Interesting read Peter and Yaron, very contextual to what I am doing these days :)
I believe both are same things, in-fact serendipity is a derivative of contextual understanding of users and recommending same.
Agree that users do not know what they like and discover same along the journey, kind of principal of duality from ‘quantum cognition’
However, there is one more fundamental component to answering “What should I show the user he would love to read?”
I believe, recommendation system are great at producing content, however long way to go. Hence Intelligent mix of content, derived from variety of methods like below and presented like Amazon’s art of making its home and product pages, is the key.
- Editorial Recommendation
- Peer Recommendation
- Machine Recommendation
- Novelty / Recency
- Probabilistic model of what various personas read in general
- Short Term Interests
and ensuring users do not go blind on some content, hence on demand option.
Learning from how eCommerce giants make people to like things and buy them instantaneously, just by great art of visual merchandising can fill the missing link. May be :)
at least we believe so.