I sure hope they take a different approach with algos here, allow them to be completely customizable per user to the point that my algo is giving me a completely different feed than yours even if we follow similar people, having different interests means showcasing some of their content and ignoring other parts of the one you donβt care for. But like I said completely customizable by the user on the client level never the protocol level. Iβm sure everyone would be happy if something like that existed..
Discussion
example: letβs say someone follows me just for my memes and doesnβt care when I post news about Ai, Bitcoin, etc. if they wanted to they could choose to not see any of that content, yet still follow me and see the memes only.
Agreed, I like the way you spelled it out and thatβs a good example.
If weβre going to build, algorithms, then I would really like to see an algorithm promoted or suggested that promotes smaller accounts looking to engage with people and even steer people towards real life meet ups.
If I have to, Iβll help build it myself one way or another.
Call it βmake friendsβ mode π€£
Facts! nostr:npub1l2vyh47mk2p0qlsku7hg0vn29faehy9hy34ygaclpn66ukqp3afqutajft can this work ? pls read thread π«
certainly
you can't force that algorithm on people but you can certainly offer it to them
I think there's a lot of appetite for that
I will also make a DVM to find people who post thought-out notes but have little engagement
yeah like an opt in option + customizable af
would you have to work with clients to do this or is this gonna be a nip or something
nostr:npub1l2vyh47mk2p0qlsku7hg0vn29faehy9hy34ygaclpn66ukqp3afqutajft π₯ yes please!!
the NIP is just NIP-90, data vending machines
Imagine if on a client like e.g. Primal, the trending topics had a pagination button and you could paginate through hundreds of different DVMs with completely different algorithms
and imagine if someone, like me, were to build a site, like vendata, were you can see and interact (and choose which one is your favorite so other clients can use that) of countless algorithms
(each one of this DVMs takes maybe 30 minutes to write)
Yes π This. Use them for new user onboarding too to help people find accounts, communities, topics, etc., to follow.
oh yeah, 10000%
How friendly is this for someone like me, simple pleb to write a DVM of my liking ? Also just had an idea, what if you create a model n train it on this data for us simple plebs to write a prompt that creates the DVM or is that out of the question ? Just trying to find a way to make it SUPER simple π
Hmm, actually that could probably work. The code for these things is actually really simple.
In fact, right now Vendata allows you to provide as an input text for an LLM.
So you will be able to tell it:
"find people that have few likes but post interesting things about photography"
and literally have that DVM as one of your columns in e.g. primal or snort or something and it would run every X time and provide you with new people to follow
Ok now that is super exciting
This is freaking awesome!
Gonna give it a try π₯
https://vendata.io <β- for anyone reading and wanting to try!
just noticed I canβt sign in with nostr:npub1getal6ykt05fsz5nqu4uld09nfj3y3qxmv8crys4aeut53unfvlqr80nfm extension, which do you recommend ?
nostr:npub1yaul8k059377u9lsu67de7y637w4jtgeuwcmh5n7788l6xnlnrgs3tvjmf check this out!! π₯
Larger than life 
All in for progress my frogβ¦
You are the ill fluddy dutch!
π«ΆπΈ
