For everyone wondering what the outbox model is
In **Nostr** (Notes and Other Stuff Transmitted by Relays), the **Outbox Model** is a proposed approach to how relays and clients interact, intended to make the network more **scalable and efficient** than the traditional "inbox model".
Here’s the breakdown:
---
### 1. **The Traditional Model (Inbox Model)**
* In the usual Nostr setup, clients connect to relays and subscribe to filters (like: *give me all notes from pubkey X*, or *all events with tag Y*).
* Relays **push** all matching events down to the client.
* This can result in:
* Relays doing a lot of work serving different clients.
* Bandwidth-heavy duplicate traffic.
* Problems scaling when many users want the same data.
---
### 2. **The Outbox Model**
* Instead of asking relays to send everything, a client **fetches directly from the author’s chosen relays** — essentially pulling from their **outbox**.
* Each user "publishes" their notes/events to one or more relays. Those relays act as their **outbox servers**.
* When another client wants someone’s notes, they fetch them from the relays the author themselves announced (via something like `relay_list` events).
* So the client aggregates data itself, rather than relying on a big relay to aggregate.
---
### 3. **Benefits**
* **Scalability**: Less load on public relays, since they mostly serve content for the authors who chose them.
* **Sovereignty**: Users decide where their data lives (their chosen relays = their "outbox").
* **Spam resistance**: A relay only needs to host and serve its own users’ data, not arbitrary spam from everyone.
* **Redundancy**: Users can publish to multiple relays for resilience.
---
### 4. **Trade-offs**
* **Client responsibility**: Clients must discover which relays to fetch from (via metadata/relay lists).
* **More connections**: A client might have to connect to many small relays instead of just a few big ones.
* **Latency**: Aggregating posts across many relays may be slower.
---
👉 In short:
The **Outbox Model** in Nostr means each relay primarily stores and serves the events of its own users, and clients "go to the source" to fetch posts, instead of depending on relays to aggregate and push everything to them.