When considering building a local-first app it's important to know which parts of the stack "just work", and which are still experimental.
In this post on the @PrototypeFund blog we give an overview of the various moving parts, using Reflection as an example: https://www.prototypefund.de/en/blog/the-road-to-a-local-first-app-ecosystem
tl;dr:
Mostly just works
✅ CRDTs (Loro, Automerge, Yrs)
✅ P2P networking (iroh)
✅ Ephemeral messaging (e.g. for cursor position)
✅ Local-first sync for eventual consistency
✅ Confidential topic discovery
✅ Token-based access (non-revocable)
WIP
🏗️ End-to-end encryption for groups
🏗️ Fine-grained, revocable access controls
🏗️ Multi-device management
🏗️ IP address privacy (Tor)
Experimental
🧪 Support nodes
🧪 Identity & contacts
🧪 Data migrations
🧪 Mesh networking (Bluetooth, LoRa)
For many types of apps, to actually replace a server you realistically need all of the above. Other apps can be built with only the things that work right now though, and there are shortcuts you can take to approximate things.
For example, this new concept for Reflection solves a lot of the pain points we found in user testing without requiring any additional local-first tech: https://github.com/p2panda/reflection/discussions/234
@modal interesting stuff. Call me a doomer, but we really need to prepare for a time where Internet is not available at all times.
The prospects of (cyber) warfare (not just from the east) or e.g. climate related catastrophe are not as abstract as they once were, and I’m increasingly getting the feeling that we should prepare to adapt to new realities.
Not saying a collapse is a given. But these initiatives feel important for our resilience as society. Very cool work.
🐡🏳️🌈
Modal Collective
Profile13115
Val Packett 🧉
Sjoerd Stendahl