• Nostr Improvement Possibility — Discussion Laboratory v0.1
  • Human Description
    • What this is
  • Core Idea
  • Event Types
  • Anchoring Replies
  • Poll Events
  • Simple Models
  • Loose Conventions
  • How to Find These Events
  • Open Questions

§Nostr Improvement Possibility — Discussion Laboratory v0.1

§Human Description

§What this is

Nostr conversations generate a lot of interesting sparks — ideas, questions, rabbit holes worth exploring.

People often go off and look into those things on their own. Sometimes they run experiments, find sources, or learn something useful. But those results rarely make it back to the original discussion.

The effect is a bit like hearing a news story with no follow-up. You know something happened, but the outcome never quite reconnects to the place where the question started.

An alternate pattern might look like this:

• someone proposes an idea
• others explore it or challenge it
• experiments or research happen
• people report what they found
• eventually someone attempts a synthesis

In practice that sequence rarely appears.

Nostr already contains many of the pieces needed for it though. This document is an attempt to see whether a few simple conventions could help those pieces line up often enough to make discussions easier to follow and build on — in the open, without requiring specialized clients.

Nothing here is strict. These are loose conventions meant to help conversations produce useful knowledge.

§Core Idea

Event kind 30023 defines a living discussion document that can evolve over time while keeping related discussion anchored to a stable address.

Replies reference the document using an a tag (30023:<pubkey>:<d-tag>), allowing experiments, reports, and discussion to remain connected even if the root document updates.

Role tags (proposal, experiment, report, synthesis) signal the purpose of replies so clients and indexing systems can organize the discussion.

Example root event:

{
  "kind": 30023,
  "tags": [
    ["d", "topic-id"],
    ["title", "Example topic"],
    ["t", "proposal"]
  ],
  "content": "",
  "created_at": 0,
  "pubkey": "...",
  "sig": "..."
}

§Event Types

Notes may include tags describing their role in the discussion.

["t","proposal"]     # introduces an idea
["t","experiment"]   # describes an attempt or test
["t","report"]       # results or observations
["t","synthesis"]    # summary of current understanding

These tags are signals rather than strict requirements.

§Anchoring Replies

Replies SHOULD reference the root document using an address reference (a tag).

kind: 1

["a","30023:<pubkey>:<d-tag>"]

A relay hint MAY be included:

["a","30023:<pubkey>:<d-tag>","wss://relay.example"]

Using the address reference ensures replies remain connected even when the root document is updated.

§Poll Events

Polls and votes MAY be used as lightweight coordination tools.

• kind:1068 — Poll event (question and options)
• kind:1018 — Vote event (response selecting an option)

Example poll:

kind: 1068

Example vote:

kind: 1018

§Simple Models

Basic exploration

root document (kind:30023)
        │
        ├─ proposal
        ├─ experiment
        │      └─ report
        └─ synthesis

Discussion and friction

root document (kind:30023)
        │
        ├─ proposal
        │
        ├─ discussion
        │      ├─ counterpoint
        │      └─ clarification
        │
        ├─ experiment
        │      └─ report
        │
        └─ synthesis

§Loose Conventions

1. Root document MAY update as understanding improves.
2. Replies SHOULD reference the root address.
3. Experiments SHOULD include evidence when possible.
4. Reports SHOULD describe what actually happened.
5. Syntheses summarize the current state of the discussion.

§How to Find These Events

Discovery mechanisms are intentionally left open for experimentation.

Possible approaches may include:

• filtering by kind:30023 root documents
• indexing replies that reference the root via the a tag
• searching for role tags such as proposal, experiment, report, or synthesis
• relay or client-specific indexing strategies

Different clients and indexing systems may experiment with their own discovery methods.

§Open Questions

Future exploration may include:

• spam handling
• web-of-trust weighting
• contributor credit
• bounty distribution
• poll coordination
• AI summarization or indexing
• how discussion events are discovered or indexed

These are not required for the discussion laboratory to function.