Can ‘taste engines’ and ‘recommendation engines’ cut through the clutter of the web? Or just serve us up an awkward hit-and-miss selection, based on the likes of our tasteless friends and the sort of clumsy clumping of artists/genres that can’t distinguish between Ziggy Bowie and Tin Machine Bowie? People often point to the system at Last.fm, but what does the research say? Some interesting quotes from the article “User Acceptance Issues in Music Recommender Systems” (2009) by Jones & Pu…

“Users perceived Last.fm’s recommendation technology as being less accurate … this is supported by post-study interviews where Last.fm users often reflected negatively on the accuracy during the post-study interviews” … “People only half agreed than ‘if similar technology existed for recommending other items (books, movies) then they would use it’.”

“Last.fm is clearly a successful website with more than ten million users. However, based on our results we believe that this does not primarily come from the recommender system which clearly poses some problems,” (Jones & Pu, 2009).