Social Science Computer Review 576 ‘Hawthorne Effect’ in The SAGE Encyclopedia of Educational Research, Measurement, and Evaluation (SAGE Publications, Inc., 2018) Hermida, Alfred, ‘Twittering the News: The Emergence of Ambient Journalism’ (2010) 4(3) Journalism practice 297 (‘Twittering the News’) Hews, Rachel and Nicolas Suzor, ‘“Scum of the Earth”: An Analysis of Prejudicial Twitter Conversations during the Baden-Clay Murder Trial’ (2017) 40(4) University of New South Wales Law Journal 1604 Hirst, Martin, Navigating Social Journalism: Friends, Enemies, Followers and Likes (Routledge, 2019) (‘Navigating Social Journalism’) Hogan, B and Anabel Quan-Haase, ‘Persistence and Change in Social Media’ (2010) 30(5) Bulletin of Science, Technology & Society 309 Hope, Robert Marsden, ‘A Constitutional Right to a Fair Trial? Implications for the Reform of the Australian Criminal Justice System’ (1996) 24 F L Rev 173 Horan, Jacqueline, Juries in the 21st Century (Federation Press, 2012) Hsieh, Hsiu-Fang and Sarah E Shannon, ‘Three Approaches to Qualitative Content Analysis’ (2005) 15(9) Qualitative Health Research 1277 Humphries, D, ‘Serious Crime News Coverage and Ideology: A Content Analysis of Crime Coverage in a Metropolitan Paper’ (1981) 27 Crime and Delinquency 191 Hunsinger, Jeremy, Lisbeth Klastrup and Matthew Allen, International Handbook of Internet Research (Springer, Springer Verlag, Springer Netherlands, 1st ed, 2010) Imrich, Dorothy J, Charles Mullin and Daniel Linz, ‘Measuring the Extent of Prejudicial Pretrial Publicity in Major American Newspapers: A Content Analysis’ (1995) 45(3) Journal of Communication 94 (‘Measuring the Extent of Prejudicial Pretrial Publicity in Major American Newspapers’) Jaconelli, Joseph, Open Justice a Critique of the Public Trial (Oxford University Press, 2002) Jameel, Tanzeela, Rukhsana Ali and Kamran Ahmed Malik, ‘Social Media as an Opinion Formulator: A Study on Implications and Recent Developments’ [2019] 2019 2nd International Conference on Computing, Mathematics and Engineering Technologies (iCoMET) Janis, Irving L, Victims of Groupthink: A Psychological Study of Foreign-Policy Decisions and Fiascoes (Houghton Mifflin, 1972) (‘Victims of Groupthink’) Twitter trials and Facebook juries 234 Johnston, Jane et al, Juries and Social Media: A Report Prepared for the Victorian Department of Justice (Australia, Standing Council on Law and Justice, 2013) 1 Jones, Rodney H and Christoph A Hafner, Understanding Digital Literacies (Routledge, 2012) Jung, Jaemin et al, ‘Intrusion of Software Robots into Journalism: The Public’s and Journalists’ Perceptions of News Written by Algorithms and Human Journalists’ (2017) 71 Computers in human behavior 291 (‘Intrusion of Software Robots into Journalism’) Jurgens, P, ‘Communities of Communication: Making Sense of the “Social” in Social Media’ (2012) 30(3–4) Journal of Technology in Human Services 186 Jury Tampering: Encyclopaedic Australian Legal Dictionary (Lexis Advance Pacific Research, 2019) Kaplan, Martin F, ‘Discussion Polarization Effects in a Modified Jury Decision Paradigm: Informational Influences’ (1977) 40(3) Sociometry 262 (‘Discussion Polarization Effects in a Modified Jury Decision Paradigm’) Kaplan, Martin F and Charles E Miller, ‘Group Decision Making and Normative versus Informational Influence: Effects of Type of Issue and Assigned Decision Rule.’ (1987) 53(2) Journal of Personality and social psychology 306 (‘Group Decision Making and Normative versus Informational Influence’) Kerbel, Matthew Robert, If It Bleeds, It Leads: An Anatomy of Television News (Westview Press, 1st Paperback ed., 2001) (‘If It Bleeds, It Leads’) Keyzer, Patrick et al, ‘The Courts and Social Media: What Do Judges and Court Workers Think?’ (2013) 25(6) Judicial Officers Bulletin 47 (‘The Courts and Social Media’) Kim, Yonghwan, Hsuan-Ting Chen and Homero Gil De Zúñiga, ‘Stumbling upon News on the Internet: Effects of Incidental News Exposure and Relative Entertainment Use on Political Engagement’ (2013) 29(6) Computers in human behavior 2607 (‘Stumbling upon News on the Internet’) Kim, Yonghwan, Shih-Hsien Hsu and Homero Gil de Zúñiga, ‘Influence of Social Media Use on Discussion Network Heterogeneity and Civic Engagement: The Moderating Role of Personality Traits’ (2013) 63(3) Journal of Communication 498 (‘Influence of Social Media Use on Discussion Network Heterogeneity and Civic Engagement’) Kitchin, R, ‘Big Data, New Epistemologies and Paradigm Shifts’ (2014) 1(1) Big Data & Society 1 Kondracki, Nancy L, Nancy S Wellman and Daniel R Amundson, ‘Content Analysis: Review of Methods and Their Applications in Nutrition Education’ (2002) 34(4) Journal of Nutrition Education and Behavior 224 (‘Content Analysis’) 235 Bibliography Kovach, Bill and Tom Rosenstiel, The Elements of Journalism: What Newspeople Should Know and the Public Should Expect (Three Rivers Press (CA), 2014) (‘The Elements of Journalism’) Kovach, Bill and Tom Rosenstiel, The Elements of Journalism: What Newspeople Should Know and the Public Should Expect (Crown Publishers, 2001) Kramer, Adam DI, Jamie E Guillory and Jeffrey T Hancock, ‘Experimental Evidence of Massive-Scale Emotional Contagion through Social Networks’ (2014) 111(24) Proceedings of the National Academy of Sciences 8788 Kramer, Geoffrey P, Norbert L Kerr and John S Carroll, ‘Pretrial Publicity, Judicial Remedies, and Jury Bias’ (1990) 14(5) Law and Human Behavior 409 Kveraga, Kestutis et al, ‘If It Bleeds, It Leads: Separating Threat from Mere Negativity’ (2014) 10(1) Social cognitive and affective neuroscience 28 (‘If It Bleeds, It Leads’) Kwak, H et al, ‘What Is Twitter, a Social Network or a News Media?’’ in Proceedings of the 19th International Conference on World Wide Web – WWW (ACM Press, 2010) 591 Kwon, K Hazel and Daegon Cho, ‘Swearing Effects on Citizen-to-Citizen Commenting Online: A Large-Scale Exploration of Political versus Nonpolitical Online News Sites’ (2017) 35(1) Social Science Computer Review 84 (‘Swearing Effects on Citizen-to-Citizen Commenting Online’) Lasorsa, Dominic L, Seth C Lewis and Avery E Holton, ‘Normalizing Twitter: Journalism Practice in an Emerging Communication Space’ (2012) 13(1) Journalism studies 19 (‘Normalizing Twitter’) Law Reform Commission of Western Australia, Discussion Paper on Contempt by Publication (No Project 93 (II), March 2002) Law Reform Commission of Western Australia, Report on Review of the Law of Contempt (No Project No 93, June 2003) LeGreco, Marianne, ‘Discourse Analysis’ in Qualitative Methodology: A Practical Guide (SAGE Publications, 2014) 67 Lessig, Lawrence, Code (Lawrence Lessig, 2006) Lewis, Seth C, ‘The Tension between Professional Control and Open Participation: Journalism and Its Boundaries’ (2012) 15(6) Information, Communication & Society 836 (‘The Tension between Professional Control and Open Participation’) Lewis, Seth C, Amy Kristin Sanders and Casey Carmody, ‘Libel by Algorithm? Automated Journalism and the Threat of Legal Liability’ (2019) 96(1) Journalism & Mass Communication Quarterly 60 (‘Libel by Algorithm?’) Twitter trials and Facebook juries 236 Lewis, Seth C and Oscar Westlund, ‘Big Data and Journalism: Epistemology, Expertise, Economics, and Ethics’ (2015) 3(3) Digital journalism 447 (‘Big Data and Journalism’) Linden, Carl-Gustav, ‘Decades of Automation in the Newsroom: Why Are There Still so Many Jobs in Journalism?’ (2017) 5(2) Digital journalism 123 (‘Decades of Automation in the Newsroom’) Lindner, Andrew M, Emma Connell and Erin Meyer, ‘Professional Journalists in “citizen” Journalism’ (2015) 18(5) Information, Communication & Society 553 Mayring, Philipp, ‘Qualitative Content Analysis’ (2000) 1(2) Forum: Qualitative Social Research Mazzella, Ronald and Alan Feingold, ‘The Effects of Physical Attractiveness, Race, Socioeconomic Status, and Gender of Defendants and Victims on Judgments of Mock Jurors: A Meta-Analysis1’ (1994) 24(15) Journal of Applied Social Psychology 1315 (‘The Effects of Physical Attractiveness, Race, Socioeconomic Status, and Gender of Defendants and Victims on Judgments of Mock Jurors’) McCay-Peet, Lori and Anabel Quan-Haase, ‘A Model of Social Media Engagement: User Profiles, Gratifications and Experiences’ in Heather L O’Brien and Luke Sloan (eds), Why Engagement Matters: Cross-Disciplinary Perspectives and Innovations on User Engagement with Digital Media (Springer-Verlag, 2016) 199 McCay-Peet, Lori and Anabel Quan-Haase, ‘What Is Social Media and What Questions Can Social Media Research Help Us Answer?’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd) 13 McEwen, Rebecca, John Eldridge and David Caruso, ‘Differential or Deferential to Media? The Effect of Prejudicial Publicity on Judge or Jury’ (2018) 22(2) The International Journal of Evidence & Proof 124 (‘Differential or Deferential to Media?’) McKimmie, Blake M and Barbara M Masser, ‘Chapter 6 The Effect of Gender in the Courtroom’ in Forensic Psychology: Concepts, Debates and Practice (Routledge, 2nd ed, 2010) McLellan, Justice P, ‘Looking inside the Jury Room. Speech Delivered at the Law Society of New South Wales Young Lawyers 2011 Annual Criminal Law Seminar’ (5 March 2011) Meehan, Kevin A, ‘The Continuing Conundrum of International Internet Jurisdiction Note’ (2008) 31 Boston College International and Comparative Law Review 345 Mencher, Melvin, News Reporting and Writing (McGraw-Hill College, 8th ed, 1999) Mitchell, Jason P, Chad S Dodson and Daniel L Schacter, ‘FMRI Evidence for the Role of Recollection in Suppressing Misattribution Errors: The Illusory Truth Effect’ (2005) 17(5) Journal of Cognitive Neuroscience 800 (‘FMRI Evidence for the Role of Recollection in Suppressing Misattribution Errors’) 237 Bibliography Moran, Gary and Brian L Cutler, ‘The Prejudicial Impact of Pretrial Publicity1’ (1991) 21(5) Journal of Applied Social Psychology 345 Morgan, David L, ‘Qualitative Content Analysis: A Guide to Paths Not Taken’ (1993) 3(1) Qualitative Health Research 112 (‘Qualitative Content Analysis’) Moseneke, Dikgang, ‘The Courtroom as TV Studio: The Case of the Oscar Pistorius Trial’ (2018) 14(4) International Journal of Law in Context 493 (‘The Courtroom as TV Studio’) Napoli, Philip M, ‘Social Media and the Public Interest: Governance of News Platforms in the Realm of Individual and Algorithmic Gatekeepers’ (2015) 39(9) Telecommunications Policy 751 (‘Social Media and the Public Interest’) Netanel, Neil Weinstock, ‘Copyright and a Democratic Civil Society’ (1996) 106(2) The Yale Law Journal 283 Nevezie, Leora and Jessica Eade, Media Consumer Survey 2018 - Australian Media and Digital Preferences - Seventh Edition (Deloitte, 2018) 1 ‘New Media Analysis’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) New South Wales Government Attorney General’s Department, Legislation, Policy and Criminal Law Review Division, Report on Access to Court Information (June 2008) New South Wales Law Reform Commission, Contempt by Publication - Discussion Paper 43 (2000) New South Wales Law Reform Commission, Contempt by Publication, Report 100 Summary (2003) New Zealand Law Commission, Juries in Criminal Trials Part Two - A Summary of the Research Findings, Preliminary Paper 37, vol 2 (1999) Ogloff, James RP and Neil Vidmar, ‘The Impact of Pretrial Publicity on Jurors: A Study to Compare the Relative Effects of Television and Print Media in a Child Sex Abuse Case’ (1994) 18(5) Law and Human Behavior 507 (‘The Impact of Pretrial Publicity on Jurors’) Ogloff, James RP and Neil Vidmar, ‘The Impact of Pretrial Publicity on Jurors: A Study to Compare the Relative Effects of Television and Print Media in a Child Sex Abuse Case’ (1994) 18(5) Law and Human Behavior 507 (‘The Impact of Pretrial Publicity on Jurors’) Osborne, M and M Dredze, ‘Facebook, Twitter and Google Plus for Breaking News: Is There a Winner?’ in Proceedings of the 8th International AAAI Conference on Web and Social Media (AAAI Press, 2014) 611 Twitter trials and Facebook juries 238 Otieno, DO and VB Matoke, ‘Social Media as Tool for Conducting Academic Research’ (2014) 4(1) International Journal of Advanced Research in Computer Science and Software Engineering 962 Palfrey, J, Interop the Promise and Perils of Highly Interconnected Systems (Basic Books, 2012) Panger, Galen, ‘Reassessing the Facebook Experiment: Critical Thinking about the Validity of Big Data Research’ (2016) 19(8) Information, Communication & Society 1108 (‘Reassessing the Facebook Experiment’) Pannam, Clifford L, ‘Trial by Jury and Section 80 of the Australian Constitution’ (1968) 6(1) The Sydney Law Review 1 Pappalardo, Kylie and Nicolas Suzor, ‘The Liability of Australian Online Intermediaries’ (2018) 40(4) Sydney Law Review 469 Pariser, Eli, The Filter Bubble: What The Internet Is Hiding From You (Penguin UK, 2011) (‘The Filter Bubble’) Park, Sora et al, Digital News Report: Australia 2018 (News and Media Research Centre, University of Canberra, 2018) 1 Pavlik, John, Gary Morgan and Bruce Henderson, ‘Information Technology: Implications for the Future of Journalism and Mass Communication Education’ [2001] Journalism and Mass Communication Education: 2001 and Beyond Payne, Geoff and Judy Payne, Key Concepts in Social Research (SAGE Publications, Ltd, 2004) Pearson, Mark, ‘Why the Public Isn’t Allowed to Know Specifics about the George Pell Case’, The Courier; Ballarat, Vic. (Ballarat, Vic., Australia, Ballarat, Vic., 27 March 2018) 0 Pearson, Mark and Caroline Graham, ‘Suppression Orders: Reskilling Journalists and the Judiciary’ (2010) 32(1) Australian Journalism Review 97 Pearson, Mark and Mark Polden, The Journalist’s Guide to Media Law: A Handbook for Communicators in a Digital World (Allen & Unwin, Fifth edition, 2015) (‘The Journalist’s Guide to Media Law’) Picard, RG et al, ‘Historical Trends and Patterns in Media Economics’ in Hand- Book of Media Management and Economics (Lawrence Erlbaum, 2006) Police, From the Deputy Commissioner of the Metropolitan, ‘Moves to Prevent Jury Tampering’, The Times (online at 15 July 2003) 239 Bibliography Pritchard, Mary E and Janice M Keenan, ‘Does Jury Deliberation Really Improve Jurors’ Memories?’ (2002) 16(5) Applied Cognitive Psychology 589 Pritchard, Mary E and Janice M Keenan, ‘Memory Monitoring in Mock Jurors’ (1999) 5(2) Journal of Experimental Psychology: Applied 152 Quan-Haase, Anabel and Lori McCay-Peet, ‘Building Interdisciplinary Social Media Research Teams: Motivations, Challenges, and Policy Frameworks’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd, 2016) 40 Quan-Haase, Anabel and AL Young, ‘Uses and Gratifications of Social Media: A Comparison of Facebook and Instant Messaging’ (2010) 30(5) Bulletin of Science, Technology and Society 350 Queensland Courts, ‘Summing up - 50.1 Flight and Other Post Offence Conduct as Demonstrating Consciousness of Guilt’ in Supreme and District Courts Criminal Directions Benchbook (2017) ‘Questions Gable Tostee Jury Wants Answered: What Was the Long Metal Object He Had?’ (online at 17 October 2016) Rainie, Lee et al, ‘Social Media and Political Engagement’ [2012] Pew Research, Washington DC ‘Random Sampling’ in Encyclopedia of Survey Research Methods (Sage Publications, Inc., 2008) ‘Random Sampling’ in The SAGE Encyclopedia of Qualitative Research Methods (SAGE Publications, Inc., 2008) Reavley, NJ and PD Pilkington, ‘Use of Twitter to Monitor Attitudes toward Depression and Schizophrenia: An Exploratory Study’ (2014) 2 PeerJ 647 Reidenberg, Joel R, ‘Technology and Internet Jurisdiction’ (2005) 153(6) University of Pennsylvania Law Review 1951 Roberts, Sarah, ‘Commercial Content Moderation: Digital Laborers’ Dirty Work’ in Safiya Umoja Noble and Brendesha M Tynes (eds), The Intersectional Internet: Race, Sex, Class and Culture Online (Peter Lang Publishing, 2016) Robson, Colin, Real World Research: A Resource for Social Scientists and Practitioner-Researchers (Blackwell Publishing, 2nd ed, 2002) Twitter trials and Facebook juries 240 Rodrick, Sharon, ‘Achieving the Aims of Open Justice-The Relationship between the Courts, the Media and the Public’ (2014) 19 Deakin L. Rev. 123 Rose, J Dafni, ‘Diverse Perspectives on the Groupthink Theory – A Literary Review’ (2011) Ryan, M and D Owen, ‘A Content Analysis of Metropolitan Newspaper Coverage of Social Issues’ (1976) 53 Journalism Quarterly 634 Salerno, Jessica M and Shari Seidman Diamond, ‘The Promise of a Cognitive Perspective on Jury Deliberation’ (2010) 17(2) Psychonomic Bulletin & Review 174 ‘Sampling, Random’ in The SAGE Encyclopedia of Communication Research Methods (SAGE Publications, Inc, 2017) Sarah T Roberts, ‘Moderating Social Media - The Content Moderators (The Agenda with Steve Paikin)’, ed The Agenda with Steve Paikin Scanlon, Christopher, Reporting and Writing: Basics of the 21st Century (Harcourt College Publishers, 2000) Schacter, Daniel L (ed), Memory Distortion: How Minds, Brains, and Societies Reconstruct the Past (Harvard University Press, Reprint edition, 1997) (‘Memory Distortion’) Schreier, Margrit, ‘Qualitative Content Analysis’ in The SAGE Handbook of Qualitative Data Analysis (SAGE Publications Ltd, 2014) 170 Schudson, Michael and Chris Anderson, ‘Objectivity, Professionalism, and Truth Seeking in Journalism’ in The Handbook of Journalism Studies (Routledge, 2009) 108 Seargeant, Philip and Caroline Tagg, ‘Social Media and the Future of Open Debate: A User-Oriented Approach to Facebook’s Filter Bubble Conundrum’ (2019) 27 Discourse, Context & Media 41 (‘Social Media and the Future of Open Debate’) Sensis, Sensis Social Media Report 2016: How Australian People and Businesses Are Using Social Media (Sensis, 2016) 1 Shadwell, Talia, ‘Gable Tostee: How an Instagramming Juror Nearly Jeopardised Trial over Warriena Wright’s Death’, Stuff (online at 21 October 2016) 241 Bibliography Shearing, Clifford and Jennifer Wood, ‘Nodal Governance, Democracy, and the New “Denizens”’ (2003) 30(3) Journal of law and society 400 Shoemaker, PJ and TP Vos, Gatekeeping Theory (Routledge, 2009) Simon, TF, F Fico and S Lacy, ‘Covering Conflict and Controversy: Measuring Balance, Fairness, Defamation’ (1989) 66 Journalism Quarterly 427 ‘Simple Random Sample’ in Encyclopedia of Survey Research Methods (Sage Publications, Inc., 2008) Singer, Jane B, ‘Who Are These Guys? The Online Challenge to the Notion of Journalistic Professionalism’ (2003) 4(2) Journalism 139 (‘Who Are These Guys?’) Spigelman, James, ‘The Internet and the Right to a Fair Trial’ (2005) 29(6) Criminal Law Journal 331 St Eve, Amy J and Michael A Zuckerman, ‘Ensuring an Impartial Jury in the Age of Social Media’ (2012) 202(2) Duke L. & Tech. Rev. 1 Steblay, Nancy et al, ‘The Impact on Juror Verdicts of Judicial Instruction to Disregard Inadmissible Evidence: A Meta-Analysis.’ (2006) 30(4) Law and Human Behaviour 469 Stoop, Ineke, ‘If It Bleeds, It Leads: The Impact of Media-Reported Events’ [2007] Measuring Attitudes Cross-Nationally. Lessons from the European Social Survey, London, SAGE 95 (‘If It Bleeds, It Leads’) Studebaker, Christina A et al, ‘Assessing Pretrial Publicity Effects: Integrating Content Analytic Results’ (2000) 24(3) Law and Human Behavior 317 (‘Assessing Pretrial Publicity Effects’) Sue, S, RE Smith and G Pedroza, ‘Authoritarianism, Pretrial Publicity, and Awareness of Bias in Simulated Jurors’ (1975) 37(3) Psychological Reports 1299 Suzor, Nicolas and Brian Fitzgerald, ‘The Legitimacy of Graduated Response Schemes in Copyright Law’ (2011) 34 UNSWLJ 1 Suzor, Nicolas Pierre, Lawless: The Secret Rules That Govern Our Digital Lives (Cambridge University Press, 2019) (‘Lawless’) Svantesson, Dan Jerker B, Private International Law and the Internet (Wolters Kluwer Law & Business, 3rd ed, 2016) Svantesson, Dan Jerker B, Solving the Internet Jurisdiction Puzzle (Oxford University Press, 2017) Tait, David, ‘Deliberating about Terrorism: Prejudice and Jury Verdicts in a Mock Terrorism Trial’ (2011) 44(3) Australian & New Zealand Journal of Criminology 387 (‘Deliberating about Terrorism’) Twitter trials and Facebook juries 242 Taylor, J, ‘Trial by Social Media: How Do You Find the Jury, Guilty or Not Guilty?’ (2019) 1(2) International Journal of Cyber Research and Education (In Press) Thayer, James Bradley, ‘The Presumption of Innocence in Criminal Cases’ (1897) 6(4) The Yale Law Journal 185 ‘Thematic Analysis’ in Encyclopedia of Case Study Research (SAGE Publications, Inc., 2010) Thomas, Cheryl, Are Juries Fair? (Ministry of Justice Research Series, 2010) Thurman, Neil, ‘Forums for Citizen Journalists? Adoption of User Generated Content Initiatives by Online News Media’ (2008) 10(1) New media & society 139 (‘Forums for Citizen Journalists?’) Thurman, Neil, Konstantin Dörr and Jessica Kunert, ‘When Reporters Get Hands-on with Robo-Writing: Professionals Consider Automated Journalism’s Capabilities and Consequences’ (2017) 5(10) Digital journalism 1240 (‘When Reporters Get Hands- on with Robo-Writing’) Tracy, K and JC Mirivel, ‘Discourse Analysis: The Practice and Practical Value of Taping, Transcribing, and Analysing Talk’ in Routledge Handbook of Applied Communication Research. New York: Routledge (2009) 153 Turel, Ofir and Hamed Qahri-Saremi, ‘Explaining Unplanned Online Media Behaviors: Dual System Theory Models of Impulsive Use and Swearing on Social Networking Sites’ (2018) 20(8) New Media & Society 3050 (‘Explaining Unplanned Online Media Behaviors’) United Kingdom Law Commission, Contempt of Court: Consultation Paper (2012) Veletsianos, G, ‘Higher Education Scholars’ Participation and Practices on Twitter’ (2012) 28(4) Journal of Computer Assisted Learning 336 Verma, Inder M, ‘Editorial Expression of Concern and Correction: Experimental Evidence of Massive-Scale Emotional Contagion through Social Networks’ (2014) 111(29) Proceedings of the National Academy of Sciences of the United States of America 10779 Vicario, Michela Del et al, ‘The Spreading of Misinformation Online’ (2016) 113(3) Proceedings of the National Academy of Sciences 554 Vitak, J, ‘Facebook as a Research Tool in the Social and Computer Sciences’ in The SAGE Handbook of Social Media Research Methods (SAGE Publications Ltd, 2016) 627 Wallace, Anne et al, ‘Courts and Social Media: Opportunities and Challenges?’ (2013) 40(7) Brief 36 243 Bibliography Warren, Justice M, ‘Open Justice in the Technological Age’ (2013) 40(1) Monash University Law Review 45 Watkins, Jerry et al, Digital News Report: Australia 2015 (News & Media Research Centre, University of Canberra and Reuters Institute for the Study of Journalism, University of Oxford, 2015) 1 Weller, Katrin et al, Twitter and Society (Peter Lang, 2014) Whealy, Anthony, ‘Contempt: Some Contemporary Thoughts’ (2008) 8(4) Judicial Review 441 Wilcox, W, ‘The Press, the Jury and the Behavioural Sciences’ [1968] Journalism Monographs 1 Wilske, Stephan and Teresa Schiller, ‘International Jurisdiction in Cyberspace: Which States May Regulate the Internet’ (1997) 50 Fed. Comm. LJ 117 (‘International Jurisdiction in Cyberspace’) Zappavigna, Michele, ‘Ambient Affiliation: A Linguistic Perspective on Twitter’ (2011) 13(5) New Media & Society 788 (‘Ambient Affiliation’) Cases Ahnee v Director of Public Prosecutions (1999) 2 AC 294 Attorney General (NSW) v Mundey (1972) 2 NSWLR 887 Attorney General (NSW) v Radio 2UE Sydney Pty Ltd (New South Wales Court of Appeal 40236/96, Priestley AJA, Meagher AJA and Powell AJA, 11 March 1998) Attorney General (NSW) v Willesee (1980) 2 NSWLR 143 Attorney General v Times Newspapers Ltd (1973) 3 ER 54 Attorney-General (NSW) v John Fairfax and Sons Ltd (1980) 1 NSWLR 362 Attorney-General (NSW) v Radio 2UE Sydney Pty Ltd and John Laws [1998] NSWSC 28 Attorney-General (NSW) v TCN Channel Nine Pty Ltd (1990) 20 NSWLR 368 Attorney-General (Qld) v Lovitt [2003] QSC 279 Attorney-General (Qld) v WIN Television (Qld) Pty Ltd [2003] QSC 157 Attorney-General (UK) v British Broadcasting Corp [1981] AC 303 Attorney-General v Guardian Newspapers (No 2) (1990) 1 AC 109 Attorney-General v Times Newspapers Ltd [1974] AC 273 Twitter trials and Facebook juries 244 Bell v Stewart (1920) 28 CLR 419 Bradley Keith Carr v R [2015] NSWCCA 186 Brimblecombe v Duncan; Ex parte Duncan (1958) 8 Qd R Brook v Evans (1860) 29 LJ Ch 616 Chamberlain v R (No 2) (1984) 153 CLR 521 Chan Kau v R [1955] AC 206 Civil Aviation Authority v Australian Broadcasting Corporation (1995) 39 NSWLR 540 Dagenais v Canadian Broadcasting Corp (1994) 120 DLR 12 Davis v Baillie [1946] VLR 486 Delbert-Evans v Davies (1945) 2 ER 167 Dietrich v R (1992) 177 CLR 292 Director of Public Prosecutions v Francis (2006) 95 SASR 302 Director of Public Prosecutions v Wran (1987) 86 FLR 92 Doe v Dowling [2017] NSWSC 202 (15 March 2017) Doe v Dowling [2017] NSWSC 1037 (10 August 2017) DPP v Pell (Sentence) [2019] VCC 260 Dupas v The Queen [2010] HCA 20 Ex parte Attorney-General; Re Truth & Sportsman Ltd [1961] SR NSW 484 Ex parte Auld; Re Consolidated Press Ltd (1936) 36 SR NSW 596 Ex parte Terrill; Re Consolidated Press Ltd (1937) 37 SR NSW 255 Fitzgibbon v Barker and others (1992) 111 FLR 191 Gallagher v Durack (1983) 152 CLR 238 Gammage v The Queen (1969) 122 CLR 444 Gilbert v The Queen (2000) 201 CLR 414 Hinch v Attorney General (Vic) (1987) 164 CLR 15 Hoch v R (1988) 165 CLR 292 John Fairfax Publications Pty Ltd v District Court (NSW) (2004) 61 NSWLR 344 245 Bibliography Kerr v O’Sullivan [1955] SASR 204 La Fontaine v R (1976) 136 CLR 62 Lodhi v R [2007] NSWCAA 360 Maxwell v Director of Public Prosecutions [1935] AC 309 McGuirk v University of NSW [2009] NSWSC 1058 Minister for Justice v West Australian Newspapers Ltd [1970] WAR 202 Murphy v R (1989) 167 CLR 94 News Digital Media Pty Ltd v Mokbel (2010) 30 VR 248 Packer v Peacock (1912) 13 CLR 577 Pennekamp v Florida (1946) 328 US 331 Prothonotary of the Supreme Court of New South Wales v Katelaris [2008] NSWSC 389 R v Australian Broadcasting Corporation [1983] Tas R 161 R v Baden-Clay (2014) QSC 154 R v Bayley [2013] VSC 313 R v Castro, Onslow & Whalley’s Case [1873] LR 9 QB 219 R v Daily Mirror (1927) 1 KB 845 R v David Syme & Co Ltd [1982] VR 173 R v Davis (1995) 57 FCR 512 R v Day & Thomson [1985] VR 261 R v Duffy; Ex parte Nash (1960) 2 QB 188 R v Dunbabin; Ex parte Williams (1935) 53 CLR 434 R v Editor of New Statesman; Ex parte Director of Public Prosecutions (1928) 44 TLR 301 R v Evening News; Ex parte Hobbs (1925) 2 KB 158 R v Falconer (1990) 171 CLR 30 R v Gable Tostee (QSC, 10 October 2016) R v Glennon (1992) 173 CLR 592 Twitter trials and Facebook juries 246 R v Gray (1865) 10 Cox CC 184 R v Grimes and Lee (1894) 15 NSWLR 209 R v Hinch [2013] VSC 520 R v Hoser & Kotabi Pty Ltd [2001] VSC 443 R v K (2003) 59 NSWLR 431 R v Karakaya (2005) 5 Cr App R R v Kray (1969) 53 Cr App R 412 R v Macfarlane; Ex parte O’Flanagan (1923) 32 CLR 518 R v McInroy (1915) 26 DLR 615 R v Meissner (Unreported, New South Wales Court of Criminal Appeal No 60170/92, 30 October 1992) R v Mullen (1938) 59 CLR 124 R v Nationwide News Pty Ltd (Unreported, Victoria Supreme Court No 6129/97, Gillard J, 22 December 1997) R v Payne (1896) 1 QB 577 R v Pearce (1992) 7 WAR 395 R v Richards; R v Bijkerk [1999] NSWCCA 114 R v Rogerson; R v McNamara (No 14) [2015] NSWSC 1157 R v Saxon, Hadfield and Western Mail Ltd [1984] WAR 283 R v Scott and Downland Publications Ltd [1972] VR 663 R v Shaun Stewart McNeil [2015] NSWSC 357 R v Shipley (1784) 99 ER 774 R v Sun Newspapers Pty Ltd (1992) 58 Crim R 281 R v Sussex Justices, ex parte McCarthy (1924) 1 KB 256 R v Truth Newspapers (unreported, Vic SC, Phillips J, 16 December 1993) R v West Australian Newspapers Ltd; Ex parte Director of Public Prosecutions (WA) (1996) 16 WAR 518 Re Coroner’s Court of Western Australia; Ex parte Porteous [2002] WASCA 144 (4 June 2002) 247 Bibliography Re Johnson (1887) 20 QB 68 Registrar of the Court of Appeal v John Fairfax Group Pty Ltd (New South Wales Court of Appeal No 40478/92, Mahoney JA, Priestley JA and Sheller JA, 21 April 1993) Registrar, Supreme Court of South Australia v Advertiser Newspaper Ltd (Unreported, Supreme Court of South Australia no 2418/95, Bollen J, 17 May 1996) Royall v R (1991) 172 CLR 378 Skaf v R [2008] NSWCCA 303 Solicitor-General v Radio Avon Ltd (1978) 1 NZLR 225 Solicitor-General v Smith (2004) NZLR 2 540 The Prothonotary v Collins (1985) 2 NSWLR 549 The Queen v Gerard Robert Baden-Clay [2016] HCA 35 The St James’s Evening Post Case [1742] 2 Atk Thomas v R (1960) 102 CLR 584 United States v Rutherford (2004) 371 F 3d 634 Van Leeuwen v R (1981) 36 ALR 591 Victoria v Australian Building Construction Employees’ & Builders’ Labourers’ Federation (1982) 152 CLR 25 Vine Products Ltd v Green [1966] Ch 484 Wade v Gilroy (1986) 10 Fam LR 793 Woolmington v DPP [1935] AC 462 Legislation Charter of Human Rights and Responsibilities Act 2006 (Vic) Commonwealth of Australia Constitution Act Convention for the Protection of Human Rights and Fundamental Freedoms (opened for signature 4 November 1950) entered into force 3 September 1953 in accordance with article 66 Crimes Act 1900 (ACT) Criminal Code Act 1899 (Qld) Twitter trials and Facebook juries 248 Criminal Code Act 1924 (Tas) Criminal Procedure Act 1986 (NSW) Criminal Procedure Act 2004 (WA) Human Rights Act 2004 International Covenant on Civil and Political Rights 1966 (adopted and opened for signature 16 December 1966,) 2200A (XXI) (entered into force 23 March 1976) Juries Act 1927 (SA) Juries Act 1962 (NT) Jury Act 1995 (Qld) Other Sources ‘A Man Accused of Jury Tampering Has Been Convicted — by a Different Jury’, miamiherald ‘Accused Balcony Killer’s New Name’, NewsComAu (4 February 2015) Adler, Simon, ‘Post No Evil | Radiolab’ ‘Are Tech Companies Becoming the Primary Legislators in International Cyberspace? - Lawfare’ Belot, political reporter Henry, ‘Prime Minister Announces New Penalties for “cowards” Who Tamper with Fruit’, ABC News (Text, 19 September 2018) Bogle, Ariel, ‘Meet the Hidden People Who Clean the Internet for You’, ABC News (Current, 29 September 2018) Bogle, Ariel, ‘Social Media Companies to Be Punished for “violent” Content after “Flawed” Laws Pass’, ABC News (Current, 4 April 2019) 249 Bibliography Byrne, Elizabeth, ‘Is Facebook Ruining the Chance of a Fair Trial?’, ABC News (Text, 20 May 2017) Constine|September 21, Josh and 2011, ‘Facebook Increases Character Limit on Posts to 5000, Rolls Out Floating Navigation Bar and More Amid Unrest’ Cooper, Adam, ‘DPP Moves to Jail Dozens of Editors, Journalists over Reports after Pell Verdict’, The Age (26 March 2019) ‘Covering Child Abuse in the Catholic Church under a Court Gag Order’, Columbia Journalism Review ‘Cracking the Code: What Facebook Really Knows about You’, Four Corners (Australian Broadcasting Corporation, 10 April 2017) Crothers, Joanna, ‘Yahoo7 Fined $300k over Article That Caused Murder Trial to Be Aborted’, ABC News (Text, 17 February 2017) Dias, Avani, ‘Mulhall Trial Aborted, Judge Blames “potentially Prejudicial” Reporting’, ABC News (Text, 20 April 2016) Dixon, Raphael, ‘Twitter Is Poised to Kill the Like Button, for Your Own Good’, ABC News (Current, 30 October 2018) Dodd, Andrew, ‘PODCAST: Pell Trial Reporters, a Judge and a Media Lawyer on Why the Suppression Order Debate Is Far from Over’, The Conversation (21 March 2019) Dwyer, Tim, ‘FactCheck: Is Australia’s Level of Media Ownership Concentration One of the Highest in the World?’, The Conversation Eustachewich, Frank Rosario, Lia, ‘Juror Dismissed for Sharing Trial Details on Facebook’, New York Post (11 April 2015) ‘Fears of Trial by Social Media in Meagher Case’ Twitter trials and Facebook juries 250 Frederick, Jeffrey T, ‘Facebook and Jury Tampering: A New Threat Posed by Social Networking Sites (SNS) to Jury Integrity’, Jury Research Services (8 August 2011) ‘Gable Tostee Could Face an Inquest into the Death of Warriena Wright | Newshub’ ‘Gable Tostee Could Face Fresh Charges for Death of Tinder Date Warriena Wright on Gold Coast’, NZ Herald (12 January 2019) ‘Gable Tostee Jury Panel Ask Judge Four Questions before Warriena Wright Murder Verdict | Daily Mail Online’ ‘Gable Tostee Jury Struggle for Unanimous Verdict’, Brisbane Times (18 October 2016) ‘Gable Tostee Not Guilty of Murder’ (20 October 2016) Gill, Jonathan M, ‘Sub Judice Contempt Regarding Publications: An Evaluation and Proposals for Reform with Particular Attention to Australia’s International Law Obligations / Jonathan M. Gill’ (1992) Grahl, T, ‘The 6 Types of Social Media’ (2013) Hall, Louise, ‘How Barrister Charles Waterstreet Caused Rogerson, McNamara Trial to Be Aborted’, The Sydney Morning Herald (15 June 2016) ‘How Does Twitter Decide What Is Trending? | ReThink Media’ ‘Instagram to Scrub Its Site of All Graphic Self-Harm Images Following Teenage Girl’s Suicide’, ABC News (Text, 8 February 2019) ‘Journalist Has Murder Jury Discharged’ (22 August 2016) 251 Bibliography ‘Jury Dynamics and Decision Making | Consensus Decision Making | Leadership & Mentoring’, Scribd Kim, Sharnie, ‘“Fearful Wright Had No Better Option” than Balcony Escape’, ABC News (Text, 14 October 2016) Kopp, Carlo and Kevin Korb, ‘We Made Deceptive Robots to See Why Fake News Spreads, and Found a Weakness’, The Conversation Lucas, Clay, ‘Dozens of Australian Journalists Threatened with Contempt of Court over Pell Stories’, The Sydney Morning Herald (26 February 2019) Ludlow, Peter, Lawrence Lessig and Tim Wu, Killswitch: The Battle to Control the Internet [Documentary] (Documentary, 2015) ‘Mark Zuckerberg Asks Governments to Control Internet Content’, ABC News (Text, 31 March 2019) Matthews, Julian, ‘How Fake News Gets into Our Minds, and What You Can Do to Resist It’, The Conversation Matthews, Julian, ‘Why Two People See the Same Thing but Have Different Memories’, The Conversation McNeal, Gregory S, ‘Controversy Over Facebook Emotional Manipulation Study Grows As Timeline Becomes More Clear’, Forbes Meyer, Robinson, ‘Did Facebook Defame Megyn Kelly?’, The Atlantic (30 August 2016) ‘Murder Trial Aborted Due to Jury Tampering’, Chronicle ‘Murder Trial Juror Discharged for Conducting Research’, ABC News (Text, 10 November 2015) Twitter trials and Facebook juries 252 Oremus, Will, ‘How Facebook’s Trending News Feature Went from Messy to Disastrous.’, Slate (30 August 2016) Oster, Jan, ‘Which Limits on Freedom of Expression Are Legitimate? Divergence of Free Speech Values in Europe and the United States’, The Net and the Nation State: Multidisciplinary Perspectives on Internet Governance (May 2017) (‘Which Limits on Freedom of Expression Are Legitimate?’) ‘Pell Gag Order Breaches as “Serious as It Gets”, Court Told’, The New Daily (15 April 2019) ‘Reading Is Believing: The Truth Effect and Source Credibility - ScienceDirect’ Rosen, Rebecca J, ‘The Thorny Combination of Old Laws and New Tech’, The Atlantic (10 November 2011) ‘Sentencing Remarks in DPP v George Pell - County Court of Victoria’ Smith, Philippa, ‘The Challenge of Drawing a Line between Objectionable Material and Freedom of Expression Online’, The Conversation ‘Social Media and Its Impact on Trials’, LawCPD.com.au (19 February 2014) Stephen, Andrew, Bernadette Kamleitner and Vincent Mitchell, ‘Your Online Privacy Depends as Much on Your Friends’ Data Habits as Your Own’, The Conversation Taylor, Tegan, ‘Help, My Nan Won’t Stop Sharing Dodgy Posts on Facebook’, ABC News (Current, 14 September 2018) ‘This Is What a Victorian Judge Told George Pell as He Sent Him to Jail for Abusing Two Choirboys’, ABC News (Text, 13 March 2019) ‘Tom Hanks Jury Appearance Results in Mistrial’, ABC News (Text, 13 September 2013) 253 Bibliography Tran, Danny, ‘Facebook, Twitter Urged to Crack down on Hate Speech in Australian Feeds’, ABC News (Text, 7 September 2017) ‘Warriena Wright’s Father Speaks for the First Time about Her Death.’, Mamamia (18 March 2017) What Facebook Knows About You Whitbourn, Michaela, ‘Blogger Shane Dowling Jailed for Contempt for Naming Tim Worner’s Alleged Lovers’, The Sydney Morning Herald (10 August 2017) Wilding, Derek and Sacha Molitorisz, ‘The Law Is Closing in on Facebook and the “Digital Gangsters”’, The Conversation