Weighted Training

idea
#1

I’d love if the training feature allowed me to weight my likes and dislikes.

For instance, I subscribe to youtube channel feeds. So I tag “speedrun” as being liked. However, I hate streaming videos, so I tag “streaming” as disliked. However, some times “streaming speedrun” shows up, and I hate “streaming” more than I like “speedrun”. I wish I could tag “streaming” as -9999 and “speedrun” as +1.

Keep in mind there are also cases where they overlap, for instance I like javascript, but I hate java. So I’d tag “java” as -1 and “javascript” as +2.

I can keep going with examples: for instance, I like books about MySQL, but I hate “how to improve mysql performance [german]” because I don’t speak german.

By the way, there’s no need to get fancy with the focus sliders: anything above 0 gets marked green, anything below 0 gets marked red, and zero stays gray (even if it has 200 points for like and 200 points for dislike).

As for UI, you wouldn’t even need to change anything, just make it so that clicking the buttons multiple times adds or subtracts a point (instead of unmarking it, which barely makes sense anyways). If you’re worried that this might confuse users, I don’t think so because it would clearly show [+3] next to the tagged text, and even in the list it could show an item as red followed by [-2] or something.

4 Likes
#2

This is THE missing feature. Here, have my vote sir.

#3

I’ve written on this subject before but I’m still convinced that the current system of green takes all is right. The implicit assumption in training is that you subscribe to a feed because you want to read it, so if there is anything about a story that you like, it gets surfaced. And with that assumption the current training system was built.

#4

Sure, but that makes the system of reds superfluous…

#5

Yes, when I subscribe to a site it’s because I want their content. But when I train something down it’s because I don’t want that particular content, in some cases I really don’t care for it at all.

A good example would be training down a podcast a site has (be it because I don’t care about it or because I already subscribe in a podcast app) yet occasionally it’ll still show up because title had a topic I had trained up but I still have no interest in it.

The same can be just as true for series of articles, authors, or tags.

1 Like
#6

One more good example where green votes should not overrule red ones: Certain RSS Feeds have Ads in between, Usually they start with the same keywords. It would be great to filter these out even if there is >1 green tag