Yesterday morning, the House of Commons Digital, Culture, Media and Sport Select Committee published its long-awaited final report into disinformation and ‘fake news’. The report – which follows a long and at times dramatic investigation – is full of interesting and insightful details about political microtargeting (the targeting of political messaging to relatively small groups of people) and the spread of disinformation.
But the report’s myopic focus on one company – Facebook – means that it misses the bigger picture – including the internet’s dominant variety of capitalism.
It is of course welcome that attention is being paid to these problems, and there is much in the Committee’s report that’s good. The report is undoubtedly right to find that Britain’s electoral laws are woefully inadequate for the age of the algorithm and are badly in need of reform. Its recommendation that inferences drawn from analysis of other data about people should be more clearly considered to be personal data likewise seems eminently sensible.
Is it ok to manipulate people to extract their money, just not for politics?
But there are also clear shortcomings. Focusing on disinformation itself as a target for regulation brings an obvious problem. By calling for interventions based on ‘harmful’ content, the report asks the Government to step into the dangerous territory of regulating lawful political conversations between people. Are private companies to be mandated to police these communications on the Government’s behalf? There are numerous good reasons why this is deeply undesirable (not to mention incompatible with human rights laws).
The biggest oversight, however, is in diagnosing disinformation as essentially a problem with Facebook, rather than a systemic issue emerging in part from the pollution of online spaces by the business model that Facebook shares with others: the surveillance and modification of human behaviour for profit.
‘Surveillance capitalism’, as it’s known, involves gathering as much data as possible about as many people as possible doing as many things as possible from as many sources as possible. These huge datasets are then algorithmically analysed so as to spot patterns and correlations from which future behaviour can be predicted. A personalised, highly dynamic, and responsive form of behavioural nudging then seeks to influence that future behaviour to drive engagement and profit for platforms and advertisers. These targeted behaviour modification tools rely on triggering cognitive biases and known short-cuts in human decision-making. Platforms and advertisers extensively experiment to find the most effective way to influence behaviour.
Without looking at surveillance capitalism, it’s impossible to understand microtargeting in its wider context. It’s impossible to understand the desires for profit and market position driving these practices. And it’s impossible to understand that the same behaviour modification tools are sold to advertisers, political parties, and anyone else who’s willing to pay. Without considering these practices within surveillance capitalism more generally, the report seems to implicitly accept that manipulating people through psychological vulnerabilities is fine if you’re doing it to extract their money, but not if you’re doing it for politics.
Notably, both Google and YouTube, its subsidiary, were largely omitted from the report. They get the odd mention, but it’s clear that the Committee was too fixated on Facebook to pay them sufficient attention. Google invented surveillance capitalism and remains arguably its foremost practitioner, with significant influence over the world’s access to information. And YouTube (also running on a surveillance business model, naturally) has serious problems of its own in terms of promoting violent extremism, disinformation, and conspiracy theories. This led the academic Zeynep Tufekci, writing in the New York Times last year, to describe YouTube and its video recommendation system as “[maybe] one of the most powerful radicalizing instruments of the 21st century”.
It's not “fake news” that’s the problem, it’s the algorithms that disseminate it
This brings us to the second aspect missed by the Committee: the increasingly prevalent algorithmic construction of reality. Take disinformation. As noted above, the report focused on false content itself. This seems to have missed one of the key routes by which an individual piece of content can become a systemic problem worthy of attention. In the grand scheme of things, a YouTube video about a wild conspiracy theory doesn’t really matter if it’s only seen by 10 people. It matters if people watching relatively innocuous content are driven towards it by YouTube’s recommendation system. It matters if it’s algorithmically promoted by YouTube and then seen by 10 million people.
Platforms might argue that they can’t be held responsible for the content they host or for the actions of their users (outside of things which are clearly illegal). But recommending content is not simply hosting it, and it is not a neutral act. Platforms selectively target content (including advertising) through their recommender systems so as to show us what they think will keep us engaged with their services, bring them revenue, and help them build market share. Make no mistake – through these platforms we do not get a true picture of what’s going on in the world. The spaces we inhabit online are viewed through the lens of corporate desires. What we encounter is algorithmically mediated to suit the platforms’ interests. While microtargeting is increasingly recognised as manipulation, this is a softer, perhaps more insidious form of corporate algorithmic influence.
Unsurprisingly, various actors have learned how to game these systems to boost the audience for their content, including conspiracy theorists and extremists. And bots and other fake accounts are often being used to take advantage of the algorithmic construction of online space to manipulate content rankings. This allows them to game trending topics so as to shape discourse more generally, and drive fringe ideas into the mainstream (a common misconception of bots holds that they are usually intended to change the opinion of real users with whom they come into contact).
The influence wielded by surveillance platforms through personalisation gives them significant means to shape the online public sphere. They are, of course, motivated by profit and duty to shareholders rather than by public good and duty to wider society. You might think that this is fine – they are, after all, private corporations. But while television, mass media, and the advertising industry have long shaped our world, never before have private companies had such influence over the construction of the everyday reality we inhabit. Never before have they exercised such influence over the private activity of individuals talking to other individuals about their lives. They do so without any democratic legitimacy, and with little transparency over their processes or accountability for their actions.
To properly address the problems of manipulation, disinformation, and violent extremism fermenting on online platforms, future regulation must properly acknowledge the role of surveillance capitalism – not just through targeting tools but in the algorithmic construction of online spaces. Future regulation should recognise that content isn’t necessarily the problem in and of itself. It must consider the active role of platforms in promoting content, and establish minimum standards for doing so (in the form of paid-for advertising or otherwise). This approach benefits from largely sidestepping much of the content regulation debate. Regulating the use of technical systems by corporations rather than intervening in communications between individuals means that people should still be free to post, view, or share anything that is not illegal. Freedom of expression demands nothing less.
Surveillance companies’ exorbitant profits and their influence on the construction of our reality is in large part driven by their use of recommender systems. That must come with responsibility in some form for what they’re algorithmically disseminating. They will argue that being more careful with recommender systems could result in lower revenues. In 2018 Google brought in $136 billion; Facebook took $56 billion. They can afford to take the hit. Perhaps that should be understood as the cost of doing business in future. This industry wouldn’t be the first to have its practices and its profits reined in by regulation for the good of society.
Because of its restricted focus, the usefulness of many of the solutions proposed in the DCMS Committee’s report is somewhat limited. That’s disappointing. But all is far from lost, and there are other directions for progress. To get there, we need to think bigger than Facebook. It’s time to acknowledge the role of surveillance capitalism in these systemic issues. It’s time to recognise that the problem isn’t just content – it’s dissemination and amplification by algorithm to maximise profit at all costs.