Building the Decentralized Social Web: From Collective Social to OpenSocial ![rw-book-cover](https://wsrv.nl/?url=https%3A%2F%2Fd3t3ozftmdmh3i.cloudfront.net%2Fstaging%2Fpodcast_uploaded_nologo%2F43243169%2F43243169-1768937223209-5abb8f753e08.jpg&w=512&h=512) šŸ”— **Source:** [Building the Decentralized Social Web: From Collective Social to OpenSocial](https://share.snipd.com/episode/6a7a43d1-28bd-4b72-912c-f262003840cf) ## Highlights - **Building Collective Social To Scratch A Personal Itch** - Brittany started Collective Social to host an async book club and to learn the AT Protocol by building a Goodreads-style app. - The work led her to create OpenSocial to share group functionality across AT Protocol apps like book clubs and event aggregation. Transcript: Brittany Ellich Yeah. So I started these, I think I'll link it in the show notes, but I started learning about at protocol when we had Nick Jerkinis on the podcast last year. And I know you and I have talked a lot about like, gosh, the problems in social media and like have vibed over a lot of like books and stuff in the past. And it really sparked my interest a lot. I'm like, wow, this, okay, this is like something that's very interesting and could make a huge difference for the world if like social media was less terrible. So I knew that I wanted to build something on AppProto just to like learn more about it. And then I started with, yeah, that Goodreads style app, Collective Social. But what I really wanted is something that we could use for our book club to like organize things like, you know, have like an async spot to collect comments or to share updates and have Like a schedule of when we're supposed to read and by when. So that's sort of what I started with. And it kind of led me down this rabbit hole where I realized like there's no real concept yet of like a group or a community shared identity within within the at proto space. I know that there's some folks looking at that now, but yeah, so that's sort of what led me down this big rabbit hole of building open social, which is actually a little bit further along Than collective social is. And we have a few use cases for it, like for the atmosphere.community website, which is where all of like the different local app proto groups are through. People can now like share their blog posts or events or something like that. And they all aggregate under one group identity, which has been really interesting, I think, to work on and build. Yeah, it's been really fun. ([TimeĀ 0:03:39](https://share.snipd.com/snip/55b77b6c-d5ba-40ec-b00e-f282b314670f)) - **User Owned Repos Shift Incentives Away From Lock In** - AT Protocol gives each user a personal repository (a PDS) so users own their posts, follows, and likes instead of a centralized app controlling data. - That portability means apps must compete on features rather than lock-in, enabling easier migration between clients like Blue Sky clones. Transcript: Brittany Ellich Yeah. So in the traditional or shall I say legacy social media space, the idea is they build this app that brings a lot of folks in. And you think like Facebook or MySpace or Twitter or whatever, you know, typically it starts out like just gaining a lot of people and a lot of excitement. And then further down the road, once people are on the platform and using it regularly, they start doing things like, how do we keep people on the platform more frequently? And like, how do we like keep users engaged so they keep scrolling and like see more ads and then start, you know, monetizing through ads? And a lot of the reason that that works in those social media apps is because it's really hard to leave, really hard. So, I mean, I still know people that like use Facebook because like it's just so hard to leave because they have that one group there or like that one group chat or, you know, you don't want To lose track of your family or something like that. Just really hard to like migrate off and all of your content and photos and everything like that are kind of trapped in that company's database um so the one major thing difference with At proto was actually created i think by folks that worked at twitter before Twitter became what it is now. It's designed so that you as the user own your own repository. And so all of your likes and follows and data posts, everything lives in your repository. ([TimeĀ 0:05:37](https://share.snipd.com/snip/e65aceb9-c36a-4ad2-a435-fe58f8ae9499)) - **How Building Made The Protocol Make Sense** - Working hands-on was what made AT Protocol vocabulary click for Brittany; concepts are essentially JSON collections in user repos rendered by any app view. - Lexicons are simple JSON schemas apps publish so other app views can reliably render records across the ecosystem. Transcript: Brittany Ellich Of made this vocabulary click for you? I think a lot of what made it really click for me was building on AppRoto, like building the apps that I was working on. Like I said, I was kind of stuck where I was like, okay, well, I want some of this data to be like aggregated under a group. And there's no concept of like a group identity yet within the AppRoto space. And there were a few different ideas out there from folks about like how to go about doing that and so I just sort of looked through some of the ideas that folks had and then I worked with Nick Jerichinus too since we worked together at GitHub and like picked his brain about like how would we do this and so we came up with this idea of like creating group dids which is the Decentralized identity so for example like I have a did which is just like an ID, that's represented throughout the entire decentralized web of some like gibberish ID name. And then I have a handle, which uses they use DNS to resolve to the handle. So I have Brittany Ellick dot com is like my handle and that represents my identity. But we also have one now called overcommitted dot dev. And I made that into a group did. So anybody can go on to OpenSocial and like become a member of that group. And then that means that they can like post content, post their own content and link it to that group. So we can show like what's everybody in the group up to at one time. And so there's a way to like basically connect those together. So I think building that and like trying to reason through that problem helped a lot and helped me realize like, all right, like a lot of this is really pretty, not crazy complicated. I think that's the beauty of it is it's like actually very simple at the end. Like it's just like a folder full of JSON. ([TimeĀ 0:09:20](https://share.snipd.com/snip/361b3868-ea8d-461e-ac6c-56afe19658b9)) - **Use Public Deadlines To Finish Side Projects** - Use external deadlines and public commitments to force progress on side projects when life gets busy. - Brittany finished OpenSocial because she submitted a talk and didn't want to be unprepared for Atmosphere Conf. Transcript: Brittany Ellich So I actually wrote up a doc about what I was planning to do a blog post which I can link it was like back in November I think where I was like figuring all of this out and then I didn't touch The project got it started and then I didn't touch the project again until like January and the thing that really motivated me to continue is I had submitted a talk proposal to Atmosphere Conf about it. And like I was primarily motivated by like the shame of not wanting to be unprepared and like not have a thing to show for the talk. So that's what got me to actually like finish it. So I'm highly motivated by being publicly shamed, I guess. It ended up going well. I enjoyed the talk. It was fun. And the talk was all about like how open social works. Erika Eggemeyer I get that. I mean, this is part of why I do book clubs is because that accountability is so important to me finishing things. Yeah. So there's no fame in using public shame. Brittany Ellich Yeah. Yeah. I'm glad my talk was chosen because I absolutely would not have finished it otherwise. Because I was very close to being like, oh, do I really want to finish this? And I did. And I was glad for doing it. And I'm glad that I went to the conference, too. It was a super good conference. So, yeah. ([TimeĀ 0:16:44](https://share.snipd.com/snip/79b3531c-e150-460f-be3b-72b8c7f6bec9)) - **Leverage AI To Ship Side Projects Faster** - Use AI tools like GitHub Copilot to reduce friction and accelerate side-project work when time is limited. - Brittany credits modern AI assistance with helping her ship OpenSocial while balancing a job and three kids. Transcript: Brittany Ellich That like ai tools got like really good and so the fact that i have like a finished side project and it's not just something that's like languishing in my to-do list forever is really due To the fact that like i had something like that to help like smooth out the friction points because otherwise like it's actually it was a lot of work to build having co-pilot available To help me like whenever I got stuck was phenomenal it made it go a lot faster than I think and like finish because I wouldn't have either way otherwise like I've got a job and three kids And other things going on. ([TimeĀ 0:18:10](https://share.snipd.com/snip/22588b83-9acb-4191-bf0a-a457cf0402d3)) - **Schedule Small Regular Blocks To Ship Projects** - Carve small, regular time blocks like team well-being days or evenings to progress side projects alongside full-time work and family. - Brittany used monthly well-being days and 8–10 p.m. creative hours to build OpenSocial around work and kids. Transcript: Brittany Ellich My current team has what's called well-being days for like once a month. We get like a Friday off. Today is actually one of those days for me. And so, yeah, I used those days pretty heavily. After bedtime that's like my turns out that's also like the time between you know 8 to 10 p.m that's when I'm like most feeling most creative and ready to do something anyway so like that Has worked out really well so I use that time a lot for building things okay so ([TimeĀ 0:18:52](https://share.snipd.com/snip/606af943-ac98-457c-b191-298de422f3a5)) - **Group Identity Implemented As A Two Way Handshake** - OpenSocial models groups as their own DID and handle so a group has a repo and members can link personal posts to the group's repo via a two-way handshake. - That handshake lets admins remove group links without deleting personal content and preserves ownership and portability. Transcript: Brittany Ellich Hasn't been like a built-in thing to the protocol yet to say like this is a group shared identity. I think people could get around that up until now by just like sharing a password to, you know, the same login essentially. And then you could like, you know, we could both post under overcommitted, you know, blue sky post or whatever, because we share a password. But thinking about that, I didn't feel like that would work at like an actual scale. You know, if I wanted to share this with everybody, I don't want to share the password to that to every single member of, you know, a community that I'm a part of or anything like that. So so since that didn't really exist, and since there was already like a lot of, you know, thinking out there, like how would, how would a shared thing work? One of the things I was looking at is like, you know, within collective social, we could just say like, all right, the group lives in a repo within, or like within the database of that collective Social app. But I really wanted something that was like on protocol so that it could be used across different apps. And in the future, it could be like extended and you could say like, all right, like we can, you know, have like the shared resource to post to blue sky or the shared resource to like work Under in Untangled or something like that. And so that is sort of where I realized like we needed some sort of an identity to like handle allowing those permissions to post under a shared resource without sharing a password with Everybody. And so the group identity themselves is just another did and handle. Like it has a group repo and a group did and a group handle, but they're like identical to an individual one. You could take any existing, you know, shared account out there in Blue Sky or whatever and turn it into this group resource so that you could use it across multiple apps. But then the way that things are shared from the actual group is there is instead of like the, you know, if you make a post and you're like, I want this post to be shared from the group's page Or something like that. Instead of instead of posting under that group's identity, you can post under your own. So you can say, hey, I'm the author of this post and then create this like two way handshake where there's a record within the group's repo that points to the record in your repo. So there's like this two way handshake basically between the content that you post and the group identifying like yes this is this is something that i'm going to show as part of our what's In our repo and that serves a couple purposes so like the group admins if you know somebody's being really spammy and they don't want to have all of these resources from someone they can Just delete the the handshake on the group side and then like the same thing, like the user can just change their own post if they're like, I don't want this shared with the group anymore. They don't have to necessarily like change anything in the group. They can just change it within their own repository and like break that connection. Erika Eggemeyer Yeah, that makes a lot of sense because yeah, that recognizes like still that personal ownership of like maybe you get even kicked out of the group or something like that but even if that Happens if I delete it on my own repository then it's still gone or like the handshake mentioning like there's this contract between you as the group and a group member and you can post It on your personal but if the group doesn't want it then like you're not it won't be recognized as records yeah it's not like it's going to delete the record or anything like that you still Brittany Ellich Own the thing that you the content that you created and you can still take that to other groups if you wanted if the one that you were in like just didn't vibe with it or whatever but yeah So that's sort of the goal. And that was what that's working now, actually, which is exciting. ([TimeĀ 0:19:50](https://share.snipd.com/snip/85efc3c5-4466-4a41-80f1-f50f6c63517e)) - **Community Feedback Doesn't Replace Implementation** - Building open specs in a community-led space means lots of feedback but limited ownership; individuals must implement proposals to make them real. - Brittany found the AT Protocol community generous with ideas and critique but realized execution still falls on contributors. Transcript: Brittany Ellich Yeah, I think it probably depends on the space that you're in. Like, I don't know if every, this is my first time really getting into any open source space in any like longer term capacity, but everybody's been super open and like willing to like Rubber duck on things. I've had folks like give me feedback like, hey, you should make your OA setup better. Like, look at this instead. I'm like, gosh, I love that feedback. That's great. Because that's just not my area of expertise in any, by any means. So I think it's, it's a really like friendly and sharing community where like, that's just sort of normal within the community. There's a lot of spaces like there's a discourse where anybody can go in and you know propose a topic and lots of folks will weigh in with their opinions if that's what you want. I think sometimes that can also be kind of I think there's also like some drawbacks to that too in some ways where like it can be really easy to be like yes things should be done this way And Then there's no real like ownership of like this person's going to go take that idea and do the thing with it. Like it's really on the individual to go make it actually happen. But I feel like everybody's been pretty like generous with their thoughts and time. ([TimeĀ 0:24:28](https://share.snipd.com/snip/62a65b2d-9466-41a5-8bb0-122293e0e10c)) - **Interoperability Encourages Cooperative App Design** - AT Protocol apps trend toward cooperation: multiple blog and content apps coordinated a shared lexicon so content interop works across clients. - Developers focus on competing UX and features rather than owning user data to attract users. Transcript: Brittany Ellich And like, this is not even like the first Goodreads type app. It's like the 10th or 20th at this point. And it seems like there's been a lot of collaboration and like design towards interoperability. Like there's a couple of different blog sites, for example, like there's Leaflet is one and Offprint, that's the one that I use. And then there's Pocket.blog. And all three of those came together and like designed a lexicon together so that like their content could easily be shared between the apps. So I think it's sort of a different way of thinking about building apps where it's not necessarily like, hey, I'm going to build this app and I need all the market share. You know, like people are going to go to the one that, you know, fits their style more or like has the features that they need. And everybody's like pretty friendly and willing to work with each other to be like, all right, let's make the best experience possible. ([TimeĀ 0:26:08](https://share.snipd.com/snip/3af9118c-37e4-42a9-b0e7-8fb99a6ed89f)) - **Permissioned Data Is The Next Big Step** - Permissioned (private) data is a major missing pillar for broader AT Protocol adoption; once available it unlocks identity-sensitive apps. - Brittany views upcoming permissioned data support as critical for expanding into non-public app types. Transcript: Brittany Ellich One of the things that they're working on this year is permissioned data. Right now, everything is just public in your repo, which like having something that is private or only available within specific settings is super important for like identity in general. I think once that comes out, that's like one of the biggest pillar. Like we need this to like really be able to expand into lots of different types of apps. But that's pretty close to landing. ([TimeĀ 0:27:50](https://share.snipd.com/snip/e678fe27-7260-4780-a297-4b6c02db30ac)) - **Decentralize Registries But Solve Trust Problems** - Decentralizing package registries like npm is attractive to avoid single-point failures and unilateral removals, but raises security and authenticity challenges. - Brittany and Erika agreed decentralizing npm scores high in value but needs solutions for rogue servers and package trust. Transcript: Erika Eggemeyer So NPM, for those who are not familiar, is a centralized package repository, and anyone can publish a package and then developers download it and anyone can also unpublish a package So there was a package called left pad um that was unpublished and caused a bunch of issues for people who relied on that what do you think bernie one to ten um realizing npf oh that's a good Brittany Ellich One so we're talking specifically about the registry correct like where all of the packages live okay um i was gonna say because npmx is came out recently and that is a really cool i'm Gonna include a link to that um yeah but that's more just a website to view them um to view the packages i think for the registry i feel like that makes a lot of sense to be decentralized yeah I feel like that's a 10 right i think it might be hard to handle security there i guess if like anybody can publish a package like how would you deal with like rogue servers that are like Trying to say like this is this package and it's not but i feel like it's it's already that true there's very little oversight yeah yeah yeah I'm with you I I think decentralizing makes Erika Eggemeyer Sense like once you publish it it's kind of like out there I mean I guess app prod is a little different we're like you you as the publisher would still own whatever you publish. So this would be a little different treatment of saying, like, once you publish it, it's, like, persistent and, like, you can't take it down. Yeah. But I think, I mean, a lot of those things are associated with, like, commit shahs and if it's open source. ([TimeĀ 0:29:17](https://share.snipd.com/snip/483d6cd3-7bbe-4687-b64d-b9ef744e85ab)) - **Personal Medical Records Would Help But Need Privacy** - Portable personal medical records could greatly improve care continuity but demand strong privacy controls and legal workarounds around HIPAA-like rules. - Brittany sees major human benefits for long-term care and mobility but recognizes security and compliance complexity. Transcript: Brittany Ellich Feel like this would solve a lot of problems in like our health care system and also would be an absolute nightmare to any lawyer that is like familiar with like hipaa things like the security Around it would have to be very very advanced but i feel like this would be really helpful for like individuals, especially, I mean, you don't live in the same place all the time. You're not going to go to the same doctor forever. So having those be portable and in like, at least in a format that's shared across multiple doctor's offices would, I think, I think that would help a lot of humans, not just like, I feel Like it would make the systems run faster and smoother. And then the problem is, is then you have to like rely on the individual to like maintain their own repository. Erika Eggemeyer Right. Because I mean, it could. I guess like, yeah, I'm with you. Like definitely this like privacy question would need to be figured out. Like you wouldn't want your medical records all public for everyone but like medical records currently are so opaque to me that I have no idea where my medical and like I don't necessarily Want to like maintain or manage them but like it would be nice for me to be like well if I did theoretically want to look up my medical history like how would I do that you know it's like it Brittany Ellich Feels like it's so protected that like even I don't know even though they're theoretically my records like who do I ask to see that like where are they yeah it's a nightmare especially For folks who have like you know need long-term care or something like that and they need all of their records in one place to like especially if they move from one place to another it's Just a huge effort to gather all the data into one thing so i think that that would be super beneficial and worthwhile but and then you'd also be like when was my last checkup i don't know Erika Eggemeyer I i think it was last year sometime it's just like there yeah yeah that's true yeah yeah so yeah i I think with the caveat that this would have more privacy controls, I'm here for decentralized Medical records and specifically personal ownership. ([TimeĀ 0:33:04](https://share.snipd.com/snip/63d2712a-0810-4adc-8242-1d544e686816)) - **Banking Is The Hardest Decentralization Problem** - Banking and real-time financial consensus are the hardest areas to decentralize due to timing, regulatory and consensus requirements. - Both hosts rated decentralizing banking lower, though personal financial record portability remains useful for tracking accounts. Transcript: Erika Eggemeyer Banking. Brittany Ellich Ooh, I think this is interesting. There's a book that I read earlier this year by Tim Berners-Lee where he was talking about something called solid protocol i'll share the book but it is is really similar to at proto but They started private first so like all of the repos are private but then like you can you to theoretically own your repository and you can bring that everywhere those seem like the main Differences from the way that it's described and i think that they actually started that at banks and like really large, like enterprise organizations in Europe, which is where he's At. So there's probably different like regulatory environments or something like that, where that is a little bit more favorable or worthwhile. But I think that there are some financial records that that would make a lot of sense. I'm curious how that would work for things like, don't know taxes or whatever like you would need to share something with the irs to be like hey here's how much i'm getting paid still or Something i don't know that sounds really complicated and like the least interesting of these different examples so i'll put that one as like a five or something i'm sure that there Erika Eggemeyer Are reasons to yeah post it in some decentralized system but it sounds hard this one this one I feel like I'm like a one on because of the timing and consensus because I think like that's When decentralization gets really tricky and like I just don't think there's any good solution like as of now like I've never heard any really good reliable solution for like decentralized Consistency and consent like consensus with these sort of like real-time transactions there's blockchain people out there saying hey we've been talking about this yeah so my my feelings Still stand um yeah that that's where i'm at i think this is like one of the one of the things that yeah i think i'm i think i'm a woman on yeah with with like i think there's room for like maybe Having a personal repository of like all my financial like things or whatever in in a repository This is my, this is like the link to my savings account and my 401k. And like, cause that stuff does get confusing to keep track of, of like, I have a 401k or like, I have a retirement account from when I was a teacher. And I'm like, I don't even know how to access that. Like I've tried to contact the school district and they like never responded. So like, I don't know if that money is just gonna like disappear like i don't know um but like that would be cool to like be able to keep track of all those different those different like Financial pieces i guess but yeah as far as banking goes the whole system i'm i'm fine to leave it as it is yeah yeah that's not not a thing that i would be like super excited to work on i don't Brittany Ellich Think but know, I'm sure there's I'm sure there's a case to be made for it. ([TimeĀ 0:35:16](https://share.snipd.com/snip/016020f3-beca-4154-aedc-8e526d6583b4))