Breaking up Facebook

Facebook Employee Handbook

It is encouraging to read that in his recent interview for The Atlantic Barack Obama identified the new malevolent information architecture facilitated by Facebook as the “single biggest threat to our democracy,” and to see growing appreciation of this threat among the Senators asking Mark Zuckerberg pointed questions about his company.

Now that this problem has finally caught the public attention, we have to have a serious conversation about the solutions. Sadly, whenever the subject of breaking Facebook apart comes up, it is only discussed in reductive terms of forcing Facebook’s acquisitions like Instagram to become independent again, and how that would or wouldn’t work out in the long run.

Naturally, Zuckerberg is going to work very hard to persuade us that it wouldn’t. As you might have guessed, I have some thoughts on the arguments against breaking up Facebook that I’ve seen so far.

Integration between Instagram and Facebook led to integrity, privacy, and security improvements for both companies

Even 8 years after the acquisition, Instagram is still significantly behind Facebook in many integrity problem areas, including hate speech. The EC Code of Conduct report that we keep seeing held up as proof that Facebook is doing better than the competition shows a massive drop in Instagram’s removal of reported illegal hate speech from 71% to 42% over the past year.

Instagram doesn’t benefit from the integration automatically, it is still a lot of hard work for Instagram’s team to keep up with what Facebook integrity tools and research have to offer.

It is also a fallacy to assume causality between the acquisition and investing in integrity. On its own, Instagram might as well have pivoted to privacy and well-being sooner, or could have adopted a more aggressive policy against misinformation and hate speech unshackled by the requirement to unconditionally platform politicians.

And what of all the other competitors? Should Facebook acquire them all for the greater justice?

Or maybe — bear with me — Facebook should start treating integrity as a public commons and share their research and tools with the entire industry, so that the competitors they didn’t yet acquire could also use it to stop bad actors on their platforms?

We work hard and make personal sacrifices for Facebook, it is offensive to suggest that integrating acquired platforms into Facebook is harmful to society.

Don’t blame the player, blame the game. Many at Facebook are just as offended as you are that all their hard work and sacrifices end up being used to promote hate and violence.

Facebook isn’t that large. Journalists pretend we have all kinds of social power and influence we don’t actually have.

I think the part where Mark and — judging by FB stock price trajectory — the rest of Facebook’s shareholders could be confident that most of the advertisers on Facebook can’t afford to join a boycott like #StopHateForProfit even for a month is a strong indicator that Facebook has enough power to no longer be accountable to the free market. Isn’t that exactly the problem that anti-trust regulation is supposed to solve?

Is it bad to have power without accountability?

Yes.

You know why all key human development indicators have been steadily improving for the past 250 years? Because humanity began getting rid of monarchs and dictators.

And no, the Oversight Board is not even close to providing accountability for Facebook’s power, neither in time (it’s not there yet and won’t be there any time soon), nor scale (a dozen or so free speech experts can’t possibly democratize decision making throughout a company of 120,000 or so workers), nor scope (even within the already very narrow scope of content moderation, its mandate has been restricted to only appeals escalations and only on the content that was removed by Facebook — not content Facebook failed to remove, not community standards, not harmful behaviors, not ranking rules, not any kind of product design decisions).

How would breaking Facebook up make anything better?

Glad you asked.

First of all, let’s get rid of the strawman in the room: taking Instagram and WhatsApp off Facebook infrastructure is breaking up Facebook along the wrong axis. You can’t just put Instagram back on AWS and Akamai, AWS and Akamai aren’t big enough.

The only way to break up Facebook that wouldn’t result in an immediate collapse of the child companies is to first break out and commoditize Facebook infrastructure (in the broadest sense of the word). Facebook’s VP of Global Affairs and Communications Nick Clegg should understand this concept well: he wrote the EU local loop unbundling regulations. The idea is to provide equal access to the infrastructure at the same price to all companies competing with the same market.

The price part is easier than it sounds. Facebook has nearly perfected the process for budgeting datacenter capacity and measuring transitive resource utilization of a product through the entire stack of backend services (as the company saying goes, “this journey is 1% finished,” but they’re moving in the right direction already).

Drawing the line between product and infrastructure is conceptually simple (anything reusable across different apps within the family is infrastructure), and the key roadblocks — the overgrown interdependence of services and the huge monolith of the core PHP monorepo — are tech debt that’s already overdue.

Balancing data sharing with respect for privacy is somewhat trickier but still tractable, and also a problem Facebook has already publicly committed to. The interoperability between Facebook, Instagram, and WhatsApp does not require losing the separation between the social graphs of these apps nor breaking their privacy policies.

With that out of the way, what are the benefits of turning Facebook into an ecosystem of independent social apps renting a shared infrastructure?

1. Power. Facebook power comes from the size and durability of its social graph. People are locked in because that’s what all their friends are using. Advertisers are locked in because that’s where most of their audience is. Politicians are locked in because that’s where most of their voters are. Even separating Instagram’s 1.3B MAP and WhatsApp’s 2.1B MAP from Facebook’s 2.6B MAP would by itself diminish Facebook’s power by creating significantly more viable alternatives for people and advertisers than Snapchat’s 230M, Twitter’s 330M, and even TikTok’s 800M.
   .
   Efficiency and scalability of Facebook’s infrastructure is a key part of its early success and continued profitability. Equal access to the same infrastructure would enable even more competing social networks reach and maintain viability, diminishing Facebook’s lock-in on people’s friends even further.
2. Blast radius. Diminished power and stronger competition wouldn’t just improve market’s ability to hold the company accountable, it would also reduce the blast radius of bad product and policy decisions.
3. Policy. If Instagram and WhatsApp could make independent decisions on community standards and political engagement, we’d be likely to see significant divergence from Facebook policy (e.g. stronger focus on safety in Instagram and on privacy in WhatsApp). They’d be able to implement a firewall between lobbyists and policy researchers. They’d have their own PACs or not have PACs at all. Maybe their CEOs might even refrain from inviting white nationalists for dinner to discuss how to make them more comfortable on their platform.
4. Product differentiation. Independent social apps are more likely to innovate and differentiate and would be less likely to choose unification over interoperability.
5. Competition. Better not just for users and advertisers, but also for workers who would have more employers to choose from and a better chance to land a full-time job instead of getting stuck in a precarious contingent contractor position for years.
6. Monetization models. Having more large financially independent social networks makes it more likely that some of them will be willing to experiment with alternative monetization models that are not driven by screen time addiction and psychological manipulation.

Last but not least, the question of breaking up Facebook should not be narrowly framed in terms of reversing the acquisitions of Instagram and WhatsApp. There is an even bigger problem with Facebook than leveraging its size to eliminate competition.

Facebook has grown large enough to be fundamentally unable to control a broad range of abuse on its platform. Facebook’s profitability comes from externalizing the cost of consequences of that abuse to the entire world.

Breaking out the acquisitions that should not have been approved in the first place is not going to be enough. New regulations have to be put in place to restore the balance between, on one hand, the level of control companies like Facebook have over people on their platforms and the profit they extract from that, and on the other hand, the burden of detecting and countering misuse of these platforms, and of mitigating and preventing public harm that comes from such misuse.