Thread

🛡️
This is a simple idea but I had never thought about it until now, so maybe some people may benefit from it: I always thought NIP-50 (search) was a hackish solution to an insurmountable problem, that of "global search". In order to search for something one must be like Google, one must index the entire network, it's centralizing and so on. And because of that it made little sense for normal relays to implement such a thing. I just now realized, after using @Gigi's
ants
advanced nostr text search
 many times, that the majority of searches are scoped by author, therefore it makes total sense for a client to query the author's outbox relays and send search queries to them directly instead of to some nameless global search engine -- which means that it makes total sense for small relays, personal relays, and random outbox relays of any kind to support search REQs.

Replies (7)

🛡️
Likewise, it also makes sense for custom relays that act as feeds to support NIP-50. It makes sense to search for things in specific relays, for example, if one is often browsing such relays' feeds, and want to recall something they saw at some point. And, of course, if a relay has a specific community or theme around it makes sense to search for it to learn what such community has said about a topic.
🛡️
In order to compete with other players in the social networks and messengers arena a powerful search engine is a must, especially in the era of AI. Reddit has a very good search engine, Telegram as well, Google made a living on it back in time etc
Another aspect where we can have a decentralized bunch of indexers is location specific data. Findings cars for ridesharing, restaurants, events and meetups, maps data, or even general posts about a location can all be indexed by indexers (or micro search engines if you want to call them that) without them knowing anything about data related to other locations View quoted note →