Trezy Who 0:04 I would like to tell you all a story about how I ended up supporting the game dev industry on Blue Sky and the Atmosphere accidentally. 0:15 But first, an introduction. 0:18 My name is Trezzi. 0:20 I've lived many lives. 0:22 I've been a professional musician, an indie filmmaker. 0:25 I've been a graphic designer, a stunt actor. 0:27 But most importantly, I've been a software engineer for more than 2 decades. 0:32 And I've been working in and around the games industry for many years. 0:37 I'm also the founder of Birbhouse Games. 0:39 It's a game studio focused on building games using web technologies. 0:46 I'm also the curator of the game dev feeds on Bluesky, the operator of the games industry labeler also on Bluesky, the author of Happyview and Debuff and the steward of the Pinner Act, the founder of Cartridge, but I might be maybe getting a little bit ahead of myself here, so we'll get into all of it. 1:08 I want to preface this story by saying that I did not start this journey that I'm about to describe with some master plan. 1:16 I didn't sit down and say I'm going to save video games. 1:20 I joined Bluesky because I was mad at Twitter. 1:24 And so I started building, because that's just sort of what I do. 1:27 And once the ball was rolling, there wasn't a whole lot of stopping it. 1:31 So I'm gonna walk you all through how I got to where I am today with all the things that I've built. 1:36 And I'm hopeful that, you know, with that discussion, I can help others get started on some new projects, maybe find direction on things that they've already begun. 1:47 But most importantly, by the end of this thing, We're going to try and distill a blueprint that you and other builders can use to make the atmosphere work for whatever industry you care about. 2:00 To kick things off, let me tell you about the game dev feeds. 2:04 When Bluesky launched custom feeds, I was stoked. 2:08 Not only were we getting algorithmic choice, but we had community control over it. 2:13 And that was such a powerful idea. 2:16 I was hooked. 2:18 I love video games and the game development community, so I thought it sure would be cool to gather all of that content in a way that has never happened on Twitter. 2:28 So I built the Game Dev Feeds. 2:30 There's the core, the Game Dev Feed itself, which is sort of the core of the community. 2:34 It acts as a town hall for our little band. 2:38 There's also game news. 2:40 We've got jobs and careers. 2:42 I also have the Screenshot Saturday and Wishlist Wednesday feeds, which are both manifest versions of hashtag trends that game devs used to use on Twitter to share their games outside of the developer community. 2:54 At first, these were nice little community tools. 2:57 There were, you know, tens of people using them. 2:59 I was very happy with that. 3:01 And then there were hundreds, and then suddenly there were thousands. 3:05 The feeds blew up, and today I have around 32,000 daily active users across the feeds. 3:12 It's probably a drop in the bucket for Indra's news feeds, but for a much smaller community like game devs, it's pretty exciting. 3:19 So I hadn't planned for it, but the community that had been gathering slowly, suddenly they decided that Bluesky, and specifically the feeds that I'd been building, made for a pretty good place to be. 3:31 Shout out, by the way, to Great Social, who hosts all of my feeds. 3:35 They've provided me with the tools that enable the feeds to thrive the way that they do today. 3:39 If it wasn't for Devin reaching out about a year ago, I don't know if I'd be up here speaking. 3:46 A bit of time passed. 3:48 Many alfs were exchanged. 3:50 And then Bluesky launched Labelers. 3:53 I thought, as I was seeing what other people were doing with them, what if I could build a labeler that helps people in the game dev industry find other professionals and track them throughout their timelines? 4:05 Studios, publishers, journalists, media creators. 4:08 It seemed like a fun little project. 4:11 So I built it. 4:12 I stood up my own Ozone instance. 4:15 I was manually reviewing every request. 4:18 It blew up. 4:18 I was completely underwater with thousands of labeling requests coming through. 4:23 But the community found it really useful. 4:25 It helped them define and understand professional relationships and trust relationships across Blue Sky. 4:32 Today, the Labeler has over 6,000 game developers, 450 studios, nearly 500 content creators in the games space, 700 games business folks, more than 150 games journalists. 4:47 In total, like we said earlier, I've applied more than 9,000 labels to games industry professionals and businesses. 4:55 The games industry, you know, it's continued to show up on the Atmosphere in a way that I really never expected. 5:02 And I didn't know it at the time, but working on the labeler and on the feeds had me working through a lot of questions that it turns out I would need to answer pretty soon. 5:13 So things settled for a while, but that brings us to last year, 2025. 5:18 That's when the delistings happened. 5:21 A radical group put pressure on payment processors like Visa and MasterCard, who in turn put pressure on game platforms, to delist adult and not safe for work games from Steam. 5:33 It happened overnight. 5:34 There was no warning. 5:35 Game developers were waking up to find their content gone, and the appeals process was scattered at best. 5:42 Now, it doesn't matter if you play or care for adult or not safe for work games. 5:48 Some people do, and regardless, those games deserve to exist, and the developers that make them deserve to be able to express their art. 5:55 But the delistings didn't stop at adult content. 5:59 Marginalized communities like LGBT developers and developers of color also got swept up. 6:05 Many of these games had nothing to do with adult content, but they were flagged or pulled because the dragnet on the delistings was wide, and the appeals process, again, was barely functional. 6:17 A few months later, the same thing happened on itch.io. 6:21 Which has long been considered a bastion for indie games, especially the weird ones. 6:27 Developers woke up to find their games gone, revenue gone, visibility completely wiped. 6:32 And because none of that data belonged to the developers, there was little they could do. 6:38 Now, remember, I'm running the Labelers. 6:40 I'm running the largest game dev feeds on Blue Sky. 6:43 I have tens of thousands of developers in this community. 6:46 They're not strangers. 6:47 They're people that I interact with regularly, people I've built a reputation with that have put their trust in me as a steward of this community. 6:56 So they started reaching out throughout all of this. 6:59 They were posting publicly and reaching out to me privately in DMs. 7:02 And the question was, what can we do to survive this? 7:06 How can we use the atmosphere to make it so that no platform and no payment processor has the power to make a developer's work disappear. 7:16 At first I said, I don't know, good luck. 7:20 I already had a lot on my plate, but it sort of stuck in my head. 7:25 I couldn't really quit thinking about it, so I did start building. 7:30 Before I get into what I actually built— yeah, I know, I do a lot of prefaces— I want to share some things that I picked up along the way. 7:39 Throughout all this work, I've been trying to convince game developers to migrate from Twitter to Bluesky onto the Atmosphere. 7:46 And in all of those conversations, I've had to explain why Bluesky and the Atmosphere mattered. 7:52 And I had to do it to people who could not care less about the protocol. 7:56 But if the blueprint that I'm handing you today is going to work for other industries, you're going to need to do the same thing. 8:03 So one thing to keep in mind with this is that while game devs are people too, they're not average users. 8:11 They're creators. 8:12 They often use social media more like a storefront than anything else. 8:15 So it's a little bit different than talking to your average individual that's just setting up their social media. 8:20 So I'm essentially speaking to them like businesses, not individuals. 8:25 With that in mind, here's the framing that I found actually clicks. 8:30 Number 1, you choose who stewards your data. 8:34 We all know this one, right? 8:35 You get to decide who's responsible for your data. 8:38 If you sign up on Bluesky, they're the steward of your data. 8:40 If you signed up on Blacksky, you can thank Rudy and his team for keeping it safe. 8:44 And if you're deeper in the atmosphere, like a lot of us in this room, uh, maybe your data is on a Raspberry Pi in your kitchen. 8:50 I don't know. 8:51 Uh, the point is, it's your choice. 8:54 The control is in your hands, and you can choose a new steward later, even if the current one doesn't want you to leave. 9:00 Steam would never. 9:03 Number 2, your data works everywhere. 9:07 If I make a post on Bluesky, any other app that supports those posts— Blacksky, Northsky, Eurosky, and a soda and about a gajillion more— can also see that post. 9:17 Users on those platforms can like it and engage with it. 9:21 And if you choose a new steward later, you can take all of that data with you because it belongs to you, not to any single platform. 9:31 And number 3, the community is smaller. 9:34 But this is real. 9:36 On Twitter, you scream into an algorithm that prioritizes exploitation over engagement. 9:41 On Bluesky, when you post about your game, the people who've opted into the game dev feed, they're there for that content, uh, so they actually see it. 9:50 Your screenshot Saturday posts get real engagement from people who genuinely care, not bots and drive-by follows. 9:58 For anyone trying to build an audience, whether that's game devs, musicians, educators, whoever, a smaller community paying attention is worth far more than a massive one that isn't. 10:10 So that's it. 10:11 That's what's been working for me. 10:12 You own your data. 10:14 It works everywhere. 10:15 And the people here are actually engaged. 10:17 That's the spiel that's been working really well for bringing these folks over to the Atmosphere. 10:23 So I'd encourage all of you to steal it, because you'll need it when the people in your industry start asking why they should care. 10:32 OK, so back to the building. 10:34 The community was asking for help, and I had an idea of how to give it. 10:38 First question I had to answer was, what does video game data look like on the Atmosphere? 10:45 Right now, every platform has its own answer. 10:48 Steam has its own data format. 10:51 Epic has its own. 10:52 And then there's IGDB, which is worth pausing on. 10:56 Because IGDB is the de facto game database for most of the industry. 11:01 Discord pulls game data from IGDB. 11:04 So does Xbox. 11:05 So does Epic, GOG, Unity. 11:08 Dozens of major companies depend on it. 11:12 And it's owned by Twitch, which is owned by Amazon. 11:18 Fun fact: Amazon also owns IMDb. 11:22 One company controls the authoritative databases for film, television, and games. 11:30 Every one of the platforms I mentioned earlier is building on data owned by a single corporation, and the developers are whose games are in the IGDB, they're guests. 11:42 They can submit corrections, but they don't own anything. 11:46 In my conversations with game developers, I found a lot of them didn't even know that their games were on the platform, and they certainly didn't know that they could edit it. 11:56 So that's a supply chain risk for an entire industry. 12:00 And for me, sitting there with a community asking, what can we do? 12:05 It felt pretty obvious where to start. 12:09 Step 1 was defining a shared language for games on the Atmosphere. 12:15 The game lexicons cover everything about a game that you'd want in any application on the network. 12:22 The game itself— title, descriptions, genres, platforms, release dates, screenshots— its offshoots, expansions, downloadable content, updates, mods, and then of course the data around it as well. 12:33 So achievements, high scores, reviews. 12:36 And credits for contributors. 12:38 But these are— because these are lexicons on the protocol, they're not locked to any single application. 12:44 The developer publishes their game data once to their own PDS, and any application on the network that understands the lexicons can display it, search it, reference it. 12:54 Right now, a game developer updating their game's description has to do it on Steam, Epic, GOG, Humble, Itch, IGDB, their own website, their press kit, their publisher's website. 13:04 With the Atmosphere, update it once, it propagates everywhere. 13:09 Well, kind of. 13:10 We have to be honest about our expectations for adoption, right? 13:14 Those other platforms aren't going to start pulling from the Atmosphere tomorrow just because the data is there. 13:20 We build the foundation though, because we have to start somewhere. 13:24 So we build the foundation, we prove the value, and then over time we either convince platforms to adopt it Or we build something so compelling that they don't really have a choice. 13:35 That's the long game. 13:37 Now lexicons by themselves are just definitions. 13:41 You still need something to make them work on the network. 13:44 Bluesky calls it an app view. 13:47 Others call it an app server or an API or a flibbertigibbet. 13:51 There's a lot of things. 13:51 The proper name and description of an app view are hotly contested, but really It's just the layer that sits between the data on the network and the people trying to use it. 14:03 If lexicons are the shared language, the app views are the applications that speak it. 14:08 Traditionally, building an app view means writing custom software from scratch for your specific use case. 14:14 You have to handle indexing records from the network. 14:17 You have to implement queries and procedure XRPCs. 14:20 You have to know what XRPCs are. 14:22 You probably need to handle authentication, and you'll likely need additional infrastructure, databases, search indexes. 14:29 That's a significant amount of work, and it's a real barrier for anyone who wants to build on the Atmosphere. 14:34 So I built HappyView. 14:36 HappyView is a lexicon-driven app view builder. 14:40 You upload your lexicons, it dynamically generates XRPCs for querying and writing data, watches the network for new records in your lexicon collections, it indexes them, and it proxies writes back to users' PDSs. 14:54 You define the schema. 14:55 You upload it. 14:56 You have a working app view. 14:57 Auth included, out of the box, totally open source. 15:01 But Happy View isn't specific to games. 15:03 It's generic. 15:04 It's open source infrastructure. 15:05 So you can use it for whatever industry you're currently looking at. 15:09 The game lexicons are the first production use case built on Happy View, but I'm hoping they will not be the last. 15:17 Now I used Happy View and the game's lexicons to create the Pinteract app view. 15:23 Some of y'all may know it by its sort of infamous handle, games.games.games.games.games. 15:31 Uh, and of course that means that the NSID for the game lexicon is games.games.games.games.games.game. 15:41 You're welcome. 15:42 Uh, shout out to Ted Han, by the way. 15:46 For pointing me at that domain and coming up with the much simpler name I mentioned just a minute ago, the Pinteract. 15:53 The Pinteract currently holds data for more than 350,000 games across all platforms. 15:59 You can think of the Pinteract as the alternative to the IGDB API. 16:03 It's the open, decentralized version of the game data layer that the whole industry has been building on. 16:09 And it's currently live and open, so feel free to go play with it. 16:12 After I'm done. 16:14 I'll be holding links to the end because I know you rascals better than to restrain yourselves from a new toy. 16:19 Now that the data layer was in place, I needed a way for everyday users, both gamers and developers, to access that data. 16:27 So that's where Cartridge comes in. 16:30 Cartridge is the first application built on the PentaRact. 16:34 It's a games database, sure, an alternative to IGDB itself, But it's also a games discovery platform for players where I'm building features to make it easier for gamers to find the games that they love. 16:46 But more importantly, and sort of back to the mission, it's where developers can go to take control of their game's data. 16:52 Game developers aren't going to interact with raw lexicons and app views and XRPCs, and they shouldn't have to. 16:59 They're building games, not infrastructure for the atmosphere. 17:04 When a developer signs in, they get a dashboard. 17:06 From there, they can create game records, update descriptions, upload screenshots, manage credits, all the stuff that they're currently doing across 10 different platforms. 17:15 Except now they're doing it once on their terms, and the data lives in their PDS. 17:23 Okay, that actually means they're doing it 11 times now, but we talked about this long game. 17:28 Now, IGDB itself, has around 350,000 games. 17:33 That may sound familiar because that's about how many games we have on cartridge. 17:37 Actually, that's exactly the same number because IGDB is the original source of all of the data that is currently in the Pinteract. 17:44 And yes, I did check. 17:46 IGDB's API allows me to do that. 17:49 But it gets better over time as developers claim and take ownership of their games. 17:55 They enrich the data and we can build a rich graph of the data between games and other things on the network, like Blue Sky posts, PopFeed.social reviews, Leaflet in Pocket articles referencing games. 18:08 And there are technical advantages, real technical advantages to using the Pinner Act app view over IGDB. 18:13 There's no Twitch credentials necessary for one thing. 18:16 Uh, you get unauthenticated API access. 18:19 We have semantic search instead of plain text matching, and we have discovery feeds that IGDB simply doesn't offer. 18:25 But the important part isn't the feature comparison. 18:28 It's what happens when a developer actually claims their game. 18:34 They verify ownership, they claim the record, and they migrate it to their PDS. 18:38 Now it's theirs. 18:39 Fully. 18:40 Their data is managed by the steward of their choice rather than the devil— I mean Amazon. 18:45 Uh, if Cartridge disappears tomorrow, their game data would still be on the network, in their PDS accessible to any application that speaks the lexicons. 18:55 That right there, that's what data sovereignty means, right? 18:59 It's not just a buzzword, it's a technical reality. 19:01 We're doing it on the Atmosphere. 19:04 Now, Cartridge isn't officially launched. 19:07 I can't tell you about tremendous user growth or anything. 19:10 There's no press releases, no big announcements, and I still have some work to finish before I'm comfortable really opening the floodgates on the application. 19:17 So the social proof of all this may feel a little bit flimsy, but what's happened during development is what really tells me that I've been on the right track. 19:26 When I started sharing my work on the Pinner Act publicly, developers started responding with use cases I hadn't considered. 19:33 Luca from the North Sky team brought up managing a game's credits and updating them regularly. 19:39 You allow games to fetch those credits and fix developers' dead names. 19:44 I didn't consider that. 19:45 I'm a straight white guy. 19:47 But for trans developers in the game industry, that's not a niche concern, right? 19:52 That's their name. 19:53 It's important. 19:55 And right now, getting a credit updated in a shipped game is a nightmare of bureaucracy and platform policies. 20:02 With developer-controlled credits on protocol, it's an update to a record owned by that contributor. 20:09 Somebody else started talking about player-owned save data, cloud saves. 20:13 And this one's a little personal for me. 20:15 City of Heroes, some of you may remember, it was a really popular MMO, massively multiplayer online game, owned by NCSoft and developed by Paragon Studios. 20:25 It was released in 2004. 20:27 I played it way too much. 20:29 And 8 years later, in 2012, NCSoft terminated the entirety of the Paragon Studios staff shut down the game, and shut down the servers. 20:38 The time invested in these characters that people had created, that they loved, the experience that players so thoroughly enjoyed, it was all gone, seemingly forever. 20:48 Years later, a community on Reddit actually revived the game. 20:52 They had a private server, and they sort of, you know, hid it for a while, but they were able to bring it back, which is incredible, except all the original My player data is gone. 21:02 I can't load up the characters that I made and fell in love with a decade ago. 21:07 They don't exist anymore. 21:08 My inner child is sad. 21:10 So is my outer child. 21:14 On the protocol, that save data could live in the player's PDS. 21:19 The game shuts down, gets revived 5 years later, 10 years later. 21:24 Your saves are still there. 21:26 Now for competitive games, that save data could live— or sorry, you need some form of verification. 21:32 You can't let people write, "I got a perfect score," into their repos when you're trying to compare scores. 21:38 So I actually use Nick Garrakine's attestation spec already to allow you to have that information stored in the players' repos where it is still non-copyable. 21:49 It's non-forgeable. 21:50 It's verifiable. 21:54 So I'm also building game engine SDKs for Unity, Unreal, and Godot, which are all game engines, so developers can interact with all this data directly from within the tools that they already use without ever having to think about the protocol underneath. 22:08 And there's more on the horizon. 22:09 Decentralized modding, game distribution, sales. 22:12 These are big unsolved problems on the Atmosphere, but the foundational pieces for doing it, at least with games, are already on the network. 22:20 Game data, developer identity, player identity, that's all ready to go. 22:24 So the question about how to do this is, is just how. 22:27 It's not if. 22:30 Okay. 22:31 I just walked you through my entire journey. 22:33 The feeds, the labeler, the delistings, and how they drove me to create the lexicons, Happy View, and Cartridge. 22:40 Now let's talk a little bit about what other atmosphere builders can hopefully take away from it. 22:47 I took a community that I was already a part of, I found the data that causes the most pain for my community, which in my case was games scattered across platforms that developers don't control, and I built tools and infrastructure that let them own it. 23:00 That's not unique to games, though. 23:02 It's a pattern, and once you see it, it's a little difficult to unsee. 23:07 Take music, for example. 23:09 An artist releases a track, and now they're managing metadata across Spotify, Apple Music, Bandcamp, SoundCloud, YouTube, Tidal. 23:15 Every platform has its own dashboard, its own format, and its own rules. 23:20 An artist gets dropped or a distributor folds, and suddenly the musician's catalog is in limbo. 23:27 Their fans can't find their music, their revenue disappears. 23:30 It's the same pattern, but it's a different industry. 23:33 How about education? 23:34 That's a weird one, right? 23:36 Your degree, your credentials, your transcripts, They're all locked within institutional silos that may or may not still exist in 20 years. 23:44 If you don't get how this is relevant, try transferring credits between universities or getting a copy of your degree after the original was destroyed. 23:53 The pattern is always the same. 23:55 Data is published to centralized third-party platforms that we don't control in formats we don't choose, and we hope that nothing goes wrong. 24:03 When it does, We have little to no recourse. 24:07 Now here's the thing. 24:08 Everything I built for games, the lexicon, the management platform, the claim and migration system, what I've built is game-specific, but the infrastructure isn't. 24:17 Happyview, which is open source, doesn't care what your data is. 24:21 It just needs lexicons. 24:23 Now let me try and make this even a little more concrete. 24:27 Let's say you want to solve this particular piece of the puzzle for the music industry. 24:30 Right? 24:30 Step 1, what data is scattered? 24:33 Track metadata, album metadata, artist credits, licensing agreements, distribution records, right? 24:39 So step 2, we define the lexicons for those things. 24:42 What's a track? 24:43 It has a title, a duration. 24:45 It has contributors and rights holders, release dates. 24:48 What's— and it's going to point to a blob that represents the audio, right? 24:52 What's an album? 24:52 It's a collection of tracks with its own metadata, like its title. 24:55 An artist is going to have a discography, collaborators, rights relationships, and maybe that's an actor in our lexicon, right? 25:03 You design those schemas for step 3, upload them to HappyView or, you know, QuickSlice if you want to try other options, and now you've got an app view that indexes music data from the network. 25:15 Step 4, you build the interface, right? 25:18 This is the easy part, right? 25:20 Where an artist logs in, claims their catalog and is now in control of their own data. 25:24 It's the same pattern I followed for games, just different domain knowledge. 25:29 So that's it. 25:30 You do this for music, for education, you do it for healthcare records, for legal documents, for anything where centralization is failing the people who create the data. 25:39 This is one pattern. 25:40 It's not the only centralization problem out there, but your domain might have a slightly different one. 25:46 Locked up credentials, gated access, licensing nightmares. 25:48 The pattern might look Slightly different, but the tooling that I've shown you today, it should be generic enough to handle pretty much all of them. 25:57 The hardest part is figuring out what data in your domain actually needs to be decentralized and then designing lexicons that accurately represent it. 26:06 That's domain expertise, not protocol expertise, and that's why it really has to come from people who know their industries, not from people outside guessing at what a community needs. 26:17 One more thing. 26:18 I've been talking about the protocol a lot because y'all are the people building on it, right? 26:24 But when I'm talking to a game developer about Cartridge or about Blue Sky, none of this comes up. 26:31 They don't know what an app view is. 26:33 Like I said earlier, they don't care about XRPCs. 26:36 They just want to manage their game data and not get screwed by big corporations. 26:41 The protocol is invisible in the product and in the conversations. 26:46 That's not an accident. 26:47 That's the whole point. 26:49 Everything I've built treats @protocol as an implementation detail, not the selling point. 26:56 It's not why a game developer would use Cartridge. 26:58 They'd use it because it solves their problems. 27:01 The protocol is just how it works under the hood. 27:04 So when you go to build for your industry, That's the mindset that's going to get people to use what you make. 27:11 Don't sell the protocol, solve the problem. 27:14 The protocol is what makes your solution possible, but it's never the pitch. 27:18 Thank you. Speaker B 27:28 That was fantastic, Trezzy. 27:30 I hope a bunch of people are lighting up their scrapers and attacking so many other industries. 27:36 The fire safety talk was one that would pair really nicely with yours because they had the same pain. Trezy Who 27:43 Thank you. 27:44 Yep.