Do you think it’s venturing too far into algo territory to pick out keywords in notes and recommend “subjects to follow” in user onboarding? 🤷♂️
Part of me thinks this is a critical part of growing nostr. 🤔
Discussion
Cheeky
Namste #[1] 🙏
No but then again who’s picking out these words? If hellthread is teaching us anything, we need some algo implemented
Nobody is picking anything for the person. We just present some general interest topics: food, travel, games, movies, etc etc… and if a note mentions a word that might indicate it belongs to one of those categories, it gets shown under a general interest topic and people can follow / unfollow it just like they can with hashtags on snort
There has to be a balance. Iris for example shows users to follow and absolutely 0 content in global unless you follow people. That doesn’t really contribute to user discovery and hinders the onboarding.
A list of topics to follow might be nice. But on damus for example search is pretty bad and there’s no way to follow hashtags. The app is awesome but that needs to be addressed too. Most people won’t care to have 3 different clients for different things.
Maybe we don’t even suggest keywords, we could let people enter time. That’s similar to what AppleNews is doing. As opposed to limiting you to some topics you can basically create your own (ex: Nostr) and are able to access related news items in a single click.
Ahh yea that’s good, finding content based on topics could be useful, should add a way users can implement their own topics as well too
I can see it as tabs you can swipe between.
Ya I think eventually there needs to be some way to enable a feed that brings you to topics/convos that match your interest.
It would also be nice to have a feed that isn’t sorted by time of post but by zaps and/or by people you follow that you could “favorite” or something, so their posts from past 24 hrs float to top of feed.
As long as it’s not persistent and does not change what people see. The thing I don’t like about the “algorithm” is that it tries to influence you by showing you only what it wants you to see for whatever reason the algorithm was designed for. It’s basically control. Nostr should be about taking back control.
I think empowering people is a huge part of any onboarding…this would be an easy and non-invasive way of doing that.
No. They are necessary for relevance with the growing volume of content.
We should not be concerned by algos. We should be concerned by algos you can't control.
Algos done right are supposed to serve the user, and help him to find more relevant content. While being in charge of his personal definition of relevance.
Client have to start experimenting with these.
Natural progression is:
1) simple sorting (by time, by zaps (zaplife.lol), by likes
2) sorting with filters
3) simple sorting, or sorting by relevance (relevanceA vs relevanceB vs C, ..) With filters
4) mixes of all of the above.
Fwiw, the vision doc for Blue Sky discusses this nicely
I’ve never seen myself on zap life so I’m not sure how accurate it is 🤷♂️
As long as it is not enabled by default (opt-in), i think it's fine for some client who will implement this features.
I think as long as it doesn't prioritize one user over another it should be good and helpful?
It would recommend topics not users
Then what possible harm could it do? As long as it isn't like Twitter and shows the people with the largest followings first I think it's good!
There's a lot of wisdom and good points of view in the common pleb.
I agree that’s why I don’t think we should recommend people with most followers. Good example is Snowden who rarely shares anything. Good guy, like him, but we don’t need to recommend the same person who posts maybe every 4 days.
Any algo features that don't erode privacy or security and are OFF by default (and can be clearly enabled/disabled quickly and easily) are fine
Yes! I want many options like this. Talking about bitcoin is cool but I also like to talk about coding, AI, and meditation and I’d like to find more folks interested in those topics! Search via #damus only seems to work with hashtags of those topics (maybe I don’t know how to search properly with damus)