Someone should connect this thread to the hole that TCTD left....
Algorithmic recommendations is okay, but human-selected materials is usually way more interesting.
This isn't necessarily a topic about code/design. We should just talk more about what we like, maybe? The tools are already there...
this.
plus, there would be quite a lot of work involved in developing a recommendations system and a lot of testing to get something that works well for the data we have. clustering based on similar likes might be interesting but would likely also introduce extra server overhead.