Ink & Switch

One-line summary: Independent research lab that named local-first software — device replica primary, server secondary — in a 2019 Onward! essay.

What it is

A lab whose 2019 essay (Kleppmann, Wiggins, van Hardenberg, McGranaghan) argues that cloud apps give collaboration but not ownership, and that CRDTs can support both. From 2026-08-21-autoresearch-best-architecture-for-a-rec-system-web-app.

Why it matters to interest-graph-recs

Their local-first contract is one cited architecture for a user-owned interest graph behind a personal recs UI: the graph on the device is authoritative; sync is optional. Not a product stack for this thread.

Key facts

• Essay title: “Local-first software: you own your data, in spite of the cloud” (Onward! 2019). From 2026-08-21-autoresearch-best-architecture-for-a-rec-system-web-app.
• Cloud apps treat the server as the primary copy and the client as a cache. Local-first swaps those roles. Same source.
• Named sync foundation: CRDTs; their JS implementation is Automerge. They do not claim Automerge fully realizes the ideals. Same source.

Open questions

• what-web-app-architecture-fits-a-user-owned-interest-graph-recommender

Sources

• 2026-08-21-autoresearch-best-architecture-for-a-rec-system-web-app

Related

• local-first-software
• user-owned-interest-graph
• solid-project