Nick Gerakines gives a short talk at ATmosphereConf Seattle 2025.
Hi, I’m going to talk about Lexicon, community, and Smoke Signal. I’ve been told that this presentation is a bit dense, so I’m going to go through about 65 slides. It’s quick — 30 seconds apiece — don’t worry. We’ll do questions and comments at the end, so until then, please enjoy each slide equally.
We’ll start with a little bit of pre-presentation housekeeping. I want to take a quick moment to appreciate the fact that we’re all here and able to do this. I feel so fortunate to work alongside so many smart people.
I’m especially grateful to people like Boris, Brian, Tom, Rudy, Paul, Ted, and the many others who have contributed directly and indirectly to this community.
I also want to acknowledge that Smoke Signal borrows from language and concepts that originate from Indigenous people. Colonialism is not a thing of the past — it’s an ongoing issue that directly and indirectly affects us all. We would be remiss to talk about growth without reflecting on and acknowledging the past. From previous talks, that’s a lot of what we’ve been doing.
Hi, I’m Nick. I’ve been in tech for a while. I’ve worked across industries — EA and Blizzard, toys at Mattel, tech and social media at Six Apart, LiveJournal, and Yahoo — and, of course, in the DevOps space at Datadog. I’ve also spent a lot of time in the AI/ML/CS space, both at previous companies and currently at GitHub, where I work on Copilot.
I have brown hair, I’m wearing a blue hoodie, and I’ve got a bushy mustache and a septum ring. I’m standing behind an iPad in front of the screen. All of the slides and content are presented with notes, and all of the images include alt text. If you need anything else to understand this presentation, please message Boris, myself, or Ted — we’ll get it to you.
Smoke Signal
In the summer of 2024, I launched Smoke Signal Events — an event and RSVP management application built on top of the AT Protocol. Today’s talk is about what went into and came out of developing Smoke Signal.
Smoke Signal was inspired by many things, but it really comes down to a question I’ve spent a lot of time thinking about:
How can communities outlive the platforms they use?
How can independence and autonomy guide the design of the tools and systems that communities depend on?
How can we model products and services that are financially stable and still respect their users?
What safeguards can we put in place to ensure people can exercise choice and use the services they want?
We ask these questions because we’ve all had experiences with this problem — walled gardens.
Platform Decay and Lessons from Meetup
Meetup started as a way to create communities emphasizing group events. In 2017, it was acquired by WeWork. The transition was rough on Meetup communities; business decisions drove many people away.
Later, in 2020, there was another sale, and again in 2022 to Bending Spoons, the company that also acquired Evernote. Both Meetup and Evernote have suffered from what I call the corporate product death cycle.
The lesson: websites come and go. Sometimes you’re too close to recognize their slow decline. For many communities, this was death by a thousand cuts.
In June 2024, Meetup announced major price changes. Small communities — especially nonprofit or casual groups — can’t and won’t pay $300 a year. It’s valid that Meetup needs to cover costs, but there’s still a big disconnect between the company and its users.
Data, Dependency, and Facebook
A 2022 study found that only 18% of U.S. social media users were confident Facebook protects their privacy and data. People acknowledge it’s a trap — “It’s how I keep up with my friends and family,” or “That’s where all my events are.”
As an organizer, I found Facebook necessary for event awareness. It was the number one source of attendees, only second to Meetup after Meetup started to decline.
On Facebook, you are both the product and the prey. You leave a trail of posts, comments, and likes wherever you go. Meta tags and labels you, and advertisers follow your scent. Your information and attention are hunted, and a lot of money is made in the process.
If you haven’t read Anil Dash’s The Purpose of a System Is What It Does, I strongly encourage it. Facebook has been thoughtfully and intentionally designed to make it hard to leave. It manipulates your attention and relationships to make itself feel necessary. For many, leaving is prohibitively difficult, sometimes even harming relationships or finances. Simply put: Facebook works exactly as designed.
The Rise, Fall, and Return of Upcoming.org
Upcoming.org launched in 2003 as a social event calendar. In the heyday of Web 2.0, it was scooped up by Yahoo in 2007 and ran for about ten years before being shut down in 2013. It was revived through crowdfunding in 2017, and the event archive was brought back online. But excitement faded, and updates became infrequent. The last major update was in June 2023 to fix a sign-in issue.
To this day, you can’t really use Upcoming effectively. Many people have poured time, energy, and money into building communities and tools to support them — and that’s what led to Smoke Signal.
Early Prototypes and the AT Protocol
My early prototypes started with ActivityPub. The Tavern and AP Events projects were implementations of group content and event management. I spent considerable time working through actors, inboxes, and interactions, even contributing upstream to better understand how it all worked.
Ultimately, those projects were shelved. ActivityPub wasn’t a good fit for what I wanted to do. Its flexibility is both its strength and weakness — when any type can be anything, you end up with many unofficial implementations, each with its own quirks.
The AT Protocol was game-changing. I wrote an article about why I wanted to use it for Smoke Signal. It emphasizes identity ownership, mobility, data portability, and user choice — all feeding into the concept of a “credible exit.”
If you love your users, you have to let them go.
Identity, Data, and the AT Protocol
The AT Protocol provides decentralized identifiers (DIDs), which support both the did:plc and did:web methods. Handle resolution, DID document verification, and service lists form the foundation for identity ownership and mobility.
Each DID document lets services like Smoke Signal know what my identities are, where to store my data, and how to verify it. Simply put: I’m known by the domain I control, and I host my data where I choose.
OAuth is employed in a clever way. Originally, app passwords were meant to grant applications API access on behalf of users. Smoke Signal was one of the first apps to fully implement OAuth for the AT Protocol. It reduces the number of clicks to log in and simplifies the user experience.
More importantly, OAuth cements identity ownership and portability. As a user, your Personal Data Server (PDS) is your home base — where you grant permissions and control credentials.
When a user visits Smoke Signal, they enter their handle, which resolves to their DID document. The app redirects them to their PDS for authentication, then back to Smoke Signal with credentials. There’s no need for a manual registration process — authentication just works across providers.
Repositories, Relays, and Lexicons
Repositories, listed in the DID document as PDS entries, serve content in a consistent, verifiable way. When someone creates an event on Smoke Signal, any app view or client can request that exact same content.
Relays are another essential component — streams of records that clients can connect to and receive AT Protocol events in near-real time. Smoke Signal isn’t the only producer or consumer of these events. Relays make interoperability much easier because apps don’t need a full view of the network to participate.
Lexicons define how records are structured. With them, records become discoverable, portable, and interoperable.
Lexicon Community
Lexicon Community doesn’t own or manage Smoke Signal — I’m just a member who contributes to it. It exists to give Smoke Signal users a credible exit if they ever want to leave my app.
Updated event records reference Smoke Signal in name and description, but the types are communally owned. They’re influenced by existing standards and refined through open discussion.
Lexicon Community’s mission is twofold:
To create an effective blueprint for collective ownership of lexicons in the ATmosphere.
To build a stable, thoughtful set of lexicons for developers, systems, and SDKs.
It’s self-governed and independent from Bluesky Social PBC. There’s a technical steering committee made up of volunteers with different backgrounds and perspectives. It has final authority over technical direction, governance, and infrastructure.
Working groups handle specific objectives — schema changes, infrastructure management, and governance reform.
Since launching in November, we’ve released several lexicons — bookmarks, events, locations, and reactions — and we’ve had great discussions about versioning, profiles, attestations, traits, and more.
Lexicon Resolution
Lexicon resolution is a recent addition to the AT Protocol spec. It allows apps to retrieve schema definitions dynamically.
When a record is encountered, its lexicon type is extracted and resolved through DNS, similar to handle resolution. The resulting repository provides the schema, enabling validation.
All released Lexicon Community lexicons can be resolved this way. Tools like Tom Sherman’s AT Proto Browser perform lexicon resolution on demand — when you view a record that matches a Lexicon Community type, you see a nice green checkmark.
Building Together
Visit Lexicon.Community
The videos from ATmosphereConf 2025 held in Seattle, Washington, are being republished along with transcripts as part of the process of preparing for ATmosphereConf 2026, taking place March 26th - 29th in Vancouver, Canada.
Follow the conference updates and register to join us!