How to beat social media's network effects

I've written several times about unbundling social media into its constituent parts. I think that "social media" should mostly be a variety of different publishing apps, reading apps, and small discussion communities, instead of a small number of large, monolithic platforms. My best piece on this is You can help unbundle social media; the most popular one is How to rebuild social media on top of RSS. Those articles cover the what and the how. Now I want to shed a bit more light on the why.

In a single word, the answer to "why" is "competition." More competition could help solve several problems with social media:

• Content moderation. Different platforms can have different moderation policies, and users can migrate to the ones they prefer.
• Incentive alignment. Social media platforms have many different kinds of users, all with different incentives. Platforms also have their own distinct incentives. The smaller the platform and the more focused their user base is, the higher the chance that incentives will be aligned. With more platforms, users can switch to platforms that cater to their own needs.
• Innovation. It's easier for smaller companies to experiment with different ways of doing things.

However, there's a reason the industry tends to gravitate towards a few, large platforms: network effects. The benefits of having lots of users dominate everything else.

The solution is interoperability. If you have a bunch of different apps that work well together, the network effects can be shared among participants instead of being concentrated in a few large players.

Interop has some downsides, though. The biggest one is that the need to coordinate can slow down innovation. If you've ever worked on a group project, you're no doubt aware that it's often faster to just do the whole thing yourself. After all, the Internet's application layer did mostly consist of a bunch of interoperable services, a few decades ago. The platforms won because they were better. The cost of interop negated the potential benefits of competition.

So how do you address that? You figure out the minimum amount of interop that's required to reach your goals.

That, by the way, is one of the biggest mental shifts I've had over the past several years. I used to think that more interop was always better, or at least, that more interop was always a worthy goal. I spent a fair amount of time wondering about how software could be architected such that interop would be easier for application developers to implement and maintain.

Now, I think of interop more as an evolutionary feature, like being tall. Being tall has advantages, for example, you can reach fruit that's higher up in a tree. But it comes with a cost: you have to consume more calories per day. Sometimes being short is better!

So we need to find the right amount of interop. How much interop is needed to get the level of competition we want? What functionality needs to work across across different apps, and what functionality can be siloed within individual apps?

I have two beliefs here:

• Subscriptions need to be interoperable. If I subscribe to you, then that subscription should remain intact even if one of us switches to a different app.
• Discussion does not need to be interoperable. If you've started a discussion on a particular app and I want to reply, it's acceptable if I have to install the same app.

This is why my "model of the universe" focuses on publishing apps (like Wordpress and Ghost), reading apps (like Gmail and RSS readers), and discussion apps (like Discord and Slack). Subscriptions are handled by publishing apps and reading apps. If you want your post to travel across the Internet, write it in a publishing app. If you want to hash out some ideas within a particular community, write your post in a discussion app. These two types of communication don't have to mix.

Reading, publishing, and discussion apps are already in use by lots of people. Let's see how far that model can be pushed.

 

Notes and further reading

• See The Standards Innovation Paradox for more details about the cost of interop.
• Geoffrey Litt is a good person to follow if you're interested in how to make interop easier.
• Protocols, Not Platforms: A Technological Approach to Free Speech started me thinking in this direction several years ago. It acknowledges the downsides of protocols but doesn't offer much of a solution. That's the gap I'm trying to fill with this article.
• Amplification and Its Discontents addresses the "how do we fix social media" question from a regulatory perspective. I think regulation and grassroots solutions should both be pursued—perhaps regulation could help push the market in the direction of the unbundled web.
• In an interview, Jeff Kossef mentions that Section 230 was intended to promote a market-based approach to content moderation, but it hasn't panned out as well as expected because network effects have negated competition.
• Besides the cost of interop, there are at least a couple more forces that push in the direction of large, centralized platforms. Shared network effects (and public goods in general) are hard to monetize, so they tend to get a suboptimal amount of investment. See Funding software innovation and Roots and ranches: centralization and the role of software startups for my thoughts. Second is the simple fact that it's cheaper to roll out a feature to a current user than it is to acquire a new user, as Ben Thompson discusses. Efficient advertising networks are important—I don't buy in to the idea that advertising is inherently evil. (In practice, in some cases, sure. Inherently, no.)
• The growth of Mastodon is encouraging. I hope the fediverse continues to gain momentum, even though I prefer to focus my own efforts elsewhere. I would probably be more interested in Mastodon if I liked writing short-form posts. But I already use the "unbundled" model myself—I write this newsletter, I put my reading time into newsletters and blogs, and I'm active on a few Slack and Discord communities—and I really like it. I left Twitter a month or so ago and haven't felt any need to replace it.
• Substack is taking the inverse approach: they're building another monolithic, VC-funded platform, while trying to offer a specific set of features which they believe will fix the main problems with social media. There's value in what they're doing, but it's not as long-term of a solution as what I'm trying to achieve. Already they've started to exhibit incentive alignment problems, and their hands-off content moderation policy conflicts with their desire to make their own brand front-and-center.