The most memorable parts of last year’s Conservative election manifesto were the ones Theresa May would rather forget. History will immortalise her social care reform as a voter-repellent “dementia tax”. The offer to give MPs a free vote on the hunting ban is a case study in how to alienate the biggest number of people with the fewest words. With such a circus of electoral self-sabotage, it is hardly surprising that one of the more interesting manifesto pledges went largely unnoticed. The Tories promised to clean up the internet: “Some people say that it is not for government to regulate when it comes to technology … We disagree.”
Not so long ago, regulating the internet was thought to be senseless. It was undesirable because the ethos of the digital age – a Californian love child of hippy liberalism and ultra-capitalism – venerated freedom. And it was impossible because the fluid character of the web would flow around clumsy analogue statecraft.
That was before the Chinese Communist party demonstrated that a government with sufficient disregard for democracy could curtail a citizen’s online exploration. It was also before the wide-open savannahs of the western web were conquered by corporate giants: Google, Amazon, Facebook, Apple. This tetrarchy – the Gafa – has a collective influence over the economic and social lives of hundreds of millions of people unparalleled in human history. It is not just the companies’ financial scale that boggles the mind (they are competing to be the first company valued at a trillion dollars) but the intimacy of their reach. They know your place, your purchases, your politics and your most personal proclivities.
When May speaks of regulating the internet she has in mind social ills conducted over the network: terrorist recruitment, paedophilia, hate speech. But any government’s ability to fix those things requires negotiation with the companies that own the platforms on which the wicked trade is done. The balance of power then depends on the credibility of a politician’s threat to sanction a business, which is a function of jurisdiction and market size. That is why Mark Zuckerberg, Facebook’s chief executive, has testified before the US Congress and taken questions from the European parliament, but ignores invitations by the Commons committee investigating data misuse in political campaigns. He cares what US and EU regulators might do. The UK doesn’t even know what its regulatory framework will be after Brexit. (And London will be copying and pasting templates set in Brussels for years to come.)
Even without the problem of border-less commerce, governments struggle to police their national patches of internet because some digital offences evade precise definition. Moderators struggle to distinguish, for example, between a racist post and a sarcastic post lampooning racists. It gets harder still when the mechanism for identifying malicious intent or distinguishing between truth and falsehood is automated. The sheer volume of content to be vetted requires algorithmic surveillance, which grants extraordinary powers to the programmers tasked with encoding moral boundaries.
Google is this week due to publish the ethical framework it uses in its application of artificial intelligence. This is meant to reassure us that the company is not striding into a dystopian future without any consideration of potentially sinister applications of its algorithms. (Given that it has previously helped the Pentagon develop AI systems for drone warfare, that reassurance is going to be a tough sell.)
At Facebook a “content standards forum” convenes every fortnight to discuss the boundaries of acceptable discourse in the Kingdom of Zuck. It is an improvisatory process. The group had to invent a “newsworthiness” exemption to cover one of the most famous images in the history of war photography. Nick Ut’s picture of Vietnamese children fleeing a napalm strike had been blocked because it featured a naked young girl.
Zuckerberg has speculated that his site might one day need something like a “supreme court” to adjudicate filth and fair comment. That sounds bombastic, but few of the other options are very appealing. It is easy to outlaw bad behaviour. Most of what people can easily agree is offensive material online is in any case already illegal offline. The difficulty comes when thousands of marginal cases have to be judged every minute. Or when every candidate in an election decides hostile press coverage is “fake news”. Is the job of arbitration then best done by a robot, a company employee or a civil servant? None seem obviously qualified.
A regulatory assault on the Gafa seems simultaneously urgent, inevitable and impossible. But these are still early days in the transformation of politics by technology. Or, rather, we must now be at the end of the phase where technology companies can pretend to be just businesses. It is inconceivable that entities in command of so much wealth and power – in control of an infrastructure that is instrumental to the routine interactions and livelihoods of hundreds of millions of people – are anything other than political.
If they had geographical boundaries they would be nations. Instead they are virtual empires with capabilities that intrude deep into national jurisdictions, rivalling the power of local elites. That has been a guaranteed formula for confrontation throughout history. There is no reason to expect it will be different this time, for all that we cannot yet imagine how such a battle is even fought.
• Rafael Behr is a Guardian staff columnist
