Just under one year ago, my cofounder Mark and I decided to integrate our end-to-end encrypted messenger, Germ Network, with Bluesky's AT Protocol. 

At the time, we had already shipped our iOS app, Germ DM, and upgraded it to standardized end-to-end encryption with Messaging Layer Security. I was attending SF happy hours and hosting events of my own to figure out our next move. Our messenger let people build multiple identities, creating new power over what they shared and how they managed and understood their relationships. People loved this idea, especially if they managed multiple social networks, say, because they had two jobs, or closely guarded private lives—and women and queer people especially understood the need to not share phone numbers—but they needed a lot of features to make the switch, and it would be a long time before we could catch up with Signal, let alone WhatsApp. 

And while our shareable “cards” let people share different information in different relationships, they were self-contained. I started to understand that sharing less is only valuable when the default is sharing more. Exchanging full social media profiles is now the normal way our relationships begin—when I meet someone, I look at their profile to see what they’ve posted lately and who we know in common. This social proof creates trust. 

My cofounder Mark reminded me that our architecture was designed for integrating with external identities. New open social protocols were gaining steam—we both were already using them. What if we pulled one of these protocol's identities into Germ? 

New open social protocols were gaining steam—we both were already using them. What if we pulled one of these protocol's identities into Germ? 

* 

[a new girl comes to town]

The stranger who came to town, the new girl in school. 

If you’re reading this post, there’s a high likelihood you already know a lot about the AT Protocol, and about computing in general. But a few years ago, I didn’t. So this post isn’t from the perspective of an expert. Instead—a classic narrative framing—it’s from my POV as an outsider. The stranger who came to town, the new girl in school. 

The reason this conceit works so well is because it allows a viewer to understand a full, busy world with its own assumptions, values, and practices. It makes an anthropologist of all of us—a goal, as Clifford Geertz said, "to make the familiar strange and the strange familiar." If you’re newer to the world of social protocols, I hope this framing helps you make more sense of this new movement in social media technology. But if open protocols are the air you breathe, perhaps this perspective will help you empathize with your next four billion unregistered users, still waiting for you behind the algorithmic doors of their favorite (least favorite?) walled garden. 

See, I’m what they call the nontechnical founder. I have a PhD, but not in computer science—in rhetoric and literacy education, and in women’s and gender studies. Before Germ I was a writing professor—and, like many millennial women, an avid user of Twitter, Facebook, and Instagram. A writer, I also blogged. As an educator I had what my professional peers called a new media teaching practice, which meant I used these tools in my classrooms and worked with my students to cultivate their skills with the media they used to communicate in their real lives. I also had a growing research area in critical internet studies, framed by my interest in race, gender, power, and virality online. 

I did not think about, read about, teach about, know about, protocols. I thought, knew, and talked about memes, social media, trust and safety, virality, DMs, the algorithm, appropriation, capital flows, the advertiser model, Big Tech and, of course, walled gardens. 

I thought, knew, and talked about memes, social media, trust and safety, virality, DMs, the algorithm, appropriation, capital flows, the advertiser model, Big Tech and, of course, walled gardens. 

I understood what a walled garden was: the idea that Facebook users only talked to other Facebook users, Discord users to other Discord users. I recognized that e-mails, phone calls, and blog posts were not limited by these artificial barriers. But when I dreamed of better social, as I increasingly did, I was thinking, basically, of a better garden. It was all I had ever known. Even my earliest days of internet browsing had been through AOL. The internet had always been framed for me by a large corporation. 

Immersed in research on the harms of social media, the findings seemed conclusive. The advertising business model, with its core incentive to keep more people on a given platform for longer, clicking highly-targeted ads, drove a wide range of harms: dangerous introduction patterns to ensnare people in dense social webs; content recommendation algorithms that highlighted incendiary, and thus addictive content; a disdain for privacy, which would obscure the platforms’ access to their core asset, user data; and careful design to prohibit easy offboarding with one’s posts, photos, and friend graphs. 

I kept reading articles with the same conclusions. But who was going to build it? Better social media obviously needed new business models that incentivized more user choice, including access to meaningfully private spaces. In particular, I was interested in fully private DMs. I had the technical literacy to understand that without end-to-end encryption, anything we wrote online was being stored and read by people other than the people we thought we were talking to. There was a reason E2EE had become the standard in global messaging for more than 3 billion people on iMessage and WhatsApp. But web 2's platform business incentives made privacy-forward social organizationally impossible. 

As a communication expert I also knew that the minimum viable product of a community is a 2-person relationship. Dating app UX was particularly interesting to me, because of the ways it was built for discretion. Once you unmatched with someone, they could never find you again. What if, when you met someone online, you could jump into a truly private message, without going anywhere? I came to believe that the MVP of a healthy 2-person relationship—one where you’re truly alone, not being spied on, not manipulated, and not coerced—was an end-to-end encrypted message based on a relationship that can be fully dissolved. 

The MVP of a healthy 2-person relationship was an end-to-end encrypted message based on a relationship that can be fully dissolved. 

The pandemic gave me time to explore this vision, and eventually, I met my cofounder, Mark (on Twitter, in fact). Mark, I would learn, also dreamed of healthier, more consensual digital connections. But his dream, from a technical standpoint at least, was bigger than mine. He had worked at Apple on iMessage, FaceTime, and Advanced Data Protection—all of their end-to-end encrypted products, and more. He knew about things I did not yet understand: about user agents, authentication, local-first storage and multi-device sync, telcos and their discontents, protocols and interoperability. Mark had a vision for an open protocol that made sending secrets as simple and universal as sending an e-mail or making a phone call.

As a privacy engineer on iMessage and FaceTime, Mark saw that phone numbers made people less safe, and their lives less convenient. People were overwhelmed with spam; were afflicted by stalking and harassment from both individuals and businesses; and wanted to manage multiple identities from a single device, but couldn’t. End-to-end encrypted communications, which traveled over the internet, were being held back by the phone number.

End-to-end encrypted communications, which traveled over the internet, were being held back by the phone number. 

We talked about the reasons that walled gardens also extended to E2EE messengers: Signal, iMessage, WhatsApp. Whereas social media profiles used walled gardens for business reasons, E2EE messengers used them for a core technological function, to authenticate conversation members and distribute the key packages that made totally private communication possible. This architecture allowed the rapid global scaling of private communications, but now it was trapping people in apps and limiting the flexibility in how they connected. 

Mark was also thinking about a new regulatory trend, epitomized by the DMA, the Digital Markets Act, from Europe. In the next few years, messaging services would be legally required to talk to each other, to interoperate. And getting there was going to be very hard for them. Serving the interoperable, secure future could be easier (not easy, just easier) if you started from scratch. And while you were starting over, you might kick phone numbers to the curb, too. 

When we met, he’d been working on a protocol for delivering ciphertexts between user-defined addresses. It mapped uncannily to my dating-app-inspired UX for messaging—but now, instead of each user having one profile card, they could have many. We got to work. 

**

Great products drive protocol usage.

By March of this year, our ATProtocol integration was well underway. Our advisor Richard Barnes, an author of the internet security protocols TLS and now MLS, suggested we go to the first ATmosphere Community Conference in Seattle. Mark and I bought our tickets and showed up. Richard was the only person there we knew. 

At the conference I learned much more about not just the ATProtocol, but Bluesky the corporation. I learned that Germ is a lot like Bluesky organizationally in that we have our open-source protocol that we want to be universally accessible, and we also have an app that will drive our business, while the protocol accelerates user acquisition.

I was also struck by Paul Frazee’s remarks on Bluesky’s theory of change, which I felt were nearly congruent with our own. Technological change drives social change, but to do so it has to thrive in the marketplace. For consumer software, the way to do that is as a startup. Further, great products drive protocol usage. Like Bluesky, Germ had also incorporated as a Public Benefit Corporation, in our case, with a chartered benefit to promote healthy communication. Like Bluesky's, Germ's users didn't need to know there is an open-source protocol underneath. But the more they get into our unique services, the more different they'd realize we are.

In Seattle, I also learned that we were part of a community.

In Seattle, I also learned that we were part of a community. There and later, we’d meet some of the other developers building experiences on top of the ATProtocol. 

We met Tori and Reed, building Skylight. 

We met Rudy, building Blacksky. 

We met Sebastian, building Flashes. 

We met Devin, building Graze. 

We met Erlend, building RoomyChat. 

We met Eli, building Stream.place.

We met Brendan and Jared, building Leaflet, where I’m blogging now. 

When I opened Flashes, my photos were there.

When I opened Skylight, my videos were there. 

On Bluesky, I could add feeds made on Graze by strangers to my main feed bar. 

I could subscribe to Blacksky’s moderation labeler to hide hate speech and misogynoir. Or, I could use their web app to see my whole feed. Friends even moved their whole handles there, changing where their data was stored and who was responsible for it. 

It was (it is) my first time in an open-source developer ecosystem. After working every day alone with Mark, and then with our tiny team, suddenly we were in this huge community of people building products that talked to each other. 

In May, we publicly announced our atproto integration, and in August, we launched it into private beta. 

CJ, the developer of Popfeed, created a custom Germ DM button to hide our links-in-bios. 

Someone else used Graze to make a feed of Germ Users, so that they could find each other and start chatting. 

The future I’d dreamed of, of healthier social media, was much closer than I thought. Mark's vision to build on an open messaging protocol made us ready to dive into an ecosystem of interoperating products—and here they were, waiting for us. Germ could build the private messaging while other teams focused on the feeds, the photos, the videos, the payments, the blogs. Together, we could—I believe we will—replace the Facebook model of single-actor media domination, and put social media on the internet again. Less doomscroll and brainrot, and back to surfing, learning, and making actual friends.

Part II of II coming soon.