Copy, Acquire, Kill— How Meta could pull off the most extraordinary pivot in tech history
Meta's sudden interest in ActivityPub and Mastodon doesn't make much sense. There are a few impossibly consistent talking points floating through the Fediverse. Each tries to explain away the oddity of Meta's presence.
I don't buy any of it. This is a plan years in the making. And we're watching Meta's biggest hurdle play out in real-time.
What if Meta's hidden objective behind the Threads-to-Mastodon initiative is a play on app.net? And, what if threads.net is a measured step towards what could be the greatest pivot in all of tech?
Put on your tinfoil hats, kids. Papa's got a story.
Over the past four years, Mark Zuckerberg has made moves that imply he's positioning Meta as a sort of Platform as a Service provider for social media companies.
This theory is in good company. Here's a 2019 tweet from "a member of Facebook's inner circle” reported by the Mostly Cloudy Newsletter:
I'm going to make a prediction: in five years, >25% of Facebook Inc's revenue will comes from selling cloud services to other UGC companies to meet global compliance needs.
If successful, Meta would usher in a new social media era as the first lords of the "social web.” The aftermath could mean that Meta gains an unprecedented concentration of power over what we see, who we interact with, and where developers can build digital communities.
Indie developers could gain a much-needed revenue stream through an ad-share program, which would be great if it weren't Meta writing the checks. We saw what happened to small businesses and aggregator websites. We saw what almost happened to corporate media after Facebook launched Instant Articles.
Everyday-users would pay the highest price. We'd be left with little more than the illusion of choice, resembling something approaching Proctor & Gamble's generic competitive strategy. We may gain the technical ability to move platforms with our followers and content in hand, but only so much as non-Meta affiliated social media exist. And perhaps not even then.
Meta must create a new market out of thin air to accomplish this monumental pivot. Such a feat could prove risky without something to compel indie developers and server admins to use Meta's shiny new infrastructure. That something could be regulation. Targeted lobbying has made it difficult for start-ups to compete in social media. Four new bills are floating around Congress. Each bill resembles Mark Zuckerberg's four-point plan.
Zuck's plan, published in the Washington Post, came one year after Cambridge Analytica and three months before the prediction tweet.
It sounds absurd that the biggest social media company would pivot from its golden goose. But when you peek into Meta's egg basket, you see it's empty. My guess is that goose was cooked a long time ago.
I'm not suggesting it happens all at once. It'll take years to fully pivot. And it's not like Meta will give away the billions of users it currently serves. But a decision has been made, and action taken.
Here's what I think happened: Zuck saw the writing on the wall, and he finally had enough— the congressional hearings, EU regulators, the impossible content moderation trap— all of it. It was time to get out of the social media game for good. Meta has the infrastructure skills and political influence to pull it off. He only needed the right angle to dip his toe in the water.
Mark Zuckerberg's Four-Point Plan #
In 2019, Mark Zuckerberg published an op-ed in the Washington Post calling for more government regulation of social media. He identified four areas needing policy— harmful content, election integrity, privacy, and data portability.
From Internet needs new rules. Let's start in these four areas, by Mark Zuckerberg, March 30th, 2019:
Finally, regulation should guarantee the principle of data portability. If you share data with one service, you should be able to move it to another. This gives people choice and enables developers to innovate and compete.
Zuckerberg then announces his support for the Data Transfer Project.
The Data Transfer Project, or Data Transfer Initiative (no one can keep the name straight), was founded in 2018. It provides a common framework for internet-based service providers to transfer user data between each other.
The project (or initiative) has some big names attached to it—Google, Microsoft, and then Meta in 2019. Twitter may have dropped out recently, as it's no longer included in press coverage.
All involved with DTP/DTI frame their efforts as a step towards users' controlling their own data. Not so much the type of data big tech may collect on us or what it's allowed to do with it. Just the swapping part. There are a lot of folks jumping from Instagram to Microsoft these days?
On Meta's engineering website, the company writes in length about the mission's importance and protocol's ease of use. But the tone for the project shifts within the context of the Fediverse.
A recent conference hosted by DTI (again with the name) announced guests from "the world's biggest ActivityPub projects" and Big Tech™. The event notes reveal that participants were concerned with trust and safety challenges that arise with data transfers on ActivityPub:
As a baseline, the ability of users to migrate their data between federated services is desirable, but any naive approach to this goal is a loaded gun.
That's... bleak. Where have all the techno-optimists gone?
It's unclear which participants shared concerns, as statements were anonymous. Perhaps DTI implemented Chatham House Rules. The event also notes that it can't provide an investigation and appeals process because the Data Transfer Initiative is "a small organization.” I hope Meta sprung for lunch.
The ACCESS Act was introduced in 2019 and reintroduced in 2022 and 2023. It would make data portability required by law. The bill is still under revision.
The RESTRICT Act, SAFE TECH Act, and Kids Online Safety Act also float around Congress. Each addresses one point in Zuckerberg's four-point plan.
The final versions of these bills could be a public good. Or, it could make starting your own social media company only possible with supplemental infrastructure. My bet is on whatever history would suggest.
Update: Since this post, I’ve discovered The Facebook Receipts, a project “designed to expose Facebook’s activity in Washington D.C. and around the globe as it attempts to influence legislation and regulation.” It’s worth a look.
threads.net is the new app.net but with ads and interoperable #
Consider a scenario where Meta leadership believed its business model was no longer viable long-term. What would their first step look like?
It might look a lot like the now-defunct app.net.
App.net was a microblogging platform that tried to compete with Twitter. App.net had a robust API that allowed 3rd party developers to build on its infrastructure. What made the start-up unique was that its founders encouraged developers to create social media platforms with their own branding, features, and experiences.
Threads.net would market itself similarly but with a few key differences— interoperability, content ownership, and a revenue split program. These flagship features make the platform attractive to small developers and server admins.
Threads.net's potential flagship features:
- Interoperability : All third-party apps are connected, or “federated”. Users can engage, follow, and migrate between apps easily.
- Content Ownership: Unlike app.net, third-party developers and users "own” their published data. Each third-party platform has a distinct URL where content lives.
- Ad revenue split: threads.net offers a turnkey ad delivery system. Qualifying third-party developers and admins can enroll in a revenue split program. If approved, they gain a dependable revenue stream.
An ad-revenue split would likely gain enthusiastic participation in the tech community. Most indie devs scrape by for years, and that's if they're lucky. The poor saps who receive VC funding get to watch their best feature copied on Instagram.
The interoperable and data storage features would also be widely supported. Martin Reece, creator of Micro.blog wrote about app.net in his online book. He suggested that what the start-up got wrong was its centralization.
From Indie Microblogging:
The app.net team got so much right — the early crowdfunding, the well-designed API, the developer story — that I didn't notice what they had left out until much later. All data lived at
app.netURLs, and when the platform was gone, all the posts and data went with it. There was no way to own your content.
Server admins would have their choice of third-party apps if they chose not to develop. For the adventurous, Meta would likely provide tools, resources, and UI kits for developing bespoke threads.net apps. Thanks to Llama, Meta's open-source AI development tool, almost anyone can build their very own social media platform. Perhaps even large influencers with the means and desire to create a branded experience.
Meta's vision for threads.net won't be a cakewalk, though. This market doesn't actually exist. Except it does, and Meta has already leached onto its network. All Mastodon server admins need is a push in the right direction.
Copy, Acquire, and Kill #
In some ways, app.net failed where Mastodon succeeded. Eugene Rochko, Mastodon's CEO, has helped cultivate an impossibly vibrant social commons. He did it with zero ads, a decentralized network, and interoperable servers. If Meta could copy Mastodon's strategy, scale the ecosystem, and then monetize, it could pivot, and kill Mastodon at the same time.
The biggest lie in the Fediverse is that Mastodon is too small for Meta to care about. By all accounts and reason, that's simply not true. Zuck is in this impossible position because of how often and indiscriminately he kills off start-ups. Universities have studied it.
Meta's Copy, Acquire, and Kill strategy is well documented. In 2020, Mark Zuckerberg faced a US Congress on allegations that Meta spies on small start-ups to identify targets for its insidious tactics.
Representative Pramila Jayapal:
"Your company uses data to spy on competitors and to copy, acquire, and kill rivals,” Jayapal added. "You've used Facebook's power to threaten smaller competitors and to ensure that you always get your way."
There are some immediately apparent reasons why Meta is interested in Mastodon. None of which relate to building a better social media for regular folks. Mastodon has experienced server admins and nine million users that Meta can siphon after it fully invades the platform.
A thousand little nodes #
My guess is Meta's plans are an open secret amongst the tech elite. A swarm of new micro blogging nodes are already buzzing around ActivityPub with great enthusiasm— Flipboard, Mozilla, Tumblr, Medium, Post.News, Pebble (previously T2), WordPress, Vivaldi, and Mammoth, to name a few. Some are even well-funded. None seem to have a viable monetization strategy. Shareholders can't buy a third house with the spirit of the open web, you guys.
As Meta pushes forward, we'll see more of these little nodes. Each will offer something unique, and Threads will be careful not to overlap. Post.news interoperable with a no-news Threads is perfect, don't you think?
But here's the question— why ActivityPub?
This is a tougher nut to crack. On the one hand, the protocol is a perfect fit for Meta's threads.net ambitions. But wouldn't the Data Transfer Project Initiative meet those needs as well? Perhaps. But DTI's "mission” isn't necessarily social, and I think Meta wants to keep it that way.
Look, there are lots of good people with the best intentions at every level of that project- er, initiative. But surely, Meta will use DTI to migrate Mastodon users over to threads.net at some point. By isolating the social graph portability to ActivityPub, Meta can swap the protocol out with something proprietary (and more restrictive) once it ditches the Fediverse.
In the end, ActivityPub still exists, but its platforms are a virtual wasteland. Some version of Mastodon lives on through threads.net. But at what cost?
Mark's Catch 22 #
Mark Zuckerberg is in a bit of a pickle right now. If he could, I think he'd be cool just swallowing up start-ups and surfing with that white shit on his face the rest of his life. I think he wants to avoid pivoting to infrastructure. Was the whole Metaverse thing his last-ditch effort to do anything but Platform as a Service? I wonder.
Meta could've offered those services years ago. Instead, it routinely gives away its server architecture. Those contributions are the least objectionable thing about Meta.
But he can't not grow the company. Wall Street won't allow it. And he can't grow it through acquisition because of anti-trust. He's stuck.
After Cambridge Analytica, Meta couldn't buy a WordPress blog from Theme Forrest without the government up their ass. Zuck's biggest fear is anti-trust breaking up Instagram and WhatsApp. If he can remove the "Gatekeeper” characterization, he might have a chance to save his empire.
Infrastructure is Zuckerberg's only play. It's a good one, too. Effectively, four companies currently run the cloud. Barriers to entry are high, and infrastructure is costly. Just ask Meta. In the last four years, they've spent billions on non-Metaverse, non-AI data center expansion.
Today, all your favorite apps run on Amazon Web Services. Soon, all your social media will run on threads.net.
This post is number 2 of a series. You can read the first post The medium is the message— Threads isn't a win for the Fediverse, and the third post How do we stop Meta in 2024? We fix the information loop
Since 2019, Meta grew from 15 to 21 data centers across the globe, with at least three more reportedly in construction. This activity is in addition to any Metaverse needs, as Meta signed a long-term contract with Amazon Web Services for its virtual reality computing. In February 2023, Data Center Frontier reported Meta halted at least three data center projects. Its architects achieved a breakthrough that extended the lifespans of its servers. Meta decided to include those updates in new centers, so it delayed construction. The improvements did not extend to AI-based servers. ↩︎