I'm Starting A Search Engine For The Fediverse - eviltoast

Hey everyone,

This isn’t an announcement, just wanted peoples thoughts on this.

I think everyone knows searching the fediverse can be better. Googling doesn’t work too well, etc. So I wanted to do my part and help out.

Indexing all posts, etc is quite a lot to handle, so I wanted to start small and just focus on video search. I’ve started indexing videos from Peertube and other video websites. (Even YouTube but this could be removed to just focus on independent sites)

I know Peertube has their own search engine for videos. I will be reaching out to them. Compared to my site I’m planning it’ll have other video sources and be easier to use.

So that leads to feedback from you guys.

  • What do you think about indexing videos posted on the fediverse and other independent platforms?
  • Are there similar services?
  • Am I just wasting my time?
  • TimLovesTech (AuDHD)(he/him)@badatbeing.social
    link
    fedilink
    English
    arrow-up
    1
    ·
    11 months ago

    Reading a post and having a bot thrashing a server indexing everything are 2 different things. If a user used the site like that they would be throttled and if repeated afterwards, banned. It is also one thing to read/interact with a site as that adds value to the site as a whole. A bot that just mass hits links cataloging everything is just a strain on the server an Admin needs to support, with no upside for the instance, as it’s a bot ingesting and no real interaction actually took place.

    • 0x1C3B00DA@kbin.social
      link
      fedilink
      arrow-up
      1
      ·
      11 months ago

      and having a bot thrashing a server indexing everything

      This is a completely separate argument and one that we already have mechanisms for. Servers can use status codes and headers to warn about rate limits and block offenders.

      It is also one thing to read/interact with a site as that adds value to the site as a whole

      A search index adds value as well; that’s why this keeps coming up. And, again, there are existing mechanisms to handle this. A robots.txt file can indicate you don’t want to be crawled and offenders can be IP blocked

      • Rednax@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        11 months ago

        Should a dedicated search not use/index ActivityPub instead of the html interface?

        If so, instances can simply defederate from search engine instances. So the point you are trying to make still holds.