I did an interview with Jay Graber back in 2019 at the Decentralized Web (DWeb) Camp a few years before she officially became the Bluesky CEO, which is currently the new hot social media platform that's a Twitter clone built on top a decentralized architecture using the Authenticated Transfer (AT) Protocol. At the time, Graber was working on a decentralized events application called Happening, but she had a lot aspirations to build out a fully-fledged decentralized social media platform. The challenge was the bootstrapping problem of getting a critical mass of users, which was a bit too much to overcome with all of the friction involved in making the onboarding and overall user experience of DWeb systems easy and intuitive. At the time, her solution was to start with a centralized onboarding process, and eventually integrate more decentralized aspects over time, which happens to be a very similar strategy to what Bluesky is currently doing.



I wanted to dig up this July 19, 2019 interview with Graber from my archives of unpublished interviews because she articulates a lot of the deeper philosophical inspirations and motivations she had back in 2019 that are being carried to their logical conclusions at her current position as CEO of Bluesky. The Facebook Cambridge Analytica scandal from March 2018 was a turning point for Graber that motivated her to start to working towards decentralized social networks that have better privacy protections and stronger agency over data, but also to work on peer-to-peer models that can be more resilient to censorship and natural disasters where our supercomputer phones are often bricked without working cell service or Internet connectivity.











In this conversation, we were able to talk abstractly about the tradeoffs between centralization versus decentralization, and how a hybrid approach is needed in order to successfully launch a DWeb applications. At the DWeb Summit 2018, I had a chance to interview one of the co-inventors of the Internet Vint Cerf, who was very bearish about the potentials of a completely decentralized web for a variety of reasons. The biggest one was probably how the economies of scale from centralized systems are just so much more efficient, more reliable, less complicated, and cheaper. Graber acknowledges some of the benefits of centralization include drastic protocol upgrades are easier when things are centralized, you can move faster in development, and it's easier to streamline the overall UX. Looking forward to the present day for how Bluesky is rolling out in a mostly centralized fashion, then all of these benefits seem to be holding true as they're still creating a critical mass of engagement.



But the real goal of decentralization is robust federation, which has the opportunity to be more resilient to single points of failure. There are challenges of avoiding the service from becoming corrupted, but there are also many opportunities to separate the data layer from the application layer as well as to create a robust and diverse number of ways that folks could interact within their ecosystem of people and content.



Fast forward to today, and many of Graber's predictions as coming to pass as there is a ton of innovation happening with developers playing with Bluesky's AT Protocol API, which is a throwback to the early days of Twitter when access to the API was more readily available for developers who helped bring lots of innovation to the social media ecosystem. It's been a really exciting time to see a similar level of excitement and enthusiasm from independent developers who are developing apps ranging from a real-time Bluesky firehose chat streaming visualizations with Firesky.tv or Skypulse, customized clients, an assortment of chatbots, and an interactive social network graph analysis tool.



The AT Protocol also has the potential to solve some of the more intractable challenges of verifiable identity using some of the Decentralized Identifi...