I’m not sure if Amazon.com was the first company to offer crowd-sourced algorithmic suggestions of books I might like to read based on my browsing and buying history but certainly Amazon was one of the first.  I’ve written about this in prior posts – http://wp.me/pn6jX-37 , http://wp.me/pn6jX-75. I think Amazon has improved its algorithmic searches (at least I don’t get offers to buy an SAT prep book since my children have not taken an SAT any more recently than three years ago) and I do find them useful – although like almost all online retailers Amazon tends to go overboard with its frequency and amount of recommendations.

Spotify.com has a fairly robust integration with social media sites like Facebook (if you allow it) and I do find it interesting to see what my friends are listening to when I happen to notice. Just last Friday I received an email from Spotify (oh joy!) – ‘Hey Mark, we found 3 artists that you may like’. Based on the history of my listening to Bill Evans, Oscar Peterson, Coleman Hawkins, John Coltrane, Django Reinhardt and several other musicians, it was suggested that I might like Thelonius Monk, Charlie Parker and George Duke. Really? Of course I like them and have been listening to and enjoying their music for many years. The intimation is that I might not know who ‘Bird’ Parker was, and I find that annoying and (to me) it makes Spotify look dumb. It’s not as if the artists in question are current artists. All of these are giants of jazz and if you know one you at least have heard of them all if not listened to them all. It would be like saying because you listened to the Yardbirds you might like Eric Clapton. Again – Really?

I understand that algorithmic suggestions are still evolving as a platform and why data aggregators would argue that the more data people are willing to share the better and more relevant the suggestions on what you might like could/would be. However unintentionally insulting the user with suggestions of artists the user might like (with no way to know if the user is familiar with) just strikes me as annoying.

What I might suggest is using the algorithm to suggest current artists that may be of interest to me based on my listening history of artists from twenty or more years ago. That might be interesting and useful.

I still really like Spotify but think it is over-reaching. What do you think?