Here's some rambling thoughts:
MOG (now Beats) was the first streaming service I ever paid for. Then I switched to Rdio, then Spotify, and now I'm currently with Google Music All Access. I've been around, you could say.
Recommendations are such a difficult thing to talk about objectively. Algorithmically or interpersonally. No person or machine really knows what's inside your head. They're all just guesses. Sometimes the machine will show you something you like, and sometimes your friends will.
In terms of algorithms suggesting music, last.fm
has been the most consistent in recommending me stuff that I actually wind up liking. I can remember thinking on more than one occasion that Rdio wasn't even using my listening history to recommend stuff. Or they were only using one artist to recommend from. Last.fm seems to build their recs off of clusters of related artists in my listening history. Much more effective in my experience.
Spotify introduced their "discovery" feature right around the time I stopped using them so much. So I can't really speak to that. I will say that GMusic's radio stations tend to be much better for me than Spotify's ever were. I don't get the feeling that Spotify does much with all the data they collect.
I understand that Beats is trying to differentiate itself in a crowded market, but does anyone think that human-curated music is really sustainable in an industry based on machine learning? Those machines don't need health plans and 401Ks. It will be interesting how long the Beats model lasts. They are a good brand in terms of mind share, but I agree that the quality of their physical products doesn't live up to the price.
And a final beef with the author of the TechCrunch piece. Since when is $10/mo. steep? Spotify, Rdio, GMusic, Rhapsody, etc. all cost $10/mo. for their premium services. And $10/mo. is a damn sight cheaper than it used to be to buy two or three CDs every month like I -- and most of my music loving friends -- used to do. $120/year is a fucking steal.