I had a asked the question of how much better is outbox than a static relay list two years ago.
No dev was interested to answer, so I did the benchmarks to find out.
Found one bug in an existing ...
that's great! did you reach any conclusions on the original question?
alphaama's outbox code is pretty straightforward if you're curious. here's a quick summary:
For each author pubkey in the query, look up their declared relays (from their kind-10002 events).
Build a map of relay → list of authors that use it. Skip relays turned off.
Each author is limited to at most "outbox_max" relays (default is 3), preferring relays with the most authors.
Your own read relays are always included with all authors as a fallback.
Authors with no known relays go into a "none" bucket that get spread across all selected relays.
alphaama's outbox code is pretty straightforward if you're curious. here's a quick summary:
For each author pubkey in the query, look up their declared relays (from their kind-10002 events).
Build a map of relay → list of authors that use it. Skip relays turned off.
Each author is limited to at most "outbox_max" relays (default is 3), preferring relays with the most authors.
Your own read relays are always included with all authors as a fallback.
Authors with no known relays go into a "none" bucket that get spread across all selected relays.
1