Project Liberty sat down with Wendy Seltzer, an advisor to the Decentralized Social Networking Protocol (DSNP).


DSNP is gathering momentum within the technical community. You’ve worked in the W3C and standards for a decade. What makes a successful open development process?

While there’s more than a bit of magic to it, a few factors can help: You need an open and transparent process, interested parties willing to give each other the benefit of good faith interactions, and participants with incentives aligned toward a technical or policy matter they need to cooperate on to meet their goals. A lot of pieces have to work together to make a standard usable, useful, and used, to help it in gaining market and mindshare and working long-term. DNSP is at an early stage.

Does standards work often start in a smaller group before opening up?

Development is often kicked off by those with a particular interest. Key to successful transition to an open standards body like W3C or IETF is a willingness to find broader agreement and in that process, to revisit decisions. There needs to be willingness to re-examine questions you thought were closed when new information is learned or participants come in with different needs. At the same time, you also need sufficient mechanisms to close off repetitive debates, to say that “unless you can bring new information, we’ve got that covered.” That’s where transparency and documented decisions help.

What’s happening in the competitive landscape DSNP is entering?

We’re seeing lots of movement in the social networking space, most obviously the turmoil at Twitter, now known as X. It’s showing people that what they thought of as a relatively stable but controlled space to exchange information just isn’t that. A proprietary platform can be changed from under us without rhyme or reason. If we don’t have open protocols or access to our own data, it’s harder to take our input elsewhere and rebuild on a new foundation. So, there’s appetite for the offer DSNP is making, and lots of other groups and entities responding. There’s the ActivityPub protocol of Mastodon and the Fediverse, a bunch of blockchain efforts, Bluesky starting out in a closed development and proposing to open up. There’s also Meta’s effort, Threads, and branches from other originally one-to-one communications like Telegram.

Lots of big platforms thought they didn’t need to federate or interoperate as everyone would be happy to give control of their social graph to a platform where it could be maintained in a nice, closed walled garden. That’s changing, but will now be the time people finally know they want more control or interoperability? And will there be sufficient economic forces to support a sustainable infrastructure? We’re seeing growing pains in other networks, as Mastodon server operators find it difficult to work through the choices of sustainability, moderation, whom to federate with, how to build a global scale of federation choices and moderation choices, and how you fund all that.

It’s like the late nineties when ‘local loop unbundling’ tried to break different services out of the telecoms stack, but it wasn’t always obvious how to make them sustainable. How do we pay for sustainable infrastructure services? That’s the big question. Or how do we make commodity markets attractive? To provide services that provide for steady, non-monopolistic revenues, if we can even persuade users to pay for those services on an ongoing basis? Investors like services that can be monopolized and have increasing returns to scale from becoming the dominant player, but we’re explicitly saying we want to build an ecosystem where there isn’t domination, where folks just make a nice profit or a nice social good out of the steady provision of services.

Are there examples of this kind of infrastructure unbundling being financially sustainable?

In the DNS space, ICANN set up contractual structures where registrars provide a templated service registering domain names. Domain name registration is roughly the same process on any domain, and yet the registrars find different business models around bundling services, offering packages, simplifying the management for users. It’s a highly competitive market, so it can be done.

Why is the blockchain central to DSNP?

I’m blockchain agnostic. I’m not going to come out as ‘Blockchain is key.’ It’s proposed as a way to do permission-less, decentralized record keeping. It outsources the governance of that to math and some assumptions that enough people with different interests will use that blockchain that it won’t fall under the control of any particular set. Is blockchain better than doing that through a governance process where people have votes on maintaining a register or where a non-profit maintains a registry? It’s different.

DSNP and the other initiatives are starting from the idea that the current way services are arranged in a centralized ‘social media stack’ isn’t inevitable or even desirable. Many different components make up the full stack of social networking. Protocol is just one. On top of the protocol people build applications and communities. The governance structures for those communities are distinct from those for the protocol itself. The users and community can set their own rules for how you use it, who can be part of the community, on what grounds someone is kicked out for being a bully or an abusive user of the application. And all of those together are important to what makes valuable space for online discussion.

One of the things we’re learning, especially in content moderation, is that global rules are hard and often not appropriate. Communities are lumpy. They have different interests and ideas about what people can say to each other. Different places have different rules about what’s permissible. Child sexual abuse images (CSAM) are impermissible everywhere, but how CSAM is recognised and prevented is still contested. Then you get into the lumpy bits of speech; what you can’t say about the king in Thailand (lèse-majesté) will look different under the US First Amendment.

This conversation has been useful to identify the parts of social media that are social, governance or determined by the business model, and how services and applications are different from the protocol. In the current concentrated stack, they all get squashed together, but they don’t have to be. Right. We can distinguish which pieces are part of the protocol and which are determined at the application layer or part of community or governance structures. DSNP isn’t the only way to effect change in the social media ecosystem. We look for how to give affordances to good behaviour and what we want to enable, but through the protocol alone we probably can’t affect all the things that make social media good or bad. We need to make sure there are hooks for those other layers to do the hard work of content moderation and community building. We need to make spaces for different ways to think about and build new ways to connect.

What’s next for DSNP?

I see good faith in the people stewarding the work and a desire to welcome participation. We still have work to do to build or find the structure in which it can best invite that. I’m looking forward to more discussion about the ecosystem it can fit into, what the DSNP can contribute and what we can adopt from work elsewhere.


Wendy was counsel to the World Wide Web Consortium (W3C), and has served on the boards of The Tor Project, Open Source Hardware Association and ICANN.