3 ms·

All they would have to do, in order to have a functional and transparent recommender 'algorithm' is to rank order articles by how many of your followers liked,

All they would have to do, in order to have a functional and transparent recommender 'algorithm' is to rank order articles by how many of your followers liked, or read, an article... so if you have 100 followers and 50 of them happened to like another article 'B' then that article would be highly recommended to people reading your article. If they knew who was reading the article (logged in) then they could simply exclude articles from the list which the person had already viewed. That would help people browse through groups that shared similar interests.

Alir3z4 7y ago
Thing is they want to put more recommended content to your face as much as possible without knowing much about you, otherwise your scroll bar will hit the bottom quickly. The recommendation engine from technical point of view is not a rocket science, they're just simply trying to tune for maximum engagement even if it costs them repeated nonsense content you don't like which tbh defeats the purpose of "recommendation". It seems there's not much of community in Medium. All viral articles are the ones that the author had a good influence over their twitter of facebook etc audience. If you already have 10s thousands on your twitter follower list that you worked hard to find and bunch of people know you for some of your other work, no matter Medium or a static generated blog, your article will go hand to hand and viral in many cases.