Comment 11 for bug 851424

I agree that a changing interface is hard to learn. How often do we look at other people's collection thought the lens? The only UI I need to learn is my own lens. On import, I would be impressed that the lens knows about my collection and assume everything works. Instead I saw half of the UI not right, so I had to ponder if my collection was wrong, or learn if the generic genres search a store -- total failure. If I saw all my genres in my collection, then saw someone else's collection, I would assume those are the genres the user has. Instead I see a set of genres that I might have learned do not map to the collection, they cannot be trusted, so I do not know which genres I can select to find music.

The problem with the current interface is that it fails more than 50% of the time for any time I use it -- it is still broken by design.
* The current design assumes there are a common set of genres in every users collection.
  * I assert every user will have genres or decades that do nothing.
* The current design assumes developers can maintain a mapping of sub-genres
  * I do not see the genres I have listed in the source code,
     but allmusic.com and wikipedia knows about my genras them,
  * The U1 music store genres are not listed in the lens genres;
    the Canonical lens dos not support the Canonical store!
  * Do I need to file a bug and harass a developer to add it?
  * Can the genres and mappings be translated?
  * Do mappings work in other cultures? are J-Pop and K-Pop always Pop?

The central issue here is that this is a filter operation over data that I know. The music lens is repeating the mistake of early desktop search apps. It wants to find everything and has present to assume the user is searching. Desktop data is not search. The user knows something exists and the UI should return the minimal set that matches the user's memory.