Information crisis hidden by opaque Big Tech

Part 6 in the Age of Anger series. By Daniel Angus

Australia’s information environment is increasingly fragmented, more personalised and less trusted. Yet control over the systems that distribute information is increasingly concentrated in the hands of a small number of global technology companies. 

This combination poses a growing challenge to democratic societies.

The 2026 Reuters Institute Digital News Report shows social media and video networks have overtaken news websites and apps as a source of online news globally. Some 54 per cent of respondents use social media as a source for news, while global trust in news has fallen to 37 per cent.

These changes are often described as media fragmentation. Audiences that once congregated around newspapers and broadcasters are now dispersed across YouTube, TikTok, Facebook, Instagram, podcasts, influencers and countless other platforms.

But fragmentation can obscure an equally important trend. While media production and consumption have fragmented, the control over the infrastructure that distributes, recommends and monetises media has become highly concentrated.

A small number of technology companies now operate the portals through which much of the world's news, political communication and public debate circulates. 

Their recommender systems shape what appears next. Their advertising systems determine which messages reach which audiences. Their moderation systems influence what remains visible, what is demoted and what disappears. This creates a paradox: an increasingly fragmented public sphere is being organised through an increasingly concentrated set of privately controlled technical systems.

This distinction matters for a deeper understanding of misinformation, polarisation and declining trust.

There is no shortage of actors willing to exploit division. Conspiracy entrepreneurs, extremist movements, political campaigners, partisan media and foreign influence operations all use digital platforms to reach audiences. Algorithmic systems can act as force multipliers, allowing provocative, emotionally charged or misleading material to travel rapidly through networks.

But measuring the prevalence and effects of these activities has become increasingly difficult.

There is a striking paradox at the centre of contemporary platform governance. The problems are highly visible at the individual level. Australians can open a social media app and immediately encounter questionable advertising, misinformation, political grievance, conspiracy theories and inflammatory content.

Yet establishing what is happening systematically, across millions of Australians, over time, is extraordinarily difficult.

Platforms possess detailed information about what we as users see, what they engage with, what they are recommended next and why. Researchers, regulators, journalists and civil society generally do not. This is why transparency is no longer enough. What Australia and the world needs is observability.

Transparency usually means platforms disclosing information about themselves. They might publish moderation statistics, maintain an advertising archive, explain elements of their algorithms or provide researchers with controlled access to selected datasets.

These measures have value, but they leave platforms deciding what can be seen.

The Meta company's Content Library illustrates the problem. It replaced CrowdTangle, a tool widely used by researchers and journalists to study public activity across Facebook and Instagram. The new controlled research environment is cumbersome, restricts what can be retrieved and analysed, and leaves independent researchers reliant on access conditions and data structures determined by Meta itself. These controlled “clean rooms” are an inadequate solution for genuine platform observation.

Observability requires the creation of independent capacity to observe how platforms operate in practice, including the experiences they produce for users.

This distinction is particularly important because platform harms cannot be reduced to individual pieces of content. Recommender systems curate sequences of content and continuously adapt those sequences in response to user behaviour. Advertising systems segment audiences and distribute different messages to different people. Moderation systems can amplify, suppress or remove material in ways that vary across communities and contexts.

Existing transparency measures based on aggregated platform data cannot adequately illuminate these systems. Regulators, researchers and civil society need legal and technical infrastructure that enables them to independently monitor platform behaviour.

“Regulators, researchers and civil society need legal and technical infrastructure that enables them to independently monitor platform behaviour”

This is where the implementation of the Australian government’s proposed Digital Duty of Care becomes critical. It is part of its response to the review of the Online Safety Act. 

Rather than regulating only specific categories of harmful content after they appear, a systemic duty could require platforms to identify and mitigate risks created through the design and operation of their services.

That includes risks arising from recommendation, advertising and moderation systems. It could also deal with broader societal harms including discrimination, polarisation and the suppression of legitimate public-interest information. However, a Digital Duty of Care without independent observability risks becoming yet another underwhelming compliance exercise.

Platforms cannot be the only institutions responsible for identifying systemic risks, deciding how those risks should be measured, determining whether their interventions work, and supplying the evidence used to assess their own compliance. Australia needs an ecosystem of observability capable of producing evidence independently of the major platforms.

That ecosystem should combine regulatory powers to obtain platform data alongside independent research infrastructure, data donation, browser and mobile measurement, secure research environments and protections for legitimate public-interest research. It should enable researchers to examine not only individual posts, but patterns of exposure: who sees what, in what sequence, under what conditions and with what consequences?

The Australian Internet Observatory provides one model for this infrastructure. It brings together data, tools, expertise and governance to enable independent study of Australians’ experiences across digital platforms. This capability could be extended to support independent scrutiny of platform compliance and provide greater democratic oversight of systems that now mediate public life.

Australia cannot effectively respond to misinformation, extremism, foreign interference, discriminatory advertising or algorithmically amplified division if it cannot establish how these phenomena move through digital systems. Nor can governments determine whether regulation is working by relying primarily on evidence supplied by the companies being regulated.

The central challenge is no longer simply making platforms more transparent. It is ensuring Australian society has the independent capacity to observe them - and if necessary make informed decisions on what (if anything) needs to be done.

Platforms should welcome that capacity too. Independent observability provides stronger evidence base not only for identifying harms, but for distinguishing systemic problems from anecdote, testing whether interventions work, and avoiding regulation driven by assumption or political pressure rather than evidence. 

If platforms want proportionate, effective and evidence-led governance, independent scrutiny and observability must be part of the solution.

Daniel Angus is professor and director of the Queensland University of Technology’s Digital Media Research Centre

This article was first published in The Klaxon and 360info.org