Local-First Computing

Local-first computing is a software design philosophy that treats the local device as primary: software should work fully offline, store data locally, and not depend on remote servers for core functionality, while still enabling collaboration and synchronisation when connectivity is available. It is a direct response to the architectural drift of the cloud era, in which user data increasingly resides on servers controlled by service providers, creating fragility (service outages render software unusable), dependency (switching costs are prohibitive because data is locked in), and loss of ownership (when a service closes, data may be inaccessible or lost entirely).

The defining text is Kleppmann, Wiggins, van Hardenberg and McGranaghan's 2019 paper "Local-first Software: You Own Your Data, in Spite of the Cloud", published in Onward!. The paper articulates seven ideals: (1) no spinners — instant responsiveness without network delay; (2) your work is not trapped in the cloud; (3) the network is optional; (4) seamless collaboration when connected; (5) the long now — data outlasts any given service and remains readable indefinitely; (6) security and privacy by default; (7) you retain ultimate ownership. The paper demonstrates that achieving these ideals simultaneously requires addressing a hard technical problem: how do you synchronise data across devices without a central server arbitrating conflicts?

The technical foundation enabling local-first software is CRDTs — Conflict-free Replicated Data Types — data structures that support concurrent editing across multiple devices without a central server by ensuring all replicas eventually converge to the same state, regardless of the order in which operations arrive. Marc Shapiro et al.'s 2011 paper formally characterised CRDTs and established their convergence properties. Martin Kleppmann's Designing Data-Intensive Applications (2017) situates them in the broader landscape of distributed systems — eventual consistency, partition tolerance, and the CAP theorem — providing the engineering context for understanding the tradeoffs in local-first architecture. The Ink & Switch research lab, which produced the original paper, has continued developing the ecosystem with open-source CRDT implementations (Automerge, Peritext) and ongoing research into collaborative local-first software patterns.

Local-first computing is not merely a technical preference — it is a political stance about the appropriate relationship between users and the software they depend on. The idealised cloud model — always-on, always-synced, always-accessible — obscures a power relationship in which service providers hold user data and can modify access terms, revoke access, or discontinue the service. Local-first software restores leverage to the user.