Today I spent some time fine-tuning the algorithm a little bit that we're implementing on scrobble.life, along with the upcoming consumer rewards it was important to get these things finished as much as we can before we start seeing results from usage. Naturally we'll be adjusting things more then, but this is what we've done today.
A fresh start
We didn't want too many past hive actions to influence what you see on scrobble.life, so the only thing taken into account for a starter on the "For You" section of /discover will be users you follow. Once you start using the site more however there'll be many different things affecting how the algo registers who's posts it recommends you in the future. We may even experiment with shortform content for the algo later but we wanna structure this properly first.
The issue that would've affected the algo how you don't want it to is the curation aspect of it all.
On scrobble.life we are also focusing on curation of posts so we can better find users worthy of curation and at the same time make it easy for consumers to recommend them for curation in the future since we are planning on making OCD nominations public soon.
When you are curating however and looking for posts you're interested in, you don't want that to affect the algo cause most of the time you'll just be looking for posts to curate. This may negatively adjust the algo simply based on the posts you're opening and reading to determine if it's genuine/worthy of curation, etc.
Thus we had to remove those sections from affecting your algorithm.
In the screenshot you can see that we also added a one-time text that tells you how things will work. For other sections however, such as /originalcontent, /awww, your "following" feed and more coming in the future will affect your "for you" feed (algo). So I want people to remember how the algo works and how it may affect their browsing experience.
We are also preparing to opensource how the algo works, even though it is quite simplistic currently, we want to ensure our readers that outside things aren't going to affect their algo the way other web2 platforms may. For instance the recent filtration of crypto content on X that hid a lot of content people wanted to see from their algo feed because of stigma/whatever reason the runners of the platform decided to stifle the algo of such content. We want people to know that only they themselves are in control of how the algo grows and operates and they will be able to select "not interested in this author/content/genre" if the algo ever overreaches or mistakingly adds things you don't want to see.
For you & following
Will be quite the same early on until you browse enough on the site which then changes the "for you" feed over time. Following will remain similar to how most other front-ends show it, descending based on post account age of users you follow.
We'll also be experimenting with lists/stars so you can have custom feeds if you don't have a lot of time to check your feed that day and are only interested in your favorites, etc.
We're also open to adding other filters on Discovery based on demand, so feel free to let us know what else you'd be interested in seeing there.
Let us know what you think about this, we're excited to see consumer rewards start to trickle in soon and how that'll work out!
Just a thought on the interest bit, could it be possible to add a preference system instead, where users can choose what they want to see? It might be a bit more of a positive approach and make their experience a bit more personal.
By choosing “not interested in this author” or “not interested in this topic,” etc., it could feel a bit unfriendly. Just a thought though. Maybe I’m wrong. Let's see what others say. 😀
Finding the right balance between discovery and authenticity is never easy.
Congratulations, may the improvements continue.
Greetings, I'll be keeping an eye on things as always, and I'll be checking the site; I've liked the updates. Can I interact with content from all of Hive through scrobblelife, or only with content uploaded through the scrobblelife website?
Did you know that the old Facebook feed worked without an algorithm and mostly just showed whatever was popular? Now, they show users whatever they decide they want to show them.
I think this is a good idea, and it would make it easier for lots of people to contribute since posting would be quick and simple, similar to Threads. But if we're linking to studies, news, or other resources as I plan to do, I think longer-form posts would work better because they provide more context around the links. Having both is good. Just a thought. :)