Thanks. I agree completely with
I don’t know what was said above as it was moderated before I read it.
I also agree that the choices should to some extent have to do with personal libraries, this makes total sense. My last sentence in the old thread was
And yes Roon can’t please everyone every time and there will be less than ideal choices. Some genres have overarching representatives that may have a very high weight. Few would argue that the Stones are wrong for Rock or Dylan for Singer/Songwriter. Choosing them may not represent my library perfectly and I might be more pleased by a more personal and still correct choice. but it’s not wrong.
However, an algorithm that chooses completely nonsensical pictures based on the apparent weighting of single albums, like the Gaga example for Roots Rock (uninfluenced by my library) has potential for improvement. I think this is obvious and can’t be argued away.
Sometimes this may be amusing or annoying, but there are cases where it can be problematic. The last years have seen many discussions about algorithms or pattern-matching “AIs” showing biases, exhibiting racism, sexism, etc. This is interesting and important stuff. Nobody accused Google developers to be racist in these cases, either, but the stuff they built could exhibit such traits. It is a topic that algorithm / AI developers must be conscious of these days.
The history of music is full of racism, of course, and people can be sensitive to this. When Paul Simon is chosen as the representative for African Traditions, it is problematic, and remains so even if a person only has Graceland from this genre in their library. But as @Rich_Borow showed above, this isn’t even the case for him. It seems that just like the Gaga case, it is simply swayed by Graceland being so successful generally.
I would argue that this is a topic that Roon should be conscious of. It may improve the choices in general and help avoid embarrassments.