How Bluesky Was Influenced by Scuttlebutt, a P2P Protocol

When Bluesky passed 20 million users earlier this week, CEO Jay Graber listed twenty facts about the Twitter/X competitor. One of them was this: “Paul, Bluesky’s CTO, built the earliest version of the Bluesky app, and the first client for Secure Scuttlebutt (SSB), a peer-to-peer networking protocol.”

Bluesky’s CTO is Paul Frazee and it’s interesting to look back on his journey as a P2P-obsessed developer, which eventually led to him playing a key role in Bluesky.

I last looked at Scuttlebutt back in July 2021, when Bluesky was just a research project within Twitter (Graber herself was only hired in August of that year). At this point, Scuttlebutt was a decentralized P2P protocol that a handful of obscure social media apps had built on top of: Manyverse, Patchwork and Planetary were a few that I tried. So Scuttlebutt was a niche part of the decentralized web, at best — Mastodon and ActivityPub were seen as the up-and-coming projects of 2021 in this domain.

But I found the design of Scuttlebutt intriguing, because it was based on content being self-hosted and only periodically sent over a peer-to-peer network. Scuttlebutt’s creator was Dominic Tarr, a New Zealander who lived on a boat and had sporadic internet coverage. So the architecture he chose — writing content on your device and only syncing it with the network whenever you connect to Wifi — was inspired by his own lifestyle. (Interestingly, this architecture matched my first blogging tool, Radio Userland in 2002, where the content lived on your computer and you synced it as needed to the network. So perhaps that’s why I liked Scuttlebutt.)

From Scuttlebutt to Beaker to Bluesky

Tarr created the Scuttlebutt project around 2014 and Paul Frazee was the first external developer to join him. In a recent Hacker News post, Frazee said that he built Patchwork, the first Scuttlebutt client (one of the apps I used to research my 2021 article). But it seems that Frazee grew disenchanted with Scuttlebutt because it wasn’t scaling from an end-user perspective. “It’s an aggressively anarchist technical model,” he noted about the protocol, but he had “serious concerns” about its ability to attract users.

Frazee took the lessons of Scuttlebutt and applied them to a new project he started called Beaker Browser, which was a P2P browser built on Chromium. I’ve actually written about this too, back in February 2018 when I ran an indie blog called Blocksplain (focused on blockchain technology). Beaker itself didn’t use blockchain, but it was clearly inspired by Scuttlebutt. Here’s how I described Beaker in early 2018:

“Beaker went ahead and developed its p2p browser without any blockchain. So where is the data stored, if not on a decentralized, distributed database? It’s stored locally, on the computers of users, and shared when requested using a p2p network.”

So, once again, you store your content on your device and it syncs with the network as needed — a local-first approach. The Beaker web browser had some other neat features. For example, you could “fork” a website and create your own version of that site. But overall, it seemed a bit too technical for the average internet user even when I reviewed it.

Long story short, Beaker Browser didn’t gain any traction either, so Frazee joined Bluesky in early 2022 and moved his attention to that project. But he wrote a postmortem of Beaker in December 2022, in which he admitted that the product was too technical. He also noted that “the heart of Beaker continues with Bluesky.” What I found most interesting, though, was his explanation for why Bluesky did not stick with the “local first” philosophy of both Scuttlebutt and Beaker. The lesson he gleaned was:

“Simplify aggressively. With Bluesky, we’ve opted for using p2p structures (IPLD) on a federated network, giving us some of the key advantages of p2p, like account portability, while retaining the performance and reliability advantages of servers. The pure p2p tech out there still has a lot of potential, but I think it’s a bad fit for large-scale social networks and sticking with it for Bluesky would’ve been a mistake.”
(Emphasis mine)

So rather than pursue the vision of a social media network where content is stored on a user’s device, with Bluesky you’re using servers — the cloud. Similar to Twitter, you could say, except it’s a federated server approach rather than centralized. Bluesky also chose to give users ownership of their identity and content; or at least the means to easily move it to another server, if they don’t like what the main server — owned by Bluesky the company — is doing. (Incidentally, the same month I reviewed Scuttlebutt, I also reviewed Decentralized Identifiers (DIDs), which became a core identity piece in Bluesky’s architecture.)

This shift in emphasis from local-first to a federated server approach makes sense, especially if users are still able to control their identity and content — an approach I actually prefer over the ActivityPub model of federated servers, where the person in charge of the server holds the keys to their users’ content.

I should add that Frazee credits Bluesky CEO Jay Graber with coming up with the hybrid P2P/federation structure for Bluesky. He wrote in his blog earlier this year:

“When Jay first contacted me, it was to discuss a hybrid of peer-to-peer and federation. This was the premise that she had pitched the company around, believing device-hosted networking to be infeasible but seeing potential in the data structures that the projects had been using.”

So We’re Back to the Social Media Rat Race?

Where I wonder if we’ve lost some of the magic of Scuttlebutt is with Bluesky’s approach to scaling, which isn’t really that different from Twitter’s. Here’s how I concluded my 2021 post about Scuttlebutt, which summarizes its appeal:

“One thing Scuttlebutt is not: a potential usurper of Facebook or Twitter. It’s never going to do that because the point of using Scuttlebutt (and its apps, like Manyverse) is to escape the rat race of mainstream social media and find a quieter, less pressured place to connect to other people. Scuttlebutt is by definition not always-on, like Twitter or Facebook. It’s sometimes-on, to give you time to enjoy life off the grid — on a boat, or otherwise.”

Bluesky is always on (provided its servers can continue to cope with the strain of 1 million new users per day!). Bluesky wants to be a mainstream social media service and replace Twitter/X for as many people as possible. Graber has been explicit about this. In a recent skeet, she wrote:

“Bluesky is a PBC (public benefit corporation) with the mission “to develop and drive large-scale adoption of technologies for open and decentralized public conversation.” From the start, our goal has been to build a decentralized social network that is usable by the mainstream.”

Nothing wrong with that, either! Certainly, I want to see a decentralized approach to social media usurp what Twitter/X has become under Elon Musk. That said, a part of me feels it’s a shame to leave behind the charms of Scuttlebutt’s quieter approach to social media.

Then again, what do I know — Scuttlebutt creator Dominic Tarr is now on Bluesky, along with André Staltz (the creator of Manyverse). Staltz stopped working on Manyverse in April this year, but he seems happy that Bluesky is moving the vision of a decentralized social media forward. “I’m very proud these folks are succeeding to make decentralized open source social nets mainstream,” he skeeted.

Time will tell whether we’ll once again crave an escape from the social media rat race, but for now many of us are excited about the rise of decentralized social media — whether it’s by using Bluesky or Mastodon, or both.

Originally published at The New Stack: https://thenewstack.io/how-bluesky-was-influenced-by-scuttlebutt-a-p2p-protocol/