Interesting debate. Here’s my take on it… which is (as are most things on the internet) my opinions, not facts…
Algorithms are good - up to a point. Like categorising music, you can figure out certain things about a track using algorithms. Like the key it is in; tempo; identify parts like verse, chorus, bridge etc. You can make inferences from it - is it “uplifting” or not. But there are limits to what an algorithm can do in terms of feature engineering. The algorithms will work better with known good features instead of trying to extract them. There is a streaming company who suggests great playlists. The algorithm may be like valence, but the source data - the music features - are human curated. No computer figuring “I think this is in an uplifting key, has an upbeat tempo…” but someone augmenting what can be machine generated with human interpretation. “This is a song about love, about failed relationships…” so the songs suggested follow themes that are not computer extractable. Maybe we are getting towards where we can do speech recognition or parse in the lyrics and then run some sort of deep text analysis on it to get to the same result.
Human curation is possible - and the sentiment here seems to be that roon should be doing it. That comes at a cost. Complaining about the monthly / annual fees now? Hire a team of folks to curate the music… it isn’t going to go well.
The other argument seems to be “why am I being exploited here?”. Go on Facebook. By virtue of being there you’re creating data to be exploited. Click like on something? Ditto. Post something? Same again. Use a search engine to search the web…
As I see it the roon approach is less nefarious than that, and you’re going into it with your eyes open. Plus it is voluntary. If you feel hard done by - fair enough. This isn’t the gig economy and we aren’t driving for a rideshare company. Contribute if you want.
I’ll leave the soapbox by the door as I show myself out 