One of the fundamental beliefs underpinning the internet is that it is the same everywhere. Two things threaten this, however, regulation and AI, and the result could end up seeing the creation of the splinternet.
One of the fundamental tenets of the internet is that it is the same everywhere. Apart from the odd bit of state intervention and national censorship (and such filtering is actually lot more widespread than most people believe and extends way beyond China’s Great Firewall) the internet is pretty much the same everywhere. You connect to websites, you boot up apps, you send and receive email, and the underlying accessible infrastructure uses common protocols and open architectures to make sure it all works.
Inevitably, this has been under threat almost since internet use first took off in the mid 1990s. The walled gardens of the likes of Facebook, the political machinations of numerous nation states, pressure from anachronistic religious leaders…all these have put significant squeezes on the concept of a single, homogenous internet. However, few of these threats have managed to change things in any meaningful way for any meaningful length of time to date. But that could be about to change as two pressures from very different directions start to increase the chances of a genuine splinternet finally emerging.
The first threat is regulation. The European Union has been flexing its muscles in the online space with increasing abandon, building on the landmark provisions of the GDPR in an attempt to corral some of the more gregarious data scraping in particular of some of the tech giants.
Via Ireland’s regulator it fined Facebook owner Meta a massive $1.2 billion back in May for transferring European’s data to the US unlawfully using a legal instrument know as standard contractual clauses (SCCs). Meta currently has until October 12 to stop relying on SCCs for such transfers, and has threatened to shut down Facebook and Instagram in Europe unless some sort of deal is reached.
This is no doubt just noise from Meta, but elsewhere, Canada passed legislation forcing Meta and Google to pay news publishers for content, to which Meta responded by removing Canadian news from Facebook and Instagram; TikTok is under constantly moving threats of bans in the US and elsewhere (and will be illegal in Montana from next year); Apple has indicated that iMessage and FaceTime could be pulled from the UK rather than accede to laws currently being debated and break end-to-end encryption, as have WhatsApp and Signal; and there’s a new battle brewing over Meta’s targeted ads and the way it processes user data, with Norway threatening to fine the company $100k per day and Meta making noises about finally actually asking users explicitly for their consent to try to sell them targeted stuff – which is nice.
“After more than five years of litigation, Meta finally comes to the conclusion that it must ask people if it’s allowed to spy on them for ads,” commented activist Max Schrems, waspishly.
And yes, of course, you can argue that all these bans and exclusions relate to platforms and not the underlying structure of the internet. But it is via platforms that many people engage with the internet, and if they become victims of internet balkanisation, however well-intentioned some of the privacy-based regulation may be, that all helps bring the splinternet into being.
Drowning in AI crap
The other huge pressure at the moment is the sheer volume of content that is being generated by Chat GPT and its merry band of AI brothers. If the first phase of the internet can be characterised very roughly as people sharing cat videos, the second one can be summed up by having to wade through thousands and thousands of five star Amazon reviews to decide whether to buy a toothbrush or not.
The whole thing can be summed up by one arresting opening paragraph on a Verge article from June titled AI is killing the old web, and the new web struggles to be born which we’ve pasted below with all the links attached because, well, it’s a lot.
“There’s the junkification of Amazon and the enshittification of TikTok. Layoffs are gutting online media. A job posting looking for an “AI editor” expects “output of 200 to 250 articles per week.” ChatGPT is being used to generate whole spam sites. Etsy is flooded with “AI-generated junk.” Chatbots cite one another in a misinformation ouroboros. LinkedIn is using AI to stimulate tired users. Snapchat and Instagram hope bots will talk to you when your friends don’t. Redditors are staging blackouts. Stack Overflow mods are on strike. The Internet Archive is fighting off data scrapers, and “AI is tearing Wikipedia apart.” The old web is dying, and the new web struggles to be born.”
Quite simply, the internet is drowning. Despite being all-singing and all-dancing nowadays thanks to an explosion of bandwidth, even though somewhere north of 80% of internet traffic is now video the mechanism beneath it all is still driven by text. Text-based metadata underlies this smorgasbord of moving images, underpins the recommendations that direct us towards it, and charts the revenue that is earned from it. And with generative AI able to produce huge amounts of this text at will, the system is collapsing.
It’s roughly analogous to alchemists actually succeeding in turning base metals into gold. That happens and the value of gold then collapses as it’s no longer got any rarity value. Text is the same. Once you can press a button and generate it at will, then you need to find a new way of sifting through what is being endlessly uploaded to the systems that drive your recommendation engines, for instance.
Of course, one answer inevitably is to set a thief to catch a thief and use AI tools to detect AI content. But these are proving to be less than reliable, currently only running at about a success rate of 80% at best, and even OpenAI quietly shut down its dedicated tool that was designed to detect AI content due to ‘a low rate of accuracy’ in July.
The answer, according to some, is to abandon the old web-model of doing things; of using advertising to underwrite such essentials as search and instead moving to one where the users pay for the service via a subscription or a series of micropayments. Instead of engaging with such AI compromised sites as Google or Amazon, instead we will engage with sites that have pre-sifted their content and come up with — often — a human curated subset of that content.
The old rallying cry from the early days of the internet was ‘information wants to be free’, which almost looks quaint now. Information? Do you mean verified facts? Genuine news? Or do you mean 1000 subtly different 5-star reviews of a dodgy toaster, 10 reasons why the Earth is flat and/or run by our lizard overlords, and a guide to self-publishing your own book using generative AI pulled together by ChatGPT and sold as a ‘book’ itself on Amazon?
It quickly becomes about the quality of information and not the quantity of it and so this is the second dimension the internet might well fracture along, giving us a splinternet of free access versus paid access; of a wild west of unregulated content and poor functionality undermined by AI spam on the one hand, and one which, at least in part, tries to be the internet of the free flow of accurate information that the early pioneers always hoped it would be on the other.
The arrival of the splinternet
Of course, we might have both fractures happen at once. The internet doesn’t have to simply bifurcate: split in two with Europe on one side and the US on the other; or of free access and AI sludge-free zones on the other. Realistically it could fracture into many many different pieces, and once either of these two processes drives a wedge into it, that provides a route for some of the other pressures that have historically looked to balkanise it to exert increasing pressure in turn.
Technology, commercialism, politics, nationalism, religion…all these have pressured the internet to spit asunder over the three decades or so of its existence. None of them have succeeded on a widespread scale as yet. But with two powerful new forces coming to bear on the internet as we know it, you can’t help feeling that the chances of it surviving in the form it is for another three years, never mind decades, are starting to look smaller with each passing month.