Thanks for the feedback. I want to speak a little bit about why we did this, since the decision was not based on graphical considerations at all.
In older versions, Roon displayed a count for streaming services and a separate count for your library. This was consistent with the concepts in that product–back then, since we enforced a regimented separation of Library and Streaming content. The counts on the album page corresponded to the counts displayed in the library and streaming tabs on the composition page, and it was crisp and consistent.
One of the big themes for 1.8 was de-siloing the library content and streaming content. Now, we try to make clean displays of “all of the stuff that I have access to”. For the “show my my stuff” use cases, we provide a library-only filter in all of these new merged views, since we know that most of our users treasure their music libraries and need a first class way to zoom in on their stuff.
After performing this de-silo-ing, the segregated counts on the album screen started to look and feel wrong. The numbers didn’t really correspond to anything tangible anymore. We thought about displaying a single merged count, but it turns out that that number is difficult to compute–we would have to load the whole list of streaming recordings, and then deduplicate it against the stuff in your library to avoid double counting the library items.
It’s ok to do this for one composition at a time when loading the composition page, but it isn’t feasible to load the equivalent ten or fifty composition pages at once to perform that deduplication when displaying an album. Either the album page would become slow, or the counts would be inaccurate.
(Roon is absolutely loaded with considerations like this–merging and deduplicating library content with streaming in a coherent way is a difficult problem, and the feasibility considerations can be arduous. The Spotify’s of the world have it easy–their entire experience is driven by one catalog of music per country. We have to cope with a private catalog of music per user plus a union of one or more streaming service catalogs, depending on what people have logged into).
Back on topic–there is something else that has always bothered us about the old composition links–the numbers were difficult to understand without a deep understanding of the norms for that type of music.
A popular song with 150 recordings is important, but an American Songbook composition with only 150 recordings is insignificant. For a large-scale classical work, 150 is a middle-of-the-road number. Not everyone is an expert in every area, and the numbers could be misleading. Gershwin’s Summertime has around 5,000 recordings, but isn’t 15 times more important than Beethoven’s 5th. So while the numbers were very accurate from a technical standpoint, they were not communicating the right thing.
Eliminating the precise number opened up the opportunity to normalize the indicator scale based on context. It’s harder to earn five discs in the American Songbook, a bit easier for a Beethoven Symphony, and easier yet for popular music. So for the most part, you can read the indicator and trust that it is identifying things that are a little bit interesting, or very interesting without having to know offhand what number constitutes “a lot” in this context.
Happy to discuss this more–just wanted to shed a little bit of light on what happened here.