Hacker News new | ask | show | jobs
by kemenaran 1324 days ago
To find someone on Mastodon, I would either enter their nickname in the search bar (no need for the domain, it will search on all instances) – or simply click on their profile on a post I like. Just like a centralized network. This tool just exists to do this automatically and in bulk.

It seems to me that finding someone on a federation is as easy as on centralized systems. Do you have a use case in mind where centralization makes it easier to find someone?
2 comments
jacooper 1324 days ago
> To find someone on Mastodon, I would either enter their nickname in the search bar (no need for the domain, it will search on all instances)

Nope, It will only show accounts followed by local accounts.

Mastodon doesn't do any kind of per server search, And your "federated feed" is basically accounts on other instances followed by local accounts.

This makes hosting a private Mastodon server, for personal or family use, incredibly hard as discovery is basically impossible.

This can be fixed by using Activity pub relays, but guess what? Mastodon's official servers don't use them, which are the servers that really matter.

Its a big part of why I'm hesitant to use Activitypub, as I believe discovery is a big part of modern social media, and Mastodon absolutely sucks at it.

This is coming from someone hosting his own Matrix server.

link
masukomi 1324 days ago
it will not search "all instances".

It will search a subset of instances.

How do we know this to be true?

a) there is no central listing of all mastodon instances, or even all public ones.

b) it would take FREAKING AGES TO COME BACK because there are so damn many, you can only make so many parallel requests at a time, and you have to process the results from all of them.

link
qchris 1324 days ago
Can you expand on b)? I'm not sure I understand why this would be difficult. If you already had a list of instances, as it seems like asynchronous clients could probably handle the scale of instances (source below list about 4k, which isn't out of scope for simultaneous async tasks), and processing the results is probably fairly simple. Especially if you were to cache that list and just do an updated request every few minutes (so not a new full-network request set every query) it seems like this wouldn't be too crazy complicated. Have I missed something?

[1] https://fediverse.party/en/mastodon/

link