Open Social: Users Own Data, Apps Aggregate

The author proposes “open social,” exemplified by the AT Protocol, to restore user ownership and portability of social data. Users keep their identity and records in personal repositories addressable by at://, while apps simply aggregate, index, and remix that data. This preserves network features like feeds and search without lock‑in, enabling forkable products and lasting data beyond any single platform.

Key Points

• Closed social traps the social graph inside proprietary databases, making meaningful data ownership and platform exit without loss effectively impossible.
• Open social (via AT Protocol) puts user data in personal repositories under user-owned handles (e.g., @alice.com), with records as signed JSON addressable by at:// URIs.
• Apps write to and read from users’ repos and can freely remix each other’s public data because the protocol itself is the API.
• Scalable aggregation is achieved by subscribing to network-wide event streams (relays) and indexing signed commits into local databases, preserving integrity and performance.
• This model enables hosting independence, product forkability, reduced cold starts, and long-term data persistence beyond any single app’s lifecycle.

Sentiment

The discussion is predominantly positive, with many commenters expressing genuine enthusiasm and describing the article as persuasive and well-written. However, there is substantive technical skepticism about privacy limitations, centralization risks around Bluesky's dominance, and pragmatic concerns about whether regular users will ever value data ownership enough to drive adoption. The discussion is constructive and informed, significantly elevated by the author's active participation in addressing technical questions.

In Agreement

• The article fundamentally changed many readers' understanding of AT Protocol, revealing it as a genuinely superior approach to ActivityPub for shared identity across multiple apps and meaningful data ownership
• The 'publish all, filter on read' philosophy empowers users while preventing platform lock-in, creating better incentive structures
• AT Protocol's lower exit costs force platforms to compete on quality rather than lock-in, potentially ending the enshittification cycle
• The open source analogy resonates: atproto could be the last migration needed, after which app competition happens on a shared protocol layer
• Building on atproto is accessible and exciting, with practical pathways for running personal data servers and appviews independently of Bluesky

Opposed

• AT Protocol's global public data model requires trusted intermediaries for aggregation, making private data fundamentally difficult unlike ActivityPub's local feed-subscription model
• Bluesky's near-total dominance among AT Protocol users creates the same concentration risk as centralized platforms; a corporate abandonment would be a social problem that technical portability cannot solve
• Domain 'ownership' is really just renting from registrars, and the DID:PLC identity system remains centralized under Bluesky's control
• Average users don't care about data ownership or portability; they want to be entertained and prefer curated walled gardens
• Each app having its own collection types limits natural interoperability compared to ActivityPub's shared types