This is part five of an ongoing, five-part series. Part one, the introduction, is here. Part two, about breaking up ad-tech companies, is here. Part three, about banning surveillance ads, is here. Part four, about opening up app stores, is here. Download this whole series as a single PDF.
Once, news organizations enthusiastically piled into social media. New platforms like Facebook and Twitter were powerful “traffic funnels,” where algorithmic recommendation systems put excerpts from news stories in front of a vast audience of new readers, who followed the links at the end of the excerpt to discover sources that became part of their regular news-diets.
As platforms like Facebook grew essential to the new companies’ business, they changed the deal. First, it was a general “deprioritization” of news posts.
This didn’t just mean that articles from news accounts were less likely to be recommended to users who didn’t subscribe to the news publisher’s account. It also meant that people who explicitly followed publishers — that is, who had explicitly directed Facebook to show them the things those publishers posted — would be less likely to see the publishers’ posts.
The public explanation for this was that Facebook was shifting to prioritizing posts from “friends,” but publishers couldn’t help but notice that their account managers at Facebook stepped up the pressure to “boost” their posts.
The implied message was, “While Facebook says it’s prioritizing posts from users’ friends in their feeds, we’re actually prioritizing posts from users’ friends, as well as posts from publishers who will pay us to put their posts back into the feeds of the readers who asked to see them.”
Facebook was betting that publishers would pay for “reach.” It’s not just Facebook: the new Twitter policy is explicit: if you want to reliably reach the people who asked to see your Tweets, you need to sign up for Twitter Blue. If you’re a media company, that will cost you $1,000/month.
Facebook parent company Meta has also doubled down on holding media companies to ransom, charging for “verification,” and with it, a soft promise that the material you post will reach the people who asked to see it.
The tech companies’ logic is simple: once end-users and publishers are locked into their platforms, any value they create for one another can — and should — be captured by the platform instead. The ideal user feed should consist of just enough of the things the user has asked to see, or is predicted to enjoy, as will keep that user on the platform.
Naturally, the news needs to be where users are. So long as the users stay put on a platform, the press will feel pressured to join them. That gives platforms leverage to charge news sources for “boosting” or “verification” or other forms of Danegeld to increase the likelihood that the news they post will reach the people who asked to see it.
This conduct — where a platform prioritizes delivering the content that makes it the most money, irrespective of its users’ wishes — violates the venerable end-to-end principle: that a platform’s first duty should be to deliver data sent by willing senders to willing recipients, as quickly and reliably as possible.
When we apply end-to-end to the internet itself, we call it “Net Neutrality”: the idea that your ISP should deliver the data to you asked for (a video from your preferred streaming service, say), not the data its investors wish you had asked for (a video from a rival streaming service owned by your ISP).
Like the internet itself, early social media was born neutral: the first social networks were simple conduits connecting users, so every user’s update was shown to that user’s followers. Gradually, these feeds were augmented with recommendation systems that helped users prioritize otherwise excessive posts from the people they followed.
But as users were switched from a default of show-me-the-things-my-friends-posted to the default of show-me-the-things-you-think-I-should-see, a powerful temptation seized platform managers: to violate the implicit contract to deliver the things users requested, and then turn to the senders (publishers, performers, creators) and demand ransom money to reach their own followers.
Paying to “boost” content is now ubiquitous. It’s another way — along with gouging on ad commissions and mobile app payments — that tech platforms misappropriate value generated by the news.
If social media platforms honored the end-to-end principle, this ransom would end. Social media users would see the things they asked to see and publishers would reach the audiences that asked to hear from them.
It's possible that we could see a law or regulation mandating end-to-end delivery for platforms, but we don't have to wait for a law. The platforms are incorrigble in their continuous violations of existing fraud, privacy, and competition laws. After years of successful delaying tactics, time has run out on the largest tech companies, which are now facing colossal fines for their bad behavior.
A fine on its own won't be enough to deter companies from breaking the rules. A fine is a price, and so long as the price is lower than the illicit profits it buys, companies will pay it.
With fines climbing, companies are apt to begin suing for peace: offering settlements to end long, bruising fights. We think these settlements will offer a fine opportunity for regulators to craft service-specific end-to-end rules.
Unlike other remedies — say, rules requiring platforms to curb harassment — an end-to-end rule is easy to administer. A no-harassment rule requires: agreement on the definition of harassment, agreement on whether a specific incident constitutes harassment, and a fact-intensive investigation of whether the platform had taken reasonable steps to prevent harassment.
By contrast, if we suspect a platform isn't reliably delivering the messages it promised to deliver, we can just send test messages and see if they arrive.
What's more, complying with an end-to-end rule doesn't require expensive engineering that would make the rule a barrier to entry for smaller, less abusive platforms. The default state of social media is delivery from senders to receivers: it's holding followers for ransom that requires extra engineering.
When tech was a competitive sector, most of the innovation was devoted to delivering more value to both end-users and publishers. Once tech became a concentrated industry dominated by bloated companies that had gorged themselves by buying up their smaller competitors, "innovation" shifted to finding new ways to misappropriate value from both users and business customers.
That's not so different from the story of telecoms itself. When telecoms curdled into a rent-seeking snarl, we built a better network, one that ran on end-to-end.
That same end-to-end principle will go a long way to unrigging social media platforms, too.