The concept and architectural outlook of the decentralized social media protocol Bluesky by Twitter

Author: Jay Graber, Bluesky CEO

Translation: Rhythm Blockbeats, Richard Lee

Editor: Richard Lee, Chain Catcher

Twitter announced the launch of a decentralized social media protocol—the Bluesky project—at the end of 2019. This month, Bluesky announced the establishment of a nonprofit LLC entity and briefly disclosed its technical architecture: combining "self-certifying protocols" with a federated delegated custody mechanism.

Bluesky CEO Jay Graber recently published an article on his personal blog about Web 3 and decentralized social network protocols, explaining the concept of "self-certifying protocols" in detail. Below are Bluesky's project disclosure announcement and Jay Graber's article:

We envision an open social media ecosystem—where developers have more opportunities to build and innovate, and users have more choices and control over the services they use and their experiences.

We see that the development and adoption of decentralized protocols is a path to establishing a strong technical foundation for achieving the above goals. Decentralization is a structural change that, by itself, is insufficient to create a healthy social media ecosystem. However, by creating an environment where developers can build freely, communities can self-govern, and users can easily switch services, decentralization can foster the innovations needed to improve public discourse.

Currently, many existing decentralized social networks can be categorized into two types: federated architecture and P2P (peer-to-peer) architecture.

Bluesky's approach is to combine the strengths of both architectures, integrating the portability of "self-certifying protocols" with a user-friendly experience under delegated custody mechanisms—so that users do not have to run their own infrastructure, while developers can build high-performance applications.

Content moderation is an important component of any online social forum, so we are committed to building a transparent, opt-in, and multi-layered reputation and moderation system. We will actively develop related tools and provide a development framework to facilitate others interested in building similar tools.

We are developing on the basis of existing protocols and technologies but are not committed to any complete stack. We see use cases for blockchain, but Bluesky is not a blockchain network, and we believe the adoption of social network protocols should be independent of any specific blockchain.

Bluesky's current focus is on building and releasing a prototype to illustrate its architecture.

What is a "self-certifying protocol"?

Recently, there has been much discussion about what Web3 really is, and my definition is: Web3 is the rights generated by users and realized through self-certifying network protocols. It is a collection of technologies that includes blockchain technology but is not limited to it. Do others see "Web3" this way? Perhaps not, but please hear me out.

The rights on the web determine who ultimately has control over the content. In the early days of the web, there was not much distinction between "users" and host operators, meaning that the host of a website was often also the publisher of its content.

As the web evolved into what is called "Web 2.0," various websites became popular, allowing users to create accounts and generate content. However, in Web 2.0, the ultimate decision-making power still rests with the website hosts, who can unilaterally modify any content—this is essentially how the current web operates.

In the transition to Web3, users can prove their identity through crypto means and publish verifiable content, and the website they are on no longer has the authority to modify that content, because the root of trust exists in the data itself, rather than in its location.

Let’s review the entire development trajectory of the web:

Web 1.0—Hosts produce content, hosts generate rights. Those who want to publish content on the web must use their own servers to host websites, which are mostly read-only and lack interfaces for users to create content or interact.

Web 2.0—Users produce content, hosts generate rights. Users can create personal accounts on websites, so they no longer need to host their own servers to publish content. Meanwhile, an increasing amount of user-generated interactive content emerges. However, these websites gradually evolve into powerful platforms that still maintain absolute control over user accounts and published content, leading to many issues.

Web 3.0—Users produce content, users generate rights. In this new model, people can publish content without hosting servers or creating accounts in someone else's database. While servers can choose whether to host someone's account or content, they do not have ultimate rights over it. "Self-certifying protocols" can achieve this, a technology built on crypto signatures and hashes.

In short, these three stages can be described as "hosted network, posted network, and signed network."

So, what is a "self-certifying protocol"?

This is a general term I use to describe protocols that have crypto user identifiers and content-addressed data.

"Crypto user identifiers" can link users to public keys, allowing users to sign with the corresponding private key, which serves as the root of trust proving the user's control over their account, rather than being a database entry recording user logins.

"Content-addressed data" means that content can be referenced by its crypto hash, which serves as a unique digital "fingerprint" for each piece of data.

Using these two technologies, the content hash signed by a user's key can directly prove the user's authorization of that content, eliminating the need for intermediaries to verify it. Data that can self-certify allows trust to reside in the data itself, rather than its location, enabling programs to move away from client-server architectures and making "user-generated rights" possible.

If Web3 is self-certifying protocols, then what role does blockchain play in Web3?

Blockchain is a self-certifying protocol that can create consensus on a global state, simulating a centralized database without being controlled by any party. "User accounts" on the blockchain are crypto key pairs used to sign transactions, while "content" consists of transaction hashes bundled in blocks, which also have their own hashes and are linked together. Bitcoin (the first blockchain) introduced a new consensus mechanism that allows mutually distrustful parties to reach agreement during transactions—this is crucial for digital currencies without intermediaries.

What protocols have self-certifying characteristics but are not part of blockchain?

Git, PGP, BitTorrent, and Tahoe-LAFs all fall into this category, having existed before Bitcoin.

Among the current non-blockchain self-certifying protocols, IPFS, Hypercore, SSB, Pergos, and Spritely all feature user keys and content addressing capabilities. In these protocols, if you can prove that a user "owns" a piece of content, you can also assert that this user is the publisher of that content.

The emergence of blockchain further clarifies the concept of ownership, establishing a globally consensus-backed timestamp ledger, confirming properties like global order and uniqueness, allowing us to prove that only one person "holds" a Bitcoin or an NFT at any given time.

From an architectural perspective, blockchain can be seen as a type of data storage in Web3, which can be quite useful for some applications, but may be cumbersome for others. From a financial and social perspective, blockchain has shifted attention and resources to this field. The important Web3 infrastructure established by blockchain includes wallets and applications that can distribute key pairs to millions of users, as well as new crypto primitive tools like zero-knowledge proofs that bring infinite possibilities to Web3.

Although discussions about Web3 are currently hotter than any other topic, we still need to develop other types of self-certifying protocols to realize the full vision of Web3.

If a self-certifying protocol allows users to directly prove the identity of content authors using keys and content addressing capabilities, then a self-certifying network protocol also possesses content linking and discovery functions. The web we use today relies on content discovery algorithms, but to enhance user experience in Web3, further refinement of algorithm functionalities is needed.

In my view, the following are things that do not yet exist or are not mature in Web3: social graphs, user profiles, identity and reputation proofs, content aggregation, indexing and discovery, as well as curation and moderation methods. However, it is perplexing that the current application of blockchain technology in Web3 remains fragmented, lacking many features we expect to see on the web.

I hope that in the future, Web3 can have a broader definition that is recognized by more people, encompassing all types of self-certifying protocols, with all blockchains being just a branch of Web3.

However, rather than focusing on these terms, it is more important to pay attention to the attributes and characteristics that play a significant role in the development of the internet. At the same time, I hope that the future web can prioritize users, allowing people to have control over content creation decisions without being limited by any centralized services.

If you agree with my view but, for some reason, do not want to use the term "Web3," we can also refer to it as "self-certifying network protocols," abbreviated as SCWP, which can more specifically encapsulate these emerging technologies.

Related Reading:

《What will decentralized social media, which Aave and Twitter want to create, look like?》

《Lens Protocol: Web3 social media endorsed by the Aave team》

《How should the tech stack for "new social" in the Web3 era be constructed?》