Some Contributions to Public Debate in Australia and Elsewhere

Continuing with the round-up of recent activity I began in my last few posts (covering new articles, new conference presentations, new research videos, and my lecture series on Gatewatching and News Curation), here’s an update on a few other writings and presentations for a more general audience.

Facebook News Ban Redux

Perhaps most timely of these, paradoxically, is the oldest: in October 2022 I was interviewed by Canadian legal scholar Michael Geist on his long-running Law Bytes podcast, about Canada’s proposed C-18 bill that is modelled closely on Australia’s controversial News Media Bargaining Code. In Australia, the NMBC resulted in Facebook blocking Australian users from accessing or posting any news on its platform for over a week, before a compromise that strongly favoured Facebook was found – and as I write this, the same is happening in Canada. I spoke to Michael about Australia’s long and tortured path towards and through the news ban, and shared our findings on what happened on Facebook during the Australian news ban (in short: live continued as usual, proving Facebook’s point that news matters a lot less to the platform than policy-makers might have thought):

Social Media Campaigning on the Voice Referendum

In other current events, my QUT Digital Media Research Centre colleagues and I have also begun to track the social media campaigns surrounding Australia’s referendum on an Indigenous Voice to Parliament, which will be held in the final quarter of 2023. We’re still at an early stage of the campaign, of course, but already the PoliDashboard project (which our colleagues at the Social Media Lab at Toronto Metropolitan University have kindly extended to cover Australia) is picking up on the intensified advertising on Facebook and Instagram, and so is our Australian Ad Observatory operated by the Centre for Automated Decision-Making and Society (ADM+S). Together with my colleague Dan Angus, I published an overview of our observations to date in The Guardian recently; subsequently, I also appeared on the Centre for Responsible Technology’s Burning Platforms podcast:

And In addition to our own article in The Guardian, I was also cited in two other articles on the Voice campaign. I have a feeling there will be a fair few more before the campaign is over.

The Long Campaign for a Robodebt Royal Commission

A little earlier, my QUT DMRC colleague Ehsan Dehghan and I had also published an article in The Conversation, this time about the long-running grassroots campaign for justice for the victims of the previous federal government’s illegal ‘Robodebt’ debt recovery scheme. Robodebt’s malfunctioning algorithms produced thousands of debt letters to innocent welfare recipients, and literally destroyed lives, yet in spite of its blatantly obvious failures it took many years of dogged campaigning for a Royal Commission of Inquiry to be established to investigate the scheme.

Building especially on Ehsan’s long-standing research into the Robodebt campaign, especially on Twitter (which was the medium of choice for grassroots political campaigning when the Robodebt scandal first emerged in 2016, even if it no longer has that standing under Elon Musk’s catastrophic leadership now), our article traces the emergence of the #NotMyDebt and #Robodebt hashtags and their growing resonance with Australian political actors and the general public, and highlights especially the critically important work of a number of community leaders – in particular the excellent Asher Wolf.

Some Reflections on the Latest Big Facebook Study (in German)

And finally, along with a number of other German colleagues I also recently provided some commentary on one of the articles in the new study of the impact of different newsfeed algorithms on Facebook, which was published as a (somewhat misleadingly titled) special issue in Science in late July 2023. Conducted in partnership with Meta, the study shows notably different patterns of engagement with political content for users exposed to a reverse-chronological as opposed to the standard algorithmically constructed newsfeed on the platform, but remains somewhat muted in its conclusions from these observations. This is perhaps because the researchers took the – unfortunate, in my view – decision to conduct the study on US users during the months before and after the highly controversial 2020 US presidential election, and therefore at a time when users would already have been overloaded with political content both on Facebook and just about anywhere else in their media and interpersonal environments. It would perhaps have been preferable to conduct the study at a quieter time in US politics (or failing that, at least a less extremely heated time).

Also frustrating is Science’s insistence on presenting this and the other articles in the collection as a special issue on ‘social media and elections’, when it’s really an isolated study on Facebook and Instagram in the 2020 US election, specifically. Neither do these platforms now represent the bulk of social media activity in the US or the world, nor is the US political system representative for elections around the world – nor indeed is the 2020 US election even representative for the longer-term experience of elections in the United States. But that’s American exceptionalism (and indeed Science’s exceptionalism within the academic publishing world) for you, I guess.

That’s not to take away from what the article we commented on, and the other articles in the same special issue that are addressing other aspects of the same larger study, actually report – there are various useful insights here, even if their applicability and generalisability beyond this particular case study is severely limited. And that’s one of the key points made in our comments for the Science Media Center, along with various other observations and critiques. Since those comments were intended for a German audience, they’re in German – but this Google Translate-produced English version should capture at least the key points with a sufficient degree of accuracy.