Or, you can click on the channel name underneath the title of an entry (Published by ___) and there should be a Subscribe/Unsubscribe button near the top of the page.
I'd like to have something like this where the up down votes started to put people into groups so that tribe A's likes were shared and tribe A's likes wouldn't affect the view of Tribe B, tribe C doesn't get affected by A or B and so on.
And it's an interesting idea! I'm not sure my unsupervised learning skills are good enough to be able to do this without creating some grotesque filter bubbles. If you have some ideas, please feel free to send me an email! I'd love to learn more.
Email: dpb@evalrss.com
When you upvote an item, you connect to other people who upvoted that item. When you downvote - you disconnect from those who upvoted it. So those who upvote useful content for you, get more weight in your recommendations.
Coincidentally, I'm doing a show HN post for it today: https://news.ycombinator.com/item?id=28405643, so will be happy to answer any questions there (instead of hijacking this post).
OP, does EvalRSS list content in reverse-chronological order? Any plans to prioritize content? What I do with LinkLonk is: the more items for a feed you upvote - the more other items from that feed are prioritized. Also, RSS readers have a notion of read/unread status. What are your thoughts on that? Is reverse-chronological ordering supposed to help the users distinguish unread content from read content?
Consider adding the number of comments each item in the list has. I added a test comment on https://www.evalrss.com/comments/9dcbaacb-96b7-4eec-8959-ba6... and there is no indication of that comment in the list.
Yes, evalRSS lists content in reverse chronological order in your feed by default. The "top" sorting options just filter out the entries submitted before the last n days and order by total votes. I think that displaying the number of comments each entry has is an excellent idea and I would definitely consider adding a read/unread status message to the entries.
To be completely honest, I have a bias against complex recommender systems because they frighten me. I'm not sure how to build one without assuming that people have some kind of static preference function, which feels like it can't be correct. So, I've been leaving content prioritization as a problem for my future self to solve. There could very well be some fancy mathematical technique to account for this, but I wasn't a good enough student to be familiar with it.